diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml index a6534f5..92a62e2 100644 --- a/.github/workflows/build.yml +++ b/.github/workflows/build.yml @@ -24,6 +24,8 @@ jobs: sudo rm -rf /var/lib/apt/lists/* sudo sed -i 's/# \(en_US.UTF-8\)/\1/' /etc/locale.gen sudo locale-gen --purge --lang en_US.UTF-8 + - name: Build SDK + run: make sdk - name: Build FunKey-OS run: make - uses: actions/upload-artifact@v2 diff --git a/.gitmodules b/.gitmodules index 4f2c72a..4bf7ba1 100644 --- a/.gitmodules +++ b/.gitmodules @@ -1,3 +1,3 @@ [submodule "buildroot"] path = buildroot - url = git://git.buildroot.net/buildroot + url = https://github.com/FunKey-Project/buildroot.git diff --git a/FunKey/Config.in b/FunKey/Config.in index 61b75af..8b1004a 100644 --- a/FunKey/Config.in +++ b/FunKey/Config.in @@ -9,3 +9,11 @@ source "$BR2_EXTERNAL_FUNKEY_PATH/package/PocketSNES/Config.in" source "$BR2_EXTERNAL_FUNKEY_PATH/package/gpsp/Config.in" source "$BR2_EXTERNAL_FUNKEY_PATH/package/dmtx-utils/Config.in" source "$BR2_EXTERNAL_FUNKEY_PATH/package/ProdScreens/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/fonts-droid/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/libini/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/libopk/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/libxdgmime/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/gmenu2x/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/agg/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/fluidlite/Config.in" +source "$BR2_EXTERNAL_FUNKEY_PATH/package/libmikmod/Config.in" diff --git a/FunKey/board/funkey/busybox.config b/FunKey/board/funkey/busybox.config index 35e3f35..70b5598 100644 --- a/FunKey/board/funkey/busybox.config +++ b/FunKey/board/funkey/busybox.config @@ -1,7 +1,7 @@ # # Automatically generated make config: don't edit -# Busybox version: 1.31.1 -# Mon Nov 9 00:34:05 2020 +# Busybox version: 1.32.0 +# Fri Jan 15 10:58:20 2021 # CONFIG_HAVE_DOT_CONFIG=y @@ -76,6 +76,7 @@ CONFIG_PREFIX="./_install" # CONFIG_DEBUG_SANITIZE is not set # CONFIG_UNIT_TEST is not set # CONFIG_WERROR is not set +# CONFIG_WARN_SIMPLE_MSG is not set CONFIG_NO_DEBUG_LIB=y # CONFIG_DMALLOC is not set # CONFIG_EFENCE is not set @@ -462,6 +463,7 @@ CONFIG_FEATURE_FIND_SIZE=y CONFIG_FEATURE_FIND_PRUNE=y CONFIG_FEATURE_FIND_QUIT=y # CONFIG_FEATURE_FIND_DELETE is not set +CONFIG_FEATURE_FIND_EMPTY=y CONFIG_FEATURE_FIND_PATH=y CONFIG_FEATURE_FIND_REGEX=y # CONFIG_FEATURE_FIND_CONTEXT is not set @@ -692,6 +694,7 @@ CONFIG_FEATURE_SETPRIV_CAPABILITY_NAMES=y CONFIG_SWITCH_ROOT=y # CONFIG_TASKSET is not set # CONFIG_FEATURE_TASKSET_FANCY is not set +# CONFIG_FEATURE_TASKSET_CPULIST is not set CONFIG_UEVENT=y CONFIG_UMOUNT=y CONFIG_FEATURE_UMOUNT_ALL=y @@ -805,6 +808,7 @@ CONFIG_MAKEDEVS=y CONFIG_FEATURE_MAKEDEVS_TABLE=y # CONFIG_MAN is not set CONFIG_MICROCOM=y +CONFIG_MIM=y CONFIG_MT=y # CONFIG_NANDWRITE is not set # CONFIG_NANDDUMP is not set @@ -1095,6 +1099,7 @@ CONFIG_SH_IS_ASH=y # CONFIG_BASH_IS_ASH is not set # CONFIG_BASH_IS_HUSH is not set CONFIG_BASH_IS_NONE=y +CONFIG_SHELL_ASH=y CONFIG_ASH=y CONFIG_ASH_OPTIMIZE_FOR_SIZE=y CONFIG_ASH_INTERNAL_GLOB=y @@ -1115,6 +1120,7 @@ CONFIG_ASH_GETOPTS=y CONFIG_ASH_CMDCMD=y # CONFIG_CTTYHACK is not set # CONFIG_HUSH is not set +# CONFIG_SHELL_HUSH is not set # CONFIG_HUSH_BASH_COMPAT is not set # CONFIG_HUSH_BRACE_EXPANSION is not set # CONFIG_HUSH_LINENO_VAR is not set @@ -1177,6 +1183,7 @@ CONFIG_FEATURE_ROTATE_LOGFILE=y CONFIG_FEATURE_REMOTE_LOG=y # CONFIG_FEATURE_SYSLOGD_DUP is not set # CONFIG_FEATURE_SYSLOGD_CFG is not set +# CONFIG_FEATURE_SYSLOGD_PRECISE_TIMESTAMPS is not set CONFIG_FEATURE_SYSLOGD_READ_BUFFER_SIZE=256 # CONFIG_FEATURE_IPC_SYSLOG is not set CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE=0 diff --git a/FunKey/board/funkey/linux.config b/FunKey/board/funkey/linux.config index 8ff92a7..5cb4c06 100644 --- a/FunKey/board/funkey/linux.config +++ b/FunKey/board/funkey/linux.config @@ -25,6 +25,7 @@ CONFIG_CPU_FREQ=y CONFIG_CPUFREQ_DT=y CONFIG_VFP=y CONFIG_NEON=y +CONFIG_KERNEL_MODE_NEON=y # CONFIG_COREDUMP is not set CONFIG_HIBERNATION=y CONFIG_PM_STD_PARTITION="/dev/mmcblk0p2" @@ -152,7 +153,7 @@ CONFIG_AUTOFS4_FS=y CONFIG_VFAT_FS=y CONFIG_TMPFS=y CONFIG_CONFIGFS_FS=y -# CONFIG_MISC_FILESYSTEMS is not set +CONFIG_SQUASHFS=y # CONFIG_NETWORK_FILESYSTEMS is not set CONFIG_NLS_CODEPAGE_437=y CONFIG_NLS_ISO8859_1=y diff --git a/FunKey/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch b/FunKey/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch new file mode 100644 index 0000000..1f27440 --- /dev/null +++ b/FunKey/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch @@ -0,0 +1,6526 @@ +From 7b128ae8c56b1055a93573004148a98465d79857 Mon Sep 17 00:00:00 2001 +From: Mizuki Asakura +Date: Sun, 17 Apr 2016 20:16:12 +0900 +Subject: [PATCH] [mod] added aarch64 bilinear implementations (ver.4.1) + +Since aarch64 has different neon syntax from aarch32 and has no +support for (older) arm-simd, +there are no SIMD accelerations for pixman on aarch64. + +We need new implementations. + +This patch also contains Ben Avions's series of patches for aarch32 +and now the benchmark results are fine to aarch64. + +Please find the result at the below ticket. + +Added: https://bugs.freedesktop.org/show_bug.cgi?id=94758 +Signed-off-by: Mizuki Asakura +--- + configure.ac | 36 +- + pixman/Makefile.am | 17 +- + pixman/pixman-arm-neon.c | 23 +- + pixman/pixman-arm.c | 6 + + pixman/pixman-arma64-neon-asm-bilinear.S | 1275 ++++++++++ + pixman/pixman-arma64-neon-asm.S | 3704 ++++++++++++++++++++++++++++++ + pixman/pixman-arma64-neon-asm.h | 1310 +++++++++++ + pixman/pixman-private.h | 7 +- + 8 files changed, 6374 insertions(+), 4 deletions(-) + create mode 100644 pixman/pixman-arma64-neon-asm-bilinear.S + create mode 100644 pixman/pixman-arma64-neon-asm.S + create mode 100644 pixman/pixman-arma64-neon-asm.h + +diff --git a/configure.ac b/configure.ac +index 6b2134e..26203a8 +--- a/configure.ac ++++ b/configure.ac +@@ -667,6 +667,40 @@ if test $enable_arm_neon = yes && test $have_arm_neon = no ; then + AC_MSG_ERROR([ARM NEON intrinsics not detected]) + fi + ++dnl ========================================================================== ++dnl Check if assembler is gas compatible and supports ARM-a64 NEON instructions ++have_arm_a64_neon=no ++AC_MSG_CHECKING(whether to use ARM A64 NEON assembler) ++xserver_save_CFLAGS=$CFLAGS ++CFLAGS="-x assembler-with-cpp $CFLAGS" ++AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ ++.text ++.arch armv8-a ++.altmacro ++prfm pldl2strm, [x0] ++xtn v0.8b, v0.8h]])], have_arm_a64_neon=yes) ++CFLAGS=$xserver_save_CFLAGS ++ ++AC_ARG_ENABLE(arm-a64-neon, ++ [AC_HELP_STRING([--disable-arm-a64-neon], ++ [disable ARM A64 NEON fast paths])], ++ [enable_arm_a64_neon=$enableval], [enable_arm_a64_neon=auto]) ++ ++if test $enable_arm_a64_neon = no ; then ++ have_arm_a64_neon=disabled ++fi ++ ++if test $have_arm_a64_neon = yes ; then ++ AC_DEFINE(USE_ARM_A64_NEON, 1, [use ARM A64_NEON assembly optimizations]) ++fi ++ ++AM_CONDITIONAL(USE_ARM_A64_NEON, test $have_arm_a64_neon = yes) ++ ++AC_MSG_RESULT($have_arm_a64_neon) ++if test $enable_arm_a64_neon = yes && test $have_arm_a64_neon4 = no ; then ++ AC_MSG_ERROR([ARM A64 NEON intrinsics not detected]) ++fi ++ + dnl =========================================================================== + dnl Check for IWMMXT + +diff --git a/pixman/Makefile.am b/pixman/Makefile.am +index 581b6f6..f1afa27 +--- a/pixman/Makefile.am ++++ b/pixman/Makefile.am +@@ -94,6 +94,21 @@ libpixman_1_la_LIBADD += libpixman-arm-neon.la + ASM_CFLAGS_arm_neon= + endif + ++# arm a64 neon code ++if USE_ARM_A64_NEON ++noinst_LTLIBRARIES += libpixman-arma64-neon.la ++libpixman_arma64_neon_la_SOURCES = \ ++ pixman-arm-neon.c \ ++ pixman-arm-common.h \ ++ pixman-arma64-neon-asm.S \ ++ pixman-arma64-neon-asm-bilinear.S \ ++ pixman-arm-asm.h \ ++ pixman-arma64-neon-asm.h ++libpixman_1_la_LIBADD += libpixman-arma64-neon.la ++ ++ASM_CFLAGS_arm_neon= ++endif ++ + # iwmmxt code + if USE_ARM_IWMMXT + libpixman_iwmmxt_la_SOURCES = pixman-mmx.c +diff --git a/pixman/pixman-arm-neon.c b/pixman/pixman-arm-neon.c +index be761c9..62c9442 100644 +--- a/pixman/pixman-arm-neon.c ++++ b/pixman/pixman-arm-neon.c +@@ -194,7 +194,7 @@ arm_neon_fill (pixman_implementation_t *imp, + uint32_t _xor) + { + /* stride is always multiple of 32bit units in pixman */ +- uint32_t byte_stride = stride * sizeof(uint32_t); ++ int32_t byte_stride = stride * sizeof(uint32_t); + + switch (bpp) + { +@@ -331,6 +331,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, a8, b5g6r5, neon_composite_over_8888_8_0565), + PIXMAN_STD_FAST_PATH (OVER, r5g6b5, a8, r5g6b5, neon_composite_over_0565_8_0565), + PIXMAN_STD_FAST_PATH (OVER, b5g6r5, a8, b5g6r5, neon_composite_over_0565_8_0565), ++ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_over_8888_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, a8r8g8b8, neon_composite_over_8888_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, r5g6b5, neon_composite_over_8888_0565), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, b5g6r5, neon_composite_over_8888_0565), +@@ -341,17 +342,33 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, null, a8r8g8b8, neon_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, null, a8b8g8r8, neon_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8, neon_composite_add_n_8_8), ++ PIXMAN_STD_FAST_PATH (ADD, solid, a8, x8r8g8b8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8r8g8b8, neon_composite_add_n_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, solid, a8, x8b8g8r8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8b8g8r8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, a8, a8, neon_composite_add_8_8_8), + PIXMAN_STD_FAST_PATH (ADD, r5g6b5, a8, r5g6b5, neon_composite_add_0565_8_0565), + PIXMAN_STD_FAST_PATH (ADD, b5g6r5, a8, b5g6r5, neon_composite_add_0565_8_0565), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, a8, x8r8g8b8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8, x8r8g8b8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, a8, x8b8g8r8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, a8, x8b8g8r8, neon_composite_add_8888_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8, a8r8g8b8, neon_composite_add_8888_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, a8, a8b8g8r8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_add_8888_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_add_8888_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, a8r8g8b8, neon_composite_add_8888_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, solid, x8r8g8b8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, solid, x8r8g8b8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, solid, x8b8g8r8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, solid, x8b8g8r8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, solid, a8r8g8b8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, solid, a8b8g8r8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, neon_composite_add_8_8), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, null, x8r8g8b8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, x8r8g8b8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, null, x8b8g8r8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, x8b8g8r8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (IN, solid, null, a8, neon_composite_in_n_8), +@@ -359,7 +376,9 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8b8g8r8, neon_composite_over_reverse_n_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, r5g6b5, neon_composite_out_reverse_8_0565), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, b5g6r5, neon_composite_out_reverse_8_0565), ++ PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, x8r8g8b8, neon_composite_out_reverse_8_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8r8g8b8, neon_composite_out_reverse_8_8888), ++ PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, x8b8g8r8, neon_composite_out_reverse_8_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8b8g8r8, neon_composite_out_reverse_8_8888), + + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, neon_8888_8888), +@@ -404,6 +423,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + + SIMPLE_BILINEAR_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, neon_8888_8888), + SIMPLE_BILINEAR_FAST_PATH (ADD, a8r8g8b8, x8r8g8b8, neon_8888_8888), ++ SIMPLE_BILINEAR_FAST_PATH (ADD, x8r8g8b8, x8r8g8b8, neon_8888_8888), + + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, neon_8888_8_8888), + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, neon_8888_8_8888), +@@ -420,6 +440,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, neon_8888_8_8888), + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, a8r8g8b8, x8r8g8b8, neon_8888_8_8888), ++ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, x8r8g8b8, x8r8g8b8, neon_8888_8_8888), + + { PIXMAN_OP_NONE }, + }; +diff --git a/pixman/pixman-arm.c b/pixman/pixman-arm.c +index 23374e4..734cbea 100644 +--- a/pixman/pixman-arm.c ++++ b/pixman/pixman-arm.c +@@ -221,5 +221,11 @@ _pixman_arm_get_implementations (pixman_implementation_t *imp) + imp = _pixman_implementation_create_arm_neon (imp); + #endif + ++#ifdef USE_ARM_A64_NEON ++ /* neon is a part of aarch64 */ ++ if (!_pixman_disabled ("arm-neon")) ++ imp = _pixman_implementation_create_arm_neon (imp); ++#endif ++ + return imp; + } +diff --git a/pixman/pixman-arma64-neon-asm-bilinear.S b/pixman/pixman-arma64-neon-asm-bilinear.S +new file mode 100644 +index 0000000..aaa4a83 +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm-bilinear.S +@@ -0,0 +1,1275 @@ ++/* ++ * Copyright © 2011 SCore Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ * Author: Taekyun Kim (tkq.kim@samsung.com) ++ */ ++ ++/* ++ * This file contains scaled bilinear scanline functions implemented ++ * using older siarhei's bilinear macro template. ++ * ++ * << General scanline function procedures >> ++ * 1. bilinear interpolate source pixels ++ * 2. load mask pixels ++ * 3. load destination pixels ++ * 4. duplicate mask to fill whole register ++ * 5. interleave source & destination pixels ++ * 6. apply mask to source pixels ++ * 7. combine source & destination pixels ++ * 8, Deinterleave final result ++ * 9. store destination pixels ++ * ++ * All registers with single number (i.e. src0, tmp0) are 64-bits registers. ++ * Registers with double numbers(src01, dst01) are 128-bits registers. ++ * All temp registers can be used freely outside the code block. ++ * Assume that symbol(register .req) OUT and MASK are defined at caller of these macro blocks. ++ * ++ * Remarks ++ * There can be lots of pipeline stalls inside code block and between code blocks. ++ * Further optimizations will be done by new macro templates using head/tail_head/tail scheme. ++ */ ++ ++/* Prevent the stack from becoming executable for no reason... */ ++#if defined(__linux__) && defined (__ELF__) ++.section .note.GNU-stack,"",%progbits ++#endif ++ ++.text ++.arch armv8-a ++.altmacro ++.p2align 2 ++ ++#include "pixman-private.h" ++#include "pixman-arm-asm.h" ++#include "pixman-arma64-neon-asm.h" ++ ++/* ++ * Bilinear macros from pixman-arm-neon-asm.S ++ */ ++ ++/* ++ * Bilinear scaling support code which tries to provide pixel fetching, color ++ * format conversion, and interpolation as separate macros which can be used ++ * as the basic building blocks for constructing bilinear scanline functions. ++ */ ++ ++.macro bilinear_load_8888 reg1, reg2, tmp ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ ld1 {®1&.2s}, [TMP1], STRIDE ++ ld1 {®2&.2s}, [TMP1] ++.endm ++ ++.macro bilinear_load_0565 reg1, reg2, tmp ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ ld1 {®2&.s}[0], [TMP1], STRIDE ++ ld1 {®2&.s}[1], [TMP1] ++ convert_four_0565_to_x888_packed reg2, reg1, reg2, tmp ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_8888 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, tmp1, tmp2 ++ ++ bilinear_load_8888 reg1, reg2, tmp1 ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ bilinear_load_8888 reg3, reg4, tmp2 ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++.endm ++ ++.macro vzip reg1, reg2 ++ zip1 v24.8b, reg1, reg2 ++ zip2 reg2, reg1, reg2 ++ mov reg1, v24.8b ++.endm ++ ++.macro vuzp reg1, reg2 ++ uzp1 v24.8b, reg1, reg2 ++ uzp2 reg2, reg1, reg2 ++ mov reg1, v24.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_0565 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, acc2lo, acc2hi ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&acc2&.s}[0], [TMP1], STRIDE ++ ld1 {&acc2&.s}[2], [TMP2], STRIDE ++ ld1 {&acc2&.s}[1], [TMP1] ++ ld1 {&acc2&.s}[3], [TMP2] ++ convert_0565_to_x888 acc2, reg3, reg2, reg1 ++ vzip ®1&.8b, ®3&.8b ++ vzip ®2&.8b, ®4&.8b ++ vzip ®3&.8b, ®4&.8b ++ vzip ®1&.8b, ®2&.8b ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_0565 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&xacc2&.s}[0], [TMP1], STRIDE ++ ld1 {&xacc2&.s}[2], [TMP2], STRIDE ++ ld1 {&xacc2&.s}[1], [TMP1] ++ ld1 {&xacc2&.s}[3], [TMP2] ++ convert_0565_to_x888 xacc2, xreg3, xreg2, xreg1 ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&yacc2&.s}[0], [TMP1], STRIDE ++ vzip &xreg1&.8b, &xreg3&.8b ++ ld1 {&yacc2&.s}[2], [TMP2], STRIDE ++ vzip &xreg2&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[1], [TMP1] ++ vzip &xreg3&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[3], [TMP2] ++ vzip &xreg1&.8b, &xreg2&.8b ++ convert_0565_to_x888 yacc2, yreg3, yreg2, yreg1 ++ umull &xacc1&.8h, &xreg1&.8b, v28.8b ++ vzip &yreg1&.8b, &yreg3&.8b ++ umlal &xacc1&.8h, &xreg2&.8b, v29.8b ++ vzip &yreg2&.8b, &yreg4&.8b ++ umull &xacc2&.8h, &xreg3&.8b, v28.8b ++ vzip &yreg3&.8b, &yreg4&.8b ++ umlal &xacc2&.8h, &xreg4&.8b, v29.8b ++ vzip &yreg1&.8b, &yreg2&.8b ++ umull &yacc1&.8h, &yreg1&.8b, v28.8b ++ umlal &yacc1&.8h, &yreg2&.8b, v29.8b ++ umull &yacc2&.8h, &yreg3&.8b, v28.8b ++ umlal &yacc2&.8h, &yreg4&.8b, v29.8b ++.endm ++ ++.macro bilinear_store_8888 numpix, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v0.2s, v1.2s}, [OUT], #16 ++.elseif numpix == 2 ++ st1 {v0.2s}, [OUT], #8 ++.elseif numpix == 1 ++ st1 {v0.s}[0], [OUT], #4 ++.else ++ .error bilinear_store_8888 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_store_0565 numpix, tmp1, tmp2 ++ vuzp v0.8b, v1.8b ++ vuzp v2.8b, v3.8b ++ vuzp v1.8b, v3.8b ++ vuzp v0.8b, v2.8b ++ convert_8888_to_0565 v2, v1, v0, v1, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v1.4h}, [OUT], #8 ++.elseif numpix == 2 ++ st1 {v1.s}[0], [OUT], #4 ++.elseif numpix == 1 ++ st1 {v1.h}[0], [OUT], #2 ++.else ++ .error bilinear_store_0565 numpix is unsupported ++.endif ++.endm ++ ++ ++/* ++ * Macros for loading mask pixels into register 'mask'. ++ * dup must be done in somewhere else. ++ */ ++.macro bilinear_load_mask_x numpix, mask ++.endm ++ ++.macro bilinear_load_mask_8 numpix, mask ++.if numpix == 4 ++ ld1 {&mask&.s}[0], [MASK], #4 ++.elseif numpix == 2 ++ ld1 {&mask&.h}[0], [MASK], #2 ++.elseif numpix == 1 ++ ld1 {&mask&.b}[0], [MASK], #1 ++.else ++ .error bilinear_load_mask_8 numpix is unsupported ++.endif ++ prfm PREFETCH_MODE, [MASK, #prefetch_offset] ++.endm ++ ++.macro bilinear_load_mask mask_fmt, numpix, mask ++ bilinear_load_mask_&mask_fmt numpix, mask ++.endm ++ ++ ++/* ++ * Macros for loading destination pixels into register 'dst0' and 'dst1'. ++ * Interleave should be done somewhere else. ++ */ ++.macro bilinear_load_dst_0565_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.if numpix == 4 ++ ld1 {&dst0&.2s, &dst1&.2s}, [OUT] ++.elseif numpix == 2 ++ ld1 {&dst0&.2s}, [OUT] ++.elseif numpix == 1 ++ ld1 {&dst0&.s}[0], [OUT] ++.else ++ .error bilinear_load_dst_8888 numpix is unsupported ++.endif ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ prfm PREFETCH_MODE, [OUT, #(prefetch_offset * 4)] ++.endm ++ ++.macro bilinear_load_dst_8888_over numpix, dst0, dst1, dst01 ++ bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888_add numpix, dst0, dst1, dst01 ++ bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst dst_fmt, op, numpix, dst0, dst1, dst01 ++ bilinear_load_dst_&dst_fmt&_&op numpix, dst0, dst1, dst01 ++.endm ++ ++/* ++ * Macros for duplicating partially loaded mask to fill entire register. ++ * We will apply mask to interleaved source pixels, that is ++ * (r0, r1, r2, r3, g0, g1, g2, g3) x (m0, m1, m2, m3, m0, m1, m2, m3) ++ * (b0, b1, b2, b3, a0, a1, a2, a3) x (m0, m1, m2, m3, m0, m1, m2, m3) ++ * So, we need to duplicate loaded mask into whole register. ++ * ++ * For two pixel case ++ * (r0, r1, x, x, g0, g1, x, x) x (m0, m1, m0, m1, m0, m1, m0, m1) ++ * (b0, b1, x, x, a0, a1, x, x) x (m0, m1, m0, m1, m0, m1, m0, m1) ++ * We can do some optimizations for this including last pixel cases. ++ */ ++.macro bilinear_duplicate_mask_x numpix, mask ++.endm ++ ++.macro bilinear_duplicate_mask_8 numpix, mask ++.if numpix == 4 ++ dup &mask&.2s, &mask&.s[0] ++.elseif numpix == 2 ++ dup &mask&.4h, &mask&.h[0] ++.elseif numpix == 1 ++ dup &mask&.8b, &mask&.b[0] ++.else ++ .error bilinear_duplicate_mask_8 is unsupported ++.endif ++.endm ++ ++.macro bilinear_duplicate_mask mask_fmt, numpix, mask ++ bilinear_duplicate_mask_&mask_fmt numpix, mask ++.endm ++ ++/* ++ * Macros for interleaving src and dst pixels to rrrr gggg bbbb aaaa form. ++ * Interleave should be done when maks is enabled or operator is 'over'. ++ */ ++.macro bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++ vuzp &src0&.8b, &src1&.8b ++ vuzp &dst0&.8b, &dst1&.8b ++ vuzp &src0&.8b, &src1&.8b ++ vuzp &dst0&.8b, &dst1&.8b ++ mov &src01&.d[1], &src1&.d[0] ++ mov &src01&.d[0], &src0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ mov &dst01&.d[0], &dst0&.d[0] ++.endm ++ ++.macro bilinear_interleave_src_dst_x_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_x_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_x_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst \ ++ mask_fmt, op, numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave_src_dst_&mask_fmt&_&op \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++ ++/* ++ * Macros for applying masks to src pixels. (see combine_mask_u() function) ++ * src, dst should be in interleaved form. ++ * mask register should be in form (m0, m1, m2, m3). ++ */ ++.macro bilinear_apply_mask_to_src_x \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++.endm ++ ++.macro bilinear_apply_mask_to_src_8 \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++ ++ umull &tmp01&.8h, &src0&.8b, &mask&.8b ++ umull &tmp23&.8h, &src1&.8b, &mask&.8b ++ /* bubbles */ ++ urshr &tmp45&.8h, &tmp01&.8h, #8 ++ urshr &tmp67&.8h, &tmp23&.8h, #8 ++ /* bubbles */ ++ raddhn &src0&.8b, &tmp45&.8h, &tmp01&.8h ++ raddhn &src1&.8b, &tmp67&.8h, &tmp23&.8h ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_apply_mask_to_src \ ++ mask_fmt, numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++ ++ bilinear_apply_mask_to_src_&mask_fmt \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++.endm ++ ++ ++/* ++ * Macros for combining src and destination pixels. ++ * Interleave or not is depending on operator 'op'. ++ */ ++.macro bilinear_combine_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++.endm ++ ++.macro bilinear_combine_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ dup &tmp8&.2s, &src1&.s[1] ++ /* bubbles */ ++ mvn &tmp8&.8b, &tmp8&.8b ++ /* bubbles */ ++ umull &tmp01&.8h, &dst0&.8b, &tmp8&.8b ++ /* bubbles */ ++ umull &tmp23&.8h, &dst1&.8b, &tmp8&.8b ++ /* bubbles */ ++ urshr &tmp45&.8h, &tmp01&.8h, #8 ++ urshr &tmp67&.8h, &tmp23&.8h, #8 ++ /* bubbles */ ++ raddhn &dst0&.8b, &tmp45&.8h, &tmp01&.8h ++ raddhn &dst1&.8b, &tmp67&.8h, &tmp23&.8h ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ /* bubbles */ ++ uqadd &src0&.8b, &dst0&.8b, &src0&.8b ++ uqadd &src1&.8b, &dst1&.8b, &src1&.8b ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_combine_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ uqadd &src0&.8b, &dst0&.8b, &src0&.8b ++ uqadd &src1&.8b, &dst1&.8b, &src1&.8b ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_combine \ ++ op, numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ bilinear_combine_&op \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++.endm ++ ++/* ++ * Macros for final deinterleaving of destination pixels if needed. ++ */ ++.macro bilinear_deinterleave numpix, dst0, dst1, dst01 ++ vuzp &dst0&.8b, &dst1&.8b ++ /* bubbles */ ++ vuzp &dst0&.8b, &dst1&.8b ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++.endm ++ ++.macro bilinear_deinterleave_dst_x_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_x_over numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_x_add numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_src numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_over numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_add numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst mask_fmt, op, numpix, dst0, dst1, dst01 ++ bilinear_deinterleave_dst_&mask_fmt&_&op numpix, dst0, dst1, dst01 ++.endm ++ ++ ++.macro bilinear_interpolate_last_pixel src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_&src_fmt v0, v1, v2 ++ bilinear_load_mask mask_fmt, 1, v4 ++ bilinear_load_dst dst_fmt, op, 1, v18, v19, v9 ++ umull v2.8h, v0.8b, v28.8b ++ umlal v2.8h, v1.8b, v29.8b ++ /* 5 cycles bubble */ ++ ushll v0.4s, v2.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v2.4h, v15.h[0] ++ umlal2 v0.4s, v2.8h, v15.h[0] ++ /* 5 cycles bubble */ ++ bilinear_duplicate_mask mask_fmt, 1, v4 ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ /* 3 cycles bubble */ ++ xtn v0.8b, v0.8h ++ /* 1 cycle bubble */ ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 1, v0, v1, v0, v18, v19, v9 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 1, v0, v1, v0, v4, \ ++ v3, v8, v10, v11 ++ bilinear_combine \ ++ op, 1, v0, v1, v0, v18, v19, v9, \ ++ v3, v8, v10, v11, v5 ++ bilinear_deinterleave_dst mask_fmt, op, 1, v0, v1, v0 ++ bilinear_store_&dst_fmt 1, v17, v18 ++.endm ++ ++.macro bilinear_interpolate_two_pixels src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_and_vertical_interpolate_two_&src_fmt \ ++ v1, v11, v18, v19, v20, v21, v22, v23 ++ bilinear_load_mask mask_fmt, 2, v4 ++ bilinear_load_dst dst_fmt, op, 2, v18, v19, v9 ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ bilinear_duplicate_mask mask_fmt, 2, v4 ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ xtn v0.8b, v0.8h ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 2, v0, v1, v0, v18, v19, v9 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 2, v0, v1, v0, v4, \ ++ v3, v8, v10, v11 ++ bilinear_combine \ ++ op, 2, v0, v1, v0, v18, v19, v9, \ ++ v3, v8, v10, v11, v5 ++ bilinear_deinterleave_dst mask_fmt, op, 2, v0, v1, v0 ++ bilinear_store_&dst_fmt 2, v16, v17 ++.endm ++ ++.macro bilinear_interpolate_four_pixels src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_and_vertical_interpolate_four_&src_fmt \ ++ v1, v11, v4, v5, v6, v7, v22, v23 \ ++ v3, v9, v16, v17, v20, v21, v18, v19 ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ sub TMP1, TMP1, STRIDE ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v3.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v3.4h, v15.h[0] ++ umlal2 v2.4s, v3.8h, v15.h[0] ++ ushll v8.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v8.4s, v9.4h, v15.h[4] ++ umlal2 v8.4s, v9.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v8.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ bilinear_load_mask mask_fmt, 4, v4 ++ bilinear_duplicate_mask mask_fmt, 4, v4 ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ xtn v0.8b, v0.8h ++ xtn v1.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ bilinear_load_dst dst_fmt, op, 4, v2, v3, v21 ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 4, v0, v1, v0, v2, v3, v11 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 4, v0, v1, v0, v4, \ ++ v6, v8, v9, v10 ++ bilinear_combine \ ++ op, 4, v0, v1, v0, v2, v3, v1, \ ++ v6, v8, v9, v10, v23 ++ bilinear_deinterleave_dst mask_fmt, op, 4, v0, v1, v0 ++ bilinear_store_&dst_fmt 4, v6, v7 ++.endm ++ ++.set BILINEAR_FLAG_USE_MASK, 1 ++.set BILINEAR_FLAG_USE_ALL_NEON_REGS, 2 ++ ++/* ++ * Main template macro for generating NEON optimized bilinear scanline functions. ++ * ++ * Bilinear scanline generator macro take folling arguments: ++ * fname - name of the function to generate ++ * src_fmt - source color format (8888 or 0565) ++ * dst_fmt - destination color format (8888 or 0565) ++ * src/dst_bpp_shift - (1 << bpp_shift) is the size of src/dst pixel in bytes ++ * process_last_pixel - code block that interpolate one pixel and does not ++ * update horizontal weight ++ * process_two_pixels - code block that interpolate two pixels and update ++ * horizontal weight ++ * process_four_pixels - code block that interpolate four pixels and update ++ * horizontal weight ++ * process_pixblock_head - head part of middle loop ++ * process_pixblock_tail - tail part of middle loop ++ * process_pixblock_tail_head - tail_head of middle loop ++ * pixblock_size - number of pixels processed in a single middle loop ++ * prefetch_distance - prefetch in the source image by that many pixels ahead ++ */ ++ ++.macro generate_bilinear_scanline_func \ ++ fname, \ ++ src_fmt, dst_fmt, src_bpp_shift, dst_bpp_shift, \ ++ bilinear_process_last_pixel, \ ++ bilinear_process_two_pixels, \ ++ bilinear_process_four_pixels, \ ++ bilinear_process_pixblock_head, \ ++ bilinear_process_pixblock_tail, \ ++ bilinear_process_pixblock_tail_head, \ ++ pixblock_size, \ ++ prefetch_distance, \ ++ flags ++ ++pixman_asm_function fname ++.if pixblock_size == 8 ++.elseif pixblock_size == 4 ++.else ++ .error unsupported pixblock size ++.endif ++ ++.if ((flags) & BILINEAR_FLAG_USE_MASK) == 0 ++ OUT .req x0 ++ TOP .req x1 ++ BOTTOM .req x2 ++ WT .req x3 ++ WWT .req w3 ++ WB .req x4 ++ WWB .req w4 ++ X .req w5 ++ UX .req w6 ++ WIDTH .req x7 ++ TMP1 .req x10 ++ WTMP1 .req w10 ++ TMP2 .req x11 ++ WTMP2 .req w11 ++ PF_OFFS .req x12 ++ TMP3 .req x13 ++ WTMP3 .req w13 ++ TMP4 .req x14 ++ WTMP4 .req w14 ++ STRIDE .req x15 ++ DUMMY .req x30 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 112 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x10, x11, [x29, -80] ++ stp x12, x13, [x29, -96] ++ stp x14, x15, [x29, -112] ++.else ++ OUT .req x0 ++ MASK .req x1 ++ TOP .req x2 ++ BOTTOM .req x3 ++ WT .req x4 ++ WWT .req w4 ++ WB .req x5 ++ WWB .req w5 ++ X .req w6 ++ UX .req w7 ++ WIDTH .req x8 ++ TMP1 .req x10 ++ WTMP1 .req w10 ++ TMP2 .req x11 ++ WTMP2 .req w11 ++ PF_OFFS .req x12 ++ TMP3 .req x13 ++ WTMP3 .req w13 ++ TMP4 .req x14 ++ WTMP4 .req w14 ++ STRIDE .req x15 ++ DUMMY .req x30 ++ ++ .set prefetch_offset, prefetch_distance ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x10, x11, [x29, -80] ++ stp x12, x13, [x29, -96] ++ stp x14, x15, [x29, -112] ++ str x8, [x29, -120] ++ ldr w8, [x29, 16] ++ sub sp, sp, 120 ++.endif ++ ++ mov WTMP1, #prefetch_distance ++ umull PF_OFFS, WTMP1, UX ++ ++ sub STRIDE, BOTTOM, TOP ++ .unreq BOTTOM ++ ++ cmp WIDTH, #0 ++ ble 300f ++ ++ dup v12.8h, X ++ dup v13.8h, UX ++ dup v28.8b, WWT ++ dup v29.8b, WWB ++ mov v25.d[0], v12.d[1] ++ mov v26.d[0], v13.d[0] ++ add v25.4h, v25.4h, v26.4h ++ mov v12.d[1], v25.d[0] ++ ++ /* ensure good destination alignment */ ++ cmp WIDTH, #1 ++ blt 100f ++ tst OUT, #(1 << dst_bpp_shift) ++ beq 100f ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ bilinear_process_last_pixel ++ sub WIDTH, WIDTH, #1 ++100: ++ add v13.8h, v13.8h, v13.8h ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ++ cmp WIDTH, #2 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 1)) ++ beq 100f ++ bilinear_process_two_pixels ++ sub WIDTH, WIDTH, #2 ++100: ++.if pixblock_size == 8 ++ cmp WIDTH, #4 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 2)) ++ beq 100f ++ bilinear_process_four_pixels ++ sub WIDTH, WIDTH, #4 ++100: ++.endif ++ subs WIDTH, WIDTH, #pixblock_size ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_process_pixblock_head ++ subs WIDTH, WIDTH, #pixblock_size ++ blt 500f ++0: ++ bilinear_process_pixblock_tail_head ++ subs WIDTH, WIDTH, #pixblock_size ++ bge 0b ++500: ++ bilinear_process_pixblock_tail ++100: ++.if pixblock_size == 8 ++ tst WIDTH, #4 ++ beq 200f ++ bilinear_process_four_pixels ++200: ++.endif ++ /* handle the remaining trailing pixels */ ++ tst WIDTH, #2 ++ beq 200f ++ bilinear_process_two_pixels ++200: ++ tst WIDTH, #1 ++ beq 300f ++ bilinear_process_last_pixel ++300: ++ ++.if ((flags) & BILINEAR_FLAG_USE_MASK) == 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x10, x11, [x29, -80] ++ ldp x12, x13, [x29, -96] ++ ldp x14, x15, [x29, -112] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x10, x11, [x29, -80] ++ ldp x12, x13, [x29, -96] ++ ldp x14, x15, [x29, -112] ++ ldr x8, [x29, -120] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++.endif ++ ret ++ ++ .unreq OUT ++ .unreq TOP ++ .unreq WT ++ .unreq WWT ++ .unreq WB ++ .unreq WWB ++ .unreq X ++ .unreq UX ++ .unreq WIDTH ++ .unreq TMP1 ++ .unreq WTMP1 ++ .unreq TMP2 ++ .unreq PF_OFFS ++ .unreq TMP3 ++ .unreq TMP4 ++ .unreq STRIDE ++.if ((flags) & BILINEAR_FLAG_USE_MASK) != 0 ++ .unreq MASK ++.endif ++ ++.endfunc ++ ++.endm ++ ++/* src_8888_8_8888 */ ++.macro bilinear_src_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_head ++ bilinear_src_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_tail_head ++ bilinear_src_8888_8_8888_process_pixblock_tail ++ bilinear_src_8888_8_8888_process_pixblock_head ++.endm ++ ++/* src_8888_8_0565 */ ++.macro bilinear_src_8888_8_0565_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_head ++ bilinear_src_8888_8_0565_process_four_pixels ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_tail_head ++ bilinear_src_8888_8_0565_process_pixblock_tail ++ bilinear_src_8888_8_0565_process_pixblock_head ++.endm ++ ++/* src_0565_8_x888 */ ++.macro bilinear_src_0565_8_x888_process_last_pixel ++ bilinear_interpolate_last_pixel 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_two_pixels ++ bilinear_interpolate_two_pixels 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_four_pixels ++ bilinear_interpolate_four_pixels 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_head ++ bilinear_src_0565_8_x888_process_four_pixels ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_tail_head ++ bilinear_src_0565_8_x888_process_pixblock_tail ++ bilinear_src_0565_8_x888_process_pixblock_head ++.endm ++ ++/* src_0565_8_0565 */ ++.macro bilinear_src_0565_8_0565_process_last_pixel ++ bilinear_interpolate_last_pixel 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_two_pixels ++ bilinear_interpolate_two_pixels 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_four_pixels ++ bilinear_interpolate_four_pixels 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_head ++ bilinear_src_0565_8_0565_process_four_pixels ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_tail_head ++ bilinear_src_0565_8_0565_process_pixblock_tail ++ bilinear_src_0565_8_0565_process_pixblock_head ++.endm ++ ++/* over_8888_8888 */ ++.macro bilinear_over_8888_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_head ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ ++ ld1 {v22.2s}, [TMP1], STRIDE ++ ld1 {v23.2s}, [TMP1] ++ asr WTMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ umull v8.8h, v22.8b, v28.8b ++ umlal v8.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP2], STRIDE ++ ld1 {v23.2s}, [TMP2] ++ asr WTMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umull v9.8h, v22.8b, v28.8b ++ umlal v9.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_tail ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ xtn v7.8b, v2.8h ++ ld1 {v2.2s, v3.2s}, [OUT] ++ prfm PREFETCH_MODE, [OUT, #(prefetch_offset * 4)] ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ dup v4.2s, v7.s[1] ++ mvn v4.8b, v4.8b ++ umull v11.8h, v2.8b, v4.8b ++ umull v2.8h, v3.8b, v4.8b ++ urshr v1.8h, v11.8h, #8 ++ urshr v10.8h, v2.8h, #8 ++ raddhn v3.8b, v10.8h, v2.8h ++ raddhn v2.8b, v1.8h, v11.8h ++ uqadd v6.8b, v2.8b, v6.8b ++ uqadd v7.8b, v3.8b, v7.8b ++ vuzp v6.8b, v7.8b ++ vuzp v6.8b, v7.8b ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_tail_head ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ umlsl v2.4s, v10.4h, v15.h[0] ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ ld1 {v20.2s}, [TMP1], STRIDE ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ ld1 {v21.2s}, [TMP1] ++ umull v8.8h, v20.8b, v28.8b ++ umlal v8.8h, v21.8b, v29.8b ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ld1 {v22.2s}, [TMP2], STRIDE ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ ld1 {v23.2s}, [TMP2] ++ umull v9.8h, v22.8b, v28.8b ++ asr WTMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ asr WTMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umlal v9.8h, v23.8b, v29.8b ++ xtn v7.8b, v2.8h ++ ld1 {v2.2s, v3.2s}, [OUT] ++ prfm PREFETCH_MODE, [OUT, PF_OFFS] ++ ld1 {v22.2s}, [TMP3], STRIDE ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ dup v4.2s, v7.s[1] ++ ld1 {v23.2s}, [TMP3] ++ mvn v4.8b, v4.8b ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ umull v11.8h, v2.8b, v4.8b ++ umull v2.8h, v3.8b, v4.8b ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ urshr v1.8h, v11.8h, #8 ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ urshr v8.8h, v2.8h, #8 ++ raddhn v3.8b, v8.8h, v2.8h ++ raddhn v2.8b, v1.8h, v11.8h ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ uqadd v6.8b, v2.8b, v6.8b ++ uqadd v7.8b, v3.8b, v7.8b ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ vuzp v6.8b, v7.8b ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ vuzp v6.8b, v7.8b ++ umlsl v1.4s, v9.4h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++/* over_8888_8_8888 */ ++.macro bilinear_over_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_four_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_head ++ bilinear_over_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_tail_head ++ bilinear_over_8888_8_8888_process_pixblock_tail ++ bilinear_over_8888_8_8888_process_pixblock_head ++.endm ++ ++/* add_8888_8888 */ ++.macro bilinear_add_8888_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_four_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_head ++ bilinear_add_8888_8888_process_four_pixels ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_tail_head ++ bilinear_add_8888_8888_process_pixblock_tail ++ bilinear_add_8888_8888_process_pixblock_head ++.endm ++ ++/* add_8888_8_8888 */ ++.macro bilinear_add_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_head ++ bilinear_add_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_tail_head ++ bilinear_add_8888_8_8888_process_pixblock_tail ++ bilinear_add_8888_8_8888_process_pixblock_head ++.endm ++ ++ ++/* Bilinear scanline functions */ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_SRC_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_src_8888_8_8888_process_last_pixel, \ ++ bilinear_src_8888_8_8888_process_two_pixels, \ ++ bilinear_src_8888_8_8888_process_four_pixels, \ ++ bilinear_src_8888_8_8888_process_pixblock_head, \ ++ bilinear_src_8888_8_8888_process_pixblock_tail, \ ++ bilinear_src_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_0565_SRC_asm_neon, \ ++ 8888, 0565, 2, 1, \ ++ bilinear_src_8888_8_0565_process_last_pixel, \ ++ bilinear_src_8888_8_0565_process_two_pixels, \ ++ bilinear_src_8888_8_0565_process_four_pixels, \ ++ bilinear_src_8888_8_0565_process_pixblock_head, \ ++ bilinear_src_8888_8_0565_process_pixblock_tail, \ ++ bilinear_src_8888_8_0565_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_8_x888_SRC_asm_neon, \ ++ 0565, 8888, 1, 2, \ ++ bilinear_src_0565_8_x888_process_last_pixel, \ ++ bilinear_src_0565_8_x888_process_two_pixels, \ ++ bilinear_src_0565_8_x888_process_four_pixels, \ ++ bilinear_src_0565_8_x888_process_pixblock_head, \ ++ bilinear_src_0565_8_x888_process_pixblock_tail, \ ++ bilinear_src_0565_8_x888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_8_0565_SRC_asm_neon, \ ++ 0565, 0565, 1, 1, \ ++ bilinear_src_0565_8_0565_process_last_pixel, \ ++ bilinear_src_0565_8_0565_process_two_pixels, \ ++ bilinear_src_0565_8_0565_process_four_pixels, \ ++ bilinear_src_0565_8_0565_process_pixblock_head, \ ++ bilinear_src_0565_8_0565_process_pixblock_tail, \ ++ bilinear_src_0565_8_0565_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_OVER_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_over_8888_8888_process_last_pixel, \ ++ bilinear_over_8888_8888_process_two_pixels, \ ++ bilinear_over_8888_8888_process_four_pixels, \ ++ bilinear_over_8888_8888_process_pixblock_head, \ ++ bilinear_over_8888_8888_process_pixblock_tail, \ ++ bilinear_over_8888_8888_process_pixblock_tail_head, \ ++ 4, 28, 0 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_OVER_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_over_8888_8_8888_process_last_pixel, \ ++ bilinear_over_8888_8_8888_process_two_pixels, \ ++ bilinear_over_8888_8_8888_process_four_pixels, \ ++ bilinear_over_8888_8_8888_process_pixblock_head, \ ++ bilinear_over_8888_8_8888_process_pixblock_tail, \ ++ bilinear_over_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_ADD_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_add_8888_8888_process_last_pixel, \ ++ bilinear_add_8888_8888_process_two_pixels, \ ++ bilinear_add_8888_8888_process_four_pixels, \ ++ bilinear_add_8888_8888_process_pixblock_head, \ ++ bilinear_add_8888_8888_process_pixblock_tail, \ ++ bilinear_add_8888_8888_process_pixblock_tail_head, \ ++ 4, 28, 0 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_ADD_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_add_8888_8_8888_process_last_pixel, \ ++ bilinear_add_8888_8_8888_process_two_pixels, \ ++ bilinear_add_8888_8_8888_process_four_pixels, \ ++ bilinear_add_8888_8_8888_process_pixblock_head, \ ++ bilinear_add_8888_8_8888_process_pixblock_tail, \ ++ bilinear_add_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK +diff --git a/pixman/pixman-arma64-neon-asm.S b/pixman/pixman-arma64-neon-asm.S +new file mode 100644 +index 0000000..18ace0e +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm.S +@@ -0,0 +1,3704 @@ ++/* ++ * Copyright © 2009 Nokia Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ */ ++ ++/* ++ * This file contains implementations of NEON optimized pixel processing ++ * functions. There is no full and detailed tutorial, but some functions ++ * (those which are exposing some new or interesting features) are ++ * extensively commented and can be used as examples. ++ * ++ * You may want to have a look at the comments for following functions: ++ * - pixman_composite_over_8888_0565_asm_neon ++ * - pixman_composite_over_n_8_0565_asm_neon ++ */ ++ ++/* Prevent the stack from becoming executable for no reason... */ ++#if defined(__linux__) && defined(__ELF__) ++.section .note.GNU-stack,"",%progbits ++#endif ++ ++.text ++.arch armv8-a ++ ++.altmacro ++.p2align 2 ++ ++#include "pixman-private.h" ++#include "pixman-arm-asm.h" ++#include "pixman-arma64-neon-asm.h" ++ ++/* Global configuration options and preferences */ ++ ++/* ++ * The code can optionally make use of unaligned memory accesses to improve ++ * performance of handling leading/trailing pixels for each scanline. ++ * Configuration variable RESPECT_STRICT_ALIGNMENT can be set to 0 for ++ * example in linux if unaligned memory accesses are not configured to ++ * generate.exceptions. ++ */ ++.set RESPECT_STRICT_ALIGNMENT, 1 ++ ++/* ++ * Set default prefetch type. There is a choice between the following options: ++ * ++ * PREFETCH_TYPE_NONE (may be useful for the ARM cores where PLD is set to work ++ * as NOP to workaround some HW bugs or for whatever other reason) ++ * ++ * PREFETCH_TYPE_SIMPLE (may be useful for simple single-issue ARM cores where ++ * advanced prefetch intruduces heavy overhead) ++ * ++ * PREFETCH_TYPE_ADVANCED (useful for superscalar cores such as ARM Cortex-A8 ++ * which can run ARM and NEON instructions simultaneously so that extra ARM ++ * instructions do not add (many) extra cycles, but improve prefetch efficiency) ++ * ++ * Note: some types of function can't support advanced prefetch and fallback ++ * to simple one (those which handle 24bpp pixels) ++ */ ++.set PREFETCH_TYPE_DEFAULT, PREFETCH_TYPE_ADVANCED ++ ++/* Prefetch distance in pixels for simple prefetch */ ++.set PREFETCH_DISTANCE_SIMPLE, 64 ++ ++/* ++ * Implementation of pixman_composite_over_8888_0565_asm_neon ++ * ++ * This function takes a8r8g8b8 source buffer, r5g6b5 destination buffer and ++ * performs OVER compositing operation. Function fast_composite_over_8888_0565 ++ * from pixman-fast-path.c does the same in C and can be used as a reference. ++ * ++ * First we need to have some NEON assembly code which can do the actual ++ * operation on the pixels and provide it to the template macro. ++ * ++ * Template macro quite conveniently takes care of emitting all the necessary ++ * code for memory reading and writing (including quite tricky cases of ++ * handling unaligned leading/trailing pixels), so we only need to deal with ++ * the data in NEON registers. ++ * ++ * NEON registers allocation in general is recommented to be the following: ++ * v0, v1, v2, v3 - contain loaded source pixel data ++ * v4, v5, v6, v7 - contain loaded destination pixels (if they are needed) ++ * v24, v25, v26, v27 - contain loading mask pixel data (if mask is used) ++ * v28, v29, v30, v31 - place for storing the result (destination pixels) ++ * ++ * As can be seen above, four 64-bit NEON registers are used for keeping ++ * intermediate pixel data and up to 8 pixels can be processed in one step ++ * for 32bpp formats (16 pixels for 16bpp, 32 pixels for 8bpp). ++ * ++ * This particular function uses the following registers allocation: ++ * v0, v1, v2, v3 - contain loaded source pixel data ++ * v4, v5 - contain loaded destination pixels (they are needed) ++ * v28, v29 - place for storing the result (destination pixels) ++ */ ++ ++/* ++ * Step one. We need to have some code to do some arithmetics on pixel data. ++ * This is implemented as a pair of macros: '*_head' and '*_tail'. When used ++ * back-to-back, they take pixel data from {v0, v1, v2, v3} and {v4, v5}, ++ * perform all the needed calculations and write the result to {v28, v29}. ++ * The rationale for having two macros and not just one will be explained ++ * later. In practice, any single monolitic function which does the work can ++ * be split into two parts in any arbitrary way without affecting correctness. ++ * ++ * There is one special trick here too. Common template macro can optionally ++ * make our life a bit easier by doing R, G, B, A color components ++ * deinterleaving for 32bpp pixel formats (and this feature is used in ++ * 'pixman_composite_over_8888_0565_asm_neon' function). So it means that ++ * instead of having 8 packed pixels in {v0, v1, v2, v3} registers, we ++ * actually use v0 register for blue channel (a vector of eight 8-bit ++ * values), v1 register for green, v2 for red and v3 for alpha. This ++ * simple conversion can be also done with a few NEON instructions: ++ * ++ * Packed to planar conversion: // vuzp8 is a wrapper macro ++ * vuzp8 v0, v1 ++ * vuzp8 v2, v3 ++ * vuzp8 v1, v3 ++ * vuzp8 v0, v2 ++ * ++ * Planar to packed conversion: // vzip8 is a wrapper macro ++ * vzip8 v0, v2 ++ * vzip8 v1, v3 ++ * vzip8 v2, v3 ++ * vzip8 v0, v1 ++ * ++ * But pixel can be loaded directly in planar format using LD4 / b NEON ++ * instruction. It is 1 cycle slower than LD1 / s, so this is not always ++ * desirable, that's why deinterleaving is optional. ++ * ++ * But anyway, here is the code: ++ */ ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_head ++ /* convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ and put data into v6 - red, v7 - green, v30 - blue */ ++ mov v4.d[1], v5.d[0] ++ shrn v6.8b, v4.8h, #8 ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ sri v6.8b, v6.8b, #5 ++ mvn v3.8b, v3.8b /* invert source alpha */ ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ /* now do alpha blending, storing results in 8-bit planar format ++ into v20 - red, v23 - green, v22 - blue */ ++ umull v10.8h, v3.8b, v6.8b ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ urshr v17.8h, v10.8h, #8 ++ urshr v18.8h, v11.8h, #8 ++ urshr v19.8h, v12.8h, #8 ++ raddhn v20.8b, v10.8h, v17.8h ++ raddhn v23.8b, v11.8h, v18.8h ++ raddhn v22.8b, v12.8h, v19.8h ++.endm ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail ++ /* ... continue alpha blending */ ++ uqadd v17.8b, v2.8b, v20.8b ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ /* convert the result to r5g6b5 and store it into {v14} */ ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* ++ * OK, now we got almost everything that we need. Using the above two ++ * macros, the work can be done right. But now we want to optimize ++ * it a bit. ARM Cortex-A8 is an in-order core, and benefits really ++ * a lot from good code scheduling and software pipelining. ++ * ++ * Let's construct some code, which will run in the core main loop. ++ * Some pseudo-code of the main loop will look like this: ++ * head ++ * while (...) { ++ * tail ++ * head ++ * } ++ * tail ++ * ++ * It may look a bit weird, but this setup allows to hide instruction ++ * latencies better and also utilize dual-issue capability more ++ * efficiently (make pairs of load-store and ALU instructions). ++ * ++ * So what we need now is a '*_tail_head' macro, which will be used ++ * in the core main loop. A trivial straightforward implementation ++ * of this macro would look like this: ++ * ++ * pixman_composite_over_8888_0565_process_pixblock_tail ++ * st1 {v28.4h, v29.4h}, [DST_W], #32 ++ * ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ * ld4 {v0.2s, v1.2s, v2.2s, v3.2s}, [SRC], #32 ++ * pixman_composite_over_8888_0565_process_pixblock_head ++ * cache_preload 8, 8 ++ * ++ * Now it also got some VLD/VST instructions. We simply can't move from ++ * processing one block of pixels to the other one with just arithmetics. ++ * The previously processed data needs to be written to memory and new ++ * data needs to be fetched. Fortunately, this main loop does not deal ++ * with partial leading/trailing pixels and can load/store a full block ++ * of pixels in a bulk. Additionally, destination buffer is already ++ * 16 bytes aligned here (which is good for performance). ++ * ++ * New things here are DST_R, DST_W, SRC and MASK identifiers. These ++ * are the aliases for ARM registers which are used as pointers for ++ * accessing data. We maintain separate pointers for reading and writing ++ * destination buffer (DST_R and DST_W). ++ * ++ * Another new thing is 'cache_preload' macro. It is used for prefetching ++ * data into CPU L2 cache and improve performance when dealing with large ++ * images which are far larger than cache size. It uses one argument ++ * (actually two, but they need to be the same here) - number of pixels ++ * in a block. Looking into 'pixman-arm-neon-asm.h' can provide some ++ * details about this macro. Moreover, if good performance is needed ++ * the code from this macro needs to be copied into '*_tail_head' macro ++ * and mixed with the rest of code for optimal instructions scheduling. ++ * We are actually doing it below. ++ * ++ * Now after all the explanations, here is the optimized code. ++ * Different instruction streams (originaling from '*_head', '*_tail' ++ * and 'cache_preload' macro) use different indentation levels for ++ * better readability. Actually taking the code from one of these ++ * indentation levels and ignoring a few LD/ST instructions would ++ * result in exactly the code from '*_head', '*_tail' or 'cache_preload' ++ * macro! ++ */ ++ ++#if 1 ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail_head ++ uqadd v17.8b, v2.8b, v20.8b ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ mov v4.d[1], v5.d[0] ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ shrn v6.8b, v4.8h, #8 ++ fetch_src_pixblock ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ PF add PF_X, PF_X, #8 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ PF tst PF_CTL, #0xF ++ sri v6.8b, v6.8b, #5 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ mvn v3.8b, v3.8b ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ umull v10.8h, v3.8b, v6.8b ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ sri v14.8h, v8.8h, #5 ++ PF cmp PF_X, ORIG_W ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ urshr v17.8h, v10.8h, #8 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ urshr v19.8h, v11.8h, #8 ++ urshr v18.8h, v12.8h, #8 ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ raddhn v20.8b, v10.8h, v17.8h ++ raddhn v23.8b, v11.8h, v19.8h ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_SRC, #1 ++10: ++ raddhn v22.8b, v12.8h, v18.8h ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++#else ++ ++/* If we did not care much about the performance, we would just use this... */ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail_head ++ pixman_composite_over_8888_0565_process_pixblock_tail ++ st1 {v14.8h}, [DST_W], #16 ++ ld1 {v4.4h, v4.5h}, [DST_R], #16 ++ fetch_src_pixblock ++ pixman_composite_over_8888_0565_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++#endif ++ ++/* ++ * And now the final part. We are using 'generate_composite_function' macro ++ * to put all the stuff together. We are specifying the name of the function ++ * which we want to get, number of bits per pixel for the source, mask and ++ * destination (0 if unused, like mask in this case). Next come some bit ++ * flags: ++ * FLAG_DST_READWRITE - tells that the destination buffer is both read ++ * and written, for write-only buffer we would use ++ * FLAG_DST_WRITEONLY flag instead ++ * FLAG_DEINTERLEAVE_32BPP - tells that we prefer to work with planar data ++ * and separate color channels for 32bpp format. ++ * The next things are: ++ * - the number of pixels processed per iteration (8 in this case, because ++ * that's the maximum what can fit into four 64-bit NEON registers). ++ * - prefetch distance, measured in pixel blocks. In this case it is 5 times ++ * by 8 pixels. That would be 40 pixels, or up to 160 bytes. Optimal ++ * prefetch distance can be selected by running some benchmarks. ++ * ++ * After that we specify some macros, these are 'default_init', ++ * 'default_cleanup' here which are empty (but it is possible to have custom ++ * init/cleanup macros to be able to save/restore some extra NEON registers ++ * like d8-d15 or do anything else) followed by ++ * 'pixman_composite_over_8888_0565_process_pixblock_head', ++ * 'pixman_composite_over_8888_0565_process_pixblock_tail' and ++ * 'pixman_composite_over_8888_0565_process_pixblock_tail_head' ++ * which we got implemented above. ++ * ++ * The last part is the NEON registers allocation scheme. ++ */ ++generate_composite_function \ ++ pixman_composite_over_8888_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_0565_process_pixblock_head ++ /* convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ and put data into v6 - red, v7 - green, v30 - blue */ ++ mov v4.d[1], v5.d[0] ++ shrn v6.8b, v4.8h, #8 ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ sri v6.8b, v6.8b, #5 ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ /* now do alpha blending, storing results in 8-bit planar format ++ into v20 - red, v23 - green, v22 - blue */ ++ umull v10.8h, v3.8b, v6.8b ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ urshr v13.8h, v10.8h, #8 ++ urshr v14.8h, v11.8h, #8 ++ urshr v15.8h, v12.8h, #8 ++ raddhn v20.8b, v10.8h, v13.8h ++ raddhn v23.8b, v11.8h, v14.8h ++ raddhn v22.8b, v12.8h, v15.8h ++.endm ++ ++.macro pixman_composite_over_n_0565_process_pixblock_tail ++ /* ... continue alpha blending */ ++ uqadd v17.8b, v2.8b, v20.8b ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ /* convert the result to r5g6b5 and store it into {v14} */ ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_n_0565_process_pixblock_tail_head ++ pixman_composite_over_n_0565_process_pixblock_tail ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ st1 {v14.8h}, [DST_W], #16 ++ pixman_composite_over_n_0565_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_over_n_0565_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++ mvn v3.8b, v3.8b /* invert source alpha */ ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_0565_asm_neon, 0, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_0565_init, \ ++ default_cleanup, \ ++ pixman_composite_over_n_0565_process_pixblock_head, \ ++ pixman_composite_over_n_0565_process_pixblock_tail, \ ++ pixman_composite_over_n_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_head ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v14.8h, v2.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v9.8h, v0.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_tail ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_tail_head ++ sri v14.8h, v8.8h, #5 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ fetch_src_pixblock ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ st1 {v14.8h}, [DST_W], #16 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ ushll v14.8h, v2.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ ushll v9.8h, v0.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_8888_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_0565_process_pixblock_head, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0565_8888_process_pixblock_head ++ mov v0.d[1], v1.d[0] ++ shrn v30.8b, v0.8h, #8 ++ shrn v29.8b, v0.8h, #3 ++ sli v0.8h, v0.8h, #5 ++ movi v31.8b, #255 ++ sri v30.8b, v30.8b, #5 ++ sri v29.8b, v29.8b, #6 ++ shrn v28.8b, v0.8h, #2 ++.endm ++ ++.macro pixman_composite_src_0565_8888_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_src_0565_8888_process_pixblock_tail_head ++ pixman_composite_src_0565_8888_process_pixblock_tail ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ fetch_src_pixblock ++ pixman_composite_src_0565_8888_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0565_8888_asm_neon, 16, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_8888_process_pixblock_head, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8_8_process_pixblock_head ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_8_8_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_add_8_8_process_pixblock_tail_head ++ fetch_src_pixblock ++ PF add PF_X, PF_X, #32 ++ PF tst PF_CTL, #0xF ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ PF beq 10f ++ PF add PF_X, PF_X, #32 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ uqadd v28.8b, v0.8b, v4.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8_8_asm_neon, 8, 0, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ uqadd v28.8b, v0.8b, v4.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_add_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_head ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++ /* do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++.endm ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ fetch_src_pixblock ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ umull v10.8h, v22.8b, v6.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_out_reverse_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_head ++ pixman_composite_out_reverse_8888_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_tail ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ fetch_src_pixblock ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ umull v10.8h, v22.8b, v6.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_over_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_process_pixblock_head ++ /* deinterleaved source pixels in {v0, v1, v2, v3} */ ++ /* inverted alpha in {v24} */ ++ /* destination pixels in {v4, v5, v6, v7} */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_process_pixblock_tail_head ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ uqadd v28.8b, v0.8b, v28.8b ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0x0F ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ PF cmp PF_X, ORIG_W ++ umull v8.8h, v24.8b, v4.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ umull v9.8h, v24.8b, v5.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v10.8h, v24.8b, v6.8b ++ PF subs PF_CTL, PF_CTL, #0x10 ++ umull v11.8h, v24.8b, v7.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_n_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_reverse_n_8888_process_pixblock_tail_head ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v12.8h, v10.8h ++ raddhn v31.8b, v13.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ ld4 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_R], #32 ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF blt 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF blt 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ umull v10.8h, v22.8b, v6.8b ++ PF blt 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_reverse_n_8888_init ++ mov v7.s[0], w4 ++ dup v4.8b, v7.b[0] ++ dup v5.8b, v7.b[1] ++ dup v6.8b, v7.b[2] ++ dup v7.8b, v7.b[3] ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_reverse_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_reverse_n_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_reverse_n_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 4, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_head ++ umull v0.8h, v24.8b, v8.8b /* IN for SRC pixels (part1) */ ++ umull v1.8h, v24.8b, v9.8b ++ umull v2.8h, v24.8b, v10.8b ++ umull v3.8h, v24.8b, v11.8b ++ mov v4.d[1], v5.d[0] ++ shrn v25.8b, v4.8h, #8 /* convert DST_R data to 32-bpp (part1) */ ++ shrn v26.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ urshr v17.8h, v0.8h, #8 /* IN for SRC pixels (part2) */ ++ urshr v18.8h, v1.8h, #8 ++ urshr v19.8h, v2.8h, #8 ++ urshr v20.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v17.8h ++ raddhn v1.8b, v1.8h, v18.8h ++ raddhn v2.8b, v2.8h, v19.8h ++ raddhn v3.8b, v3.8h, v20.8h ++ sri v25.8b, v25.8b, #5 /* convert DST_R data to 32-bpp (part2) */ ++ sri v26.8b, v26.8b, #6 ++ mvn v3.8b, v3.8b ++ shrn v30.8b, v4.8h, #2 ++ umull v18.8h, v3.8b, v25.8b /* now do alpha blending */ ++ umull v19.8h, v3.8b, v26.8b ++ umull v20.8h, v3.8b, v30.8b ++.endm ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_tail ++ /* 3 cycle bubble (after vmull.u8) */ ++ urshr v5.8h, v18.8h, #8 ++ urshr v6.8h, v19.8h, #8 ++ urshr v7.8h, v20.8h, #8 ++ raddhn v17.8b, v18.8h, v5.8h ++ raddhn v19.8b, v19.8h, v6.8h ++ raddhn v18.8b, v20.8h, v7.8h ++ uqadd v5.8b, v2.8b, v17.8b ++ /* 1 cycle bubble */ ++ uqadd v6.8b, v0.8b, v18.8b ++ uqadd v7.8b, v1.8b, v19.8b ++ ushll v14.8h, v5.8b, #7 /* convert to 16bpp */ ++ sli v14.8h, v14.8h, #1 ++ ushll v18.8h, v7.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v6.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ sri v14.8h, v18.8h, #5 ++ /* 1 cycle bubble */ ++ sri v14.8h, v19.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_tail_head ++#if 0 ++ ld1 {v4.8h}, [DST_R], #16 ++ shrn v25.8b, v4.8h, #8 ++ fetch_mask_pixblock ++ shrn v26.8b, v4.8h, #3 ++ fetch_src_pixblock ++ umull v22.8h, v24.8b, v10.8b ++ urshr v13.8h, v18.8h, #8 ++ urshr v11.8h, v19.8h, #8 ++ urshr v15.8h, v20.8h, #8 ++ raddhn v17.8b, v18.8h, v13.8h ++ raddhn v19.8b, v19.8h, v11.8h ++ raddhn v18.8b, v20.8h, v15.8h ++ uqadd v17.8b, v2.8b, v17.8b ++ umull v21.8h, v24.8b, v9.8b ++ uqadd v18.8b, v0.8b, v18.8b ++ uqadd v19.8b, v1.8b, v19.8b ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ umull v20.8h, v24.8b, v8.8b ++ ushll v18.8h, v18.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v19.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ sri v14.8h, v18.8h, #5 ++ umull v23.8h, v24.8b, v11.8b ++ sri v14.8h, v19.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ ++ cache_preload 8, 8 ++ ++ sli v4.8h, v4.8h, #5 ++ urshr v16.8h, v20.8h, #8 ++ urshr v17.8h, v21.8h, #8 ++ urshr v18.8h, v22.8h, #8 ++ urshr v19.8h, v23.8h, #8 ++ raddhn v0.8b, v20.8h, v16.8h ++ raddhn v1.8b, v21.8h, v17.8h ++ raddhn v2.8b, v22.8h, v18.8h ++ raddhn v3.8b, v23.8h, v19.8h ++ sri v25.8b, v25.8b, #5 ++ sri v26.8b, v26.8b, #6 ++ mvn v3.8b, v3.8b ++ shrn v30.8b, v4.8h, #2 ++ st1 {v14.8h}, [DST_W], #16 ++ umull v18.8h, v3.8b, v25.8b ++ umull v19.8h, v3.8b, v26.8b ++ umull v20.8h, v3.8b, v30.8b ++#else ++ pixman_composite_over_8888_8_0565_process_pixblock_tail ++ st1 {v28.4h, v29.4h}, [DST_W], #16 ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ fetch_mask_pixblock ++ fetch_src_pixblock ++ pixman_composite_over_8888_8_0565_process_pixblock_head ++#endif ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8_0565_asm_neon, 32, 8, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* ++ * This function needs a special initialization of solid mask. ++ * Solid source pixel data is fetched from stack at ARGS_STACK_OFFSET ++ * offset, split into color components and replicated in d8-d11 ++ * registers. Additionally, this function needs all the NEON registers, ++ * so it has to save d8-d15 registers which are callee saved according ++ * to ABI. These registers are restored from 'cleanup' macro. All the ++ * other NEON registers are caller saved, so can be clobbered freely ++ * without introducing any problems. ++ */ ++.macro pixman_composite_over_n_8_0565_init ++ mov v11.s[0], w4 ++ dup v8.8b, v11.b[0] ++ dup v9.8b, v11.b[1] ++ dup v10.8b, v11.b[2] ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_0565_asm_neon, 0, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_0565_init, \ ++ pixman_composite_over_n_8_0565_cleanup, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_n_0565_init ++ mov v24.s[0], w6 ++ dup v24.8b, v24.b[3] ++.endm ++ ++.macro pixman_composite_over_8888_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_n_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_8888_n_0565_init, \ ++ pixman_composite_over_8888_n_0565_cleanup, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_tail_head ++ st1 {v0.4h, v1.4h, v2.4h, v3.4h}, [DST_W], #32 ++ fetch_src_pixblock ++ cache_preload 16, 16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0565_0565_asm_neon, 16, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 16, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_0565_process_pixblock_head, \ ++ pixman_composite_src_0565_0565_process_pixblock_tail, \ ++ pixman_composite_src_0565_0565_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_8_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_8_process_pixblock_tail_head ++ st1 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_W], 32 ++.endm ++ ++.macro pixman_composite_src_n_8_init ++ mov v0.s[0], w4 ++ dup v3.8b, v0.b[0] ++ dup v2.8b, v0.b[0] ++ dup v1.8b, v0.b[0] ++ dup v0.8b, v0.b[0] ++.endm ++ ++.macro pixman_composite_src_n_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_asm_neon, 0, 0, 8, \ ++ FLAG_DST_WRITEONLY, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_8_init, \ ++ pixman_composite_src_n_8_cleanup, \ ++ pixman_composite_src_n_8_process_pixblock_head, \ ++ pixman_composite_src_n_8_process_pixblock_tail, \ ++ pixman_composite_src_n_8_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_0565_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_0565_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_0565_process_pixblock_tail_head ++ st1 {v0.4h, v1.4h, v2.4h, v3.4h}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_src_n_0565_init ++ mov v0.s[0], w4 ++ dup v3.4h, v0.h[0] ++ dup v2.4h, v0.h[0] ++ dup v1.4h, v0.h[0] ++ dup v0.4h, v0.h[0] ++.endm ++ ++.macro pixman_composite_src_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_0565_asm_neon, 0, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 16, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_0565_init, \ ++ pixman_composite_src_n_0565_cleanup, \ ++ pixman_composite_src_n_0565_process_pixblock_head, \ ++ pixman_composite_src_n_0565_process_pixblock_tail, \ ++ pixman_composite_src_n_0565_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_src_n_8888_init ++ mov v0.s[0], w4 ++ dup v3.2s, v0.s[0] ++ dup v2.2s, v0.s[0] ++ dup v1.2s, v0.s[0] ++ dup v0.2s, v0.s[0] ++.endm ++ ++.macro pixman_composite_src_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_8888_init, \ ++ pixman_composite_src_n_8888_cleanup, \ ++ pixman_composite_src_n_8888_process_pixblock_head, \ ++ pixman_composite_src_n_8888_process_pixblock_tail, \ ++ pixman_composite_src_n_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++ fetch_src_pixblock ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_8888_process_pixblock_head, \ ++ pixman_composite_src_8888_8888_process_pixblock_tail, \ ++ pixman_composite_src_8888_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_head ++ orr v0.8b, v0.8b, v4.8b ++ orr v1.8b, v1.8b, v4.8b ++ orr v2.8b, v2.8b, v4.8b ++ orr v3.8b, v3.8b, v4.8b ++.endm ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++ fetch_src_pixblock ++ orr v0.8b, v0.8b, v4.8b ++ orr v1.8b, v1.8b, v4.8b ++ orr v2.8b, v2.8b, v4.8b ++ orr v3.8b, v3.8b, v4.8b ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_src_x888_8888_init ++ movi v4.2s, #0xff, lsl 24 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_x888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ pixman_composite_src_x888_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_src_x888_8888_process_pixblock_head, \ ++ pixman_composite_src_x888_8888_process_pixblock_tail, \ ++ pixman_composite_src_x888_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_head ++ /* expecting solid source in {v0, v1, v2, v3} */ ++ /* mask is in v24 (v25, v26, v27 are unused) */ ++ ++ /* in */ ++ umull v8.8h, v24.8b, v0.8b ++ umull v9.8h, v24.8b, v1.8b ++ umull v10.8h, v24.8b, v2.8b ++ umull v11.8h, v24.8b, v3.8b ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_tail ++ rshrn v28.8b, v8.8h, #8 ++ rshrn v29.8b, v9.8h, #8 ++ rshrn v30.8b, v10.8h, #8 ++ rshrn v31.8b, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_tail_head ++ fetch_mask_pixblock ++ PF add PF_X, PF_X, #8 ++ rshrn v28.8b, v8.8h, #8 ++ PF tst PF_CTL, #0x0F ++ rshrn v29.8b, v9.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ rshrn v30.8b, v10.8h, #8 ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ rshrn v31.8b, v11.8h, #8 ++ PF cmp PF_X, ORIG_W ++ umull v8.8h, v24.8b, v0.8b ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v9.8h, v24.8b, v1.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v10.8h, v24.8b, v2.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v11.8h, v24.8b, v3.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_src_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_src_n_8_8888_init, \ ++ pixman_composite_src_n_8_8888_cleanup, \ ++ pixman_composite_src_n_8_8888_process_pixblock_head, \ ++ pixman_composite_src_n_8_8888_process_pixblock_tail, \ ++ pixman_composite_src_n_8_8888_process_pixblock_tail_head, \ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_head ++ umull v0.8h, v24.8b, v16.8b ++ umull v1.8h, v25.8b, v16.8b ++ umull v2.8h, v26.8b, v16.8b ++ umull v3.8h, v27.8b, v16.8b ++ ursra v0.8h, v0.8h, #8 ++ ursra v1.8h, v1.8h, #8 ++ ursra v2.8h, v2.8h, #8 ++ ursra v3.8h, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_tail ++ rshrn v28.8b, v0.8h, #8 ++ rshrn v29.8b, v1.8h, #8 ++ rshrn v30.8b, v2.8h, #8 ++ rshrn v31.8b, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_tail_head ++ fetch_mask_pixblock ++ PF add PF_X, PF_X, #8 ++ rshrn v28.8b, v0.8h, #8 ++ PF tst PF_CTL, #0x0F ++ rshrn v29.8b, v1.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ rshrn v30.8b, v2.8h, #8 ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ rshrn v31.8b, v3.8h, #8 ++ PF cmp PF_X, ORIG_W ++ umull v0.8h, v24.8b, v16.8b ++ PF lsl DUMMY, PF_X, mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v1.8h, v25.8b, v16.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v2.8h, v26.8b, v16.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v3.8h, v27.8b, v16.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v0.8h, v0.8h, #8 ++ ursra v1.8h, v1.8h, #8 ++ ursra v2.8h, v2.8h, #8 ++ ursra v3.8h, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_init ++ mov v16.s[0], w4 ++ dup v16.8b, v16.b[3] ++.endm ++ ++.macro pixman_composite_src_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_WRITEONLY, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_src_n_8_8_init, \ ++ pixman_composite_src_n_8_8_cleanup, \ ++ pixman_composite_src_n_8_8_process_pixblock_head, \ ++ pixman_composite_src_n_8_8_process_pixblock_tail, \ ++ pixman_composite_src_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_head ++ /* expecting deinterleaved source data in {v8, v9, v10, v11} */ ++ /* v8 - blue, v9 - green, v10 - red, v11 - alpha */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ /* mask is in v24 (v25, v26, v27 are unused) */ ++ ++ /* in */ ++ umull v12.8h, v24.8b, v8.8b ++ umull v13.8h, v24.8b, v9.8b ++ umull v14.8h, v24.8b, v10.8b ++ umull v15.8h, v24.8b, v11.8b ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v0.8b, v12.8h, v16.8h ++ raddhn v1.8b, v13.8h, v17.8h ++ raddhn v2.8b, v14.8h, v18.8h ++ raddhn v3.8b, v15.8h, v19.8h ++ mvn v25.8b, v3.8b /* get inverted alpha */ ++ /* source: v0 - blue, v1 - green, v2 - red, v3 - alpha */ ++ /* destination: v4 - blue, v5 - green, v6 - red, v7 - alpha */ ++ /* now do alpha blending */ ++ umull v12.8h, v25.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v25.8b, v6.8b ++ umull v15.8h, v25.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_tail ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_tail_head ++ urshr v16.8h, v12.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v17.8h, v13.8h, #8 ++ fetch_mask_pixblock ++ urshr v18.8h, v14.8h, #8 ++ PF add PF_X, PF_X, #8 ++ urshr v19.8h, v15.8h, #8 ++ PF tst PF_CTL, #0x0F ++ raddhn v28.8b, v16.8h, v12.8h ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ raddhn v29.8b, v17.8h, v13.8h ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v30.8b, v18.8h, v14.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v31.8b, v19.8h, v15.8h ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ umull v16.8h, v24.8b, v8.8b ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v17.8h, v24.8b, v9.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v18.8h, v24.8b, v10.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v19.8h, v24.8b, v11.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v28.8b, v0.8b, v28.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ urshr v12.8h, v16.8h, #8 ++ urshr v13.8h, v17.8h, #8 ++ urshr v14.8h, v18.8h, #8 ++ urshr v15.8h, v19.8h, #8 ++ raddhn v0.8b, v16.8h, v12.8h ++ raddhn v1.8b, v17.8h, v13.8h ++ raddhn v2.8b, v18.8h, v14.8h ++ raddhn v3.8b, v19.8h, v15.8h ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ mvn v25.8b, v3.8b ++ umull v12.8h, v25.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v25.8b, v6.8b ++ umull v15.8h, v25.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_init ++ mov v11.s[0], w4 ++ dup v8.8b, v11.b[0] ++ dup v9.8b, v11.b[1] ++ dup v10.8b, v11.b[2] ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_8888_init, \ ++ pixman_composite_over_n_8_8888_cleanup, \ ++ pixman_composite_over_n_8_8888_process_pixblock_head, \ ++ pixman_composite_over_n_8_8888_process_pixblock_tail, \ ++ pixman_composite_over_n_8_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8_8_process_pixblock_head ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v8.8b ++ umull v2.8h, v26.8b, v8.8b ++ umull v3.8h, v27.8b, v8.8b ++ urshr v10.8h, v0.8h, #8 ++ urshr v11.8h, v1.8h, #8 ++ urshr v12.8h, v2.8h, #8 ++ urshr v13.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v10.8h ++ raddhn v1.8b, v1.8h, v11.8h ++ raddhn v2.8b, v2.8h, v12.8h ++ raddhn v3.8b, v3.8h, v13.8h ++ mvn v24.8b, v0.8b ++ mvn v25.8b, v1.8b ++ mvn v26.8b, v2.8b ++ mvn v27.8b, v3.8b ++ umull v10.8h, v24.8b, v4.8b ++ umull v11.8h, v25.8b, v5.8b ++ umull v12.8h, v26.8b, v6.8b ++ umull v13.8h, v27.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8_process_pixblock_tail ++ urshr v14.8h, v10.8h, #8 ++ urshr v15.8h, v11.8h, #8 ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ raddhn v28.8b, v14.8h, v10.8h ++ raddhn v29.8b, v15.8h, v11.8h ++ raddhn v30.8b, v16.8h, v12.8h ++ raddhn v31.8b, v17.8h, v13.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_n_8_8_process_pixblock_tail_head ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_n_8_8_process_pixblock_tail ++ fetch_mask_pixblock ++ cache_preload 32, 32 ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ pixman_composite_over_n_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_n_8_8_init ++ mov v8.s[0], w4 ++ dup v8.8b, v8.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_8_init, \ ++ pixman_composite_over_n_8_8_cleanup, \ ++ pixman_composite_over_n_8_8_process_pixblock_head, \ ++ pixman_composite_over_n_8_8_process_pixblock_tail, \ ++ pixman_composite_over_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_head ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} ++ * dest in {v4, v5, v6, v7 } ++ * mask in {v24, v25, v26, v27} ++ * output: updated src in {v0, v1, v2, v3 } ++ * updated mask in {v24, v25, v26, v3 } ++ */ ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ umull v3.8h, v27.8b, v11.8b ++ umull v12.8h, v11.8b, v25.8b ++ umull v13.8h, v11.8b, v24.8b ++ umull v14.8h, v11.8b, v26.8b ++ urshr v15.8h, v0.8h, #8 ++ urshr v16.8h, v1.8h, #8 ++ urshr v17.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v15.8h ++ raddhn v1.8b, v1.8h, v16.8h ++ raddhn v2.8b, v2.8h, v17.8h ++ urshr v15.8h, v13.8h, #8 ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v14.8h, #8 ++ urshr v18.8h, v3.8h, #8 ++ raddhn v24.8b, v13.8h, v15.8h ++ raddhn v25.8b, v12.8h, v16.8h ++ raddhn v26.8b, v14.8h, v17.8h ++ raddhn v3.8b, v3.8h, v18.8h ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in {v28, v29, v30, v31} ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ mvn v26.8b, v26.8b ++ mvn v27.8b, v3.8b ++ umull v12.8h, v24.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v26.8b, v6.8b ++ umull v15.8h, v27.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_tail ++ /* ... continue 'combine_over_ca' replacement */ ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_tail_head ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ fetch_mask_pixblock ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ cache_preload 8, 8 ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_init ++ mov v13.s[0], w4 ++ dup v8.8b, v13.b[0] ++ dup v9.8b, v13.b[1] ++ dup v10.8b, v13.b[2] ++ dup v11.8b, v13.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_8888_ca_asm_neon, 0, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_8888_ca_init, \ ++ pixman_composite_over_n_8888_8888_ca_cleanup, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_head, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_head ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} [B, G, R, A] ++ * mask in {v24, v25, v26} [B, G, R] ++ * output: updated src in {v0, v1, v2 } [B, G, R] ++ * updated mask in {v24, v25, v26} [B, G, R] ++ */ ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ umull v12.8h, v11.8b, v24.8b ++ umull v13.8h, v11.8b, v25.8b ++ umull v14.8h, v11.8b, v26.8b ++ urshr v15.8h, v0.8h, #8 ++ urshr v16.8h, v1.8h, #8 ++ urshr v17.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v15.8h ++ raddhn v1.8b, v1.8h, v16.8h ++ raddhn v2.8b, v2.8h, v17.8h ++ urshr v19.8h, v12.8h, #8 ++ urshr v20.8h, v13.8h, #8 ++ urshr v21.8h, v14.8h, #8 ++ raddhn v24.8b, v12.8h, v19.8h ++ raddhn v25.8b, v13.8h, v20.8h ++ /* ++ * convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ * and put data into v16 - blue, v17 - green, v18 - red ++ */ ++ mov v4.d[1], v5.d[0] ++ shrn v17.8b, v4.8h, #3 ++ shrn v18.8b, v4.8h, #8 ++ raddhn v26.8b, v14.8h, v21.8h ++ sli v4.8h, v4.8h, #5 ++ sri v18.8b, v18.8b, #5 ++ sri v17.8b, v17.8b, #6 ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in v16 - blue, v17 - green, v18 - red ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ shrn v16.8b, v4.8h, #2 ++ mvn v26.8b, v26.8b ++ umull v5.8h, v16.8b, v24.8b ++ umull v6.8h, v17.8b, v25.8b ++ umull v7.8h, v18.8b, v26.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_tail ++ /* ... continue 'combine_over_ca' replacement */ ++ urshr v13.8h, v5.8h, #8 ++ urshr v14.8h, v6.8h, #8 ++ urshr v15.8h, v7.8h, #8 ++ raddhn v16.8b, v13.8h, v5.8h ++ raddhn v17.8b, v14.8h, v6.8h ++ raddhn v18.8b, v15.8h, v7.8h ++ uqadd v16.8b, v0.8b, v16.8b ++ uqadd v17.8b, v1.8b, v17.8b ++ uqadd v18.8b, v2.8b, v18.8b ++ /* ++ * convert the results in v16, v17, v18 to r5g6b5 and store ++ * them into {v14} ++ */ ++ ushll v14.8h, v18.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v12.8h, v17.8b, #7 ++ sli v12.8h, v12.8h, #1 ++ ushll v13.8h, v16.8b, #7 ++ sli v13.8h, v13.8h, #1 ++ sri v14.8h, v12.8h, #5 ++ sri v14.8h, v13.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_tail_head ++ fetch_mask_pixblock ++ urshr v13.8h, v5.8h, #8 ++ urshr v14.8h, v6.8h, #8 ++ ld1 {v4.8h}, [DST_R], #16 ++ urshr v15.8h, v7.8h, #8 ++ raddhn v16.8b, v13.8h, v5.8h ++ raddhn v17.8b, v14.8h, v6.8h ++ raddhn v18.8b, v15.8h, v7.8h ++ mov v5.d[0], v4.d[1] ++ /* process_pixblock_head */ ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} [B, G, R, A] ++ * mask in {v24, v25, v26} [B, G, R] ++ * output: updated src in {v0, v1, v2 } [B, G, R] ++ * updated mask in {v24, v25, v26} [B, G, R] ++ */ ++ uqadd v16.8b, v0.8b, v16.8b ++ uqadd v17.8b, v1.8b, v17.8b ++ uqadd v18.8b, v2.8b, v18.8b ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ /* ++ * convert the result in v16, v17, v18 to r5g6b5 and store ++ * it into {v14} ++ */ ++ ushll v14.8h, v18.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v18.8h, v16.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v17.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ umull v12.8h, v11.8b, v24.8b ++ sri v14.8h, v19.8h, #5 ++ umull v13.8h, v11.8b, v25.8b ++ umull v15.8h, v11.8b, v26.8b ++ sri v14.8h, v18.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ cache_preload 8, 8 ++ urshr v16.8h, v0.8h, #8 ++ urshr v17.8h, v1.8h, #8 ++ urshr v18.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v16.8h ++ raddhn v1.8b, v1.8h, v17.8h ++ raddhn v2.8b, v2.8h, v18.8h ++ urshr v19.8h, v12.8h, #8 ++ urshr v20.8h, v13.8h, #8 ++ urshr v21.8h, v15.8h, #8 ++ raddhn v24.8b, v12.8h, v19.8h ++ raddhn v25.8b, v13.8h, v20.8h ++ /* ++ * convert 8 r5g6b5 pixel data from {v4, v5} to planar ++ * 8-bit format and put data into v16 - blue, v17 - green, ++ * v18 - red ++ */ ++ mov v4.d[1], v5.d[0] ++ shrn v17.8b, v4.8h, #3 ++ shrn v18.8b, v4.8h, #8 ++ raddhn v26.8b, v15.8h, v21.8h ++ sli v4.8h, v4.8h, #5 ++ sri v17.8b, v17.8b, #6 ++ sri v18.8b, v18.8b, #5 ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in v16 - blue, v17 - green, v18 - red ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ shrn v16.8b, v4.8h, #2 ++ mvn v26.8b, v26.8b ++ umull v5.8h, v16.8b, v24.8b ++ umull v6.8h, v17.8b, v25.8b ++ umull v7.8h, v18.8b, v26.8b ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_init ++ mov v13.s[0], w4 ++ dup v8.8b, v13.b[0] ++ dup v9.8b, v13.b[1] ++ dup v10.8b, v13.b[2] ++ dup v11.8b, v13.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_0565_ca_asm_neon, 0, 32, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_0565_ca_init, \ ++ pixman_composite_over_n_8888_0565_ca_cleanup, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_head, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_in_n_8_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ umull v8.8h, v4.8b, v3.8b ++ umull v9.8h, v5.8b, v3.8b ++ umull v10.8h, v6.8b, v3.8b ++ umull v11.8h, v7.8b, v3.8b ++.endm ++ ++.macro pixman_composite_in_n_8_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v28.8b, v8.8h, v14.8h ++ raddhn v29.8b, v9.8h, v15.8h ++ raddhn v30.8b, v10.8h, v12.8h ++ raddhn v31.8b, v11.8h, v13.8h ++.endm ++ ++.macro pixman_composite_in_n_8_process_pixblock_tail_head ++ pixman_composite_in_n_8_process_pixblock_tail ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ cache_preload 32, 32 ++ pixman_composite_in_n_8_process_pixblock_head ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_in_n_8_init ++ mov v3.s[0], w4 ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_in_n_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_in_n_8_asm_neon, 0, 0, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_in_n_8_init, \ ++ pixman_composite_in_n_8_cleanup, \ ++ pixman_composite_in_n_8_process_pixblock_head, \ ++ pixman_composite_in_n_8_process_pixblock_tail, \ ++ pixman_composite_in_n_8_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++.macro pixman_composite_add_n_8_8_process_pixblock_head ++ /* expecting source data in {v8, v9, v10, v11} */ ++ /* v8 - blue, v9 - green, v10 - red, v11 - alpha */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ /* mask is in v24, v25, v26, v27 */ ++ umull v0.8h, v24.8b, v11.8b ++ umull v1.8h, v25.8b, v11.8b ++ umull v2.8h, v26.8b, v11.8b ++ umull v3.8h, v27.8b, v11.8b ++ urshr v12.8h, v0.8h, #8 ++ urshr v13.8h, v1.8h, #8 ++ urshr v14.8h, v2.8h, #8 ++ urshr v15.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v12.8h ++ raddhn v1.8b, v1.8h, v13.8h ++ raddhn v2.8b, v2.8h, v14.8h ++ raddhn v3.8b, v3.8h, v15.8h ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_n_8_8_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_n_8_8_process_pixblock_tail_head ++ pixman_composite_add_n_8_8_process_pixblock_tail ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ fetch_mask_pixblock ++ cache_preload 32, 32 ++ pixman_composite_add_n_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_add_n_8_8_init ++ mov v11.s[0], w4 ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_add_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_n_8_8_init, \ ++ pixman_composite_add_n_8_8_cleanup, \ ++ pixman_composite_add_n_8_8_process_pixblock_head, \ ++ pixman_composite_add_n_8_8_process_pixblock_tail, \ ++ pixman_composite_add_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8_8_8_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* mask in {v24, v25, v26, v27} */ ++ umull v8.8h, v24.8b, v0.8b ++ umull v9.8h, v25.8b, v1.8b ++ umull v10.8h, v26.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ urshr v0.8h, v8.8h, #8 ++ urshr v1.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v0.8b, v0.8h, v8.8h ++ raddhn v1.8b, v1.8h, v9.8h ++ raddhn v2.8b, v12.8h, v10.8h ++ raddhn v3.8b, v13.8h, v11.8h ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_8_8_8_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_8_8_8_process_pixblock_tail_head ++ pixman_composite_add_8_8_8_process_pixblock_tail ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ fetch_mask_pixblock ++ fetch_src_pixblock ++ cache_preload 32, 32 ++ pixman_composite_add_8_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_add_8_8_8_init ++.endm ++ ++.macro pixman_composite_add_8_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8_8_8_asm_neon, 8, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_8_8_8_init, \ ++ pixman_composite_add_8_8_8_cleanup, \ ++ pixman_composite_add_8_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* mask in {v24, v25, v26, v27} */ ++ umull v8.8h, v27.8b, v0.8b ++ umull v9.8h, v27.8b, v1.8b ++ umull v10.8h, v27.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ /* 1 cycle bubble */ ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_tail ++ /* 2 cycle bubble */ ++ rshrn v28.8b, v8.8h, #8 ++ rshrn v29.8b, v9.8h, #8 ++ rshrn v30.8b, v10.8h, #8 ++ rshrn v31.8b, v11.8h, #8 ++ uqadd v28.8b, v4.8b, v28.8b ++ uqadd v29.8b, v5.8b, v29.8b ++ uqadd v30.8b, v6.8b, v30.8b ++ uqadd v31.8b, v7.8b, v31.8b ++.endm ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ rshrn v28.8b, v8.8h, #8 ++ fetch_mask_pixblock ++ rshrn v29.8b, v9.8h, #8 ++ umull v8.8h, v27.8b, v0.8b ++ rshrn v30.8b, v10.8h, #8 ++ umull v9.8h, v27.8b, v1.8b ++ rshrn v31.8b, v11.8h, #8 ++ umull v10.8h, v27.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ uqadd v28.8b, v4.8b, v28.8b ++ uqadd v29.8b, v5.8b, v29.8b ++ uqadd v30.8b, v6.8b, v30.8b ++ uqadd v31.8b, v7.8b, v31.8b ++ ursra v8.8h, v8.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ ursra v9.8h, v9.8h, #8 ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v10.8h, v10.8h, #8 ++ ++ cache_preload 8, 8 ++ ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8888_8888_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_add_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8_8888_asm_neon, 32, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_n_8_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_add_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_n_8_8888_init, \ ++ pixman_composite_add_n_8_8888_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_n_8888_init ++ mov v27.s[0], w6 ++ dup v27.8b, v27.b[3] ++.endm ++ ++.macro pixman_composite_add_8888_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_n_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_8888_n_8888_init, \ ++ pixman_composite_add_8888_n_8888_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* solid mask is in v15 */ ++ ++ /* 'in' */ ++ umull v11.8h, v15.8b, v3.8b ++ umull v10.8h, v15.8b, v2.8b ++ umull v9.8h, v15.8b, v1.8b ++ umull v8.8h, v15.8b, v0.8b ++ urshr v16.8h, v11.8h, #8 ++ urshr v14.8h, v10.8h, #8 ++ urshr v13.8h, v9.8h, #8 ++ urshr v12.8h, v8.8h, #8 ++ raddhn v3.8b, v11.8h, v16.8h ++ raddhn v2.8b, v10.8h, v14.8h ++ raddhn v1.8b, v9.8h, v13.8h ++ raddhn v0.8b, v8.8h, v12.8h ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ urshr v16.8h, v8.8h, #8 ++ urshr v17.8h, v9.8h, #8 ++ urshr v18.8h, v10.8h, #8 ++ urshr v19.8h, v11.8h, #8 ++ raddhn v28.8b, v16.8h, v8.8h ++ raddhn v29.8b, v17.8h, v9.8h ++ raddhn v30.8b, v18.8h, v10.8h ++ raddhn v31.8b, v19.8h, v11.8h ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8888_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_out_reverse_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_head ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_tail ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_init ++ mov v15.s[0], w6 ++ dup v15.8b, v15.b[3] ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_n_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_8888_n_8888_init, \ ++ pixman_composite_over_8888_n_8888_cleanup, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8888_8888_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_over_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_8_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8_8888_asm_neon, 32, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_tail_head ++ st3 {v0.8b, v1.8b, v2.8b}, [DST_W], #24 ++ fetch_src_pixblock ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_0888_asm_neon, 24, 0, 24, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_0888_process_pixblock_head, \ ++ pixman_composite_src_0888_0888_process_pixblock_tail, \ ++ pixman_composite_src_0888_0888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_head ++ mov v31.8b, v2.8b ++ mov v2.8b, v0.8b ++ mov v0.8b, v31.8b ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_tail_head ++ st4 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_W], #32 ++ fetch_src_pixblock ++ mov v31.8b, v2.8b ++ mov v2.8b, v0.8b ++ mov v0.8b, v31.8b ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_init ++ eor v3.8b, v3.8b, v3.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_8888_rev_asm_neon, 24, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ pixman_composite_src_0888_8888_rev_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_head, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_tail, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_head ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v2.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_tail ++ ushll v14.8h, v0.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_tail_head ++ ushll v14.8h, v0.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ fetch_src_pixblock ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ st1 {v14.8h}, [DST_W], #16 ++ ushll v9.8h, v2.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_0565_rev_asm_neon, 24, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_head, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_tail, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_head ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++.endm ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v30.8b, v11.8h, v8.8h ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v28.8b, v13.8h, v10.8h ++.endm ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_tail_head ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v31.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ fetch_src_pixblock ++ raddhn v30.8b, v11.8h, v8.8h ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v28.8b, v13.8h, v10.8h ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_pixbuf_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_head, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_tail, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_head ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++.endm ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v28.8b, v11.8h, v8.8h ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v30.8b, v13.8h, v10.8h ++.endm ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_tail_head ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ fetch_src_pixblock ++ raddhn v28.8b, v11.8h, v8.8h ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v30.8b, v13.8h, v10.8h ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_rpixbuf_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_head, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_tail, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v4.d[0], v8.d[0] ++ mov v4.d[1], v9.d[0] ++ mov v13.d[0], v10.d[0] ++ mov v13.d[1], v11.d[0] ++ convert_0565_to_x888 v4, v2, v1, v0 ++ convert_0565_to_x888 v13, v6, v5, v4 ++ /* source pixel data is in {v0, v1, v2, XX} */ ++ /* destination pixel data is in {v4, v5, v6, XX} */ ++ mvn v7.8b, v15.8b ++ umull v10.8h, v15.8b, v2.8b ++ umull v9.8h, v15.8b, v1.8b ++ umull v8.8h, v15.8b, v0.8b ++ umull v11.8h, v7.8b, v4.8b ++ umull v12.8h, v7.8b, v5.8b ++ umull v13.8h, v7.8b, v6.8b ++ urshr v19.8h, v10.8h, #8 ++ urshr v18.8h, v9.8h, #8 ++ urshr v17.8h, v8.8h, #8 ++ raddhn v2.8b, v10.8h, v19.8h ++ raddhn v1.8b, v9.8h, v18.8h ++ raddhn v0.8b, v8.8h, v17.8h ++.endm ++ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_tail ++ urshr v17.8h, v11.8h, #8 ++ urshr v18.8h, v12.8h, #8 ++ urshr v19.8h, v13.8h, #8 ++ raddhn v28.8b, v17.8h, v11.8h ++ raddhn v29.8b, v18.8h, v12.8h ++ raddhn v30.8b, v19.8h, v13.8h ++ uqadd v0.8b, v0.8b, v28.8b ++ uqadd v1.8b, v1.8b, v29.8b ++ uqadd v2.8b, v2.8b, v30.8b ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v30, v13 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_tail_head ++ fetch_mask_pixblock ++ pixman_composite_over_0565_8_0565_process_pixblock_tail ++ fetch_src_pixblock ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_over_0565_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_0565_8_0565_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_0565_n_0565_init ++ mov v15.s[0], w6 ++ dup v15.8b, v15.b[3] ++.endm ++ ++.macro pixman_composite_over_0565_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_0565_n_0565_asm_neon, 16, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_0565_n_0565_init, \ ++ pixman_composite_over_0565_n_0565_cleanup, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v4.d[0], v8.d[0] ++ mov v4.d[1], v9.d[0] ++ mov v13.d[0], v10.d[0] ++ mov v13.d[1], v11.d[0] ++ convert_0565_to_x888 v4, v2, v1, v0 ++ convert_0565_to_x888 v13, v6, v5, v4 ++ /* source pixel data is in {v0, v1, v2, XX} */ ++ /* destination pixel data is in {v4, v5, v6, XX} */ ++ umull v9.8h, v15.8b, v2.8b ++ umull v8.8h, v15.8b, v1.8b ++ umull v7.8h, v15.8b, v0.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v11.8h, v8.8h, #8 ++ urshr v10.8h, v7.8h, #8 ++ raddhn v2.8b, v9.8h, v12.8h ++ raddhn v1.8b, v8.8h, v11.8h ++ raddhn v0.8b, v7.8h, v10.8h ++.endm ++ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_tail ++ uqadd v0.8b, v0.8b, v4.8b ++ uqadd v1.8b, v1.8b, v5.8b ++ uqadd v2.8b, v2.8b, v6.8b ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v30, v13 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_tail_head ++ fetch_mask_pixblock ++ pixman_composite_add_0565_8_0565_process_pixblock_tail ++ fetch_src_pixblock ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_add_0565_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_0565_8_0565_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v12.d[0], v10.d[0] ++ mov v12.d[1], v11.d[0] ++ convert_0565_to_x888 v12, v6, v5, v4 ++ /* destination pixel data is in {v4, v5, v6, xx} */ ++ mvn v24.8b, v15.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v0.8b, v11.8h, v8.8h ++ raddhn v1.8b, v12.8h, v9.8h ++ raddhn v2.8b, v13.8h, v10.8h ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v12, v3 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_tail_head ++ fetch_src_pixblock ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_out_reverse_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_out_reverse_8_0565_asm_neon, 8, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_head, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 15, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_head ++ /* src is in v0 */ ++ /* destination pixel data is in {v4, v5, v6, v7} */ ++ mvn v1.8b, v0.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v1.8b, v4.8b ++ umull v9.8h, v1.8b, v5.8b ++ umull v10.8h, v1.8b, v6.8b ++ umull v11.8h, v1.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v12.8h, v10.8h ++ raddhn v31.8b, v13.8h, v11.8h ++ /* 32bpp result is in {v28, v29, v30, v31} */ ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ cache_preload 8, 8 ++ pixman_composite_out_reverse_8_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_out_reverse_8_8888_asm_neon, 8, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_8888_OVER_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_0565_OVER_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_0565_SRC_asm_neon, 32, 0, 16, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_0565_process_pixblock_head, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail_head, \ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_0565_8888_SRC_asm_neon, 16, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_8888_process_pixblock_head, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail_head ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_8_0565_OVER_asm_neon, 32, 8, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_0565_8_0565_OVER_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* ++ * Bilinear scaling support code which tries to provide pixel fetching, color ++ * format conversion, and interpolation as separate macros which can be used ++ * as the basic building blocks for constructing bilinear scanline functions. ++ */ ++ ++.macro bilinear_load_8888 reg1, reg2, tmp ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ ld1 {®1&.2s}, [TMP1], STRIDE ++ ld1 {®2&.2s}, [TMP1] ++.endm ++ ++.macro bilinear_load_0565 reg1, reg2, tmp ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ ld1 {®2&.s}[0], [TMP1], STRIDE ++ ld1 {®2&.s}[1], [TMP1] ++ convert_four_0565_to_x888_packed reg2, reg1, reg2, tmp ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_8888 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, tmp1, tmp2 ++ ++ bilinear_load_8888 reg1, reg2, tmp1 ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ bilinear_load_8888 reg3, reg4, tmp2 ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++.endm ++ ++.macro vzip reg1, reg2 ++ umov TMP4, v31.d[0] ++ zip1 v31.8b, reg1, reg2 ++ zip2 reg2, reg1, reg2 ++ mov reg1, v31.8b ++ mov v31.d[0], TMP4 ++.endm ++ ++.macro vuzp reg1, reg2 ++ umov TMP4, v31.d[0] ++ uzp1 v31.8b, reg1, reg2 ++ uzp2 reg2, reg1, reg2 ++ mov reg1, v31.8b ++ mov v31.d[0], TMP4 ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_0565 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, acc2lo, acc2hi ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&acc2&.s}[0], [TMP1], STRIDE ++ ld1 {&acc2&.s}[2], [TMP2], STRIDE ++ ld1 {&acc2&.s}[1], [TMP1] ++ ld1 {&acc2&.s}[3], [TMP2] ++ convert_0565_to_x888 acc2, reg3, reg2, reg1 ++ vzip ®1&.8b, ®3&.8b ++ vzip ®2&.8b, ®4&.8b ++ vzip ®3&.8b, ®4&.8b ++ vzip ®1&.8b, ®2&.8b ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_0565 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&xacc2&.s}[0], [TMP1], STRIDE ++ ld1 {&xacc2&.s}[2], [TMP2], STRIDE ++ ld1 {&xacc2&.s}[1], [TMP1] ++ ld1 {&xacc2&.s}[3], [TMP2] ++ convert_0565_to_x888 xacc2, xreg3, xreg2, xreg1 ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&yacc2&.s}[0], [TMP1], STRIDE ++ vzip &xreg1&.8b, &xreg3&.8b ++ ld1 {&yacc2&.s}[2], [TMP2], STRIDE ++ vzip &xreg2&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[1], [TMP1] ++ vzip &xreg3&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[3], [TMP2] ++ vzip &xreg1&.8b, &xreg2&.8b ++ convert_0565_to_x888 yacc2, yreg3, yreg2, yreg1 ++ umull &xacc1&.8h, &xreg1&.8b, v28.8b ++ vzip &yreg1&.8b, &yreg3&.8b ++ umlal &xacc1&.8h, &xreg2&.8b, v29.8b ++ vzip &yreg2&.8b, &yreg4&.8b ++ umull &xacc2&.8h, &xreg3&.8b, v28.8b ++ vzip &yreg3&.8b, &yreg4&.8b ++ umlal &xacc2&.8h, &xreg4&.8b, v29.8b ++ vzip &yreg1&.8b, &yreg2&.8b ++ umull &yacc1&.8h, &yreg1&.8b, v28.8b ++ umlal &yacc1&.8h, &yreg2&.8b, v29.8b ++ umull &yacc2&.8h, &yreg3&.8b, v28.8b ++ umlal &yacc2&.8h, &yreg4&.8b, v29.8b ++.endm ++ ++.macro bilinear_store_8888 numpix, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v0.2s, v1.2s}, [OUT], #16 ++.elseif numpix == 2 ++ st1 {v0.2s}, [OUT], #8 ++.elseif numpix == 1 ++ st1 {v0.s}[0], [OUT], #4 ++.else ++ .error bilinear_store_8888 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_store_0565 numpix, tmp1, tmp2 ++ vuzp v0.8b, v1.8b ++ vuzp v2.8b, v3.8b ++ vuzp v1.8b, v3.8b ++ vuzp v0.8b, v2.8b ++ convert_8888_to_0565 v2, v1, v0, v1, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v1.4h}, [OUT], #8 ++.elseif numpix == 2 ++ st1 {v1.s}[0], [OUT], #4 ++.elseif numpix == 1 ++ st1 {v1.h}[0], [OUT], #2 ++.else ++ .error bilinear_store_0565 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_interpolate_last_pixel src_fmt, dst_fmt ++ bilinear_load_&src_fmt v0, v1, v2 ++ umull v2.8h, v0.8b, v28.8b ++ umlal v2.8h, v1.8b, v29.8b ++ /* 5 cycles bubble */ ++ ushll v0.4s, v2.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v2.4h, v15.h[0] ++ umlal2 v0.4s, v2.8h, v15.h[0] ++ /* 5 cycles bubble */ ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ /* 3 cycles bubble */ ++ xtn v0.8b, v0.8h ++ /* 1 cycle bubble */ ++ bilinear_store_&dst_fmt 1, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_two_pixels src_fmt, dst_fmt ++ bilinear_load_and_vertical_interpolate_two_&src_fmt \ ++ v1, v11, v2, v3, v20, v21, v22, v23 ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ xtn v0.8b, v0.8h ++ bilinear_store_&dst_fmt 2, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_four_pixels src_fmt, dst_fmt ++ bilinear_load_and_vertical_interpolate_four_&src_fmt \ ++ v1, v11, v14, v20, v16, v17, v22, v23 \ ++ v3, v9, v24, v25, v26, v27, v18, v19 ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ sub TMP1, TMP1, STRIDE ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v3.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v3.4h, v15.h[0] ++ umlal2 v2.4s, v3.8h, v15.h[0] ++ ushll v8.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ prfm PREFETCH_MODE, [TMP2, PF_OFFS] ++ umlsl v8.4s, v9.4h, v15.h[4] ++ umlal2 v8.4s, v9.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v8.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ xtn v0.8b, v0.8h ++ xtn v1.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ bilinear_store_&dst_fmt 4, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_head ++.else ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_tail ++.endif ++.endm ++ ++.macro bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_tail_head ++.else ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_head ++.else ++ bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_tail src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_tail ++.else ++ bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_tail_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_tail_head ++.else ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.endif ++.endm ++ ++.set BILINEAR_FLAG_UNROLL_4, 0 ++.set BILINEAR_FLAG_UNROLL_8, 1 ++.set BILINEAR_FLAG_USE_ALL_NEON_REGS, 2 ++ ++/* ++ * Main template macro for generating NEON optimized bilinear scanline ++ * functions. ++ * ++ * Bilinear scanline scaler macro template uses the following arguments: ++ * fname - name of the function to generate ++ * src_fmt - source color format (8888 or 0565) ++ * dst_fmt - destination color format (8888 or 0565) ++ * bpp_shift - (1 << bpp_shift) is the size of source pixel in bytes ++ * prefetch_distance - prefetch in the source image by that many ++ * pixels ahead ++ */ ++ ++.macro generate_bilinear_scanline_func fname, src_fmt, dst_fmt, \ ++ src_bpp_shift, dst_bpp_shift, \ ++ prefetch_distance, flags ++ ++pixman_asm_function fname ++ OUT .req x0 ++ TOP .req x1 ++ BOTTOM .req x2 ++ WT .req x3 ++ WB .req x4 ++ X .req x5 ++ UX .req x6 ++ WIDTH .req x7 ++ TMP1 .req x8 ++ TMP2 .req x9 ++ PF_OFFS .req x10 ++ TMP3 .req x11 ++ TMP4 .req x12 ++ STRIDE .req x13 ++ ++ sxtw x3, w3 ++ sxtw x4, w4 ++ sxtw x5, w5 ++ sxtw x6, w6 ++ sxtw x7, w7 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 112 /* push all registers */ ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ stp x8, x9, [x29, -80] ++ stp x10, x11, [x29, -96] ++ stp x12, x13, [x29, -112] ++ ++ mov PF_OFFS, #prefetch_distance ++ mul PF_OFFS, PF_OFFS, UX ++ ++ subs STRIDE, BOTTOM, TOP ++ .unreq BOTTOM ++ ++ cmp WIDTH, #0 ++ ble 300f ++ ++ dup v12.8h, w5 ++ dup v13.8h, w6 ++ dup v28.8b, w3 ++ dup v29.8b, w4 ++ mov v25.d[0], v12.d[1] ++ mov v26.d[0], v13.d[0] ++ add v25.4h, v25.4h, v26.4h ++ mov v12.d[1], v25.d[0] ++ ++ /* ensure good destination alignment */ ++ cmp WIDTH, #1 ++ blt 100f ++ tst OUT, #(1 << dst_bpp_shift) ++ beq 100f ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ bilinear_interpolate_last_pixel src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #1 ++100: ++ add v13.8h, v13.8h, v13.8h ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ++ cmp WIDTH, #2 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 1)) ++ beq 100f ++ bilinear_interpolate_two_pixels src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #2 ++100: ++.if ((flags) & BILINEAR_FLAG_UNROLL_8) != 0 ++/*********** 8 pixels per iteration *****************/ ++ cmp WIDTH, #4 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 2)) ++ beq 100f ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #4 ++100: ++ subs WIDTH, WIDTH, #8 ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_interpolate_eight_pixels_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #8 ++ blt 500f ++1000: ++ bilinear_interpolate_eight_pixels_tail_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #8 ++ bge 1000b ++500: ++ bilinear_interpolate_eight_pixels_tail src_fmt, dst_fmt ++100: ++ tst WIDTH, #4 ++ beq 200f ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++200: ++.else ++/*********** 4 pixels per iteration *****************/ ++ subs WIDTH, WIDTH, #4 ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #4 ++ blt 500f ++1000: ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #4 ++ bge 1000b ++500: ++ bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++100: ++/****************************************************/ ++.endif ++ /* handle the remaining trailing pixels */ ++ tst WIDTH, #2 ++ beq 200f ++ bilinear_interpolate_two_pixels src_fmt, dst_fmt ++200: ++ tst WIDTH, #1 ++ beq 300f ++ bilinear_interpolate_last_pixel src_fmt, dst_fmt ++300: ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -104] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret ++ ++ .unreq OUT ++ .unreq TOP ++ .unreq WT ++ .unreq WB ++ .unreq X ++ .unreq UX ++ .unreq WIDTH ++ .unreq TMP1 ++ .unreq TMP2 ++ .unreq PF_OFFS ++ .unreq TMP3 ++ .unreq TMP4 ++ .unreq STRIDE ++.endfunc ++ ++.endm ++ ++/*****************************************************************************/ ++ ++.set have_bilinear_interpolate_four_pixels_8888_8888, 1 ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_head ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ ++ ld1 {v22.2s}, [TMP1], STRIDE ++ ld1 {v23.2s}, [TMP1] ++ asr TMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ umull v8.8h, v22.8b, v28.8b ++ umlal v8.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP2], STRIDE ++ ld1 {v23.2s}, [TMP2] ++ asr TMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umull v9.8h, v22.8b, v28.8b ++ umlal v9.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++.endm ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_tail ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ xtn v7.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_tail_head ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ ld1 {v20.2s}, [TMP1], STRIDE ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ ld1 {v21.2s}, [TMP1] ++ umull v8.8h, v20.8b, v28.8b ++ umlal v8.8h, v21.8b, v29.8b ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v4.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ld1 {v22.2s}, [TMP2], STRIDE ++ shrn2 v4.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ld1 {v23.2s}, [TMP2] ++ umull v9.8h, v22.8b, v28.8b ++ asr TMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ asr TMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umlal v9.8h, v23.8b, v29.8b ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ xtn v6.8b, v0.8h ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ xtn v7.8b, v4.8h ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ add v12.8h, v12.8h, v13.8h ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++.endm ++ ++/*****************************************************************************/ ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_SRC_asm_neon, 8888, 8888, \ ++ 2, 2, 28, BILINEAR_FLAG_UNROLL_4 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_0565_SRC_asm_neon, 8888, 0565, \ ++ 2, 1, 28, BILINEAR_FLAG_UNROLL_8 | BILINEAR_FLAG_USE_ALL_NEON_REGS ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_x888_SRC_asm_neon, 0565, 8888, \ ++ 1, 2, 28, BILINEAR_FLAG_UNROLL_4 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_0565_SRC_asm_neon, 0565, 0565, \ ++ 1, 1, 28, BILINEAR_FLAG_UNROLL_4 +diff --git a/pixman/pixman-arma64-neon-asm.h b/pixman/pixman-arma64-neon-asm.h +new file mode 100644 +index 0000000..0389d12 +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm.h +@@ -0,0 +1,1310 @@ ++/* ++ * Copyright © 2009 Nokia Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ */ ++ ++/* ++ * This file contains a macro ('generate_composite_function') which can ++ * construct 2D image processing functions, based on a common template. ++ * Any combinations of source, destination and mask images with 8bpp, ++ * 16bpp, 24bpp, 32bpp color formats are supported. ++ * ++ * This macro takes care of: ++ * - handling of leading and trailing unaligned pixels ++ * - doing most of the work related to L2 cache preload ++ * - encourages the use of software pipelining for better instructions ++ * scheduling ++ * ++ * The user of this macro has to provide some configuration parameters ++ * (bit depths for the images, prefetch distance, etc.) and a set of ++ * macros, which should implement basic code chunks responsible for ++ * pixels processing. See 'pixman-armv8-neon-asm.S' file for the usage ++ * examples. ++ * ++ * TODO: ++ * - try overlapped pixel method (from Ian Rickards) when processing ++ * exactly two blocks of pixels ++ * - maybe add an option to do reverse scanline processing ++ */ ++ ++/* ++ * Bit flags for 'generate_composite_function' macro which are used ++ * to tune generated functions behavior. ++ */ ++.set FLAG_DST_WRITEONLY, 0 ++.set FLAG_DST_READWRITE, 1 ++.set FLAG_DEINTERLEAVE_32BPP, 2 ++ ++/* ++ * Constants for selecting preferable prefetch type. ++ */ ++.set PREFETCH_TYPE_NONE, 0 /* No prefetch at all */ ++.set PREFETCH_TYPE_SIMPLE, 1 /* A simple, fixed-distance-ahead prefetch */ ++.set PREFETCH_TYPE_ADVANCED, 2 /* Advanced fine-grained prefetch */ ++ ++/* ++ * prefetch mode ++ * available modes are: ++ * pldl1keep ++ * pldl1strm ++ * pldl2keep ++ * pldl2strm ++ * pldl3keep ++ * pldl3strm ++ */ ++#define PREFETCH_MODE pldl1keep ++ ++/* ++ * Definitions of supplementary pixld/pixst macros (for partial load/store of ++ * pixel data). ++ */ ++ ++.macro pixldst1 op, elem_size, reg1, mem_operand, abits ++ op {v®1&.&elem_size}, [&mem_operand&], #8 ++.endm ++ ++.macro pixldst2 op, elem_size, reg1, reg2, mem_operand, abits ++ op {v®1&.&elem_size, v®2&.&elem_size}, [&mem_operand&], #16 ++.endm ++ ++.macro pixldst4 op, elem_size, reg1, reg2, reg3, reg4, mem_operand, abits ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size, v®4&.&elem_size}, [&mem_operand&], #32 ++.endm ++ ++.macro pixldst0 op, elem_size, reg1, idx, mem_operand, abits, bytes ++ op {v®1&.&elem_size}[idx], [&mem_operand&], #&bytes& ++.endm ++ ++.macro pixldst3 op, elem_size, reg1, reg2, reg3, mem_operand ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size}, [&mem_operand&], #24 ++.endm ++ ++.macro pixldst30 op, elem_size, reg1, reg2, reg3, idx, mem_operand ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size}[idx], [&mem_operand&], #3 ++.endm ++ ++.macro pixldst numbytes, op, elem_size, basereg, mem_operand, abits ++.if numbytes == 32 ++ .if elem_size==32 ++ pixldst4 op, 2s, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst4 op, 4h, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .else ++ pixldst4 op, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .endif ++.elseif numbytes == 16 ++ .if elem_size==32 ++ pixldst2 op, 2s, %(basereg+2), %(basereg+3), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst2 op, 4h, %(basereg+2), %(basereg+3), mem_operand, abits ++ .else ++ pixldst2 op, 8b, %(basereg+2), %(basereg+3), mem_operand, abits ++ .endif ++.elseif numbytes == 8 ++ .if elem_size==32 ++ pixldst1 op, 2s, %(basereg+1), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst1 op, 4h, %(basereg+1), mem_operand, abits ++ .else ++ pixldst1 op, 8b, %(basereg+1), mem_operand, abits ++ .endif ++.elseif numbytes == 4 ++ .if !RESPECT_STRICT_ALIGNMENT || (elem_size == 32) ++ pixldst0 op, s, %(basereg+0), 1, mem_operand, abits, 4 ++ .elseif elem_size == 16 ++ pixldst0 op, h, %(basereg+0), 2, mem_operand, abits, 2 ++ pixldst0 op, h, %(basereg+0), 3, mem_operand, abits, 2 ++ .else ++ pixldst0 op, b, %(basereg+0), 4, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 5, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 6, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 7, mem_operand, abits, 1 ++ .endif ++.elseif numbytes == 2 ++ .if !RESPECT_STRICT_ALIGNMENT || (elem_size == 16) ++ pixldst0 op, h, %(basereg+0), 1, mem_operand, abits, 2 ++ .else ++ pixldst0 op, b, %(basereg+0), 2, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 3, mem_operand, abits, 1 ++ .endif ++.elseif numbytes == 1 ++ pixldst0 op, b, %(basereg+0), 1, mem_operand, abits, 1 ++.else ++ .error "unsupported size: numbytes" ++.endif ++.endm ++ ++.macro pixld numpix, bpp, basereg, mem_operand, abits=0 ++.if bpp > 0 ++.if (bpp == 32) && (numpix == 8) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ pixldst4 ld4, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++.elseif (bpp == 24) && (numpix == 8) ++ pixldst3 ld3, 8b, %(basereg+3), %(basereg+4), %(basereg+5), mem_operand ++.elseif (bpp == 24) && (numpix == 4) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 4, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 5, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 6, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 7, mem_operand ++.elseif (bpp == 24) && (numpix == 2) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 2, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 3, mem_operand ++.elseif (bpp == 24) && (numpix == 1) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 1, mem_operand ++.else ++ pixldst %(numpix * bpp / 8), ld1, %(bpp), basereg, mem_operand, abits ++.endif ++.endif ++.endm ++ ++.macro pixst numpix, bpp, basereg, mem_operand, abits=0 ++.if bpp > 0 ++.if (bpp == 32) && (numpix == 8) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ pixldst4 st4, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++.elseif (bpp == 24) && (numpix == 8) ++ pixldst3 st3, 8b, %(basereg+3), %(basereg+4), %(basereg+5), mem_operand ++.elseif (bpp == 24) && (numpix == 4) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 4, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 5, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 6, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 7, mem_operand ++.elseif (bpp == 24) && (numpix == 2) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 2, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 3, mem_operand ++.elseif (bpp == 24) && (numpix == 1) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 1, mem_operand ++.elseif numpix * bpp == 32 && abits == 32 ++ pixldst 4, st1, 32, basereg, mem_operand, abits ++.elseif numpix * bpp == 16 && abits == 16 ++ pixldst 2, st1, 16, basereg, mem_operand, abits ++.else ++ pixldst %(numpix * bpp / 8), st1, %(bpp), basereg, mem_operand, abits ++.endif ++.endif ++.endm ++ ++.macro pixld_a numpix, bpp, basereg, mem_operand ++.if (bpp * numpix) <= 128 ++ pixld numpix, bpp, basereg, mem_operand, %(bpp * numpix) ++.else ++ pixld numpix, bpp, basereg, mem_operand, 128 ++.endif ++.endm ++ ++.macro pixst_a numpix, bpp, basereg, mem_operand ++.if (bpp * numpix) <= 128 ++ pixst numpix, bpp, basereg, mem_operand, %(bpp * numpix) ++.else ++ pixst numpix, bpp, basereg, mem_operand, 128 ++.endif ++.endm ++ ++/* ++ * Pixel fetcher for nearest scaling (needs TMP1, TMP2, VX, UNIT_X register ++ * aliases to be defined) ++ */ ++.macro pixld1_s elem_size, reg1, mem_operand ++.if elem_size == 16 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #1 ++ ld1 {v®1&.h}[0], [TMP1] ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ ld1 {v®1&.h}[1], [TMP2] ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #1 ++ ld1 {v®1&.h}[2], [TMP1] ++ ld1 {v®1&.h}[3], [TMP2] ++.elseif elem_size == 32 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #2 ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #2 ++ ld1 {v®1&.s}[0], [TMP1] ++ ld1 {v®1&.s}[1], [TMP2] ++.else ++ .error "unsupported" ++.endif ++.endm ++ ++.macro pixld2_s elem_size, reg1, reg2, mem_operand ++.if 0 /* elem_size == 32 */ ++ mov TMP1, VX, asr #16 ++ add VX, VX, UNIT_X, asl #1 ++ add TMP1, mem_operand, TMP1, asl #2 ++ mov TMP2, VX, asr #16 ++ sub VX, VX, UNIT_X ++ add TMP2, mem_operand, TMP2, asl #2 ++ ld1 {v®1&.s}[0], [TMP1] ++ mov TMP1, VX, asr #16 ++ add VX, VX, UNIT_X, asl #1 ++ add TMP1, mem_operand, TMP1, asl #2 ++ ld1 {v®2&.s}[0], [TMP2, :32] ++ mov TMP2, VX, asr #16 ++ add VX, VX, UNIT_X ++ add TMP2, mem_operand, TMP2, asl #2 ++ ld1 {v®1&.s}[1], [TMP1] ++ ld1 {v®2&.s}[1], [TMP2] ++.else ++ pixld1_s elem_size, reg1, mem_operand ++ pixld1_s elem_size, reg2, mem_operand ++.endif ++.endm ++ ++.macro pixld0_s elem_size, reg1, idx, mem_operand ++.if elem_size == 16 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ ld1 {v®1&.h}[idx], [TMP1] ++.elseif elem_size == 32 ++ asr DUMMY, VX, #16 ++ mov TMP1, DUMMY ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #2 ++ ld1 {v®1&.s}[idx], [TMP1] ++.endif ++.endm ++ ++.macro pixld_s_internal numbytes, elem_size, basereg, mem_operand ++.if numbytes == 32 ++ pixld2_s elem_size, %(basereg+4), %(basereg+5), mem_operand ++ pixld2_s elem_size, %(basereg+6), %(basereg+7), mem_operand ++ pixdeinterleave elem_size, %(basereg+4) ++.elseif numbytes == 16 ++ pixld2_s elem_size, %(basereg+2), %(basereg+3), mem_operand ++.elseif numbytes == 8 ++ pixld1_s elem_size, %(basereg+1), mem_operand ++.elseif numbytes == 4 ++ .if elem_size == 32 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++ .elseif elem_size == 16 ++ pixld0_s elem_size, %(basereg+0), 2, mem_operand ++ pixld0_s elem_size, %(basereg+0), 3, mem_operand ++ .else ++ pixld0_s elem_size, %(basereg+0), 4, mem_operand ++ pixld0_s elem_size, %(basereg+0), 5, mem_operand ++ pixld0_s elem_size, %(basereg+0), 6, mem_operand ++ pixld0_s elem_size, %(basereg+0), 7, mem_operand ++ .endif ++.elseif numbytes == 2 ++ .if elem_size == 16 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++ .else ++ pixld0_s elem_size, %(basereg+0), 2, mem_operand ++ pixld0_s elem_size, %(basereg+0), 3, mem_operand ++ .endif ++.elseif numbytes == 1 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++.else ++ .error "unsupported size: numbytes" ++.endif ++.endm ++ ++.macro pixld_s numpix, bpp, basereg, mem_operand ++.if bpp > 0 ++ pixld_s_internal %(numpix * bpp / 8), %(bpp), basereg, mem_operand ++.endif ++.endm ++ ++.macro vuzp8 reg1, reg2 ++ umov DUMMY, v16.d[0] ++ uzp1 v16.8b, v®1&.8b, v®2&.8b ++ uzp2 v®2&.8b, v®1&.8b, v®2&.8b ++ mov v®1&.8b, v16.8b ++ mov v16.d[0], DUMMY ++.endm ++ ++.macro vzip8 reg1, reg2 ++ umov DUMMY, v16.d[0] ++ zip1 v16.8b, v®1&.8b, v®2&.8b ++ zip2 v®2&.8b, v®1&.8b, v®2&.8b ++ mov v®1&.8b, v16.8b ++ mov v16.d[0], DUMMY ++.endm ++ ++/* deinterleave B, G, R, A channels for eight 32bpp pixels in 4 registers */ ++.macro pixdeinterleave bpp, basereg ++.if (bpp == 32) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ vuzp8 %(basereg+0), %(basereg+1) ++ vuzp8 %(basereg+2), %(basereg+3) ++ vuzp8 %(basereg+1), %(basereg+3) ++ vuzp8 %(basereg+0), %(basereg+2) ++.endif ++.endm ++ ++/* interleave B, G, R, A channels for eight 32bpp pixels in 4 registers */ ++.macro pixinterleave bpp, basereg ++.if (bpp == 32) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ vzip8 %(basereg+0), %(basereg+2) ++ vzip8 %(basereg+1), %(basereg+3) ++ vzip8 %(basereg+2), %(basereg+3) ++ vzip8 %(basereg+0), %(basereg+1) ++.endif ++.endm ++ ++/* ++ * This is a macro for implementing cache preload. The main idea is that ++ * cache preload logic is mostly independent from the rest of pixels ++ * processing code. It starts at the top left pixel and moves forward ++ * across pixels and can jump across scanlines. Prefetch distance is ++ * handled in an 'incremental' way: it starts from 0 and advances to the ++ * optimal distance over time. After reaching optimal prefetch distance, ++ * it is kept constant. There are some checks which prevent prefetching ++ * unneeded pixel lines below the image (but it still can prefetch a bit ++ * more data on the right side of the image - not a big issue and may ++ * be actually helpful when rendering text glyphs). Additional trick is ++ * the use of LDR instruction for prefetch instead of PLD when moving to ++ * the next line, the point is that we have a high chance of getting TLB ++ * miss in this case, and PLD would be useless. ++ * ++ * This sounds like it may introduce a noticeable overhead (when working with ++ * fully cached data). But in reality, due to having a separate pipeline and ++ * instruction queue for NEON unit in ARM Cortex-A8, normal ARM code can ++ * execute simultaneously with NEON and be completely shadowed by it. Thus ++ * we get no performance overhead at all (*). This looks like a very nice ++ * feature of Cortex-A8, if used wisely. We don't have a hardware prefetcher, ++ * but still can implement some rather advanced prefetch logic in software ++ * for almost zero cost! ++ * ++ * (*) The overhead of the prefetcher is visible when running some trivial ++ * pixels processing like simple copy. Anyway, having prefetch is a must ++ * when working with the graphics data. ++ */ ++.macro PF a, x:vararg ++.if (PREFETCH_TYPE_CURRENT == PREFETCH_TYPE_ADVANCED) ++ a x ++.endif ++.endm ++ ++.macro cache_preload std_increment, boost_increment ++.if (src_bpp_shift >= 0) || (dst_r_bpp != 0) || (mask_bpp_shift >= 0) ++.if std_increment != 0 ++ PF add PF_X, PF_X, #std_increment ++.endif ++ PF tst PF_CTL, #0xF ++ PF beq 71f ++ PF add PF_X, PF_X, #boost_increment ++ PF sub PF_CTL, PF_CTL, #1 ++71: ++ PF cmp PF_X, ORIG_W ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++.endif ++ PF ble 71f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++71: ++ PF ble 72f ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++.endif ++72: ++.endif ++.endm ++ ++.macro cache_preload_simple ++.if (PREFETCH_TYPE_CURRENT == PREFETCH_TYPE_SIMPLE) ++.if src_bpp > 0 ++ prfm PREFETCH_MODE, [SRC, #(PREFETCH_DISTANCE_SIMPLE * src_bpp / 8)] ++.endif ++.if dst_r_bpp > 0 ++ prfm PREFETCH_MODE, [DST_R, #(PREFETCH_DISTANCE_SIMPLE * dst_r_bpp / 8)] ++.endif ++.if mask_bpp > 0 ++ prfm PREFETCH_MODE, [MASK, #(PREFETCH_DISTANCE_SIMPLE * mask_bpp / 8)] ++.endif ++.endif ++.endm ++ ++.macro fetch_mask_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++.endm ++ ++/* ++ * Macro which is used to process leading pixels until destination ++ * pointer is properly aligned (at 16 bytes boundary). When destination ++ * buffer uses 16bpp format, this is unnecessary, or even pointless. ++ */ ++.macro ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++.if dst_w_bpp != 24 ++ tst DST_R, #0xF ++ beq 52f ++ ++.if src_bpp > 0 || mask_bpp > 0 || dst_r_bpp > 0 ++.irp lowbit, 1, 2, 4, 8, 16 ++local skip1 ++.if (dst_w_bpp <= (lowbit * 8)) && ((lowbit * 8) < (pixblock_size * dst_w_bpp)) ++.if lowbit < 16 /* we don't need more than 16-byte alignment */ ++ tst DST_R, #lowbit ++ beq 51f ++.endif ++ pixld_src (lowbit * 8 / dst_w_bpp), src_bpp, src_basereg, SRC ++ pixld (lowbit * 8 / dst_w_bpp), mask_bpp, mask_basereg, MASK ++.if dst_r_bpp > 0 ++ pixld_a (lowbit * 8 / dst_r_bpp), dst_r_bpp, dst_r_basereg, DST_R ++.else ++ add DST_R, DST_R, #lowbit ++.endif ++ PF add PF_X, PF_X, #(lowbit * 8 / dst_w_bpp) ++ sub W, W, #(lowbit * 8 / dst_w_bpp) ++51: ++.endif ++.endr ++.endif ++ pixdeinterleave src_bpp, src_basereg ++ pixdeinterleave mask_bpp, mask_basereg ++ pixdeinterleave dst_r_bpp, dst_r_basereg ++ ++ process_pixblock_head ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++ process_pixblock_tail ++ ++ pixinterleave dst_w_bpp, dst_w_basereg ++ ++.irp lowbit, 1, 2, 4, 8, 16 ++.if (dst_w_bpp <= (lowbit * 8)) && ((lowbit * 8) < (pixblock_size * dst_w_bpp)) ++.if lowbit < 16 /* we don't need more than 16-byte alignment */ ++ tst DST_W, #lowbit ++ beq 51f ++.endif ++.if src_bpp == 0 && mask_bpp == 0 && dst_r_bpp == 0 ++ sub W, W, #(lowbit * 8 / dst_w_bpp) ++.endif ++ pixst_a (lowbit * 8 / dst_w_bpp), dst_w_bpp, dst_w_basereg, DST_W ++51: ++.endif ++.endr ++.endif ++52: ++.endm ++ ++/* ++ * Special code for processing up to (pixblock_size - 1) remaining ++ * trailing pixels. As SIMD processing performs operation on ++ * pixblock_size pixels, anything smaller than this has to be loaded ++ * and stored in a special way. Loading and storing of pixel data is ++ * performed in such a way that we fill some 'slots' in the NEON ++ * registers (some slots naturally are unused), then perform compositing ++ * operation as usual. In the end, the data is taken from these 'slots' ++ * and saved to memory. ++ * ++ * cache_preload_flag - allows to suppress prefetch if ++ * set to 0 ++ * dst_aligned_flag - selects whether destination buffer ++ * is aligned ++ */ ++.macro process_trailing_pixels cache_preload_flag, \ ++ dst_aligned_flag, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ tst W, #(pixblock_size - 1) ++ beq 52f ++.if src_bpp > 0 || mask_bpp > 0 || dst_r_bpp > 0 ++.irp chunk_size, 16, 8, 4, 2, 1 ++.if pixblock_size > chunk_size ++ tst W, #chunk_size ++ beq 51f ++ pixld_src chunk_size, src_bpp, src_basereg, SRC ++ pixld chunk_size, mask_bpp, mask_basereg, MASK ++.if dst_aligned_flag != 0 ++ pixld_a chunk_size, dst_r_bpp, dst_r_basereg, DST_R ++.else ++ pixld chunk_size, dst_r_bpp, dst_r_basereg, DST_R ++.endif ++.if cache_preload_flag != 0 ++ PF add PF_X, PF_X, #chunk_size ++.endif ++51: ++.endif ++.endr ++.endif ++ pixdeinterleave src_bpp, src_basereg ++ pixdeinterleave mask_bpp, mask_basereg ++ pixdeinterleave dst_r_bpp, dst_r_basereg ++ ++ process_pixblock_head ++.if cache_preload_flag != 0 ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++.endif ++ process_pixblock_tail ++ pixinterleave dst_w_bpp, dst_w_basereg ++.irp chunk_size, 16, 8, 4, 2, 1 ++.if pixblock_size > chunk_size ++ tst W, #chunk_size ++ beq 51f ++.if dst_aligned_flag != 0 ++ pixst_a chunk_size, dst_w_bpp, dst_w_basereg, DST_W ++.else ++ pixst chunk_size, dst_w_bpp, dst_w_basereg, DST_W ++.endif ++51: ++.endif ++.endr ++52: ++.endm ++ ++/* ++ * Macro, which performs all the needed operations to switch to the next ++ * scanline and start the next loop iteration unless all the scanlines ++ * are already processed. ++ */ ++.macro advance_to_next_scanline start_of_loop_label ++ mov W, ORIG_W ++ add DST_W, DST_W, DST_STRIDE, lsl #dst_bpp_shift ++.if src_bpp != 0 ++ add SRC, SRC, SRC_STRIDE, lsl #src_bpp_shift ++.endif ++.if mask_bpp != 0 ++ add MASK, MASK, MASK_STRIDE, lsl #mask_bpp_shift ++.endif ++.if (dst_w_bpp != 24) ++ sub DST_W, DST_W, W, lsl #dst_bpp_shift ++.endif ++.if (src_bpp != 24) && (src_bpp != 0) ++ sub SRC, SRC, W, lsl #src_bpp_shift ++.endif ++.if (mask_bpp != 24) && (mask_bpp != 0) ++ sub MASK, MASK, W, lsl #mask_bpp_shift ++.endif ++ subs H, H, #1 ++ mov DST_R, DST_W ++ bge start_of_loop_label ++.endm ++ ++/* ++ * Registers are allocated in the following way by default: ++ * v0, v1, v2, v3 - reserved for loading source pixel data ++ * v4, v5, v6, v7 - reserved for loading destination pixel data ++ * v24, v25, v26, v27 - reserved for loading mask pixel data ++ * v28, v29, v30, v31 - final destination pixel data for writeback to memory ++ */ ++.macro generate_composite_function fname, \ ++ src_bpp_, \ ++ mask_bpp_, \ ++ dst_w_bpp_, \ ++ flags, \ ++ pixblock_size_, \ ++ prefetch_distance, \ ++ init, \ ++ cleanup, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head, \ ++ dst_w_basereg_ = 28, \ ++ dst_r_basereg_ = 4, \ ++ src_basereg_ = 0, \ ++ mask_basereg_ = 24 ++ ++ pixman_asm_function fname ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 232 /* push all registers */ ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ stp x8, x9, [x29, -80] ++ stp x10, x11, [x29, -96] ++ stp x12, x13, [x29, -112] ++ stp x14, x15, [x29, -128] ++ stp x16, x17, [x29, -144] ++ stp x18, x19, [x29, -160] ++ stp x20, x21, [x29, -176] ++ stp x22, x23, [x29, -192] ++ stp x24, x25, [x29, -208] ++ stp x26, x27, [x29, -224] ++ str x28, [x29, -232] ++ ++/* ++ * Select prefetch type for this function. If prefetch distance is ++ * set to 0 or one of the color formats is 24bpp, SIMPLE prefetch ++ * has to be used instead of ADVANCED. ++ */ ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_DEFAULT ++.if prefetch_distance == 0 ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_NONE ++.elseif (PREFETCH_TYPE_CURRENT > PREFETCH_TYPE_SIMPLE) && \ ++ ((src_bpp_ == 24) || (mask_bpp_ == 24) || (dst_w_bpp_ == 24)) ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_SIMPLE ++.endif ++ ++/* ++ * Make some macro arguments globally visible and accessible ++ * from other macros ++ */ ++ .set src_bpp, src_bpp_ ++ .set mask_bpp, mask_bpp_ ++ .set dst_w_bpp, dst_w_bpp_ ++ .set pixblock_size, pixblock_size_ ++ .set dst_w_basereg, dst_w_basereg_ ++ .set dst_r_basereg, dst_r_basereg_ ++ .set src_basereg, src_basereg_ ++ .set mask_basereg, mask_basereg_ ++ ++ .macro pixld_src x:vararg ++ pixld x ++ .endm ++ .macro fetch_src_pixblock ++ pixld_src pixblock_size, src_bpp, \ ++ (src_basereg - pixblock_size * src_bpp / 64), SRC ++ .endm ++/* ++ * Assign symbolic names to registers ++ */ ++ W .req x0 /* width (is updated during processing) */ ++ H .req x1 /* height (is updated during processing) */ ++ DST_W .req x2 /* destination buffer pointer for writes */ ++ DST_STRIDE .req x3 /* destination image stride */ ++ SRC .req x4 /* source buffer pointer */ ++ SRC_STRIDE .req x5 /* source image stride */ ++ MASK .req x6 /* mask pointer */ ++ MASK_STRIDE .req x7 /* mask stride */ ++ ++ DST_R .req x8 /* destination buffer pointer for reads */ ++ ++ PF_CTL .req x9 /* combined lines counter and prefetch */ ++ /* distance increment counter */ ++ PF_X .req x10 /* pixel index in a scanline for current */ ++ /* pretetch position */ ++ PF_SRC .req x11 /* pointer to source scanline start */ ++ /* for prefetch purposes */ ++ PF_DST .req x12 /* pointer to destination scanline start */ ++ /* for prefetch purposes */ ++ PF_MASK .req x13 /* pointer to mask scanline start */ ++ /* for prefetch purposes */ ++ ++ ORIG_W .req x14 /* saved original width */ ++ DUMMY .req x15 /* temporary register */ ++ ++ sxtw x0, w0 ++ sxtw x1, w1 ++ sxtw x3, w3 ++ sxtw x5, w5 ++ sxtw x7, w7 ++ ++ .set mask_bpp_shift, -1 ++.if src_bpp == 32 ++ .set src_bpp_shift, 2 ++.elseif src_bpp == 24 ++ .set src_bpp_shift, 0 ++.elseif src_bpp == 16 ++ .set src_bpp_shift, 1 ++.elseif src_bpp == 8 ++ .set src_bpp_shift, 0 ++.elseif src_bpp == 0 ++ .set src_bpp_shift, -1 ++.else ++ .error "requested src bpp (src_bpp) is not supported" ++.endif ++.if mask_bpp == 32 ++ .set mask_bpp_shift, 2 ++.elseif mask_bpp == 24 ++ .set mask_bpp_shift, 0 ++.elseif mask_bpp == 8 ++ .set mask_bpp_shift, 0 ++.elseif mask_bpp == 0 ++ .set mask_bpp_shift, -1 ++.else ++ .error "requested mask bpp (mask_bpp) is not supported" ++.endif ++.if dst_w_bpp == 32 ++ .set dst_bpp_shift, 2 ++.elseif dst_w_bpp == 24 ++ .set dst_bpp_shift, 0 ++.elseif dst_w_bpp == 16 ++ .set dst_bpp_shift, 1 ++.elseif dst_w_bpp == 8 ++ .set dst_bpp_shift, 0 ++.else ++ .error "requested dst bpp (dst_w_bpp) is not supported" ++.endif ++ ++.if (((flags) & FLAG_DST_READWRITE) != 0) ++ .set dst_r_bpp, dst_w_bpp ++.else ++ .set dst_r_bpp, 0 ++.endif ++.if (((flags) & FLAG_DEINTERLEAVE_32BPP) != 0) ++ .set DEINTERLEAVE_32BPP_ENABLED, 1 ++.else ++ .set DEINTERLEAVE_32BPP_ENABLED, 0 ++.endif ++ ++.if prefetch_distance < 0 || prefetch_distance > 15 ++ .error "invalid prefetch distance (prefetch_distance)" ++.endif ++ ++ PF mov PF_X, #0 ++ mov DST_R, DST_W ++ ++.if src_bpp == 24 ++ sub SRC_STRIDE, SRC_STRIDE, W ++ sub SRC_STRIDE, SRC_STRIDE, W, lsl #1 ++.endif ++.if mask_bpp == 24 ++ sub MASK_STRIDE, MASK_STRIDE, W ++ sub MASK_STRIDE, MASK_STRIDE, W, lsl #1 ++.endif ++.if dst_w_bpp == 24 ++ sub DST_STRIDE, DST_STRIDE, W ++ sub DST_STRIDE, DST_STRIDE, W, lsl #1 ++.endif ++ ++/* ++ * Setup advanced prefetcher initial state ++ */ ++ PF mov PF_SRC, SRC ++ PF mov PF_DST, DST_R ++ PF mov PF_MASK, MASK ++ /* PF_CTL = prefetch_distance | ((h - 1) << 4) */ ++ PF lsl DUMMY, H, #4 ++ PF mov PF_CTL, DUMMY ++ PF add PF_CTL, PF_CTL, #(prefetch_distance - 0x10) ++ ++ init ++ subs H, H, #1 ++ mov ORIG_W, W ++ blt 9f ++ cmp W, #(pixblock_size * 2) ++ blt 800f ++/* ++ * This is the start of the pipelined loop, which if optimized for ++ * long scanlines ++ */ ++0: ++ ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ /* Implement "head (tail_head) ... (tail_head) tail" loop pattern */ ++ pixld_a pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ PF add PF_X, PF_X, #pixblock_size ++ process_pixblock_head ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++ subs W, W, #(pixblock_size * 2) ++ blt 200f ++ ++100: ++ process_pixblock_tail_head ++ cache_preload_simple ++ subs W, W, #pixblock_size ++ bge 100b ++ ++200: ++ process_pixblock_tail ++ pixst_a pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++ ++ /* Process the remaining trailing pixels in the scanline */ ++ process_trailing_pixels 1, 1, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ advance_to_next_scanline 0b ++ ++ cleanup ++1000: ++ /* pop all registers */ ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -112] ++ ldp x14, x15, [x29, -128] ++ ldp x16, x17, [x29, -144] ++ ldp x18, x19, [x29, -160] ++ ldp x20, x21, [x29, -176] ++ ldp x22, x23, [x29, -192] ++ ldp x24, x25, [x29, -208] ++ ldp x26, x27, [x29, -224] ++ ldr x28, [x29, -232] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++/* ++ * This is the start of the loop, designed to process images with small width ++ * (less than pixblock_size * 2 pixels). In this case neither pipelining ++ * nor prefetch are used. ++ */ ++800: ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [SRC, DUMMY] ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [DST_R, DUMMY] ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [MASK, DUMMY] ++.endif ++ /* Process exactly pixblock_size pixels if needed */ ++ tst W, #pixblock_size ++ beq 100f ++ pixld pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ process_pixblock_head ++ process_pixblock_tail ++ pixst pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++100: ++ /* Process the remaining trailing pixels in the scanline */ ++ process_trailing_pixels 0, 0, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ advance_to_next_scanline 800b ++9: ++ cleanup ++ /* pop all registers */ ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -112] ++ ldp x14, x15, [x29, -128] ++ ldp x16, x17, [x29, -144] ++ ldp x18, x19, [x29, -160] ++ ldp x20, x21, [x29, -176] ++ ldp x22, x23, [x29, -192] ++ ldp x24, x25, [x29, -208] ++ ldp x26, x27, [x29, -224] ++ ldr x28, [x29, -232] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .purgem fetch_src_pixblock ++ .purgem pixld_src ++ ++ .unreq SRC ++ .unreq MASK ++ .unreq DST_R ++ .unreq DST_W ++ .unreq ORIG_W ++ .unreq W ++ .unreq H ++ .unreq SRC_STRIDE ++ .unreq DST_STRIDE ++ .unreq MASK_STRIDE ++ .unreq PF_CTL ++ .unreq PF_X ++ .unreq PF_SRC ++ .unreq PF_DST ++ .unreq PF_MASK ++ .unreq DUMMY ++ .endfunc ++.endm ++ ++/* ++ * A simplified variant of function generation template for a single ++ * scanline processing (for implementing pixman combine functions) ++ */ ++.macro generate_composite_function_scanline use_nearest_scaling, \ ++ fname, \ ++ src_bpp_, \ ++ mask_bpp_, \ ++ dst_w_bpp_, \ ++ flags, \ ++ pixblock_size_, \ ++ init, \ ++ cleanup, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head, \ ++ dst_w_basereg_ = 28, \ ++ dst_r_basereg_ = 4, \ ++ src_basereg_ = 0, \ ++ mask_basereg_ = 24 ++ ++ pixman_asm_function fname ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_NONE ++ ++/* ++ * Make some macro arguments globally visible and accessible ++ * from other macros ++ */ ++ .set src_bpp, src_bpp_ ++ .set mask_bpp, mask_bpp_ ++ .set dst_w_bpp, dst_w_bpp_ ++ .set pixblock_size, pixblock_size_ ++ .set dst_w_basereg, dst_w_basereg_ ++ .set dst_r_basereg, dst_r_basereg_ ++ .set src_basereg, src_basereg_ ++ .set mask_basereg, mask_basereg_ ++ ++.if use_nearest_scaling != 0 ++ /* ++ * Assign symbolic names to registers for nearest scaling ++ */ ++ W .req x0 ++ DST_W .req x1 ++ SRC .req x2 ++ VX .req x3 ++ UNIT_X .req x4 ++ SRC_WIDTH_FIXED .req x5 ++ MASK .req x6 ++ TMP1 .req x8 ++ TMP2 .req x9 ++ DST_R .req x10 ++ DUMMY .req x30 ++ ++ .macro pixld_src x:vararg ++ pixld_s x ++ .endm ++ ++ sxtw x0, w0 ++ sxtw x3, w3 ++ sxtw x4, w4 ++ sxtw x5, w5 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 88 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x8, x9, [x29, -80] ++ str x10, [x29, -88] ++.else ++ /* ++ * Assign symbolic names to registers ++ */ ++ W .req x0 /* width (is updated during processing) */ ++ DST_W .req x1 /* destination buffer pointer for writes */ ++ SRC .req x2 /* source buffer pointer */ ++ MASK .req x3 /* mask pointer */ ++ DST_R .req x4 /* destination buffer pointer for reads */ ++ DUMMY .req x30 ++ ++ .macro pixld_src x:vararg ++ pixld x ++ .endm ++ ++ sxtw x0, w0 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 64 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++.endif ++ ++.if (((flags) & FLAG_DST_READWRITE) != 0) ++ .set dst_r_bpp, dst_w_bpp ++.else ++ .set dst_r_bpp, 0 ++.endif ++.if (((flags) & FLAG_DEINTERLEAVE_32BPP) != 0) ++ .set DEINTERLEAVE_32BPP_ENABLED, 1 ++.else ++ .set DEINTERLEAVE_32BPP_ENABLED, 0 ++.endif ++ ++ .macro fetch_src_pixblock ++ pixld_src pixblock_size, src_bpp, \ ++ (src_basereg - pixblock_size * src_bpp / 64), SRC ++ .endm ++ ++ init ++ mov DST_R, DST_W ++ ++ cmp W, #pixblock_size ++ blt 800f ++ ++ ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ subs W, W, #pixblock_size ++ blt 700f ++ ++ /* Implement "head (tail_head) ... (tail_head) tail" loop pattern */ ++ pixld_a pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ process_pixblock_head ++ subs W, W, #pixblock_size ++ blt 200f ++100: ++ process_pixblock_tail_head ++ subs W, W, #pixblock_size ++ bge 100b ++200: ++ process_pixblock_tail ++ pixst_a pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++700: ++ /* Process the remaining trailing pixels in the scanline (dst aligned) */ ++ process_trailing_pixels 0, 1, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ cleanup ++.if use_nearest_scaling != 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldr x10, [x29, -96] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++.endif ++800: ++ /* Process the remaining trailing pixels in the scanline (dst unaligned) */ ++ process_trailing_pixels 0, 0, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ cleanup ++.if use_nearest_scaling != 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldr x10, [x29, -88] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .unreq DUMMY ++ .unreq DST_R ++ .unreq SRC ++ .unreq W ++ .unreq VX ++ .unreq UNIT_X ++ .unreq TMP1 ++ .unreq TMP2 ++ .unreq DST_W ++ .unreq MASK ++ .unreq SRC_WIDTH_FIXED ++ ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .unreq DUMMY ++ .unreq SRC ++ .unreq MASK ++ .unreq DST_R ++ .unreq DST_W ++ .unreq W ++.endif ++ ++ .purgem fetch_src_pixblock ++ .purgem pixld_src ++ ++ .endfunc ++.endm ++ ++.macro generate_composite_function_single_scanline x:vararg ++ generate_composite_function_scanline 0, x ++.endm ++ ++.macro generate_composite_function_nearest_scanline x:vararg ++ generate_composite_function_scanline 1, x ++.endm ++ ++/* Default prologue/epilogue, nothing special needs to be done */ ++ ++.macro default_init ++.endm ++ ++.macro default_cleanup ++.endm ++ ++/* ++ * Prologue/epilogue variant which additionally saves/restores v8-v15 ++ * registers (they need to be saved/restored by callee according to ABI). ++ * This is required if the code needs to use all the NEON registers. ++ */ ++ ++.macro default_init_need_all_regs ++.endm ++ ++.macro default_cleanup_need_all_regs ++.endm ++ ++/******************************************************************************/ ++ ++/* ++ * Conversion of 8 r5g6b6 pixels packed in 128-bit register (in) ++ * into a planar a8r8g8b8 format (with a, r, g, b color components ++ * stored into 64-bit registers out_a, out_r, out_g, out_b respectively). ++ * ++ * Warning: the conversion is destructive and the original ++ * value (in) is lost. ++ */ ++.macro convert_0565_to_8888 in, out_a, out_r, out_g, out_b ++ shrn &out_r&.8b, &in&.8h, #8 ++ shrn &out_g&.8b, &in&.8h, #3 ++ sli &in&.8h, &in&.8h, #5 ++ movi &out_a&.8b, #255 ++ sri &out_r&.8b, &out_r&.8b, #5 ++ sri &out_g&.8b, &out_g&.8b, #6 ++ shrn &out_b&.8b, &in&.8h, #2 ++.endm ++ ++.macro convert_0565_to_x888 in, out_r, out_g, out_b ++ shrn &out_r&.8b, &in&.8h, #8 ++ shrn &out_g&.8b, &in&.8h, #3 ++ sli &in&.8h, &in&.8h, #5 ++ sri &out_r&.8b, &out_r&.8b, #5 ++ sri &out_g&.8b, &out_g&.8b, #6 ++ shrn &out_b&.8b, &in&.8h, #2 ++.endm ++ ++/* ++ * Conversion from planar a8r8g8b8 format (with a, r, g, b color components ++ * in 64-bit registers in_a, in_r, in_g, in_b respectively) into 8 r5g6b6 ++ * pixels packed in 128-bit register (out). Requires two temporary 128-bit ++ * registers (tmp1, tmp2) ++ */ ++.macro convert_8888_to_0565 in_r, in_g, in_b, out, tmp1, tmp2 ++ ushll &tmp1&.8h, &in_g&.8b, #7 ++ shl &tmp1&.8h, &tmp1&.8h, #1 ++ ushll &out&.8h, &in_r&.8b, #7 ++ shl &out&.8h, &out&.8h, #1 ++ ushll &tmp2&.8h, &in_b&.8b, #7 ++ shl &tmp2&.8h, &tmp2&.8h, #1 ++ sri &out&.8h, &tmp1&.8h, #5 ++ sri &out&.8h, &tmp2&.8h, #11 ++.endm ++ ++/* ++ * Conversion of four r5g6b5 pixels (in) to four x8r8g8b8 pixels ++ * returned in (out0, out1) registers pair. Requires one temporary ++ * 64-bit register (tmp). 'out1' and 'in' may overlap, the original ++ * value from 'in' is lost ++ */ ++.macro convert_four_0565_to_x888_packed in, out0, out1, tmp ++ shl &out0&.4h, &in&.4h, #5 /* G top 6 bits */ ++ shl &tmp&.4h, &in&.4h, #11 /* B top 5 bits */ ++ sri &in&.4h, &in&.4h, #5 /* R is ready in top bits */ ++ sri &out0&.4h, &out0&.4h, #6 /* G is ready in top bits */ ++ sri &tmp&.4h, &tmp&.4h, #5 /* B is ready in top bits */ ++ ushr &out1&.4h, &in&.4h, #8 /* R is in place */ ++ sri &out0&.4h, &tmp&.4h, #8 /* G & B is in place */ ++ zip1 &tmp&.4h, &out0&.4h, &out1&.4h /* everything is in place */ ++ zip2 &out1&.4h, &out0&.4h, &out1&.4h ++ mov &out0&.d[0], &tmp&.d[0] ++.endm +diff --git a/pixman/pixman-private.h b/pixman/pixman-private.h +index 73a5414..22c8ccc +--- a/pixman/pixman-private.h ++++ b/pixman/pixman-private.h +@@ -607,6 +607,11 @@ pixman_implementation_t * + _pixman_implementation_create_arm_neon (pixman_implementation_t *fallback); + #endif + ++#ifdef USE_ARM_A64_NEON ++pixman_implementation_t * ++_pixman_implementation_create_arm_neon (pixman_implementation_t *fallback); ++#endif ++ + #ifdef USE_MIPS_DSPR2 + pixman_implementation_t * + _pixman_implementation_create_mips_dspr2 (pixman_implementation_t *fallback); +-- +2.8.0 + diff --git a/FunKey/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch b/FunKey/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch new file mode 100644 index 0000000..46e9524 --- /dev/null +++ b/FunKey/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch @@ -0,0 +1,26 @@ +Merge this bug as it can affect performance : +https://github.com/OpenDingux/SDL/commit/e51100dce8da9099278dac9f5affbecf6396378b + +--- a/src/audio/alsa/SDL_alsa_audio.c ++++ b/src/audio/alsa/SDL_alsa_audio.c +@@ -479,6 +479,10 @@ + return(-1); + } + ++ /* Switch to blocking mode for playback */ ++ /* Note: this must happen before hw/sw params are set. */ ++ SDL_NAME(snd_pcm_nonblock)(pcm_handle, 0); ++ + /* Figure out what the hardware is capable of */ + snd_pcm_hw_params_alloca(&hwparams); + status = SDL_NAME(snd_pcm_hw_params_any)(pcm_handle, hwparams); +@@ -611,9 +615,6 @@ + } + SDL_memset(mixbuf, spec->silence, spec->size); + +- /* Switch to blocking mode for playback */ +- SDL_NAME(snd_pcm_nonblock)(pcm_handle, 0); +- + /* We're ready to rock and roll. :-) */ + return(0); + } diff --git a/FunKey/board/funkey/patches/sdl/sdl-fix-kb-input.patch b/FunKey/board/funkey/patches/sdl/sdl-fix-kb-input.patch new file mode 100644 index 0000000..8f7db83 --- /dev/null +++ b/FunKey/board/funkey/patches/sdl/sdl-fix-kb-input.patch @@ -0,0 +1,22 @@ +diff --git a/src/video/fbcon/SDL_fbevents.c b/src/video/fbcon/SDL_fbevents.c +index 5e369a4..549a7ad 100644 +--- a/src/video/fbcon/SDL_fbevents.c ++++ b/src/video/fbcon/SDL_fbevents.c +@@ -270,17 +270,6 @@ int FB_OpenKeyboard(_THIS) + fprintf(stderr, "vtpath = %s, fd = %d\n", + vtpath, keyboard_fd); + #endif /* DEBUG_KEYBOARD */ +- +- /* This needs to be our controlling tty +- so that the kernel ioctl() calls work +- */ +- if ( keyboard_fd >= 0 ) { +- tty0_fd = open("/dev/tty", O_RDWR, 0); +- if ( tty0_fd >= 0 ) { +- ioctl(tty0_fd, TIOCNOTTY, 0); +- close(tty0_fd); +- } +- } + } + } + if ( keyboard_fd < 0 ) { diff --git a/FunKey/board/funkey/rootfs-overlay/etc/fstab b/FunKey/board/funkey/rootfs-overlay/etc/fstab index 47ade62..a729a33 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/fstab +++ b/FunKey/board/funkey/rootfs-overlay/etc/fstab @@ -8,4 +8,4 @@ tmpfs /run tmpfs mode=0755,nosuid,nodev 0 0 sysfs /sys sysfs defaults 0 0 /dev/mmcblk0p3 none swap sw 0 0 configfs /sys/kernel/config configfs rw,relatime 0 0 -/dev/mmcblk0p4 /mnt vfat rw,relatime,fmask=0022,dmask=0022,iocharset=iso8859-1,shortname=mixed,errors=remount-ro 0 0 +/dev/mmcblk0p4 /mnt vfat rw,relatime,fmask=0022,dmask=0022,iocharset=iso8859-1,shortname=mixed 0 0 diff --git a/FunKey/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf b/FunKey/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf index 11a4e43..b3500da 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf +++ b/FunKey/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf @@ -1,6 +1,6 @@ -################################## -# Funkey GPIO keymap config file # -################################## +#################################### +# Funkey GPIO keymap config file # +#################################### # Format: # # - First all GPIO Pin numbers must be declared (integers separated by commas) @@ -21,19 +21,20 @@ -################################### -# Pins declaration: +######################## +# Pins declaration # +######################## 0,1,2,3,4,6,7,10*,11,12,13,14,15 -################################### -# Mapping: +############### +# Mapping # +############### #7, KEYBOARD, KEY_F, KEY_F, Fn #7+6, KEYBOARD, KEY_K, KEY_K, Select 7, KEYBOARD, KEY_K, KEY_K, Select 6, KEYBOARD, KEY_S, KEY_S, Start 3, KEYBOARD, KEY_U, KEY_U, Up -7+3, KEYBOARD, KEY_P, KEY_P, Quick Save 4, KEYBOARD, KEY_L, KEY_L, Left 7+4, KEYBOARD, KEY_J, KEY_J, Aspect ratio factor -- 1, KEYBOARD, KEY_D, KEY_D, Down @@ -49,18 +50,13 @@ 13, KEYBOARD, KEY_Y, KEY_Y, Y 11, KEYBOARD, KEY_X, KEY_X, X +7+3, SHELL_COMMAND, snap, Fn+Start, Snapshot 7+12, SHELL_COMMAND, quick_action_volume_up, Fn+Y, Volume++ 7+13, SHELL_COMMAND, quick_action_volume_down, Fn+A, Volume-- 7+11, SHELL_COMMAND, quick_action_bright_up, Fn+B, Brightness++ 7+14, SHELL_COMMAND, quick_action_bright_down, Fn+X, Brightness-- -10, SHELL_COMMAND, sched_shutdown 1 & signal_usr1_to_emulators, N_OE, Quick save and Poweroff because of N_OE +10, SHELL_COMMAND, sched_shutdown 0.1, N_OE, Instant Play save and Poweroff because of N_OE 7+15+2, SHELL_COMMAND, display_notif_system_stats, Fn+L1+L2, display system cpu and ram usage -#7+12, KEYBOARD, KEY_G, KEY_G, Brightness++ -#7+14, KEYBOARD, KEY_E, KEY_E, Volume-- -#7+13, KEYBOARD, KEY_W, KEY_W, Brightness-- -#7+11, KEYBOARD, KEY_C, KEY_C, Volume++ -#10, KEYBOARD, KEY_T, KEY_T, Should Poweroff because N_OE_received -#7+15, KEYBOARD, KEY_Q, 7+11, Launch menu diff --git a/FunKey/board/funkey/rootfs-overlay/etc/init.d/S01first_boot b/FunKey/board/funkey/rootfs-overlay/etc/init.d/S01first_boot index a04a285..590b8cb 100755 --- a/FunKey/board/funkey/rootfs-overlay/etc/init.d/S01first_boot +++ b/FunKey/board/funkey/rootfs-overlay/etc/init.d/S01first_boot @@ -4,20 +4,10 @@ THIS=$(basename $0) case "$1" in start) - - # Check is SWAP partition already created - fdisk -l /dev/mmcblk0 | grep "Linux swap" > /dev/null - if [ $? -ne 0 ]; then - first_boot - exit $? - fi - - # Check is share partition already created - fdisk -l /dev/mmcblk0 | grep "W95 FAT32" > /dev/null - if [ $? -ne 0 ]; then - first_boot - exit $? - fi + first_boot_ok=$(fw_printenv -n first_boot_ok 2>/dev/null) + if ! [ ! "${first_boot_ok}" -ne "${first_boot_ok}" ] 2> /dev/null; then + first_boot + fi ;; stop) ;; diff --git a/FunKey/board/funkey/rootfs-overlay/etc/inittab b/FunKey/board/funkey/rootfs-overlay/etc/inittab index 005aff9..b0d4598 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/inittab +++ b/FunKey/board/funkey/rootfs-overlay/etc/inittab @@ -38,7 +38,7 @@ null::sysinit:/bin/ln -sf /proc/self/fd/2 /dev/stderr #::ctrlaltdel:/sbin/reboot # Stuff to do before rebooting -::shutdown:/etc/init.d/rcK -::shutdown:/sbin/swapoff -a -::shutdown:/bin/umount -r / -::shutdown:/bin/umount -r /mnt +#::shutdown:/etc/init.d/rcK +#::shutdown:/sbin/swapoff -a +#::shutdown:/bin/umount -r / +#::shutdown:/bin/umount -r /mnt diff --git a/FunKey/board/funkey/rootfs-overlay/etc/issue b/FunKey/board/funkey/rootfs-overlay/etc/issue index c3d1cfd..6acbb23 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/issue +++ b/FunKey/board/funkey/rootfs-overlay/etc/issue @@ -5,6 +5,6 @@ |___| |_____|__|__||__|\__||_____|___ | FUN ON A KEYCHAIN |_____| ----------------------------------------------------- - Version 1.1.0 + Version 2.0.0 ----------------------------------------------------- diff --git a/FunKey/board/funkey/rootfs-overlay/etc/os-release b/FunKey/board/funkey/rootfs-overlay/etc/os-release index 5e259a6..625c497 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/os-release +++ b/FunKey/board/funkey/rootfs-overlay/etc/os-release @@ -1,12 +1,12 @@ NAME="FunKey-OS" -VERSION="1.1.0 (Quacking Quagga)" +VERSION="2.0.0 (Rowdy Rabbit)" ID=funkey ID_LIKE=buildroot -PRETTY_NAME="FunKey-OS 1.1.0" -VERSION_ID="1.1.0" +PRETTY_NAME="FunKey-OS 2.0.0" +VERSION_ID="2.0.0.0" HOME_URL="https://www.funkey-project.com/" SUPPORT_URL="https://www.funkey-project.com/" BUG_REPORT_URL="https://www.funkey-project.com/" PRIVACY_POLICY_URL="https://www.funkey-project.com" -VERSION_CODENAME=Quacking -UBUNTU_CODENAME=Quacking +VERSION_CODENAME=Rowdy +UBUNTU_CODENAME=Rowdy diff --git a/FunKey/board/funkey/rootfs-overlay/etc/profile b/FunKey/board/funkey/rootfs-overlay/etc/profile index 8a3692e..831311f 100755 --- a/FunKey/board/funkey/rootfs-overlay/etc/profile +++ b/FunKey/board/funkey/rootfs-overlay/etc/profile @@ -11,7 +11,6 @@ fi export PAGER='/bin/more' export EDITOR='/usr/bin/nano' export SDL_NOMOUSE=1 -export MEDNAFEN_HOME=/tmp/funkey/.mednafen # Source configuration files from /etc/profile.d for i in /etc/profile.d/*.sh ; do diff --git a/FunKey/board/funkey/rootfs-overlay/etc/sw-versions b/FunKey/board/funkey/rootfs-overlay/etc/sw-versions index fc6545a..44869b1 100644 --- a/FunKey/board/funkey/rootfs-overlay/etc/sw-versions +++ b/FunKey/board/funkey/rootfs-overlay/etc/sw-versions @@ -1 +1 @@ -rootfs 1.1.0 +rootfs 2.0.0 diff --git a/FunKey/board/funkey/rootfs-overlay/media b/FunKey/board/funkey/rootfs-overlay/media new file mode 120000 index 0000000..cca5abd --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/media @@ -0,0 +1 @@ +/mnt \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/opk/.empty b/FunKey/board/funkey/rootfs-overlay/opk/.empty new file mode 100644 index 0000000..e69de29 diff --git a/FunKey/board/funkey/rootfs-overlay/root/.gmenu2x b/FunKey/board/funkey/rootfs-overlay/root/.gmenu2x new file mode 120000 index 0000000..117526c --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/root/.gmenu2x @@ -0,0 +1 @@ +/mnt/funkey/.gmenu2x \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/root/.profile b/FunKey/board/funkey/rootfs-overlay/root/.profile index eb33949..cad4256 100755 --- a/FunKey/board/funkey/rootfs-overlay/root/.profile +++ b/FunKey/board/funkey/rootfs-overlay/root/.profile @@ -18,6 +18,14 @@ alias l='ls $LS_OPTIONS -lA' # alias cp='cp -i' # alias mv='mv -i' +# Relocate HOME into the r/w partition +export HOME=/mnt/FunKey +mkdir -p "${HOME}" +export MEDNAFEN_HOME=$HOME/.mednafen +mkdir -p "${MEDNAFEN_HOME}" +export GMENU2X_HOME="$HOME/.gmenu2x" +mkdir -p "${GMENU2X_HOME}" + # Resize the console to the terminal dimensions resize() { if [[ -t 0 && $# -eq 0 ]]; then @@ -37,11 +45,9 @@ resize() { } -# Start ampli if necessary -echo "Start audio amplifier if necessary" -if [[ "$(volume_get)" -ne "0" ]]; then - start_audio_amp 1 >/dev/null 2>&1 -fi +# Start ampli +echo "Start audio amplifier" +start_audio_amp 1 >/dev/null 2>&1 # Force unmute sound card and reset volume echo "Force unmute sound card and reset volume" diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gamegear_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gamegear_launch.sh index ad784a4..8364468 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gamegear_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gamegear_launch.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -fs 1 -gg.stretch full "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -fs 1 -gg.stretch full "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gb_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gb_launch.sh index e720655..4246680 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gb_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gb_launch.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -sdlgnuboy --syncrtc "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +sdlgnuboy --syncrtc "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_gpsp.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_gpsp.sh index 583d447..fb4d910 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_gpsp.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_gpsp.sh @@ -1,6 +1,10 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID cd ${HOME} -gpsp "$1" +gpsp "$1"& +record_pid $! +wait $! +erase_pid + diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_mednafen.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_mednafen.sh index 583d447..88dd220 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_mednafen.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/gba_launch_mednafen.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -gpsp "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +gpsp "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/lynx_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/lynx_launch.sh index b49b791..300941f 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/lynx_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/lynx_launch.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -fs 1 -lynx.stretch full "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -fs 1 -lynx.stretch full "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/mame_launch_mednafen.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/mame_launch_mednafen.sh index cc29289..cca6afe 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/mame_launch_mednafen.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/mame_launch_mednafen.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -sound 1 -soundrate 22050 -soundbufsize 100 -vdriver sdl -frameskip 1 -fs 0 "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -sound 1 -soundrate 22050 -soundbufsize 100 -vdriver sdl -frameskip 1 -fs 0 "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/megadrive_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/megadrive_launch.sh index 0186681..cf3805a 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/megadrive_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/megadrive_launch.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -PicoDrive "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +PicoDrive "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/nes_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/nes_launch.sh index 0555e12..e241779 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/nes_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/nes_launch.sh @@ -1,7 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -fceux "$1" - +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +fceux "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/ngp_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/ngp_launch.sh index 0e50eba..54e6eb9 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/ngp_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/ngp_launch.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -fs 1 -ngp.stretch full "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -fs 1 -ngp.stretch full "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/pce_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/pce_launch.sh index 9bd08eb..6c3f914 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/pce_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/pce_launch.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -fs 1 -force_module pce_fast -pce_fast.stretch full "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -fs 1 -force_module pce_fast -pce_fast.stretch full "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_mednafen.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_mednafen.sh index 3afc12d..b81a321 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_mednafen.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_mednafen.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -pcsx -frameskip -cdfile "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +pcsx -frameskip -cdfile "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_pcsx.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_pcsx.sh index f6c3e48..b2a0613 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_pcsx.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/psone_launch_pcsx.sh @@ -1,7 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -#pcsx -frameskip -cdfile "$1" -pcsx -cdfile "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +pcsx -cdfile "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/sms_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/sms_launch.sh index 0186681..cf3805a 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/sms_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/sms_launch.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -PicoDrive "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +PicoDrive "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch.sh index 5a5e718..e0b39cf 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -psnes "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +psnes "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch_mednafen.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch_mednafen.sh index 5a5e718..e0b39cf 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch_mednafen.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/snes_launch_mednafen.sh @@ -1,6 +1,8 @@ #!/bin/sh -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -psnes "$1" +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +psnes "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/wonderswan_launch.sh b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/wonderswan_launch.sh index b274169..69f7ca2 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/wonderswan_launch.sh +++ b/FunKey/board/funkey/rootfs-overlay/usr/games/launchers/wonderswan_launch.sh @@ -1,8 +1,10 @@ #!/bin/sh -# Security cp /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ -export HOME=/tmp/funkey -mkdir -p ${HOME} -cd ${HOME} -mednafen -fs 1 -wswan.stretch full "$1" + +# Launch the process in background, record the PID into a file, wait +# for the process to terminate and erase the recorded PID +mednafen -fs 1 -wswan.stretch full "$1"& +record_pid $! +wait $! +erase_pid diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/lib/utils b/FunKey/board/funkey/rootfs-overlay/usr/local/lib/utils index a770cca..ec02b68 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/lib/utils +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/lib/utils @@ -17,6 +17,14 @@ if [ $? -ne 0 ]; then warn "$@" exit $return_code } + + die_notif () { + local return_code=$1 + shift + warn "$@" + notif "$@" + exit $return_code + } fi notif () { diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/assembly_tests b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/assembly_tests index 1b9f2c2..12a92e5 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/assembly_tests +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/assembly_tests @@ -16,6 +16,10 @@ if [ -f ${LOCK_FILE} ]; then fi touch ${LOCK_FILE} +## Register ourself as the running FunKey task for receiving USR1 +## signal for shutting down +echo $$ > "/var/run/funkey.pid" + ## Binaries PROD_SCREEN_BIN="/usr/local/sbin/funkey_prod_screens" GET_PROC_UID="/usr/local/sbin/get_sid" @@ -28,7 +32,7 @@ stop_loop=false proc_uid=$($GET_PROC_UID) ## Defines -VERSION="1.00" +VERSION="1.01" LOG_FILE="/root/logs/assembly_tests/assy_tests_${proc_uid}.log" [ -d $(dirname $LOG_FILE) ] || mkdir -p $(dirname $LOG_FILE) MAGNET_DETECTED_FILE="/root/.assembly_tests_magnet_detected" @@ -40,6 +44,9 @@ function function_magnet_detected_ok { ## Kill scheduled shutdown pkill sched_shutdown + ## Kill the funkey_prod_screen binary + killall -s USR1 "${PROD_SCREEN_BIN} > /dev/null 2>&1 + ## Write magnet_detected file if $test_failed; then echo "1" > $MAGNET_DETECTED_FILE @@ -58,6 +65,9 @@ function function_magnet_detected_ok { ## Function called when SIGUSR1 is caught while NOT waiting for it function function_magnet_detected_ko { echo "ERROR: Caught SIGUSR1 signal (magnet detected!)" + + ## Kill the funkey_prod_screen binary + killall -s USR1 "${PROD_SCREEN_BIN} > /dev/null 2>&1 sync } diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/brightness_get b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/brightness_get index d664293..399bc37 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/brightness_get +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/brightness_get @@ -12,7 +12,7 @@ fi brightness=$(fw_printenv -n brightness 2>/dev/null) if ! [ ! "${brightness}" -ne "${brightness}" ] 2> /dev/null; then brightness=${BRIGHTNESS_DEFAULT_VALUE} - fw_setenv brightness ${brightness} + fw_setenv brightness ${brightness} fi echo ${brightness} exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/cancel_sched_powerdown b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/cancel_sched_powerdown new file mode 100755 index 0000000..2952a44 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/cancel_sched_powerdown @@ -0,0 +1,11 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +if [ ${#} != 0 ]; then + echo "Usage: $(basename ${0})" + exit 1 +fi +pkill sched_powerdown +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/display_notif_system_stats b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/display_notif_system_stats index 5c4638d..a3cbd27 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/display_notif_system_stats +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/display_notif_system_stats @@ -1,5 +1,5 @@ #!/bin/sh -notif_set 0 "Getting system stats..." +notif_set 0 " Getting system stats..." killall -s USR1 system_stats exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/erase_pid b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/erase_pid new file mode 100755 index 0000000..afb3597 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/erase_pid @@ -0,0 +1,12 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +# Check args +if [ ${#} -ne 0 ]; then + echo "Usage: $(basename ${0})" + exit 1 +fi +rm -f /var/run/funkey.pid +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/first_boot b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/first_boot index 2ab0721..03b022a 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/first_boot +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/first_boot @@ -14,31 +14,40 @@ SELF=$(basename $0) root_part=$(cat /proc/cmdline | sed -n 's|^.*root=\([^ ]*\).*|\1|p') root_part_num=${root_part#/dev/mmcblk0p} if [ "${root_part_num}" -eq 1 ]; then - die 0 "recovery mode" + die_notif 0 "recovery mode" elif [ "${root_part_num}" = "{$root_part}" ]; then - die 1 "${root_part} is not an SD card partition, aborting" + die_notif 1 "${root_part} is not an SD card partition, aborting" elif [ "${root_part_num}" -ne 2 ]; then - die 2 "unknown partition layout, aborting" + die_notif 2 "unknown partition layout, aborting" fi let swap_part_num=${root_part_num}+1 swap_part=/dev/mmcblk0p${swap_part_num} -let share_part_num=${swap_part_num}+1 -share_part=/dev/mmcblk0p${share_part_num} +let usb_part_num=${swap_part_num}+1 +usb_part=/dev/mmcblk0p${usb_part_num} check_root_id () { - [ $(id -u) -ne 0 ] && die 3 "this script must be run as root, aborting" + [ $(id -u) -ne 0 ] && die_notif 3 "this script must be run as root, aborting" return 0 } resize_rootfs_partition () { + # Check if root partition is already resized + local rootfs_part_line=$(fdisk -l /dev/mmcblk0 2>/dev/null | grep ${root_part}) + set ${rootfs_part_line} + local rootfs_part_size=${6} + if [ "${rootfs_part_size}" = "1G" ]; then + info "root partition is already resized" + return 0 + fi + # Check that the last partition is the rootfs partition local last_part_line=$(fdisk -l /dev/mmcblk0 2>/dev/null | tail -n 1) set ${last_part_line} local last_part_num=${1#/dev/mmcblk0p} local part_start=${3} if [ "${last_part_num}" != "${root_part_num}" ]; then - die 4 "rootfs is not the last partition. Don't know how to expand, aborting" + die_notif 4 "rootfs is not the last partition. Don't know how to expand, aborting" fi # Remove (temporarily) the rootfs partition @@ -55,39 +64,50 @@ w EOF # Mark the rootfs partition as bootable - sfdisk -A /dev/mmcblk0 ${root_part_num} >/dev/null 2>&1 || die 7 "cannot make the rootfs partition bootable, aborting" + sfdisk -A /dev/mmcblk0 ${root_part_num} >/dev/null 2>&1 || die_notif 7 "cannot make the rootfs partition bootable, aborting" - return 0 -} + # Reload the partition table + partprobe /dev/mmcblk0 >/dev/null 2>&1 || die_notif 8 "cannot reload the partition table, aborting" -reload_partition_table () { - partprobe /dev/mmcblk0 >/dev/null 2>&1 || die 9 "cannot reload the partition table, aborting" return 0 } resize_rootfs_filesystem () { + local rootfs_line=$(df | grep /dev/root) + set ${rootfs_line} + local rootfs_size=${2} + if [ ${rootfs_size} -gt 1000000 ]; then + info "rootfs already resized" + return 0 + fi rw - resize2fs ${root_part} >/dev/null 2>&1 || die 10 "cannot resize the root filesystem, aborting" + resize2fs ${root_part} >/dev/null 2>&1 || die_notif 9 "cannot resize the root filesystem, aborting" ro return 0 } create_swap () { - mount | grep -q ${share_part} - if [ $? -ne 0 ]; then - # Check that the last partition is the rootfs partition - local last_part_line=$(fdisk -l /dev/mmcblk0 2>/dev/null | tail -n 1) - set ${last_part_line} - local last_part_num=${1#/dev/mmcblk0p} - if [ "$last_part_num" != "$root_part_num" ]; then - die 11 "rootfs is not the last partition. Don't know how to create the backing store partition" - fi + # Check if swap partition already exists + fdisk -l /dev/mmcblk0 2>/dev/null | grep "Linux swap" >/dev/null 2>&1 + if [ $? -eq 0 ]; then + info "swap partition already exists" + else + mount | grep -q ${usb_part} + if [ $? -ne 0 ]; then - # Create an additional linux swap partition - let swap_part_num=${last_part_num}+1 - swap_part=/dev/mmcblk0p${swap_part_num} - fdisk /dev/mmcblk0 >/dev/null 2>&1 </dev/null | tail -n 1) + set ${last_part_line} + local last_part_num=${1#/dev/mmcblk0p} + if [ "$last_part_num" != "$root_part_num" ]; then + die_notif 10 "rootfs is not the last partition. Don't know how to create the backing store partition" + fi + + # Create an additional linux swap partition + let swap_part_num=${last_part_num}+1 + swap_part=/dev/mmcblk0p${swap_part_num} + fdisk /dev/mmcblk0 >/dev/null 2>&1 </dev/null 2>&1 if [ $? -ne 0 ]; then - die 14 "cannot create swap file, aborting" + die_notif 11 "cannot create swap file, aborting" fi + + # Enable swap + swapon -a >/dev/null 2>&1 || die_notif 12 "cannot enable swap file, aborting" fi return 0 } -enable_swap () { - swapon -a >/dev/null 2>&1 || die 15 "cannot enable swap file, aborting" - return 0 -} +create_usb_partition () { -create_backing_store_partition () { - mount | grep -q ${share_part} + # Check if the USB partition already exists + fdisk -l /dev/mmcblk0 2>/dev/null | grep "W95 FAT32" >/dev/null 2>&1 + if [ $? -eq 0 ]; then + info "USB partition already exists" + return 0 + fi + + mount | grep -q ${usb_part} if [ $? -ne 0 ]; then # Check that the last partition is the swap partition @@ -120,76 +154,86 @@ create_backing_store_partition () { set ${last_part_line} local last_part_num=${1#/dev/mmcblk0p} if [ "${last_part_num}" != "${swap_part_num}" ]; then - die 15 "rootfs is not the last partition. Don't know how to create the backing store partition" + die_notif 13 "rootfs is not the last partition. Don't know how to create the backing store partition" fi - # Create an additional FAT32 share partition that fills the disk - let share_part_num=${last_part_num}+1 - share_part=/dev/mmcblk0p${share_part_num} + # Create an additional FAT32 USB partition that fills the disk + let usb_part_num=${last_part_num}+1 + usb_part=/dev/mmcblk0p${usb_part_num} fdisk /dev/mmcblk0 >/dev/null 2>&1 </dev/null 2>&1 || die_notif 14 "cannot reload the partition table, aborting" + return 0 } -format_backing_store_partition () { +format_usb_partition () { + + # Check if the USB partition is already mounted + mount | grep /mnt > /dev/null 2>&1 + if [ $? -eq 0 ]; then + info "USB partition already mounted" + return 0 + fi # Format the backing store as FAT32 - mkfs.vfat ${share_part} >/dev/null 2>&1 || die 17 "cannot format the backing store partition" + mkfs.vfat ${usb_part} >/dev/null 2>&1 || die_notif 15 "cannot format the backing store partition" return 0 } -copy_files_to_store_partition () { - # Add file to force assembly tests - - - mount /mnt/ || die 18 "Cannot mount /mnt" +copy_files_to_usb_partition () { + + # Check if the USB partition is already mounted + mount | grep /mnt > /dev/null 2>&1 + if [ $? -ne 0 ]; then + mount /mnt/ || die_notif 16 "Cannot mount /mnt" + fi unzip -q -o /usr/local/share/mnt_freware_games.zip -d /mnt/ - umount /mnt/ || die 20 "Cannot unmount /mnt" + mkdir -p /mnt/Emulators + set +f + cp -f /usr/games/opk/*.opk /mnt/Emulators/ + set -f + umount /mnt/ || die_notif 17 "Cannot unmount /mnt" return 0 } check_root_id notif " FIRST BOOT DETECTED" -notif " 1/9 RESIZE ROOT PARTITION" +notif " 1/6 RESIZE ROOT PARTITION" resize_rootfs_partition -notif " 2/9 RELOAD ROOT PARTITION" -reload_partition_table - -notif " 3/9 RESIZE ROOT FILESYSTEM" +notif " 2/6 RESIZE ROOT FILESYSTEM" resize_rootfs_filesystem -notif " 4/9 CREATE SWAP" +notif " 3/6 CREATE SWAP" create_swap -notif " 5/9 ENABLE SWAP" -enable_swap +notif " 4/6 CREATE USB PARTITION" +create_usb_partition -notif " 6/9 CREATE USB PARTITION" -create_backing_store_partition +notif " 5/6 FORMAT USB PARTITION" +format_usb_partition -notif " 7/9 RELOAD PARTITION TABLE" -reload_partition_table - -notif " 8/9 FORMAT USB PARTITION" -format_backing_store_partition - -notif " 9/9 COPY FILES TO ^ USB PARTITION" -copy_files_to_store_partition +notif " 6/6 COPY FILES TO ^ USB PARTITION" +copy_files_to_usb_partition notif " FIRST BOOT SETUP FINISHED!" +fw_setenv first_boot_ok 1 + sleep 1 clear_notif diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/get_launcher b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/get_launcher new file mode 100755 index 0000000..0911e80 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/get_launcher @@ -0,0 +1,25 @@ +#!/bin/sh + +# Check args +if [ ${#} -ne 0 ]; then + echo "Usage: $(basename ${0})" + exit 1 +fi + +# Launcher File +LAUNCHER_FILE=$HOME/Launchers/launcher.txt +DEFAULT_LAUNCHER=retrofe + +# Check that file exists +[ -f "$LAUNCHER_FILE" ] || set_launcher $DEFAULT_LAUNCHER >/dev/null 2>&1 + +# Check Launcher +launcher=$(cat "$LAUNCHER_FILE" | head -1) + +# Check not empty +[ -z "$LAUNCHER_FILE" ] && launcher=$DEFAULT_LAUNCHER; set_launcher $launcher >/dev/null 2>&1 + +# Return launcher name +echo $launcher + +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/instant_play b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/instant_play new file mode 100755 index 0000000..42db9a0 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/instant_play @@ -0,0 +1,37 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +# Check args +if [ ${#} -eq 0 ]; then + echo "Usage: $(basename ${0}) args..." + exit 1 +fi + +INSTANT_PLAY_FILE="/mnt/instant_play" + +# Write quick load file args +echo -n "" > "${INSTANT_PLAY_FILE}" +for arg in "$@"; do + if $(echo "${arg}" | egrep -q '[[:space:]]'); then + + # Add quotes around arguments containing spaces + echo -n "\"${arg}\" " >> "${INSTANT_PLAY_FILE}" + else + echo -n "${arg} " >> "${INSTANT_PLAY_FILE}" + fi +done + +# Add the magic sauce to launch the process in background, record the +# PID into a file, wait for the process to terminate and erase the +# recorded PID +cat << EOF >> "${INSTANT_PLAY_FILE}" +& +record_pid \$! +wait \$! +erase_pid +EOF + +# Now terminate gracefully +exec shutdown_funkey diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/kill_emulators b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/kill_emulators deleted file mode 100755 index 7e5efd8..0000000 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/kill_emulators +++ /dev/null @@ -1,4 +0,0 @@ -#!/bin/sh - -killall mednafen gpsp psnes pcsx sdlgnuboy PicoDriveBin -exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/record_pid b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/record_pid new file mode 100755 index 0000000..0f98bfc --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/record_pid @@ -0,0 +1,12 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +# Check args +if [ ${#} -ne 1 ]; then + echo "Usage: $(basename ${0}) pid" + exit 1 +fi +echo ${1} > /var/run/funkey.pid +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/sched_shutdown b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/sched_shutdown index 00ff51f..3fa1732 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/sched_shutdown +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/sched_shutdown @@ -1,15 +1,22 @@ #!/bin/sh -if [ ${#} != 1 ]; then - echo "Usage: $0 seconds_before_shutdown" +# Uncomment the following line to get debug info +#set -x + +if [ ${#} != 1 -o "${1}" -eq 0 ]; then + echo "Usage: $(basename ${0}) grace_delay" exit 1 fi -nb_secs_to_wait=$1 +# Send USR1 signal to the running FunKey process to warn about +# impending shutdown +pkill -USR1 -F /var/run/funkey.pid > /dev/null 2>&1 -# Wait $nb_secs_to_wait seconds to catch signal USR2 -# If the signal is caught, then it means a process canceled this shutdown -sleep ${nb_secs_to_wait} +# Delay for the given grace period seconds to catch signal USR2. +# If the signal is caught, then it means the running FunKey process +# canceled this shutdown and will handle it by itself. +sleep ${1} -# Too late to cancel: init shutdown +# Delay expired, initiate shutdown shutdown_funkey + diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/set_launcher b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/set_launcher new file mode 100755 index 0000000..1d20bd9 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/set_launcher @@ -0,0 +1,18 @@ +#!/bin/sh + +# Check number of args +if [ ${#} -ne 1 ]; then + echo "Usage: $(basename ${0}) launcher" + exit 1 +fi + +# Launcher File +LAUNCHER_FILE=$HOME/Launchers/launcher.txt +mkdir -p "$(dirname "$LAUNCHER_FILE")" + +# Check Launcher +NEW_LAUNCHER=${1} +echo "Setting launcher: ${NEW_LAUNCHER}" +echo ${NEW_LAUNCHER} > "$LAUNCHER_FILE" + +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/share b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/share index 8e29850..f63d8a5 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/share +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/share @@ -48,7 +48,8 @@ mount_share () { fi # Create the directory structure if required - (cd /mnt; mkdir -p "Atari lynx" "Game Boy" "Game Boy Color" "Game Boy Advance" "Game Gear" "Neo Geo Pocket" "NES" "PS1" "PS1/bios" "Sega Genesis" "Sega Master System" "SNES" "WonderSwan" "PCE-TurboGrafx") + #(cd /mnt; mkdir -p "Applications" "Emulators" "Games" "Atari lynx" "Game Boy" "Game Boy Color" "Game Boy Advance" "Game Gear" "Neo Geo Pocket" "NES" "PS1" "PS1/bios" "Sega Genesis" "Sega Master System" "SNES" "WonderSwan" "PCE-TurboGrafx") + (cd /mnt; mkdir -p "Emulators" "Atari lynx" "Game Boy" "Game Boy Color" "Game Boy Advance" "Game Gear" "Neo Geo Pocket" "NES" "PS1" "PS1/bios" "Sega Genesis" "Sega Master System" "SNES" "WonderSwan" "PCE-TurboGrafx") # Check if there is a firmware update file if [ -f /mnt/FunKey-*.fwu ]; then diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/shutdown_funkey b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/shutdown_funkey index 9f1dc45..54ed4ca 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/shutdown_funkey +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/shutdown_funkey @@ -1,5 +1,9 @@ #!/bin/sh +# Notify system, reboot in progress +REBOOTING_FILE="/run/rebooting" +touch $REBOOTING_FILE + # Notif fullscreen "Shutting down" notif_set 0 "^^^^^^^^ SHUTTING DOWN...^^^^^^^^" @@ -9,5 +13,8 @@ start_audio_amp 0 >/dev/null 2>&1 # Force Read Only ro +# Unmount writeable partition to force +umount -r /mnt + # Poweroff poweroff diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/signal_usr1_to_emulators b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/signal_usr1_to_emulators deleted file mode 100755 index 0d2aa02..0000000 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/signal_usr1_to_emulators +++ /dev/null @@ -1,23 +0,0 @@ -#!/bin/sh -# This should replaced by storing the correct PID before -# launching an emulator and signaling only this one. - -RUN_ENV_VAR=$(fw_printenv -n assembly_tests 2>/dev/null) -if [ "x${RUN_ENV_VAR}" == "x1" ]; then - # First this one - killall -s USR1 assembly_tests > /dev/null 2>&1 - - # Then this one - killall -s USR1 funkey_prod_screens > /dev/null 2>&1 -fi - -# Send signal to all PCSX first (time critical) -killall -s USR1 pcsx > /dev/null 2>&1 - -# Send signal to all other emulators -killall -s USR1 gpsp psnes sdlgnuboy PicoDrive mednafen fceux > /dev/null 2>&1 - -# Send signal to launcher -killall -s USR1 retrofe > /dev/null 2>&1 - -exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/snap b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/snap new file mode 100755 index 0000000..7a246a5 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/snap @@ -0,0 +1,33 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +# Check args +if [ ${#} -ne 0 ]; then + echo "Usage: $(basename ${0})" + exit 1 +fi + +# Lock file (necessary since fbgrab must run in bg not to block the buttons while gaming) +LOCK_FILE="/var/lock/snap.lock" +if [ -f "${LOCK_FILE}" ]; then + echo "${LOCK_FILE} already exists" + exit 1 +fi +touch "${LOCK_FILE}" + +# Increment name and save snapshot +SNAPSHOT_EXT=PNG +SNAPSHOT_DIR=$HOME/snapshots +mkdir -p "${SNAPSHOT_DIR}" +last=$(cd ${SNAPSHOT_DIR}; ls IMG_*.${SNAPSHOT_EXT} 2> /dev/null | tail -1 | sed 's/^IMG_0*\([0-9]\+\)\.'${SNAPSHOT_EXT}'$/\1/') +let last=${last}+1 +snapshot_file=$(printf "IMG_%04d.${SNAPSHOT_EXT}" $last) +notif_set 2 " SCREEENSHOT ${snapshot_file}" +fbgrab "${SNAPSHOT_DIR}/${snapshot_file}" >/dev/null 2>&1 & + +# Remove lock file +rm -f "${LOCK_FILE}" + +exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_audio_amp b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_audio_amp index 79f7f18..50034e0 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_audio_amp +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_audio_amp @@ -6,10 +6,16 @@ if [ ${#} -ne 1 ]; then exit 1 fi -# Check Enable arg +# Check enable arg enable=${1} if [ ${enable} -eq 1 ]; then - echo "Turning audio amplifier ON" + + # Turn ON only if volume is not null + if [ "$(volume_get)" -ne "0" ]; then + echo "Turning audio amplifier ON" + else + exit 0 + fi elif [ ${enable} -eq 0 ]; then echo "Turning audio amplifier OFF" else diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_launcher b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_launcher index bcf2b43..1c10d96 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_launcher +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/start_launcher @@ -1,46 +1,75 @@ #!/bin/sh -LOCK_FILE=/var/lock/launcher.lock -PREVENT_LAUNCHER_FILE=/mnt/prevent_launcher -PREVENT_LAUNCHER_FILE2=/boot/prevent_launcher -QUICK_LOAD_FILE=/mnt/quick_load_cmd +# Uncomment the following line to get debug info +#set -x -if [ -f ${LOCK_FILE} ]; then +LOCK_FILE="/var/lock/launcher.lock" +INSTANT_PLAY_FILE="/mnt/instant_play" +PREVENT_LAUNCHER_FILE="/mnt/prevent_launcher" +REBOOTING_FILE="/run/rebooting" + + +if [ -f "${LOCK_FILE}" ]; then echo "${LOCK_FILE} already exists" exit 1 fi -touch ${LOCK_FILE} +touch "${LOCK_FILE}" -mkdir -p ${MEDNAFEN_HOME} -cp /usr/games/lynxboot.img /usr/games/mednafen-09x.cfg ${MEDNAFEN_HOME}/ +# Sanity cmd: in case these files do not exist +mkdir -p "${MEDNAFEN_HOME}" +cp "/usr/games/lynxboot.img" "/usr/games/mednafen-09x.cfg" "${MEDNAFEN_HOME}/" -# Launch Previous Game -if [ -f ${QUICK_LOAD_FILE} ]; then - command=$(cat ${QUICK_LOAD_FILE}) - echo "Found quick load file, restarting previous game with command:" - echo ${command} - rm ${QUICK_LOAD_FILE} - eval ${command} +# Launch Previous Game if any +if [ -f "${INSTANT_PLAY_FILE}" ]; then + echo "Found Instant Play file, restarting previous game with command: "$(head -n 1 "${INSTANT_PLAY_FILE}") + source "${INSTANT_PLAY_FILE}" + rm -f "${INSTANT_PLAY_FILE}" termfix_all fi -# Loop to launch launcher indefinitely +# Then loop to launch the launcher indefinitely while true; do # Check if prevent launcher file present - if [ -f ${PREVENT_LAUNCHER_FILE} ]; then - echo "Found file: ${PREVENT_LAUNCHER_FILE}, not launching launcher" - sleep 5 - elif [ -f ${PREVENT_LAUNCHER_FILE2} ]; then - echo "Found file: ${PREVENT_LAUNCHER_FILE2}, not launching launcher" - sleep 5 + if [ -f "${PREVENT_LAUNCHER_FILE}" ]; then + echo "${PREVENT_LAUNCHER_FILE} file found, not starting launcher" + sleep 5 else - # Launch Retrofe - retrofe + LAUNCHER=$(get_launcher) - # In case retrofe quits with errors, clear graphic VT - termfix_all + if [ ${LAUNCHER} == "gmenu2x" ]; then + + # Launch gmenu2x + gmenu2x& + elif [ ${LAUNCHER} == "retrofe" ]; then + + # Launch Retrofe + retrofe& + else + DEFAULT_LAUNCHER=retrofe + echo "Not recognized launcher: $LAUNCHER, setting $DEFAULT_LAUNCHER" + set_launcher $DEFAULT_LAUNCHER + fi + + # Record the PID into a file, wait for the + # process to terminate and erase the recorded PID + record_pid $! + wait $! + erase_pid + + # In case retrofe quits with errors, clear graphic VT + termfix_all + fi + + # WD to prevent 100% CPU + sleep 0.5 + + # Exit if console rebooting + if [ -f $REBOOTING_FILE ]; then + break fi done -rm ${LOCK_FILE} + +# Remove lock file and exit +rm "${LOCK_FILE}" exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_info b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_info new file mode 100755 index 0000000..cd1fca7 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_info @@ -0,0 +1,18 @@ +#!/bin/sh + +. /etc/os-release +ROOTFS_VERSION=`echo $VERSION | sed -e 's/.*-g/g/'` +ROOTFS_DATE=`date -r /etc/os-release '+%a %b %d %Y' ` + +KERNEL_VERSION=`uname -r` +KERNEL_DATE=`uname -v |cut -d ' ' -f 3-5,8` +PROCESSOR=`cat /proc/cpuinfo |head -1 |sed 's/^.*: //' ` +RAM=`cat /proc/meminfo | head -1 |sed 's/^MemTotal: \+//' ` +SWAP=`awk 'BEGIN { getline } { print "\t"$1": "$3" kB" }' /proc/swaps 2>/dev/null` +[ "$SWAP" ] && SWAP="Swap: $SWAP\n" + +echo -e "Kernel version: $KERNEL_VERSION\nCompiled: $KERNEL_DATE\n +Root FS version: $ROOTFS_VERSION\nCompiled: $ROOTFS_DATE\n +Processor: $PROCESSOR\nUsable RAM: $RAM\n$SWAP +Network interfaces:" +ip -o -4 address list | awk '{if ($2 != "lo") { printf "%7s %s\n", $2, gensub("/.*$", "", "g", $4) ; outlines++ } } END { if (outlines == 0) print " (none)" }' diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_stats b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_stats index a6a091a..4e6ad97 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_stats +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/system_stats @@ -20,16 +20,21 @@ while true; do if [ ${perform} -eq 1 ]; then # Compute stats - cpu=$(printf "%.1f\n" $(mpstat -P ALL $UPDATE_PERIOD 1 | tail -1 | awk '{print 100-$12}')) - ram_mem=$(printf "%.1f\n" $(free | grep Mem | awk '{print $3/$2 * 100.0}')) - ram_swap=$(printf "%.1f\n" $(free | grep Swap | awk '{print $3/$2 * 100.0}')) + cpu=$(printf "%.0f\n" $(mpstat -P ALL $UPDATE_PERIOD 1 | tail -1 | awk '{print 100-$12}')) + ram_mem=$(printf "%.0f\n" $(free | grep Mem | awk '{print $3/$2 * 100.0}')) + ram_swap=$(printf "%.0f\n" $(free | grep Swap | awk '{print $3/$2 * 100.0}')) + ip_addr=$(ifconfig usb0 | grep "inet " | awk -F'[: ]+' '{ print $4 }') # Notif if [ ${notif_dirty} -eq 1 ]; then notif_clear notif_dirty=0 else - notif_set 0 "CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%" + if [ "x${ip_addr}" != "x" ]; then + notif_set 0 " CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%^IP:${ip_addr}" + else + notif_set 0 " CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%" + fi fi else sleep ${UPDATE_PERIOD} diff --git a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/volume_set b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/volume_set index bebc413..41c9619 100755 --- a/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/volume_set +++ b/FunKey/board/funkey/rootfs-overlay/usr/local/sbin/volume_set @@ -21,13 +21,6 @@ volume_scaled=$(echo "a = $volume_percent * (63 - $vol_mini) / 100 + $vol_mini + # Get current value current_volume=$(volume_get) -# Turn on/off audio amplifier if necessary -if [ ${current_volume} -eq 0 -a ${volume_percent} -ne 0 ]; then - start_audio_amp 1 -elif [ ${current_volume} -ne 0 -a ${volume_percent} -eq 0 ]; then - start_audio_amp 0 -fi - # Set new volume amixer -q sset 'Headphone' ${volume_scaled} unmute @@ -35,4 +28,12 @@ amixer -q sset 'Headphone' ${volume_scaled} unmute if [ ${?} -eq 0 -a ${current_volume} -ne ${volume_percent} ]; then fw_setenv volume ${volume_percent} fi + +# Turn on/off audio amplifier if necessary +if [ ${current_volume} -eq 0 -a ${volume_percent} -ne 0 ]; then + start_audio_amp 1 +elif [ ${current_volume} -ne 0 -a ${volume_percent} -eq 0 ]; then + start_audio_amp 0 +fi + exit 0 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/fonts/truetype b/FunKey/board/funkey/rootfs-overlay/usr/share/fonts/truetype new file mode 120000 index 0000000..945c9b4 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/fonts/truetype @@ -0,0 +1 @@ +. \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/about.txt b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/about.txt new file mode 100644 index 0000000..7f8db0e --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/about.txt @@ -0,0 +1,20 @@ +GMenu2X has been created by Massimiliano "Ryo" Torromeo, and is released under the GPL-v2 license. + +This version is maintained by the Qi-Hardware and OpenDingux crew. The credits, as well as the latest source code can be found here: +http://projects.qi-hardware.com/index.php/p/gmenu2x + +Credits for the translations: +---- +English & Italian by Massimiliano Torromeo +French by Paul Cercueil +Danish by claus +Dutch by superfly +Spanish by pedator +Portuguese (Portugal) by NightShadow +Slovak by Jozef +Swedish by Esslan and Micket +German by fusion_power, johnnysnet and Waldteufel +Finnish by Jontte and Atte +Norwegian by Lithium Flower +Russian by XaMMaX90 +Polish by Artur Rojek diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/gmenu2x.conf b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/gmenu2x.conf new file mode 100644 index 0000000..3d6e76c --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/gmenu2x.conf @@ -0,0 +1,7 @@ +videoBpp=16 +backlightTimeout=0 +opkPlatforms="funkey-s" +brightnessSysfs="/sys/class/backlight/backlight" +powerSupplySysfs="/sys/class/power_supply/axp20x-usb" +batterySysfs="/sys/class/power_supply/axp20x-battery" +section=1 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/input.conf b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/input.conf new file mode 100644 index 0000000..4341eb1 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/input.conf @@ -0,0 +1,11 @@ +accept=keyboard,97 +cancel=keyboard,98 +altleft=keyboard,109 +altright=keyboard,110 +menu=keyboard,107 +settings=keyboard,115 +up=keyboard,117 +down=keyboard,100 +left=keyboard,108 +right=keyboard,114 +home=keyboard,113 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Blue.png b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Blue.png new file mode 100644 index 0000000..8ce6fed Binary files /dev/null and b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Blue.png differ diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Orange.png b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Orange.png new file mode 100644 index 0000000..9cb8459 Binary files /dev/null and b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Orange.png differ diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Red.png b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Red.png new file mode 100644 index 0000000..941eb86 Binary files /dev/null and b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Abstract Red.png differ diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Cubes.png b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Cubes.png new file mode 100644 index 0000000..63ab98d Binary files /dev/null and b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/Cubes.png differ diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/planet.png b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/planet.png new file mode 100644 index 0000000..643e6ad Binary files /dev/null and b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/skins/240x240/Default/wallpapers/planet.png differ diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Basque b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Basque new file mode 100644 index 0000000..1895e61 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Basque @@ -0,0 +1,129 @@ +Settings=Aukerak +Configure GMenu2X's options=GMenu2X aukerak konfiguratu +Activate Usb on SD=Sd-aren usb-a aktibatu +Activate Usb on Nand=Nand-aren usb-a aktibatu +Info about GMenu2X=GMenu2X-ri buruzkoak erakutsi +About=Honi buruz... +Add section=Sekzioa gehitu +Rename section=Sekzioaren izena aldatu +Delete section=Sekzioa ezabatu +Scan for applications and games=Jokuak eta programak bilatu +applications=Programak +Edit link=Esteka aldatu +Title=Izenburua +Link title=Estekaren izenburua +Description=Azalpena +Link description=Azalpenaren esteka +Section=sekzioa +The section this link belongs to=Esteka honen sekzioa da +Icon=Ikonoa +Select an icon for the link: $1=Estekarentzako ikonoa aukeratu: $1 +Manual=Eskuliburua +Select a graphic/textual manual or a readme=Aukeratu eskuliburua(testua edo grafikoa) +Cpu clock frequency to set when launching this link=Programa hau abiarazteko PUZ abiadura ezarri +Volume to set for this link=Esteka honentzako bolumena ezarri +Parameters=Parametroak +Parameters to pass to the application=Programara pasatuko diren parametroak +Selector Directory=Selektorearen karpeta +Directory to scan for the selector=Selektorearekin eskaneatuko den direktorioa +Selector Browser=Selektorearen arakatzailea +Allow the selector to change directory=Selektoreari direktorioz aldatzea baimendu +Selector Filter=Selektorearen iragazkia +Filter for the selector (Separate values with a comma)=Selektorearentzako iragazkiak(Balioak komekin banandu) +Selector Screenshots=Selektorearen pantaila-argazkia +Directory of the screenshots for the selector=Selektorearen pantaila-argazkien direktorioa +Selector Aliases=Selektorearen aliasak +File containing a list of aliases for the selector=Aliasen zerrenda duen artxiboaren izena +Don't Leave=Ez irten +Don't quit GMenu2X when launching this link=Lotura hau abiaraztean ez amaitu Gmenu2x +Save last selection=Azken aukera gogoratu +Save the last selected link and section on exit=Azken aukera eta esteka gorde irtetean +Clock for GMenu2X=GMenu2X-ren erlojua +Set the cpu working frequency when running GMenu2X=Gmenu2x-rentzako PUZ abiadura ezarri +Maximum overclock=Overclock muga +Set the maximum overclock for launching links=Ezarri daitekeen overclockik handiena +Global Volume=Bolumen orokorra +Set the default volume for the gp2x soundcard=Aurrezarritako bolumen maila +Output logs=log artxiboak +Logs the output of the links. Use the Log Viewer to read them.=esteken log-ak gorde. Log irakurlea erabili irakurtzeko. +Number of columns=Zutabe zenbakia +Set the number of columns of links to display on a page=Orri bakoitzeko erakutsiko diren zutabeak +Number of rows=Ilara zenbakia +Set the number of rows of links to display on a page=Orri bakoitzeko erakutsiko diren ilarak +Top Bar Color=Goiko barraren kolorea +Color of the top bar=Goian dagoen barraren kolorea +Bottom Bar Color=Beheko barraren kolorea +Color of the bottom bar=Behean dagoen barraren kolorea +Selection Color=aukeratutakoaren kolorea +Color of the selection and other interface details=Aukeratutako testuaren kolorea eta interfacearen beste aukera batzuk +You should disable Usb Networking to do this.=Usb sarea desaktibatu beharko zenuke. +Operation not permitted.=Baimendu gabeko operazioa. +Language=Hizkuntza +Set the language used by GMenu2X=Aukeratu Gmenu2x-ren hizkuntza +Increase=Handitu +Decrease=txikitu +Change color component=Kolore konponentea aldatu +Increase value=Balioa handitu +Decrease value=Balioa txikitu +Switch=Aldatu +Change value=Balioa aldatu +Edit=Editatu +Clear=Garbitu +Select a directory=Direktorioa aukeratu +Select a file=Artxiboa aukeratu ezazu +Clock (default: 200)=Maiztasuna (Aurrezarritakoa: 200) +Volume (default: -1)=bolumena (Aurrezarritakoa: -1) +Enter folder=Karpetan sartu +Confirm=Berretsi +Enter folder/Confirm=karpetan sartu/Berretsi +Up one folder=Karpeta bat gora +Select an application=Programa aukeratu +Space=espazio-barra +Shift=Shift +Cancel=Cancelar +OK=Ok +Backspace=Backspace +Skin=Maskara +Set the skin used by GMenu2X=GMenu2X-ren maskara aukeratu +Add link in $1=esteka sortu... $1 +Edit $1=Aldatu $1 +Delete $1 link=honen esteka ezabatu $1 +Deleting $1=Ezabatzen $1 +Are you sure?=Ziur zaude? +Insert a name for the new section=Sekzio berrirako izena sartu +Insert a new name for this section=Sekzio honetarako izen berria sartu +Yes=Bai +No=Ez +You will lose all the links in this section.=Sekzio honetako esteka guztiak galduko dira +Exit=Irten +Link Scanner=Esteka bilatzailea +Scanning SD filesystem...=SD txartela arakatzen... +Scanning NAND filesystem...=NAND memoria arakatzen... +$1 files found.=$1 aurkitutako artxiboak(s). +Creating links...=Estekak sortzen... +$1 links created.=$1 esteka sorturik(s). +Version $1 (Build date: $2)=Bertsioa $1 (data: $2) +Log Viewer=Log irakurlea +Displays last launched program's output=Abiarazitako azken programa erakutsi +Do you want to delete the log file?=Log-ak ezabatu nahi dituzu? +USB Enabled (SD)=USB-a piztuta (SD) +USB Enabled (Nand)=USB-a piztuta (Nand) +Turn off=Desaktibatu +Launching $1=Abiarazten $1 +Change page=Orria aldatu +Page=Orria +Scroll=scroll +Untitled=Izenburu gabekoa +Change GMenu2X wallpaper=GMenu2X-ren atzeko irudia aldatu +Activate/deactivate tv-out=Tv-out piztu/itzali +Select wallpaper=aukeratu atzekaldeko irudia +Gamma=Gama +Set gp2x gamma value (default: 10)=Gp2x-ren gama balioa ezarri (aurrezarritakoa: 10) +Tv-Out encoding=tv-out-ren kodeketa +Encoding of the tv-out signal=TV-out-ren seinalearen kodeketa +Tweak RAM Timings=Ram denborak hobetu +This usually speeds up the application at the cost of stability=programa azkarrago doa estabilitatearen truke. +Gamma (default: 0)=Gama (aurrezarritakoa: 0) +Gamma value to set when launching this link=Esteka hau abiarazteko erabiltzen den gama balioa +Wrapper=Itzuli +Explicitly relaunch GMenu2X after this link's execution ends=Amaitzerakoan gmenu2x birkargatu diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Catalan b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Catalan new file mode 100644 index 0000000..87d25ab --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Catalan @@ -0,0 +1,137 @@ +Settings=Preferències +Configure GMenu2X's options=Configura les opcions del GMenu2X +Activate Usb on SD=Activa USB per la SD +Activate Usb on Nand=Activa USB per la Nand +Info about GMenu2X=Informació del GMenu2X +About=Informació +Add section=Afegir secció +Rename section=Re anomenar secció +Delete section=Eliminar secció +Scan for applications and games=Buscar aplicacions i jocs +applications=Aplicacions +Edit link=Editar enllaç +Title=Títol +Link title=Títol de l'enllaç +Description=Descripció +Link description=Descripció de l'enllaç +Section=Secció +The section this link belongs to=Secció a la que pertany l'enllaç +Icon=Icona +Select an icon for the link: $1=Selecciona una icona per l'enllaç: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Selecciona un manual gràfic/text o un "readme" +Cpu clock frequency to set when launching this link=Ajust del rellotge de la cpu per aquest enllaç +Volume to set for this link=Ajust del volum de l'enllaç +Parameters=Paràmetres +Parameters to pass to the application=Paràmetres que s'envien a l'aplicació +Selector Directory=Directori del Selector +Directory to scan for the selector=Directori a explorar amb el selector +Selector Browser=Explorador del selector +Allow the selector to change directory=Permetre al selector canviar de directori +Selector Filter=Filtre del selector +Filter for the selector (Separate values with a comma)=Filtre per el selector (Separar valors amb comes) +Selector Screenshots=Captures de pantalla del selector +Directory of the screenshots for the selector=Directori de captures de pantalla per el selector +Selector Aliases=Alias del selector +File containing a list of aliases for the selector=Fitxer que conté la llista d'alias per el selector +Explicitly relaunch GMenu2X after this link's execution ends=Força recarregar el GMenu2X a l'acabar l'execució de l'enllaç +Don't Leave=No sortir +Don't quit GMenu2X when launching this link=No tancar GMenu2X al carregar aquest enllaç +Save last selection=Recordar l'última selecció +Save the last selected link and section on exit=Recordar l'última secció i enllaç seleccionat al sortir +Clock for GMenu2X=Rellotge per al GMenu2X +Set the cpu working frequency when running GMenu2X=Ajustar la freqüència de treball de la cpu a l'executar GMenu2X +Maximum overclock=Overclock màxim +Set the maximum overclock for launching links=Ajustar al màxim overclock per a carregar enllaços +Global Volume=Volum global +Set the default volume for the gp2x soundcard=Ajusta el volum per defecte del so a la gp2x +Output logs=Fitxers de Log +Logs the output of the links. Use the Log Viewer to read them.=Enregistra els Logs dels enllaços. Usa el lector de registres per llegir-los. +Number of columns=Número de columnes +Set the number of columns of links to display on a page=Ajusta el número de columnes d'enllaços a mostrar per pàgina +Number of rows=Número de línies +Set the number of rows of links to display on a page=Ajusta el número de línies d'enllaços a mostrar per pàgina +Top Bar Color=Color de barra superior +Color of the top bar=Color de la barra superior +Bottom Bar Color=Color de barra inferior +Color of the bottom bar=Color de la barra inferior +Selection Color=Color selecció +Color of the selection and other interface details=Color de la selecció i altres detalls de la interfície +You should disable Usb Networking to do this.=Ha de desactivar la Xarxa per USB per fer això. +Operation not permitted.=Operació no permesa. +Language=Idioma +Set the language used by GMenu2X=Ajusta l'idioma utilitzat al GMenu2X +Increase=Augmentar +Decrease=Reduïr +Change color component=Canviar component cromàtic +Increase value=Incrementar valor +Decrease value=Reduir valor +Switch=Canviar +Change value=Canviar valor +Edit=Modificar +Clear=Netejar +Select a directory=Selecciona un directori +Select a file=Selecciona un fitxer +Clock (default: 200)=Freqüència (predeterminada: 200) +Volume (default: -1)=Volum (predeterminat: -1) +Enter folder=Entrar a la carpeta +Wrapper=Retornar +Confirm=Confirmar +Enter folder/Confirm=Entrar a la carpeta/Confirmar +Up one folder=Pujar una carpeta +Select an application=Selecciona un programa +Space=Espai +Shift=Majúscules +Cancel=Cancel·lar +OK=Acceptar +Backspace=Retrocés +Skin=Tema +Set the skin used by GMenu2X=Selecciona el tema a utilitzar al GMenu2X +Add link in $1=Afegir enllaç a $1 +Edit $1=Modificar $1 +Delete $1 link=Eliminar l'enllaç de $1 +Deleting $1=Eliminant $1 +Are you sure?=Estàs segur? +Insert a name for the new section=Introduir nom per a la nova secció +Insert a new name for this section=Introduir nou nom per a la secció +Yes=Si +No=No +You will lose all the links in this section.=Es perdran tots els enllaços d'aquesta secció. +Exit=Sortir +Link Scanner=Buscador d'enllaços +Scanning SD filesystem...=Explorant el sistema de fitxers de la SD... +Scanning NAND filesystem...=Explorant el sistema de fitxers de la NAND... +$1 files found.=$1 Fitxer(s) trobat(s). +Creating links...=Creant enllaços... +$1 links created.=$1 enllaç(os) creat(s). +Version $1 (Build date: $2)=Versió $1 (Compilació: $2) +Log Viewer=Visor de Logs +Displays last launched program's output=Mostra la sortida de l'últim programa executat +Do you want to delete the log file?=¿Desitja eliminar el fitxer de registre de successos? +USB Enabled (SD)=USB Activat (SD) +USB Enabled (Nand)=USB Activat (Nand) +Turn off=Desactivar +Launching $1=Executant $1 +Change page=Canviar pàgina +Page=Pàgina +Scroll=Desplaçament +Untitled=Sense títol +Change GMenu2X wallpaper=Canvia el fons del GMenu2X +Activate/deactivate tv-out=Activa/desactiva la sortida de TV +Select wallpaper=Selecciona la imatge de fons +Gamma=Gamma +Set gp2x gamma value (default: 10)=Ajustar el valor gama de la gp2x (predeterminat: 10) +Tv-Out encoding=Codificació de sortida de TV +Encoding of the tv-out signal=Codificació de la senyal de sortida de TV +Tweak RAM Timings=Modifica la sincronització de RAM +This usually speeds up the application at the cost of stability=Normalment accelera l'aplicació a costa de l'estabilitat +Gamma (default: 0)=Gamma (predeterminat: 0) +Gamma value to set when launching this link=Valor de gamma que utilitzarà a l'executar aquest enllaç +Wallpaper=Fons d'escriptori +Configure skin=Configura el Tema +Message Box Color=Color de caixa de text +Message Box Border Color=Color de la vora de la caixa de text +Message Box Selection Color=Color de la selecció de la caixa de text +Background color of the message box=Color de fons de la caixa de text +Border color of the message box=Color de la vora de la caixa de text +Color of the selection of the message box=Color de la selecció de la caixa de text diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (CN) b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (CN) new file mode 100644 index 0000000..33b8924 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (CN) @@ -0,0 +1,199 @@ +Add section=新建分组 +Rename section=重命名分组 +Delete section=删除分组 +Scan for applications and games=搜索程序和游戏 +applications=程序 +Edit link=编辑程序链接 +Title=程序标题 +Link title=设置程序名称 +Description=程序概述 +Link description=设置程序的简要概述 +Section=程序分组 +The section this link belongs to=选择程序所属的分组 +Icon=程序图标 +Select an icon for the link=选择程序的图标 +Manual=程序手册 +Select a manual or README file=选择程序使用手册或说明文档 +CPU Clock=CPU运行频率 +CPU clock frequency when launching this link=运行程序时CPU的运行频率(Mhz) +Volume to set for this link=为程序设置要使用的卷 +Parameters=运行参数 +Parameters to pass to the application=设置程序的运行参数 +Selector Directory=选择程序目录 +Directory to scan for the selector=选择程序默认加载目录(如:游戏rom存放目录) +Selector Browser=程序浏览器 +Allow the selector to change directory=允许程序选择器改变目录 +Selector Filter=选择程序筛选器 +Filter file type (separate with commas)=选择程序筛选(用逗号分割) +Selector Screenshots=选择截图目录 +Directory of the screenshots for the selector=选择程序截图目录 +Selector Aliases=程序选择器别名 +File containing a list of aliases for the selector=包含选择器别名列表的文件 +Backdrop=背景壁纸 +Select an image backdrop=选择程序的背景壁纸 +Wrapper=重新载入系统 +Relaunch GMenu2X after this link's execution ends=结束当前程序时重新加载系统 +Don't Leave=不退出程序 +Don't quit GMenu2X when launching this link=运行程序时不注销系统 +Save last selection=记录最后选择的分组 +Save the last selected link and section on exit=记录最后运行的程序和分组 +Clock for GMenu2X=运行频率 +Set the cpu working frequency when running GMenu2X=设置CPU运行频率 +Maximum overclock=最大超频 +Set the maximum overclock for launching links=设置此软件链接最大超频时钟频率 +Global Volume=全局音量 +Set the default volume for the gp2x soundcard=选择默认输出音量 +Output logs=输出日志 +Logs the output of the links. Use the Log Viewer to read them.=记录并输出日志 +Top/Section Bar=标题栏颜色 +Color of the top and section bar=设置标题栏的颜色 +List Body=列表项颜色 +Color of the list body=设置列表项目的颜色 +Bottom Bar=底部栏颜色 +Color of the bottom bar=设置底部栏颜色 +Selection=选中项颜色 +Color of the selection and other interface details=当前被选中项目的颜色 +Message Box=对话框颜色 +Background color of the message box=设置对话框的背景颜色 +Msg Box Border=对话框边框颜色 +Border color of the message box=设置对话框边框的颜色 +Msg Box Selection=对话框选中项颜色 +Color of the selection of the message box=设置对话框中当前被选中项目的颜色 +Font=文字颜色 +Color of the font=设置系统文字的颜色 +Font Outline=文字轮廓 +Color of the font's outline=设置文字轮廓的颜色 +Alt Font=备选文字 +Color of the alternative font=设置备选文字的颜色 +Alt Font Outline=备选文字轮廓 +Color of the alternative font outline=设置备选文字轮廓的颜色 +Font Size=文字大小 +Size of text font=设置文本文字的尺寸 +Title font size=标题文字大小 +Size of title's text font=设置标题文本文字的尺寸 +Top bar height=标题栏高度 +Height of top bar=设置标题栏的高度值 +Bottom bar height=底部栏高度 +Height of bottom bar=设置底部栏的高度值 +Link Height=程序项高度 +Height of link item=设置程序项的高度值 +Section bar size=分组栏大小 +Size of section bar=设置分组栏的尺寸 +Language=系统语言 +Set the language used by GMenu2X=设置系统界面语言 +Component=项目 +Increase=增大 +Decrease=减小 +Change color component=改变颜色组合 +Increase value=增大 +Decrease value=减小 +Switch=切换 +Change value=改变值 +Edit=编辑 +Clear=清除 +Select a directory=选择一个目录 +Select a file=选择一个文件 +Enter folder=输入文件夹 +Confirm=确认 +Enter folder/Confirm=输入文件夹/确认 +Folder up=上层文件夹 +Select=选择 +Back=返回 +Space=空格 +Keys=切换键盘 +Press=选择 +Cancel=取消 +Backspace=删除 +Skin=界面主题 +Set the skin used by GMenu2X=为系统选择一个界面主题 +Add link in $1=添加链接 $1 +Edit $1=编辑 $1 +Delete $1=删除 $1 +Delete $1 link=删除链接 $1 +Deleting $1=正在删除 $1 +Are you sure?=是否确定? +Insert a name for the new section=为新分组输入名称 +Insert a new name for this section=为此分组输入新名称 +No=否 +All links in this section will be removed.=您将删除此分组所有程序链接 +Exit=退出 +Link Scanner=程序搜索器 +Scanning SD filesystem...=搜索TF卡文件系统... +Scanning NAND filesystem...=搜索内部存储文件系统... +$1 files found.=$1 个文件被找到 +Creating links...=正在创建新链接... +$1 links created.=$1 链接已创建 +Charging=正在充电 +Log Viewer=日志阅读器 +Displays last launched program's output=显示最后运行的程序输出日志 +Do you want to delete the log file?=您是否想删除此日志文件? +Turn off=关闭 +Launching $1=正在运行 $1 +Change page=换页 +Page=页 +Scroll=滚动 +Untitled=无标题 +Change GMenu2X wallpaper=更换系统壁纸 +Select an image to use as a wallpaper=选择一张图片用作壁纸 +Explorer=文件管理器 +Launch an application= +File Dialog=文件对话框 +Wallpaper=系统壁纸 +Set background image= +Configure skin= +Umount=卸载外置TF卡 +Umount external SD= +About=关于 +Poweroff=关机 +Default=默认值 +Wallpaper selection=壁纸选择窗口 +Select an image from the list, to use as a wallpaper=从下列选择一个图片作为壁纸 +Do you want to umount external sdcard ?=确认卸载扩展TF卡吗? +Poweroff or reboot the device?=关机或重启设备吗？ +Reboot=重启 +Rebooting=设备重启中... +Settings=系统设置 +Configure settings= +Image Browser=图像浏览器 +Info about system= +Loading=系统加载中··· +Suspend=闲置超时关闭屏幕 +Power=电源选项 +Power menu= +Umount external SD card?=卸载外置TF卡吗？ +Yes=是 +Complete!=已完成! +OK=确定 +Umount Test=卸载测试 +Select an application=选择想要运行的程序 +Date & Time=时间设置 +Set system's date time=设置系统时间 +Section bar postition=标题栏位置 +Set the position of the Section Bar=设置主界面标题栏位置 +Battery profile=电池类型 +Set the battery discharge profile=设置电池类型 +Skin backdrops=程序壁纸 +Automatic load backdrops from skin pack=自动从默认主题包加载程序背景壁纸 +Screen timeout=闲置锁屏时间 +Seconds to turn display off if inactive=设置闲置超时关闭屏幕时间(秒) +Power timeout=闲置关机时间 +Minutes to poweroff system if inactive=设置闲置超时关机时间(分钟) +Backlight=背光亮度 +Set LCD backlight=设置屏幕背光亮度 +Global volume=全局音量 +Set the default volume for the soundcard=设置系统全局默认音量 +TV-out=电视输出 +TV-out signal encoding=设置电视输出信号类型 +TV-out connected.=电视视频线已连接 +Continue?=确定继续吗? +Delete=删除 +Add link=新增程序链接 +Link scanner=搜索程序链接 +Select USB mode:=选择USB连接模式： +Charger=仅充电 +USB Drive=USB驱动器 +USB Drive Connected=USB驱动器已连接 +Battery Logger=电池记录器 +Log battery power to battery.csv= +Del battery.csv=删除电池电量记录文件 +File Manager=文件管理器 \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (TW) b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (TW) new file mode 100644 index 0000000..3ef6c15 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Chinese (TW) @@ -0,0 +1,238 @@ +Settings=設定 +Configure GMenu2X's options=調整GMenu2X的設定 +Activate Usb on SD=啟動USB模式 (掛載SD卡) +Activate Usb on Nand=啟動USB模式 (掛載NAND) +Info about GMenu2X=有關GMenu2X的資訊 +About=關於 +Add section=加入區段 +Rename section=重新命名區段 +Delete section=刪除區段 +Scan for applications and games=掃描遊戲和應用程式 +applications=應用程式 +Edit link=編輯連結 +Title=標題 +Link title=連結標題 +Description=說明 +Link description=連結說明 +Section=區段 +The section this link belongs to=這個連結的區段屬於 +Icon=圖示 +Select an icon for the link: $1=選擇連結的圖示: $1 +Manual=說明 +Select a graphic/textual manual or a readme=請選擇圖片(或文字)的說明書 +Cpu clock frequency to set when launching this link=啟動這個連結時的CPU頻率 +Volume to set for this link=啟動這個連結時的音量 +Parameters=參數 +Parameters to pass to the application=要傳遞到這個應用程式中的參數 +Selector Directory=選擇器資料夾 +Directory to scan for the selector=選擇區段中欲掃描的資料夾 +Selector Browser=選擇器瀏覽器 +Allow the selector to change directory=允許選擇器改變資料夾 +Selector Filter=選擇器過濾器 +Filter for the selector (Separate values with a comma)=選擇器的過濾器(以逗號分隔) +Selector Screenshots=選擇器截圖 +Directory of the screenshots for the selector=選擇器用截圖資料夾 +Selector Aliases=選擇器別名 +File containing a list of aliases for the selector=File containing a list of aliases for the selector +Explicitly relaunch GMenu2X after this link's execution ends=Explicitly relaunch GMenu2X after this link's execution ends +Don't Leave=別離開 +Don't quit GMenu2X when launching this link=當啟動這個連結時請勿離開GMenu2X +Save last selection=記住最後的選擇 +Save the last selected link and section on exit=Save the last selected link and section on exit +Clock for GMenu2X=GMenu2X的時脈 +Set the cpu working frequency when running GMenu2X=設定GMenu2X執行時的CPU時脈 +Maximum overclock=最大超頻 +Set the maximum overclock for launching links=設定啟動連結時的最大超頻量 +Global Volume=全域音量 +Set the default volume for the gp2x soundcard=設定GP2X的預設音量 +Output logs=輸出logs +Logs the output of the links. Use the Log Viewer to read them.=Logs the output of the links. Use the Log Viewer to read them. +Number of columns=列數 +Set the number of columns of links to display on a page=設定一頁中的連結列數 +Number of rows=行數 +Set the number of rows of links to display on a page=設定一頁中的連結行數 +Top Bar Color=上面Bar的顏色 +Color of the top bar=上面Bar的顏色 +Bottom Bar Color=下面Bar的顏色 +Color of the bottom bar=下面Bar的顏色 +Selection Color=選項顏色 +Color of the selection and other interface details=選項和其他介面細節的顏色 +You should disable Usb Networking to do this.=在執行此操作前應先關閉USB網路 +Operation not permitted.=不允許的操作 +Language=語言 +Set the language used by GMenu2X=設定GMenu2X使用的語言 +Increase=增加 +Decrease=減少 +Change color component=調整顏色分量 +Increase value=增加數值 +Decrease value=減少數值 +Switch=切換 +Change value=切換數值 +Edit=編輯 +Clear=清除 +Select a directory=選擇資料夾 +Select a file=選擇檔案 +Clock (default: 200)=時脈 (預設: 200) +Volume (default: -1)=音量 (預設: -1) +Wrapper=Wrapper +Enter folder=進入資料夾 +Confirm=確認 +Enter folder/Confirm=進入資料夾/確認 +Up one folder=返回上層 +Select an application=選擇應用程式 +Space=空白 +Shift=Shift +Cancel=取消 +OK=OK +Backspace=Backspace +Skin=顏色 +Set the skin used by GMenu2X=設定Gmenu2X的顏色 +Add link in $1=於 $1 中加入連結 +Edit $1=編輯 $1 +Delete $1 link=刪除 $1 連結 +Deleting $1=正在刪除 $1 +Are you sure?=你確定嗎? +Insert a name for the new section=插入這個新區段中的名稱 +Insert a new name for this section=插入這個區段中的新名稱 +Yes=是 +No=否 +You will lose all the links in this section.=你將會遺失所有此區段中的連結. +Exit=離開 +Link Scanner=連結掃描器 +Scanning SD filesystem...=掃描SD中... +Scanning NAND filesystem...=掃描NAND中... +$1 files found.=找到$1 個檔案. +Creating links...=創建連結中... +$1 links created.=創建$1 個連結 +Version $1 (Build date: $2)=版本 $1 (建置日期: $2) +Log Viewer=Log檢視器 +Displays last launched program's output=顯示最後啟動程式的輸出 +Do you want to delete the log file?=是否刪除Log檔? +USB Enabled (SD)=開啟USB (SD) +USB Enabled (Nand)=開啟USB (Nand) +Turn off=關閉 +Launching $1=啟動 $1 +Change page=換頁 +Page=頁 +Scroll=捲動 +Untitled=未命名 +Change GMenu2X wallpaper=更換GMenu2X桌布 +Activate/deactivate tv-out=切換TV輸出 +Select wallpaper=選擇桌布 +Gamma=Gamma +Set gp2x gamma value (default: 10)=設定GP2X的Gamma value (預設: 10) +Tv-Out encoding=電視輸出格式 +Encoding of the tv-out signal=電視輸出訊號的格式 +Tweak RAM Timings=微調RAM timings +This usually speeds up the application at the cost of stability=這通常會以穩定性為代價換取更高效能 +Gamma (default: 0)=Gamma (預設: 0) +Gamma value to set when launching this link=啟動這個連結時的Gamma值 +Explorer=檔案總管 +Launch an application=執行程式 +File Dialog=檔案視窗 +Wallpaper=桌布 +Configure skin=設定顯示顏色 +Umount=卸載 +Umount external SD=卸載外部SDCard +Poweroff=關機 +Poweroff/Reboot device=機器關機/重新開機 +TV=電視 +Default="預設值" +Message Box Color=對話盒顏色 +Message Box Border Color=對話盒外框顏色 +Message Box Selection Color=對話盒選擇顏色 +Font Color=字形顏色 +Font Outline Color=字形外框顏色 +Wallpaper selection=桌布選擇視窗 +Select an image from the list, to use as a wallpaper=從下面列表選擇桌布 +Do you want to umount external sdcard ?=確認卸載外部SDCard ? +Reboot=重新開機 +Poweroff or reboot device ?=關機或重新開機 ? +Jinyong=金庸 +Jinyong game=金庸群俠傳 +PAL=仙劍 +PAL game=仙劍奇俠傳 +PAL(MHV) game=仙劍奇俠傳(夢幻版) +Rockbot=洛克機器人 +GMenu2X=設定 +File Manager=檔案管理器 +DOS Emulator=DOS模擬器 +Arcade Emulator(A320)=街機模擬器(A320) +Arcade Emulator(GCW0)=街機模擬器(GCW0) +FC/NES Emulator=FC/NES模擬器 +Hugo Emulator=PCE模擬器 +PS1 Emulator=PS1模擬器 +SMS/SMD Emulator=SMS/SMD模擬器 +SNES Emulator=SNES模擬器 +JavaME Emulator=JavaME模擬器 +MAME Emulator=街機模擬器 +PCE Emulator=PCE模擬器 +Comic Viewer=漫畫閱讀軟體 +Text Editor=文字編輯器 +Music Player=音樂播放器 +Image Viewer=圖片閱讀軟體 +ColecoVision Emulator=ColecoVision模擬器 +Atari 2600 Emulator=Atari 2600模擬器 +Atari 800/5200 Emulator=Atari 800/5200模擬器 +Amstrad CPC Emulator=Amstrad CPC模擬器 +GB/GBC Emulator=GB/GBC模擬器 +Arcade Emulator=街機模擬器 +GBA Emulator=GBA模擬器 +Intellivision Emulator=Intellivision模擬器 +PC98 Emulator=PC98模擬器 +OnScripter Emulator=OnScripter模擬器 +WS/WSC Emulator=WS/WSC模擬器 +Atari 7800 Emulator=Atari 7800模擬器 +Neo Geo Pocket Emulator=Neo Geo Pocket模擬器 +ScummVM Emulator=ScummVM模擬器 +SMS/GG Emulator=SMS/GG模擬器 +ZX Spectrum Emulator=ZX Spectrum模擬器 +Amiga500 Emulator=Amiga500模擬器 +Commodore Emulator=Commodore模擬器 +Atari Lynx Emulator=Atari Lynx模擬器 +Arkanoid Game=Arkanoid遊戲 +Apricots Game=Apricots遊戲 +ASCIIpOrtal Game=ASCIIpOrtal遊戲 +Bermuda Syndrome Game=Bermuda Syndrome遊戲 +Boulder Game=Boulder遊戲 +Cave Story Game=洞窟物語(英文) +Cave Story Game(CN)=洞窟物語(簡體) +Cave Story Game(TW)=洞窟物語(繁體) +Chocolate Doom Game=Chocolate Doom遊戲 +Chroma Game=Chroma遊戲 +CZDoom Game=CZDoom遊戲 +Digger Game=Digger遊戲 +FreeDroid Classic Game=FreeDroid Classic遊戲 +Ganbare-Natsuki-San Game=Ganbare-Natsuki-San遊戲 +HHeretic Game=HHeretic遊戲 +HHexen Game=HHexen遊戲 +Homing Fever Game=Homing Fever遊戲 +JinYong Legend Game=金庸群俠傳(復刻版) +JinYong v1.2 Game=金庸群俠傳(蒼龍逐日1.2) +Just4Qix Game=Just4Qix遊戲 +KETM Game=KETM遊戲 +Last Mission Game=Last Mission遊戲 +Meritous Game=Meritous遊戲 +Meteoroid3D Game=Meteoroid3D遊戲 +Mr. Driller Game=Mr. Driller遊戲 +Noiz2sa Game=Noiz2sa遊戲 +OpenTitus Game=OpenTitus遊戲 +Pang Game=Pang遊戲 +ProfaDeluxe Game=ProfaDeluxe遊戲 +Puzzletube Game=Puzzletube遊戲 +Quake Game=Quake遊戲 +Rockbot Game=洛克機器人 +SDLPAL Game=仙劍奇俠傳 +SDLPAL MHV Game=仙劍奇俠傳(夢幻版) +Shifty Pills Game=Shifty Pills遊戲 +Shisen Seki Game=Shisen Seki遊戲 +Snowman Game=Snowman遊戲 +SORR Game=怒之鐵拳 +Spartak Chess Game=Spartak Chess遊戲 +Spout Game=Spout遊戲 +Sonic Robo Blast 2 Game=Sonic Robo Blast 2遊戲 +Super Transball 2 Game=Super Transball 2遊戲 +SuperTux Game=SuperTux遊戲 +Triple Trapled Game=Triple Trapled遊戲 +Wizznic Game=Wizznic遊戲 +Wolfenstein 3D Game=Wolfenstein 3D遊戲 diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Danish b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Danish new file mode 100644 index 0000000..9be85f5 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Danish @@ -0,0 +1,129 @@ +settings=Indstillinger +Configure GMenu2X's options=Konfigurer GMenu2X's Indstillinger +Activate Usb on SD=Aktiver Usb på SD +Activate Usb on Nand=Aktiver Usb på Nand +Info about GMenu2X=Information om GMenu2X +Activate/deactivate tv-out=Aktiver/deaktiver tv-udgang +Exit GMenu2X to the official frontend=Lukker GMenu2X +Change GMenu2X wallpaper=Skift baggrund +About=Om +Add section=Tilføj sektion +Rename section=Ændre navn på sektion +Delete section=Slet sektion +Scan for applications and games=Skan hukommelsen for applikationer og spil +applications=applikationer +Edit link=Rediger genveje +Title=Titel +Link title=Genvejs titel +Description=Beskrivelse +Link description=Genvejs beskrivelse +Section=Seektion +The section this link belongs to=Sektionen for denne genvej +Icon=Ikon +Select an icon for the link=Vælg et ikon til denne genvej +Manual=Manual +Select a graphic/textual manual or a readme=Vælg en grafisk/tekstbaseret manual eller en readme fil +Cpu clock frequency to set when launching this link=Cpu-clockfrekvens indstilling for denne genvej +Volume to set for this link=Lydstyrke indstilling for denne genvej +Parameters=Parametre +Parameters to pass to the application=Angiv parametre for applikationen +Selector Directory=Selector oversigt +Directory to scan for the selector=Angiv Mappe som selector skal skanne +Selector Browser=Selector Browser +Allow the selector to change directory=Tillad selector at ændre mappe +Selector Filter=Selector filter +Filter for the selector (Separate values with a comma)=Filter til selector (separer værdier med komma) +Selector Screenshots= Selector Screenshots +Directory of the screenshots for the selector=Mappe med Screenshots af selector +Selector Aliases=Selector alias +File containing a list of aliases for the selector=Fil som indeholder en liste over alias for selector +Explicitly relaunch GMenu2X after this link's execution ends=Tving GMenu2X til at genstarte når denne genvej køres +Don't Leave=Forlad ikke +Don't quit GMenu2X when launching this link=Afslut ikke GMenu2X når denne genvej startes +Save last selection=Gem sidste ændring +Save the last selected link and section on exit= Gem sidst valgte genvej og sektion ved afslutning +Clock for GMenu2X=Clockfrekvens for GMenu2X +Set the cpu working frequency when running GMenu2X=Indstil cpu-clockfrekvens for GMenu2X +Maximum overclock=Maksimal clockfrekvens +Set the maximum overclock for launching links=Indstil maksimal clockfrekvens ved opstart af genvej +Global Volume=Global lydstyrke +Set the default volume for the gp2x soundcard=Indstil standard lydstyrke for gp2x lydkort +Output logs=Vis logs +Logs the output of the links. Use the Log Viewer to read them.=Danner logs for genvejene. Anvend Vis log for at åbne dem. +Number of columns=Antal spalter +Set the number of columns of links to display on a page=Angiv antallet af spalter for genveje per side +Number of rows=Antal rækker +Set the number of rows of links to display on a page= Angiv antallet af rækker for genveje per side +Top Bar Color=Øverste bjælkes farve +Color of the top bar= Øverste bjælkes farve +Bottom Bar Color=Nederste bjælkes farve +Color of the bottom bar=Nederste bjælkes farve +Selection Color=Markørens farve +Color of the selection and other interface details= Markøren og grænseflades farve +You should disable Usb Networking to do this.=Du bør fravælge USB netværket nå du vælger dette +Operation not permitted.=Dette er ikke tilladt. +Language=Sprog +Set the language used by GMenu2X=Indstil sprog der anvendes i GMenu2X +Increase=Op +Decrease=Ned +Change color component=Ændre farven på komponent +Increase value=Op +Decrease value=Ned +Switch=Ændre +Change value=Ændre værdi +Edit=Rediger +Clear=Ryd +Select a directory=Vælg en mappe +Select a file=Vælg en fil +Clock (default: 200)=Clockfrekvens (normal: 200) +Volume (default: -1)=Lydstyrke (normal: -1) +Wrapper=Wrapper +Enter folder=Åbn mappe +Confirm=Bekræft +Enter folder/Confirm=Åbn mappe/Bekræft +Up one folder=Tilbage +Select an application=Vælg en applikation +Space=Mellemrum +Shift=Skift +Cancel=Afbryd +OK=OK +Backspace=Slet +Skin=Tema +Set the skin used by GMenu2X =Angiv tema for GMenu2X +Add link in $1=Tilføj genvej i $1 +Edit $1=Rediger $1 +Delete $1 link=Slet $1 +Deleting $1=Sletter $1 +Are you sure?=Er du sikker? +Insert a name for the new section=Angiv navn for den nye sektion +Insert a new name for this section=Angiv nyt navn for denne sektion +Yes=Ja +No=Nej +You will lose all the links in this section.=Du vil miste alle genveje i denne sektion. +Exit=Afslut +Link Scanner=Skan genveje +Scanning SD filesystem...=Skanner SD filsystem... +Scanning NAND filesystem...=Skanner NAND filsystem... +$1 files found.=$1 fil(er) fundet. +Creating links...=Opretter genveje... +$1 links created.=$1 genvej(e) oprettet. +Version $1 (Build date: $2)=Version $1 (den: $2) +Log Viewer=Vis log +Displays last launched program's output=Vis log fra sidst kørte program +Do you want to delete the log file?=Vil du slette denne log fil? +USB Enabled (SD)=USB Aktiveret (SD) +USB Enabled (Nand)=USB Aktiveret (Nand) +Turn off=Afbryd +Launching $1=Starter $1 +Change page=Skift side +Page=Side +Scroll=Rulle +Untitled=Ikke navngivet +Wallpaper=Baggrund +Configure skin=Konfigurer tema +Message Box Color=Farve på Konfigurations vinduet +Message Box Border Color= Farve på Konfig vinduets kant +Message Box Selection Color=Konfig vinduets markør farve +Background color of the message box= Konfigurations vinduets baggrundsfarve +Border color of the message box=Farve på Konfigurations vinduets kant +Color of the selection of the message box=Farven på markøren i Konfigurations vinduet diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Dutch b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Dutch new file mode 100644 index 0000000..eaa3062 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Dutch @@ -0,0 +1,118 @@ +Settings=Instellingen +Configure GMenu2X's options=Instellingen van GMenu2X +Activate Usb on SD=Activeer USB op SD +Activate Usb on Nand=Activeer USB op Nand +Info about GMenu2X=Informatie over GMenu2X +About=Over +Add section=Groep toevoegen +Rename section=Groep hernoemen +Delete section=Groep verwijderen +Scan for applications and games=Zoek applicaties en spellen +applications=applicaties +Edit link=Wijzig snelkoppeling +Title=Naam +Link title=Naam van de snelkoppeling +Description=Omschrijving +Link description=Omschrijving van de snelkoppeling +Section=Groep +The section this link belongs to=De groep waartoe deze snelkoppeling behoort +Icon=Pictogram +Select an icon for the link: $1=Selecteer een pictogram voor de snelkoppeling: $1 +Manual=Handleiding +Select a graphic/textual manual or a readme=Selecteer een grafische/tekstuele handleiding of een readme/leesmij +Cpu clock frequency to set when launching this link=Kloksnelheid voor het starten van deze snelkoppeling +Volume to set for this link=Volume-instelling voor deze snelkoppeling +Parameters=Parameters +Parameters to pass to the application=Parameters om door te geven aan de applicatie +Selector Directory=Kiezer map +Directory to scan for the selector=Map met bestanden voor de kiezer +Selector Browser=Kiezer navigatie +Allow the selector to change directory=Wisselen van map mogelijk maken in kiezer +Selector Filter=Kiezer filter +Filter for the selector (Separate values with a comma)=Filter voor de kiezer (komma-gescheiden) +Selector Screenshots=Kiezer schermafdrukken +Directory of the screenshots for the selector=Map met schermafdrukken voor de kiezer +Selector Aliases=Kiezer aliassen +File containing a list of aliases for the selector=Bestand met aliassen voor de kiezer +Explicitly relaunch GMenu2X after this link's execution ends=GMenux2X altijd herstarten na uitvoeren snelkoppeling +Don't Leave=Niet verlaten +Don't quit GMenu2X when launching this link=GMenu2X niet verlaten bij uitvoeren snelkoppeling +Save last selection=Bewaar laatste keuze +Save the last selected link and section on exit=Bewaar de laatst gekozen snelkoppeling en groep +Clock for GMenu2X=Kloksnelheid voor GMenu2X +Set the cpu working frequency when running GMenu2X=Stel de kloksnelheid in voor het draaien van Gmenu2X +Maximum overclock=Maximale overkloksnelheid +Set the maximum overclock for launching links=Stel maximaal toegestane overkloksnelheid in +Global Volume=Hoofdvolume +Set the default volume for the gp2x soundcard=Stel het standaardvolume in voor de gp2x geluidskaart +Output logs=Uitvoer naar logboek +Logs the output of the links. Use the Log Viewer to read them.=Logt de uitvoer van snelkoppelingen. Gebruik de log viewer om de log te lezen. +Number of columns=Aantal kolommen +Set the number of columns of links to display on a page=Stel het aantal getoonde pictogrammen in (horizontaal) +Number of rows=Aantal rijen +Set the number of rows of links to display on a page=Stel het aantal getoonde pictogrammen in (verticaal) +Top Bar Color=Kleur bovenste balk +Color of the top bar=Kleur van de bovenste balk +Bottom Bar Color=Kleur onderste balk +Color of the bottom bar=Kleur van de onderste balk +Selection Color=Kleur selectie +Color of the selection and other interface details=Kleur van de selectie en andere interface details +You should disable Usb Networking to do this.=Zet USB Netwerk uit om dit te gebruiken. +Operation not permitted.=Handeling niet toegestaan. +Language=Taal +Set the language used by GMenu2X=Stel de taal van GMenu2X in +Increase=Verhoog +Decrease=Verlaag +Change color component=Wijzig kleur component +Increase value=Verhoog waarde +Decrease value=Verlaag waarde +Switch=Wissel +Change value=Wijzig waarde +Edit=Bewerk +Clear=Wis +Select a directory=Selecteer een map +Select a file=Selecteer een bestand +Clock (default: 200)=Kloksnelheid (standaard: 200) +Volume (default: -1)=Volume (standaard: -1) +Wrapper=Schil +Enter folder=Open map +Confirm=Bevestig +Enter folder/Confirm=Open map/Bevestig +Up one folder=Map omhoog +Select an application=Selecteer een applicatie +Space=Spatie +Shift=Shift +Cancel=Annuleer +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Kies de skin voor GMenu2X +Add link in $1=Voeg snelkoppeling toe in $1 +Edit $1=Wijzig $1 +Delete $1 link=Verwijder snelkoppeling $1 +Deleting $1=Verwijdert $1 +Are you sure?=Weet u het zeker? +Insert a name for the new section=Geef een naam voor de nieuwe groep +Insert a new name for this section=Geef een nieuwe naam voor deze groep +Yes=Ja +No=Nee +You will lose all the links in this section.=Alle snelkoppelingen in deze groep worden gewist. +Exit=Verlaat +Link Scanner=Snelkoppeling-Scanner +Scanning SD filesystem...=Scant het SD-bestandssysteem... +Scanning NAND filesystem...=Scant het NAND-bestandssysteem... +$1 files found.=$1 bestand(en) gevonden. +Creating links...=Snelkoppelingen aanmaken... +$1 links created.=$1 Snelkoppeling(en) aangemaakt. +Version $1 (Build date: $2)=Versie $1 (Datum build: $2) +Log Viewer=Logbestand lezer +Displays last launched program's output=Toont de uitvoer van het laatst gestarte programma +Do you want to delete the log file?=Wilt u het logbestand verwijderen? +USB Enabled (SD)=USB geactiveerd (SD) +USB Enabled (Nand)=USB geactiveerd (Nand) +Turn off=Zet uit +Launching $1=Start $1 +Change page=Wijzig pagina +Page=Pagina +Scroll=Scroll +Untitled=Naamloos diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Finnish b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Finnish new file mode 100644 index 0000000..9158178 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Finnish @@ -0,0 +1,117 @@ +Settings=Asetukset +Configure GMenu2X's options=Muuta GMenu2X:n asetuksia +Activate Usb on SD=Aktivoi USB SD-kortille +Activate Usb on Nand=Aktivoi USB NAND-muistille +Info about GMenu2X=Tietoa GMenu2X:st� +About=Tietoa +Add section=Lis�� v�lilehti +Rename section=Nime� v�lilehti uudelleen +Delete section=Poista v�lilehti +Scan for applications and games=Etsi ohjelmia ja pelej� +applications=ohjelmat +Edit link=Muokkaa linkki� +Title=Otsikko +Link title=Linkin otsikko +Description=Kuvaus +Link description=Linkin kuvaus +Section=V�lilehti +The section this link belongs to=V�lilehti johon t�m� linkki kuuluu +Icon=Kuvake +Select an icon for the link: $1=Valitse kuvake linkille: $1 +Manual=Ohjetiedosto +Select a graphic/textual manual or a readme=Valitse graafinen/tekstipohjainen ohjetiedosto +Cpu clock frequency to set when launching this link=CPU kellotaajuus t�m�n linkin k�ynnistyksess� +Volume to set for this link=��nenvoimakkuus t�lle linkille +Parameters=Parametrit +Parameters to pass to the application=Ohjelmalle annettavat parametrit +Selector Directory=Ohjelmanvalitsimen hakemisto +Directory to scan for the selector=Hakemisto joka skannataan ohjelmanvalitsimelle +Selector Browser=Ohjelmanvalitsimen selain +Allow the selector to change directory=Anna ohjelmanvalitsimen vaihtaa hakemistoa +Selector Filter=Ohjelmavalitsimen filtteri +Filter for the selector (Separate values with a comma)=Filtteri ohjelmanvalitsimelle (Eroita arvot pilkulla) +Selector Screenshots=Kuvakaappaukset ohjelmanvalitsimesta +Directory of the screenshots for the selector=Ohjelmanvalitsimen kuvakaappausten hakemisto +Selector Aliases=Ohjelmanvalitsimen peitenimet +File containing a list of aliases for the selector=Tiedosto, joka sis�lt�� listan peitenimist� ohjelmavalitsimelle +Explicitly relaunch GMenu2X after this link's execution ends=K�ynnist� GMenu2X uudelleen kun linkin ajo on suoritettu +Don't Leave=�l� poistu +Don't quit GMenu2X when launching this link=�l� sulje GMenu2X:�� kun linkki k�ynnistet��n +Save last selection=Muista viimeisin valinta +Save the last selected link and section on exit=Muista viimeisin valinta ja v�lilehti poistuttaessa +Clock for GMenu2X=Kellotaajuus GMenu2X:lle +Set the cpu working frequency when running GMenu2X=S��d� CPU kellotaajuutta GMenu2X:lle +Maximum overclock=Ylikellotusrajoitus +Set the maximum overclock for launching links=S��d� suurin mahdollinen ylikellotus k�ynnistett�ess� linkkej� +Global Volume=Yleinen ��nenvoimakkuus +Set the default volume fo the gp2x soundcard=S��d� perus��nenvoimakkuus gp2x:n ��nikortille +Output logs=Tulosteloki +Logs the output of the links. Use the Log Viewer to read them.=Kirjoita linkkien tuloste lokiin. K�yt� lokilukijaa niiden lukemiseen. +Number of columns=Sarakkeiden lukum��r� +Set the number of columns of links to display on a page=Aseta linkkisarakkeiden lukum��r� sivulla +Number of rows=Rivien lukum��r� +Set the number of rows of links to display on a page=Aseta linkkirivien lukum��r� sivulla +Top Bar Color=V�ri yl�palkille +Color of the top bar=Yl�palkin v�ri +Color of the bottom bar=Alapalkin v�ri +Selection Color=Valinnan v�ri +Color of the selection and other interface details=Valinnan ja muiden ykstiyiskohtien v�ri +You should disable Usb Networking to do this.=Usb Networking:in pit�� olla poissa k�yt�st� jotta voit tehd� t�m�n. +Operation not permitted.=Toiminto ei ole sallittu. +Language=Kieli +Set the language used by GMenu2X=Valitse GMenu2X:n k�ytt�m� kieli +Increase=Lis�� +Decrease=V�henn� +Change color component=Vaihda v�rikomponenttia +Increase value=Nosta arvoa +Decrease value=Laske arvoa +Switch=Vaihda +Change value=Vaihda arvoa +Edit=Muokkaa +Clear=Tyhjenn� +Select a directory=Valitse hakemisto +Select a file=Valitse tiedosto +Clock (default: 200)=Kellotaajuus (oletusarvo: 200) +Volume (default: -1)=��nenvoimakkuus (oletusarvo: -1) +Wrapper=Wrapperi +Enter folder=Avaa kansio +Confirm=Vahvista +Enter folder/Confirm=Avaa kansio/Vahvista +Up one folder=Yksi hakemisto yl�sp�in +Select an application=Valitse ohjelma +Space=V�lily�nti +Shift=Vaihto +Cancel=Peruuta +OK=OK +Backspace=Askelpalautin +Skin=Teema +Set the skin used by GMenu2X=Aseta GMenu2X:n k�ytt�m� teema +Add link in $1=Lis�� linkki v�lilehteen $1 +Edit $1=Muokkaa v�lilehte� $1 +Delete $1 link=Poista v�lilehti $1 +Deleting $1=Poistetaan v�lilehte� $1 +Are you sure?=Oletko varma? +Insert a name for the new section=Anna uuden v�lilehden nimi +Insert a new name for this section=Anna uusi nimi t�lle v�lilehdelle +Yes=Kyll� +No=Ei +You will lose all the links in this section.=Menet�t kaikki t�ss� v�lilehdess� olevat linkit. +Exit=Poistu +Link Scanner=Linkkiskanneri +Scanning SD filesystem...=Skannataan SD-tiedostoj�rjestelm��... +Scanning NAND filesystem...=Skannataan NAND-tiedostoj�rjestelm��... +$1 files found.=$1 tiedosto(a) l�ydetty. +Creating links...=Luodaan linkkej�... +$1 links created.=$1 linkki(�) luotu. +Version $1 (Build date: $2)=Versio $1 (K��nt�p�iv�m��r�: $2) +Log Viewer=Lokilukija +Displays last launched program's output=N�ytt�� viimeksi k�ynnistetyn ohjelman tulosteen +Do you want to delete the log file?=Haluatko poistaa logitiedoston? +USB Enabled (SD)=USB Aktivoitu (SD) +USB Enabled (Nand)=USB Aktivoitu (Nand) +Turn off=Sammuta +Launching $1=K�ynnistet��n $1 +Change page=Vaihda sivua +Page=Sivu +Scroll=Vierit� +Untitled=Nime�m�t�n diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/French b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/French new file mode 100644 index 0000000..f3d6081 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/French @@ -0,0 +1,129 @@ +Settings=Configurations +Configure GMenu2X's options=Configurer les options de GMenu2X +Activate Usb on SD=Activer l'Usb sur la SD +Activate Usb on Nand=Activer l'Usb sur la Nand +Info about GMenu2X=Information sur GMenu2X +About=A propos de +Add section=Ajouter une section +Rename section=Renommer une section +Delete section=Supprimer une section +Scan for applications and games=Rechercher des applications et des jeux +applications=applications +Edit link=Editer un lien +Title=Titre +Link title=Titre du lien +Description=Description +Link description=Description du lien +Section=Section +The section this link belongs to=La section à laquelle appartient ce lien +Icon=Icône +Select an icon for the link: $1=Selectionner une icône pour le lien: $1 +Manual=Manuel +Select a graphic/textual manual or a readme=Selectionner un manuel graphique/textuel ou un readme +Cpu clock frequency to set when launching this link=Fréquence d'horloge CPU à définir lorsqu'on lance ce lien +Volume to set for this link=Volume à définir pour ce lien +Parameters=Paramètres +Parameters to pass to the application=Paramètres à donner à l'application +Selector Directory=Répertoire de Selector +Directory to scan for the selector=Répertoire à rechercher pour le selector +Selector Browser=Explorateur de Selector +Allow the selector to change directory=Laisser le Selector de changer le répertoire +Selector Filter=Filtre de Selector +Filter for the selector (Separate values with a comma)=Filtre pour le Selector (Séparer les valeurs avec une virgule) +Selector Screenshots=Les captures d'écran de Selector +Directory of the screenshots for the selector=Répertoire des captures d'écran pour le selector +Selector Aliases=Aliases de Selector +File containing a list of aliases for the selector=Fichier contenant une liste d'aliases pour selector +Explicitly relaunch GMenu2X after this link's execution ends=Relancer explicitement GMenu2X après la fin de l'execution de ce lien +Don't Leave=Ne quitter pas +Don't quit GMenu2X when launching this link=Ne pas quitter GMenu2X lorsqu'on lance ce lien +Save last selection=Sauvegarder la dernière sélection +Save the last selected link and section on exit=Sauvegarder le dernier lien sélectionné et section en sortant +Clock for GMenu2X=Horloge pour GMenu2X +Set the cpu working frequency when running GMenu2X=Définir la féquence de fonctionnement du CPU lorsqu'on utilise GMenu2X +Maximum overclock=Overclock maximum +Set the maximum overclock for launching links=Définir l'overclock maximum pour lancer des liens +Global Volume=Volume global +Set the default volume for the gp2x soundcard=Définir le volume par défaut pour la carte sonore de la GP2x +Output logs=Logs de sortie +Logs the output of the links. Use the Log Viewer to read them.=Loguer la sortie des liens. Utiliser le lecteur de log pour les lire +Number of columns=Nombre de colonnes +Set the number of columns of links to display on a page=Définir le nombre de colonnes de liens à afficher sur une page +Number of rows=Nombres de rangées +Set the number of rows of links to display on a page=Définir le nombre de rangées de liens à afficher sur une page +Top Bar Color=Couleur de la bar supérieur +Color of the top bar=Couleur de la bar supérieur +Bottom Bar Color=Couleur de la bar inférieur +Color of the bottom bar=Couleur de la bar inférieur +Selection Color=Couleur de sélection +Color of the selection and other interface details=Couleur de la sélection et des autres détails de l'interface +You should disable Usb Networking to do this.=Vous devez désactiver le réseau Usb pour faire ceci. +Operation not permitted.=Opération non permise +Language=Langue +Set the language used by GMenu2X=Définir la langue utilisée par GMenu2X +Increase=Augmenter +Decrease=Diminuer +Change color component=Changer la couleur composante +Increase value=Augmenter la valeur +Decrease value=Diminuer la valeur +Switch=Changer +Change value=Changer la valeur +Edit=Editer +Clear=Effacer +Select a directory=Sélectionner un répertoire +Select a file=Sélectionner un fichier +Clock (default: 200)=Fréquence (par défaut: 200) +Volume (default: -1)=Volume (par défaut: -1) +Wrapper=Wrapper +Enter folder=Enter dans le répertoire +Confirm=Confirmer +Enter folder/Confirm=Entrer un répertoire/Confirmer +Up one folder=Remonter d'un répertoire +Select an application=Sélectionner une application +Space=Espace +Shift=Majuscule +Cancel=Annuler +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Définir la skin utilisé par GMenu2X +Add link in $1=Ajouter un lien dans $1 +Edit $1=Editer $1 +Delete $1 link=Effacer le lien $1 +Deleting $1=Effacement de $1 +Are you sure?=Êtes vous sûr? +Insert a name for the new section=Saisir un nom pour cette nouvelle section +Insert a new name for this section=Saisir un nouveau nom pour cette section +Yes=Oui +No=Non +You will lose all the links in this section.=Vous perdrez tous les liens dans cette section. +Exit=Quitter +Link Scanner=Scanneur de lien +Scanning SD filesystem...=Scanne du sytème de fichier de la SD... +Scanning NAND filesystem...=Scanne du sytème de fichier de la NAND... +$1 files found.=$1 fichiers trouvés. +Creating links...=Création de liens... +$1 links created.=$1 liens créés. +Version $1 (Build date: $2)=Version $1 (Date de compilation: $2) +Log Viewer=Visualisateur de log +Displays last launched program's output=Afficher la sortie du dernier programme lancé +Do you want to delete the log file?=Voulez vous effacer le fichier de log? +USB Enabled (SD)=USB Activé (SD) +USB Enabled (Nand)=USB Activé (Nand) +Turn off=Désactiver +Launching $1=Lancement de $1 +Change page=Changer de page +Page=Page +Scroll=Défilement +Untitled=Sans titre +Change GMenu2X wallpaper=Changer l'arrière-plan de GMenu2X +Activate/deactivate tv-out=Activer/desactiver la sortie TV. +Select wallpaper=Sélectionner l'arrière-plan +Gamma=Gamma +Set gp2x gamma value (default: 10)=Définir la valeur de gamma de la gp2x (par défaut: 10) +Tv-Out encoding=Encodage de la sortie TV +Encoding of the tv-out signal=Encodage du signal de la sortie TV +Tweak RAM Timings=Modification des timings de la RAM +This usually speeds up the application at the cost of stability=Ceci accélère, normalement, l'application mais en contre partie de la stabilité +Gamma (default: 0)=Gamma (par défaut: 0) +Gamma value to set when launching this link=Valeur de gamma à définir lors du lancement de ce lien \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/German b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/German new file mode 100644 index 0000000..c34ba94 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/German @@ -0,0 +1,129 @@ +Settings=Einstellungen +Configure GMenu2X's options=Optionen des GMenu2X konfigurieren +Activate Usb on SD=Aktiviert USB für die SD-Karte +Activate Usb on Nand=Aktiviert USB für den Nand-Speicher +Info about GMenu2X=Informationen über GMenu2X +About=Über +Add section=Sektion hinzufügen +Rename section=Sektion umbenennen +Delete section=Sektion löschen +Scan for applications and games= Nach Anwendungen und Spielen scannen +applications=Anwendungen +Edit link=Link bearbeiten +Title=Titel +Link title=Linktitel +Description=Beschreibung +Link description=Beschreibung für diesen Link +Section=Sektion +The section this link belongs to=Zum Link gehörende Sektion +Icon=Icon +Select an icon for the link: $1=Icon für diesen Link wählen: $1 +Manual=Anleitung +Select a graphic/textual manual or a readme=Wähle eine Graphik/Text Anleitung oder Readme +Cpu clock frequency to set when launching this link=CPU-Takt, mit welchem dieser Link gestartet wird +Volume to set for this link=Lautstärke für diesen Link +Parameters=Parameter +Parameters to pass to the application=Parameter, die an die Anwendung übergeben werden +Selector Directory=Selector Verzeichnis +Directory to scan for the selector=Vom Selector zu scannendes Verzeichnis +Selector Browser=Selector Browser +Allow the selector to change directory=Erlaube dem Selector, das Verzeichnis zu wechseln +Selector Filter=Selector Filter +Filter for the selector (Separate values with a comma)=Filter für den Selector (Werte mit Komma trennen) +Selector Screenshots=Selector Screenshots +Directory of the screenshots for the selector=Screenshot-Verzeichnis für den Selector +Selector Aliases=Selector Alternativnamen +File containing a list of aliases for the selector=Datei mit Liste von Alternativnamen für den Selector +Explicitly relaunch GMenu2X after this link's execution ends=GMenu2X nach Beenden der Anwendung umgehend Neustarten +Don't Leave=Nicht verlassen +Don't quit GMenu2X when launching this link=Beim starten dieses Links GMenu2X nicht beenden +Save last selection=Letzte Auswahl speichern +Save the last selected link and section on exit=Speichert den zuletzt gewählten Link und die Sektion beim Beenden +Clock for GMenu2X=Taktfrequenz für GMenu2X +Set the cpu working frequency when running GMenu2X=Stellt den CPU Arbeitstakt für GMenu2X ein +Maximum overclock=Maximale Übertaktung +Set the maximum overclock for launching links=Einstellen der maximalen Taktfrequenz zum Starten von Links +Global Volume=Allgemeine Lautstärke +Set the default volume for the gp2x soundcard=Einstellen der Standardlautstärke des GP2X Soundchips +Output logs=Ausgabeprotokolle +Logs the output of the links. Use the Log Viewer to read them.=Protokolliert Ausgabe der Links. Benutz den Log Viewer zum Lesen. +Number of columns=Anzahl der Spalten +Set the number of columns of links to display on a page=Anzahl der Spalten mit Links, pro Seite +Number of rows=Anzahl der Zeilen +Set the number of rows of links to display on a page=Anzahl der Zeilen mit Links, pro Seite +Top Bar Color=Farbe der Kopfleiste +Color of the top bar= Stellt Farbe und Transparenz der oberen Menüleiste ein +Bottom Bar Color=Farbe der Fußleiste +Color of the bottom bar=Stellt Farbe und Transparenz der unteren Menüleiste ein +Selection Color=Farbe der Auswahl +Color of the selection and other interface details=Farbe der Auswahl-Hervorhebung und anderer Interface-Details +You should disable Usb Networking to do this.=Du solltest USB Networking deaktivieren um dies zu tun. +Operation not permitted.=Operation nicht gestattet. +Language=Sprache +Set the language used by GMenu2X=Einstellen der GMenu2X-Sprache +Increase=Erhöhen +Decrease=Verringern +Change color component=Farbkomponente ändern +Increase value=Wert erhöhen +Decrease value=Wert verringern +Switch=Wechseln +Change value=Wert ändern +Edit=Bearbeiten +Clear=Leeren +Select a directory=Verzeichnis wählen +Select a file=Datei wählen +Clock (default: 200)=Taktfrequenz (Standard: 200) +Volume (default: -1)=Lautstärke (Standard: -1) +Wrapper=Wrapper +Enter folder=Ordner öffnen +Confirm=Bestätigen +Enter folder/Confirm=Ordner öffnen/Bestätigen +Up one folder=Einen Ordner Aufwärts +Select an application=Anwendung wählen +Space=Leer +Shift=Umschalt +Cancel=Abbrechen +OK=OK +Backspace=Rücktaste +Skin=Skin +Set the skin used by GMenu2X=Skin für GMenu2X auswählen +Add link in $1=Link in "$1" hinzufügen +Edit $1="$1" bearbeiten +Delete $1 link=Link "$1" löschen +Deleting $1="$1" wird gelöscht +Are you sure?=Sind Sie sicher? +Insert a name for the new section=Name für die neue Sektion +Insert a new name for this section=Neuer Name für diese Sektion +Yes=Ja +No=Nein +You will lose all the links in this section.=Alle Links in dieser Sektion gehen verloren. +Exit=Beenden +Link Scanner=Links suchen +Scanning SD filesystem...=SD-Karte wird durchsucht... +Scanning NAND filesystem...=NAND-Speicher wird durchsucht... +$1 files found.=$1 Datei(en) gefunden. +Creating links...=Verknüpfungen werden erstellt. +$1 links created.=$1 Verknüpfung(en) erstellt. +Version $1 (Build date: $2)=Version $1 vom $2 +Log Viewer=Log-Viewer +Displays last launched program's output=Zeigt die Ausgabe der zuletzt gestarteten Anwendung an +Do you want to delete the log file?=Möchten Sie diese Log-Datei löschen? +USB Enabled (SD)=USB aktiviert (SD) +USB Enabled (Nand)=USB aktiviert (Nand) +Turn off=Abschalten +Launching $1=$1 wird gestartet +Change page=Seite wechseln +Page=Seite +Scroll=Scrollen +Untitled=Unbenannt +Change GMenu2X wallpaper=Ändert das GMenu2X Hintergrundbild +Activate/deactivate tv-out=Aktiviert/deaktiviert TV-O ut +Select wallpaper=Hintergrundbild wählen +Gamma=Gamma +Set gp2x gamma value (default: 10)=Setzt den GP2X Gamma-Wert (Standard: 10) +Tv-Out encoding=TV-Out Signal +Encoding of the tv-out signal=Einstellen der TV-Out Fernsehnorm +Tweak RAM Timings=Schnellere RAM-Timings +This usually speeds up the application at the cost of stability=Kann Anwendung beschleunigen, auf Kosten der Stabilität +Gamma (default: 0)=Gamma (Standard: 0) +Gamma value to set when launching this link=Bein Starten dieses Links benutzter Gamma-Wert diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Italian b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Italian new file mode 100644 index 0000000..bad8d40 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Italian @@ -0,0 +1,142 @@ +Settings=Impostazioni +Configure GMenu2X's options=Configura le opzioni di GMenu2X +Activate Usb on SD=Attiva USB sulla SD +Activate Usb on Nand=Attiva USB sulla Nand +Info about GMenu2X=Informazioni su GMenu2X +About=Informazioni +Add section=Aggiungi sezione +Rename section=Rinomina sezione +Delete section=Elimina sezione +Scan for applications and games=Cerca applicazioni e giochi +applications=applicazioni +Edit link: $1=Modifica collegamento: $1 +Title=Titolo +Link title=Titolo collegamento +Description=Descrizione +Link description=Descrizione collegamento +Section=Sezione +The section this link belongs to=La sezione alla quale appartiene questo collegamento +Icon=Icona +Select an icon for the link: $1=Seleziona un'icona per il collegamento: $1 +Manual=Manuale +Select a graphic/textual manual or a readme=Seleziona un manuale grafico/testuale o un readme +Cpu clock frequency to set when launching this link=Clock della cpu da impostare quando si lancia questo collegamento +Volume to set for this link=Volume da impostare per questo collegamento +Parameters=Parametri +Parameters to pass to the application=Parametri da passare all'applicazione +Selector Directory=Directory del selettore +Directory to scan for the selector=Directory da utilizzare con il selettore +Selector Browser=Browser del selettore +Allow the selector to change directory=Permetti al selettore di cambiare directory +Selector Filter=Filtro del selettore +Filter for the selector (Separate values with a comma)=Filtro per il selettore (Separa i valori con virgola) +Selector Screenshots=Screenshot del selettore +Directory of the screenshots for the selector=Directory contenente gli screenshot per il selettore +Selector Aliases=Alias del selettore +File containing a list of aliases for the selector=File contenente una lista di alias per il selettore +Explicitly relaunch GMenu2X after this link's execution ends=Rilancia esplicitamente GMenu2X dopo l'esecuzione del collegamento +Don't Leave=Non lasciare +Don't quit GMenu2X when launching this link=Non terminare GMenu2X quando viene lanciato questo collegamento +Save last selection=Salva ultima selezione +Save the last selected link and section on exit=Salva l'ultimo collegamento e sezione usati quando si esce +Clock for GMenu2X=Clock per GMenu2X +Set the cpu working frequency when running GMenu2X=Imposta la frequenza di lavoro per GMenu2X +Maximum overclock=Overclock massimo +Set the maximum overclock for launching links=Imposta la frequenza massima per i collegamenti +Global Volume=Volume globale +Set the default volume for the gp2x soundcard=Imposta il volume standard per la gp2x +Output logs=Log dell'output +Logs the output of the links. Use the Log Viewer to read them.=Tiene traccia dell'output dei collegamenti. Usa il Visualizzatore di Log per leggerlo. +Number of columns=Numero di colonne +Set the number of columns of links to display on a page=Imposta il numero di colonne di collegamenti da visualizzare in una pagina +Number of rows=Numero di righe +Set the number of rows of links to display on a page=Imposta il numero di righe di collegamenti da visualizzare in una pagina +Top Bar Color=Colore barra superiore +Color of the top bar=Colore della barra superiore +Bottom Bar Color=Colore barra inferiore +Color of the bottom bar=Colore della barra inferiore +Selection Color=Colore selezione +Color of the selection and other interface details=Colore della selezione e altri dettagli dell'interfaccia +You should disable Usb Networking to do this.=Dovresti disattivare le impostazioni di rete per farlo. +Operation not permitted.=Operazione non consentita. +Language=Lingua +Set the language used by GMenu2X=Imposta la lingua usata da GMenu2X +Increase=Aumenta +Decrease=Riduci +Change color component=Cambia componente cromatico +Increase value=Incrementa valore +Decrease value=Decrementa valore +Switch=Cambia +Change value=Cambia valore +Edit=Modifica +Clear=Svuota +Select a directory=Seleziona una directory +Select a file=Seleziona un file +Clock (default: 200)=Frequenza (predefinito: 200) +Volume (default: -1)=Volume (predefinito: -1) +Wrapper=Involucro +Enter folder=Entra nella cartella +Confirm=Conferma +Enter folder/Confirm=Entra nella cartella/Conferma +Up one folder=Sali di una cartella +Select an application=Seleziona un'applicazione +Space=Spazio +Shift=Maiusc +Cancel=Annulla +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Imposta la skin usata da GMenu2X +Add link in $1=Aggiungi collegamento in $1 +Edit $1=Modifica $1 +Delete $1 link=Elimina il collegamento $1 +Deleting $1=Rimozione di $1 +Are you sure?=Sei sicuro? +Insert a name for the new section=Inserisci un nome per la nuova sezione +Insert a new name for this section=Inserisci un nuovo nome per questa sezione +Yes=Si +No=No +You will lose all the links in this section.=Perderai tutti i collegamenti in questa sezione. +Exit=Esci +Link Scanner=Scanner di collegamenti +Scanning SD filesystem...=Scansione del filesystem della SD... +Scanning NAND filesystem...=Scansione del filesystem della NAND... +$1 files found.=$1 file trovati. +Creating links...=Creazione collegamenti... +$1 links created.=$1 collegamenti creati. +Version $1 (Build date: $2)=Versione $1 (Data compilazione: $2) +Log Viewer=Visualizzatore di log +Displays last launched program's output=Visualizza l'output dell'ultimo programma eseguito +Do you want to delete the log file?=Vuoi eliminare il file di log? +USB Enabled (SD)=USB Attivata (SD) +USB Enabled (Nand)=USB Attivata (Nand) +Turn off=Disattiva +Launching $1=Esecuzione di $1 +Change page=Cambia pagina +Page=Pagina +Scroll=Scorri +Untitled=Senza titolo +Change GMenu2X wallpaper=Cambia lo sfondo di GMenu2X +Activate/deactivate tv-out=Attiva/disattiva tv-out +Select wallpaper=Seleziona sfondo +Gamma=Gamma +Set gp2x gamma value (default: 10)=Imposta il valore gamma della gp2x (predefinito: 10) +Tv-Out encoding=Codifica uscita tv +Encoding of the tv-out signal=Codifica del segnale dell'uscita tv +Tweak RAM Timings=Modifica i timings della RAM +This usually speeds up the application at the cost of stability=Comporta solitamente un miglioramento delle performance al costo di stabilit� +Gamma (default: 0)=Gamma (predefinito: 0) +Gamma value to set when launching this link=Valore di gamma da impostare quando si lancia questo collegamento +Wallpaper=Sfondo +Configure skin=Configura skin +Message Box Color=Colore Finestra Messaggi +Message Box Border Color=Colore Bordo Finestra Messaggi +Message Box Selection Color=Color Selezione Finestra Messaggi +Background color of the message box=Colore di sfondo della finestra dei messaggi +Border color of the message box=Colore del bordo della finestra dei messaggi +Color of the selection of the message box=Colore della selezione della finestra dei messaggi + +Show root=Mostra radice +Show root folder in the file selection dialogs=Mostra la cartella radice nelle finestre di selezione di file +Change keys=Cambia tasti +Launch an application=Esegue un'applicazione diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Korean b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Korean new file mode 100644 index 0000000..c11d508 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Korean @@ -0,0 +1,133 @@ +Settings=설정 +Configure GMenu2X's options=GMenu2X의 옵션 구성하기 +Activate Usb on SD=SD Usb 활성화 +Activate Usb on Nand=Nand Usb 활성화 +Info about GMenu2X=GMenu2X에 대한 정보 +About=대하여 +Add section=섹션 추가 +Rename section=섹션 이름 변경 +Delete section=섹션 삭제 +Scan for applications and games=응용 프로그램 및 게임 스캔 +applications=응용 프로그램 +Edit link=링크 수정 +Title=제목 +Link title=링크 제목 +Description=기술 +Link description=링크 설명 +Section=섹션 +The section this link belongs to=이 링크가 속한 섹션 +Icon=아이콘 +Select an icon for the link: $1=링크 아이콘을 선택하십시오 : $1 +Manual=설명서 +Select a graphic/textual manual or a readme=그래픽 / 텍스트 설명서 또는 추가 정보 선택 +Cpu clock frequency to set when launching this link=이 링크를 시작할 때 설정할 CPU 클럭 주파수 +Volume to set for this link=이 링크에 설정할 볼륨 +Parameters=매개 변수 +Parameters to pass to the application=응용 프로그램에 전달할 매개 변수 +Selector Directory=선택기 디렉토리 +Directory to scan for the selector=선택기를 검색 할 디렉토리 +Selector Browser=선택기 브라우저 +Allow the selector to change directory=선택기에서 디렉토리를 변경할 수 있도록 허용 +Selector Filter=선택기 필터 +Filter for the selector (Separate values with a comma)=선택기 필터 (쉼표로 값 구분) +Selector Screenshots=선택기 스크린 샷 +Directory of the screenshots for the selector=선택기의 스크린 샷 디렉토리 +Selector Aliases=선택자 별칭 +File containing a list of aliases for the selector=선택자에 대한 별칭 목록이 포함 된 파일 +Explicitly relaunch GMenu2X after this link's execution ends=이 링크 실행이 끝난 후 GMenu2X를 명시 적으로 다시 실행하십시오. +Don't Leave=떠나지 마라. +Don't quit GMenu2X when launching this link=이 링크를 실행할 때 GMenu2X를 종료하지 마십시오. +Save last selection=마지막 선택 저장 +Save the last selected link and section on exit=종료시 마지막으로 선택한 링크 및 섹션 저장 +Clock for GMenu2X=GMenu2X 용 시계 +Set the cpu working frequency when running GMenu2X=GMenu2X를 실행할 때 CPU 작동 주파수 설정 +Maximum overclock=최대 오버 클럭 +Set the maximum overclock for launching links=링크 실행을위한 최대 오버 클럭 설정 +Global Volume=글로벌 볼륨 +Set the default volume for the gp2x soundcard=gp2x 사운드 카드의 기본 볼륨 설정 +Output logs=출력 로그 +Logs the output of the links. Use the Log Viewer to read them.=링크 출력을 기록합니다. Log Viewer를 사용하여 읽을 수 있습니다. +Number of columns=열 수 +Set the number of columns of links to display on a page=페이지에 표시 할 링크 열 수 설정 +Number of rows=행 수 +Set the number of rows of links to display on a page=페이지에 표시 할 링크 행 수 설정 +Top Bar Color=탑 바 색상 +Color of the top bar=위쪽 막대의 색 +Bottom Bar Color=하단 바 색상 +Color of the bottom bar=하단 막대의 색상 +Selection Color=선택 색상 +Color of the selection and other interface details=선택 색상 및 기타 인터페이스 세부 정보 +You should disable Usb Networking to do this.=이를 위해서는 Usb 네트워킹을 비활성화해야합니다. +Operation not permitted.=작동이 허용되지 않습니다. +Language=언어 +Set the language used by GMenu2X=GMenu2X에서 사용하는 언어 설정 +Increase=증가 +Decrease=감소 +Change color component=색상 구성 요소 변경 +Increase value=볼륨 높이기 +Decrease value=볼륨 줄이기 +Switch=스위치 +Change value=값 변경 +Edit=편집 +Clear=클리어 +Select a directory=디렉토리 선택 +Select a file=파일 선택 +Clock (default: 200)=시계 (기본값 : 200) +Volume (default: -1)=볼륨 (기본값 : -1) +Wrapper=레퍼 +Enter folder=폴더 입력 +Confirm=확인 +Enter folder/Confirm=폴더 입력 / 확인 +Up one folder=한 폴더 위로 +Select an application=응용 프로그램 선택 +Space=공간 +Shift=시프트 +Cancel=취소 +OK=승인 +Backspace=백스페이스 +Skin=스킨 +Set the skin used by GMenu2X=GMenu2X에서 사용하는 스킨 설정 +Add link in $1=$1에 링크 추가 +Edit $1=$1 수정 +Delete $1 link=$1 링크 삭제 +Deleting $1=$1 삭제 중 +Are you sure?=확실합니까? +Insert a name for the new section=새 섹션의 이름 삽입 +Insert a new name for this section=이 섹션의 새 이름 삽입 +Yes=예 +No=아니오 +You will lose all the links in this section.=이 섹션의 모든 링크가 사라집니다. +Exit=나가기 +Link Scanner=링크 스캐너 +Scanning SD filesystem...=SD 파일 시스템 스캔 중 ... +Scanning NAND filesystem...=NAND 파일 시스템 스캔 중 ... +$1 files found.=$1 개의 파일을 찾았습니다. +Creating links...=링크 생성 중 ... +$1 links created.=$1 링크가 생성되었습니다. +Version $1 (Build date: $2)=버전 $1 (빌드 날짜 : $2) +Log Viewer=로그 뷰어 +Displays last launched program's output=마지막으로 실행 된 프로그램의 출력을 표시합니다. +Do you want to delete the log file?=로그 파일을 삭제 하시겠습니까? +USB Enabled (SD)=USB 사용 (SD) +USB Enabled (Nand)=USB 사용 가능 (Nand) +Turn off=끄다 +Launching $1=$1 시작 +Change page=페이지 변경 +Page=페이지 +Scroll=스크롤 +Untitled=제목 없음 +Change GMenu2X wallpaper=변경 GMenu2X의 벽지 +Activate/deactivate tv-out=TV 출력 활성화 / 비활성화 +Select wallpaper=월페이퍼 선택 +Gamma=감마 +Set gp2x gamma value (default: 10)=gp2x 감마 값 설정 (기본값 : 10) +Tv-Out encoding=TV 출력 인코딩 +Encoding of the tv-out signal=TV 출력 신호 인코딩 +Tweak RAM Timings=RAM 타이밍 조정 +This usually speeds up the application at the cost of stability=이것은 대개 안정성을 희생시키면서 애플리케이션 속도를 높입니다. +Gamma (default: 0)=감마 (기본값 : 0) +Gamma value to set when launching this link=이 링크를 시작할 때 설정할 감마 값 +ON=온 +OFF=오프 +File Browser=파일 탐색기 +Directory Browser=디렉토리 브라우저 \ No newline at end of file diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Norwegian b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Norwegian new file mode 100644 index 0000000..ddb9c13 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Norwegian @@ -0,0 +1,118 @@ +Settings=Instillinger +Configure GMenu2X's options=Konfigurer GMenu2X's innstillinger +Activate Usb on SD=Aktiver USB på SD +Activate Usb on Nand=Aktiver USB på Nand +Info about GMenu2X=Info om GMenu2X +About=Om +Add section=Legg til avdeling +Rename section=Gi nytt navn på avdelning +Delete section=Slett avdelning +Scan for applications and games=Skann etter applikasjoner +applications=applikasjoner +Edit link=Rediger link +Title=Tittel +Link title=Linktittel +Description=Beskrivelse +Link description=Linkbeskrivelse +Section=Avdeling +The section this link belongs to=Avdelningen som denne linken tilhører +Icon=Ikon +Select an icon for the link: $1=Velg et ikon til linken: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Velg en grafisk/tekstbasert manual eller en readme +Cpu clock frequency to set when launching this link=CPU-klokk som skal settes når denne linken startes +Volume to set for this link=Volum som skal settes til denne linken +Parameters=Parametrer +Parameters to pass to the application=Parametrer som skal settes til applikasjonen +Selector Directory=Selectormappe +Directory to scan for the selector=Mappe som selector skal skanne +Selector Browser=Selectorutforsker +Allow the selector to change directory=Tillat selector å bytte mappe +Selector Filter=Selectorfilter +Filter for the selector (Separate values with a comma)=Filter for selector (separer verdiene med et komma) +Selector Screenshots=Screenshots for Selector +Directory of the screenshots for the selector=Mappe med screenshots til selector +Selector Aliases=Selectoralias +File containing a list of aliases for the selector=Fil som inneholder en liste med alias for selector +Explicitly relaunch GMenu2X after this link's execution ends=Tving GMenu2X i å starte når denne linken er ferdigkjørt +Don't Leave=Forlat ikke +Don't quit GMenu2X when launching this link=Ikke avslutt GMenu2X når denne linken kjøres +Save last selection=Husk siste markering +Save the last selected link and section on exit=Husk siste link og avdeling ved avslutning +Clock for GMenu2X=Klokkefrekvens for GMenu2X +Set the cpu working frequency when running GMenu2X=Sett CPU-frekvensen for GMenu2X +Maximum overclock=Maks overklokk +Set the maximum overclock for launching links=Sett maks overklokk for linker +Global Volume=Mastervolum +Set the default volume fo the gp2x soundcard=Sett overall mastervolum +Output logs=Skriv logg +Logs the output of the links. Use the Log Viewer to read them.=Skriver ut logg for linkene. Bruk loggleseren for å lese dem. +Number of columns=Antall spalter +Set the number of columns of links to display on a page=Velg antall spalter med linker som skal vises per side +Number of rows=Antall rader +Set the number of rows of links to display on a page=Velg antall rader med linker som skal vises per side +Top Bar Color=Øverste felts farge +Color of the top bar=Farge på det øverste feltet +Bottom Bar Color=Nederste felts farge +Color of the bottom bar=Färge på det nederste feltet +Selection Color=Markørfarge +Color of the selection and other interface details=Farge på markøren og andre deler av grensesnittet +You should disable Usb Networking to do this.=Du bør slå av USB-nettverket når du gjør dette. +Operation not permitted.=Utillat operasjon. +Language=Språk +Set the language used by GMenu2X=Still inn språk for GMenu2X +Increase=Øk +Decrease=Minsk +Change color component=Endre fargekomponent +Increase value=Øk verdi +Decrease value=Minsk verdi +Switch=Endre +Change value=Endre verdi +Edit=Rediger +Clear=Rens +Select a directory=Velg en mappe +Select a file=Velg en fil +Clock (default: 200)=Klokkefrekvens (standard: 200) +Volume (default: -1)=Volum (standard: -1) +Wrapper=Wrapper +Enter folder=Åpne mappe +Confirm=Bekreft +Enter folder/Confirm=Åpne mappe/Bekrefte +Up one folder=Opp en mappe +Select an application=Velg en applikasjon +Space=Mellomrom +Shift=Skift +Cancel=Avbryt +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Velg hvilket skin GMenu2X skal bruke +Add link in $1=Lag link i $1 +Edit $1=Rediger $1 +Delete $1 link=Slett $1 +Deleting $1=Sletter $1 +Are you sure?=Er du sikker? +Insert a name for the new section=Sett navn for den nye avdelningen +Insert a new name for this section=Sett navn for denne avdelningen +Yes=Ja +No=Nei +You will lose all the links in this section.=Du vil miste alle linkene i denne avdelingen. +Exit=Avslutt +Link Scanner=Linkskanner +Scanning SD filesystem...=Skanner igjennom minnekortet... +Scanning NAND filesystem...=Skanner igjennom NAND... +$1 files found.=$1 fil(er) funnet. +Creating links...=Lager linker... +$1 links created.=$1 link(er) er lagd. +Version $1 (Build date: $2)=Versjon $1 (Bygd den: $2) +Log Viewer=Loggleser +Displays last launched program's output=Vis utdata fra siste startede program +Do you want to delete the log file?=Vil du slette loggen? +USB Enabled (SD)=USB Aktivert (SD) +USB Enabled (Nand)=USB Aktivert (NAND) +Turn off=Slå av +Launching $1=Starter $1 +Change page=Bytt side +Page=Side +Scroll=Rull +Untitled=Uten navn diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Brazil) b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Brazil) new file mode 100644 index 0000000..5af1487 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Brazil) @@ -0,0 +1,119 @@ +Settings=Configurações +Configure GMenu2X's options=Configurar opções do GMenu2X +Activate Usb on SD=Ativar USB para SD +Activate Usb on Nand=Ativar USB para Nand +Info about GMenu2X=Informação sobre GMenu2X +About=Sobre +Add section=Adicionar Seção +Rename section=Renomear Seção +Delete section=Excluir Seção +Scan for applications and games=Procurar aplicativos e jogos +applications=aplicativos +Edit link=Editar Link +Title=Título +Link title=Título do Link +Description=Descrição +Link description=Descrição do Link +Section=Seção +The section this link belongs to=Seção a qual pertence este link +Icon=Ícone +Select an icon for the link: $1=Selecione um ícone para o link: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Selecionar um manual gráfico e/ou de texto +Cpu clock frequency to set when launching this link=Frequência de relógio do CPU ao executar este link +Volume to set for this link=Ajustar o volume para este link +Parameters=Parâmetros +Parameters to pass to the application=Adicionar parâmetros ao aplicativo +Selector Directory=Diretório do Seletor +Directory to scan for the selector=Diretório a explorar com o Seletor +Selector Browser=Navegar Diretórios +Allow the selector to change directory=Permitir ao Seletor que mude de diretório +Selector Filter=Filtro de Arquivos +Filter for the selector (Separate values with a comma)=Filtro do Seletor (Separar valores com virgulas) +Selector Screenshots=Capturas de tela +Directory of the screenshots for the selector=Diretório das capturas de tela do Seletor +Selector Aliases=Alias do Seletor +File containing a list of aliases for the selector=Arquivo que contém a lista de alias para o Seletor +Backdrop=Plano de fundo +Select an image backdrop=Escolha a imagem de fundo +Explicitly relaunch GMenu2X after this link's execution ends=Executar novamente o GMenu2x após término deste programa +Don't Leave=Não Sair +Don't quit GMenu2X when launching this link=Não sair do GMenu2X ao executar este link +Save last selection=Lembrar a última seleção +Save the last selected link and section on exit=Lembrar link e seção selecionadas ao sair +Clock for GMenu2X=Relógio no GMenu2X +Set the cpu working frequency when running GMenu2X=Define a frequência da CPU durante a execução do GMenu2X +Maximum overclock=Overclock máximo +Set the maximum overclock for launching links=Define o overclock máximo ao executar um link +Global Volume=Volume global +Set the default volume for the soundcard=Define o volume padrão +Output logs=Logs de Saída +Logs the output of the links. Use the Log Viewer to read them.=Registra a saída dos links. Usar o Leitor de Logs para consultar. +Number of columns=Número de colunas +Set the number of columns of links to display on a page=Define o número de colunas de links por página +Number of rows=Número de filas +Set the number of rows of links to display on a page=Define o número de linhas de links por página +Top Bar=Barra superior +Color of the top bar=Cor da barra superior +Bottom Bar=Barra inferior +Color of the bottom bar=Cor da barra inferior +Selection=Seleção +Color of the selection and other interface details=Cor da seleção e outros detalhes da interface +You should disable Usb Networking to do this.=A função de Rede via USB deve ser desativada para executar este comando. +Operation not permitted.=Operação não permitida. +Language=Idioma +Set the language used by GMenu2X=Define o idioma usado no GMenu2X +Increase=Aumentar +Decrease=Diminuir +Change color component=Alterar componente da cor +Increase value=Aumentar valor +Decrease value=Diminuir Valor +Switch=Trocar +Change value=Mudar Valor +Edit=Editar +Clear=Limpar +Select a directory=Selecionar diretório +Select a file=Selecionar arquivo +Clock (default: 200)=Frequência (padrão: 200) +Volume (default: -1)=Volume (padrão: -1) +Wrapper=Invólucro +Enter folder=Entrar na Pasta +Confirm=Confirmar +Enter folder/Confirm=Entrar na Pasta/Confirmar +Up one folder=Subir uma pasta +Select an application=Selecionar aplicativo +Space=Espaço +Shift=Shift +Cancel=Cancelar +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Selecionar a Skin a usar pelo GMenu2X +Add link=Adicionar link +Edit $1=Editar $1 +Delete $1=Excluir $1 +Deleting $1=Excluindo $1 +Are you sure?=Tem certeza? +Insert a name for the new section=Insira o nome da nova seção +Insert a new name for this section=Insira um novo nome para esta seção +Yes=Sim +No=Não +You will lose all the links in this section.=Todos os links desta seção serão removidos. +Exit=Sair +Scanning SD filesystem...=Buscando arquivos no SD... +Scanning NAND filesystem...=Buscando arquivos na NAND... +$1 files found.=$1 arquivos encontrados. +Creating links...=Criando Links... +$1 links created.=$1 Links criados. +Version $1 (Build date: $2)=Versão $1 (Data de compilação: $2) +Log Viewer=Leitor de Logs +Displays last launched program's output=Mostra a saída do último programa executado +Do you want to delete the log file?=Deseja excluir o arquivo de log? +USB Enabled (SD)=USB Activado (SD) +USB Enabled (Nand)=USB Activado (Nand) +Turn off=Desligar +Launching $1=Executando $1 +Change page=Mudar página +Page=Página +Scroll=Rolar +Untitled=Sem título diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Portugal) b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Portugal) new file mode 100644 index 0000000..70c9084 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Portuguese (Portugal) @@ -0,0 +1,118 @@ +Settings= Configurações +Configure GMenu2X's options=Configurar opções do GMenu2X +Activate Usb on SD=Activar USB para SD +Activate Usb on Nand=Activar USB para Nand +Info about GMenu2X=Informação sobre GMenu2X +About=Sobre +Add section=Adicionar Secção +Rename section= Renomear Secção +Delete section= Eliminar Secção +Scan for applications and games=Procurar aplicações e jogos +applications=aplicações +Edit link=Editar Link +Title=Título +Link title=Título do Link +Description=Descrição +Link description=Descrição do Link +Section=Secção +The section this link belongs to=A secção a que pertence este link +Icon=Ícone +Select an icon for the link: $1=Seleccionar um ícone para o link: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Seleccionar um manual gráfico e/ou de texto +Cpu clock frequency to set when launching this link=Frequência de relógio do CPU ao lançar este link +Volume to set for this link=Ajustar o volume para este link +Parameters=Parâmetros +Parameters to pass to the application=Parâmetros a enviar para a aplicação +Selector Directory=Directório do Selector +Directory to scan for the selector=Directório a explorar com o selector +Selector Browser=Explorador do selector +Allow the selector to change directory=Permitir ao selector que mude de directório +Selector Filter=Filtro do selector +Filter for the selector (Separate values with a comma)=Filtro do selector (Separar valores com virgulas) +Selector Screenshots=Capturas de ecrã do selector +Directory of the screenshots for the selector=Directório das capturas de ecrã do selector +Selector Aliases=Alias do selector +File containing a list of aliases for the selector=Arquivo que contém a lista de alias para o selector +Explicitly relaunch GMenu2X after this link's execution ends=Forçar relançamento do GMenu2x após fim da execução deste link +Don't Leave=Não Sair +Don't quit GMenu2X when launching this link=Não sair do GMenu2X ao lançar este link +Save last selection=Gravar a última selecção +Save the last selected link and section on exit=Recordar link e secção seleccionadas ao sair +Clock for GMenu2X=Relógio no GMenu2X +Set the cpu working frequency when running GMenu2X=Ajustar a frequência do CPU durante a execução do GMenu2X +Maximum overclock=Overclock máximo +Set the maximum overclock for launching links=Ajustar o overclock máximo ao lançar um link +Global Volume=Volume global +Set the default volume fo the gp2x soundcard=Ajustar o volume por defeito da gp2x +Output logs=Logs de Output +Logs the output of the links. Use the Log Viewer to read them.=Regista o output dos links. Usar o Leitor de Logs para consultar. +Number of columns=Número de colunas +Set the number of columns of links to display on a page=Ajustar o número de colunas (de links) por página +Number of rows=Número de filas +Set the number of rows of links to display on a page=Ajustar o número de filas (de links) por página +Top Bar Color=Cor da barra superior +Color of the top bar= Cor da barra superior +Bottom Bar Color= Cor da barra inferior +Color of the bottom bar= Cor da barra inferior +Selection Color=Cor da selecção +Color of the selection and other interface details=Cor da selecção e outros detalhes do interface +You should disable Usb Networking to do this.=Deve desactivar a função Networking por USB para executar este comando. +Operation not permitted.=Operação não permitida. +Language=Idioma +Set the language used by GMenu2X=Ajustar o idioma usado no GMenu2X +Increase=Aumentar +Decrease=Reduzir +Change color component=Alterar componente da cor +Increase value=Aumentar valor +Decrease value=Reduzir Valor +Switch=Mudar +Change value=Mudar Valor +Edit=Editar +Clear=Eliminar +Select a directory=Seleccionar directório +Select a file=Seleccionar ficheiro +Clock (default: 200)=Frequência (predefinido: 200) +Volume (default: -1)=Volume (predefinido: -1) +Wrapper=Invólucro +Enter folder=Entrar na Pasta +Confirm=Confirmar +Enter folder/Confirm=Entrar na Pasta/Confirmar +Up one folder=Subir uma pasta +Select an application=Seleccionar uma aplicação +Space=Espaço +Shift=Shift +Cancel=Cancelar +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Seleccionar a Skin a usar pelo GMenu2X +Add link in $1=Adicionar link em $1 +Edit $1=Modificar $1 +Delete $1 link=Eliminar o link $1 +Deleting $1=Removendo $1 +Are you sure?=Tem a certeza? +Insert a name for the new section=Insira o nome da nova secção +Insert a new name for this section=Insira um novo nome para esta secção +Yes=Sim +No=Não +You will lose all the links in this section.=Perderá todos os links desta secção. +Exit=Sair +Link Scanner=Pesquisador de links +Scanning SD filesystem...=A pesquisar o filesystem do SD... +Scanning NAND filesystem...=A pesquisar o filesystem do NAND... +$1 files found.=$1 ficheiros encontrados. +Creating links...=A criar Links... +$1 links created.=$1 Links criados. +Version $1 (Build date: $2)=Versão $1 (Data de compilação: $2) +Log Viewer=Leitor de Logs +Displays last launched program's output=Mostra o output do último programa executado +Do you want to delete the log file?=Deseja eliminar o ficheiro de log? +USB Enabled (SD)=USB Activado (SD) +USB Enabled (Nand)=USB Activado (Nand) +Turn off=Desligar +Launching $1= A Executar $1 +Change page=Mudar página +Page=Página +Scroll=Scroll +Untitled=Sem título diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Russian b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Russian new file mode 100644 index 0000000..d53bd4f --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Russian @@ -0,0 +1,132 @@ +Settings=Настройки +Configure GMenu2X's options=Изменить настройки GMenu2X +Activate Usb on SD=Активировать SD через USB +Activate Usb on Nand=Активировать NAND через USB +Info about GMenu2X=Информация о GMenu2X +About=Информация +Add section=Добавить секцию +Rename section=Переименовать секцию +Delete section=Удалить секцию +Scan for applications and games= Поиск игр и приложений +applications=Приложения +Edit link=Изменить ссылку +Title=Заголовок +Link title=Имя ссылки +Description=Описание +Link description=Описание ссылки +Section=Секция +The section this link belongs to=Секция, которой принадлежит ссылка +Icon=Иконка +Select an icon for the link: $1=Выберите иконку к ссылке: $1 +Manual=Инструкция +Select a graphic/textual manual or a readme=Выберите текстовую инструкцию +Cpu clock frequency to set when launching this link=Частота CPU при запуске данной ссылки +Volume to set for this link=Громкость установленная для этой ссылки +Parameters=Параметры +Parameters to pass to the application=Параметры для передачи приложению +Selector Directory=Папка проводника +Directory to scan for the selector=Папка для сканирования проводником +Selector Browser=Выбрать браузером +Allow the selector to change directory=Разрешить проводнику выбирать папку +Selector Filter=Выбрать фильтр +Filter for the selector (Separate values with a comma)=Фильтр для проводника +Selector Screenshots=Обзор скриншотов +Directory of the screenshots for the selector=Папка с скриншотами для проводника +Selector Aliases=Обзор списков с именами +File containing a list of aliases for the selector=Файл, содержащий список имён-псевдонимов +Explicitly relaunch GMenu2X after this link's execution ends=Перезапуск GMenu2X после завершения выполнения ссылки +Don't Leave=Не покидать +Don't quit GMenu2X when launching this link=Не выключать Gmenu2X когда запускается эта ссылка +Save last selection=Сохранять последней выбор +Save the last selected link and section on exit=Сохранение последней выбранной ссылки и секции при выключение +Clock for GMenu2X=Частота CPU для Gmenu2X +Set the cpu working frequency when running GMenu2X=Устанавливает частоту CPU пока запущен Gmenu2X +Maximum overclock=Максимальная частота CPU +Set the maximum overclock for launching links=Максимальная частота CPU для запуска ссылок +Global Volume=Громкость +Set the default volume for the gp2x soundcard=Устанавливает громкость для звуковой карты GP2X +Output logs=Отчёты +Logs the output of the links. Use the Log Viewer to read them.=Создавать отчёты ссылок +Number of columns=Количество столбцов +Set the number of columns of links to display on a page=Установите количество столбцов для отображения на странице +Number of rows=Количество колонок +Set the number of rows of links to display on a page=Установите количество колонок для отображения на странице +Top Bar Color=Цвет панели сверху +Color of the top bar=Выберите цвет панели сверху +Bottom Bar Color=Цвет панели внизу +Color of the bottom bar= Выберите цвет панели внизу +Selection Color=Цвет панели выбора +Color of the selection and other interface details=Выберите цвет панели выбора +You should disable Usb Networking to do this.=Вы должны выключить USB Networking чтобы сделать это. +Operation not permitted.=Операция не разрешена. +Language=Язык +Set the language used by GMenu2X=Выберите язык интерфейса +Increase=Прибавить +Decrease=Убавить +Change color component=Изменить компонент цвета +Increase value=Увеличить значение +Decrease value=Уменьшить значение +Switch=Переключить +Change value=Изменить значение +Edit=Изменить +Clear=Очистить +Select a directory=Выбрать папку +Select a file=Выбрать файл +Clock (default: 200)=Частота CPU (Стандарт: 200) +Volume (default: -1)=Громкость (Стандарт: -1) +Wrapper=Перезапуск +Enter folder=Открыть папку +Confirm=Подтвердить +Enter folder/Confirm=Выбрать папку/подтвердить +Up one folder=Назад на одну папку +Select an application=Выберите приложение +Space=Пробел +Shift=Shift +Cancel=Выход +OK=OK +Backspace=Стереть +Skin=Скин +Set the skin used by GMenu2X=Выберите скин для GMenu2X +Add link in $1=Добавить ссылку в $1 +Edit $1=Именить $1 +Delete $1 link=Удалить ссылку $1 +Deleting $1=Удаление $1 +Are you sure?=Вы уверены? +Insert a name for the new section=Впишите имя для новой секции +Insert a new name for this section=Впишите новое имя для этой секции +Yes=Да +No=Нет +You will lose all the links in this section.=Вы потеряете все ссылки в этой секции +Exit=Выход +Link Scanner=Поиск ссылок +Scanning SD filesystem...=Сканирование SD... +Scanning NAND filesystem...=Сканирование NAND... +$1 files found.=$1 Файлов найдено. +Creating links...=Создание ссылок... +$1 links created.=$1 ссылок создано. +Version $1 (Build date: $2)=Версия $1 (Дата сборки: $2) +Log Viewer=Отчёты +Displays last launched program's output=Отображает последний запущенный отчёт программы +Do you want to delete the log file?=Вы хотите удалить этот отчёт? +USB Enabled (SD)=USB включен (SD) +USB Enabled (Nand)=USB включен (Nand) +Turn off=Выключить +Launching $1=Запуск $1... +Change page=Изменить страницу +Page=Страница +Scroll=Прокрутка +Untitled=Не названный + + +Change GMenu2X wallpaper=Изменить обои Gmenu2X +Activate/deactivate tv-out=Активировать/дезактивировать ТВ-выход +Select wallpaper=Выберите обои +Gamma=Гамма +Set gp2x gamma value (default: 10)=Значение гаммы экрана GP2X (стандарт: 10) +Tv-Out encoding=Технология вывода на ТВ +Encoding of the tv-out signal=Шифровка ТВ сигнала +Tweak RAM Timings=Изменение параметров RAM +This usually speeds up the application at the cost of stability=Это обычно убыстряет приложение +Gamma (default: 0)=Гамма (стандарт: 0) +Gamma value to set when launching this link=Значение гаммы при запуске этой ссылки + diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Slovak b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Slovak new file mode 100644 index 0000000..80da006 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Slovak @@ -0,0 +1,137 @@ +Settings=Nastavenia +Configure GMenu2X's options=Nastaviť voľby pre GMenu2X +Activate Usb on SD=Aktivovať USB pre SD kartu +Activate Usb on Nand=Aktivovať USB pre pamäť Nand +Info about GMenu2X=Informácie o GMenu2X +About=O programe +Add section=Pridať sekciu +Rename section=Premenovať sekciu +Delete section=Vymazať sekciu +Scan for applications and games= Hľadať aplikácie a hry +applications=aplikácie +Edit link=Upraviť odkaz +Title=Názov +Link title=Názov odkazu +Description=Popis +Link description=Popis pre odkaz +Section=Sekcia +The section this link belongs to=Sekcia, do ktorej patrí tento odkaz +Icon=Ikona +Select an icon for the link=Vyberte ikonu pre tento odkaz +Manual=Návod +Select a graphic/textual manual or a readme=Vyberte grafický/textový návod alebo readme +Cpu clock frequency to set when launching this link=Taktovacia frekvencia procesora, s ktorou bude spustený odkaz +Volume to set for this link=Nastavenie hlasitosti pre tento odkaz +Parameters=Parametre +Parameters to pass to the application=Parametre, ktoré majú byť predané aplikácii +Selector Directory=Adresár selektora +Directory to scan for the selector=Adresár, v ktorom má byť hľadaný selektor +Selector Browser=Prehliadač selektora +Allow the selector to change directory=Povolí selektorovi zmeniť adresár +Selector Filter=Filter pre selektor +Filter for the selector (Separate values with a comma)=Filter pre selektor (hodnoty oddeľujte čiarkou) +Selector Screenshots=Snímky obrazovky selektora +Directory of the screenshots for the selector=Adresár so snímkami obrazovky selektora +Selector Aliases=Aliasy selektora +File containing a list of aliases for the selector=Súbor obsahujúci zoznam aliasov pre selektor +Explicitly relaunch GMenu2X after this link's execution ends=Explicitne opätovne spustiť GMenu2X po ukončení spustenia tohto odkazu +Don't Leave=Neopúšťať +Don't quit GMenu2X when launching this link=Neukončovať GMenu2X pri spúšťaní tohto odkazu +Save last selection=Ulož posledný výber +Save the last selected link and section on exit=Ulož naposledy vybraný odkaz a sekciu pri ukončení +Clock for GMenu2X=Takt. frekvencia pre GMenu2X +Set the cpu working frequency when running GMenu2X=Nastavte frekvenciu cpu počas behu GMenu2X +Maximum overclock=Maximálne pretaktovanie +Set the maximum overclock for launching links=Nastavte maximálne pretaktovanie pre spúšťanie odkazov +Global Volume=Globálna hlasitosť +Set the default volume for the gp2x soundcard=Nastavte východziu hlasitosť pre zvukovú kartu gp2x +Output logs=Výstupné logy +Logs the output of the links. Use the Log Viewer to read them.=Loguje výstup odkazov. Na prezretie použite Log Viewer. +Number of columns=Počet stĺpcov +Set the number of columns of links to display on a page=Nastavte počet stĺpcov pre odkazy zobrazené na stránke +Number of rows=Počet riadkov +Set the number of rows of links to display on a page=Počet riadkov odkazov zobrazených na stránke +Top Bar Color=Farba hornej lišty +Color of the top bar= Farba hornej lišty +Bottom Bar Color=Farba spodnej lišty +Color of the bottom bar=Farba spodnej lišty +Selection Color=Farba výberu +Color of the selection and other interface details=Farba výberu a iných detailov interfacu +You should disable Usb Networking to do this.=Pre vykonanie tejto operácie by ste mali deaktivovať Usb sieťovanie. +Operation not permitted.=Operácia nepovolená. +Language=Jazyk +Set the language used by GMenu2X=Nastavte jazyk pre GMenu2X +Increase=Zvýšiť +Decrease=Znížiť +Change color component=Zmeniť farebnú zložku +Increase value=Zvýšiť hodnotu +Decrease value=Znížiť hodnotu +Switch=Prepnúť +Change value=Zmeniť hodnotu +Edit=Upraviť +Clear=Vyčistiť +Select a directory=Vyberte adresár +Select a file=Vyberte súbor +Clock (default: 200)=Takt (štandardne: 200) +Volume (default: -1)=Hlasitosť (štandardne: -1) +Wrapper=Obaľovač +Enter folder=Zadajte priečinok +Confirm=Potvrdiť +Enter folder/Confirm=Zadajte priečinok/Potvrdiť +Up one folder=O jeden priečinok vyššie +Select an application=Vyberte aplikáciu +Space=Medzera +Shift=Shift +Cancel=Zrušiť +OK=OK +Backspace=Backspace +Skin=Skin +Set the skin used by GMenu2X=Nastavte skin pre GMenu2X +Add link in $1=Pridať odkaz do $1 +Edit $1=Upraviť $1 +Delete $1 link=Vymazať odkaz na $1 +Deleting $1=Mažem $1 +Are you sure?=Ste si istý? +Insert a name for the new section=Zadajte názov novej sekcie +Insert a new name for this section=Zadajte nový názov pre túto sekciu +Yes=Áno +No=Nie +You will lose all the links in this section.=Stratíte všetky odkazy v tejto sekcii. +Exit=Ukončiť +Link Scanner=Vyhľadávač odkazov +Scanning SD filesystem...=Prehľadávam súborový systém na SD karte... +Scanning NAND filesystem...=Prehľadávam súborový systém na pamäti NAND... +$1 files found.=$1 súbor(ov) nájdených. +Creating links...=Vytváram odkazy... +$1 links created.=$1 odkazov vytvorených. +Version $1 (Build date: $2)=Verzia $1 (dátum zostavenia: $2) +Log Viewer=Prehliadač log súborov +Displays last launched program's output=Zobrazuje výstup naposledy spusteného súboru +Do you want to delete the log file?=Želáte si vymazať log súbor? +USB Enabled (SD)=USB aktivované (SD) +USB Enabled (Nand)=USB aktivované (Nand) +Turn off=Vypnúť +Launching $1=Spúšťam $1 +Change page=Zmeniť stránku +Page=Stránka +Scroll=Skrolovať +Untitled=Bez mena +Change GMenu2X wallpaper=Zmeniť pozadie GMenu2X +Activate/deactivate tv-out=Aktivovať/deaktivovať výstup na TV +Select wallpaper=Vyberte pozadie +Gamma=Gamma +Set gp2x gamma value (default: 10)=Nastavte hodnotu gamma (implic: 10) +Tv-Out encoding=Kódovanie výstupu na TV +Encoding of the tv-out signal=Kódovanie televízneho signálu +Tweak RAM Timings=Upraviť časovanie RAM +This usually speeds up the application at the cost of stability=Toto nastavenie zvyčajne zrýchli aplikáciu na úkor stability +Gamma (default: 0)=Gamma (implic: 0) +Gamma value to set when launching this link=Hodnota gamma pri spúšťaní tohto odkazu +Wallpaper=Pozadie +Configure skin=Nastaviť skin +Message Box Color=Farba textového okna +Message Box Border Color=Farba okraja textového okna +Message Box Selection Color=Farba výberu textového okna +Background color of the message box=Farba pozadia textového okna +Border color of the message box=Farba okraja textového okna +Color of the selection of the message box=Farba výberu textového okna diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Spanish b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Spanish new file mode 100644 index 0000000..74f5a8e --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Spanish @@ -0,0 +1,129 @@ +Settings=Ajustes +Configure GMenu2X's options=Configura las opciones de GMenu2X +Activate Usb on SD=Activa USB para SD +Activate Usb on Nand=Activa USB para Nand +Info about GMenu2X=Información sobre GMenu2X +About=Información +Add section=Añadir sección +Rename section=Renombrar sección +Delete section=Eliminar sección +Scan for applications and games=Buscar aplicaciones y juegos +applications=aplicaciones +Edit link=Editar enlace +Title=Título +Link title=Título de enlace +Description=Descripción +Link description=Descripción de enlace +Section=Sección +The section this link belongs to=La sección a la que pertenece el enlace +Icon=Icono +Select an icon for the link: $1=Seleccione un icono para el enlace: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Seleccione un manual gráfico/texto o un leeme +Cpu clock frequency to set when launching this link=Ajuste del reloj de la cpu para cargar este enlace +Volume to set for this link=Ajuste del volumen para este enlace +Parameters=Parámetros +Parameters to pass to the application=Parámetros que enviar a la aplicación +Selector Directory=Directorio del Selector +Directory to scan for the selector=Directorio que explorar con el selector +Selector Browser=Explorador del selector +Allow the selector to change directory=Permitir al selector cambiar de directorio +Selector Filter=Filtro del selector +Filter for the selector (Separate values with a comma)=Filtro para el selector (Separar valores con comas) +Selector Screenshots=Capturas de pantalla del selector +Directory of the screenshots for the selector=Directorio de capturas de pantalla para el selector +Selector Aliases=Alias del selector +File containing a list of aliases for the selector=Archivo contenedor de la lista de alias para el selector +Don't Leave=No salir +Don't quit GMenu2X when launching this link=No terminar GMenu2X al cargar este enlace +Save last selection=Recordar última selección +Save the last selected link and section on exit=Recordar última sección y enlace seleccionado al salir +Clock for GMenu2X=Reloj para GMenu2X +Set the cpu working frequency when running GMenu2X=Ajuste la frecuencia de trabajo de cpu al ejecutar GMenu2X +Maximum overclock=Overclock máximo +Set the maximum overclock for launching links=Ajuste el máximo overclock para cargar enlaces +Global Volume=Volumen global +Set the default volume for the gp2x soundcard=Ajuste el volumen por defecto del sonido en gp2x +Output logs=Archivos de registro de sucesos +Logs the output of the links. Use the Log Viewer to read them.=Registra los sucesos de los enlaces. Usa el lector de registros para leerlos. +Number of columns=Número de columnas +Set the number of columns of links to display on a page=Ajuste el número de columnas de enlaces que mostrar por página +Number of rows=Número de líneas +Set the number of rows of links to display on a page=Ajuste el número de líneas de enlaces que mostrar por página +Top Bar Color=Color de barra superior +Color of the top bar=Color de la barra superior +Bottom Bar Color=Color de barra inferior +Color of the bottom bar=Color de la barra inferior +Selection Color=Color de selección +Color of the selection and other interface details=Color de la selección y otros detalles del interfaz +You should disable Usb Networking to do this.=Debe desactivar Red por USB para hacer esto. +Operation not permitted.=Operación no permitida. +Language=Idioma +Set the language used by GMenu2X=Ajuste el idioma usado en GMenu2X +Increase=Aumentar +Decrease=Reducir +Change color component=Cambiar componente cromático +Increase value=Incrementar valor +Decrease value=Reducir valor +Switch=Cambiar +Change value=Cambiar valor +Edit=Modificar +Clear=Limpiar +Select a directory=Selecciona un directorio +Select a file=Selecciona un archivo +Clock (default: 200)=Frecuencia (predefinida: 200) +Volume (default: -1)=Volumen (predefinida: -1) +Enter folder=Entrar en carpeta +Confirm=Confirmar +Enter folder/Confirm=Entrar en carpeta/Confirmar +Up one folder=Subir una carpeta +Select an application=Seleccionar un programa +Space=Espacio +Shift=Mayúsculas +Cancel=Cancelar +OK=Aceptar +Backspace=Retroceso +Skin=Máscara +Set the skin used by GMenu2X=Seleccione la máscara usada en GMenu2X +Add link in $1=Añadir enlace en $1 +Edit $1=Modificar $1 +Delete $1 link=Eliminar el enlace de $1 +Deleting $1=Eliminando $1 +Are you sure?=¿Estás seguro? +Insert a name for the new section=Insertar nombre para la nueva sección +Insert a new name for this section=Insertar nuevo nombre para la sección +Yes=Si +No=No +You will lose all the links in this section.=Se perderán todos los enlaces de esta sección. +Exit=Salir +Link Scanner=Buscador de enlaces +Scanning SD filesystem...=Explorando sistema de archivos de SD... +Scanning NAND filesystem...=Explorando sistema de archivos de NAND... +$1 files found.=$1 archivo(s) encontrado(s). +Creating links...=Creando enlaces... +$1 links created.=$1 enlace(s) creado(s). +Version $1 (Build date: $2)=Versión $1 (Compilación: $2) +Log Viewer=Lector de Registro de Sucesos +Displays last launched program's output=Muestra la salida del último programa ejecutado +Do you want to delete the log file?=¿Desea eliminar el archivo de registro de sucesos? +USB Enabled (SD)=USB Activado (SD) +USB Enabled (Nand)=USB Activado (Nand) +Turn off=Desactivar +Launching $1=Ejecutando $1 +Change page=Cambiar página +Page=Página +Scroll=Desplazamiento +Untitled=Sin titulo +Change GMenu2X wallpaper=Cambia el fondo de GMenu2X +Activate/deactivate tv-out=Activa/desactiva salida de tv +Select wallpaper=Selecciona imagen de fondo +Gamma=Gama +Set gp2x gamma value (default: 10)=Ajusta el valor gama de la gp2x (predefinido: 10) +Tv-Out encoding=Codificación de salida de tv +Encoding of the tv-out signal=Codificación de la señal de salida de tv +Tweak RAM Timings=Modifica la sincronización de RAM +This usually speeds up the application at the cost of stability=Normalmente acelera la aplicación a costa de la estabilidad +Gamma (default: 0)=Gama (predefinido: 0) +Gamma value to set when launching this link=Valor de gama que usar al lanzar este enlace +Wrapper=Retornar +Explicitly relaunch GMenu2X after this link's execution ends=Forzar recarga de GMenu2X al terminar ejecución del enlace diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Swedish b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Swedish new file mode 100644 index 0000000..5501464 --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Swedish @@ -0,0 +1,119 @@ +Settings=Inställningar +Configure GMenu2X's options=Konfigurera GMenu2X's inställningar +Activate Usb on SD=Aktivera Usb på SD +Activate Usb on Nand=Aktivera Usb på Nand +Info about GMenu2X=Info om GMenu2X +About=Om +Add section=Lägg till avdelning +Rename section=Byt namn på avdelning +Delete section=Ta bort avdelning +Scan for applications and games=Scanna efter applikationer +applications=applikationer +Edit link=Redigera länk +Title=Titel +Link title=Länktitel +Description=Beskrivning +Link description=Länkbeskrivning +Section=Avdelning +The section this link belongs to=Avdelningen som den här länken tillhör +Icon=Ikon +Select an icon for the link: $1=Välj en ikon till länken: $1 +Manual=Manual +Select a graphic/textual manual or a readme=Välj en grafisk/textbaserad manual eller en readme +Cpu clock frequency to set when launching this link=Cpu-klockfrekvens att ändra till när denna länk körs +Volume to set for this link=Volymen som skall sättas för den här länken +Parameters=Parametrar +Parameters to pass to the application=Parametrar som skall skickas till applikationen +Selector Directory=Selectorkatalog +Directory to scan for the selector=Katalog som selector skall scanna +Selector Browser=Selectorutforskare +Allow the selector to change directory=Tillåt selector att byta katalog +Selector Filter=Selectorfilter +Filter for the selector (Separate values with a comma)=Filter för selector (separera värdena med kommatecken) +Selector Screenshots=Skärmdumpar till Selector +Directory of the screenshots for the selector=Katalog med skärmdumpar till selector +Selector Aliases=Selectoralias +File containing a list of aliases for the selector=Fil som innehåller en lista med alias för selector +Explicitly relaunch GMenu2X after this link's execution ends=Tvinga GMenu2X att starta om när denna länk körts klart +Don't Leave=Lämna inte +Don't quit GMenu2X when launching this link=Avsluta inte GMenu2X när denna länk körs +Save last selection=Spara senaste markeringen +Save the last selected link and section on exit=Spara senaste länk och avdelning vid avslut +Clock for GMenu2X=Klockfrekvens för GMenu2X +Set the cpu working frequency when running GMenu2X=Ställ in cpu-klockfrekvensen för GMenu2X +Maximum overclock=Maximal överklockning +Set the maximum overclock for launching links=Sätt maximal överklockning vid start av länk +Global Volume=Global volym +Set the default volume fo the gp2x soundcard=Sätt förinställd volym på gp2x ljudkortet +Output logs=logg utskrift +Logs the output of the links. Use the Log Viewer to read them.=Skriver ut loggar för länkarna. Använd loggläsaren för att läsa dem. +Number of columns=Antal spalter +Set the number of columns of links to display on a page=Välj antal spalter med länkar som skall visas per sida +Number of rows=Antal rader +Set the number of rows of links to display on a page=Välj antal rader med länkar som skall visas per sida +Top Bar Color=Översta fältets färg +Color of the top bar=Färg på det översta fältet +Bottom Bar Color=Nedersta fältets färg +Color of the bottom bar=Färg på det nedersta fältet +Selection Color=Markörfärg +Color of the selection and other interface details=Färg på markören och andra delar av gränssnittet +You should disable Usb Networking to do this.=Du bör slå av usb-nätverket när du gör detta. +Operation not permitted.=Otillåten användning. +Language=Språk +Set the language used by GMenu2X=Ställ in språk för GMenu2X +Increase=Öka +Decrease=Minska +Change color component=Ändra färg komponent +Increase value=Öka värde +Decrease value=Minska värde +Switch=Ändra +Change value=Ändra värde +Edit=Redigera +Clear=Rensa +Select a directory=Välj en katalog +Select a file=Välj en fil +Clock (default: 200)=Klockfrekvens (förinställd: 200) +Volume (default: -1)=Volym (förinställd: -1) +Wrapper=Wrapper +Enter folder=Öppna katalog +Confirm=Bekräfta +Enter folder/Confirm=Öppna katalog/Bekräfta +Up one folder=Upp en katalog +Select an application=Välj en applikation +Space=Mellanslag +Shift=Skift +Cancel=Avbryt +OK=OK +Backspace=Backsteg +Skin=Skin +Set the skin used by GMenu2X=Välj vilket skin GMenu2X ska använda +Add link in $1=Skapa länk i $1 +Edit $1=Redigera $1 +Delete $1 link=Ta bort $1 +Deleting $1=Tar bort $1 +Are you sure?=Är du säker? +Insert a name for the new section=Skriv in namn för den nya avdelningen +Insert a new name for this section=Skriv in namn för den här avdelningen +Yes=Ja +No=Nej +You will lose all the links in this section.=Du kommer att förlora alla länkar i den här +avdelningen. +Exit=Avsluta +Link Scanner=Länkscanner +Scanning SD filesystem...=Scannar SD filsystem... +Scanning NAND filesystem...=Scannar NAND filsystem... +$1 files found.=$1 fil(er) funna. +Creating links...=Skapar länkar... +$1 links created.=$1 länk(ar) skapade. +Version $1 (Build date: $2)=Version $1 (Skapad den: $2) +Log Viewer=Logg läsare +Displays last launched program's output=Visa utdata från senast körda program +Do you want to delete the log file?=Vill du ta bort logg filen? +USB Enabled (SD)=USB Aktiverad (SD) +USB Enabled (Nand)=USB Aktiverad (Nand) +Turn off=Stäng av +Launching $1=Startar $1 +Change page=Byt sida +Page=Sida +Scroll=Rulla +Untitled=Obetitlad diff --git a/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Turkish b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Turkish new file mode 100644 index 0000000..7554ced --- /dev/null +++ b/FunKey/board/funkey/rootfs-overlay/usr/share/gmenu2x/translations/Turkish @@ -0,0 +1,133 @@ +Settings=Ayarlar +Configure GMenu2X's options=GMenu2X'in ayarlarini degistir +Activate Usb on SD=SD Karti için USB baglantisini aktive et +Activate Usb on Nand=Nand Bellegi için USB baglantisini aktive et +Info about GMenu2X=GMenu2X hakkinda bilgi +About=Hakkinda +Add section=Bölüm ekle +Rename section=Bölümü yeniden adlandir +Delete section=Bölümü sil +Scan for applications and games=Oyun ve program taramasi +applications=programlar +Edit link=Link'i düzenle +Title=Baslik +Link title=Link basligi +Description=Açiklama +Link description=Link açiklamasi +Section=Bölüm +The section this link belongs to=Bu linkin ait oldugu bölüm +Icon=Icon +Select an icon for the link: $1=$1 Linki için bir simge seçin: +Manual=Kullanim kilavuzu +Select a graphic/textual manual or a readme=Grafik/Text bir kullanim kilavuzu veya readme dosyasi seçin +Cpu clock frequency to set when launching this link=Bu linki baslatirken kullanilacak CPU hizi +Volume to set for this link=Bu linkin ses seviyesi +Parameters=Parametreler +Parameters to pass to the application=Programa baslatirken aktarilacak parametreler +Selector Directory=Seçim klasörü +Directory to scan for the selector=Seçim için taranacak klasör +Selector Browser=Dosya seçimi +Allow the selector to change directory=Dosyas seçiminin klasör degistirmesine izin ver +Selector Filter=Seçim filtresi +Filter for the selector (Separate values with a comma)=Seçim için filtre (Verileri virgül ile ayirin) +Selector Screenshots=Seçim Ekran görüntüleri +Directory of the screenshots for the selector=Seçim için ekran görüntülerinin klasörü +Selector Aliases=Seçim için alternatif isimler (alias) +File containing a list of aliases for the selector=Seçim için alternatif isimleri içeren dosya adi +Explicitly relaunch GMenu2X after this link's execution ends=GMenu2X'i bu linkin çalismasi bittiginde özellikle tekrar baslat +Don't Leave=Çikma +Don't quit GMenu2X when launching this link=Bu linki baslatirken GMenu2X'i durdurma +Save last selection=Son seçimi sakla +Save the last selected link and section on exit=Son seçilen link ve seçimi çikista sakla +Clock for GMenu2X=GMenu2X için CPU hizi +Set the cpu working frequency when running GMenu2X=GMenu2X'in çalistigi CPU hizi +Maximum overclock=Maksimum overclock hizi +Set the maximum overclock for launching links=Linkler baslatilirken kullanilacak en yüksek CPU hizini belirleyin +Global Volume=Genel ses seviyesi +Set the default volume for the gp2x soundcard=GP2X ses kartinin standart ses seviyesini belirleyin +Output logs=Çikti kayitlari +Logs the output of the links. Use the Log Viewer to read them.=Linklerin çiktilarini kaydeder. Kayit gösterici ile okuyabilirsiniz. +Number of columns=Sütun sayisi +Set the number of columns of links to display on a page=Bir sayfada gösterilecek sütun sayisini belirleyin +Number of rows=Satir sayisi +Set the number of rows of links to display on a page=Bir sayfada gösterilecek satir sayisini belirleyin +Top Bar Color=Baslik çubugunun rengi +Color of the top bar=Baslik çubugunun rengini ve saydamligini belirler +Bottom Bar Color=Statü çubugunun rengi +Color of the bottom bar=Statü çubugunun rengini ve saydamligini belirler +Selection Color=Seçim rengi +Color of the selection and other interface details=Seçim rengi ve baska arabirim detaylarinin rengi +You should disable Usb Networking to do this.=Bunu yapmadan önce USB-Ag destegini kapatmalisiniz. +Operation not permitted.=Isleme izin verilmedi. +Language=Dil +Set the language used by GMenu2X=GMenu2X'in kullanacagi dili seçin +Increase=Artir +Decrease=Azalt +Change color component=Renk ögesini degistirin +Increase value=Degeri artir +Decrease value=Degeri azalt +Switch=Degistir +Change value=Degeri degistir +Edit=Düzenle +Clear=Temizle +Select a directory=Bir klasör seç +Select a file=Bir dosya seç +Clock (default: 200)=Hiz (Standart: 200) +Volume (default: -1)=Ses seviyesi (Standart: -1) +Wrapper=GMenu2X'i Tekrar baslat +Enter folder=Klasöre gir +Confirm=Onayla +Enter folder/Confirm=Klasöre gir / Onayla +Up one folder=Bir klasör yukari +Select an application=Bir program seçin +Space=Bosluk +Shift=Shift +Cancel=Vazgeç +OK=OK +Backspace=Geri +Skin=Skin +Set the skin used by GMenu2X=GMenu2X'in kullanacagi Skin'i seçin +Add link in $1=$1 bölümüne link ekle +Edit $1=$1 linkini düzenle +Delete $1 link=$1 linkini sil +Deleting $1=$1 linki silinecek +Are you sure?=Emin misiniz? +Insert a name for the new section=Yeni bölüm için bir isim girin +Insert a new name for this section=Bu bölüm için yeni isim girin +Yes=Evet +No=Hayir +You will lose all the links in this section.=Bu bölümdeki tüm linkleri kaybedeceksiniz. +Exit=Çikis +Link Scanner=Link taramasi +Scanning SD filesystem...=SD karti taraniyor... +Scanning NAND filesystem...=NAND bellegi taraniyor... +$1 files found.=$1 dosya bulundu. +Creating links...=Linkler yaratiliyor. +$1 links created.=$1 link yaratildi. +Version $1 (Build date: $2)=Version $1 - Derleme: $2 +Log Viewer=Kayit Gösterici +Displays last launched program's output=Son çalistirilan programin çiktisini gösterir +Do you want to delete the log file?=Kayit dosyasini silmek istiyor musunuz? +USB Enabled (SD)=USB aktive edildi (SD) +USB Enabled (Nand)=USB aktive edildi (Nand) +Turn off=Kapat +Launching $1=$1 baslatiliyor +Change page=Sayfa degistir +Page=Sayfa +Scroll=Kaydir +Untitled=Basliksiz +Change GMenu2X wallpaper=GMenu2X arka plan resmini degistir +Activate/deactivate tv-out=TV-Out aktive/deaktive et +Select wallpaper=Arka plan resmi seç +Gamma=Gamma +Set gp2x gamma value (default: 10)=gp2x gamma degerini degistir (Standart: 10) +Tv-Out encoding=TV-Out Sinyali +Encoding of the tv-out signal=TV-Out formati ayari +Tweak RAM Timings=Hafiza hizlandirmasi +This usually speeds up the application at the cost of stability=Programlari hizlandirir ancak stabilitelerini düsürür +Gamma (default: 0)=Gamma (Standart: 0) +Gamma value to set when launching this link=Bu linki çalistirirken kullanilacak gamma degeri +ON=Açik +OFF=Kapali +File Browser=Dosya seçici +Directory Browser=Klasör seçici diff --git a/FunKey/board/funkey/sw-description b/FunKey/board/funkey/sw-description index dd72e7c..d411479 100644 --- a/FunKey/board/funkey/sw-description +++ b/FunKey/board/funkey/sw-description @@ -1,6 +1,6 @@ software = { - version = "1.1.0"; + version = "2.0.0"; hardware-compatibility = [ "#RE:^Rev\.[D-E]$" ]; diff --git a/FunKey/board/funkey/update_partition b/FunKey/board/funkey/update_partition index bbbc887..82f46b1 100755 --- a/FunKey/board/funkey/update_partition +++ b/FunKey/board/funkey/update_partition @@ -9,20 +9,37 @@ root_mount=/tmp/rootfs do_preinst() { - notif " 1/3 EXTRACT FIRMWARE UPDATE..^DO NOT TURN OFF THE CONSOLE" + notif " 1/4 EXTRACT FIRMWARE UPDATE..^DO NOT TURN OFF THE CONSOLE" exit 0 } do_postinst() { - notif " 2/3 RESIZE ROOT FILESYSTEM^DO NOT TURN OFF THE CONSOLE" + notif " 2/4 RESIZE ROOT FILESYSTEM^DO NOT TURN OFF THE CONSOLE" resize2fs ${root_part} if [ $? -ne 0 ]; then - notif " CANNOT RESIZE ROOT ^FILESYSTEM" + notif " CANNOT RESIZE ROOT^FILESYSTEM" exit 1 fi + notif " 3/4 COPY OPKS TO USB MOUNT^DO NOT TURN OFF THE CONSOLE" + folder_opks_emulators=/mnt/Emulators + if [ ! -d "$folder_opks_emulators" ]; then + mkdir -p "$folder_opks_emulators" + mkdir -p ${root_mount} + mount -t ext4 ${root_part} ${root_mount} + if [ $? -ne 0 ]; then + notif "CANNOT MOUNT ROOT^FILESYSTEM" + exit 1 + fi + cp -f ${root_mount}/usr/games/opk/*.opk "$folder_opks_emulators" + umount ${root_mount} + if [ $? -ne 0 ]; then + notif "CANNOT UNMOUNT ROOT^FILESYSTEM" + exit 1 + fi + fi for file in $(ls /mnt/FunKey-*.fwu); do - notif " 3/3 ERASE UPDATE FILE^DO NOT TURN OFF THE CONSOLE" + notif " 4/4 ERASE UPDATE FILE^DO NOT TURN OFF THE CONSOLE" rm -f "${file}" done exit 0 diff --git a/FunKey/configs/funkey_defconfig b/FunKey/configs/funkey_defconfig index 49ed628..f35b796 100644 --- a/FunKey/configs/funkey_defconfig +++ b/FunKey/configs/funkey_defconfig @@ -1,10 +1,20 @@ BR2_arm=y BR2_cortex_a7=y +BR2_ARM_FPU_VFPV4=y BR2_DL_DIR="../download" BR2_CCACHE=y +BR2_OPTIMIZE_FAST=y +BR2_SHARED_STATIC_LIBS=y BR2_GLOBAL_PATCH_DIR="$(BR2_EXTERNAL_FUNKEY_PATH)/board/funkey/patches" BR2_TOOLCHAIN_EXTERNAL=y +BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y +BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y +BR2_TOOLCHAIN_EXTERNAL_URL="https://github.com/FunKey-Project/FunKey-OS/releases/download/FunKey-OS-2.0.0/FunKey-sdk-2.0.0.tar.gz" +BR2_TOOLCHAIN_EXTERNAL_HEADERS_4_14=y +BR2_TOOLCHAIN_EXTERNAL_CUSTOM_MUSL=y +BR2_TOOLCHAIN_EXTERNAL_CXX=y BR2_TOOLCHAIN_EXTERNAL_GDB_SERVER_COPY=y +BR2_TARGET_OPTIMIZATION="-fno-PIC -march=armv7-a+neon-vfpv4 -mtune=cortex-a7 -mfpu=neon-vfpv4 -mvectorize-with-neon-quad -D__ARM_NEON__" BR2_TARGET_GENERIC_HOSTNAME="FunKey" BR2_TARGET_GENERIC_ISSUE="Welcome to Buildroot for the FunKey" BR2_ROOTFS_DEVICE_TABLE="$(BR2_EXTERNAL_FUNKEY_PATH)/board/funkey/device_table.txt" @@ -42,6 +52,7 @@ BR2_PACKAGE_ALSA_UTILS_BAT=y BR2_PACKAGE_ALSA_UTILS_IECSET=y BR2_PACKAGE_ALSA_UTILS_SPEAKER_TEST=y BR2_PACKAGE_AUMIX=y +BR2_PACKAGE_MPG123=y BR2_PACKAGE_DOSFSTOOLS=y BR2_PACKAGE_DOSFSTOOLS_FATLABEL=y BR2_PACKAGE_DOSFSTOOLS_FSCK_FAT=y @@ -49,10 +60,11 @@ BR2_PACKAGE_DOSFSTOOLS_MKFS_FAT=y BR2_PACKAGE_E2FSPROGS=y # BR2_PACKAGE_E2FSPROGS_FSCK is not set BR2_PACKAGE_E2FSPROGS_RESIZE2FS=y -BR2_PACKAGE_SDL_GFX=y +BR2_PACKAGE_FBGRAB=y BR2_PACKAGE_SDL_IMAGE_GIF=y BR2_PACKAGE_SDL_IMAGE_JPEG=y BR2_PACKAGE_SDL_IMAGE_PNG=y +BR2_PACKAGE_SDL_NET=y BR2_PACKAGE_SDL_SOUND=y BR2_PACKAGE_SDL_SOUND_PLAYSOUND=y BR2_PACKAGE_GPTFDISK=y @@ -62,14 +74,25 @@ BR2_PACKAGE_PARTED=y BR2_PACKAGE_SYSSTAT=y BR2_PACKAGE_UBOOT_TOOLS=y BR2_PACKAGE_UBOOT_TOOLS_MKENVIMAGE=y +BR2_PACKAGE_LUA=y +BR2_PACKAGE_LIBSAMPLERATE=y +BR2_PACKAGE_LIBSNDFILE=y +BR2_PACKAGE_OPENAL=y +BR2_PACKAGE_TINYALSA=y +BR2_PACKAGE_TREMOR=y BR2_PACKAGE_LIBARCHIVE=y BR2_PACKAGE_LIBARCHIVE_BSDTAR=y BR2_PACKAGE_LIBARCHIVE_BSDCPIO=y BR2_PACKAGE_LIBARCHIVE_BSDCAT=y +BR2_PACKAGE_LIBCONFIG=y BR2_PACKAGE_LIBQRENCODE=y BR2_PACKAGE_LIBQRENCODE_TOOLS=y +BR2_PACKAGE_PIXMAN=y +BR2_PACKAGE_TINYXML2=y BR2_PACKAGE_LIBNL=y BR2_PACKAGE_LIBRSYNC=y +BR2_PACKAGE_FMT=y +BR2_PACKAGE_ICU=y BR2_PACKAGE_DHCPCD=y BR2_PACKAGE_DROPBEAR=y BR2_PACKAGE_PROCPS_NG=y @@ -80,9 +103,16 @@ BR2_PACKAGE_UTIL_LINUX_BINARIES=y BR2_PACKAGE_NANO=y BR2_TARGET_ROOTFS_EXT2=y BR2_TARGET_ROOTFS_EXT2_4=y -BR2_TARGET_ROOTFS_EXT2_SIZE="100M" +BR2_TARGET_ROOTFS_EXT2_SIZE="160M" # BR2_TARGET_ROOTFS_TAR is not set BR2_PACKAGE_HOST_DOSFSTOOLS=y BR2_PACKAGE_HOST_DTC=y BR2_PACKAGE_HOST_MTOOLS=y +BR2_PACKAGE_HOST_SQUASHFS=y BR2_PACKAGE_HOST_UBOOT_TOOLS=y +BR2_PACKAGE_LIBOPK=y +BR2_PACKAGE_LIBXDGMIME=y +BR2_PACKAGE_GMENU2X=y +BR2_PACKAGE_AGG=y +BR2_PACKAGE_FLUIDLITE=y +BR2_PACKAGE_LIBMIKMOD=y diff --git a/FunKey/package/FCEUX/FCEUX.mk b/FunKey/package/FCEUX/FCEUX.mk index 0522bf3..132e462 100644 --- a/FunKey/package/FCEUX/FCEUX.mk +++ b/FunKey/package/FCEUX/FCEUX.mk @@ -4,7 +4,7 @@ # ################################################################################ -FCEUX_VERSION = fceux-FunKey-1.00 +FCEUX_VERSION = fceux-FunKey-1.1.0 FCEUX_SITE_METHOD = git FCEUX_SITE = https://github.com/FunKey-Project/fceux.git FCEUX_LICENSE = GPL-2.0 @@ -12,25 +12,7 @@ FCEUX_LICENSE_FILES = COPYING FCEUX_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -FCEUX_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -FCEUX_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -FCEUX_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -FCEUX_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -FCEUX_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -FCEUX_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +FCEUX_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations FCEUX_CFLAGS += -ggdb -O3 FCEUX_CFLAGS += -Wno-write-strings -Wno-sign-compare @@ -65,5 +47,10 @@ define FCEUX_INSTALL_TARGET_CMDS $(INSTALL) -m 0755 $(@D)/fceux/fceux $(TARGET_DIR)/usr/games/fceux endef +define FCEUX_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(FCEUX_PKGDIR)/opk/nes $(TARGET_DIR)/usr/games/opk/nes_fceux_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +FCEUX_POST_INSTALL_TARGET_HOOKS += FCEUX_CREATE_OPK $(eval $(generic-package)) diff --git a/FunKey/package/FCEUX/opk/nes/nes.funkey-s.desktop b/FunKey/package/FCEUX/opk/nes/nes.funkey-s.desktop new file mode 100644 index 0000000..191ad6d --- /dev/null +++ b/FunKey/package/FCEUX/opk/nes/nes.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=NES +Comment=FC/NES Emulator +Icon=nes +Exec=/usr/games/launchers/nes_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/NES +SelectorFilter=nes,NES diff --git a/FunKey/package/FCEUX/opk/nes/nes.png b/FunKey/package/FCEUX/opk/nes/nes.png new file mode 100644 index 0000000..04ea8bb Binary files /dev/null and b/FunKey/package/FCEUX/opk/nes/nes.png differ diff --git a/FunKey/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk b/FunKey/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk index af98396..4e85222 100644 --- a/FunKey/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk +++ b/FunKey/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk @@ -4,7 +4,7 @@ # ################################################################################ -FUNKEY_GPIO_MAPPING_VERSION = FunKey-GPIO-Mapping-FunKey-1.00 +FUNKEY_GPIO_MAPPING_VERSION = FunKey-GPIO-Mapping-FunKey-1.1.0 FUNKEY_GPIO_MAPPING_SITE_METHOD = git FUNKEY_GPIO_MAPPING_SITE = https://github.com/FunKey-Project/FunKey-GPIO-Mapping.git FUNKEY_GPIO_MAPPING_SITE_LICENSE = GPL-2.1+ diff --git a/FunKey/package/PCSX-ReARMed/PCSX-ReARMed.mk b/FunKey/package/PCSX-ReARMed/PCSX-ReARMed.mk index 6bf4d61..46d70ae 100644 --- a/FunKey/package/PCSX-ReARMed/PCSX-ReARMed.mk +++ b/FunKey/package/PCSX-ReARMed/PCSX-ReARMed.mk @@ -4,7 +4,7 @@ # ################################################################################ -PCSX_REARMED_VERSION = pcsx_rearmed-FunKey-1.00 +PCSX_REARMED_VERSION = pcsx_rearmed-FunKey-1.1.0 PCSX_REARMED_SITE_METHOD = git PCSX_REARMED_SITE = https://github.com/FunKey-Project/pcsx_rearmed.git PCSX_REARMED_LICENSE = GPL-2.0 @@ -12,25 +12,10 @@ PCSX_REARMED_LICENSE_FILES = COPYING PCSX_REARMED_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -PCSX_REARMED_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -PCSX_REARMED_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -PCSX_REARMED_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -PCSX_REARMED_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -PCSX_REARMED_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif +PCSX_REARMED_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -PCSX_REARMED_CONF_OPTS += --enable-neon --gpu=neon -PCSX_REARMED_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad + PCSX_REARMED_CONF_OPTS += --enable-neon --gpu=neon endif PCSX_REARMED_CONF_OPTS += --sound-drivers=sdl @@ -62,4 +47,10 @@ define PCSX_REARMED_INSTALL_TARGET_CMDS $(INSTALL) -m 0755 $(@D)/pcsx $(TARGET_DIR)/usr/games/pcsx endef +define PCSX_REARMED_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(PCSX_REARMED_PKGDIR)/opk/ps1 $(TARGET_DIR)/usr/games/opk/ps1_pcsx_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +PCSX_REARMED_POST_INSTALL_TARGET_HOOKS += PCSX_REARMED_CREATE_OPK + $(eval $(generic-package)) diff --git a/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.funkey-s.desktop b/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.funkey-s.desktop new file mode 100644 index 0000000..c2d3861 --- /dev/null +++ b/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=PS1 +Comment=PS1 Emulator +Icon=ps1 +Exec=/usr/games/launchers/psone_launch_pcsx.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/PS1 +SelectorFilter=bin,BIN,cue,CUE,pbp,PBP diff --git a/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.png b/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.png new file mode 100644 index 0000000..66181c7 Binary files /dev/null and b/FunKey/package/PCSX-ReARMed/opk/ps1/ps1.png differ diff --git a/FunKey/package/PocketSNES/PocketSNES.mk b/FunKey/package/PocketSNES/PocketSNES.mk index 7753f9d..265d486 100644 --- a/FunKey/package/PocketSNES/PocketSNES.mk +++ b/FunKey/package/PocketSNES/PocketSNES.mk @@ -4,7 +4,7 @@ # ################################################################################ -POCKETSNES_VERSION = PocketSNES-FunKey-1.00 +POCKETSNES_VERSION = PocketSNES-FunKey-1.1.0 POCKETSNES_SITE_METHOD = git POCKETSNES_SITE = https://github.com/FunKey-Project/PocketSNES.git POCKETSNES_LICENSE = GPL-2.0 @@ -12,25 +12,7 @@ POCKETSNES_LICENSE_FILES = COPYING POCKETSNES_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -POCKETSNES_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -POCKETSNES_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -POCKETSNES_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -POCKETSNES_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -POCKETSNES_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -POCKETSNES_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +POCKETSNES_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations POCKETSNES_SDL_CFLAGS += $(shell $(STAGING_DIR)/usr/bin/sdl-config --cflags) POCKETSNES_SDL_LIBS += $(shell $(STAGING_DIR)/usr/bin/sdl-config --libs) @@ -71,4 +53,10 @@ define POCKETSNES_INSTALL_TARGET_CMDS $(INSTALL) -m 0755 $(@D)/psnes $(TARGET_DIR)/usr/games/psnes endef +define POCKETSNES_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(POCKETSNES_PKGDIR)/opk/snes $(TARGET_DIR)/usr/games/opk/snes_pocketsnes_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +POCKETSNES_POST_INSTALL_TARGET_HOOKS += POCKETSNES_CREATE_OPK + $(eval $(generic-package)) diff --git a/FunKey/package/PocketSNES/opk/snes/snes.funkey-s.desktop b/FunKey/package/PocketSNES/opk/snes/snes.funkey-s.desktop new file mode 100644 index 0000000..f7668dd --- /dev/null +++ b/FunKey/package/PocketSNES/opk/snes/snes.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=SNES +Comment=SNES Emulator +Icon=snes +Exec=/usr/games/launchers/snes_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/SNES +SelectorFilter=sfc,SFC,smc,SMC diff --git a/FunKey/package/PocketSNES/opk/snes/snes.png b/FunKey/package/PocketSNES/opk/snes/snes.png new file mode 100644 index 0000000..d3f327a Binary files /dev/null and b/FunKey/package/PocketSNES/opk/snes/snes.png differ diff --git a/FunKey/package/ProdScreens/ProdScreens.mk b/FunKey/package/ProdScreens/ProdScreens.mk index e14548f..0037a14 100644 --- a/FunKey/package/ProdScreens/ProdScreens.mk +++ b/FunKey/package/ProdScreens/ProdScreens.mk @@ -4,7 +4,7 @@ # ################################################################################ -PRODSCREENS_VERSION = FunKey-ProdScreens-FunKey-1.00 +PRODSCREENS_VERSION = FunKey-ProdScreens-FunKey-1.1.0 PRODSCREENS_SITE_METHOD = git PRODSCREENS_SITE = https://github.com/FunKey-Project/FunKey-ProdScreens.git PRODSCREENS_SITE_LICENSE = GPL-2.1+ diff --git a/FunKey/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch b/FunKey/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch new file mode 100644 index 0000000..eaf0467 --- /dev/null +++ b/FunKey/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch @@ -0,0 +1,81 @@ +From efd33aad5e69f36ab343b1f28839a55db4538104 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 10:55:37 +0100 +Subject: [PATCH 01/15] Fix non-terminating loop conditions when len=1 + +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; + } +--- + include/agg_renderer_outline_aa.h | 8 ++++---- + include/agg_renderer_outline_image.h | 4 ++-- + 2 files changed, 6 insertions(+), 6 deletions(-) + +diff --git a/include/agg_renderer_outline_aa.h b/include/agg_renderer_outline_aa.h +index ce25a2e..cb2aa00 100644 +--- a/include/agg_renderer_outline_aa.h ++++ b/include/agg_renderer_outline_aa.h +@@ -1659,7 +1659,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -1726,7 +1726,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +@@ -1798,7 +1798,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -1811,7 +1811,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +diff --git a/include/agg_renderer_outline_image.h b/include/agg_renderer_outline_image.h +index fbfac10..66d2b9a 100644 +--- a/include/agg_renderer_outline_image.h ++++ b/include/agg_renderer_outline_image.h +@@ -969,7 +969,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -982,7 +982,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0001-autogen.patch b/FunKey/package/agg/0001-autogen.patch new file mode 100644 index 0000000..b773f12 --- /dev/null +++ b/FunKey/package/agg/0001-autogen.patch @@ -0,0 +1,15 @@ +Author: Andrea Veri +Description: Disable configure's run from the autogen file. + +Index: agg-2.5+dfsg1/autogen.sh +=================================================================== +--- a/autogen.sh 2007-10-11 00:06:16.000000000 +0200 ++++ b/autogen.sh 2012-05-01 16:57:37.916862783 +0200 +@@ -18,6 +18,6 @@ + automake --foreign --add-missing --ignore-deps + + # and finally invoke our new configure +-./configure $* ++[ -n "$NOCONFIGURE" ] || ./configure $* + + # end diff --git a/FunKey/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch b/FunKey/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch new file mode 100644 index 0000000..4fe7434 --- /dev/null +++ b/FunKey/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch @@ -0,0 +1,40 @@ +From e269fe9b62af6fe314cebe0ee7a6d6d1a4a84d1c Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 11:03:26 +0100 +Subject: [PATCH 02/15] Cure recursion by aborting if the co-ordinates are to + big to handle + +--- + include/agg_rasterizer_cells_aa.h | 9 ++++++++- + 1 file changed, 8 insertions(+), 1 deletion(-) + +diff --git a/include/agg_rasterizer_cells_aa.h b/include/agg_rasterizer_cells_aa.h +index d3bb138..3a616d9 100644 +--- a/include/agg_rasterizer_cells_aa.h ++++ b/include/agg_rasterizer_cells_aa.h +@@ -40,7 +40,8 @@ + #define AGG_RASTERIZER_CELLS_AA_INCLUDED + + #include +-#include ++#include ++#include + #include "agg_math.h" + #include "agg_array.h" + +@@ -333,6 +334,12 @@ namespace agg + { + int cx = (x1 + x2) >> 1; + int cy = (y1 + y2) >> 1; ++ ++ // Bail if values are so large they are likely to wrap ++ if ((std::abs(x1) >= std::numeric_limits::max()/2) || (std::abs(y1) >= std::numeric_limits::max()/2) || ++ (std::abs(x2) >= std::numeric_limits::max()/2) || (std::abs(y2) >= std::numeric_limits::max()/2)) ++ return; ++ + line(x1, y1, cx, cy); + line(cx, cy, x2, y2); + } +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0002-caca.patch b/FunKey/package/agg/0002-caca.patch new file mode 100644 index 0000000..f98a573 --- /dev/null +++ b/FunKey/package/agg/0002-caca.patch @@ -0,0 +1,34 @@ +--- a/configure.in ++++ b/configure.in +@@ -39,7 +39,7 @@ + # used as platform library in examples: + # todo, make the PREFERED_PLATFORM selectable, after the set of possible + # Platforms to link the examples have been evaluated. +-PREFERED_PLATFORM=X11 ++PREFERED_PLATFORM=sdl + case "$host" in + *darwin* ) + OSX_LIBS="-framework Carbon -framework QuickTime" +@@ -120,9 +120,7 @@ + if test "$no_x" = "yes"; then + AC_MSG_WARN([*** X11 not found! Omitting X11 layer.]) + fi +-AM_CONDITIONAL(ENABLE_X11,[test x$no_x = x -a xno != x$enable_platform -a x$win32_host != xyes]) +-AC_SUBST(x_includes) +-AC_SUBST(x_libraries) ++AM_CONDITIONAL(ENABLE_X11,0) + dnl ############################################### + + dnl Settung up library version + +--- a/include/agg_renderer_outline_aa.h ++++ b/include/agg_renderer_outline_aa.h +@@ -1375,7 +1375,7 @@ + //--------------------------------------------------------------------- + void profile(const line_profile_aa& prof) { m_profile = &prof; } + const line_profile_aa& profile() const { return *m_profile; } +- line_profile_aa& profile() { return *m_profile; } ++// line_profile_aa& profile() { return *m_profile; } + + //--------------------------------------------------------------------- + int subpixel_width() const { return m_profile->subpixel_width(); } diff --git a/FunKey/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch b/FunKey/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch new file mode 100644 index 0000000..b12684d --- /dev/null +++ b/FunKey/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch @@ -0,0 +1,30 @@ +From 032d5342430f4c5dfbc34a2817d67386a14fd51b Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 11:40:49 +0100 +Subject: [PATCH 03/15] Get coordinates from previous vertex if last command is + path_cmd_end_poly + +--- + include/agg_path_storage.h | 6 ++++++ + 1 file changed, 6 insertions(+) + +diff --git a/include/agg_path_storage.h b/include/agg_path_storage.h +index 7be7393..8922fc8 100644 +--- a/include/agg_path_storage.h ++++ b/include/agg_path_storage.h +@@ -878,6 +878,12 @@ namespace agg + *x += x2; + *y += y2; + } ++ else if (!is_stop(m_vertices.last_command()) && ++ is_vertex(m_vertices.prev_vertex(&x2, &y2))) ++ { ++ *x += x2; ++ *y += y2; ++ } + } + } + +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch b/FunKey/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch new file mode 100644 index 0000000..0cecaf7 --- /dev/null +++ b/FunKey/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch @@ -0,0 +1,138 @@ +From b9c4b1c72b4ad6b24c37f402d3eec39ef393b0eb Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 14:17:43 +0100 +Subject: [PATCH 04/15] Make rasterizer_outline_aa ignore close_polygon when + vertex count < 3 + +--- + include/agg_rasterizer_outline_aa.h | 107 ++++++++++++++++++------------------ + 1 file changed, 52 insertions(+), 55 deletions(-) + +diff --git a/include/agg_rasterizer_outline_aa.h b/include/agg_rasterizer_outline_aa.h +index 4d6dd57..24301d5 100644 +--- a/include/agg_rasterizer_outline_aa.h ++++ b/include/agg_rasterizer_outline_aa.h +@@ -333,68 +333,65 @@ namespace agg + int y2; + int lprev; + +- if(close_polygon) ++ if(close_polygon && (m_src_vertices.size() >= 3)) + { +- if(m_src_vertices.size() >= 3) ++ dv.idx = 2; ++ ++ v = &m_src_vertices[m_src_vertices.size() - 1]; ++ x1 = v->x; ++ y1 = v->y; ++ lprev = v->len; ++ ++ v = &m_src_vertices[0]; ++ x2 = v->x; ++ y2 = v->y; ++ dv.lcurr = v->len; ++ line_parameters prev(x1, y1, x2, y2, lprev); ++ ++ v = &m_src_vertices[1]; ++ dv.x1 = v->x; ++ dv.y1 = v->y; ++ dv.lnext = v->len; ++ dv.curr = line_parameters(x2, y2, dv.x1, dv.y1, dv.lcurr); ++ ++ v = &m_src_vertices[dv.idx]; ++ dv.x2 = v->x; ++ dv.y2 = v->y; ++ dv.next = line_parameters(dv.x1, dv.y1, dv.x2, dv.y2, dv.lnext); ++ ++ dv.xb1 = 0; ++ dv.yb1 = 0; ++ dv.xb2 = 0; ++ dv.yb2 = 0; ++ ++ switch(m_line_join) + { +- dv.idx = 2; +- +- v = &m_src_vertices[m_src_vertices.size() - 1]; +- x1 = v->x; +- y1 = v->y; +- lprev = v->len; +- +- v = &m_src_vertices[0]; +- x2 = v->x; +- y2 = v->y; +- dv.lcurr = v->len; +- line_parameters prev(x1, y1, x2, y2, lprev); +- +- v = &m_src_vertices[1]; +- dv.x1 = v->x; +- dv.y1 = v->y; +- dv.lnext = v->len; +- dv.curr = line_parameters(x2, y2, dv.x1, dv.y1, dv.lcurr); +- +- v = &m_src_vertices[dv.idx]; +- dv.x2 = v->x; +- dv.y2 = v->y; +- dv.next = line_parameters(dv.x1, dv.y1, dv.x2, dv.y2, dv.lnext); +- +- dv.xb1 = 0; +- dv.yb1 = 0; +- dv.xb2 = 0; +- dv.yb2 = 0; +- +- switch(m_line_join) +- { +- case outline_no_join: +- dv.flags = 3; +- break; ++ case outline_no_join: ++ dv.flags = 3; ++ break; + +- case outline_miter_join: +- case outline_round_join: +- dv.flags = +- (prev.diagonal_quadrant() == dv.curr.diagonal_quadrant()) | +- ((dv.curr.diagonal_quadrant() == dv.next.diagonal_quadrant()) << 1); +- break; ++ case outline_miter_join: ++ case outline_round_join: ++ dv.flags = ++ (prev.diagonal_quadrant() == dv.curr.diagonal_quadrant()) | ++ ((dv.curr.diagonal_quadrant() == dv.next.diagonal_quadrant()) << 1); ++ break; + +- case outline_miter_accurate_join: +- dv.flags = 0; +- break; +- } ++ case outline_miter_accurate_join: ++ dv.flags = 0; ++ break; ++ } + +- if((dv.flags & 1) == 0 && m_line_join != outline_round_join) +- { +- bisectrix(prev, dv.curr, &dv.xb1, &dv.yb1); +- } ++ if((dv.flags & 1) == 0 && m_line_join != outline_round_join) ++ { ++ bisectrix(prev, dv.curr, &dv.xb1, &dv.yb1); ++ } + +- if((dv.flags & 2) == 0 && m_line_join != outline_round_join) +- { +- bisectrix(dv.curr, dv.next, &dv.xb2, &dv.yb2); +- } +- draw(dv, 0, m_src_vertices.size()); ++ if((dv.flags & 2) == 0 && m_line_join != outline_round_join) ++ { ++ bisectrix(dv.curr, dv.next, &dv.xb2, &dv.yb2); + } ++ draw(dv, 0, m_src_vertices.size()); + } + else + { +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0005-Remove-VC-6-workaround.patch b/FunKey/package/agg/0005-Remove-VC-6-workaround.patch new file mode 100644 index 0000000..f38f7c4 --- /dev/null +++ b/FunKey/package/agg/0005-Remove-VC-6-workaround.patch @@ -0,0 +1,52 @@ +From b8c43fb0ba13af0cc2b1050f48f81d76d2fdf0c7 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:04:05 +0100 +Subject: [PATCH 05/15] Remove VC++ 6 workaround + +--- + include/agg_renderer_scanline.h | 29 +---------------------------- + 1 file changed, 1 insertion(+), 28 deletions(-) + +diff --git a/include/agg_renderer_scanline.h b/include/agg_renderer_scanline.h +index c3bb6f0..c27ca60 100644 +--- a/include/agg_renderer_scanline.h ++++ b/include/agg_renderer_scanline.h +@@ -79,34 +79,7 @@ namespace agg + sl.reset(ras.min_x(), ras.max_x()); + while(ras.sweep_scanline(sl)) + { +- //render_scanline_aa_solid(sl, ren, ren_color); +- +- // This code is equivalent to the above call (copy/paste). +- // It's just a "manual" optimization for old compilers, +- // like Microsoft Visual C++ v6.0 +- //------------------------------- +- int y = sl.y(); +- unsigned num_spans = sl.num_spans(); +- typename Scanline::const_iterator span = sl.begin(); +- +- for(;;) +- { +- int x = span->x; +- if(span->len > 0) +- { +- ren.blend_solid_hspan(x, y, (unsigned)span->len, +- ren_color, +- span->covers); +- } +- else +- { +- ren.blend_hline(x, y, (unsigned)(x - span->len - 1), +- ren_color, +- *(span->covers)); +- } +- if(--num_spans == 0) break; +- ++span; +- } ++ render_scanline_aa_solid(sl, ren, ren_color); + } + } + } +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch b/FunKey/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch new file mode 100644 index 0000000..f1e465b --- /dev/null +++ b/FunKey/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch @@ -0,0 +1,85 @@ +From 9422570f4e099a834fc43619f7b2a7eb6b442e25 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:31:01 +0100 +Subject: [PATCH 06/15] Implement grain-merge blending mode (GIMP) + +--- + include/agg_pixfmt_rgba.h | 42 ++++++++++++++++++++++++++++++++++++++++-- + 1 file changed, 40 insertions(+), 2 deletions(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 79d10dc..f576ce4 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1401,9 +1401,46 @@ namespace agg + } + }; + ++ //================================================comp_op_rgba_grain_merge ++ template struct comp_op_rgba_grain_merge ++ { ++ typedef ColorT color_type; ++ typedef Order order_type; ++ typedef typename color_type::value_type value_type; ++ typedef typename color_type::calc_type calc_type; ++ typedef typename color_type::long_type long_type; ++ enum base_scale_e ++ { ++ base_shift = color_type::base_shift, ++ base_mask = color_type::base_mask ++ }; + ++ // E = I + M - 128 ++ static AGG_INLINE void blend_pix(value_type* p, ++ unsigned sr, unsigned sg, unsigned sb, ++ unsigned sa, unsigned cover) ++ { + +- ++ if(cover < 255) ++ { ++ sr = (sr * cover + 255) >> 8; ++ sg = (sg * cover + 255) >> 8; ++ sb = (sb * cover + 255) >> 8; ++ sa = (sa * cover + 255) >> 8; ++ } ++ if(sa) ++ { ++ calc_type da = p[Order::A]; ++ int dr = sr + p[Order::R] - 128; ++ int dg = sg + p[Order::G] - 128; ++ int db = sb + p[Order::B] - 128; ++ p[Order::R] = (value_type)(dr < 0 ? 0 : (dr > 255 ? 255 : dr)); ++ p[Order::G] = (value_type)(dg < 0 ? 0 : (dg > 255 ? 255 : dg)); ++ p[Order::B] = (value_type)(db < 0 ? 0 : (db > 255 ? 255 : db)); ++ p[Order::A] = (value_type)(sa + da - ((sa * da + base_mask) >> base_shift)); ++ } ++ } ++ }; + + //======================================================comp_op_table_rgba + template struct comp_op_table_rgba +@@ -1451,6 +1488,7 @@ namespace agg + comp_op_rgba_contrast ::blend_pix, + comp_op_rgba_invert ::blend_pix, + comp_op_rgba_invert_rgb ::blend_pix, ++ comp_op_rgba_grain_merge::blend_pix, + 0 + }; + +@@ -1486,6 +1524,7 @@ namespace agg + comp_op_contrast, //----comp_op_contrast + comp_op_invert, //----comp_op_invert + comp_op_invert_rgb, //----comp_op_invert_rgb ++ comp_op_grain_merge, //----comp_op_grain_merge + + end_of_comp_op_e + }; +@@ -2908,4 +2947,3 @@ namespace agg + } + + #endif +- +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch b/FunKey/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch new file mode 100644 index 0000000..cafb36e --- /dev/null +++ b/FunKey/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch @@ -0,0 +1,85 @@ +From abd440342e166a90d08610bf5b31d2a8357eafbe Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:43:18 +0100 +Subject: [PATCH 07/15] Implement grain-extract blending mode (GIMP) + +--- + include/agg_pixfmt_rgba.h | 48 +++++++++++++++++++++++++++++++++++++++++++++++ + 1 file changed, 48 insertions(+) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index f576ce4..42f0a05 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1442,6 +1442,52 @@ namespace agg + } + }; + ++ //==============================================comp_op_rgba_grain_extract ++ template struct comp_op_rgba_grain_extract ++ { ++ typedef ColorT color_type; ++ typedef Order order_type; ++ typedef typename color_type::value_type value_type; ++ typedef typename color_type::calc_type calc_type; ++ typedef typename color_type::long_type long_type; ++ enum base_scale_e ++ { ++ base_shift = color_type::base_shift, ++ base_mask = color_type::base_mask ++ }; ++ ++ // E = I - M + 128 ++ static AGG_INLINE void blend_pix(value_type* p, ++ unsigned sr, unsigned sg, unsigned sb, ++ unsigned sa, unsigned cover) ++ { ++ calc_type da = (p[Order::A] * sa + 255) >> 8; ++ ++ int dr = p[Order::R] - sr + 128; ++ int dg = p[Order::G] - sg + 128; ++ int db = p[Order::B] - sb + 128; ++ ++ dr = dr < 0 ? 0 : (dr > 255 ? 255 : dr); ++ dg = dg < 0 ? 0 : (dg > 255 ? 255 : dg); ++ db = db < 0 ? 0 : (db > 255 ? 255 : db); ++ ++ p[Order::A] = da; ++ ++ if(da < 255) ++ { ++ p[Order::R] = (dr * da + 255) >> 8; ++ p[Order::G] = (dg * da + 255) >> 8; ++ p[Order::B] = (db * da + 255) >> 8; ++ } ++ else ++ { ++ p[Order::R] = dr; ++ p[Order::G] = dg; ++ p[Order::B] = db; ++ } ++ } ++ }; ++ + //======================================================comp_op_table_rgba + template struct comp_op_table_rgba + { +@@ -1489,6 +1535,7 @@ namespace agg + comp_op_rgba_invert ::blend_pix, + comp_op_rgba_invert_rgb ::blend_pix, + comp_op_rgba_grain_merge::blend_pix, ++ comp_op_rgba_grain_extract::blend_pix, + 0 + }; + +@@ -1525,6 +1572,7 @@ namespace agg + comp_op_invert, //----comp_op_invert + comp_op_invert_rgb, //----comp_op_invert_rgb + comp_op_grain_merge, //----comp_op_grain_merge ++ comp_op_grain_extract, //----comp_op_grain_extract + + end_of_comp_op_e + }; +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch b/FunKey/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch new file mode 100644 index 0000000..0ed92ee --- /dev/null +++ b/FunKey/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch @@ -0,0 +1,36 @@ +From 2688af280836b95908d3cfd6915510d55de673b8 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:15:01 +0100 +Subject: [PATCH 08/15] Declare multiplication and division operators as const + +--- + include/agg_trans_affine.h | 8 ++++---- + 1 file changed, 4 insertions(+), 4 deletions(-) + +diff --git a/include/agg_trans_affine.h b/include/agg_trans_affine.h +index a662099..2f602a0 100644 +--- a/include/agg_trans_affine.h ++++ b/include/agg_trans_affine.h +@@ -216,15 +216,15 @@ namespace agg + } + + // Multiply the matrix by another one and return +- // the result in a separete matrix. +- trans_affine operator * (const trans_affine& m) ++ // the result in a separate matrix. ++ trans_affine operator * (const trans_affine& m) const + { + return trans_affine(*this).multiply(m); + } + + // Multiply the matrix by inverse of another one +- // and return the result in a separete matrix. +- trans_affine operator / (const trans_affine& m) ++ // and return the result in a separate matrix. ++ trans_affine operator / (const trans_affine& m) const + { + return trans_affine(*this).multiply_inv(m); + } +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0009-Add-a-static-identity-transformation.patch b/FunKey/package/agg/0009-Add-a-static-identity-transformation.patch new file mode 100644 index 0000000..01555cb --- /dev/null +++ b/FunKey/package/agg/0009-Add-a-static-identity-transformation.patch @@ -0,0 +1,37 @@ +From be9ed90897bc43b4547a3a1f8046827caaf13b4c Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:15:36 +0100 +Subject: [PATCH 09/15] Add a static identity transformation + +--- + include/agg_trans_affine.h | 1 + + src/agg_trans_affine.cpp | 1 + + 2 files changed, 2 insertions(+) + +diff --git a/include/agg_trans_affine.h b/include/agg_trans_affine.h +index 2f602a0..67fe5ca 100644 +--- a/include/agg_trans_affine.h ++++ b/include/agg_trans_affine.h +@@ -92,6 +92,7 @@ namespace agg + //---------------------------------------------------------------------- + struct trans_affine + { ++ static const trans_affine identity; + double sx, shy, shx, sy, tx, ty; + + //------------------------------------------ Construction +diff --git a/src/agg_trans_affine.cpp b/src/agg_trans_affine.cpp +index aca18c2..b3d9bc0 100644 +--- a/src/agg_trans_affine.cpp ++++ b/src/agg_trans_affine.cpp +@@ -28,6 +28,7 @@ + + namespace agg + { ++ const trans_affine trans_affine::identity; + + //------------------------------------------------------------------------ + const trans_affine& trans_affine::parl_to_parl(const double* src, +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0010-Add-renderer_scanline_aa_alpha.patch b/FunKey/package/agg/0010-Add-renderer_scanline_aa_alpha.patch new file mode 100644 index 0000000..b0be258 --- /dev/null +++ b/FunKey/package/agg/0010-Add-renderer_scanline_aa_alpha.patch @@ -0,0 +1,193 @@ +From 749c8cd11e9e6f81e93ae5ce19258431722b6bdf Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:43:25 +0100 +Subject: [PATCH 10/15] Add renderer_scanline_aa_alpha + +--- + include/agg_pixfmt_rgba.h | 24 +++++++++++++- + include/agg_renderer_base.h | 28 ++++++++++++++++ + include/agg_renderer_scanline.h | 71 +++++++++++++++++++++++++++++++++++++++++ + 3 files changed, 122 insertions(+), 1 deletion(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 42f0a05..6c4bc37 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -2247,7 +2247,6 @@ namespace agg + } + + +- + //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, + unsigned len, +@@ -2751,6 +2750,29 @@ namespace agg + } + + //-------------------------------------------------------------------- ++ void blend_color_hspan_alpha(int x, int y, unsigned len, ++ const color_type* colors, ++ value_type alpha, ++ const int8u* covers, ++ int8u cover) ++ { ++ value_type* p = (value_type*)m_rbuf->row_ptr(x, y, len) + (x << 2); ++ do ++ { ++ blender_type::blend_pix(m_comp_op, ++ p, ++ (colors->r * alpha + 255) >> 8, ++ (colors->g * alpha + 255) >> 8, ++ (colors->b * alpha + 255) >> 8, ++ (colors->a * alpha + 255) >> 8, ++ covers ? *covers++ : cover); ++ p += 4; ++ ++colors; ++ } ++ while(--len); ++ } ++ ++ //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, unsigned len, + const color_type* colors, + const int8u* covers, +diff --git a/include/agg_renderer_base.h b/include/agg_renderer_base.h +index 1808944..25f07c3 100644 +--- a/include/agg_renderer_base.h ++++ b/include/agg_renderer_base.h +@@ -37,6 +37,7 @@ namespace agg + public: + typedef PixelFormat pixfmt_type; + typedef typename pixfmt_type::color_type color_type; ++ typedef typename pixfmt_type::color_type::value_type value_type; + typedef typename pixfmt_type::row_data row_data; + + //-------------------------------------------------------------------- +@@ -383,6 +384,33 @@ namespace agg + } + + //-------------------------------------------------------------------- ++ void blend_color_hspan_alpha(int x, int y, int len, ++ const color_type* colors, ++ value_type alpha, ++ const cover_type* covers, ++ cover_type cover = agg::cover_full) ++ { ++ if(y > ymax()) return; ++ if(y < ymin()) return; ++ ++ if(x < xmin()) ++ { ++ int d = xmin() - x; ++ len -= d; ++ if(len <= 0) return; ++ if(covers) covers += d; ++ colors += d; ++ x = xmin(); ++ } ++ if(x + len > xmax()) ++ { ++ len = xmax() - x + 1; ++ if(len <= 0) return; ++ } ++ m_ren->blend_color_hspan_alpha(x, y, len, colors, alpha, covers, cover); ++ } ++ ++ //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, int len, + const color_type* colors, + const cover_type* covers, +diff --git a/include/agg_renderer_scanline.h b/include/agg_renderer_scanline.h +index c27ca60..4fcb557 100644 +--- a/include/agg_renderer_scanline.h ++++ b/include/agg_renderer_scanline.h +@@ -156,6 +156,35 @@ namespace agg + } + } + ++ //================================================render_scanline_aa_alpha ++ template ++ void render_scanline_aa_alpha(const Scanline& sl, BaseRenderer& ren, ++ SpanAllocator& alloc, SpanGenerator& span_gen, ++ unsigned alpha) ++ { ++ int y = sl.y(); ++ ++ unsigned num_spans = sl.num_spans(); ++ typename Scanline::const_iterator span = sl.begin(); ++ for(;;) ++ { ++ int x = span->x; ++ int len = span->len; ++ const typename Scanline::cover_type* covers = span->covers; ++ ++ if(len < 0) len = -len; ++ typename BaseRenderer::color_type* colors = alloc.allocate(len); ++ span_gen.generate(colors, x, y, len); ++ ren.blend_color_hspan_alpha(x, y, len, colors, alpha, ++ (span->len < 0) ? 0 : covers, *covers); ++ ++ if(--num_spans == 0) break; ++ ++span; ++ } ++ } ++ ++ + //=====================================================render_scanlines_aa + template +@@ -216,8 +245,50 @@ namespace agg + }; + + ++ //==============================================renderer_scanline_aa_alpha ++ template ++ class renderer_scanline_aa_alpha ++ { ++ public: ++ typedef BaseRenderer base_ren_type; ++ typedef SpanAllocator alloc_type; ++ typedef SpanGenerator span_gen_type; + ++ //-------------------------------------------------------------------- ++ renderer_scanline_aa_alpha() : m_ren(0), m_alloc(0), m_span_gen(0), m_alpha(1.0) {} ++ renderer_scanline_aa_alpha(base_ren_type& ren, ++ alloc_type& alloc, ++ span_gen_type& span_gen, ++ unsigned alpha) : ++ m_ren(&ren), ++ m_alloc(&alloc), ++ m_span_gen(&span_gen), ++ m_alpha(alpha) ++ {} ++ void attach(base_ren_type& ren, ++ alloc_type& alloc, ++ span_gen_type& span_gen) ++ { ++ m_ren = &ren; ++ m_alloc = &alloc; ++ m_span_gen = &span_gen; ++ } + ++ //-------------------------------------------------------------------- ++ void prepare() { m_span_gen->prepare(); } ++ ++ //-------------------------------------------------------------------- ++ template void render(const Scanline& sl) ++ { ++ render_scanline_aa_alpha(sl, *m_ren, *m_alloc, *m_span_gen, m_alpha); ++ } ++ ++ private: ++ base_ren_type* m_ren; ++ alloc_type* m_alloc; ++ span_gen_type* m_span_gen; ++ unsigned m_alpha; ++ }; + + + //===============================================render_scanline_bin_solid +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch b/FunKey/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch new file mode 100644 index 0000000..2a0d198 --- /dev/null +++ b/FunKey/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch @@ -0,0 +1,58 @@ +From 0ec68d7f5695403eccac75025ba7f6f7ecf1814e Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:49:08 +0100 +Subject: [PATCH 11/15] Avoid division by zero in color-burn mode + +FIXME: re-work using latest math from http://www.w3.org/TR/SVGCompositing/ +--- + include/agg_pixfmt_rgba.h | 21 ++++++++++++++++++--- + 1 file changed, 18 insertions(+), 3 deletions(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 6c4bc37..5d6b511 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1027,6 +1027,21 @@ namespace agg + // Dca' = Sa.(Sca.Da + Dca.Sa - Sa.Da)/Sca + Sca.(1 - Da) + Dca.(1 - Sa) + // + // Da' = Sa + Da - Sa.Da ++ ++ ++ // http://www.w3.org/TR/SVGCompositing/ ++ // if Sca == 0 and Dca == Da ++ // Dca' = Sa × Da + Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Sa × Da + Dca × (1 - Sa) ++ // = Da = Dca ++ // otherwise if Sca == 0 ++ // Dca' = Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Dca × (1 - Sa) ++ // otherwise if Sca > 0 ++ // Dca' = Sa × Da - Sa × Da × min(1, (1 - Dca/Da) × Sa/Sca) + Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Sa × Da × (1 - min(1, (1 - Dca/Da) × Sa/Sca)) + Sca × (1 - Da) + Dca × (1 - Sa) ++ ++ // sa * da * (255 - std::min(255, (255 - p[0]/da)*(sa/(sc*sa)) + + static AGG_INLINE void blend_pix(value_type* p, + unsigned sr, unsigned sg, unsigned sb, + unsigned sa, unsigned cover) +@@ -1056,15 +1071,15 @@ namespace agg + + p[Order::R] = (value_type)(((srda + drsa <= sada) ? + sr * d1a + dr * s1a : +- sa * (srda + drsa - sada) / sr + sr * d1a + dr * s1a + base_mask) >> base_shift); ++ (sr > 0 ? sa * (srda + drsa - sada) / sr + sr * d1a + dr * s1a + base_mask : 0)) >> base_shift); + + p[Order::G] = (value_type)(((sgda + dgsa <= sada) ? + sg * d1a + dg * s1a : +- sa * (sgda + dgsa - sada) / sg + sg * d1a + dg * s1a + base_mask) >> base_shift); ++ (sg > 0 ? sa * (sgda + dgsa - sada) / sg + sg * d1a + dg * s1a + base_mask : 0)) >> base_shift); + + p[Order::B] = (value_type)(((sbda + dbsa <= sada) ? + sb * d1a + db * s1a : +- sa * (sbda + dbsa - sada) / sb + sb * d1a + db * s1a + base_mask) >> base_shift); ++ (sb > 0 ? sa * (sbda + dbsa - sada) / sb + sb * d1a + db * s1a + base_mask : 0)) >> base_shift); + + p[Order::A] = (value_type)(sa + da - ((sa * da + base_mask) >> base_shift)); + } +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch b/FunKey/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch new file mode 100644 index 0000000..b3e641e --- /dev/null +++ b/FunKey/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch @@ -0,0 +1,26 @@ +From bf0e0b71360cfbc690a29f4abe15d7b9b61b8479 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:11:54 +0100 +Subject: [PATCH 12/15] Avoid pixel artifacts when compositing + +Change src_over alpha to avoid pixel artifacts by reordering computations. +--- + include/agg_pixfmt_rgba.h | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 5d6b511..bb255cd 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -346,7 +346,7 @@ namespace agg + p[Order::R] = (value_type)(sr + ((p[Order::R] * s1a + base_mask) >> base_shift)); + p[Order::G] = (value_type)(sg + ((p[Order::G] * s1a + base_mask) >> base_shift)); + p[Order::B] = (value_type)(sb + ((p[Order::B] * s1a + base_mask) >> base_shift)); +- p[Order::A] = (value_type)(sa + p[Order::A] - ((sa * p[Order::A] + base_mask) >> base_shift)); ++ p[Order::A] = (value_type)(sa + ((p[Order::A] * s1a + base_mask) >> base_shift)); + } + }; + +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch b/FunKey/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch new file mode 100644 index 0000000..9deb904 --- /dev/null +++ b/FunKey/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch @@ -0,0 +1,93 @@ +From 6f1ab5f4b470bcf4e7e72aac6e2f7f6ee3e7b424 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:16:42 +0100 +Subject: [PATCH 13/15] Modify agg conv classes to allow access to the original + geometry type + +--- + include/agg_conv_adaptor_vcgen.h | 2 ++ + include/agg_conv_adaptor_vpgen.h | 1 + + include/agg_conv_clip_polygon.h | 1 + + include/agg_conv_clip_polyline.h | 1 + + include/agg_conv_smooth_poly1.h | 2 ++ + 5 files changed, 7 insertions(+) + +diff --git a/include/agg_conv_adaptor_vcgen.h b/include/agg_conv_adaptor_vcgen.h +index 7bd9b07..fef4579 100644 +--- a/include/agg_conv_adaptor_vcgen.h ++++ b/include/agg_conv_adaptor_vcgen.h +@@ -38,6 +38,7 @@ namespace agg + + void rewind(unsigned) {} + unsigned vertex(double*, double*) { return path_cmd_stop; } ++ unsigned type() const { return 0; } + }; + + +@@ -73,6 +74,7 @@ namespace agg + } + + unsigned vertex(double* x, double* y); ++ unsigned type() const { return m_source->type(); } + + private: + // Prohibit copying +diff --git a/include/agg_conv_adaptor_vpgen.h b/include/agg_conv_adaptor_vpgen.h +index dca9415..a39102d 100644 +--- a/include/agg_conv_adaptor_vpgen.h ++++ b/include/agg_conv_adaptor_vpgen.h +@@ -42,6 +42,7 @@ namespace agg + + void rewind(unsigned path_id); + unsigned vertex(double* x, double* y); ++ unsigned type() const { return m_source->type(); } + + private: + conv_adaptor_vpgen(const conv_adaptor_vpgen&); +diff --git a/include/agg_conv_clip_polygon.h b/include/agg_conv_clip_polygon.h +index 3c34590..e417a7d 100644 +--- a/include/agg_conv_clip_polygon.h ++++ b/include/agg_conv_clip_polygon.h +@@ -60,6 +60,7 @@ namespace agg + double y1() const { return base_type::vpgen().y1(); } + double x2() const { return base_type::vpgen().x2(); } + double y2() const { return base_type::vpgen().y2(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_clip_polygon(const conv_clip_polygon&); +diff --git a/include/agg_conv_clip_polyline.h b/include/agg_conv_clip_polyline.h +index d45067f..0de4b57 100644 +--- a/include/agg_conv_clip_polyline.h ++++ b/include/agg_conv_clip_polyline.h +@@ -60,6 +60,7 @@ namespace agg + double y1() const { return base_type::vpgen().y1(); } + double x2() const { return base_type::vpgen().x2(); } + double y2() const { return base_type::vpgen().y2(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_clip_polyline(const conv_clip_polyline&); +diff --git a/include/agg_conv_smooth_poly1.h b/include/agg_conv_smooth_poly1.h +index 15f7f8d..0956c4e 100644 +--- a/include/agg_conv_smooth_poly1.h ++++ b/include/agg_conv_smooth_poly1.h +@@ -48,6 +48,7 @@ namespace agg + + void smooth_value(double v) { base_type::generator().smooth_value(v); } + double smooth_value() const { return base_type::generator().smooth_value(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_smooth_poly1(const conv_smooth_poly1&); +@@ -70,6 +71,7 @@ namespace agg + + void smooth_value(double v) { m_smooth.generator().smooth_value(v); } + double smooth_value() const { return m_smooth.generator().smooth_value(); } ++ unsigned type() const { return m_smooth.type(); } + + private: + conv_smooth_poly1_curve(const conv_smooth_poly1_curve&); +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch b/FunKey/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch new file mode 100644 index 0000000..547b0d2 --- /dev/null +++ b/FunKey/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch @@ -0,0 +1,30 @@ +From 6433a64f4cd41e88499386b0b7c7ae05d30683b8 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:33:32 +0100 +Subject: [PATCH 14/15] Avoid potential zero division resulting in nan in + agg::gamma_linear + +--- + include/agg_gamma_functions.h | 6 +++++- + 1 file changed, 5 insertions(+), 1 deletion(-) + +diff --git a/include/agg_gamma_functions.h b/include/agg_gamma_functions.h +index fa38a45..beb0c04 100644 +--- a/include/agg_gamma_functions.h ++++ b/include/agg_gamma_functions.h +@@ -94,7 +94,11 @@ namespace agg + { + if(x < m_start) return 0.0; + if(x > m_end) return 1.0; +- return (x - m_start) / (m_end - m_start); ++ double delta = m_end - m_start; ++ // avoid nan from potential zero division ++ // https://github.com/mapnik/mapnik/issues/761 ++ if (delta <= 0.0) return 0.0; ++ return (x - m_start) / delta; + } + + private: +-- +1.8.1.4 + diff --git a/FunKey/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch b/FunKey/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch new file mode 100644 index 0000000..6214bd6 --- /dev/null +++ b/FunKey/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch @@ -0,0 +1,24 @@ +From ca818d4dcd428c5560fc3c341fbaf427a7485e32 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:34:37 +0100 +Subject: [PATCH 15/15] Ensure first value in the gamma table is always zero + +--- + include/agg_gamma_functions.h | 1 + + 1 file changed, 1 insertion(+) + +diff --git a/include/agg_gamma_functions.h b/include/agg_gamma_functions.h +index beb0c04..b8eda52 100644 +--- a/include/agg_gamma_functions.h ++++ b/include/agg_gamma_functions.h +@@ -49,6 +49,7 @@ namespace agg + + double operator() (double x) const + { ++ if (x == 0.0) return 0.0; + return pow(x, m_gamma); + } + +-- +1.8.1.4 + diff --git a/FunKey/package/agg/CREATE_FILES.patch b/FunKey/package/agg/CREATE_FILES.patch new file mode 100644 index 0000000..1a78125 --- /dev/null +++ b/FunKey/package/agg/CREATE_FILES.patch @@ -0,0 +1,14 @@ +--- a/README.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/README 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac + +--- a/NEWS.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/NEWS 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac + +--- a/AUTHORS.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/AUTHORS 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac diff --git a/FunKey/package/agg/Config.in b/FunKey/package/agg/Config.in new file mode 100644 index 0000000..a842098 --- /dev/null +++ b/FunKey/package/agg/Config.in @@ -0,0 +1,13 @@ +config BR2_PACKAGE_AGG + bool "agg" + depends on BR2_INSTALL_LIBSTDCPP + select BR2_PACKAGE_SDL + help + The Anti-Grain Geometry project. A High Quality 2D Graphics Rendering + Engine for C++. + We select the SDL backend by default. + + http://www.antigrain.com/index.html + +comment "agg needs a toolchain with C++ support" + depends on !BR2_INSTALL_LIBSTDCPP diff --git a/FunKey/package/agg/agg-2.4-depends.patch b/FunKey/package/agg/agg-2.4-depends.patch new file mode 100644 index 0000000..f5506e2 --- /dev/null +++ b/FunKey/package/agg/agg-2.4-depends.patch @@ -0,0 +1,36 @@ +--- agg-2.4.orig/font_freetype/Makefile.am 2005-10-18 11:45:40.000000000 +0100 ++++ agg-2.4/font_freetype/Makefile.am 2006-07-10 15:11:55.000000000 +0100 +@@ -4,8 +4,9 @@ + agginclude_HEADERS = agg_font_freetype.h + lib_LTLIBRARIES = libaggfontfreetype.la + +-libaggfontfreetype_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ @FREETYPE_LIBS@ ++libaggfontfreetype_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ + libaggfontfreetype_la_SOURCES = agg_font_freetype.cpp + libaggfontfreetype_la_CXXFLAGS = -I$(top_srcdir)/include @FREETYPE_CFLAGS@ ++libaggfontfreetype_la_LIBADD = ../src/libagg.la @FREETYPE_LIBS@ + endif + +--- agg-2.4.orig/src/platform/sdl/Makefile.am 2005-10-17 23:49:35.000000000 +0100 ++++ agg-2.4/src/platform/sdl/Makefile.am 2006-07-10 15:11:55.000000000 +0100 +@@ -5,6 +5,6 @@ + libaggplatformsdl_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ + libaggplatformsdl_la_SOURCES = agg_platform_support.cpp + libaggplatformsdl_la_CXXFLAGS = -I$(top_srcdir)/include @SDL_CFLAGS@ +-libaggplatformsdl_la_LIBADD = @SDL_LIBS@ ++libaggplatformsdl_la_LIBADD = ../../libagg.la @SDL_LIBS@ + endif + +--- agg-2.5.orig/src/platform/X11/Makefile.am 2006-12-11 00:59:45.000000000 +0000 ++++ agg-2.5/src/platform/X11/Makefile.am 2007-01-07 14:07:39.000000000 +0000 +@@ -1,8 +1,8 @@ + if ENABLE_X11 + lib_LTLIBRARIES = libaggplatformX11.la + +-libaggplatformX11_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ -L@x_libraries@ ++libaggplatformX11_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ @X_LDFLAGS@ + libaggplatformX11_la_SOURCES = agg_platform_support.cpp + libaggplatformX11_la_CXXFLAGS = -I$(top_srcdir)/include -I@x_includes@ +-libaggplatformX11_la_LIBADD = -lX11 ++libaggplatformX11_la_LIBADD = ../../libagg.la -lX11 + endif diff --git a/FunKey/package/agg/agg-2.5-autotools.patch b/FunKey/package/agg/agg-2.5-autotools.patch new file mode 100644 index 0000000..1272b65 --- /dev/null +++ b/FunKey/package/agg/agg-2.5-autotools.patch @@ -0,0 +1,11 @@ +--- a/configure.in 2013-02-22 09:30:00.000000000 -0600 ++++ b/configure.in 2013-02-22 09:30:49.030777571 -0600 +@@ -8,7 +8,7 @@ + AC_PROG_CC + AC_PROG_CXX + AC_ISC_POSIX +-AM_C_PROTOTYPES ++#AM_C_PROTOTYPES + if test "x$U" != "x"; then + AC_MSG_ERROR(Compiler not ANSI compliant) + fi diff --git a/FunKey/package/agg/agg-2.5-pkgconfig.patch b/FunKey/package/agg/agg-2.5-pkgconfig.patch new file mode 100644 index 0000000..a303bfb --- /dev/null +++ b/FunKey/package/agg/agg-2.5-pkgconfig.patch @@ -0,0 +1,10 @@ +--- agg-2.5/libagg.pc.in.orig 2007-01-07 13:58:28.000000000 +0000 ++++ agg-2.5/libagg.pc.in 2007-01-07 14:02:40.000000000 +0000 +@@ -6,5 +6,6 @@ + Name: libagg + Description: Anti Grain Geometry - A High Quality Rendering Engine for C++ + Version: @VERSION@ +-Libs: -L${libdir} -Wl,-rpath,${exec_prefix}/lib -lagg ++Requires.private: freetype2 ++Libs: -L${libdir} -lagg + Cflags: -I${includedir} diff --git a/FunKey/package/agg/agg.mk b/FunKey/package/agg/agg.mk new file mode 100644 index 0000000..ecf5749 --- /dev/null +++ b/FunKey/package/agg/agg.mk @@ -0,0 +1,32 @@ +############################################################################### +# +# agg +# +############################################################################### + +AGG_VERSION = 2.5 +AGG_SOURCE = agg-$(AGG_VERSION).tar.gz +AGG_SITE = https://ftp.osuosl.org/pub/blfs/8.0/a +AGG_LICENSE = GPLv3+ +AGG_LICENSE_FILES = COPYING +AGG_INSTALL_STAGING = YES +AGG_AUTORECONF = YES + +AGG_DEPENDENCIES = host-pkgconf sdl + +AGG_CONF_OPTS = \ + --with-sdl-prefix=$(STAGING_DIR)/usr \ + --disable-sdltest + +AGG_CONF_OPTS += \ + --with-x=NO \ + --disable-examples --disable-gpc + +ifeq ($(BR2_PACKAGE_FREETYPE),y) +AGG_DEPENDENCIES += freetype +AGG_CONF_OPTS += --enable-freetype +else +AGG_CONF_OPTS += --disable-freetype +endif + +$(eval $(autotools-package)) diff --git a/FunKey/package/fluidlite/0001-fluidlite.patch b/FunKey/package/fluidlite/0001-fluidlite.patch new file mode 100644 index 0000000..9dc01c4 --- /dev/null +++ b/FunKey/package/fluidlite/0001-fluidlite.patch @@ -0,0 +1,11 @@ +--- a/CMakeLists.txt ++++ b/CMakeLists.txt +@@ -125,7 +125,7 @@ + endif() + endif() + +-option(FLUIDLITE_BUILD_SHARED "Build shared library" TRUE) ++option(FLUIDLITE_BUILD_SHARED "Build shared library" FALSE) + if(FLUIDLITE_BUILD_SHARED) + add_library(${PROJECT_NAME} SHARED ${SOURCES}) + diff --git a/FunKey/package/fluidlite/Config.in b/FunKey/package/fluidlite/Config.in new file mode 100644 index 0000000..bdfb7e0 --- /dev/null +++ b/FunKey/package/fluidlite/Config.in @@ -0,0 +1,5 @@ +config BR2_PACKAGE_FLUIDLITE + bool "FluidLite" + depends on BR2_USE_MMU + help + FluidLite is a very light version of FluidSynth. diff --git a/FunKey/package/fluidlite/fluidsynth.hash b/FunKey/package/fluidlite/fluidsynth.hash new file mode 100644 index 0000000..d39b0e8 --- /dev/null +++ b/FunKey/package/fluidlite/fluidsynth.hash @@ -0,0 +1,3 @@ +# Locally computed +sha256 ef4d008f9fe2fa9a48135505d42dd7e8e9cc4d7494a4b13d6caa13adb5c61ff8 1.0.tar.gz +sha256 a5564e99fd7f49e1344152a8c5bc1d420d5f973b30e010946764db0b5b9e668c LICENSE diff --git a/FunKey/package/fluidlite/fluidsynth.mk b/FunKey/package/fluidlite/fluidsynth.mk new file mode 100644 index 0000000..5cd24b7 --- /dev/null +++ b/FunKey/package/fluidlite/fluidsynth.mk @@ -0,0 +1,19 @@ +################################################################################ +# +# FLUIDLITE +# +################################################################################ + +FLUIDLITE_VERSION = fdd05bad03cdb24d1f78b5fe3453842890c1b0e8 +FLUIDLITE_SITE = $(call github,gcw0,FluidLite,$(FLUIDLITE_VERSION)) +FLUIDLITE_LICENSE = LGPL-2.1+ +FLUIDLITE_LICENSE_FILES = LICENSE +FLUIDLITE_INSTALL_STAGING = YES +FLUIDLITE_DEPENDENCIES = + +# Disable the shared library for static only build +ifeq ($(BR2_STATIC_LIBS),y) +FLUIDLITE_CONF_OPTS += -DDFLUIDLITE_BUILD_SHARED=FALSE +endif + +$(eval $(cmake-package)) diff --git a/FunKey/package/fonts-droid/Config.in b/FunKey/package/fonts-droid/Config.in new file mode 100644 index 0000000..a964265 --- /dev/null +++ b/FunKey/package/fonts-droid/Config.in @@ -0,0 +1,4 @@ +config BR2_PACKAGE_FONTS_DROID + bool "fonts-droid" + help + Android (AOSP) "droid" family of fonts diff --git a/FunKey/package/fonts-droid/fonts-droid.mk b/FunKey/package/fonts-droid/fonts-droid.mk new file mode 100644 index 0000000..630f438 --- /dev/null +++ b/FunKey/package/fonts-droid/fonts-droid.mk @@ -0,0 +1,28 @@ +################################################################################ +# +# fonts-droid +# +################################################################################ + +FONTS_DROID_VERSION = 074990596701553b8b51ff22290453de522f0d15 +FONTS_DROID_SITE = https://android.googlesource.com/platform/frameworks/base/+archive/$(FONTS_DROID_VERSION)/data +FONTS_DROID_SOURCE = fonts.tar.gz +FONTS_DROID_LICENSE = Apache-2.0 + +FONTS_DROID_STRIP_COMPONENTS = 0 + +# We cannot verify the hash because googlesource.com produces an archive +# with a different hash on every request. +# +# This still issues a warning. +BR_NO_CHECK_HASH_FOR += $(FONTS_DROID_SOURCE) + +define FONTS_DROID_INSTALL_TARGET_CMDS + mkdir -p $(TARGET_DIR)/usr/share/fonts/droid/ + install -m 0644 $(@D)/NOTICE $(@D)/DroidSansFallback.ttf \ + $(TARGET_DIR)/usr/share/fonts/droid/ + install -m 0644 $(@D)/NOTICE $(@D)/DroidSansFallbackFull.ttf \ + $(TARGET_DIR)/usr/share/fonts/droid/ +endef + +$(eval $(generic-package)) diff --git a/FunKey/package/gmenu2x/Config.in b/FunKey/package/gmenu2x/Config.in new file mode 100644 index 0000000..833c465 --- /dev/null +++ b/FunKey/package/gmenu2x/Config.in @@ -0,0 +1,25 @@ +config BR2_PACKAGE_GMENU2X + bool "gmenu2x" + select BR2_PACKAGE_LIBPNG + select BR2_PACKAGE_SDL + select BR2_PACKAGE_SDL_TTF + select BR2_PACKAGE_SDL_GFX + select BR2_PACKAGE_DEJAVU + select BR2_PACKAGE_FONTS_DROID + help + Application menu for devices with a 320x240 screen and buttons. + + This is a forked version with cleanups, targeted at the Ben NanoNote + and the Dingoo A320/A330. + + http://projects.qi-hardware.com/index.php/p/gmenu2x + +if BR2_PACKAGE_GMENU2X + +config BR2_PACKAGE_GMENU2X_SHOW_CLOCK + bool "Show clock" + +config BR2_PACKAGE_GMENU2X_CPUFREQ + bool "Support CPU frequency control" + +endif diff --git a/FunKey/package/gmenu2x/gmenu2x.mk b/FunKey/package/gmenu2x/gmenu2x.mk new file mode 100644 index 0000000..f73e8c0 --- /dev/null +++ b/FunKey/package/gmenu2x/gmenu2x.mk @@ -0,0 +1,37 @@ +############################################################# +# +# gmenu2x +# +############################################################# +GMENU2X_VERSION = gmenu2x-FunKey-1.0.0 +GMENU2X_SITE_METHOD = git +GMENU2X_SITE = https://github.com/FunKey-Project/gmenu2x.git +GMENU2X_LICENSE = GPL-2.0 + +GMENU2X_DEPENDENCIES = sdl sdl_ttf sdl_gfx dejavu libpng fonts-droid + +GMENU2X_CONF_OPTS = -DBIND_CONSOLE=ON + +ifeq ($(BR2_PACKAGE_GMENU2X_SHOW_CLOCK),y) +GMENU2X_CONF_OPTS += -DCLOCK=ON +else +GMENU2X_CONF_OPTS += -DCLOCK=OFF +endif + +ifeq ($(BR2_PACKAGE_GMENU2X_CPUFREQ),y) +GMENU2X_CONF_OPTS += -DCPUFREQ=ON +else +GMENU2X_CONF_OPTS += -DCPUFREQ=OFF +endif + +GMENU2X_CONF_OPTS += -DSCREEN_WIDTH=240 -DSCREEN_HEIGHT=240 -DSCREEN_DEPTH=16 + +ifeq ($(BR2_PACKAGE_LIBOPK),y) +GMENU2X_DEPENDENCIES += libopk +endif + +ifeq ($(BR2_PACKAGE_LIBXDGMIME),y) +GMENU2X_DEPENDENCIES += libxdgmime +endif + +$(eval $(cmake-package)) diff --git a/FunKey/package/gnuboy/gnuboy.mk b/FunKey/package/gnuboy/gnuboy.mk index 830f056..dfc4e71 100644 --- a/FunKey/package/gnuboy/gnuboy.mk +++ b/FunKey/package/gnuboy/gnuboy.mk @@ -4,7 +4,7 @@ # ################################################################################ -GNUBOY_VERSION = gnuboy-FunKey-1.00 +GNUBOY_VERSION = gnuboy-FunKey-1.1.0 GNUBOY_SITE_METHOD = git GNUBOY_SITE = https://github.com/FunKey-Project/gnuboy.git GNUBOY_LICENSE = GPL-2.0 @@ -12,25 +12,7 @@ GNUBOY_LICENSE_FILES = COPYING GNUBOY_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -GNUBOY_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -GNUBOY_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -GNUBOY_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -GNUBOY_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -GNUBOY_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -GNUBOY_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +GNUBOY_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations GNUBOY_CFLAGS += -ggdb -O3 @@ -43,4 +25,11 @@ GNUBOY_CONF_OPTS += --without-fb \ GNUBOY_CONF_ENV += SDL_CONFIG="$(STAGING_DIR)/usr/bin/sdl-config" +define GNUBOY_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(GNUBOY_PKGDIR)/opk/gb $(TARGET_DIR)/usr/games/opk/gb_gnuboy_funkey-s.opk -all-root -noappend -no-exports -no-xattrs + $(HOST_DIR)/usr/bin/mksquashfs $(GNUBOY_PKGDIR)/opk/gbc $(TARGET_DIR)/usr/games/opk/gbc_gnuboy_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +GNUBOY_POST_INSTALL_TARGET_HOOKS += GNUBOY_CREATE_OPK + $(eval $(autotools-package)) diff --git a/FunKey/package/gnuboy/opk/gb/gb.funkey-s.desktop b/FunKey/package/gnuboy/opk/gb/gb.funkey-s.desktop new file mode 100644 index 0000000..a6e56d7 --- /dev/null +++ b/FunKey/package/gnuboy/opk/gb/gb.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=GameBoy +Comment=Game Boy Emulator +Icon=gb +Exec=/usr/games/launchers/gb_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Game Boy +SelectorFilter=gb,GB diff --git a/FunKey/package/gnuboy/opk/gb/gb.png b/FunKey/package/gnuboy/opk/gb/gb.png new file mode 100644 index 0000000..7b7d823 Binary files /dev/null and b/FunKey/package/gnuboy/opk/gb/gb.png differ diff --git a/FunKey/package/gnuboy/opk/gbc/gbc.funkey-s.desktop b/FunKey/package/gnuboy/opk/gbc/gbc.funkey-s.desktop new file mode 100644 index 0000000..30f3c65 --- /dev/null +++ b/FunKey/package/gnuboy/opk/gbc/gbc.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=GameBoy Col +Comment=Game Boy Color Emulator +Icon=gbc +Exec=/usr/games/launchers/gb_launch.sh %f +Terminal=false +X-Od-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Game Boy Color +SelectorFilter=gbc,GBC diff --git a/FunKey/package/gnuboy/opk/gbc/gbc.png b/FunKey/package/gnuboy/opk/gbc/gbc.png new file mode 100644 index 0000000..24eea74 Binary files /dev/null and b/FunKey/package/gnuboy/opk/gbc/gbc.png differ diff --git a/FunKey/package/gpsp/gpsp.mk b/FunKey/package/gpsp/gpsp.mk index 743d032..19e2721 100644 --- a/FunKey/package/gpsp/gpsp.mk +++ b/FunKey/package/gpsp/gpsp.mk @@ -4,7 +4,7 @@ # ################################################################################ -GPSP_VERSION = gpsp-FunKey-1.00 +GPSP_VERSION = gpsp-FunKey-1.1.0 GPSP_SITE_METHOD = git GPSP_SITE = https://github.com/FunKey-Project/gpsp.git GPSP_LICENSE = GPL-2.0 @@ -12,25 +12,7 @@ GPSP_LICENSE_FILES = COPYING.DOC GPSP_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -GPSP_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -GPSP_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -GPSP_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -GPSP_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -GPSP_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -GPSP_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +GPSP_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations GPSP_SDL_CFLAGS += $(shell $(STAGING_DIR)/usr/bin/sdl-config --cflags) GPSP_SDL_LIBS += $(shell $(STAGING_DIR)/usr/bin/sdl-config --libs) @@ -58,4 +40,10 @@ define GPSP_INSTALL_TARGET_CMDS $(INSTALL) -m 0644 $(@D)/game_config.txt $(TARGET_DIR)/usr/games/game_config.txt endef +define GPSP_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(GPSP_PKGDIR)/opk/gba $(TARGET_DIR)/usr/games/opk/gba_gpsp_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +GPSP_POST_INSTALL_TARGET_HOOKS += GPSP_CREATE_OPK + $(eval $(generic-package)) diff --git a/FunKey/package/gpsp/opk/gba/gba.funkey-s.desktop b/FunKey/package/gpsp/opk/gba/gba.funkey-s.desktop new file mode 100644 index 0000000..44ec461 --- /dev/null +++ b/FunKey/package/gpsp/opk/gba/gba.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=GameBoy Adv +Comment=Game Boy Advance Emulator +Icon=gba +Exec=/usr/games/launchers/gba_launch_gpsp.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Game Boy Advance +SelectorFilter=gba,GBA diff --git a/FunKey/package/gpsp/opk/gba/gba.png b/FunKey/package/gpsp/opk/gba/gba.png new file mode 100644 index 0000000..77d3eab Binary files /dev/null and b/FunKey/package/gpsp/opk/gba/gba.png differ diff --git a/FunKey/package/libini/Config.in b/FunKey/package/libini/Config.in new file mode 100644 index 0000000..a42967b --- /dev/null +++ b/FunKey/package/libini/Config.in @@ -0,0 +1,6 @@ +config BR2_PACKAGE_LIBINI + bool "libini" + help + Tiny library to help reading INI files. + + https://github.com/pcercuei/libini diff --git a/FunKey/package/libini/libini.mk b/FunKey/package/libini/libini.mk new file mode 100644 index 0000000..33c1736 --- /dev/null +++ b/FunKey/package/libini/libini.mk @@ -0,0 +1,13 @@ +############################################################# +# +# libini +# +############################################################# +LIBINI_VERSION = libini-FunKey-1.0.0 +LIBINI_SITE_METHOD = git +LIBINI_SITE = https://github.com/FunKey-Project/libini.git +LIBINI_LICENSE = LGPL-2.1 + +LIBINI_INSTALL_STAGING = YES + +$(eval $(cmake-package)) diff --git a/FunKey/package/libmikmod/Config.in b/FunKey/package/libmikmod/Config.in new file mode 100644 index 0000000..909cc00 --- /dev/null +++ b/FunKey/package/libmikmod/Config.in @@ -0,0 +1,7 @@ +config BR2_PACKAGE_LIBMIKMOD + bool "libmikmod" + help + Mikmod is a module player and library supporting many + tracker formats, including mod, s3m, it, and xm. + + http://mikmod.shlomifish.org/ diff --git a/FunKey/package/libmikmod/libmikmod.mk b/FunKey/package/libmikmod/libmikmod.mk new file mode 100644 index 0000000..cce5190 --- /dev/null +++ b/FunKey/package/libmikmod/libmikmod.mk @@ -0,0 +1,21 @@ +############################################################# +# +# libmikmod +# +############################################################# +LIBMIKMOD_VERSION:=3.3.11.1 +LIBMIKMOD_SITE:=http://sourceforge.net/projects/mikmod/files/libmikmod/$(LIBMIKMOD_VERSION) + +LIBMIKMOD_CONF_OPTS = --localstatedir=/var + +LIBMIKMOD_LIBTOOL_PATCH = NO +LIBMIKMOD_INSTALL_STAGING = YES + +LIBMIKMOD_CONFIG_SCRIPTS = libmikmod-config + +define LIBMIKMOD_REMOVE_LIBMIKMOD_CONFIG +mv $(TARGET_DIR)/usr/bin/libmikmod-config $(HOST_DIR)/bin/ +endef +LIBMIKMOD_POST_INSTALL_TARGET_HOOKS += LIBMIKMOD_REMOVE_LIBMIKMOD_CONFIG + +$(eval $(autotools-package)) diff --git a/FunKey/package/libopk/Config.in b/FunKey/package/libopk/Config.in new file mode 100644 index 0000000..db3bce9 --- /dev/null +++ b/FunKey/package/libopk/Config.in @@ -0,0 +1,8 @@ +config BR2_PACKAGE_LIBOPK + bool "libopk" + select BR2_PACKAGE_ZLIB + select BR2_PACKAGE_LIBINI + help + Library to handle OPK packages. + + https://github.com/pcercuei/libopk diff --git a/FunKey/package/libopk/libopk.mk b/FunKey/package/libopk/libopk.mk new file mode 100644 index 0000000..7f767a7 --- /dev/null +++ b/FunKey/package/libopk/libopk.mk @@ -0,0 +1,14 @@ +############################################################# +# +# libopk +# +############################################################# +LIBOPK_VERSION = libopk-FunKey-1.0.0 +LIBOPK_SITE_METHOD = git +LIBOPK_SITE = https://github.com/FunKey-Project/libopk.git + +LIBOPK_DEPENDENCIES = libini zlib + +LIBOPK_INSTALL_STAGING = YES + +$(eval $(cmake-package)) diff --git a/FunKey/package/libxdgmime/Config.in b/FunKey/package/libxdgmime/Config.in new file mode 100644 index 0000000..f1d4fc0 --- /dev/null +++ b/FunKey/package/libxdgmime/Config.in @@ -0,0 +1,12 @@ +config BR2_PACKAGE_LIBXDGMIME + bool "libxdgmime" + depends on BR2_USE_WCHAR # shared-mime-info + select BR2_PACKAGE_SHARED_MIME_INFO + help + Simple library that parses the proposed MIME spec + listed at http://freedesktop.org/. + + https://github.com/pcercuei/libxdgmime + +comment "libxdgmime requires a toolchain with WCHAR support" + depends on !BR2_USE_WCHAR diff --git a/FunKey/package/libxdgmime/libxdgmime.mk b/FunKey/package/libxdgmime/libxdgmime.mk new file mode 100644 index 0000000..509703d --- /dev/null +++ b/FunKey/package/libxdgmime/libxdgmime.mk @@ -0,0 +1,30 @@ +############################################################# +# +# libxdgmime +# +############################################################# +LIBXDGMIME_VERSION = libxdgmime-FunKey-1.0.0 +LIBXDGMIME_SITE_METHOD = git +LIBXDGMIME_SITE = https://github.com/FunKey-Project/libxdgmime.git +LIBXDGMIME_DEPENDENCIES = shared-mime-info +LIBXDGMIME_LICENCE = LGPL-2.1+ or AFL-2.1 + +LIBXDGMIME_INSTALL_STAGING = YES + +LIBXDGMIME_MAKE_ENV = CFLAGS="$(TARGET_CFLAGS)" LDFLAGS="$(TARGET_LDFLAGS)" \ + CROSS_COMPILE="$(TARGET_CROSS)" PREFIX=/usr \ + PLATFORM="$(BR2_VENDOR)" + +define LIBXDGMIME_BUILD_CMDS + $(LIBXDGMIME_MAKE_ENV) $(MAKE) -C $(@D) +endef + +define LIBXDGMIME_INSTALL_STAGING_CMDS + $(LIBXDGMIME_MAKE_ENV) DESTDIR="$(STAGING_DIR)" $(MAKE) -C $(@D) install +endef + +define LIBXDGMIME_INSTALL_TARGET_CMDS + $(LIBXDGMIME_MAKE_ENV) DESTDIR="$(TARGET_DIR)" $(MAKE) -C $(@D) install-lib +endef + +$(eval $(generic-package)) diff --git a/FunKey/package/mednafen/mednafen.mk b/FunKey/package/mednafen/mednafen.mk index 991044f..0ce2272 100644 --- a/FunKey/package/mednafen/mednafen.mk +++ b/FunKey/package/mednafen/mednafen.mk @@ -4,7 +4,7 @@ # ################################################################################ -MEDNAFEN_VERSION = mednafen-git-FunKey-1.00 +MEDNAFEN_VERSION = mednafen-git-FunKey-1.1.0 MEDNAFEN_SITE_METHOD = git MEDNAFEN_SITE = https://github.com/FunKey-Project/mednafen-git.git MEDNAFEN_LICENSE = GPL-2.0+ @@ -12,36 +12,31 @@ MEDNAFEN_LICENSE_FILES = COPYING MEDNAFEN_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -MEDNAFEN_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -MEDNAFEN_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -MEDNAFEN_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -MEDNAFEN_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -MEDNAFEN_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -MEDNAFEN_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +MEDNAFEN_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard #MEDNAFEN_AUTORECONF = YES MEDNAFEN_CFLAGS += -ggdb -O3 +MEDNAFEN_CFLAGS += -DFUNKEY_FAST_BLIT #MEDNAFEN_LDFLAGS += -lSDL_ttf -lSDL_image -MEDNAFEN_CONF_OPTS += CFLAGS="$(MEDNAFEN_CFLAGS)" +MEDNAFEN_CONF_OPTS += CXXFLAGS="$(MEDNAFEN_CFLAGS)" #MEDNAFEN_CONF_OPTS += LDFLAGS="$(MEDNAFEN_LDFLAGS)" MEDNAFEN_CONF_OPTS += --prefix=/usr/local --bindir=/usr/games --without-libsndfile +MEDNAFEN_CONF_OPTS += --disable-ss --disable-ssfplay --disable-fancy-scalers +#MEDNAFEN_CONF_OPTS += --disable-nes --disable-gba --disable-psx --disable-snes MEDNAFEN_CONF_ENV += SDL_CONFIG="$(STAGING_DIR)/usr/bin/sdl-config" +define MEDNAFEN_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(MEDNAFEN_PKGDIR)/opk/gamegear $(TARGET_DIR)/usr/games/opk/gamegear_mednafen_funkey-s.opk -all-root -noappend -no-exports -no-xattrs + $(HOST_DIR)/usr/bin/mksquashfs $(MEDNAFEN_PKGDIR)/opk/lynx $(TARGET_DIR)/usr/games/opk/lynx_mednafen_funkey-s.opk -all-root -noappend -no-exports -no-xattrs + $(HOST_DIR)/usr/bin/mksquashfs $(MEDNAFEN_PKGDIR)/opk/ngp $(TARGET_DIR)/usr/games/opk/ngp_mednafen_funkey-s.opk -all-root -noappend -no-exports -no-xattrs + $(HOST_DIR)/usr/bin/mksquashfs $(MEDNAFEN_PKGDIR)/opk/pce $(TARGET_DIR)/usr/games/opk/pce_mednaefn_funkey-s.opk -all-root -noappend -no-exports -no-xattrs + $(HOST_DIR)/usr/bin/mksquashfs $(MEDNAFEN_PKGDIR)/opk/wonderswan $(TARGET_DIR)/usr/games/opk/wonderswan_mednafen_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +MEDNAFEN_POST_INSTALL_TARGET_HOOKS += MEDNAFEN_CREATE_OPK + $(eval $(autotools-package)) diff --git a/FunKey/package/mednafen/opk/gamegear/gamegear.funkey-s.desktop b/FunKey/package/mednafen/opk/gamegear/gamegear.funkey-s.desktop new file mode 100644 index 0000000..bbc5a28 --- /dev/null +++ b/FunKey/package/mednafen/opk/gamegear/gamegear.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=Game Gear +Comment=Game Gear Emulator +Icon=gamegear +Exec=/usr/games/launchers/gamegear_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Game Gear +SelectorFilter=zip,ZIP,gg,GG diff --git a/FunKey/package/mednafen/opk/gamegear/gamegear.png b/FunKey/package/mednafen/opk/gamegear/gamegear.png new file mode 100644 index 0000000..c6a0f24 Binary files /dev/null and b/FunKey/package/mednafen/opk/gamegear/gamegear.png differ diff --git a/FunKey/package/mednafen/opk/lynx/lynx.funkey-s.desktop b/FunKey/package/mednafen/opk/lynx/lynx.funkey-s.desktop new file mode 100644 index 0000000..3216c4f --- /dev/null +++ b/FunKey/package/mednafen/opk/lynx/lynx.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=Lynx +Comment=Atari Lynx Emulator +Icon=lynx +Exec=/usr/games/launchers/lynx_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Atari lynx +SelectorFilter=zip,ZIP,lnx,LNX diff --git a/FunKey/package/mednafen/opk/lynx/lynx.png b/FunKey/package/mednafen/opk/lynx/lynx.png new file mode 100644 index 0000000..bb144e9 Binary files /dev/null and b/FunKey/package/mednafen/opk/lynx/lynx.png differ diff --git a/FunKey/package/mednafen/opk/ngp/ngp.funkey-s.desktop b/FunKey/package/mednafen/opk/ngp/ngp.funkey-s.desktop new file mode 100644 index 0000000..76365aa --- /dev/null +++ b/FunKey/package/mednafen/opk/ngp/ngp.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=Neo Geo +Comment=Neo Geo Pocket Emulator +Icon=ngp +Exec=/usr/games/launchers/ngp_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Neo Geo Pocket +SelectorFilter=zip,ZIP,ngp,NGP diff --git a/FunKey/package/mednafen/opk/ngp/ngp.png b/FunKey/package/mednafen/opk/ngp/ngp.png new file mode 100644 index 0000000..29d9c0e Binary files /dev/null and b/FunKey/package/mednafen/opk/ngp/ngp.png differ diff --git a/FunKey/package/mednafen/opk/pce/pce.funkey-s.desktop b/FunKey/package/mednafen/opk/pce/pce.funkey-s.desktop new file mode 100644 index 0000000..355dcd9 --- /dev/null +++ b/FunKey/package/mednafen/opk/pce/pce.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=PC Engine +Comment=PCE-TurboGrafx Emulator +Icon=pce +Exec=/usr/games/launchers/pce_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/PCE-TurboGrafx +SelectorFilter=zip,ZIP,pce,PCE,sgx,SGX diff --git a/FunKey/package/mednafen/opk/pce/pce.png b/FunKey/package/mednafen/opk/pce/pce.png new file mode 100644 index 0000000..e1a0f8c Binary files /dev/null and b/FunKey/package/mednafen/opk/pce/pce.png differ diff --git a/FunKey/package/mednafen/opk/wonderswan/wonderswan.funkey-s.desktop b/FunKey/package/mednafen/opk/wonderswan/wonderswan.funkey-s.desktop new file mode 100644 index 0000000..ec4280b --- /dev/null +++ b/FunKey/package/mednafen/opk/wonderswan/wonderswan.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=WonderSwan +Comment=WonderSwan/WonderSwan Color Emulator +Icon=wonderswan +Exec=/usr/games/launchers/wonderswan_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/WonderSwan +SelectorFilter=zip,ZIP,wsc,WSC diff --git a/FunKey/package/mednafen/opk/wonderswan/wonderswan.png b/FunKey/package/mednafen/opk/wonderswan/wonderswan.png new file mode 100644 index 0000000..265f1c0 Binary files /dev/null and b/FunKey/package/mednafen/opk/wonderswan/wonderswan.png differ diff --git a/FunKey/package/picodrive/opk/megadrive/megadrive.funkey-s.desktop b/FunKey/package/picodrive/opk/megadrive/megadrive.funkey-s.desktop new file mode 100644 index 0000000..0f2f544 --- /dev/null +++ b/FunKey/package/picodrive/opk/megadrive/megadrive.funkey-s.desktop @@ -0,0 +1,12 @@ +[Desktop Entry] +Type=Application +Name=MegaDrive +Comment=Sega Genesis/MegaDrive Emulator +Icon=megadrive +Exec=/usr/games/launchers/megadrive_launch.sh %f +Terminal=false +X-OD-NeedsDownscaling=true +Categories=emulators; +selectorbrowser=true +SelectorDir=/mnt/Sega Genesis +SelectorFilter=zip,ZIP,md,MD diff --git a/FunKey/package/picodrive/opk/megadrive/megadrive.png b/FunKey/package/picodrive/opk/megadrive/megadrive.png new file mode 100644 index 0000000..585dd22 Binary files /dev/null and b/FunKey/package/picodrive/opk/megadrive/megadrive.png differ diff --git a/FunKey/package/picodrive/picodrive.mk b/FunKey/package/picodrive/picodrive.mk index f6b7070..1b17d4f 100644 --- a/FunKey/package/picodrive/picodrive.mk +++ b/FunKey/package/picodrive/picodrive.mk @@ -4,7 +4,7 @@ # ################################################################################ -PICODRIVE_VERSION = picodrive-FunKey-1.00 +PICODRIVE_VERSION = picodrive-FunKey-1.1.0 PICODRIVE_SITE_METHOD = git PICODRIVE_SITE = https://github.com/FunKey-Project/picodrive.git PICODRIVE_LICENSE = MAME @@ -12,25 +12,7 @@ PICODRIVE_LICENSE_FILES = COPYING PICODRIVE_DEPENDENCIES = sdl sdl_image sdl_mixer sdl_ttf zlib -PICODRIVE_CFLAGS = $(TARGET_CFLAGS) - -ifeq ($(BR2_ARM_CPU_ARMV7A),y) -PICODRIVE_CFLAGS += -march=armv7-a -endif - -ifeq ($(BR2_GCC_TARGET_CPU),"cortex-a7") -PICODRIVE_CFLAGS += -mtune=cortex-a7 -endif - -ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"hard") -PICODRIVE_CFLAGS += -mfloat-abi=hard -ffast-math -funsafe-math-optimizations -else ifeq ($(BR2_GCC_TARGET_FLOAT_ABI),"soft") -PICODRIVE_CFLAGS += -mfloat-abi=soft -ffast-math -funsafe-math-optimizations -endif - -ifeq ($(BR2_ARM_CPU_HAS_NEON),y) -PICODRIVE_CFLAGS += -D__ARM_NEON__ -mfpu=neon -mvectorize-with-neon-quad -endif +PICODRIVE_CFLAGS = $(TARGET_CFLAGS) $(subst $\",,$(BR2_TARGET_OPTIMIZATION)) -mfloat-abi=hard -ffast-math -funsafe-math-optimizations PICODRIVE_CONF_OPTS += --platform=generic --sound-drivers=sdl PICODRIVE_CFLAGS += -ggdb -O3 @@ -62,5 +44,10 @@ define PICODRIVE_INSTALL_TARGET_CMDS $(INSTALL) -m 0755 $(@D)/PicoDrive $(TARGET_DIR)/usr/games/ endef +define PICODRIVE_CREATE_OPK + $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games/opk + $(HOST_DIR)/usr/bin/mksquashfs $(PICODRIVE_PKGDIR)/opk/megadrive $(TARGET_DIR)/usr/games/opk/megadrive_picodrive_funkey-s.opk -all-root -noappend -no-exports -no-xattrs +endef +PICODRIVE_POST_INSTALL_TARGET_HOOKS += PICODRIVE_CREATE_OPK $(eval $(generic-package)) diff --git a/FunKey/package/retrofe/retrofe.mk b/FunKey/package/retrofe/retrofe.mk index ab89168..82a9f51 100644 --- a/FunKey/package/retrofe/retrofe.mk +++ b/FunKey/package/retrofe/retrofe.mk @@ -4,7 +4,7 @@ # ################################################################################ -RETROFE_VERSION = RetroFE-FunKey-1.00 +RETROFE_VERSION = RetroFE-FunKey-1.1.0 RETROFE_SITE_METHOD = git RETROFE_SITE = https://github.com/FunKey-Project/RetroFE.git RETROFE_DEPENDENCIES = gstreamer1 gst1-plugins-base sdl sdl_image sdl_mixer sdl_sound sdl_ttf libglib2 sqlite zlib @@ -15,9 +15,16 @@ RETROFE_SUBDIR = RetroFE/Source RETROFE_SUPPORTS_IN_SOURCE_BUILD = NO RETROFE_CONF_OPTS += -DVERSION_MAJOR=0 -DVERSION_MINOR=0 -DVERSION_BUILD=0 +ifeq ($(BR2_PACKAGE_LIBMIKMOD),y) +RETROFE_DEPENDENCIES += libmikmod +RETROFE_CONF_OPTS += -DLIBMIKMOD=1 +endif + define RETROFE_INSTALL_TARGET_CMDS $(INSTALL) -d -m 0755 $(TARGET_DIR)/usr/games $(INSTALL) -m 0755 $(@D)/RetroFE/Build/retrofe $(TARGET_DIR)/usr/games/retrofe endef +TARGET_CFLAGS += -O3 + $(eval $(cmake-package)) diff --git a/Makefile b/Makefile index 83cf861..86a6c02 100644 --- a/Makefile +++ b/Makefile @@ -57,6 +57,26 @@ fun: buildroot Recovery/output/.config FunKey/output/.config @$(call MESSAGE,"Making fun in FunKey") @$(BRMAKE) BR2_EXTERNAL=../FunKey O=../FunKey/output +sdk: buildroot SDK/output/.config + @$(call MESSAGE,"Making FunKey SDK") + @$(BRMAKE) BR2_EXTERNAL=../SDK O=../SDK/output prepare-sdk + @$(call MESSAGE,"Generating SDK tarball") + export LC_ALL=C; \ + SDK=FunKey-sdk-$(shell cat FunKey/board/funkey/rootfs-overlay/etc/sw-versions | cut -f 2); \ + grep -lr "$(shell pwd)/SDK/output/host" SDK/output/host | while read -r FILE ; do \ + if file -b --mime-type "$${FILE}" | grep -q '^text/'; then \ + sed -i "s|$(shell pwd)/SDK/output/host|/opt/$${SDK}|g" "$${FILE}"; \ + fi; \ + done; \ + mkdir -p images; \ + tar czf "images/$${SDK}.tar.gz" \ + --owner=0 --group=0 --numeric-owner \ + --transform="s#^$(patsubst /%,%,$(shell pwd))/SDK/output/host#$${SDK}#" \ + -C / "$(patsubst /%,%,$(shell pwd))/SDK/output/host"; \ + rm -f download/toolchain-external-custom/$${SDK}.tar.gz; \ + mkdir -p download/toolchain-external-custom; \ + ln -s ../../images/$${SDK}.tar.gz download/toolchain-external-custom/ + FunKey/%: FunKey/output/.config @$(call MESSAGE,"Making $(notdir $@) in $(subst /,,$(dir $@))") @$(BR) BR2_EXTERNAL=../FunKey O=../FunKey/output $(notdir $@) @@ -65,12 +85,17 @@ Recovery/%: Recovery/output/.config @$(call MESSAGE,"Making $(notdir $@) in $(subst /,,$(dir $@))") @$(BR) BR2_EXTERNAL=../Recovery O=../Recovery/output $(notdir $@) +SDK/%: SDK/output/.config + @$(call MESSAGE,"Making $(notdir $@) in $(subst /,,$(dir $@))") + @$(BR) BR2_EXTERNAL=../SDK O=../SDK/output $(notdir $@) + #%: FunKey/output/.config # @$(call MESSAGE,"Making $@ in FunKey") # @$(BR) BR2_EXTERNAL=../FunKey O=../FunKey/output $@ source: @$(call MESSAGE,"Getting sources") + @$(BR) BR2_EXTERNAL=../SDK O=../SDK/output source @$(BR) BR2_EXTERNAL=../Recovery O=../Recovery/output source @$(BR) BR2_EXTERNAL=../FunKey O=../FunKey/output source @@ -108,6 +133,8 @@ update: fun defconfig: @$(call MESSAGE,"Updating default configs") + @$(call MESSAGE,"Updating default configs in SDK") + @$(BR) BR2_EXTERNAL=../SDK O=../SDK/output savedefconfig @$(call MESSAGE,"Updating default configs in Recovery") @$(BR) BR2_EXTERNAL=../Recovery O=../Recovery/output savedefconfig linux-update-defconfig uboot-update-defconfig busybox-update-config @$(call MESSAGE,"Updating default configs in FunKey") @@ -115,6 +142,7 @@ defconfig: clean: @$(call MESSAGE,"Clean everything") + @$(BR) BR2_EXTERNAL=../SDK O=../SDK/output distclean @$(BR) BR2_EXTERNAL=../Recovery O=../Recovery/output distclean @$(BR) BR2_EXTERNAL=../FunKey O=../FunKey/output distclean @rm -f br.log @@ -132,3 +160,8 @@ Recovery/output/.config: @$(call MESSAGE,"Configure Recovery") @mkdir -p Recovery/board/funkey/patches @$(BR) BR2_EXTERNAL=../Recovery O=../Recovery/output recovery_defconfig + +SDK/output/.config: + @$(call MESSAGE,"Configure SDK") + @mkdir -p SDK/board/funkey/patches + @$(BR) BR2_EXTERNAL=../SDK O=../SDK/output funkey_defconfig diff --git a/Recovery/board/funkey/busybox.config b/Recovery/board/funkey/busybox.config index 6d28ca2..bc03308 100644 --- a/Recovery/board/funkey/busybox.config +++ b/Recovery/board/funkey/busybox.config @@ -1,7 +1,7 @@ # # Automatically generated make config: don't edit -# Busybox version: 1.31.1 -# Sun Nov 8 23:55:55 2020 +# Busybox version: 1.32.0 +# Fri Jan 15 10:35:44 2021 # CONFIG_HAVE_DOT_CONFIG=y @@ -76,6 +76,7 @@ CONFIG_PREFIX="./_install" # CONFIG_DEBUG_SANITIZE is not set # CONFIG_UNIT_TEST is not set # CONFIG_WERROR is not set +# CONFIG_WARN_SIMPLE_MSG is not set CONFIG_NO_DEBUG_LIB=y # CONFIG_DMALLOC is not set # CONFIG_EFENCE is not set @@ -462,6 +463,7 @@ CONFIG_FEATURE_FIND_SIZE=y CONFIG_FEATURE_FIND_PRUNE=y CONFIG_FEATURE_FIND_QUIT=y # CONFIG_FEATURE_FIND_DELETE is not set +CONFIG_FEATURE_FIND_EMPTY=y CONFIG_FEATURE_FIND_PATH=y CONFIG_FEATURE_FIND_REGEX=y # CONFIG_FEATURE_FIND_CONTEXT is not set @@ -692,6 +694,7 @@ CONFIG_FEATURE_SETPRIV_CAPABILITY_NAMES=y CONFIG_SWITCH_ROOT=y # CONFIG_TASKSET is not set # CONFIG_FEATURE_TASKSET_FANCY is not set +# CONFIG_FEATURE_TASKSET_CPULIST is not set CONFIG_UEVENT=y CONFIG_UMOUNT=y CONFIG_FEATURE_UMOUNT_ALL=y @@ -805,6 +808,7 @@ CONFIG_MAKEDEVS=y CONFIG_FEATURE_MAKEDEVS_TABLE=y # CONFIG_MAN is not set CONFIG_MICROCOM=y +CONFIG_MIM=y CONFIG_MT=y # CONFIG_NANDWRITE is not set # CONFIG_NANDDUMP is not set @@ -1095,6 +1099,7 @@ CONFIG_SH_IS_ASH=y # CONFIG_BASH_IS_ASH is not set # CONFIG_BASH_IS_HUSH is not set CONFIG_BASH_IS_NONE=y +CONFIG_SHELL_ASH=y CONFIG_ASH=y CONFIG_ASH_OPTIMIZE_FOR_SIZE=y CONFIG_ASH_INTERNAL_GLOB=y @@ -1115,6 +1120,7 @@ CONFIG_ASH_GETOPTS=y CONFIG_ASH_CMDCMD=y # CONFIG_CTTYHACK is not set # CONFIG_HUSH is not set +# CONFIG_SHELL_HUSH is not set # CONFIG_HUSH_BASH_COMPAT is not set # CONFIG_HUSH_BRACE_EXPANSION is not set # CONFIG_HUSH_LINENO_VAR is not set @@ -1177,6 +1183,7 @@ CONFIG_FEATURE_ROTATE_LOGFILE=y CONFIG_FEATURE_REMOTE_LOG=y # CONFIG_FEATURE_SYSLOGD_DUP is not set # CONFIG_FEATURE_SYSLOGD_CFG is not set +# CONFIG_FEATURE_SYSLOGD_PRECISE_TIMESTAMPS is not set CONFIG_FEATURE_SYSLOGD_READ_BUFFER_SIZE=256 # CONFIG_FEATURE_IPC_SYSLOG is not set CONFIG_FEATURE_IPC_SYSLOG_BUFFER_SIZE=0 diff --git a/Recovery/board/funkey/linux.config b/Recovery/board/funkey/linux.config index 20f59df..8993f06 100644 --- a/Recovery/board/funkey/linux.config +++ b/Recovery/board/funkey/linux.config @@ -25,6 +25,7 @@ CONFIG_CPU_FREQ=y CONFIG_CPUFREQ_DT=y CONFIG_VFP=y CONFIG_NEON=y +CONFIG_KERNEL_MODE_NEON=y # CONFIG_COREDUMP is not set CONFIG_HIBERNATION=y CONFIG_PM_STD_PARTITION="/dev/mmcblk0p2" @@ -153,7 +154,7 @@ CONFIG_AUTOFS4_FS=y CONFIG_VFAT_FS=y CONFIG_TMPFS=y CONFIG_CONFIGFS_FS=y -# CONFIG_MISC_FILESYSTEMS is not set +CONFIG_SQUASHFS=y # CONFIG_NETWORK_FILESYSTEMS is not set CONFIG_NLS_CODEPAGE_437=y CONFIG_NLS_ISO8859_1=y diff --git a/Recovery/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf b/Recovery/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf index ecb81ba..e5cbd8d 100644 --- a/Recovery/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf +++ b/Recovery/board/funkey/rootfs-overlay/etc/funkey_gpio_mapping.conf @@ -49,6 +49,7 @@ 13, KEYBOARD, KEY_Y, KEY_Y, Y 11, KEYBOARD, KEY_X, KEY_X, X +7+6, SHELL_COMMAND, snap, Fn+Start, Snapshot 7+12, SHELL_COMMAND, quick_action_volume_up, Fn+Y, Volume++ 7+13, SHELL_COMMAND, quick_action_volume_down, Fn+A, Volume-- 7+11, SHELL_COMMAND, quick_action_bright_up, Fn+B, Brightness++ diff --git a/Recovery/board/funkey/rootfs-overlay/etc/issue b/Recovery/board/funkey/rootfs-overlay/etc/issue index afbd169..62edbca 100644 --- a/Recovery/board/funkey/rootfs-overlay/etc/issue +++ b/Recovery/board/funkey/rootfs-overlay/etc/issue @@ -5,6 +5,6 @@ |___| |_____|__|__||__|\__||_____|___ | FUN ON A KEYCHAIN |_____| ----------------------------------------------------- - Version 1.1.0 (Recovery) + Version 2.0.0 (Recovery) ----------------------------------------------------- diff --git a/Recovery/board/funkey/rootfs-overlay/etc/os-release b/Recovery/board/funkey/rootfs-overlay/etc/os-release index 2483f0b..8a1fa48 100644 --- a/Recovery/board/funkey/rootfs-overlay/etc/os-release +++ b/Recovery/board/funkey/rootfs-overlay/etc/os-release @@ -1,12 +1,12 @@ NAME="FunKey-OS Recovery" -VERSION="1.1.0 (Quacking Quagga)" +VERSION="2.0.0 (Rowdy Rabbit)" ID=funkey-recovery ID_LIKE=buildroot -PRETTY_NAME="FunKey-OS Recovery 1.1.0" -VERSION_ID="1.1.0" +PRETTY_NAME="FunKey-OS Recovery 2.0.0" +VERSION_ID="2.0.0" HOME_URL="https://www.funkey-project.com/" SUPPORT_URL="https://www.funkey-project.com/" BUG_REPORT_URL="https://www.funkey-project.com/" PRIVACY_POLICY_URL="https://www.funkey-project.com" -VERSION_CODENAME=Quacking -UBUNTU_CODENAME=Quacking +VERSION_CODENAME=Rowdy +UBUNTU_CODENAME=Rowdy diff --git a/Recovery/board/funkey/rootfs-overlay/etc/sw-versions b/Recovery/board/funkey/rootfs-overlay/etc/sw-versions index 6019c23..4455861 100644 --- a/Recovery/board/funkey/rootfs-overlay/etc/sw-versions +++ b/Recovery/board/funkey/rootfs-overlay/etc/sw-versions @@ -1 +1 @@ -Recovery 1.1.0 +Recovery 2.0.0 diff --git a/Recovery/board/funkey/rootfs-overlay/media b/Recovery/board/funkey/rootfs-overlay/media new file mode 120000 index 0000000..35ec3b9 --- /dev/null +++ b/Recovery/board/funkey/rootfs-overlay/media @@ -0,0 +1 @@ +/ \ No newline at end of file diff --git a/Recovery/board/funkey/rootfs-overlay/opk/.empty b/Recovery/board/funkey/rootfs-overlay/opk/.empty new file mode 100644 index 0000000..e69de29 diff --git a/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/menu b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/menu index 1048ec7..88b8b89 100755 --- a/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/menu +++ b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/menu @@ -16,8 +16,8 @@ menu_display () { case ${entry} in 0) - # Version - message=" VERSION" + # Information + message=" INFO" ;; 1) @@ -61,20 +61,32 @@ menu_display () { 6) # Factory tests enable/disable - if [ -e /mnt/.assembly_tests ]; then - message=" FACTORY TESTS DISABLE" + RUN_ENV_VAR=$(fw_printenv -n assembly_tests 2>/dev/null) + if [ "x${RUN_ENV_VAR}" == "x1" ]; then + message=" FACTORY TESTS DISABLE" else - message=" FACTORY TESTS ENABLE" + message=" FACTORY TESTS ENABLE" fi ;; 7) + # First boot enable/disable + FIRST_BOOT_OK_ENV_VAR=$(fw_printenv -n first_boot_ok 2>/dev/null) + if [ "x${FIRST_BOOT_OK_ENV_VAR}" == "x1" ]; then + message=" FIRST BOOT DISABLE" + else + message=" FACTORY TESTS ENABLE" + fi + ;; + + 8) + # Exit Recovery message=" EXIT RECOVERY" ;; - 8) + 9) # Shutdown message=" SHUTDOWN" @@ -90,7 +102,7 @@ menu_run () { case ${entry} in 0) - # Version + # Information rootfs_mount=/tmp/rootfs mkdir -p ${rootfs_mount} mount -t ext4 -o ro /dev/mmcblk0p2 ${rootfs_mount} @@ -101,7 +113,8 @@ menu_run () { fi umount ${rootfs_mount} version_recovery=$(grep Recovery /etc/sw-versions | cut -f 2) - notif "${message}^^ Recovery: ${version_recovery}^ rootfs : ${version_rootfs}" + ip_addr=$(ifconfig usb0 | grep "inet " | awk -F'[: ]+' '{ print $4 }') + notif "${message}^^ Recovery: ${version_recovery}^ rootfs : ${version_rootfs}^ IP addr : ${ip_addr}" ;; 1) @@ -224,22 +237,36 @@ menu_run () { # Factory tests enable/disable RUN_ENV_VAR=$(fw_printenv -n assembly_tests 2>/dev/null) - if [ "x${RUN_ENV_VAR}" == "x1" ]; then - fw_setenv assembly_tests 0 - message=" FACTORY TESTS ENABLE" - else - fw_setenv assembly_tests 1 - message=" FACTORY TESTS DISABLE" - fi - notif "${message}" + if [ "x${RUN_ENV_VAR}" == "x1" ]; then + fw_setenv assembly_tests 0 + message=" FACTORY TESTS ENABLE" + else + fw_setenv assembly_tests 1 + message=" FACTORY TESTS DISABLE" + fi + notif "${message}" ;; 7) + + # First boot enable/disable + FIRST_BOOT_OK_ENV_VAR=$(fw_printenv -n first_boot_ok 2>/dev/null) + if [ "x${FIRST_BOOT_OK_ENV_VAR}" == "x1" ]; then + fw_setenv first_boot_ok + message=" FIRST BOOT ENABLE" + else + fw_setenv first_boot_ok 1 + message=" FIRST BOOT DISABLE" + fi + notif "${message}" + ;; + + 8) notif "^^^^^^^^ RESTARTING...^^^^^^^^" normal_mode ;; - 8) + 9) notif "^^^^^^^^ SHUTDOWN...^^^^^^^^" poweroff ;; diff --git a/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/snap b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/snap new file mode 100755 index 0000000..7a246a5 --- /dev/null +++ b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/snap @@ -0,0 +1,33 @@ +#!/bin/sh + +# Uncomment the following line to get debug info +#set -x + +# Check args +if [ ${#} -ne 0 ]; then + echo "Usage: $(basename ${0})" + exit 1 +fi + +# Lock file (necessary since fbgrab must run in bg not to block the buttons while gaming) +LOCK_FILE="/var/lock/snap.lock" +if [ -f "${LOCK_FILE}" ]; then + echo "${LOCK_FILE} already exists" + exit 1 +fi +touch "${LOCK_FILE}" + +# Increment name and save snapshot +SNAPSHOT_EXT=PNG +SNAPSHOT_DIR=$HOME/snapshots +mkdir -p "${SNAPSHOT_DIR}" +last=$(cd ${SNAPSHOT_DIR}; ls IMG_*.${SNAPSHOT_EXT} 2> /dev/null | tail -1 | sed 's/^IMG_0*\([0-9]\+\)\.'${SNAPSHOT_EXT}'$/\1/') +let last=${last}+1 +snapshot_file=$(printf "IMG_%04d.${SNAPSHOT_EXT}" $last) +notif_set 2 " SCREEENSHOT ${snapshot_file}" +fbgrab "${SNAPSHOT_DIR}/${snapshot_file}" >/dev/null 2>&1 & + +# Remove lock file +rm -f "${LOCK_FILE}" + +exit 0 diff --git a/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/system_stats b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/system_stats index a6a091a..4e6ad97 100755 --- a/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/system_stats +++ b/Recovery/board/funkey/rootfs-overlay/usr/local/sbin/system_stats @@ -20,16 +20,21 @@ while true; do if [ ${perform} -eq 1 ]; then # Compute stats - cpu=$(printf "%.1f\n" $(mpstat -P ALL $UPDATE_PERIOD 1 | tail -1 | awk '{print 100-$12}')) - ram_mem=$(printf "%.1f\n" $(free | grep Mem | awk '{print $3/$2 * 100.0}')) - ram_swap=$(printf "%.1f\n" $(free | grep Swap | awk '{print $3/$2 * 100.0}')) + cpu=$(printf "%.0f\n" $(mpstat -P ALL $UPDATE_PERIOD 1 | tail -1 | awk '{print 100-$12}')) + ram_mem=$(printf "%.0f\n" $(free | grep Mem | awk '{print $3/$2 * 100.0}')) + ram_swap=$(printf "%.0f\n" $(free | grep Swap | awk '{print $3/$2 * 100.0}')) + ip_addr=$(ifconfig usb0 | grep "inet " | awk -F'[: ]+' '{ print $4 }') # Notif if [ ${notif_dirty} -eq 1 ]; then notif_clear notif_dirty=0 else - notif_set 0 "CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%" + if [ "x${ip_addr}" != "x" ]; then + notif_set 0 " CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%^IP:${ip_addr}" + else + notif_set 0 " CPU:${cpu}%% RAM:${ram_mem}%% SWAP:${ram_swap}%%" + fi fi else sleep ${UPDATE_PERIOD} diff --git a/Recovery/configs/recovery_defconfig b/Recovery/configs/recovery_defconfig index 7570988..ee12e1c 100644 --- a/Recovery/configs/recovery_defconfig +++ b/Recovery/configs/recovery_defconfig @@ -1,10 +1,19 @@ BR2_arm=y BR2_cortex_a7=y +BR2_ARM_FPU_VFPV4=y BR2_DL_DIR="../download" BR2_CCACHE=y +BR2_OPTIMIZE_FAST=y +BR2_SHARED_STATIC_LIBS=y BR2_GLOBAL_PATCH_DIR="$(BR2_EXTERNAL_RECOVERY_PATH)/board/funkey/patches" BR2_TOOLCHAIN_EXTERNAL=y -BR2_TOOLCHAIN_EXTERNAL_GDB_SERVER_COPY=y +BR2_TOOLCHAIN_EXTERNAL_CUSTOM=y +BR2_TOOLCHAIN_EXTERNAL_DOWNLOAD=y +BR2_TOOLCHAIN_EXTERNAL_URL="https://github.com/FunKey-Project/FunKey-OS/releases/download/FunKey-OS-2.0.0/FunKey-sdk-2.0.0.tar.gz" +BR2_TOOLCHAIN_EXTERNAL_HEADERS_4_14=y +BR2_TOOLCHAIN_EXTERNAL_CUSTOM_MUSL=y +BR2_TOOLCHAIN_EXTERNAL_CXX=y +BR2_TARGET_OPTIMIZATION="-fno-PIC -march=armv7-a+neon-vfpv4 -mtune=cortex-a7 -mfpu=neon-vfpv4" BR2_TARGET_GENERIC_HOSTNAME="FunKey" BR2_TARGET_GENERIC_ISSUE="Welcome to Recovery Buildroot for the FunKey" BR2_ROOTFS_DEVICE_TABLE="$(BR2_EXTERNAL_RECOVERY_PATH)/board/funkey/device_table.txt" diff --git a/Recovery/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk b/Recovery/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk index af98396..36958ed 100644 --- a/Recovery/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk +++ b/Recovery/package/FunKey-GPIO-Mapping/FunKey-GPIO-Mapping.mk @@ -4,7 +4,7 @@ # ################################################################################ -FUNKEY_GPIO_MAPPING_VERSION = FunKey-GPIO-Mapping-FunKey-1.00 +FUNKEY_GPIO_MAPPING_VERSION = e2b637f FUNKEY_GPIO_MAPPING_SITE_METHOD = git FUNKEY_GPIO_MAPPING_SITE = https://github.com/FunKey-Project/FunKey-GPIO-Mapping.git FUNKEY_GPIO_MAPPING_SITE_LICENSE = GPL-2.1+ diff --git a/SDK/Config.in b/SDK/Config.in new file mode 100644 index 0000000..984b74f --- /dev/null +++ b/SDK/Config.in @@ -0,0 +1,7 @@ +source "$BR2_EXTERNAL_SDK_PATH/package/dmtx-utils/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/libini/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/libopk/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/libxdgmime/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/agg/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/fluidlite/Config.in" +source "$BR2_EXTERNAL_SDK_PATH/package/libmikmod/Config.in" diff --git a/SDK/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch b/SDK/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch new file mode 100644 index 0000000..1f27440 --- /dev/null +++ b/SDK/board/funkey/patches/pixman/0002-aarch64NEON_asm.patch @@ -0,0 +1,6526 @@ +From 7b128ae8c56b1055a93573004148a98465d79857 Mon Sep 17 00:00:00 2001 +From: Mizuki Asakura +Date: Sun, 17 Apr 2016 20:16:12 +0900 +Subject: [PATCH] [mod] added aarch64 bilinear implementations (ver.4.1) + +Since aarch64 has different neon syntax from aarch32 and has no +support for (older) arm-simd, +there are no SIMD accelerations for pixman on aarch64. + +We need new implementations. + +This patch also contains Ben Avions's series of patches for aarch32 +and now the benchmark results are fine to aarch64. + +Please find the result at the below ticket. + +Added: https://bugs.freedesktop.org/show_bug.cgi?id=94758 +Signed-off-by: Mizuki Asakura +--- + configure.ac | 36 +- + pixman/Makefile.am | 17 +- + pixman/pixman-arm-neon.c | 23 +- + pixman/pixman-arm.c | 6 + + pixman/pixman-arma64-neon-asm-bilinear.S | 1275 ++++++++++ + pixman/pixman-arma64-neon-asm.S | 3704 ++++++++++++++++++++++++++++++ + pixman/pixman-arma64-neon-asm.h | 1310 +++++++++++ + pixman/pixman-private.h | 7 +- + 8 files changed, 6374 insertions(+), 4 deletions(-) + create mode 100644 pixman/pixman-arma64-neon-asm-bilinear.S + create mode 100644 pixman/pixman-arma64-neon-asm.S + create mode 100644 pixman/pixman-arma64-neon-asm.h + +diff --git a/configure.ac b/configure.ac +index 6b2134e..26203a8 +--- a/configure.ac ++++ b/configure.ac +@@ -667,6 +667,40 @@ if test $enable_arm_neon = yes && test $have_arm_neon = no ; then + AC_MSG_ERROR([ARM NEON intrinsics not detected]) + fi + ++dnl ========================================================================== ++dnl Check if assembler is gas compatible and supports ARM-a64 NEON instructions ++have_arm_a64_neon=no ++AC_MSG_CHECKING(whether to use ARM A64 NEON assembler) ++xserver_save_CFLAGS=$CFLAGS ++CFLAGS="-x assembler-with-cpp $CFLAGS" ++AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ ++.text ++.arch armv8-a ++.altmacro ++prfm pldl2strm, [x0] ++xtn v0.8b, v0.8h]])], have_arm_a64_neon=yes) ++CFLAGS=$xserver_save_CFLAGS ++ ++AC_ARG_ENABLE(arm-a64-neon, ++ [AC_HELP_STRING([--disable-arm-a64-neon], ++ [disable ARM A64 NEON fast paths])], ++ [enable_arm_a64_neon=$enableval], [enable_arm_a64_neon=auto]) ++ ++if test $enable_arm_a64_neon = no ; then ++ have_arm_a64_neon=disabled ++fi ++ ++if test $have_arm_a64_neon = yes ; then ++ AC_DEFINE(USE_ARM_A64_NEON, 1, [use ARM A64_NEON assembly optimizations]) ++fi ++ ++AM_CONDITIONAL(USE_ARM_A64_NEON, test $have_arm_a64_neon = yes) ++ ++AC_MSG_RESULT($have_arm_a64_neon) ++if test $enable_arm_a64_neon = yes && test $have_arm_a64_neon4 = no ; then ++ AC_MSG_ERROR([ARM A64 NEON intrinsics not detected]) ++fi ++ + dnl =========================================================================== + dnl Check for IWMMXT + +diff --git a/pixman/Makefile.am b/pixman/Makefile.am +index 581b6f6..f1afa27 +--- a/pixman/Makefile.am ++++ b/pixman/Makefile.am +@@ -94,6 +94,21 @@ libpixman_1_la_LIBADD += libpixman-arm-neon.la + ASM_CFLAGS_arm_neon= + endif + ++# arm a64 neon code ++if USE_ARM_A64_NEON ++noinst_LTLIBRARIES += libpixman-arma64-neon.la ++libpixman_arma64_neon_la_SOURCES = \ ++ pixman-arm-neon.c \ ++ pixman-arm-common.h \ ++ pixman-arma64-neon-asm.S \ ++ pixman-arma64-neon-asm-bilinear.S \ ++ pixman-arm-asm.h \ ++ pixman-arma64-neon-asm.h ++libpixman_1_la_LIBADD += libpixman-arma64-neon.la ++ ++ASM_CFLAGS_arm_neon= ++endif ++ + # iwmmxt code + if USE_ARM_IWMMXT + libpixman_iwmmxt_la_SOURCES = pixman-mmx.c +diff --git a/pixman/pixman-arm-neon.c b/pixman/pixman-arm-neon.c +index be761c9..62c9442 100644 +--- a/pixman/pixman-arm-neon.c ++++ b/pixman/pixman-arm-neon.c +@@ -194,7 +194,7 @@ arm_neon_fill (pixman_implementation_t *imp, + uint32_t _xor) + { + /* stride is always multiple of 32bit units in pixman */ +- uint32_t byte_stride = stride * sizeof(uint32_t); ++ int32_t byte_stride = stride * sizeof(uint32_t); + + switch (bpp) + { +@@ -331,6 +331,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, a8, b5g6r5, neon_composite_over_8888_8_0565), + PIXMAN_STD_FAST_PATH (OVER, r5g6b5, a8, r5g6b5, neon_composite_over_0565_8_0565), + PIXMAN_STD_FAST_PATH (OVER, b5g6r5, a8, b5g6r5, neon_composite_over_0565_8_0565), ++ PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_over_8888_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, a8r8g8b8, neon_composite_over_8888_8888_8888), + PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, r5g6b5, neon_composite_over_8888_0565), + PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, b5g6r5, neon_composite_over_8888_0565), +@@ -341,17 +342,33 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER, x8r8g8b8, null, a8r8g8b8, neon_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (OVER, x8b8g8r8, null, a8b8g8r8, neon_composite_src_x888_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8, neon_composite_add_n_8_8), ++ PIXMAN_STD_FAST_PATH (ADD, solid, a8, x8r8g8b8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8r8g8b8, neon_composite_add_n_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, solid, a8, x8b8g8r8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, solid, a8, a8b8g8r8, neon_composite_add_n_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, a8, a8, neon_composite_add_8_8_8), + PIXMAN_STD_FAST_PATH (ADD, r5g6b5, a8, r5g6b5, neon_composite_add_0565_8_0565), + PIXMAN_STD_FAST_PATH (ADD, b5g6r5, a8, b5g6r5, neon_composite_add_0565_8_0565), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, a8, x8r8g8b8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8, x8r8g8b8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, a8, x8b8g8r8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, a8, x8b8g8r8, neon_composite_add_8888_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8, a8r8g8b8, neon_composite_add_8888_8_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, a8, a8b8g8r8, neon_composite_add_8888_8_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_add_8888_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, x8r8g8b8, neon_composite_add_8888_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, a8r8g8b8, neon_composite_add_8888_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, solid, x8r8g8b8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, solid, x8r8g8b8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, solid, x8b8g8r8, neon_composite_add_8888_n_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, solid, x8b8g8r8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, solid, a8r8g8b8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, solid, a8b8g8r8, neon_composite_add_8888_n_8888), + PIXMAN_STD_FAST_PATH (ADD, a8, null, a8, neon_composite_add_8_8), ++ PIXMAN_STD_FAST_PATH (ADD, x8r8g8b8, null, x8r8g8b8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, x8r8g8b8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, x8b8g8r8, null, x8b8g8r8, neon_composite_add_8888_8888), ++ PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, x8b8g8r8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8r8g8b8, null, a8r8g8b8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (ADD, a8b8g8r8, null, a8b8g8r8, neon_composite_add_8888_8888), + PIXMAN_STD_FAST_PATH (IN, solid, null, a8, neon_composite_in_n_8), +@@ -359,7 +376,9 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER_REVERSE, solid, null, a8b8g8r8, neon_composite_over_reverse_n_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, r5g6b5, neon_composite_out_reverse_8_0565), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, b5g6r5, neon_composite_out_reverse_8_0565), ++ PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, x8r8g8b8, neon_composite_out_reverse_8_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8r8g8b8, neon_composite_out_reverse_8_8888), ++ PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, x8b8g8r8, neon_composite_out_reverse_8_8888), + PIXMAN_STD_FAST_PATH (OUT_REVERSE, a8, null, a8b8g8r8, neon_composite_out_reverse_8_8888), + + SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, a8r8g8b8, neon_8888_8888), +@@ -404,6 +423,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + + SIMPLE_BILINEAR_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, neon_8888_8888), + SIMPLE_BILINEAR_FAST_PATH (ADD, a8r8g8b8, x8r8g8b8, neon_8888_8888), ++ SIMPLE_BILINEAR_FAST_PATH (ADD, x8r8g8b8, x8r8g8b8, neon_8888_8888), + + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, neon_8888_8_8888), + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, neon_8888_8_8888), +@@ -420,6 +440,7 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = + + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, a8r8g8b8, a8r8g8b8, neon_8888_8_8888), + SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, a8r8g8b8, x8r8g8b8, neon_8888_8_8888), ++ SIMPLE_BILINEAR_A8_MASK_FAST_PATH (ADD, x8r8g8b8, x8r8g8b8, neon_8888_8_8888), + + { PIXMAN_OP_NONE }, + }; +diff --git a/pixman/pixman-arm.c b/pixman/pixman-arm.c +index 23374e4..734cbea 100644 +--- a/pixman/pixman-arm.c ++++ b/pixman/pixman-arm.c +@@ -221,5 +221,11 @@ _pixman_arm_get_implementations (pixman_implementation_t *imp) + imp = _pixman_implementation_create_arm_neon (imp); + #endif + ++#ifdef USE_ARM_A64_NEON ++ /* neon is a part of aarch64 */ ++ if (!_pixman_disabled ("arm-neon")) ++ imp = _pixman_implementation_create_arm_neon (imp); ++#endif ++ + return imp; + } +diff --git a/pixman/pixman-arma64-neon-asm-bilinear.S b/pixman/pixman-arma64-neon-asm-bilinear.S +new file mode 100644 +index 0000000..aaa4a83 +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm-bilinear.S +@@ -0,0 +1,1275 @@ ++/* ++ * Copyright © 2011 SCore Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ * Author: Taekyun Kim (tkq.kim@samsung.com) ++ */ ++ ++/* ++ * This file contains scaled bilinear scanline functions implemented ++ * using older siarhei's bilinear macro template. ++ * ++ * << General scanline function procedures >> ++ * 1. bilinear interpolate source pixels ++ * 2. load mask pixels ++ * 3. load destination pixels ++ * 4. duplicate mask to fill whole register ++ * 5. interleave source & destination pixels ++ * 6. apply mask to source pixels ++ * 7. combine source & destination pixels ++ * 8, Deinterleave final result ++ * 9. store destination pixels ++ * ++ * All registers with single number (i.e. src0, tmp0) are 64-bits registers. ++ * Registers with double numbers(src01, dst01) are 128-bits registers. ++ * All temp registers can be used freely outside the code block. ++ * Assume that symbol(register .req) OUT and MASK are defined at caller of these macro blocks. ++ * ++ * Remarks ++ * There can be lots of pipeline stalls inside code block and between code blocks. ++ * Further optimizations will be done by new macro templates using head/tail_head/tail scheme. ++ */ ++ ++/* Prevent the stack from becoming executable for no reason... */ ++#if defined(__linux__) && defined (__ELF__) ++.section .note.GNU-stack,"",%progbits ++#endif ++ ++.text ++.arch armv8-a ++.altmacro ++.p2align 2 ++ ++#include "pixman-private.h" ++#include "pixman-arm-asm.h" ++#include "pixman-arma64-neon-asm.h" ++ ++/* ++ * Bilinear macros from pixman-arm-neon-asm.S ++ */ ++ ++/* ++ * Bilinear scaling support code which tries to provide pixel fetching, color ++ * format conversion, and interpolation as separate macros which can be used ++ * as the basic building blocks for constructing bilinear scanline functions. ++ */ ++ ++.macro bilinear_load_8888 reg1, reg2, tmp ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ ld1 {®1&.2s}, [TMP1], STRIDE ++ ld1 {®2&.2s}, [TMP1] ++.endm ++ ++.macro bilinear_load_0565 reg1, reg2, tmp ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ ld1 {®2&.s}[0], [TMP1], STRIDE ++ ld1 {®2&.s}[1], [TMP1] ++ convert_four_0565_to_x888_packed reg2, reg1, reg2, tmp ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_8888 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, tmp1, tmp2 ++ ++ bilinear_load_8888 reg1, reg2, tmp1 ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ bilinear_load_8888 reg3, reg4, tmp2 ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++.endm ++ ++.macro vzip reg1, reg2 ++ zip1 v24.8b, reg1, reg2 ++ zip2 reg2, reg1, reg2 ++ mov reg1, v24.8b ++.endm ++ ++.macro vuzp reg1, reg2 ++ uzp1 v24.8b, reg1, reg2 ++ uzp2 reg2, reg1, reg2 ++ mov reg1, v24.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_0565 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, acc2lo, acc2hi ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&acc2&.s}[0], [TMP1], STRIDE ++ ld1 {&acc2&.s}[2], [TMP2], STRIDE ++ ld1 {&acc2&.s}[1], [TMP1] ++ ld1 {&acc2&.s}[3], [TMP2] ++ convert_0565_to_x888 acc2, reg3, reg2, reg1 ++ vzip ®1&.8b, ®3&.8b ++ vzip ®2&.8b, ®4&.8b ++ vzip ®3&.8b, ®4&.8b ++ vzip ®1&.8b, ®2&.8b ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_0565 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&xacc2&.s}[0], [TMP1], STRIDE ++ ld1 {&xacc2&.s}[2], [TMP2], STRIDE ++ ld1 {&xacc2&.s}[1], [TMP1] ++ ld1 {&xacc2&.s}[3], [TMP2] ++ convert_0565_to_x888 xacc2, xreg3, xreg2, xreg1 ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&yacc2&.s}[0], [TMP1], STRIDE ++ vzip &xreg1&.8b, &xreg3&.8b ++ ld1 {&yacc2&.s}[2], [TMP2], STRIDE ++ vzip &xreg2&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[1], [TMP1] ++ vzip &xreg3&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[3], [TMP2] ++ vzip &xreg1&.8b, &xreg2&.8b ++ convert_0565_to_x888 yacc2, yreg3, yreg2, yreg1 ++ umull &xacc1&.8h, &xreg1&.8b, v28.8b ++ vzip &yreg1&.8b, &yreg3&.8b ++ umlal &xacc1&.8h, &xreg2&.8b, v29.8b ++ vzip &yreg2&.8b, &yreg4&.8b ++ umull &xacc2&.8h, &xreg3&.8b, v28.8b ++ vzip &yreg3&.8b, &yreg4&.8b ++ umlal &xacc2&.8h, &xreg4&.8b, v29.8b ++ vzip &yreg1&.8b, &yreg2&.8b ++ umull &yacc1&.8h, &yreg1&.8b, v28.8b ++ umlal &yacc1&.8h, &yreg2&.8b, v29.8b ++ umull &yacc2&.8h, &yreg3&.8b, v28.8b ++ umlal &yacc2&.8h, &yreg4&.8b, v29.8b ++.endm ++ ++.macro bilinear_store_8888 numpix, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v0.2s, v1.2s}, [OUT], #16 ++.elseif numpix == 2 ++ st1 {v0.2s}, [OUT], #8 ++.elseif numpix == 1 ++ st1 {v0.s}[0], [OUT], #4 ++.else ++ .error bilinear_store_8888 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_store_0565 numpix, tmp1, tmp2 ++ vuzp v0.8b, v1.8b ++ vuzp v2.8b, v3.8b ++ vuzp v1.8b, v3.8b ++ vuzp v0.8b, v2.8b ++ convert_8888_to_0565 v2, v1, v0, v1, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v1.4h}, [OUT], #8 ++.elseif numpix == 2 ++ st1 {v1.s}[0], [OUT], #4 ++.elseif numpix == 1 ++ st1 {v1.h}[0], [OUT], #2 ++.else ++ .error bilinear_store_0565 numpix is unsupported ++.endif ++.endm ++ ++ ++/* ++ * Macros for loading mask pixels into register 'mask'. ++ * dup must be done in somewhere else. ++ */ ++.macro bilinear_load_mask_x numpix, mask ++.endm ++ ++.macro bilinear_load_mask_8 numpix, mask ++.if numpix == 4 ++ ld1 {&mask&.s}[0], [MASK], #4 ++.elseif numpix == 2 ++ ld1 {&mask&.h}[0], [MASK], #2 ++.elseif numpix == 1 ++ ld1 {&mask&.b}[0], [MASK], #1 ++.else ++ .error bilinear_load_mask_8 numpix is unsupported ++.endif ++ prfm PREFETCH_MODE, [MASK, #prefetch_offset] ++.endm ++ ++.macro bilinear_load_mask mask_fmt, numpix, mask ++ bilinear_load_mask_&mask_fmt numpix, mask ++.endm ++ ++ ++/* ++ * Macros for loading destination pixels into register 'dst0' and 'dst1'. ++ * Interleave should be done somewhere else. ++ */ ++.macro bilinear_load_dst_0565_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.if numpix == 4 ++ ld1 {&dst0&.2s, &dst1&.2s}, [OUT] ++.elseif numpix == 2 ++ ld1 {&dst0&.2s}, [OUT] ++.elseif numpix == 1 ++ ld1 {&dst0&.s}[0], [OUT] ++.else ++ .error bilinear_load_dst_8888 numpix is unsupported ++.endif ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ prfm PREFETCH_MODE, [OUT, #(prefetch_offset * 4)] ++.endm ++ ++.macro bilinear_load_dst_8888_over numpix, dst0, dst1, dst01 ++ bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst_8888_add numpix, dst0, dst1, dst01 ++ bilinear_load_dst_8888 numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_load_dst dst_fmt, op, numpix, dst0, dst1, dst01 ++ bilinear_load_dst_&dst_fmt&_&op numpix, dst0, dst1, dst01 ++.endm ++ ++/* ++ * Macros for duplicating partially loaded mask to fill entire register. ++ * We will apply mask to interleaved source pixels, that is ++ * (r0, r1, r2, r3, g0, g1, g2, g3) x (m0, m1, m2, m3, m0, m1, m2, m3) ++ * (b0, b1, b2, b3, a0, a1, a2, a3) x (m0, m1, m2, m3, m0, m1, m2, m3) ++ * So, we need to duplicate loaded mask into whole register. ++ * ++ * For two pixel case ++ * (r0, r1, x, x, g0, g1, x, x) x (m0, m1, m0, m1, m0, m1, m0, m1) ++ * (b0, b1, x, x, a0, a1, x, x) x (m0, m1, m0, m1, m0, m1, m0, m1) ++ * We can do some optimizations for this including last pixel cases. ++ */ ++.macro bilinear_duplicate_mask_x numpix, mask ++.endm ++ ++.macro bilinear_duplicate_mask_8 numpix, mask ++.if numpix == 4 ++ dup &mask&.2s, &mask&.s[0] ++.elseif numpix == 2 ++ dup &mask&.4h, &mask&.h[0] ++.elseif numpix == 1 ++ dup &mask&.8b, &mask&.b[0] ++.else ++ .error bilinear_duplicate_mask_8 is unsupported ++.endif ++.endm ++ ++.macro bilinear_duplicate_mask mask_fmt, numpix, mask ++ bilinear_duplicate_mask_&mask_fmt numpix, mask ++.endm ++ ++/* ++ * Macros for interleaving src and dst pixels to rrrr gggg bbbb aaaa form. ++ * Interleave should be done when maks is enabled or operator is 'over'. ++ */ ++.macro bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++ vuzp &src0&.8b, &src1&.8b ++ vuzp &dst0&.8b, &dst1&.8b ++ vuzp &src0&.8b, &src1&.8b ++ vuzp &dst0&.8b, &dst1&.8b ++ mov &src01&.d[1], &src1&.d[0] ++ mov &src01&.d[0], &src0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ mov &dst01&.d[0], &dst0&.d[0] ++.endm ++ ++.macro bilinear_interleave_src_dst_x_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_x_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_x_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst_8_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_interleave_src_dst \ ++ mask_fmt, op, numpix, src0, src1, src01, dst0, dst1, dst01 ++ ++ bilinear_interleave_src_dst_&mask_fmt&_&op \ ++ numpix, src0, src1, src01, dst0, dst1, dst01 ++.endm ++ ++ ++/* ++ * Macros for applying masks to src pixels. (see combine_mask_u() function) ++ * src, dst should be in interleaved form. ++ * mask register should be in form (m0, m1, m2, m3). ++ */ ++.macro bilinear_apply_mask_to_src_x \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++.endm ++ ++.macro bilinear_apply_mask_to_src_8 \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++ ++ umull &tmp01&.8h, &src0&.8b, &mask&.8b ++ umull &tmp23&.8h, &src1&.8b, &mask&.8b ++ /* bubbles */ ++ urshr &tmp45&.8h, &tmp01&.8h, #8 ++ urshr &tmp67&.8h, &tmp23&.8h, #8 ++ /* bubbles */ ++ raddhn &src0&.8b, &tmp45&.8h, &tmp01&.8h ++ raddhn &src1&.8b, &tmp67&.8h, &tmp23&.8h ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_apply_mask_to_src \ ++ mask_fmt, numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++ ++ bilinear_apply_mask_to_src_&mask_fmt \ ++ numpix, src0, src1, src01, mask, \ ++ tmp01, tmp23, tmp45, tmp67 ++.endm ++ ++ ++/* ++ * Macros for combining src and destination pixels. ++ * Interleave or not is depending on operator 'op'. ++ */ ++.macro bilinear_combine_src \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++.endm ++ ++.macro bilinear_combine_over \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ dup &tmp8&.2s, &src1&.s[1] ++ /* bubbles */ ++ mvn &tmp8&.8b, &tmp8&.8b ++ /* bubbles */ ++ umull &tmp01&.8h, &dst0&.8b, &tmp8&.8b ++ /* bubbles */ ++ umull &tmp23&.8h, &dst1&.8b, &tmp8&.8b ++ /* bubbles */ ++ urshr &tmp45&.8h, &tmp01&.8h, #8 ++ urshr &tmp67&.8h, &tmp23&.8h, #8 ++ /* bubbles */ ++ raddhn &dst0&.8b, &tmp45&.8h, &tmp01&.8h ++ raddhn &dst1&.8b, &tmp67&.8h, &tmp23&.8h ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++ /* bubbles */ ++ uqadd &src0&.8b, &dst0&.8b, &src0&.8b ++ uqadd &src1&.8b, &dst1&.8b, &src1&.8b ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_combine_add \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ uqadd &src0&.8b, &dst0&.8b, &src0&.8b ++ uqadd &src1&.8b, &dst1&.8b, &src1&.8b ++ mov &src01&.d[0], &src0&.d[0] ++ mov &src01&.d[1], &src1&.d[0] ++.endm ++ ++.macro bilinear_combine \ ++ op, numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++ ++ bilinear_combine_&op \ ++ numpix, src0, src1, src01, dst0, dst1, dst01, \ ++ tmp01, tmp23, tmp45, tmp67, tmp8 ++.endm ++ ++/* ++ * Macros for final deinterleaving of destination pixels if needed. ++ */ ++.macro bilinear_deinterleave numpix, dst0, dst1, dst01 ++ vuzp &dst0&.8b, &dst1&.8b ++ /* bubbles */ ++ vuzp &dst0&.8b, &dst1&.8b ++ mov &dst01&.d[0], &dst0&.d[0] ++ mov &dst01&.d[1], &dst1&.d[0] ++.endm ++ ++.macro bilinear_deinterleave_dst_x_src numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_x_over numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_x_add numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_src numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_over numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst_8_add numpix, dst0, dst1, dst01 ++ bilinear_deinterleave numpix, dst0, dst1, dst01 ++.endm ++ ++.macro bilinear_deinterleave_dst mask_fmt, op, numpix, dst0, dst1, dst01 ++ bilinear_deinterleave_dst_&mask_fmt&_&op numpix, dst0, dst1, dst01 ++.endm ++ ++ ++.macro bilinear_interpolate_last_pixel src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_&src_fmt v0, v1, v2 ++ bilinear_load_mask mask_fmt, 1, v4 ++ bilinear_load_dst dst_fmt, op, 1, v18, v19, v9 ++ umull v2.8h, v0.8b, v28.8b ++ umlal v2.8h, v1.8b, v29.8b ++ /* 5 cycles bubble */ ++ ushll v0.4s, v2.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v2.4h, v15.h[0] ++ umlal2 v0.4s, v2.8h, v15.h[0] ++ /* 5 cycles bubble */ ++ bilinear_duplicate_mask mask_fmt, 1, v4 ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ /* 3 cycles bubble */ ++ xtn v0.8b, v0.8h ++ /* 1 cycle bubble */ ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 1, v0, v1, v0, v18, v19, v9 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 1, v0, v1, v0, v4, \ ++ v3, v8, v10, v11 ++ bilinear_combine \ ++ op, 1, v0, v1, v0, v18, v19, v9, \ ++ v3, v8, v10, v11, v5 ++ bilinear_deinterleave_dst mask_fmt, op, 1, v0, v1, v0 ++ bilinear_store_&dst_fmt 1, v17, v18 ++.endm ++ ++.macro bilinear_interpolate_two_pixels src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_and_vertical_interpolate_two_&src_fmt \ ++ v1, v11, v18, v19, v20, v21, v22, v23 ++ bilinear_load_mask mask_fmt, 2, v4 ++ bilinear_load_dst dst_fmt, op, 2, v18, v19, v9 ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ bilinear_duplicate_mask mask_fmt, 2, v4 ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ xtn v0.8b, v0.8h ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 2, v0, v1, v0, v18, v19, v9 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 2, v0, v1, v0, v4, \ ++ v3, v8, v10, v11 ++ bilinear_combine \ ++ op, 2, v0, v1, v0, v18, v19, v9, \ ++ v3, v8, v10, v11, v5 ++ bilinear_deinterleave_dst mask_fmt, op, 2, v0, v1, v0 ++ bilinear_store_&dst_fmt 2, v16, v17 ++.endm ++ ++.macro bilinear_interpolate_four_pixels src_fmt, mask_fmt, dst_fmt, op ++ bilinear_load_and_vertical_interpolate_four_&src_fmt \ ++ v1, v11, v4, v5, v6, v7, v22, v23 \ ++ v3, v9, v16, v17, v20, v21, v18, v19 ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ sub TMP1, TMP1, STRIDE ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v3.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v3.4h, v15.h[0] ++ umlal2 v2.4s, v3.8h, v15.h[0] ++ ushll v8.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v8.4s, v9.4h, v15.h[4] ++ umlal2 v8.4s, v9.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v8.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ bilinear_load_mask mask_fmt, 4, v4 ++ bilinear_duplicate_mask mask_fmt, 4, v4 ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ xtn v0.8b, v0.8h ++ xtn v1.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ bilinear_load_dst dst_fmt, op, 4, v2, v3, v21 ++ bilinear_interleave_src_dst \ ++ mask_fmt, op, 4, v0, v1, v0, v2, v3, v11 ++ bilinear_apply_mask_to_src \ ++ mask_fmt, 4, v0, v1, v0, v4, \ ++ v6, v8, v9, v10 ++ bilinear_combine \ ++ op, 4, v0, v1, v0, v2, v3, v1, \ ++ v6, v8, v9, v10, v23 ++ bilinear_deinterleave_dst mask_fmt, op, 4, v0, v1, v0 ++ bilinear_store_&dst_fmt 4, v6, v7 ++.endm ++ ++.set BILINEAR_FLAG_USE_MASK, 1 ++.set BILINEAR_FLAG_USE_ALL_NEON_REGS, 2 ++ ++/* ++ * Main template macro for generating NEON optimized bilinear scanline functions. ++ * ++ * Bilinear scanline generator macro take folling arguments: ++ * fname - name of the function to generate ++ * src_fmt - source color format (8888 or 0565) ++ * dst_fmt - destination color format (8888 or 0565) ++ * src/dst_bpp_shift - (1 << bpp_shift) is the size of src/dst pixel in bytes ++ * process_last_pixel - code block that interpolate one pixel and does not ++ * update horizontal weight ++ * process_two_pixels - code block that interpolate two pixels and update ++ * horizontal weight ++ * process_four_pixels - code block that interpolate four pixels and update ++ * horizontal weight ++ * process_pixblock_head - head part of middle loop ++ * process_pixblock_tail - tail part of middle loop ++ * process_pixblock_tail_head - tail_head of middle loop ++ * pixblock_size - number of pixels processed in a single middle loop ++ * prefetch_distance - prefetch in the source image by that many pixels ahead ++ */ ++ ++.macro generate_bilinear_scanline_func \ ++ fname, \ ++ src_fmt, dst_fmt, src_bpp_shift, dst_bpp_shift, \ ++ bilinear_process_last_pixel, \ ++ bilinear_process_two_pixels, \ ++ bilinear_process_four_pixels, \ ++ bilinear_process_pixblock_head, \ ++ bilinear_process_pixblock_tail, \ ++ bilinear_process_pixblock_tail_head, \ ++ pixblock_size, \ ++ prefetch_distance, \ ++ flags ++ ++pixman_asm_function fname ++.if pixblock_size == 8 ++.elseif pixblock_size == 4 ++.else ++ .error unsupported pixblock size ++.endif ++ ++.if ((flags) & BILINEAR_FLAG_USE_MASK) == 0 ++ OUT .req x0 ++ TOP .req x1 ++ BOTTOM .req x2 ++ WT .req x3 ++ WWT .req w3 ++ WB .req x4 ++ WWB .req w4 ++ X .req w5 ++ UX .req w6 ++ WIDTH .req x7 ++ TMP1 .req x10 ++ WTMP1 .req w10 ++ TMP2 .req x11 ++ WTMP2 .req w11 ++ PF_OFFS .req x12 ++ TMP3 .req x13 ++ WTMP3 .req w13 ++ TMP4 .req x14 ++ WTMP4 .req w14 ++ STRIDE .req x15 ++ DUMMY .req x30 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 112 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x10, x11, [x29, -80] ++ stp x12, x13, [x29, -96] ++ stp x14, x15, [x29, -112] ++.else ++ OUT .req x0 ++ MASK .req x1 ++ TOP .req x2 ++ BOTTOM .req x3 ++ WT .req x4 ++ WWT .req w4 ++ WB .req x5 ++ WWB .req w5 ++ X .req w6 ++ UX .req w7 ++ WIDTH .req x8 ++ TMP1 .req x10 ++ WTMP1 .req w10 ++ TMP2 .req x11 ++ WTMP2 .req w11 ++ PF_OFFS .req x12 ++ TMP3 .req x13 ++ WTMP3 .req w13 ++ TMP4 .req x14 ++ WTMP4 .req w14 ++ STRIDE .req x15 ++ DUMMY .req x30 ++ ++ .set prefetch_offset, prefetch_distance ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x10, x11, [x29, -80] ++ stp x12, x13, [x29, -96] ++ stp x14, x15, [x29, -112] ++ str x8, [x29, -120] ++ ldr w8, [x29, 16] ++ sub sp, sp, 120 ++.endif ++ ++ mov WTMP1, #prefetch_distance ++ umull PF_OFFS, WTMP1, UX ++ ++ sub STRIDE, BOTTOM, TOP ++ .unreq BOTTOM ++ ++ cmp WIDTH, #0 ++ ble 300f ++ ++ dup v12.8h, X ++ dup v13.8h, UX ++ dup v28.8b, WWT ++ dup v29.8b, WWB ++ mov v25.d[0], v12.d[1] ++ mov v26.d[0], v13.d[0] ++ add v25.4h, v25.4h, v26.4h ++ mov v12.d[1], v25.d[0] ++ ++ /* ensure good destination alignment */ ++ cmp WIDTH, #1 ++ blt 100f ++ tst OUT, #(1 << dst_bpp_shift) ++ beq 100f ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ bilinear_process_last_pixel ++ sub WIDTH, WIDTH, #1 ++100: ++ add v13.8h, v13.8h, v13.8h ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ++ cmp WIDTH, #2 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 1)) ++ beq 100f ++ bilinear_process_two_pixels ++ sub WIDTH, WIDTH, #2 ++100: ++.if pixblock_size == 8 ++ cmp WIDTH, #4 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 2)) ++ beq 100f ++ bilinear_process_four_pixels ++ sub WIDTH, WIDTH, #4 ++100: ++.endif ++ subs WIDTH, WIDTH, #pixblock_size ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_process_pixblock_head ++ subs WIDTH, WIDTH, #pixblock_size ++ blt 500f ++0: ++ bilinear_process_pixblock_tail_head ++ subs WIDTH, WIDTH, #pixblock_size ++ bge 0b ++500: ++ bilinear_process_pixblock_tail ++100: ++.if pixblock_size == 8 ++ tst WIDTH, #4 ++ beq 200f ++ bilinear_process_four_pixels ++200: ++.endif ++ /* handle the remaining trailing pixels */ ++ tst WIDTH, #2 ++ beq 200f ++ bilinear_process_two_pixels ++200: ++ tst WIDTH, #1 ++ beq 300f ++ bilinear_process_last_pixel ++300: ++ ++.if ((flags) & BILINEAR_FLAG_USE_MASK) == 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x10, x11, [x29, -80] ++ ldp x12, x13, [x29, -96] ++ ldp x14, x15, [x29, -112] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x10, x11, [x29, -80] ++ ldp x12, x13, [x29, -96] ++ ldp x14, x15, [x29, -112] ++ ldr x8, [x29, -120] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++.endif ++ ret ++ ++ .unreq OUT ++ .unreq TOP ++ .unreq WT ++ .unreq WWT ++ .unreq WB ++ .unreq WWB ++ .unreq X ++ .unreq UX ++ .unreq WIDTH ++ .unreq TMP1 ++ .unreq WTMP1 ++ .unreq TMP2 ++ .unreq PF_OFFS ++ .unreq TMP3 ++ .unreq TMP4 ++ .unreq STRIDE ++.if ((flags) & BILINEAR_FLAG_USE_MASK) != 0 ++ .unreq MASK ++.endif ++ ++.endfunc ++ ++.endm ++ ++/* src_8888_8_8888 */ ++.macro bilinear_src_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 8888, src ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_head ++ bilinear_src_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_8888_8_8888_process_pixblock_tail_head ++ bilinear_src_8888_8_8888_process_pixblock_tail ++ bilinear_src_8888_8_8888_process_pixblock_head ++.endm ++ ++/* src_8888_8_0565 */ ++.macro bilinear_src_8888_8_0565_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 0565, src ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_head ++ bilinear_src_8888_8_0565_process_four_pixels ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_8888_8_0565_process_pixblock_tail_head ++ bilinear_src_8888_8_0565_process_pixblock_tail ++ bilinear_src_8888_8_0565_process_pixblock_head ++.endm ++ ++/* src_0565_8_x888 */ ++.macro bilinear_src_0565_8_x888_process_last_pixel ++ bilinear_interpolate_last_pixel 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_two_pixels ++ bilinear_interpolate_two_pixels 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_four_pixels ++ bilinear_interpolate_four_pixels 0565, 8, 8888, src ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_head ++ bilinear_src_0565_8_x888_process_four_pixels ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_0565_8_x888_process_pixblock_tail_head ++ bilinear_src_0565_8_x888_process_pixblock_tail ++ bilinear_src_0565_8_x888_process_pixblock_head ++.endm ++ ++/* src_0565_8_0565 */ ++.macro bilinear_src_0565_8_0565_process_last_pixel ++ bilinear_interpolate_last_pixel 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_two_pixels ++ bilinear_interpolate_two_pixels 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_four_pixels ++ bilinear_interpolate_four_pixels 0565, 8, 0565, src ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_head ++ bilinear_src_0565_8_0565_process_four_pixels ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_tail ++.endm ++ ++.macro bilinear_src_0565_8_0565_process_pixblock_tail_head ++ bilinear_src_0565_8_0565_process_pixblock_tail ++ bilinear_src_0565_8_0565_process_pixblock_head ++.endm ++ ++/* over_8888_8888 */ ++.macro bilinear_over_8888_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, x, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_head ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ ++ ld1 {v22.2s}, [TMP1], STRIDE ++ ld1 {v23.2s}, [TMP1] ++ asr WTMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ umull v8.8h, v22.8b, v28.8b ++ umlal v8.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP2], STRIDE ++ ld1 {v23.2s}, [TMP2] ++ asr WTMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umull v9.8h, v22.8b, v28.8b ++ umlal v9.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_tail ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ xtn v7.8b, v2.8h ++ ld1 {v2.2s, v3.2s}, [OUT] ++ prfm PREFETCH_MODE, [OUT, #(prefetch_offset * 4)] ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ dup v4.2s, v7.s[1] ++ mvn v4.8b, v4.8b ++ umull v11.8h, v2.8b, v4.8b ++ umull v2.8h, v3.8b, v4.8b ++ urshr v1.8h, v11.8h, #8 ++ urshr v10.8h, v2.8h, #8 ++ raddhn v3.8b, v10.8h, v2.8h ++ raddhn v2.8b, v1.8h, v11.8h ++ uqadd v6.8b, v2.8b, v6.8b ++ uqadd v7.8b, v3.8b, v7.8b ++ vuzp v6.8b, v7.8b ++ vuzp v6.8b, v7.8b ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++.macro bilinear_over_8888_8888_process_pixblock_tail_head ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ asr WTMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ umlsl v2.4s, v10.4h, v15.h[0] ++ asr WTMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ ld1 {v20.2s}, [TMP1], STRIDE ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ ld1 {v21.2s}, [TMP1] ++ umull v8.8h, v20.8b, v28.8b ++ umlal v8.8h, v21.8b, v29.8b ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ld1 {v22.2s}, [TMP2], STRIDE ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ ld1 {v23.2s}, [TMP2] ++ umull v9.8h, v22.8b, v28.8b ++ asr WTMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ asr WTMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umlal v9.8h, v23.8b, v29.8b ++ xtn v7.8b, v2.8h ++ ld1 {v2.2s, v3.2s}, [OUT] ++ prfm PREFETCH_MODE, [OUT, PF_OFFS] ++ ld1 {v22.2s}, [TMP3], STRIDE ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ vuzp v6.8b, v7.8b ++ vuzp v2.8b, v3.8b ++ dup v4.2s, v7.s[1] ++ ld1 {v23.2s}, [TMP3] ++ mvn v4.8b, v4.8b ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ umull v11.8h, v2.8b, v4.8b ++ umull v2.8h, v3.8b, v4.8b ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ urshr v1.8h, v11.8h, #8 ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ urshr v8.8h, v2.8h, #8 ++ raddhn v3.8b, v8.8h, v2.8h ++ raddhn v2.8b, v1.8h, v11.8h ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ uqadd v6.8b, v2.8b, v6.8b ++ uqadd v7.8b, v3.8b, v7.8b ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ vuzp v6.8b, v7.8b ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ vuzp v6.8b, v7.8b ++ umlsl v1.4s, v9.4h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++/* over_8888_8_8888 */ ++.macro bilinear_over_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_four_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++ bilinear_interpolate_two_pixels 8888, 8, 8888, over ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_head ++ bilinear_over_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_over_8888_8_8888_process_pixblock_tail_head ++ bilinear_over_8888_8_8888_process_pixblock_tail ++ bilinear_over_8888_8_8888_process_pixblock_head ++.endm ++ ++/* add_8888_8888 */ ++.macro bilinear_add_8888_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_four_pixels ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++ bilinear_interpolate_two_pixels 8888, x, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_head ++ bilinear_add_8888_8888_process_four_pixels ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_add_8888_8888_process_pixblock_tail_head ++ bilinear_add_8888_8888_process_pixblock_tail ++ bilinear_add_8888_8888_process_pixblock_head ++.endm ++ ++/* add_8888_8_8888 */ ++.macro bilinear_add_8888_8_8888_process_last_pixel ++ bilinear_interpolate_last_pixel 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_two_pixels ++ bilinear_interpolate_two_pixels 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_four_pixels ++ bilinear_interpolate_four_pixels 8888, 8, 8888, add ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_head ++ bilinear_add_8888_8_8888_process_four_pixels ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_tail ++.endm ++ ++.macro bilinear_add_8888_8_8888_process_pixblock_tail_head ++ bilinear_add_8888_8_8888_process_pixblock_tail ++ bilinear_add_8888_8_8888_process_pixblock_head ++.endm ++ ++ ++/* Bilinear scanline functions */ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_SRC_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_src_8888_8_8888_process_last_pixel, \ ++ bilinear_src_8888_8_8888_process_two_pixels, \ ++ bilinear_src_8888_8_8888_process_four_pixels, \ ++ bilinear_src_8888_8_8888_process_pixblock_head, \ ++ bilinear_src_8888_8_8888_process_pixblock_tail, \ ++ bilinear_src_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_0565_SRC_asm_neon, \ ++ 8888, 0565, 2, 1, \ ++ bilinear_src_8888_8_0565_process_last_pixel, \ ++ bilinear_src_8888_8_0565_process_two_pixels, \ ++ bilinear_src_8888_8_0565_process_four_pixels, \ ++ bilinear_src_8888_8_0565_process_pixblock_head, \ ++ bilinear_src_8888_8_0565_process_pixblock_tail, \ ++ bilinear_src_8888_8_0565_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_8_x888_SRC_asm_neon, \ ++ 0565, 8888, 1, 2, \ ++ bilinear_src_0565_8_x888_process_last_pixel, \ ++ bilinear_src_0565_8_x888_process_two_pixels, \ ++ bilinear_src_0565_8_x888_process_four_pixels, \ ++ bilinear_src_0565_8_x888_process_pixblock_head, \ ++ bilinear_src_0565_8_x888_process_pixblock_tail, \ ++ bilinear_src_0565_8_x888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_8_0565_SRC_asm_neon, \ ++ 0565, 0565, 1, 1, \ ++ bilinear_src_0565_8_0565_process_last_pixel, \ ++ bilinear_src_0565_8_0565_process_two_pixels, \ ++ bilinear_src_0565_8_0565_process_four_pixels, \ ++ bilinear_src_0565_8_0565_process_pixblock_head, \ ++ bilinear_src_0565_8_0565_process_pixblock_tail, \ ++ bilinear_src_0565_8_0565_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_OVER_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_over_8888_8888_process_last_pixel, \ ++ bilinear_over_8888_8888_process_two_pixels, \ ++ bilinear_over_8888_8888_process_four_pixels, \ ++ bilinear_over_8888_8888_process_pixblock_head, \ ++ bilinear_over_8888_8888_process_pixblock_tail, \ ++ bilinear_over_8888_8888_process_pixblock_tail_head, \ ++ 4, 28, 0 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_OVER_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_over_8888_8_8888_process_last_pixel, \ ++ bilinear_over_8888_8_8888_process_two_pixels, \ ++ bilinear_over_8888_8_8888_process_four_pixels, \ ++ bilinear_over_8888_8_8888_process_pixblock_head, \ ++ bilinear_over_8888_8_8888_process_pixblock_tail, \ ++ bilinear_over_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_ADD_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_add_8888_8888_process_last_pixel, \ ++ bilinear_add_8888_8888_process_two_pixels, \ ++ bilinear_add_8888_8888_process_four_pixels, \ ++ bilinear_add_8888_8888_process_pixblock_head, \ ++ bilinear_add_8888_8888_process_pixblock_tail, \ ++ bilinear_add_8888_8888_process_pixblock_tail_head, \ ++ 4, 28, 0 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8_8888_ADD_asm_neon, \ ++ 8888, 8888, 2, 2, \ ++ bilinear_add_8888_8_8888_process_last_pixel, \ ++ bilinear_add_8888_8_8888_process_two_pixels, \ ++ bilinear_add_8888_8_8888_process_four_pixels, \ ++ bilinear_add_8888_8_8888_process_pixblock_head, \ ++ bilinear_add_8888_8_8888_process_pixblock_tail, \ ++ bilinear_add_8888_8_8888_process_pixblock_tail_head, \ ++ 4, 28, BILINEAR_FLAG_USE_MASK +diff --git a/pixman/pixman-arma64-neon-asm.S b/pixman/pixman-arma64-neon-asm.S +new file mode 100644 +index 0000000..18ace0e +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm.S +@@ -0,0 +1,3704 @@ ++/* ++ * Copyright © 2009 Nokia Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ */ ++ ++/* ++ * This file contains implementations of NEON optimized pixel processing ++ * functions. There is no full and detailed tutorial, but some functions ++ * (those which are exposing some new or interesting features) are ++ * extensively commented and can be used as examples. ++ * ++ * You may want to have a look at the comments for following functions: ++ * - pixman_composite_over_8888_0565_asm_neon ++ * - pixman_composite_over_n_8_0565_asm_neon ++ */ ++ ++/* Prevent the stack from becoming executable for no reason... */ ++#if defined(__linux__) && defined(__ELF__) ++.section .note.GNU-stack,"",%progbits ++#endif ++ ++.text ++.arch armv8-a ++ ++.altmacro ++.p2align 2 ++ ++#include "pixman-private.h" ++#include "pixman-arm-asm.h" ++#include "pixman-arma64-neon-asm.h" ++ ++/* Global configuration options and preferences */ ++ ++/* ++ * The code can optionally make use of unaligned memory accesses to improve ++ * performance of handling leading/trailing pixels for each scanline. ++ * Configuration variable RESPECT_STRICT_ALIGNMENT can be set to 0 for ++ * example in linux if unaligned memory accesses are not configured to ++ * generate.exceptions. ++ */ ++.set RESPECT_STRICT_ALIGNMENT, 1 ++ ++/* ++ * Set default prefetch type. There is a choice between the following options: ++ * ++ * PREFETCH_TYPE_NONE (may be useful for the ARM cores where PLD is set to work ++ * as NOP to workaround some HW bugs or for whatever other reason) ++ * ++ * PREFETCH_TYPE_SIMPLE (may be useful for simple single-issue ARM cores where ++ * advanced prefetch intruduces heavy overhead) ++ * ++ * PREFETCH_TYPE_ADVANCED (useful for superscalar cores such as ARM Cortex-A8 ++ * which can run ARM and NEON instructions simultaneously so that extra ARM ++ * instructions do not add (many) extra cycles, but improve prefetch efficiency) ++ * ++ * Note: some types of function can't support advanced prefetch and fallback ++ * to simple one (those which handle 24bpp pixels) ++ */ ++.set PREFETCH_TYPE_DEFAULT, PREFETCH_TYPE_ADVANCED ++ ++/* Prefetch distance in pixels for simple prefetch */ ++.set PREFETCH_DISTANCE_SIMPLE, 64 ++ ++/* ++ * Implementation of pixman_composite_over_8888_0565_asm_neon ++ * ++ * This function takes a8r8g8b8 source buffer, r5g6b5 destination buffer and ++ * performs OVER compositing operation. Function fast_composite_over_8888_0565 ++ * from pixman-fast-path.c does the same in C and can be used as a reference. ++ * ++ * First we need to have some NEON assembly code which can do the actual ++ * operation on the pixels and provide it to the template macro. ++ * ++ * Template macro quite conveniently takes care of emitting all the necessary ++ * code for memory reading and writing (including quite tricky cases of ++ * handling unaligned leading/trailing pixels), so we only need to deal with ++ * the data in NEON registers. ++ * ++ * NEON registers allocation in general is recommented to be the following: ++ * v0, v1, v2, v3 - contain loaded source pixel data ++ * v4, v5, v6, v7 - contain loaded destination pixels (if they are needed) ++ * v24, v25, v26, v27 - contain loading mask pixel data (if mask is used) ++ * v28, v29, v30, v31 - place for storing the result (destination pixels) ++ * ++ * As can be seen above, four 64-bit NEON registers are used for keeping ++ * intermediate pixel data and up to 8 pixels can be processed in one step ++ * for 32bpp formats (16 pixels for 16bpp, 32 pixels for 8bpp). ++ * ++ * This particular function uses the following registers allocation: ++ * v0, v1, v2, v3 - contain loaded source pixel data ++ * v4, v5 - contain loaded destination pixels (they are needed) ++ * v28, v29 - place for storing the result (destination pixels) ++ */ ++ ++/* ++ * Step one. We need to have some code to do some arithmetics on pixel data. ++ * This is implemented as a pair of macros: '*_head' and '*_tail'. When used ++ * back-to-back, they take pixel data from {v0, v1, v2, v3} and {v4, v5}, ++ * perform all the needed calculations and write the result to {v28, v29}. ++ * The rationale for having two macros and not just one will be explained ++ * later. In practice, any single monolitic function which does the work can ++ * be split into two parts in any arbitrary way without affecting correctness. ++ * ++ * There is one special trick here too. Common template macro can optionally ++ * make our life a bit easier by doing R, G, B, A color components ++ * deinterleaving for 32bpp pixel formats (and this feature is used in ++ * 'pixman_composite_over_8888_0565_asm_neon' function). So it means that ++ * instead of having 8 packed pixels in {v0, v1, v2, v3} registers, we ++ * actually use v0 register for blue channel (a vector of eight 8-bit ++ * values), v1 register for green, v2 for red and v3 for alpha. This ++ * simple conversion can be also done with a few NEON instructions: ++ * ++ * Packed to planar conversion: // vuzp8 is a wrapper macro ++ * vuzp8 v0, v1 ++ * vuzp8 v2, v3 ++ * vuzp8 v1, v3 ++ * vuzp8 v0, v2 ++ * ++ * Planar to packed conversion: // vzip8 is a wrapper macro ++ * vzip8 v0, v2 ++ * vzip8 v1, v3 ++ * vzip8 v2, v3 ++ * vzip8 v0, v1 ++ * ++ * But pixel can be loaded directly in planar format using LD4 / b NEON ++ * instruction. It is 1 cycle slower than LD1 / s, so this is not always ++ * desirable, that's why deinterleaving is optional. ++ * ++ * But anyway, here is the code: ++ */ ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_head ++ /* convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ and put data into v6 - red, v7 - green, v30 - blue */ ++ mov v4.d[1], v5.d[0] ++ shrn v6.8b, v4.8h, #8 ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ sri v6.8b, v6.8b, #5 ++ mvn v3.8b, v3.8b /* invert source alpha */ ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ /* now do alpha blending, storing results in 8-bit planar format ++ into v20 - red, v23 - green, v22 - blue */ ++ umull v10.8h, v3.8b, v6.8b ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ urshr v17.8h, v10.8h, #8 ++ urshr v18.8h, v11.8h, #8 ++ urshr v19.8h, v12.8h, #8 ++ raddhn v20.8b, v10.8h, v17.8h ++ raddhn v23.8b, v11.8h, v18.8h ++ raddhn v22.8b, v12.8h, v19.8h ++.endm ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail ++ /* ... continue alpha blending */ ++ uqadd v17.8b, v2.8b, v20.8b ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ /* convert the result to r5g6b5 and store it into {v14} */ ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* ++ * OK, now we got almost everything that we need. Using the above two ++ * macros, the work can be done right. But now we want to optimize ++ * it a bit. ARM Cortex-A8 is an in-order core, and benefits really ++ * a lot from good code scheduling and software pipelining. ++ * ++ * Let's construct some code, which will run in the core main loop. ++ * Some pseudo-code of the main loop will look like this: ++ * head ++ * while (...) { ++ * tail ++ * head ++ * } ++ * tail ++ * ++ * It may look a bit weird, but this setup allows to hide instruction ++ * latencies better and also utilize dual-issue capability more ++ * efficiently (make pairs of load-store and ALU instructions). ++ * ++ * So what we need now is a '*_tail_head' macro, which will be used ++ * in the core main loop. A trivial straightforward implementation ++ * of this macro would look like this: ++ * ++ * pixman_composite_over_8888_0565_process_pixblock_tail ++ * st1 {v28.4h, v29.4h}, [DST_W], #32 ++ * ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ * ld4 {v0.2s, v1.2s, v2.2s, v3.2s}, [SRC], #32 ++ * pixman_composite_over_8888_0565_process_pixblock_head ++ * cache_preload 8, 8 ++ * ++ * Now it also got some VLD/VST instructions. We simply can't move from ++ * processing one block of pixels to the other one with just arithmetics. ++ * The previously processed data needs to be written to memory and new ++ * data needs to be fetched. Fortunately, this main loop does not deal ++ * with partial leading/trailing pixels and can load/store a full block ++ * of pixels in a bulk. Additionally, destination buffer is already ++ * 16 bytes aligned here (which is good for performance). ++ * ++ * New things here are DST_R, DST_W, SRC and MASK identifiers. These ++ * are the aliases for ARM registers which are used as pointers for ++ * accessing data. We maintain separate pointers for reading and writing ++ * destination buffer (DST_R and DST_W). ++ * ++ * Another new thing is 'cache_preload' macro. It is used for prefetching ++ * data into CPU L2 cache and improve performance when dealing with large ++ * images which are far larger than cache size. It uses one argument ++ * (actually two, but they need to be the same here) - number of pixels ++ * in a block. Looking into 'pixman-arm-neon-asm.h' can provide some ++ * details about this macro. Moreover, if good performance is needed ++ * the code from this macro needs to be copied into '*_tail_head' macro ++ * and mixed with the rest of code for optimal instructions scheduling. ++ * We are actually doing it below. ++ * ++ * Now after all the explanations, here is the optimized code. ++ * Different instruction streams (originaling from '*_head', '*_tail' ++ * and 'cache_preload' macro) use different indentation levels for ++ * better readability. Actually taking the code from one of these ++ * indentation levels and ignoring a few LD/ST instructions would ++ * result in exactly the code from '*_head', '*_tail' or 'cache_preload' ++ * macro! ++ */ ++ ++#if 1 ++ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail_head ++ uqadd v17.8b, v2.8b, v20.8b ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ mov v4.d[1], v5.d[0] ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ shrn v6.8b, v4.8h, #8 ++ fetch_src_pixblock ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ PF add PF_X, PF_X, #8 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ PF tst PF_CTL, #0xF ++ sri v6.8b, v6.8b, #5 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ mvn v3.8b, v3.8b ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ umull v10.8h, v3.8b, v6.8b ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ sri v14.8h, v8.8h, #5 ++ PF cmp PF_X, ORIG_W ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ urshr v17.8h, v10.8h, #8 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ urshr v19.8h, v11.8h, #8 ++ urshr v18.8h, v12.8h, #8 ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ raddhn v20.8b, v10.8h, v17.8h ++ raddhn v23.8b, v11.8h, v19.8h ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_SRC, #1 ++10: ++ raddhn v22.8b, v12.8h, v18.8h ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++#else ++ ++/* If we did not care much about the performance, we would just use this... */ ++.macro pixman_composite_over_8888_0565_process_pixblock_tail_head ++ pixman_composite_over_8888_0565_process_pixblock_tail ++ st1 {v14.8h}, [DST_W], #16 ++ ld1 {v4.4h, v4.5h}, [DST_R], #16 ++ fetch_src_pixblock ++ pixman_composite_over_8888_0565_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++#endif ++ ++/* ++ * And now the final part. We are using 'generate_composite_function' macro ++ * to put all the stuff together. We are specifying the name of the function ++ * which we want to get, number of bits per pixel for the source, mask and ++ * destination (0 if unused, like mask in this case). Next come some bit ++ * flags: ++ * FLAG_DST_READWRITE - tells that the destination buffer is both read ++ * and written, for write-only buffer we would use ++ * FLAG_DST_WRITEONLY flag instead ++ * FLAG_DEINTERLEAVE_32BPP - tells that we prefer to work with planar data ++ * and separate color channels for 32bpp format. ++ * The next things are: ++ * - the number of pixels processed per iteration (8 in this case, because ++ * that's the maximum what can fit into four 64-bit NEON registers). ++ * - prefetch distance, measured in pixel blocks. In this case it is 5 times ++ * by 8 pixels. That would be 40 pixels, or up to 160 bytes. Optimal ++ * prefetch distance can be selected by running some benchmarks. ++ * ++ * After that we specify some macros, these are 'default_init', ++ * 'default_cleanup' here which are empty (but it is possible to have custom ++ * init/cleanup macros to be able to save/restore some extra NEON registers ++ * like d8-d15 or do anything else) followed by ++ * 'pixman_composite_over_8888_0565_process_pixblock_head', ++ * 'pixman_composite_over_8888_0565_process_pixblock_tail' and ++ * 'pixman_composite_over_8888_0565_process_pixblock_tail_head' ++ * which we got implemented above. ++ * ++ * The last part is the NEON registers allocation scheme. ++ */ ++generate_composite_function \ ++ pixman_composite_over_8888_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_0565_process_pixblock_head ++ /* convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ and put data into v6 - red, v7 - green, v30 - blue */ ++ mov v4.d[1], v5.d[0] ++ shrn v6.8b, v4.8h, #8 ++ shrn v7.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ sri v6.8b, v6.8b, #5 ++ sri v7.8b, v7.8b, #6 ++ shrn v30.8b, v4.8h, #2 ++ /* now do alpha blending, storing results in 8-bit planar format ++ into v20 - red, v23 - green, v22 - blue */ ++ umull v10.8h, v3.8b, v6.8b ++ umull v11.8h, v3.8b, v7.8b ++ umull v12.8h, v3.8b, v30.8b ++ urshr v13.8h, v10.8h, #8 ++ urshr v14.8h, v11.8h, #8 ++ urshr v15.8h, v12.8h, #8 ++ raddhn v20.8b, v10.8h, v13.8h ++ raddhn v23.8b, v11.8h, v14.8h ++ raddhn v22.8b, v12.8h, v15.8h ++.endm ++ ++.macro pixman_composite_over_n_0565_process_pixblock_tail ++ /* ... continue alpha blending */ ++ uqadd v17.8b, v2.8b, v20.8b ++ uqadd v18.8b, v0.8b, v22.8b ++ uqadd v19.8b, v1.8b, v23.8b ++ /* convert the result to r5g6b5 and store it into {v14} */ ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v8.8h, v19.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v18.8b, #7 ++ sli v9.8h, v9.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_n_0565_process_pixblock_tail_head ++ pixman_composite_over_n_0565_process_pixblock_tail ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ st1 {v14.8h}, [DST_W], #16 ++ pixman_composite_over_n_0565_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_over_n_0565_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++ mvn v3.8b, v3.8b /* invert source alpha */ ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_0565_asm_neon, 0, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_0565_init, \ ++ default_cleanup, \ ++ pixman_composite_over_n_0565_process_pixblock_head, \ ++ pixman_composite_over_n_0565_process_pixblock_tail, \ ++ pixman_composite_over_n_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_head ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v14.8h, v2.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v9.8h, v0.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_tail ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_src_8888_0565_process_pixblock_tail_head ++ sri v14.8h, v8.8h, #5 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ fetch_src_pixblock ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ st1 {v14.8h}, [DST_W], #16 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ ushll v14.8h, v2.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ ushll v9.8h, v0.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_8888_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_0565_process_pixblock_head, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0565_8888_process_pixblock_head ++ mov v0.d[1], v1.d[0] ++ shrn v30.8b, v0.8h, #8 ++ shrn v29.8b, v0.8h, #3 ++ sli v0.8h, v0.8h, #5 ++ movi v31.8b, #255 ++ sri v30.8b, v30.8b, #5 ++ sri v29.8b, v29.8b, #6 ++ shrn v28.8b, v0.8h, #2 ++.endm ++ ++.macro pixman_composite_src_0565_8888_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_src_0565_8888_process_pixblock_tail_head ++ pixman_composite_src_0565_8888_process_pixblock_tail ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ fetch_src_pixblock ++ pixman_composite_src_0565_8888_process_pixblock_head ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0565_8888_asm_neon, 16, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_8888_process_pixblock_head, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8_8_process_pixblock_head ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_8_8_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_add_8_8_process_pixblock_tail_head ++ fetch_src_pixblock ++ PF add PF_X, PF_X, #32 ++ PF tst PF_CTL, #0xF ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ PF beq 10f ++ PF add PF_X, PF_X, #32 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ uqadd v28.8b, v0.8b, v4.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8_8_asm_neon, 8, 0, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ uqadd v28.8b, v0.8b, v4.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_add_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_head ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++ /* do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++.endm ++ ++.macro pixman_composite_out_reverse_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ fetch_src_pixblock ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ umull v10.8h, v22.8b, v6.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_out_reverse_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_head ++ pixman_composite_out_reverse_8888_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_tail ++ pixman_composite_out_reverse_8888_8888_process_pixblock_tail ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ fetch_src_pixblock ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++ umull v10.8h, v22.8b, v6.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_over_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_process_pixblock_head ++ /* deinterleaved source pixels in {v0, v1, v2, v3} */ ++ /* inverted alpha in {v24} */ ++ /* destination pixels in {v4, v5, v6, v7} */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_process_pixblock_tail_head ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v16.8h, v10.8h, #8 ++ urshr v17.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v16.8h, v10.8h ++ raddhn v31.8b, v17.8h, v11.8h ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ uqadd v28.8b, v0.8b, v28.8b ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0x0F ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ PF cmp PF_X, ORIG_W ++ umull v8.8h, v24.8b, v4.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ umull v9.8h, v24.8b, v5.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v10.8h, v24.8b, v6.8b ++ PF subs PF_CTL, PF_CTL, #0x10 ++ umull v11.8h, v24.8b, v7.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_n_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_reverse_n_8888_process_pixblock_tail_head ++ urshr v14.8h, v8.8h, #8 ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v30.8b, v12.8h, v10.8h ++ raddhn v31.8b, v13.8h, v11.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ ld4 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_R], #32 ++ mvn v22.8b, v3.8b ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF blt 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v8.8h, v22.8b, v4.8b ++ PF blt 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v9.8h, v22.8b, v5.8b ++ umull v10.8h, v22.8b, v6.8b ++ PF blt 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ umull v11.8h, v22.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_reverse_n_8888_init ++ mov v7.s[0], w4 ++ dup v4.8b, v7.b[0] ++ dup v5.8b, v7.b[1] ++ dup v6.8b, v7.b[2] ++ dup v7.8b, v7.b[3] ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_reverse_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_reverse_n_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_reverse_n_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 4, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_head ++ umull v0.8h, v24.8b, v8.8b /* IN for SRC pixels (part1) */ ++ umull v1.8h, v24.8b, v9.8b ++ umull v2.8h, v24.8b, v10.8b ++ umull v3.8h, v24.8b, v11.8b ++ mov v4.d[1], v5.d[0] ++ shrn v25.8b, v4.8h, #8 /* convert DST_R data to 32-bpp (part1) */ ++ shrn v26.8b, v4.8h, #3 ++ sli v4.8h, v4.8h, #5 ++ urshr v17.8h, v0.8h, #8 /* IN for SRC pixels (part2) */ ++ urshr v18.8h, v1.8h, #8 ++ urshr v19.8h, v2.8h, #8 ++ urshr v20.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v17.8h ++ raddhn v1.8b, v1.8h, v18.8h ++ raddhn v2.8b, v2.8h, v19.8h ++ raddhn v3.8b, v3.8h, v20.8h ++ sri v25.8b, v25.8b, #5 /* convert DST_R data to 32-bpp (part2) */ ++ sri v26.8b, v26.8b, #6 ++ mvn v3.8b, v3.8b ++ shrn v30.8b, v4.8h, #2 ++ umull v18.8h, v3.8b, v25.8b /* now do alpha blending */ ++ umull v19.8h, v3.8b, v26.8b ++ umull v20.8h, v3.8b, v30.8b ++.endm ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_tail ++ /* 3 cycle bubble (after vmull.u8) */ ++ urshr v5.8h, v18.8h, #8 ++ urshr v6.8h, v19.8h, #8 ++ urshr v7.8h, v20.8h, #8 ++ raddhn v17.8b, v18.8h, v5.8h ++ raddhn v19.8b, v19.8h, v6.8h ++ raddhn v18.8b, v20.8h, v7.8h ++ uqadd v5.8b, v2.8b, v17.8b ++ /* 1 cycle bubble */ ++ uqadd v6.8b, v0.8b, v18.8b ++ uqadd v7.8b, v1.8b, v19.8b ++ ushll v14.8h, v5.8b, #7 /* convert to 16bpp */ ++ sli v14.8h, v14.8h, #1 ++ ushll v18.8h, v7.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v6.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ sri v14.8h, v18.8h, #5 ++ /* 1 cycle bubble */ ++ sri v14.8h, v19.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_over_8888_8_0565_process_pixblock_tail_head ++#if 0 ++ ld1 {v4.8h}, [DST_R], #16 ++ shrn v25.8b, v4.8h, #8 ++ fetch_mask_pixblock ++ shrn v26.8b, v4.8h, #3 ++ fetch_src_pixblock ++ umull v22.8h, v24.8b, v10.8b ++ urshr v13.8h, v18.8h, #8 ++ urshr v11.8h, v19.8h, #8 ++ urshr v15.8h, v20.8h, #8 ++ raddhn v17.8b, v18.8h, v13.8h ++ raddhn v19.8b, v19.8h, v11.8h ++ raddhn v18.8b, v20.8h, v15.8h ++ uqadd v17.8b, v2.8b, v17.8b ++ umull v21.8h, v24.8b, v9.8b ++ uqadd v18.8b, v0.8b, v18.8b ++ uqadd v19.8b, v1.8b, v19.8b ++ ushll v14.8h, v17.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ umull v20.8h, v24.8b, v8.8b ++ ushll v18.8h, v18.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v19.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ sri v14.8h, v18.8h, #5 ++ umull v23.8h, v24.8b, v11.8b ++ sri v14.8h, v19.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ ++ cache_preload 8, 8 ++ ++ sli v4.8h, v4.8h, #5 ++ urshr v16.8h, v20.8h, #8 ++ urshr v17.8h, v21.8h, #8 ++ urshr v18.8h, v22.8h, #8 ++ urshr v19.8h, v23.8h, #8 ++ raddhn v0.8b, v20.8h, v16.8h ++ raddhn v1.8b, v21.8h, v17.8h ++ raddhn v2.8b, v22.8h, v18.8h ++ raddhn v3.8b, v23.8h, v19.8h ++ sri v25.8b, v25.8b, #5 ++ sri v26.8b, v26.8b, #6 ++ mvn v3.8b, v3.8b ++ shrn v30.8b, v4.8h, #2 ++ st1 {v14.8h}, [DST_W], #16 ++ umull v18.8h, v3.8b, v25.8b ++ umull v19.8h, v3.8b, v26.8b ++ umull v20.8h, v3.8b, v30.8b ++#else ++ pixman_composite_over_8888_8_0565_process_pixblock_tail ++ st1 {v28.4h, v29.4h}, [DST_W], #16 ++ ld1 {v4.4h, v5.4h}, [DST_R], #16 ++ fetch_mask_pixblock ++ fetch_src_pixblock ++ pixman_composite_over_8888_8_0565_process_pixblock_head ++#endif ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8_0565_asm_neon, 32, 8, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* ++ * This function needs a special initialization of solid mask. ++ * Solid source pixel data is fetched from stack at ARGS_STACK_OFFSET ++ * offset, split into color components and replicated in d8-d11 ++ * registers. Additionally, this function needs all the NEON registers, ++ * so it has to save d8-d15 registers which are callee saved according ++ * to ABI. These registers are restored from 'cleanup' macro. All the ++ * other NEON registers are caller saved, so can be clobbered freely ++ * without introducing any problems. ++ */ ++.macro pixman_composite_over_n_8_0565_init ++ mov v11.s[0], w4 ++ dup v8.8b, v11.b[0] ++ dup v9.8b, v11.b[1] ++ dup v10.8b, v11.b[2] ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_0565_asm_neon, 0, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_0565_init, \ ++ pixman_composite_over_n_8_0565_cleanup, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_n_0565_init ++ mov v24.s[0], w6 ++ dup v24.8b, v24.b[3] ++.endm ++ ++.macro pixman_composite_over_8888_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_n_0565_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_8888_n_0565_init, \ ++ pixman_composite_over_8888_n_0565_cleanup, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0565_0565_process_pixblock_tail_head ++ st1 {v0.4h, v1.4h, v2.4h, v3.4h}, [DST_W], #32 ++ fetch_src_pixblock ++ cache_preload 16, 16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0565_0565_asm_neon, 16, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 16, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_0565_process_pixblock_head, \ ++ pixman_composite_src_0565_0565_process_pixblock_tail, \ ++ pixman_composite_src_0565_0565_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_8_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_8_process_pixblock_tail_head ++ st1 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_W], 32 ++.endm ++ ++.macro pixman_composite_src_n_8_init ++ mov v0.s[0], w4 ++ dup v3.8b, v0.b[0] ++ dup v2.8b, v0.b[0] ++ dup v1.8b, v0.b[0] ++ dup v0.8b, v0.b[0] ++.endm ++ ++.macro pixman_composite_src_n_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_asm_neon, 0, 0, 8, \ ++ FLAG_DST_WRITEONLY, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_8_init, \ ++ pixman_composite_src_n_8_cleanup, \ ++ pixman_composite_src_n_8_process_pixblock_head, \ ++ pixman_composite_src_n_8_process_pixblock_tail, \ ++ pixman_composite_src_n_8_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_0565_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_0565_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_0565_process_pixblock_tail_head ++ st1 {v0.4h, v1.4h, v2.4h, v3.4h}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_src_n_0565_init ++ mov v0.s[0], w4 ++ dup v3.4h, v0.h[0] ++ dup v2.4h, v0.h[0] ++ dup v1.4h, v0.h[0] ++ dup v0.4h, v0.h[0] ++.endm ++ ++.macro pixman_composite_src_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_0565_asm_neon, 0, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 16, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_0565_init, \ ++ pixman_composite_src_n_0565_cleanup, \ ++ pixman_composite_src_n_0565_process_pixblock_head, \ ++ pixman_composite_src_n_0565_process_pixblock_tail, \ ++ pixman_composite_src_n_0565_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_n_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_n_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_src_n_8888_init ++ mov v0.s[0], w4 ++ dup v3.2s, v0.s[0] ++ dup v2.2s, v0.s[0] ++ dup v1.2s, v0.s[0] ++ dup v0.2s, v0.s[0] ++.endm ++ ++.macro pixman_composite_src_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8888_asm_neon, 0, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 0, /* prefetch distance */ \ ++ pixman_composite_src_n_8888_init, \ ++ pixman_composite_src_n_8888_cleanup, \ ++ pixman_composite_src_n_8888_process_pixblock_head, \ ++ pixman_composite_src_n_8888_process_pixblock_tail, \ ++ pixman_composite_src_n_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_8888_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++ fetch_src_pixblock ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_8888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_8888_process_pixblock_head, \ ++ pixman_composite_src_8888_8888_process_pixblock_tail, \ ++ pixman_composite_src_8888_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_head ++ orr v0.8b, v0.8b, v4.8b ++ orr v1.8b, v1.8b, v4.8b ++ orr v2.8b, v2.8b, v4.8b ++ orr v3.8b, v3.8b, v4.8b ++.endm ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_x888_8888_process_pixblock_tail_head ++ st1 {v0.2s, v1.2s, v2.2s, v3.2s}, [DST_W], #32 ++ fetch_src_pixblock ++ orr v0.8b, v0.8b, v4.8b ++ orr v1.8b, v1.8b, v4.8b ++ orr v2.8b, v2.8b, v4.8b ++ orr v3.8b, v3.8b, v4.8b ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_src_x888_8888_init ++ movi v4.2s, #0xff, lsl 24 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_x888_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ pixman_composite_src_x888_8888_init, \ ++ default_cleanup, \ ++ pixman_composite_src_x888_8888_process_pixblock_head, \ ++ pixman_composite_src_x888_8888_process_pixblock_tail, \ ++ pixman_composite_src_x888_8888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_head ++ /* expecting solid source in {v0, v1, v2, v3} */ ++ /* mask is in v24 (v25, v26, v27 are unused) */ ++ ++ /* in */ ++ umull v8.8h, v24.8b, v0.8b ++ umull v9.8h, v24.8b, v1.8b ++ umull v10.8h, v24.8b, v2.8b ++ umull v11.8h, v24.8b, v3.8b ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_tail ++ rshrn v28.8b, v8.8h, #8 ++ rshrn v29.8b, v9.8h, #8 ++ rshrn v30.8b, v10.8h, #8 ++ rshrn v31.8b, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_process_pixblock_tail_head ++ fetch_mask_pixblock ++ PF add PF_X, PF_X, #8 ++ rshrn v28.8b, v8.8h, #8 ++ PF tst PF_CTL, #0x0F ++ rshrn v29.8b, v9.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ rshrn v30.8b, v10.8h, #8 ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ rshrn v31.8b, v11.8h, #8 ++ PF cmp PF_X, ORIG_W ++ umull v8.8h, v24.8b, v0.8b ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v9.8h, v24.8b, v1.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v10.8h, v24.8b, v2.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v11.8h, v24.8b, v3.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_src_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_src_n_8_8888_init, \ ++ pixman_composite_src_n_8_8888_cleanup, \ ++ pixman_composite_src_n_8_8888_process_pixblock_head, \ ++ pixman_composite_src_n_8_8888_process_pixblock_tail, \ ++ pixman_composite_src_n_8_8888_process_pixblock_tail_head, \ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_head ++ umull v0.8h, v24.8b, v16.8b ++ umull v1.8h, v25.8b, v16.8b ++ umull v2.8h, v26.8b, v16.8b ++ umull v3.8h, v27.8b, v16.8b ++ ursra v0.8h, v0.8h, #8 ++ ursra v1.8h, v1.8h, #8 ++ ursra v2.8h, v2.8h, #8 ++ ursra v3.8h, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_tail ++ rshrn v28.8b, v0.8h, #8 ++ rshrn v29.8b, v1.8h, #8 ++ rshrn v30.8b, v2.8h, #8 ++ rshrn v31.8b, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_process_pixblock_tail_head ++ fetch_mask_pixblock ++ PF add PF_X, PF_X, #8 ++ rshrn v28.8b, v0.8h, #8 ++ PF tst PF_CTL, #0x0F ++ rshrn v29.8b, v1.8h, #8 ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ rshrn v30.8b, v2.8h, #8 ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ rshrn v31.8b, v3.8h, #8 ++ PF cmp PF_X, ORIG_W ++ umull v0.8h, v24.8b, v16.8b ++ PF lsl DUMMY, PF_X, mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v1.8h, v25.8b, v16.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v2.8h, v26.8b, v16.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v3.8h, v27.8b, v16.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v0.8h, v0.8h, #8 ++ ursra v1.8h, v1.8h, #8 ++ ursra v2.8h, v2.8h, #8 ++ ursra v3.8h, v3.8h, #8 ++.endm ++ ++.macro pixman_composite_src_n_8_8_init ++ mov v16.s[0], w4 ++ dup v16.8b, v16.b[3] ++.endm ++ ++.macro pixman_composite_src_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_WRITEONLY, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_src_n_8_8_init, \ ++ pixman_composite_src_n_8_8_cleanup, \ ++ pixman_composite_src_n_8_8_process_pixblock_head, \ ++ pixman_composite_src_n_8_8_process_pixblock_tail, \ ++ pixman_composite_src_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_head ++ /* expecting deinterleaved source data in {v8, v9, v10, v11} */ ++ /* v8 - blue, v9 - green, v10 - red, v11 - alpha */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ /* mask is in v24 (v25, v26, v27 are unused) */ ++ ++ /* in */ ++ umull v12.8h, v24.8b, v8.8b ++ umull v13.8h, v24.8b, v9.8b ++ umull v14.8h, v24.8b, v10.8b ++ umull v15.8h, v24.8b, v11.8b ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v0.8b, v12.8h, v16.8h ++ raddhn v1.8b, v13.8h, v17.8h ++ raddhn v2.8b, v14.8h, v18.8h ++ raddhn v3.8b, v15.8h, v19.8h ++ mvn v25.8b, v3.8b /* get inverted alpha */ ++ /* source: v0 - blue, v1 - green, v2 - red, v3 - alpha */ ++ /* destination: v4 - blue, v5 - green, v6 - red, v7 - alpha */ ++ /* now do alpha blending */ ++ umull v12.8h, v25.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v25.8b, v6.8b ++ umull v15.8h, v25.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_tail ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_process_pixblock_tail_head ++ urshr v16.8h, v12.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v17.8h, v13.8h, #8 ++ fetch_mask_pixblock ++ urshr v18.8h, v14.8h, #8 ++ PF add PF_X, PF_X, #8 ++ urshr v19.8h, v15.8h, #8 ++ PF tst PF_CTL, #0x0F ++ raddhn v28.8b, v16.8h, v12.8h ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++10: ++ raddhn v29.8b, v17.8h, v13.8h ++ PF beq 10f ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v30.8b, v18.8h, v14.8h ++ PF cmp PF_X, ORIG_W ++ raddhn v31.8b, v19.8h, v15.8h ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++ umull v16.8h, v24.8b, v8.8b ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++ umull v17.8h, v24.8b, v9.8b ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++10: ++ umull v18.8h, v24.8b, v10.8b ++ PF ble 10f ++ PF subs PF_CTL, PF_CTL, #0x10 ++10: ++ umull v19.8h, v24.8b, v11.8b ++ PF ble 10f ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++10: ++ uqadd v28.8b, v0.8b, v28.8b ++ PF ble 10f ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++10: ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ urshr v12.8h, v16.8h, #8 ++ urshr v13.8h, v17.8h, #8 ++ urshr v14.8h, v18.8h, #8 ++ urshr v15.8h, v19.8h, #8 ++ raddhn v0.8b, v16.8h, v12.8h ++ raddhn v1.8b, v17.8h, v13.8h ++ raddhn v2.8b, v18.8h, v14.8h ++ raddhn v3.8b, v19.8h, v15.8h ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ mvn v25.8b, v3.8b ++ umull v12.8h, v25.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v25.8b, v6.8b ++ umull v15.8h, v25.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8888_init ++ mov v11.s[0], w4 ++ dup v8.8b, v11.b[0] ++ dup v9.8b, v11.b[1] ++ dup v10.8b, v11.b[2] ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_8888_init, \ ++ pixman_composite_over_n_8_8888_cleanup, \ ++ pixman_composite_over_n_8_8888_process_pixblock_head, \ ++ pixman_composite_over_n_8_8888_process_pixblock_tail, \ ++ pixman_composite_over_n_8_8888_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8_8_process_pixblock_head ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v8.8b ++ umull v2.8h, v26.8b, v8.8b ++ umull v3.8h, v27.8b, v8.8b ++ urshr v10.8h, v0.8h, #8 ++ urshr v11.8h, v1.8h, #8 ++ urshr v12.8h, v2.8h, #8 ++ urshr v13.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v10.8h ++ raddhn v1.8b, v1.8h, v11.8h ++ raddhn v2.8b, v2.8h, v12.8h ++ raddhn v3.8b, v3.8h, v13.8h ++ mvn v24.8b, v0.8b ++ mvn v25.8b, v1.8b ++ mvn v26.8b, v2.8b ++ mvn v27.8b, v3.8b ++ umull v10.8h, v24.8b, v4.8b ++ umull v11.8h, v25.8b, v5.8b ++ umull v12.8h, v26.8b, v6.8b ++ umull v13.8h, v27.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8_8_process_pixblock_tail ++ urshr v14.8h, v10.8h, #8 ++ urshr v15.8h, v11.8h, #8 ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ raddhn v28.8b, v14.8h, v10.8h ++ raddhn v29.8b, v15.8h, v11.8h ++ raddhn v30.8b, v16.8h, v12.8h ++ raddhn v31.8b, v17.8h, v13.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_n_8_8_process_pixblock_tail_head ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_n_8_8_process_pixblock_tail ++ fetch_mask_pixblock ++ cache_preload 32, 32 ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ pixman_composite_over_n_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_n_8_8_init ++ mov v8.s[0], w4 ++ dup v8.8b, v8.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8_8_init, \ ++ pixman_composite_over_n_8_8_cleanup, \ ++ pixman_composite_over_n_8_8_process_pixblock_head, \ ++ pixman_composite_over_n_8_8_process_pixblock_tail, \ ++ pixman_composite_over_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_head ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} ++ * dest in {v4, v5, v6, v7 } ++ * mask in {v24, v25, v26, v27} ++ * output: updated src in {v0, v1, v2, v3 } ++ * updated mask in {v24, v25, v26, v3 } ++ */ ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ umull v3.8h, v27.8b, v11.8b ++ umull v12.8h, v11.8b, v25.8b ++ umull v13.8h, v11.8b, v24.8b ++ umull v14.8h, v11.8b, v26.8b ++ urshr v15.8h, v0.8h, #8 ++ urshr v16.8h, v1.8h, #8 ++ urshr v17.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v15.8h ++ raddhn v1.8b, v1.8h, v16.8h ++ raddhn v2.8b, v2.8h, v17.8h ++ urshr v15.8h, v13.8h, #8 ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v14.8h, #8 ++ urshr v18.8h, v3.8h, #8 ++ raddhn v24.8b, v13.8h, v15.8h ++ raddhn v25.8b, v12.8h, v16.8h ++ raddhn v26.8b, v14.8h, v17.8h ++ raddhn v3.8b, v3.8h, v18.8h ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in {v28, v29, v30, v31} ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ mvn v26.8b, v26.8b ++ mvn v27.8b, v3.8b ++ umull v12.8h, v24.8b, v4.8b ++ umull v13.8h, v25.8b, v5.8b ++ umull v14.8h, v26.8b, v6.8b ++ umull v15.8h, v27.8b, v7.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_tail ++ /* ... continue 'combine_over_ca' replacement */ ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_process_pixblock_tail_head ++ urshr v16.8h, v12.8h, #8 ++ urshr v17.8h, v13.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ urshr v18.8h, v14.8h, #8 ++ urshr v19.8h, v15.8h, #8 ++ raddhn v28.8b, v16.8h, v12.8h ++ raddhn v29.8b, v17.8h, v13.8h ++ raddhn v30.8b, v18.8h, v14.8h ++ raddhn v31.8b, v19.8h, v15.8h ++ fetch_mask_pixblock ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++ cache_preload 8, 8 ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_init ++ mov v13.s[0], w4 ++ dup v8.8b, v13.b[0] ++ dup v9.8b, v13.b[1] ++ dup v10.8b, v13.b[2] ++ dup v11.8b, v13.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8888_8888_ca_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_8888_ca_asm_neon, 0, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_8888_ca_init, \ ++ pixman_composite_over_n_8888_8888_ca_cleanup, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_head, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_8888_ca_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_head ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} [B, G, R, A] ++ * mask in {v24, v25, v26} [B, G, R] ++ * output: updated src in {v0, v1, v2 } [B, G, R] ++ * updated mask in {v24, v25, v26} [B, G, R] ++ */ ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ umull v12.8h, v11.8b, v24.8b ++ umull v13.8h, v11.8b, v25.8b ++ umull v14.8h, v11.8b, v26.8b ++ urshr v15.8h, v0.8h, #8 ++ urshr v16.8h, v1.8h, #8 ++ urshr v17.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v15.8h ++ raddhn v1.8b, v1.8h, v16.8h ++ raddhn v2.8b, v2.8h, v17.8h ++ urshr v19.8h, v12.8h, #8 ++ urshr v20.8h, v13.8h, #8 ++ urshr v21.8h, v14.8h, #8 ++ raddhn v24.8b, v12.8h, v19.8h ++ raddhn v25.8b, v13.8h, v20.8h ++ /* ++ * convert 8 r5g6b5 pixel data from {v4} to planar 8-bit format ++ * and put data into v16 - blue, v17 - green, v18 - red ++ */ ++ mov v4.d[1], v5.d[0] ++ shrn v17.8b, v4.8h, #3 ++ shrn v18.8b, v4.8h, #8 ++ raddhn v26.8b, v14.8h, v21.8h ++ sli v4.8h, v4.8h, #5 ++ sri v18.8b, v18.8b, #5 ++ sri v17.8b, v17.8b, #6 ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in v16 - blue, v17 - green, v18 - red ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ shrn v16.8b, v4.8h, #2 ++ mvn v26.8b, v26.8b ++ umull v5.8h, v16.8b, v24.8b ++ umull v6.8h, v17.8b, v25.8b ++ umull v7.8h, v18.8b, v26.8b ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_tail ++ /* ... continue 'combine_over_ca' replacement */ ++ urshr v13.8h, v5.8h, #8 ++ urshr v14.8h, v6.8h, #8 ++ urshr v15.8h, v7.8h, #8 ++ raddhn v16.8b, v13.8h, v5.8h ++ raddhn v17.8b, v14.8h, v6.8h ++ raddhn v18.8b, v15.8h, v7.8h ++ uqadd v16.8b, v0.8b, v16.8b ++ uqadd v17.8b, v1.8b, v17.8b ++ uqadd v18.8b, v2.8b, v18.8b ++ /* ++ * convert the results in v16, v17, v18 to r5g6b5 and store ++ * them into {v14} ++ */ ++ ushll v14.8h, v18.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v12.8h, v17.8b, #7 ++ sli v12.8h, v12.8h, #1 ++ ushll v13.8h, v16.8b, #7 ++ sli v13.8h, v13.8h, #1 ++ sri v14.8h, v12.8h, #5 ++ sri v14.8h, v13.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_process_pixblock_tail_head ++ fetch_mask_pixblock ++ urshr v13.8h, v5.8h, #8 ++ urshr v14.8h, v6.8h, #8 ++ ld1 {v4.8h}, [DST_R], #16 ++ urshr v15.8h, v7.8h, #8 ++ raddhn v16.8b, v13.8h, v5.8h ++ raddhn v17.8b, v14.8h, v6.8h ++ raddhn v18.8b, v15.8h, v7.8h ++ mov v5.d[0], v4.d[1] ++ /* process_pixblock_head */ ++ /* ++ * 'combine_mask_ca' replacement ++ * ++ * input: solid src (n) in {v8, v9, v10, v11} [B, G, R, A] ++ * mask in {v24, v25, v26} [B, G, R] ++ * output: updated src in {v0, v1, v2 } [B, G, R] ++ * updated mask in {v24, v25, v26} [B, G, R] ++ */ ++ uqadd v16.8b, v0.8b, v16.8b ++ uqadd v17.8b, v1.8b, v17.8b ++ uqadd v18.8b, v2.8b, v18.8b ++ umull v0.8h, v24.8b, v8.8b ++ umull v1.8h, v25.8b, v9.8b ++ umull v2.8h, v26.8b, v10.8b ++ /* ++ * convert the result in v16, v17, v18 to r5g6b5 and store ++ * it into {v14} ++ */ ++ ushll v14.8h, v18.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ ushll v18.8h, v16.8b, #7 ++ sli v18.8h, v18.8h, #1 ++ ushll v19.8h, v17.8b, #7 ++ sli v19.8h, v19.8h, #1 ++ umull v12.8h, v11.8b, v24.8b ++ sri v14.8h, v19.8h, #5 ++ umull v13.8h, v11.8b, v25.8b ++ umull v15.8h, v11.8b, v26.8b ++ sri v14.8h, v18.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ cache_preload 8, 8 ++ urshr v16.8h, v0.8h, #8 ++ urshr v17.8h, v1.8h, #8 ++ urshr v18.8h, v2.8h, #8 ++ raddhn v0.8b, v0.8h, v16.8h ++ raddhn v1.8b, v1.8h, v17.8h ++ raddhn v2.8b, v2.8h, v18.8h ++ urshr v19.8h, v12.8h, #8 ++ urshr v20.8h, v13.8h, #8 ++ urshr v21.8h, v15.8h, #8 ++ raddhn v24.8b, v12.8h, v19.8h ++ raddhn v25.8b, v13.8h, v20.8h ++ /* ++ * convert 8 r5g6b5 pixel data from {v4, v5} to planar ++ * 8-bit format and put data into v16 - blue, v17 - green, ++ * v18 - red ++ */ ++ mov v4.d[1], v5.d[0] ++ shrn v17.8b, v4.8h, #3 ++ shrn v18.8b, v4.8h, #8 ++ raddhn v26.8b, v15.8h, v21.8h ++ sli v4.8h, v4.8h, #5 ++ sri v17.8b, v17.8b, #6 ++ sri v18.8b, v18.8b, #5 ++ /* ++ * 'combine_over_ca' replacement ++ * ++ * output: updated dest in v16 - blue, v17 - green, v18 - red ++ */ ++ mvn v24.8b, v24.8b ++ mvn v25.8b, v25.8b ++ shrn v16.8b, v4.8h, #2 ++ mvn v26.8b, v26.8b ++ umull v5.8h, v16.8b, v24.8b ++ umull v6.8h, v17.8b, v25.8b ++ umull v7.8h, v18.8b, v26.8b ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_init ++ mov v13.s[0], w4 ++ dup v8.8b, v13.b[0] ++ dup v9.8b, v13.b[1] ++ dup v10.8b, v13.b[2] ++ dup v11.8b, v13.b[3] ++.endm ++ ++.macro pixman_composite_over_n_8888_0565_ca_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_n_8888_0565_ca_asm_neon, 0, 32, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_n_8888_0565_ca_init, \ ++ pixman_composite_over_n_8888_0565_ca_cleanup, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_head, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_tail, \ ++ pixman_composite_over_n_8888_0565_ca_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_in_n_8_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ umull v8.8h, v4.8b, v3.8b ++ umull v9.8h, v5.8b, v3.8b ++ umull v10.8h, v6.8b, v3.8b ++ umull v11.8h, v7.8b, v3.8b ++.endm ++ ++.macro pixman_composite_in_n_8_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v28.8b, v8.8h, v14.8h ++ raddhn v29.8b, v9.8h, v15.8h ++ raddhn v30.8b, v10.8h, v12.8h ++ raddhn v31.8b, v11.8h, v13.8h ++.endm ++ ++.macro pixman_composite_in_n_8_process_pixblock_tail_head ++ pixman_composite_in_n_8_process_pixblock_tail ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ cache_preload 32, 32 ++ pixman_composite_in_n_8_process_pixblock_head ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_in_n_8_init ++ mov v3.s[0], w4 ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_in_n_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_in_n_8_asm_neon, 0, 0, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_in_n_8_init, \ ++ pixman_composite_in_n_8_cleanup, \ ++ pixman_composite_in_n_8_process_pixblock_head, \ ++ pixman_composite_in_n_8_process_pixblock_tail, \ ++ pixman_composite_in_n_8_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++.macro pixman_composite_add_n_8_8_process_pixblock_head ++ /* expecting source data in {v8, v9, v10, v11} */ ++ /* v8 - blue, v9 - green, v10 - red, v11 - alpha */ ++ /* and destination data in {v4, v5, v6, v7} */ ++ /* mask is in v24, v25, v26, v27 */ ++ umull v0.8h, v24.8b, v11.8b ++ umull v1.8h, v25.8b, v11.8b ++ umull v2.8h, v26.8b, v11.8b ++ umull v3.8h, v27.8b, v11.8b ++ urshr v12.8h, v0.8h, #8 ++ urshr v13.8h, v1.8h, #8 ++ urshr v14.8h, v2.8h, #8 ++ urshr v15.8h, v3.8h, #8 ++ raddhn v0.8b, v0.8h, v12.8h ++ raddhn v1.8b, v1.8h, v13.8h ++ raddhn v2.8b, v2.8h, v14.8h ++ raddhn v3.8b, v3.8h, v15.8h ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_n_8_8_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_n_8_8_process_pixblock_tail_head ++ pixman_composite_add_n_8_8_process_pixblock_tail ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ fetch_mask_pixblock ++ cache_preload 32, 32 ++ pixman_composite_add_n_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_add_n_8_8_init ++ mov v11.s[0], w4 ++ dup v11.8b, v11.b[3] ++.endm ++ ++.macro pixman_composite_add_n_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_n_8_8_asm_neon, 0, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_n_8_8_init, \ ++ pixman_composite_add_n_8_8_cleanup, \ ++ pixman_composite_add_n_8_8_process_pixblock_head, \ ++ pixman_composite_add_n_8_8_process_pixblock_tail, \ ++ pixman_composite_add_n_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8_8_8_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* mask in {v24, v25, v26, v27} */ ++ umull v8.8h, v24.8b, v0.8b ++ umull v9.8h, v25.8b, v1.8b ++ umull v10.8h, v26.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ urshr v0.8h, v8.8h, #8 ++ urshr v1.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v0.8b, v0.8h, v8.8h ++ raddhn v1.8b, v1.8h, v9.8h ++ raddhn v2.8b, v12.8h, v10.8h ++ raddhn v3.8b, v13.8h, v11.8h ++ uqadd v28.8b, v0.8b, v4.8b ++ uqadd v29.8b, v1.8b, v5.8b ++ uqadd v30.8b, v2.8b, v6.8b ++ uqadd v31.8b, v3.8b, v7.8b ++.endm ++ ++.macro pixman_composite_add_8_8_8_process_pixblock_tail ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_8_8_8_process_pixblock_tail_head ++ pixman_composite_add_8_8_8_process_pixblock_tail ++ st1 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ld1 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ fetch_mask_pixblock ++ fetch_src_pixblock ++ cache_preload 32, 32 ++ pixman_composite_add_8_8_8_process_pixblock_head ++.endm ++ ++.macro pixman_composite_add_8_8_8_init ++.endm ++ ++.macro pixman_composite_add_8_8_8_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8_8_8_asm_neon, 8, 8, 8, \ ++ FLAG_DST_READWRITE, \ ++ 32, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_8_8_8_init, \ ++ pixman_composite_add_8_8_8_cleanup, \ ++ pixman_composite_add_8_8_8_process_pixblock_head, \ ++ pixman_composite_add_8_8_8_process_pixblock_tail, \ ++ pixman_composite_add_8_8_8_process_pixblock_tail_head ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* mask in {v24, v25, v26, v27} */ ++ umull v8.8h, v27.8b, v0.8b ++ umull v9.8h, v27.8b, v1.8b ++ umull v10.8h, v27.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ /* 1 cycle bubble */ ++ ursra v8.8h, v8.8h, #8 ++ ursra v9.8h, v9.8h, #8 ++ ursra v10.8h, v10.8h, #8 ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_tail ++ /* 2 cycle bubble */ ++ rshrn v28.8b, v8.8h, #8 ++ rshrn v29.8b, v9.8h, #8 ++ rshrn v30.8b, v10.8h, #8 ++ rshrn v31.8b, v11.8h, #8 ++ uqadd v28.8b, v4.8b, v28.8b ++ uqadd v29.8b, v5.8b, v29.8b ++ uqadd v30.8b, v6.8b, v30.8b ++ uqadd v31.8b, v7.8b, v31.8b ++.endm ++ ++.macro pixman_composite_add_8888_8888_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ rshrn v28.8b, v8.8h, #8 ++ fetch_mask_pixblock ++ rshrn v29.8b, v9.8h, #8 ++ umull v8.8h, v27.8b, v0.8b ++ rshrn v30.8b, v10.8h, #8 ++ umull v9.8h, v27.8b, v1.8b ++ rshrn v31.8b, v11.8h, #8 ++ umull v10.8h, v27.8b, v2.8b ++ umull v11.8h, v27.8b, v3.8b ++ uqadd v28.8b, v4.8b, v28.8b ++ uqadd v29.8b, v5.8b, v29.8b ++ uqadd v30.8b, v6.8b, v30.8b ++ uqadd v31.8b, v7.8b, v31.8b ++ ursra v8.8h, v8.8h, #8 ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ ursra v9.8h, v9.8h, #8 ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ ursra v10.8h, v10.8h, #8 ++ ++ cache_preload 8, 8 ++ ++ ursra v11.8h, v11.8h, #8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8888_8888_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_add_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++generate_composite_function \ ++ pixman_composite_add_8888_8_8888_asm_neon, 32, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_n_8_8888_init ++ mov v3.s[0], w4 ++ dup v0.8b, v3.b[0] ++ dup v1.8b, v3.b[1] ++ dup v2.8b, v3.b[2] ++ dup v3.8b, v3.b[3] ++.endm ++ ++.macro pixman_composite_add_n_8_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_n_8_8888_asm_neon, 0, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_n_8_8888_init, \ ++ pixman_composite_add_n_8_8888_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_8888_n_8888_init ++ mov v27.s[0], w6 ++ dup v27.8b, v27.b[3] ++.endm ++ ++.macro pixman_composite_add_8888_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_8888_n_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_add_8888_n_8888_init, \ ++ pixman_composite_add_8888_n_8888_cleanup, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_head, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail, \ ++ pixman_composite_add_8888_8888_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 27 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++ /* expecting source data in {v0, v1, v2, v3} */ ++ /* destination data in {v4, v5, v6, v7} */ ++ /* solid mask is in v15 */ ++ ++ /* 'in' */ ++ umull v11.8h, v15.8b, v3.8b ++ umull v10.8h, v15.8b, v2.8b ++ umull v9.8h, v15.8b, v1.8b ++ umull v8.8h, v15.8b, v0.8b ++ urshr v16.8h, v11.8h, #8 ++ urshr v14.8h, v10.8h, #8 ++ urshr v13.8h, v9.8h, #8 ++ urshr v12.8h, v8.8h, #8 ++ raddhn v3.8b, v11.8h, v16.8h ++ raddhn v2.8b, v10.8h, v14.8h ++ raddhn v1.8b, v9.8h, v13.8h ++ raddhn v0.8b, v8.8h, v12.8h ++ mvn v24.8b, v3.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++ umull v11.8h, v24.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ urshr v16.8h, v8.8h, #8 ++ urshr v17.8h, v9.8h, #8 ++ urshr v18.8h, v10.8h, #8 ++ urshr v19.8h, v11.8h, #8 ++ raddhn v28.8b, v16.8h, v8.8h ++ raddhn v29.8b, v17.8h, v9.8h ++ raddhn v30.8b, v18.8h, v10.8h ++ raddhn v31.8b, v19.8h, v11.8h ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8888_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_out_reverse_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_head ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_tail ++ pixman_composite_out_reverse_8888_n_8888_process_pixblock_tail ++ uqadd v28.8b, v0.8b, v28.8b ++ uqadd v29.8b, v1.8b, v29.8b ++ uqadd v30.8b, v2.8b, v30.8b ++ uqadd v31.8b, v3.8b, v31.8b ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_n_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_init ++ mov v15.s[0], w6 ++ dup v15.8b, v15.b[3] ++.endm ++ ++.macro pixman_composite_over_8888_n_8888_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_n_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_8888_n_8888_init, \ ++ pixman_composite_over_8888_n_8888_cleanup, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_8888_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8888_8888_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++generate_composite_function_single_scanline \ ++ pixman_composite_scanline_over_mask_asm_neon, 32, 32, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 12 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_8888_8_8888_process_pixblock_tail_head ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ pixman_composite_over_8888_n_8888_process_pixblock_tail ++ fetch_src_pixblock ++ cache_preload 8, 8 ++ fetch_mask_pixblock ++ pixman_composite_over_8888_n_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_8888_8_8888_asm_neon, 32, 8, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_n_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_8888_process_pixblock_tail_head \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_head ++.endm ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0888_0888_process_pixblock_tail_head ++ st3 {v0.8b, v1.8b, v2.8b}, [DST_W], #24 ++ fetch_src_pixblock ++ cache_preload 8, 8 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_0888_asm_neon, 24, 0, 24, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_0888_process_pixblock_head, \ ++ pixman_composite_src_0888_0888_process_pixblock_tail, \ ++ pixman_composite_src_0888_0888_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_head ++ mov v31.8b, v2.8b ++ mov v2.8b, v0.8b ++ mov v0.8b, v31.8b ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_tail ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_process_pixblock_tail_head ++ st4 {v0.8b, v1.8b, v2.8b, v3.8b}, [DST_W], #32 ++ fetch_src_pixblock ++ mov v31.8b, v2.8b ++ mov v2.8b, v0.8b ++ mov v0.8b, v31.8b ++ cache_preload 8, 8 ++.endm ++ ++.macro pixman_composite_src_0888_8888_rev_init ++ eor v3.8b, v3.8b, v3.8b ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_8888_rev_asm_neon, 24, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ pixman_composite_src_0888_8888_rev_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_head, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_tail, \ ++ pixman_composite_src_0888_8888_rev_process_pixblock_tail_head, \ ++ 0, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_head ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ ushll v9.8h, v2.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_tail ++ ushll v14.8h, v0.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++.macro pixman_composite_src_0888_0565_rev_process_pixblock_tail_head ++ ushll v14.8h, v0.8b, #7 ++ sli v14.8h, v14.8h, #1 ++ fetch_src_pixblock ++ sri v14.8h, v8.8h, #5 ++ sri v14.8h, v9.8h, #11 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++ ushll v8.8h, v1.8b, #7 ++ sli v8.8h, v8.8h, #1 ++ st1 {v14.8h}, [DST_W], #16 ++ ushll v9.8h, v2.8b, #7 ++ sli v9.8h, v9.8h, #1 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_0888_0565_rev_asm_neon, 24, 0, 16, \ ++ FLAG_DST_WRITEONLY, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_head, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_tail, \ ++ pixman_composite_src_0888_0565_rev_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_head ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++.endm ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v30.8b, v11.8h, v8.8h ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v28.8b, v13.8h, v10.8h ++.endm ++ ++.macro pixman_composite_src_pixbuf_8888_process_pixblock_tail_head ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v31.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ fetch_src_pixblock ++ raddhn v30.8b, v11.8h, v8.8h ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v28.8b, v13.8h, v10.8h ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_pixbuf_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_head, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_tail, \ ++ pixman_composite_src_pixbuf_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_head ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++.endm ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v28.8b, v11.8h, v8.8h ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v30.8b, v13.8h, v10.8h ++.endm ++ ++.macro pixman_composite_src_rpixbuf_8888_process_pixblock_tail_head ++ urshr v11.8h, v8.8h, #8 ++ mov v30.8b, v31.8b ++ mov v31.8b, v3.8b ++ mov v3.8b, v30.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ fetch_src_pixblock ++ raddhn v28.8b, v11.8h, v8.8h ++ PF add PF_X, PF_X, #8 ++ PF tst PF_CTL, #0xF ++ PF beq 10f ++ PF add PF_X, PF_X, #8 ++ PF sub PF_CTL, PF_CTL, #1 ++10: ++ raddhn v29.8b, v12.8h, v9.8h ++ raddhn v30.8b, v13.8h, v10.8h ++ umull v8.8h, v3.8b, v0.8b ++ umull v9.8h, v3.8b, v1.8b ++ umull v10.8h, v3.8b, v2.8b ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++ PF cmp PF_X, ORIG_W ++ PF lsl DUMMY, PF_X, src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++ PF ble 10f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++ PF ble 10f ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++10: ++.endm ++ ++generate_composite_function \ ++ pixman_composite_src_rpixbuf_8888_asm_neon, 32, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 10, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_head, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_tail, \ ++ pixman_composite_src_rpixbuf_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 0, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v4.d[0], v8.d[0] ++ mov v4.d[1], v9.d[0] ++ mov v13.d[0], v10.d[0] ++ mov v13.d[1], v11.d[0] ++ convert_0565_to_x888 v4, v2, v1, v0 ++ convert_0565_to_x888 v13, v6, v5, v4 ++ /* source pixel data is in {v0, v1, v2, XX} */ ++ /* destination pixel data is in {v4, v5, v6, XX} */ ++ mvn v7.8b, v15.8b ++ umull v10.8h, v15.8b, v2.8b ++ umull v9.8h, v15.8b, v1.8b ++ umull v8.8h, v15.8b, v0.8b ++ umull v11.8h, v7.8b, v4.8b ++ umull v12.8h, v7.8b, v5.8b ++ umull v13.8h, v7.8b, v6.8b ++ urshr v19.8h, v10.8h, #8 ++ urshr v18.8h, v9.8h, #8 ++ urshr v17.8h, v8.8h, #8 ++ raddhn v2.8b, v10.8h, v19.8h ++ raddhn v1.8b, v9.8h, v18.8h ++ raddhn v0.8b, v8.8h, v17.8h ++.endm ++ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_tail ++ urshr v17.8h, v11.8h, #8 ++ urshr v18.8h, v12.8h, #8 ++ urshr v19.8h, v13.8h, #8 ++ raddhn v28.8b, v17.8h, v11.8h ++ raddhn v29.8b, v18.8h, v12.8h ++ raddhn v30.8b, v19.8h, v13.8h ++ uqadd v0.8b, v0.8b, v28.8b ++ uqadd v1.8b, v1.8b, v29.8b ++ uqadd v2.8b, v2.8b, v30.8b ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v30, v13 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_over_0565_8_0565_process_pixblock_tail_head ++ fetch_mask_pixblock ++ pixman_composite_over_0565_8_0565_process_pixblock_tail ++ fetch_src_pixblock ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_over_0565_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_0565_8_0565_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_over_0565_n_0565_init ++ mov v15.s[0], w6 ++ dup v15.8b, v15.b[3] ++.endm ++ ++.macro pixman_composite_over_0565_n_0565_cleanup ++.endm ++ ++generate_composite_function \ ++ pixman_composite_over_0565_n_0565_asm_neon, 16, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ pixman_composite_over_0565_n_0565_init, \ ++ pixman_composite_over_0565_n_0565_cleanup, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v4.d[0], v8.d[0] ++ mov v4.d[1], v9.d[0] ++ mov v13.d[0], v10.d[0] ++ mov v13.d[1], v11.d[0] ++ convert_0565_to_x888 v4, v2, v1, v0 ++ convert_0565_to_x888 v13, v6, v5, v4 ++ /* source pixel data is in {v0, v1, v2, XX} */ ++ /* destination pixel data is in {v4, v5, v6, XX} */ ++ umull v9.8h, v15.8b, v2.8b ++ umull v8.8h, v15.8b, v1.8b ++ umull v7.8h, v15.8b, v0.8b ++ urshr v12.8h, v9.8h, #8 ++ urshr v11.8h, v8.8h, #8 ++ urshr v10.8h, v7.8h, #8 ++ raddhn v2.8b, v9.8h, v12.8h ++ raddhn v1.8b, v8.8h, v11.8h ++ raddhn v0.8b, v7.8h, v10.8h ++.endm ++ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_tail ++ uqadd v0.8b, v0.8b, v4.8b ++ uqadd v1.8b, v1.8b, v5.8b ++ uqadd v2.8b, v2.8b, v6.8b ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v30, v13 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_add_0565_8_0565_process_pixblock_tail_head ++ fetch_mask_pixblock ++ pixman_composite_add_0565_8_0565_process_pixblock_tail ++ fetch_src_pixblock ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_add_0565_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_add_0565_8_0565_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_add_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_head ++ /* mask is in v15 */ ++ mov v12.d[0], v10.d[0] ++ mov v12.d[1], v11.d[0] ++ convert_0565_to_x888 v12, v6, v5, v4 ++ /* destination pixel data is in {v4, v5, v6, xx} */ ++ mvn v24.8b, v15.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v24.8b, v4.8b ++ umull v9.8h, v24.8b, v5.8b ++ umull v10.8h, v24.8b, v6.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_tail ++ urshr v11.8h, v8.8h, #8 ++ urshr v12.8h, v9.8h, #8 ++ urshr v13.8h, v10.8h, #8 ++ raddhn v0.8b, v11.8h, v8.8h ++ raddhn v1.8b, v12.8h, v9.8h ++ raddhn v2.8b, v13.8h, v10.8h ++ /* 32bpp result is in {v0, v1, v2, XX} */ ++ convert_8888_to_0565 v2, v1, v0, v14, v12, v3 ++ mov v28.d[0], v14.d[0] ++ mov v29.d[0], v14.d[1] ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8_0565_process_pixblock_tail_head ++ fetch_src_pixblock ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail ++ ld1 {v10.4h, v11.4h}, [DST_R], #16 ++ cache_preload 8, 8 ++ pixman_composite_out_reverse_8_0565_process_pixblock_head ++ st1 {v14.8h}, [DST_W], #16 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_out_reverse_8_0565_asm_neon, 8, 0, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_head, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 15, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_head ++ /* src is in v0 */ ++ /* destination pixel data is in {v4, v5, v6, v7} */ ++ mvn v1.8b, v0.8b /* get inverted alpha */ ++ /* now do alpha blending */ ++ umull v8.8h, v1.8b, v4.8b ++ umull v9.8h, v1.8b, v5.8b ++ umull v10.8h, v1.8b, v6.8b ++ umull v11.8h, v1.8b, v7.8b ++.endm ++ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_tail ++ urshr v14.8h, v8.8h, #8 ++ urshr v15.8h, v9.8h, #8 ++ urshr v12.8h, v10.8h, #8 ++ urshr v13.8h, v11.8h, #8 ++ raddhn v28.8b, v14.8h, v8.8h ++ raddhn v29.8b, v15.8h, v9.8h ++ raddhn v30.8b, v12.8h, v10.8h ++ raddhn v31.8b, v13.8h, v11.8h ++ /* 32bpp result is in {v28, v29, v30, v31} */ ++.endm ++ ++/* TODO: expand macros and do better instructions scheduling */ ++.macro pixman_composite_out_reverse_8_8888_process_pixblock_tail_head ++ fetch_src_pixblock ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail ++ ld4 {v4.8b, v5.8b, v6.8b, v7.8b}, [DST_R], #32 ++ cache_preload 8, 8 ++ pixman_composite_out_reverse_8_8888_process_pixblock_head ++ st4 {v28.8b, v29.8b, v30.8b, v31.8b}, [DST_W], #32 ++.endm ++ ++generate_composite_function \ ++ pixman_composite_out_reverse_8_8888_asm_neon, 8, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ 5, /* prefetch distance */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_head, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail, \ ++ pixman_composite_out_reverse_8_8888_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 0 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_8888_OVER_asm_neon, 32, 0, 32, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_8888_process_pixblock_head, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail, \ ++ pixman_composite_over_8888_8888_process_pixblock_tail_head ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_0565_OVER_asm_neon, 32, 0, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_over_8888_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 0, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_0565_SRC_asm_neon, 32, 0, 16, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_8888_0565_process_pixblock_head, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail, \ ++ pixman_composite_src_8888_0565_process_pixblock_tail_head, \ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_0565_8888_SRC_asm_neon, 16, 0, 32, \ ++ FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init, \ ++ default_cleanup, \ ++ pixman_composite_src_0565_8888_process_pixblock_head, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail, \ ++ pixman_composite_src_0565_8888_process_pixblock_tail_head ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_8888_8_0565_OVER_asm_neon, 32, 8, 16, \ ++ FLAG_DST_READWRITE | FLAG_DEINTERLEAVE_32BPP, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_head, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_8888_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 4, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 24 /* mask_basereg */ ++ ++generate_composite_function_nearest_scanline \ ++ pixman_scaled_nearest_scanline_0565_8_0565_OVER_asm_neon, 16, 8, 16, \ ++ FLAG_DST_READWRITE, \ ++ 8, /* number of pixels, processed in a single block */ \ ++ default_init_need_all_regs, \ ++ default_cleanup_need_all_regs, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_head, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail, \ ++ pixman_composite_over_0565_8_0565_process_pixblock_tail_head, \ ++ 28, /* dst_w_basereg */ \ ++ 10, /* dst_r_basereg */ \ ++ 8, /* src_basereg */ \ ++ 15 /* mask_basereg */ ++ ++/******************************************************************************/ ++ ++/* ++ * Bilinear scaling support code which tries to provide pixel fetching, color ++ * format conversion, and interpolation as separate macros which can be used ++ * as the basic building blocks for constructing bilinear scanline functions. ++ */ ++ ++.macro bilinear_load_8888 reg1, reg2, tmp ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ ld1 {®1&.2s}, [TMP1], STRIDE ++ ld1 {®2&.2s}, [TMP1] ++.endm ++ ++.macro bilinear_load_0565 reg1, reg2, tmp ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ ld1 {®2&.s}[0], [TMP1], STRIDE ++ ld1 {®2&.s}[1], [TMP1] ++ convert_four_0565_to_x888_packed reg2, reg1, reg2, tmp ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_8888 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, tmp1, tmp2 ++ ++ bilinear_load_8888 reg1, reg2, tmp1 ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ bilinear_load_8888 reg3, reg4, tmp2 ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi ++ bilinear_load_and_vertical_interpolate_two_8888 \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++.endm ++ ++.macro vzip reg1, reg2 ++ umov TMP4, v31.d[0] ++ zip1 v31.8b, reg1, reg2 ++ zip2 reg2, reg1, reg2 ++ mov reg1, v31.8b ++ mov v31.d[0], TMP4 ++.endm ++ ++.macro vuzp reg1, reg2 ++ umov TMP4, v31.d[0] ++ uzp1 v31.8b, reg1, reg2 ++ uzp2 reg2, reg1, reg2 ++ mov reg1, v31.8b ++ mov v31.d[0], TMP4 ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_two_0565 \ ++ acc1, acc2, reg1, reg2, reg3, reg4, acc2lo, acc2hi ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&acc2&.s}[0], [TMP1], STRIDE ++ ld1 {&acc2&.s}[2], [TMP2], STRIDE ++ ld1 {&acc2&.s}[1], [TMP1] ++ ld1 {&acc2&.s}[3], [TMP2] ++ convert_0565_to_x888 acc2, reg3, reg2, reg1 ++ vzip ®1&.8b, ®3&.8b ++ vzip ®2&.8b, ®4&.8b ++ vzip ®3&.8b, ®4&.8b ++ vzip ®1&.8b, ®2&.8b ++ umull &acc1&.8h, ®1&.8b, v28.8b ++ umlal &acc1&.8h, ®2&.8b, v29.8b ++ umull &acc2&.8h, ®3&.8b, v28.8b ++ umlal &acc2&.8h, ®4&.8b, v29.8b ++.endm ++ ++.macro bilinear_load_and_vertical_interpolate_four_0565 \ ++ xacc1, xacc2, xreg1, xreg2, xreg3, xreg4, xacc2lo, xacc2hi \ ++ yacc1, yacc2, yreg1, yreg2, yreg3, yreg4, yacc2lo, yacc2hi ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&xacc2&.s}[0], [TMP1], STRIDE ++ ld1 {&xacc2&.s}[2], [TMP2], STRIDE ++ ld1 {&xacc2&.s}[1], [TMP1] ++ ld1 {&xacc2&.s}[3], [TMP2] ++ convert_0565_to_x888 xacc2, xreg3, xreg2, xreg1 ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #1 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #1 ++ ld1 {&yacc2&.s}[0], [TMP1], STRIDE ++ vzip &xreg1&.8b, &xreg3&.8b ++ ld1 {&yacc2&.s}[2], [TMP2], STRIDE ++ vzip &xreg2&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[1], [TMP1] ++ vzip &xreg3&.8b, &xreg4&.8b ++ ld1 {&yacc2&.s}[3], [TMP2] ++ vzip &xreg1&.8b, &xreg2&.8b ++ convert_0565_to_x888 yacc2, yreg3, yreg2, yreg1 ++ umull &xacc1&.8h, &xreg1&.8b, v28.8b ++ vzip &yreg1&.8b, &yreg3&.8b ++ umlal &xacc1&.8h, &xreg2&.8b, v29.8b ++ vzip &yreg2&.8b, &yreg4&.8b ++ umull &xacc2&.8h, &xreg3&.8b, v28.8b ++ vzip &yreg3&.8b, &yreg4&.8b ++ umlal &xacc2&.8h, &xreg4&.8b, v29.8b ++ vzip &yreg1&.8b, &yreg2&.8b ++ umull &yacc1&.8h, &yreg1&.8b, v28.8b ++ umlal &yacc1&.8h, &yreg2&.8b, v29.8b ++ umull &yacc2&.8h, &yreg3&.8b, v28.8b ++ umlal &yacc2&.8h, &yreg4&.8b, v29.8b ++.endm ++ ++.macro bilinear_store_8888 numpix, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v0.2s, v1.2s}, [OUT], #16 ++.elseif numpix == 2 ++ st1 {v0.2s}, [OUT], #8 ++.elseif numpix == 1 ++ st1 {v0.s}[0], [OUT], #4 ++.else ++ .error bilinear_store_8888 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_store_0565 numpix, tmp1, tmp2 ++ vuzp v0.8b, v1.8b ++ vuzp v2.8b, v3.8b ++ vuzp v1.8b, v3.8b ++ vuzp v0.8b, v2.8b ++ convert_8888_to_0565 v2, v1, v0, v1, tmp1, tmp2 ++.if numpix == 4 ++ st1 {v1.4h}, [OUT], #8 ++.elseif numpix == 2 ++ st1 {v1.s}[0], [OUT], #4 ++.elseif numpix == 1 ++ st1 {v1.h}[0], [OUT], #2 ++.else ++ .error bilinear_store_0565 numpix is unsupported ++.endif ++.endm ++ ++.macro bilinear_interpolate_last_pixel src_fmt, dst_fmt ++ bilinear_load_&src_fmt v0, v1, v2 ++ umull v2.8h, v0.8b, v28.8b ++ umlal v2.8h, v1.8b, v29.8b ++ /* 5 cycles bubble */ ++ ushll v0.4s, v2.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v2.4h, v15.h[0] ++ umlal2 v0.4s, v2.8h, v15.h[0] ++ /* 5 cycles bubble */ ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ /* 3 cycles bubble */ ++ xtn v0.8b, v0.8h ++ /* 1 cycle bubble */ ++ bilinear_store_&dst_fmt 1, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_two_pixels src_fmt, dst_fmt ++ bilinear_load_and_vertical_interpolate_two_&src_fmt \ ++ v1, v11, v2, v3, v20, v21, v22, v23 ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ xtn v0.8b, v0.8h ++ bilinear_store_&dst_fmt 2, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_four_pixels src_fmt, dst_fmt ++ bilinear_load_and_vertical_interpolate_four_&src_fmt \ ++ v1, v11, v14, v20, v16, v17, v22, v23 \ ++ v3, v9, v24, v25, v26, v27, v18, v19 ++ prfm PREFETCH_MODE, [TMP1, PF_OFFS] ++ sub TMP1, TMP1, STRIDE ++ ushll v0.4s, v1.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v1.4h, v15.h[0] ++ umlal2 v0.4s, v1.8h, v15.h[0] ++ ushll v10.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v10.4s, v11.4h, v15.h[4] ++ umlal2 v10.4s, v11.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v3.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v3.4h, v15.h[0] ++ umlal2 v2.4s, v3.8h, v15.h[0] ++ ushll v8.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ prfm PREFETCH_MODE, [TMP2, PF_OFFS] ++ umlsl v8.4s, v9.4h, v15.h[4] ++ umlal2 v8.4s, v9.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v10.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v8.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ xtn v0.8b, v0.8h ++ xtn v1.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ bilinear_store_&dst_fmt 4, v3, v4 ++.endm ++ ++.macro bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_head ++.else ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_tail ++.endif ++.endm ++ ++.macro bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_four_pixels_&src_fmt&_&dst_fmt&_tail_head ++.else ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_head ++.else ++ bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_tail src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_tail ++.else ++ bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++.endif ++.endm ++ ++.macro bilinear_interpolate_eight_pixels_tail_head src_fmt, dst_fmt ++.ifdef have_bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt ++ bilinear_interpolate_eight_pixels_&src_fmt&_&dst_fmt&_tail_head ++.else ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++.endif ++.endm ++ ++.set BILINEAR_FLAG_UNROLL_4, 0 ++.set BILINEAR_FLAG_UNROLL_8, 1 ++.set BILINEAR_FLAG_USE_ALL_NEON_REGS, 2 ++ ++/* ++ * Main template macro for generating NEON optimized bilinear scanline ++ * functions. ++ * ++ * Bilinear scanline scaler macro template uses the following arguments: ++ * fname - name of the function to generate ++ * src_fmt - source color format (8888 or 0565) ++ * dst_fmt - destination color format (8888 or 0565) ++ * bpp_shift - (1 << bpp_shift) is the size of source pixel in bytes ++ * prefetch_distance - prefetch in the source image by that many ++ * pixels ahead ++ */ ++ ++.macro generate_bilinear_scanline_func fname, src_fmt, dst_fmt, \ ++ src_bpp_shift, dst_bpp_shift, \ ++ prefetch_distance, flags ++ ++pixman_asm_function fname ++ OUT .req x0 ++ TOP .req x1 ++ BOTTOM .req x2 ++ WT .req x3 ++ WB .req x4 ++ X .req x5 ++ UX .req x6 ++ WIDTH .req x7 ++ TMP1 .req x8 ++ TMP2 .req x9 ++ PF_OFFS .req x10 ++ TMP3 .req x11 ++ TMP4 .req x12 ++ STRIDE .req x13 ++ ++ sxtw x3, w3 ++ sxtw x4, w4 ++ sxtw x5, w5 ++ sxtw x6, w6 ++ sxtw x7, w7 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 112 /* push all registers */ ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ stp x8, x9, [x29, -80] ++ stp x10, x11, [x29, -96] ++ stp x12, x13, [x29, -112] ++ ++ mov PF_OFFS, #prefetch_distance ++ mul PF_OFFS, PF_OFFS, UX ++ ++ subs STRIDE, BOTTOM, TOP ++ .unreq BOTTOM ++ ++ cmp WIDTH, #0 ++ ble 300f ++ ++ dup v12.8h, w5 ++ dup v13.8h, w6 ++ dup v28.8b, w3 ++ dup v29.8b, w4 ++ mov v25.d[0], v12.d[1] ++ mov v26.d[0], v13.d[0] ++ add v25.4h, v25.4h, v26.4h ++ mov v12.d[1], v25.d[0] ++ ++ /* ensure good destination alignment */ ++ cmp WIDTH, #1 ++ blt 100f ++ tst OUT, #(1 << dst_bpp_shift) ++ beq 100f ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ bilinear_interpolate_last_pixel src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #1 ++100: ++ add v13.8h, v13.8h, v13.8h ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ++ cmp WIDTH, #2 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 1)) ++ beq 100f ++ bilinear_interpolate_two_pixels src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #2 ++100: ++.if ((flags) & BILINEAR_FLAG_UNROLL_8) != 0 ++/*********** 8 pixels per iteration *****************/ ++ cmp WIDTH, #4 ++ blt 100f ++ tst OUT, #(1 << (dst_bpp_shift + 2)) ++ beq 100f ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++ sub WIDTH, WIDTH, #4 ++100: ++ subs WIDTH, WIDTH, #8 ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_interpolate_eight_pixels_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #8 ++ blt 500f ++1000: ++ bilinear_interpolate_eight_pixels_tail_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #8 ++ bge 1000b ++500: ++ bilinear_interpolate_eight_pixels_tail src_fmt, dst_fmt ++100: ++ tst WIDTH, #4 ++ beq 200f ++ bilinear_interpolate_four_pixels src_fmt, dst_fmt ++200: ++.else ++/*********** 4 pixels per iteration *****************/ ++ subs WIDTH, WIDTH, #4 ++ blt 100f ++ asr PF_OFFS, PF_OFFS, #(16 - src_bpp_shift) ++ bilinear_interpolate_four_pixels_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #4 ++ blt 500f ++1000: ++ bilinear_interpolate_four_pixels_tail_head src_fmt, dst_fmt ++ subs WIDTH, WIDTH, #4 ++ bge 1000b ++500: ++ bilinear_interpolate_four_pixels_tail src_fmt, dst_fmt ++100: ++/****************************************************/ ++.endif ++ /* handle the remaining trailing pixels */ ++ tst WIDTH, #2 ++ beq 200f ++ bilinear_interpolate_two_pixels src_fmt, dst_fmt ++200: ++ tst WIDTH, #1 ++ beq 300f ++ bilinear_interpolate_last_pixel src_fmt, dst_fmt ++300: ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -104] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret ++ ++ .unreq OUT ++ .unreq TOP ++ .unreq WT ++ .unreq WB ++ .unreq X ++ .unreq UX ++ .unreq WIDTH ++ .unreq TMP1 ++ .unreq TMP2 ++ .unreq PF_OFFS ++ .unreq TMP3 ++ .unreq TMP4 ++ .unreq STRIDE ++.endfunc ++ ++.endm ++ ++/*****************************************************************************/ ++ ++.set have_bilinear_interpolate_four_pixels_8888_8888, 1 ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_head ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ ++ ld1 {v22.2s}, [TMP1], STRIDE ++ ld1 {v23.2s}, [TMP1] ++ asr TMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ umull v8.8h, v22.8b, v28.8b ++ umlal v8.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP2], STRIDE ++ ld1 {v23.2s}, [TMP2] ++ asr TMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umull v9.8h, v22.8b, v28.8b ++ umlal v9.8h, v23.8b, v29.8b ++ ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++.endm ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_tail ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ add v12.8h, v12.8h, v13.8h ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v2.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ shrn2 v2.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ xtn v6.8b, v0.8h ++ xtn v7.8b, v2.8h ++ add v12.8h, v12.8h, v13.8h ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++.endm ++ ++.macro bilinear_interpolate_four_pixels_8888_8888_tail_head ++ asr TMP1, X, #16 ++ add X, X, UX ++ add TMP1, TOP, TMP1, lsl #2 ++ asr TMP2, X, #16 ++ add X, X, UX ++ add TMP2, TOP, TMP2, lsl #2 ++ umlal2 v1.4s, v9.8h, v15.h[4] ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ushll v2.4s, v10.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v2.4s, v10.4h, v15.h[0] ++ umlal2 v2.4s, v10.8h, v15.h[0] ++ ushll v3.4s, v11.4h, #BILINEAR_INTERPOLATION_BITS ++ ld1 {v20.2s}, [TMP1], STRIDE ++ umlsl v3.4s, v11.4h, v15.h[4] ++ umlal2 v3.4s, v11.8h, v15.h[4] ++ ld1 {v21.2s}, [TMP1] ++ umull v8.8h, v20.8b, v28.8b ++ umlal v8.8h, v21.8b, v29.8b ++ shrn v0.4h, v0.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn2 v0.8h, v1.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ shrn v4.4h, v2.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ ld1 {v22.2s}, [TMP2], STRIDE ++ shrn2 v4.8h, v3.4s, #(2 * BILINEAR_INTERPOLATION_BITS) ++ add v12.8h, v12.8h, v13.8h ++ ld1 {v23.2s}, [TMP2] ++ umull v9.8h, v22.8b, v28.8b ++ asr TMP3, X, #16 ++ add X, X, UX ++ add TMP3, TOP, TMP3, lsl #2 ++ asr TMP4, X, #16 ++ add X, X, UX ++ add TMP4, TOP, TMP4, lsl #2 ++ umlal v9.8h, v23.8b, v29.8b ++ ld1 {v22.2s}, [TMP3], STRIDE ++ ushr v15.8h, v12.8h, #(16 - BILINEAR_INTERPOLATION_BITS) ++ ld1 {v23.2s}, [TMP3] ++ umull v10.8h, v22.8b, v28.8b ++ umlal v10.8h, v23.8b, v29.8b ++ xtn v6.8b, v0.8h ++ ushll v0.4s, v8.4h, #BILINEAR_INTERPOLATION_BITS ++ xtn v7.8b, v4.8h ++ umlsl v0.4s, v8.4h, v15.h[0] ++ umlal2 v0.4s, v8.8h, v15.h[0] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ ld1 {v16.2s}, [TMP4], STRIDE ++ add v12.8h, v12.8h, v13.8h ++ ld1 {v17.2s}, [TMP4] ++ prfm PREFETCH_MODE, [TMP4, PF_OFFS] ++ umull v11.8h, v16.8b, v28.8b ++ umlal v11.8h, v17.8b, v29.8b ++ st1 {v6.2s, v7.2s}, [OUT], #16 ++ ushll v1.4s, v9.4h, #BILINEAR_INTERPOLATION_BITS ++ umlsl v1.4s, v9.4h, v15.h[4] ++.endm ++ ++/*****************************************************************************/ ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_8888_SRC_asm_neon, 8888, 8888, \ ++ 2, 2, 28, BILINEAR_FLAG_UNROLL_4 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_8888_0565_SRC_asm_neon, 8888, 0565, \ ++ 2, 1, 28, BILINEAR_FLAG_UNROLL_8 | BILINEAR_FLAG_USE_ALL_NEON_REGS ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_x888_SRC_asm_neon, 0565, 8888, \ ++ 1, 2, 28, BILINEAR_FLAG_UNROLL_4 ++ ++generate_bilinear_scanline_func \ ++ pixman_scaled_bilinear_scanline_0565_0565_SRC_asm_neon, 0565, 0565, \ ++ 1, 1, 28, BILINEAR_FLAG_UNROLL_4 +diff --git a/pixman/pixman-arma64-neon-asm.h b/pixman/pixman-arma64-neon-asm.h +new file mode 100644 +index 0000000..0389d12 +--- /dev/null ++++ b/pixman/pixman-arma64-neon-asm.h +@@ -0,0 +1,1310 @@ ++/* ++ * Copyright © 2009 Nokia Corporation ++ * ++ * Permission is hereby granted, free of charge, to any person obtaining a ++ * copy of this software and associated documentation files (the "Software"), ++ * to deal in the Software without restriction, including without limitation ++ * the rights to use, copy, modify, merge, publish, distribute, sublicense, ++ * and/or sell copies of the Software, and to permit persons to whom the ++ * Software is furnished to do so, subject to the following conditions: ++ * ++ * The above copyright notice and this permission notice (including the next ++ * paragraph) shall be included in all copies or substantial portions of the ++ * Software. ++ * ++ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR ++ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, ++ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL ++ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER ++ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING ++ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER ++ * DEALINGS IN THE SOFTWARE. ++ * ++ * Author: Siarhei Siamashka (siarhei.siamashka@nokia.com) ++ */ ++ ++/* ++ * This file contains a macro ('generate_composite_function') which can ++ * construct 2D image processing functions, based on a common template. ++ * Any combinations of source, destination and mask images with 8bpp, ++ * 16bpp, 24bpp, 32bpp color formats are supported. ++ * ++ * This macro takes care of: ++ * - handling of leading and trailing unaligned pixels ++ * - doing most of the work related to L2 cache preload ++ * - encourages the use of software pipelining for better instructions ++ * scheduling ++ * ++ * The user of this macro has to provide some configuration parameters ++ * (bit depths for the images, prefetch distance, etc.) and a set of ++ * macros, which should implement basic code chunks responsible for ++ * pixels processing. See 'pixman-armv8-neon-asm.S' file for the usage ++ * examples. ++ * ++ * TODO: ++ * - try overlapped pixel method (from Ian Rickards) when processing ++ * exactly two blocks of pixels ++ * - maybe add an option to do reverse scanline processing ++ */ ++ ++/* ++ * Bit flags for 'generate_composite_function' macro which are used ++ * to tune generated functions behavior. ++ */ ++.set FLAG_DST_WRITEONLY, 0 ++.set FLAG_DST_READWRITE, 1 ++.set FLAG_DEINTERLEAVE_32BPP, 2 ++ ++/* ++ * Constants for selecting preferable prefetch type. ++ */ ++.set PREFETCH_TYPE_NONE, 0 /* No prefetch at all */ ++.set PREFETCH_TYPE_SIMPLE, 1 /* A simple, fixed-distance-ahead prefetch */ ++.set PREFETCH_TYPE_ADVANCED, 2 /* Advanced fine-grained prefetch */ ++ ++/* ++ * prefetch mode ++ * available modes are: ++ * pldl1keep ++ * pldl1strm ++ * pldl2keep ++ * pldl2strm ++ * pldl3keep ++ * pldl3strm ++ */ ++#define PREFETCH_MODE pldl1keep ++ ++/* ++ * Definitions of supplementary pixld/pixst macros (for partial load/store of ++ * pixel data). ++ */ ++ ++.macro pixldst1 op, elem_size, reg1, mem_operand, abits ++ op {v®1&.&elem_size}, [&mem_operand&], #8 ++.endm ++ ++.macro pixldst2 op, elem_size, reg1, reg2, mem_operand, abits ++ op {v®1&.&elem_size, v®2&.&elem_size}, [&mem_operand&], #16 ++.endm ++ ++.macro pixldst4 op, elem_size, reg1, reg2, reg3, reg4, mem_operand, abits ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size, v®4&.&elem_size}, [&mem_operand&], #32 ++.endm ++ ++.macro pixldst0 op, elem_size, reg1, idx, mem_operand, abits, bytes ++ op {v®1&.&elem_size}[idx], [&mem_operand&], #&bytes& ++.endm ++ ++.macro pixldst3 op, elem_size, reg1, reg2, reg3, mem_operand ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size}, [&mem_operand&], #24 ++.endm ++ ++.macro pixldst30 op, elem_size, reg1, reg2, reg3, idx, mem_operand ++ op {v®1&.&elem_size, v®2&.&elem_size, v®3&.&elem_size}[idx], [&mem_operand&], #3 ++.endm ++ ++.macro pixldst numbytes, op, elem_size, basereg, mem_operand, abits ++.if numbytes == 32 ++ .if elem_size==32 ++ pixldst4 op, 2s, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst4 op, 4h, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .else ++ pixldst4 op, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++ .endif ++.elseif numbytes == 16 ++ .if elem_size==32 ++ pixldst2 op, 2s, %(basereg+2), %(basereg+3), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst2 op, 4h, %(basereg+2), %(basereg+3), mem_operand, abits ++ .else ++ pixldst2 op, 8b, %(basereg+2), %(basereg+3), mem_operand, abits ++ .endif ++.elseif numbytes == 8 ++ .if elem_size==32 ++ pixldst1 op, 2s, %(basereg+1), mem_operand, abits ++ .elseif elem_size==16 ++ pixldst1 op, 4h, %(basereg+1), mem_operand, abits ++ .else ++ pixldst1 op, 8b, %(basereg+1), mem_operand, abits ++ .endif ++.elseif numbytes == 4 ++ .if !RESPECT_STRICT_ALIGNMENT || (elem_size == 32) ++ pixldst0 op, s, %(basereg+0), 1, mem_operand, abits, 4 ++ .elseif elem_size == 16 ++ pixldst0 op, h, %(basereg+0), 2, mem_operand, abits, 2 ++ pixldst0 op, h, %(basereg+0), 3, mem_operand, abits, 2 ++ .else ++ pixldst0 op, b, %(basereg+0), 4, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 5, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 6, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 7, mem_operand, abits, 1 ++ .endif ++.elseif numbytes == 2 ++ .if !RESPECT_STRICT_ALIGNMENT || (elem_size == 16) ++ pixldst0 op, h, %(basereg+0), 1, mem_operand, abits, 2 ++ .else ++ pixldst0 op, b, %(basereg+0), 2, mem_operand, abits, 1 ++ pixldst0 op, b, %(basereg+0), 3, mem_operand, abits, 1 ++ .endif ++.elseif numbytes == 1 ++ pixldst0 op, b, %(basereg+0), 1, mem_operand, abits, 1 ++.else ++ .error "unsupported size: numbytes" ++.endif ++.endm ++ ++.macro pixld numpix, bpp, basereg, mem_operand, abits=0 ++.if bpp > 0 ++.if (bpp == 32) && (numpix == 8) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ pixldst4 ld4, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++.elseif (bpp == 24) && (numpix == 8) ++ pixldst3 ld3, 8b, %(basereg+3), %(basereg+4), %(basereg+5), mem_operand ++.elseif (bpp == 24) && (numpix == 4) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 4, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 5, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 6, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 7, mem_operand ++.elseif (bpp == 24) && (numpix == 2) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 2, mem_operand ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 3, mem_operand ++.elseif (bpp == 24) && (numpix == 1) ++ pixldst30 ld3, b, %(basereg+0), %(basereg+1), %(basereg+2), 1, mem_operand ++.else ++ pixldst %(numpix * bpp / 8), ld1, %(bpp), basereg, mem_operand, abits ++.endif ++.endif ++.endm ++ ++.macro pixst numpix, bpp, basereg, mem_operand, abits=0 ++.if bpp > 0 ++.if (bpp == 32) && (numpix == 8) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ pixldst4 st4, 8b, %(basereg+4), %(basereg+5), \ ++ %(basereg+6), %(basereg+7), mem_operand, abits ++.elseif (bpp == 24) && (numpix == 8) ++ pixldst3 st3, 8b, %(basereg+3), %(basereg+4), %(basereg+5), mem_operand ++.elseif (bpp == 24) && (numpix == 4) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 4, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 5, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 6, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 7, mem_operand ++.elseif (bpp == 24) && (numpix == 2) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 2, mem_operand ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 3, mem_operand ++.elseif (bpp == 24) && (numpix == 1) ++ pixldst30 st3, b, %(basereg+0), %(basereg+1), %(basereg+2), 1, mem_operand ++.elseif numpix * bpp == 32 && abits == 32 ++ pixldst 4, st1, 32, basereg, mem_operand, abits ++.elseif numpix * bpp == 16 && abits == 16 ++ pixldst 2, st1, 16, basereg, mem_operand, abits ++.else ++ pixldst %(numpix * bpp / 8), st1, %(bpp), basereg, mem_operand, abits ++.endif ++.endif ++.endm ++ ++.macro pixld_a numpix, bpp, basereg, mem_operand ++.if (bpp * numpix) <= 128 ++ pixld numpix, bpp, basereg, mem_operand, %(bpp * numpix) ++.else ++ pixld numpix, bpp, basereg, mem_operand, 128 ++.endif ++.endm ++ ++.macro pixst_a numpix, bpp, basereg, mem_operand ++.if (bpp * numpix) <= 128 ++ pixst numpix, bpp, basereg, mem_operand, %(bpp * numpix) ++.else ++ pixst numpix, bpp, basereg, mem_operand, 128 ++.endif ++.endm ++ ++/* ++ * Pixel fetcher for nearest scaling (needs TMP1, TMP2, VX, UNIT_X register ++ * aliases to be defined) ++ */ ++.macro pixld1_s elem_size, reg1, mem_operand ++.if elem_size == 16 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #1 ++ ld1 {v®1&.h}[0], [TMP1] ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ ld1 {v®1&.h}[1], [TMP2] ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #1 ++ ld1 {v®1&.h}[2], [TMP1] ++ ld1 {v®1&.h}[3], [TMP2] ++.elseif elem_size == 32 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #2 ++ asr TMP2, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP2, mem_operand, TMP2, lsl #2 ++ ld1 {v®1&.s}[0], [TMP1] ++ ld1 {v®1&.s}[1], [TMP2] ++.else ++ .error "unsupported" ++.endif ++.endm ++ ++.macro pixld2_s elem_size, reg1, reg2, mem_operand ++.if 0 /* elem_size == 32 */ ++ mov TMP1, VX, asr #16 ++ add VX, VX, UNIT_X, asl #1 ++ add TMP1, mem_operand, TMP1, asl #2 ++ mov TMP2, VX, asr #16 ++ sub VX, VX, UNIT_X ++ add TMP2, mem_operand, TMP2, asl #2 ++ ld1 {v®1&.s}[0], [TMP1] ++ mov TMP1, VX, asr #16 ++ add VX, VX, UNIT_X, asl #1 ++ add TMP1, mem_operand, TMP1, asl #2 ++ ld1 {v®2&.s}[0], [TMP2, :32] ++ mov TMP2, VX, asr #16 ++ add VX, VX, UNIT_X ++ add TMP2, mem_operand, TMP2, asl #2 ++ ld1 {v®1&.s}[1], [TMP1] ++ ld1 {v®2&.s}[1], [TMP2] ++.else ++ pixld1_s elem_size, reg1, mem_operand ++ pixld1_s elem_size, reg2, mem_operand ++.endif ++.endm ++ ++.macro pixld0_s elem_size, reg1, idx, mem_operand ++.if elem_size == 16 ++ asr TMP1, VX, #16 ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #1 ++ ld1 {v®1&.h}[idx], [TMP1] ++.elseif elem_size == 32 ++ asr DUMMY, VX, #16 ++ mov TMP1, DUMMY ++ adds VX, VX, UNIT_X ++ bmi 55f ++5: subs VX, VX, SRC_WIDTH_FIXED ++ bpl 5b ++55: ++ add TMP1, mem_operand, TMP1, lsl #2 ++ ld1 {v®1&.s}[idx], [TMP1] ++.endif ++.endm ++ ++.macro pixld_s_internal numbytes, elem_size, basereg, mem_operand ++.if numbytes == 32 ++ pixld2_s elem_size, %(basereg+4), %(basereg+5), mem_operand ++ pixld2_s elem_size, %(basereg+6), %(basereg+7), mem_operand ++ pixdeinterleave elem_size, %(basereg+4) ++.elseif numbytes == 16 ++ pixld2_s elem_size, %(basereg+2), %(basereg+3), mem_operand ++.elseif numbytes == 8 ++ pixld1_s elem_size, %(basereg+1), mem_operand ++.elseif numbytes == 4 ++ .if elem_size == 32 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++ .elseif elem_size == 16 ++ pixld0_s elem_size, %(basereg+0), 2, mem_operand ++ pixld0_s elem_size, %(basereg+0), 3, mem_operand ++ .else ++ pixld0_s elem_size, %(basereg+0), 4, mem_operand ++ pixld0_s elem_size, %(basereg+0), 5, mem_operand ++ pixld0_s elem_size, %(basereg+0), 6, mem_operand ++ pixld0_s elem_size, %(basereg+0), 7, mem_operand ++ .endif ++.elseif numbytes == 2 ++ .if elem_size == 16 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++ .else ++ pixld0_s elem_size, %(basereg+0), 2, mem_operand ++ pixld0_s elem_size, %(basereg+0), 3, mem_operand ++ .endif ++.elseif numbytes == 1 ++ pixld0_s elem_size, %(basereg+0), 1, mem_operand ++.else ++ .error "unsupported size: numbytes" ++.endif ++.endm ++ ++.macro pixld_s numpix, bpp, basereg, mem_operand ++.if bpp > 0 ++ pixld_s_internal %(numpix * bpp / 8), %(bpp), basereg, mem_operand ++.endif ++.endm ++ ++.macro vuzp8 reg1, reg2 ++ umov DUMMY, v16.d[0] ++ uzp1 v16.8b, v®1&.8b, v®2&.8b ++ uzp2 v®2&.8b, v®1&.8b, v®2&.8b ++ mov v®1&.8b, v16.8b ++ mov v16.d[0], DUMMY ++.endm ++ ++.macro vzip8 reg1, reg2 ++ umov DUMMY, v16.d[0] ++ zip1 v16.8b, v®1&.8b, v®2&.8b ++ zip2 v®2&.8b, v®1&.8b, v®2&.8b ++ mov v®1&.8b, v16.8b ++ mov v16.d[0], DUMMY ++.endm ++ ++/* deinterleave B, G, R, A channels for eight 32bpp pixels in 4 registers */ ++.macro pixdeinterleave bpp, basereg ++.if (bpp == 32) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ vuzp8 %(basereg+0), %(basereg+1) ++ vuzp8 %(basereg+2), %(basereg+3) ++ vuzp8 %(basereg+1), %(basereg+3) ++ vuzp8 %(basereg+0), %(basereg+2) ++.endif ++.endm ++ ++/* interleave B, G, R, A channels for eight 32bpp pixels in 4 registers */ ++.macro pixinterleave bpp, basereg ++.if (bpp == 32) && (DEINTERLEAVE_32BPP_ENABLED != 0) ++ vzip8 %(basereg+0), %(basereg+2) ++ vzip8 %(basereg+1), %(basereg+3) ++ vzip8 %(basereg+2), %(basereg+3) ++ vzip8 %(basereg+0), %(basereg+1) ++.endif ++.endm ++ ++/* ++ * This is a macro for implementing cache preload. The main idea is that ++ * cache preload logic is mostly independent from the rest of pixels ++ * processing code. It starts at the top left pixel and moves forward ++ * across pixels and can jump across scanlines. Prefetch distance is ++ * handled in an 'incremental' way: it starts from 0 and advances to the ++ * optimal distance over time. After reaching optimal prefetch distance, ++ * it is kept constant. There are some checks which prevent prefetching ++ * unneeded pixel lines below the image (but it still can prefetch a bit ++ * more data on the right side of the image - not a big issue and may ++ * be actually helpful when rendering text glyphs). Additional trick is ++ * the use of LDR instruction for prefetch instead of PLD when moving to ++ * the next line, the point is that we have a high chance of getting TLB ++ * miss in this case, and PLD would be useless. ++ * ++ * This sounds like it may introduce a noticeable overhead (when working with ++ * fully cached data). But in reality, due to having a separate pipeline and ++ * instruction queue for NEON unit in ARM Cortex-A8, normal ARM code can ++ * execute simultaneously with NEON and be completely shadowed by it. Thus ++ * we get no performance overhead at all (*). This looks like a very nice ++ * feature of Cortex-A8, if used wisely. We don't have a hardware prefetcher, ++ * but still can implement some rather advanced prefetch logic in software ++ * for almost zero cost! ++ * ++ * (*) The overhead of the prefetcher is visible when running some trivial ++ * pixels processing like simple copy. Anyway, having prefetch is a must ++ * when working with the graphics data. ++ */ ++.macro PF a, x:vararg ++.if (PREFETCH_TYPE_CURRENT == PREFETCH_TYPE_ADVANCED) ++ a x ++.endif ++.endm ++ ++.macro cache_preload std_increment, boost_increment ++.if (src_bpp_shift >= 0) || (dst_r_bpp != 0) || (mask_bpp_shift >= 0) ++.if std_increment != 0 ++ PF add PF_X, PF_X, #std_increment ++.endif ++ PF tst PF_CTL, #0xF ++ PF beq 71f ++ PF add PF_X, PF_X, #boost_increment ++ PF sub PF_CTL, PF_CTL, #1 ++71: ++ PF cmp PF_X, ORIG_W ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, PF_X, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_SRC, DUMMY] ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, PF_X, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_DST, DUMMY] ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, PF_X, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [PF_MASK, DUMMY] ++.endif ++ PF ble 71f ++ PF sub PF_X, PF_X, ORIG_W ++ PF subs PF_CTL, PF_CTL, #0x10 ++71: ++ PF ble 72f ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF ldrsb DUMMY, [PF_SRC, DUMMY] ++ PF add PF_SRC, PF_SRC, #1 ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF ldrsb DUMMY, [PF_DST, DUMMY] ++ PF add PF_DST, PF_DST, #1 ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF ldrsb DUMMY, [PF_MASK, DUMMY] ++ PF add PF_MASK, PF_MASK, #1 ++.endif ++72: ++.endif ++.endm ++ ++.macro cache_preload_simple ++.if (PREFETCH_TYPE_CURRENT == PREFETCH_TYPE_SIMPLE) ++.if src_bpp > 0 ++ prfm PREFETCH_MODE, [SRC, #(PREFETCH_DISTANCE_SIMPLE * src_bpp / 8)] ++.endif ++.if dst_r_bpp > 0 ++ prfm PREFETCH_MODE, [DST_R, #(PREFETCH_DISTANCE_SIMPLE * dst_r_bpp / 8)] ++.endif ++.if mask_bpp > 0 ++ prfm PREFETCH_MODE, [MASK, #(PREFETCH_DISTANCE_SIMPLE * mask_bpp / 8)] ++.endif ++.endif ++.endm ++ ++.macro fetch_mask_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++.endm ++ ++/* ++ * Macro which is used to process leading pixels until destination ++ * pointer is properly aligned (at 16 bytes boundary). When destination ++ * buffer uses 16bpp format, this is unnecessary, or even pointless. ++ */ ++.macro ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++.if dst_w_bpp != 24 ++ tst DST_R, #0xF ++ beq 52f ++ ++.if src_bpp > 0 || mask_bpp > 0 || dst_r_bpp > 0 ++.irp lowbit, 1, 2, 4, 8, 16 ++local skip1 ++.if (dst_w_bpp <= (lowbit * 8)) && ((lowbit * 8) < (pixblock_size * dst_w_bpp)) ++.if lowbit < 16 /* we don't need more than 16-byte alignment */ ++ tst DST_R, #lowbit ++ beq 51f ++.endif ++ pixld_src (lowbit * 8 / dst_w_bpp), src_bpp, src_basereg, SRC ++ pixld (lowbit * 8 / dst_w_bpp), mask_bpp, mask_basereg, MASK ++.if dst_r_bpp > 0 ++ pixld_a (lowbit * 8 / dst_r_bpp), dst_r_bpp, dst_r_basereg, DST_R ++.else ++ add DST_R, DST_R, #lowbit ++.endif ++ PF add PF_X, PF_X, #(lowbit * 8 / dst_w_bpp) ++ sub W, W, #(lowbit * 8 / dst_w_bpp) ++51: ++.endif ++.endr ++.endif ++ pixdeinterleave src_bpp, src_basereg ++ pixdeinterleave mask_bpp, mask_basereg ++ pixdeinterleave dst_r_bpp, dst_r_basereg ++ ++ process_pixblock_head ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++ process_pixblock_tail ++ ++ pixinterleave dst_w_bpp, dst_w_basereg ++ ++.irp lowbit, 1, 2, 4, 8, 16 ++.if (dst_w_bpp <= (lowbit * 8)) && ((lowbit * 8) < (pixblock_size * dst_w_bpp)) ++.if lowbit < 16 /* we don't need more than 16-byte alignment */ ++ tst DST_W, #lowbit ++ beq 51f ++.endif ++.if src_bpp == 0 && mask_bpp == 0 && dst_r_bpp == 0 ++ sub W, W, #(lowbit * 8 / dst_w_bpp) ++.endif ++ pixst_a (lowbit * 8 / dst_w_bpp), dst_w_bpp, dst_w_basereg, DST_W ++51: ++.endif ++.endr ++.endif ++52: ++.endm ++ ++/* ++ * Special code for processing up to (pixblock_size - 1) remaining ++ * trailing pixels. As SIMD processing performs operation on ++ * pixblock_size pixels, anything smaller than this has to be loaded ++ * and stored in a special way. Loading and storing of pixel data is ++ * performed in such a way that we fill some 'slots' in the NEON ++ * registers (some slots naturally are unused), then perform compositing ++ * operation as usual. In the end, the data is taken from these 'slots' ++ * and saved to memory. ++ * ++ * cache_preload_flag - allows to suppress prefetch if ++ * set to 0 ++ * dst_aligned_flag - selects whether destination buffer ++ * is aligned ++ */ ++.macro process_trailing_pixels cache_preload_flag, \ ++ dst_aligned_flag, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ tst W, #(pixblock_size - 1) ++ beq 52f ++.if src_bpp > 0 || mask_bpp > 0 || dst_r_bpp > 0 ++.irp chunk_size, 16, 8, 4, 2, 1 ++.if pixblock_size > chunk_size ++ tst W, #chunk_size ++ beq 51f ++ pixld_src chunk_size, src_bpp, src_basereg, SRC ++ pixld chunk_size, mask_bpp, mask_basereg, MASK ++.if dst_aligned_flag != 0 ++ pixld_a chunk_size, dst_r_bpp, dst_r_basereg, DST_R ++.else ++ pixld chunk_size, dst_r_bpp, dst_r_basereg, DST_R ++.endif ++.if cache_preload_flag != 0 ++ PF add PF_X, PF_X, #chunk_size ++.endif ++51: ++.endif ++.endr ++.endif ++ pixdeinterleave src_bpp, src_basereg ++ pixdeinterleave mask_bpp, mask_basereg ++ pixdeinterleave dst_r_bpp, dst_r_basereg ++ ++ process_pixblock_head ++.if cache_preload_flag != 0 ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++.endif ++ process_pixblock_tail ++ pixinterleave dst_w_bpp, dst_w_basereg ++.irp chunk_size, 16, 8, 4, 2, 1 ++.if pixblock_size > chunk_size ++ tst W, #chunk_size ++ beq 51f ++.if dst_aligned_flag != 0 ++ pixst_a chunk_size, dst_w_bpp, dst_w_basereg, DST_W ++.else ++ pixst chunk_size, dst_w_bpp, dst_w_basereg, DST_W ++.endif ++51: ++.endif ++.endr ++52: ++.endm ++ ++/* ++ * Macro, which performs all the needed operations to switch to the next ++ * scanline and start the next loop iteration unless all the scanlines ++ * are already processed. ++ */ ++.macro advance_to_next_scanline start_of_loop_label ++ mov W, ORIG_W ++ add DST_W, DST_W, DST_STRIDE, lsl #dst_bpp_shift ++.if src_bpp != 0 ++ add SRC, SRC, SRC_STRIDE, lsl #src_bpp_shift ++.endif ++.if mask_bpp != 0 ++ add MASK, MASK, MASK_STRIDE, lsl #mask_bpp_shift ++.endif ++.if (dst_w_bpp != 24) ++ sub DST_W, DST_W, W, lsl #dst_bpp_shift ++.endif ++.if (src_bpp != 24) && (src_bpp != 0) ++ sub SRC, SRC, W, lsl #src_bpp_shift ++.endif ++.if (mask_bpp != 24) && (mask_bpp != 0) ++ sub MASK, MASK, W, lsl #mask_bpp_shift ++.endif ++ subs H, H, #1 ++ mov DST_R, DST_W ++ bge start_of_loop_label ++.endm ++ ++/* ++ * Registers are allocated in the following way by default: ++ * v0, v1, v2, v3 - reserved for loading source pixel data ++ * v4, v5, v6, v7 - reserved for loading destination pixel data ++ * v24, v25, v26, v27 - reserved for loading mask pixel data ++ * v28, v29, v30, v31 - final destination pixel data for writeback to memory ++ */ ++.macro generate_composite_function fname, \ ++ src_bpp_, \ ++ mask_bpp_, \ ++ dst_w_bpp_, \ ++ flags, \ ++ pixblock_size_, \ ++ prefetch_distance, \ ++ init, \ ++ cleanup, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head, \ ++ dst_w_basereg_ = 28, \ ++ dst_r_basereg_ = 4, \ ++ src_basereg_ = 0, \ ++ mask_basereg_ = 24 ++ ++ pixman_asm_function fname ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 232 /* push all registers */ ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], #32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], #32 ++ stp x8, x9, [x29, -80] ++ stp x10, x11, [x29, -96] ++ stp x12, x13, [x29, -112] ++ stp x14, x15, [x29, -128] ++ stp x16, x17, [x29, -144] ++ stp x18, x19, [x29, -160] ++ stp x20, x21, [x29, -176] ++ stp x22, x23, [x29, -192] ++ stp x24, x25, [x29, -208] ++ stp x26, x27, [x29, -224] ++ str x28, [x29, -232] ++ ++/* ++ * Select prefetch type for this function. If prefetch distance is ++ * set to 0 or one of the color formats is 24bpp, SIMPLE prefetch ++ * has to be used instead of ADVANCED. ++ */ ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_DEFAULT ++.if prefetch_distance == 0 ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_NONE ++.elseif (PREFETCH_TYPE_CURRENT > PREFETCH_TYPE_SIMPLE) && \ ++ ((src_bpp_ == 24) || (mask_bpp_ == 24) || (dst_w_bpp_ == 24)) ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_SIMPLE ++.endif ++ ++/* ++ * Make some macro arguments globally visible and accessible ++ * from other macros ++ */ ++ .set src_bpp, src_bpp_ ++ .set mask_bpp, mask_bpp_ ++ .set dst_w_bpp, dst_w_bpp_ ++ .set pixblock_size, pixblock_size_ ++ .set dst_w_basereg, dst_w_basereg_ ++ .set dst_r_basereg, dst_r_basereg_ ++ .set src_basereg, src_basereg_ ++ .set mask_basereg, mask_basereg_ ++ ++ .macro pixld_src x:vararg ++ pixld x ++ .endm ++ .macro fetch_src_pixblock ++ pixld_src pixblock_size, src_bpp, \ ++ (src_basereg - pixblock_size * src_bpp / 64), SRC ++ .endm ++/* ++ * Assign symbolic names to registers ++ */ ++ W .req x0 /* width (is updated during processing) */ ++ H .req x1 /* height (is updated during processing) */ ++ DST_W .req x2 /* destination buffer pointer for writes */ ++ DST_STRIDE .req x3 /* destination image stride */ ++ SRC .req x4 /* source buffer pointer */ ++ SRC_STRIDE .req x5 /* source image stride */ ++ MASK .req x6 /* mask pointer */ ++ MASK_STRIDE .req x7 /* mask stride */ ++ ++ DST_R .req x8 /* destination buffer pointer for reads */ ++ ++ PF_CTL .req x9 /* combined lines counter and prefetch */ ++ /* distance increment counter */ ++ PF_X .req x10 /* pixel index in a scanline for current */ ++ /* pretetch position */ ++ PF_SRC .req x11 /* pointer to source scanline start */ ++ /* for prefetch purposes */ ++ PF_DST .req x12 /* pointer to destination scanline start */ ++ /* for prefetch purposes */ ++ PF_MASK .req x13 /* pointer to mask scanline start */ ++ /* for prefetch purposes */ ++ ++ ORIG_W .req x14 /* saved original width */ ++ DUMMY .req x15 /* temporary register */ ++ ++ sxtw x0, w0 ++ sxtw x1, w1 ++ sxtw x3, w3 ++ sxtw x5, w5 ++ sxtw x7, w7 ++ ++ .set mask_bpp_shift, -1 ++.if src_bpp == 32 ++ .set src_bpp_shift, 2 ++.elseif src_bpp == 24 ++ .set src_bpp_shift, 0 ++.elseif src_bpp == 16 ++ .set src_bpp_shift, 1 ++.elseif src_bpp == 8 ++ .set src_bpp_shift, 0 ++.elseif src_bpp == 0 ++ .set src_bpp_shift, -1 ++.else ++ .error "requested src bpp (src_bpp) is not supported" ++.endif ++.if mask_bpp == 32 ++ .set mask_bpp_shift, 2 ++.elseif mask_bpp == 24 ++ .set mask_bpp_shift, 0 ++.elseif mask_bpp == 8 ++ .set mask_bpp_shift, 0 ++.elseif mask_bpp == 0 ++ .set mask_bpp_shift, -1 ++.else ++ .error "requested mask bpp (mask_bpp) is not supported" ++.endif ++.if dst_w_bpp == 32 ++ .set dst_bpp_shift, 2 ++.elseif dst_w_bpp == 24 ++ .set dst_bpp_shift, 0 ++.elseif dst_w_bpp == 16 ++ .set dst_bpp_shift, 1 ++.elseif dst_w_bpp == 8 ++ .set dst_bpp_shift, 0 ++.else ++ .error "requested dst bpp (dst_w_bpp) is not supported" ++.endif ++ ++.if (((flags) & FLAG_DST_READWRITE) != 0) ++ .set dst_r_bpp, dst_w_bpp ++.else ++ .set dst_r_bpp, 0 ++.endif ++.if (((flags) & FLAG_DEINTERLEAVE_32BPP) != 0) ++ .set DEINTERLEAVE_32BPP_ENABLED, 1 ++.else ++ .set DEINTERLEAVE_32BPP_ENABLED, 0 ++.endif ++ ++.if prefetch_distance < 0 || prefetch_distance > 15 ++ .error "invalid prefetch distance (prefetch_distance)" ++.endif ++ ++ PF mov PF_X, #0 ++ mov DST_R, DST_W ++ ++.if src_bpp == 24 ++ sub SRC_STRIDE, SRC_STRIDE, W ++ sub SRC_STRIDE, SRC_STRIDE, W, lsl #1 ++.endif ++.if mask_bpp == 24 ++ sub MASK_STRIDE, MASK_STRIDE, W ++ sub MASK_STRIDE, MASK_STRIDE, W, lsl #1 ++.endif ++.if dst_w_bpp == 24 ++ sub DST_STRIDE, DST_STRIDE, W ++ sub DST_STRIDE, DST_STRIDE, W, lsl #1 ++.endif ++ ++/* ++ * Setup advanced prefetcher initial state ++ */ ++ PF mov PF_SRC, SRC ++ PF mov PF_DST, DST_R ++ PF mov PF_MASK, MASK ++ /* PF_CTL = prefetch_distance | ((h - 1) << 4) */ ++ PF lsl DUMMY, H, #4 ++ PF mov PF_CTL, DUMMY ++ PF add PF_CTL, PF_CTL, #(prefetch_distance - 0x10) ++ ++ init ++ subs H, H, #1 ++ mov ORIG_W, W ++ blt 9f ++ cmp W, #(pixblock_size * 2) ++ blt 800f ++/* ++ * This is the start of the pipelined loop, which if optimized for ++ * long scanlines ++ */ ++0: ++ ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ /* Implement "head (tail_head) ... (tail_head) tail" loop pattern */ ++ pixld_a pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ PF add PF_X, PF_X, #pixblock_size ++ process_pixblock_head ++ cache_preload 0, pixblock_size ++ cache_preload_simple ++ subs W, W, #(pixblock_size * 2) ++ blt 200f ++ ++100: ++ process_pixblock_tail_head ++ cache_preload_simple ++ subs W, W, #pixblock_size ++ bge 100b ++ ++200: ++ process_pixblock_tail ++ pixst_a pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++ ++ /* Process the remaining trailing pixels in the scanline */ ++ process_trailing_pixels 1, 1, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ advance_to_next_scanline 0b ++ ++ cleanup ++1000: ++ /* pop all registers */ ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -112] ++ ldp x14, x15, [x29, -128] ++ ldp x16, x17, [x29, -144] ++ ldp x18, x19, [x29, -160] ++ ldp x20, x21, [x29, -176] ++ ldp x22, x23, [x29, -192] ++ ldp x24, x25, [x29, -208] ++ ldp x26, x27, [x29, -224] ++ ldr x28, [x29, -232] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++/* ++ * This is the start of the loop, designed to process images with small width ++ * (less than pixblock_size * 2 pixels). In this case neither pipelining ++ * nor prefetch are used. ++ */ ++800: ++.if src_bpp_shift >= 0 ++ PF lsl DUMMY, SRC_STRIDE, #src_bpp_shift ++ PF prfm PREFETCH_MODE, [SRC, DUMMY] ++.endif ++.if dst_r_bpp != 0 ++ PF lsl DUMMY, DST_STRIDE, #dst_bpp_shift ++ PF prfm PREFETCH_MODE, [DST_R, DUMMY] ++.endif ++.if mask_bpp_shift >= 0 ++ PF lsl DUMMY, MASK_STRIDE, #mask_bpp_shift ++ PF prfm PREFETCH_MODE, [MASK, DUMMY] ++.endif ++ /* Process exactly pixblock_size pixels if needed */ ++ tst W, #pixblock_size ++ beq 100f ++ pixld pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ process_pixblock_head ++ process_pixblock_tail ++ pixst pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++100: ++ /* Process the remaining trailing pixels in the scanline */ ++ process_trailing_pixels 0, 0, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ advance_to_next_scanline 800b ++9: ++ cleanup ++ /* pop all registers */ ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldp x10, x11, [x29, -96] ++ ldp x12, x13, [x29, -112] ++ ldp x14, x15, [x29, -128] ++ ldp x16, x17, [x29, -144] ++ ldp x18, x19, [x29, -160] ++ ldp x20, x21, [x29, -176] ++ ldp x22, x23, [x29, -192] ++ ldp x24, x25, [x29, -208] ++ ldp x26, x27, [x29, -224] ++ ldr x28, [x29, -232] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .purgem fetch_src_pixblock ++ .purgem pixld_src ++ ++ .unreq SRC ++ .unreq MASK ++ .unreq DST_R ++ .unreq DST_W ++ .unreq ORIG_W ++ .unreq W ++ .unreq H ++ .unreq SRC_STRIDE ++ .unreq DST_STRIDE ++ .unreq MASK_STRIDE ++ .unreq PF_CTL ++ .unreq PF_X ++ .unreq PF_SRC ++ .unreq PF_DST ++ .unreq PF_MASK ++ .unreq DUMMY ++ .endfunc ++.endm ++ ++/* ++ * A simplified variant of function generation template for a single ++ * scanline processing (for implementing pixman combine functions) ++ */ ++.macro generate_composite_function_scanline use_nearest_scaling, \ ++ fname, \ ++ src_bpp_, \ ++ mask_bpp_, \ ++ dst_w_bpp_, \ ++ flags, \ ++ pixblock_size_, \ ++ init, \ ++ cleanup, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head, \ ++ dst_w_basereg_ = 28, \ ++ dst_r_basereg_ = 4, \ ++ src_basereg_ = 0, \ ++ mask_basereg_ = 24 ++ ++ pixman_asm_function fname ++ .set PREFETCH_TYPE_CURRENT, PREFETCH_TYPE_NONE ++ ++/* ++ * Make some macro arguments globally visible and accessible ++ * from other macros ++ */ ++ .set src_bpp, src_bpp_ ++ .set mask_bpp, mask_bpp_ ++ .set dst_w_bpp, dst_w_bpp_ ++ .set pixblock_size, pixblock_size_ ++ .set dst_w_basereg, dst_w_basereg_ ++ .set dst_r_basereg, dst_r_basereg_ ++ .set src_basereg, src_basereg_ ++ .set mask_basereg, mask_basereg_ ++ ++.if use_nearest_scaling != 0 ++ /* ++ * Assign symbolic names to registers for nearest scaling ++ */ ++ W .req x0 ++ DST_W .req x1 ++ SRC .req x2 ++ VX .req x3 ++ UNIT_X .req x4 ++ SRC_WIDTH_FIXED .req x5 ++ MASK .req x6 ++ TMP1 .req x8 ++ TMP2 .req x9 ++ DST_R .req x10 ++ DUMMY .req x30 ++ ++ .macro pixld_src x:vararg ++ pixld_s x ++ .endm ++ ++ sxtw x0, w0 ++ sxtw x3, w3 ++ sxtw x4, w4 ++ sxtw x5, w5 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 88 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ stp x8, x9, [x29, -80] ++ str x10, [x29, -88] ++.else ++ /* ++ * Assign symbolic names to registers ++ */ ++ W .req x0 /* width (is updated during processing) */ ++ DST_W .req x1 /* destination buffer pointer for writes */ ++ SRC .req x2 /* source buffer pointer */ ++ MASK .req x3 /* mask pointer */ ++ DST_R .req x4 /* destination buffer pointer for reads */ ++ DUMMY .req x30 ++ ++ .macro pixld_src x:vararg ++ pixld x ++ .endm ++ ++ sxtw x0, w0 ++ ++ stp x29, x30, [sp, -16]! ++ mov x29, sp ++ sub sp, sp, 64 ++ sub x29, x29, 64 ++ st1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ st1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++.endif ++ ++.if (((flags) & FLAG_DST_READWRITE) != 0) ++ .set dst_r_bpp, dst_w_bpp ++.else ++ .set dst_r_bpp, 0 ++.endif ++.if (((flags) & FLAG_DEINTERLEAVE_32BPP) != 0) ++ .set DEINTERLEAVE_32BPP_ENABLED, 1 ++.else ++ .set DEINTERLEAVE_32BPP_ENABLED, 0 ++.endif ++ ++ .macro fetch_src_pixblock ++ pixld_src pixblock_size, src_bpp, \ ++ (src_basereg - pixblock_size * src_bpp / 64), SRC ++ .endm ++ ++ init ++ mov DST_R, DST_W ++ ++ cmp W, #pixblock_size ++ blt 800f ++ ++ ensure_destination_ptr_alignment process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ subs W, W, #pixblock_size ++ blt 700f ++ ++ /* Implement "head (tail_head) ... (tail_head) tail" loop pattern */ ++ pixld_a pixblock_size, dst_r_bpp, \ ++ (dst_r_basereg - pixblock_size * dst_r_bpp / 64), DST_R ++ fetch_src_pixblock ++ pixld pixblock_size, mask_bpp, \ ++ (mask_basereg - pixblock_size * mask_bpp / 64), MASK ++ process_pixblock_head ++ subs W, W, #pixblock_size ++ blt 200f ++100: ++ process_pixblock_tail_head ++ subs W, W, #pixblock_size ++ bge 100b ++200: ++ process_pixblock_tail ++ pixst_a pixblock_size, dst_w_bpp, \ ++ (dst_w_basereg - pixblock_size * dst_w_bpp / 64), DST_W ++700: ++ /* Process the remaining trailing pixels in the scanline (dst aligned) */ ++ process_trailing_pixels 0, 1, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ cleanup ++.if use_nearest_scaling != 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldr x10, [x29, -96] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++.endif ++800: ++ /* Process the remaining trailing pixels in the scanline (dst unaligned) */ ++ process_trailing_pixels 0, 0, \ ++ process_pixblock_head, \ ++ process_pixblock_tail, \ ++ process_pixblock_tail_head ++ ++ cleanup ++.if use_nearest_scaling != 0 ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ ldp x8, x9, [x29, -80] ++ ldr x10, [x29, -88] ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .unreq DUMMY ++ .unreq DST_R ++ .unreq SRC ++ .unreq W ++ .unreq VX ++ .unreq UNIT_X ++ .unreq TMP1 ++ .unreq TMP2 ++ .unreq DST_W ++ .unreq MASK ++ .unreq SRC_WIDTH_FIXED ++ ++.else ++ sub x29, x29, 64 ++ ld1 {v8.8b, v9.8b, v10.8b, v11.8b}, [x29], 32 ++ ld1 {v12.8b, v13.8b, v14.8b, v15.8b}, [x29], 32 ++ mov sp, x29 ++ ldp x29, x30, [sp], 16 ++ ret /* exit */ ++ ++ .unreq DUMMY ++ .unreq SRC ++ .unreq MASK ++ .unreq DST_R ++ .unreq DST_W ++ .unreq W ++.endif ++ ++ .purgem fetch_src_pixblock ++ .purgem pixld_src ++ ++ .endfunc ++.endm ++ ++.macro generate_composite_function_single_scanline x:vararg ++ generate_composite_function_scanline 0, x ++.endm ++ ++.macro generate_composite_function_nearest_scanline x:vararg ++ generate_composite_function_scanline 1, x ++.endm ++ ++/* Default prologue/epilogue, nothing special needs to be done */ ++ ++.macro default_init ++.endm ++ ++.macro default_cleanup ++.endm ++ ++/* ++ * Prologue/epilogue variant which additionally saves/restores v8-v15 ++ * registers (they need to be saved/restored by callee according to ABI). ++ * This is required if the code needs to use all the NEON registers. ++ */ ++ ++.macro default_init_need_all_regs ++.endm ++ ++.macro default_cleanup_need_all_regs ++.endm ++ ++/******************************************************************************/ ++ ++/* ++ * Conversion of 8 r5g6b6 pixels packed in 128-bit register (in) ++ * into a planar a8r8g8b8 format (with a, r, g, b color components ++ * stored into 64-bit registers out_a, out_r, out_g, out_b respectively). ++ * ++ * Warning: the conversion is destructive and the original ++ * value (in) is lost. ++ */ ++.macro convert_0565_to_8888 in, out_a, out_r, out_g, out_b ++ shrn &out_r&.8b, &in&.8h, #8 ++ shrn &out_g&.8b, &in&.8h, #3 ++ sli &in&.8h, &in&.8h, #5 ++ movi &out_a&.8b, #255 ++ sri &out_r&.8b, &out_r&.8b, #5 ++ sri &out_g&.8b, &out_g&.8b, #6 ++ shrn &out_b&.8b, &in&.8h, #2 ++.endm ++ ++.macro convert_0565_to_x888 in, out_r, out_g, out_b ++ shrn &out_r&.8b, &in&.8h, #8 ++ shrn &out_g&.8b, &in&.8h, #3 ++ sli &in&.8h, &in&.8h, #5 ++ sri &out_r&.8b, &out_r&.8b, #5 ++ sri &out_g&.8b, &out_g&.8b, #6 ++ shrn &out_b&.8b, &in&.8h, #2 ++.endm ++ ++/* ++ * Conversion from planar a8r8g8b8 format (with a, r, g, b color components ++ * in 64-bit registers in_a, in_r, in_g, in_b respectively) into 8 r5g6b6 ++ * pixels packed in 128-bit register (out). Requires two temporary 128-bit ++ * registers (tmp1, tmp2) ++ */ ++.macro convert_8888_to_0565 in_r, in_g, in_b, out, tmp1, tmp2 ++ ushll &tmp1&.8h, &in_g&.8b, #7 ++ shl &tmp1&.8h, &tmp1&.8h, #1 ++ ushll &out&.8h, &in_r&.8b, #7 ++ shl &out&.8h, &out&.8h, #1 ++ ushll &tmp2&.8h, &in_b&.8b, #7 ++ shl &tmp2&.8h, &tmp2&.8h, #1 ++ sri &out&.8h, &tmp1&.8h, #5 ++ sri &out&.8h, &tmp2&.8h, #11 ++.endm ++ ++/* ++ * Conversion of four r5g6b5 pixels (in) to four x8r8g8b8 pixels ++ * returned in (out0, out1) registers pair. Requires one temporary ++ * 64-bit register (tmp). 'out1' and 'in' may overlap, the original ++ * value from 'in' is lost ++ */ ++.macro convert_four_0565_to_x888_packed in, out0, out1, tmp ++ shl &out0&.4h, &in&.4h, #5 /* G top 6 bits */ ++ shl &tmp&.4h, &in&.4h, #11 /* B top 5 bits */ ++ sri &in&.4h, &in&.4h, #5 /* R is ready in top bits */ ++ sri &out0&.4h, &out0&.4h, #6 /* G is ready in top bits */ ++ sri &tmp&.4h, &tmp&.4h, #5 /* B is ready in top bits */ ++ ushr &out1&.4h, &in&.4h, #8 /* R is in place */ ++ sri &out0&.4h, &tmp&.4h, #8 /* G & B is in place */ ++ zip1 &tmp&.4h, &out0&.4h, &out1&.4h /* everything is in place */ ++ zip2 &out1&.4h, &out0&.4h, &out1&.4h ++ mov &out0&.d[0], &tmp&.d[0] ++.endm +diff --git a/pixman/pixman-private.h b/pixman/pixman-private.h +index 73a5414..22c8ccc +--- a/pixman/pixman-private.h ++++ b/pixman/pixman-private.h +@@ -607,6 +607,11 @@ pixman_implementation_t * + _pixman_implementation_create_arm_neon (pixman_implementation_t *fallback); + #endif + ++#ifdef USE_ARM_A64_NEON ++pixman_implementation_t * ++_pixman_implementation_create_arm_neon (pixman_implementation_t *fallback); ++#endif ++ + #ifdef USE_MIPS_DSPR2 + pixman_implementation_t * + _pixman_implementation_create_mips_dspr2 (pixman_implementation_t *fallback); +-- +2.8.0 + diff --git a/SDK/board/funkey/patches/sdl/0001-pixel-alpha-multiply.patch b/SDK/board/funkey/patches/sdl/0001-pixel-alpha-multiply.patch new file mode 100644 index 0000000..29e6af0 --- /dev/null +++ b/SDK/board/funkey/patches/sdl/0001-pixel-alpha-multiply.patch @@ -0,0 +1,295 @@ + SDL_blit_A.c | 270 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ + 1 file changed, 270 insertions(+) + +diff --git a/src/video/SDL_blit_A.c b/src/video/SDL_blit_A.c +index 219cdcc..d8e921e 100644 +--- a/src/video/SDL_blit_A.c ++++ b/src/video/SDL_blit_A.c +@@ -27,3 +34,270 @@ ++ ++/*! ++\brief Alpha adjustment table for custom blitter. ++ ++The table provides values for a modified, non-linear ++transfer function which maintain brightness. ++ ++*/ ++const unsigned int GFX_ALPHA_ADJUST_ARRAY[256] = { ++ 0, /* 0 */ ++ 15, /* 1 */ ++ 22, /* 2 */ ++ 27, /* 3 */ ++ 31, /* 4 */ ++ 35, /* 5 */ ++ 39, /* 6 */ ++ 42, /* 7 */ ++ 45, /* 8 */ ++ 47, /* 9 */ ++ 50, /* 10 */ ++ 52, /* 11 */ ++ 55, /* 12 */ ++ 57, /* 13 */ ++ 59, /* 14 */ ++ 61, /* 15 */ ++ 63, /* 16 */ ++ 65, /* 17 */ ++ 67, /* 18 */ ++ 69, /* 19 */ ++ 71, /* 20 */ ++ 73, /* 21 */ ++ 74, /* 22 */ ++ 76, /* 23 */ ++ 78, /* 24 */ ++ 79, /* 25 */ ++ 81, /* 26 */ ++ 82, /* 27 */ ++ 84, /* 28 */ ++ 85, /* 29 */ ++ 87, /* 30 */ ++ 88, /* 31 */ ++ 90, /* 32 */ ++ 91, /* 33 */ ++ 93, /* 34 */ ++ 94, /* 35 */ ++ 95, /* 36 */ ++ 97, /* 37 */ ++ 98, /* 38 */ ++ 99, /* 39 */ ++ 100, /* 40 */ ++ 102, /* 41 */ ++ 103, /* 42 */ ++ 104, /* 43 */ ++ 105, /* 44 */ ++ 107, /* 45 */ ++ 108, /* 46 */ ++ 109, /* 47 */ ++ 110, /* 48 */ ++ 111, /* 49 */ ++ 112, /* 50 */ ++ 114, /* 51 */ ++ 115, /* 52 */ ++ 116, /* 53 */ ++ 117, /* 54 */ ++ 118, /* 55 */ ++ 119, /* 56 */ ++ 120, /* 57 */ ++ 121, /* 58 */ ++ 122, /* 59 */ ++ 123, /* 60 */ ++ 124, /* 61 */ ++ 125, /* 62 */ ++ 126, /* 63 */ ++ 127, /* 64 */ ++ 128, /* 65 */ ++ 129, /* 66 */ ++ 130, /* 67 */ ++ 131, /* 68 */ ++ 132, /* 69 */ ++ 133, /* 70 */ ++ 134, /* 71 */ ++ 135, /* 72 */ ++ 136, /* 73 */ ++ 137, /* 74 */ ++ 138, /* 75 */ ++ 139, /* 76 */ ++ 140, /* 77 */ ++ 141, /* 78 */ ++ 141, /* 79 */ ++ 142, /* 80 */ ++ 143, /* 81 */ ++ 144, /* 82 */ ++ 145, /* 83 */ ++ 146, /* 84 */ ++ 147, /* 85 */ ++ 148, /* 86 */ ++ 148, /* 87 */ ++ 149, /* 88 */ ++ 150, /* 89 */ ++ 151, /* 90 */ ++ 152, /* 91 */ ++ 153, /* 92 */ ++ 153, /* 93 */ ++ 154, /* 94 */ ++ 155, /* 95 */ ++ 156, /* 96 */ ++ 157, /* 97 */ ++ 158, /* 98 */ ++ 158, /* 99 */ ++ 159, /* 100 */ ++ 160, /* 101 */ ++ 161, /* 102 */ ++ 162, /* 103 */ ++ 162, /* 104 */ ++ 163, /* 105 */ ++ 164, /* 106 */ ++ 165, /* 107 */ ++ 165, /* 108 */ ++ 166, /* 109 */ ++ 167, /* 110 */ ++ 168, /* 111 */ ++ 168, /* 112 */ ++ 169, /* 113 */ ++ 170, /* 114 */ ++ 171, /* 115 */ ++ 171, /* 116 */ ++ 172, /* 117 */ ++ 173, /* 118 */ ++ 174, /* 119 */ ++ 174, /* 120 */ ++ 175, /* 121 */ ++ 176, /* 122 */ ++ 177, /* 123 */ ++ 177, /* 124 */ ++ 178, /* 125 */ ++ 179, /* 126 */ ++ 179, /* 127 */ ++ 180, /* 128 */ ++ 181, /* 129 */ ++ 182, /* 130 */ ++ 182, /* 131 */ ++ 183, /* 132 */ ++ 184, /* 133 */ ++ 184, /* 134 */ ++ 185, /* 135 */ ++ 186, /* 136 */ ++ 186, /* 137 */ ++ 187, /* 138 */ ++ 188, /* 139 */ ++ 188, /* 140 */ ++ 189, /* 141 */ ++ 190, /* 142 */ ++ 190, /* 143 */ ++ 191, /* 144 */ ++ 192, /* 145 */ ++ 192, /* 146 */ ++ 193, /* 147 */ ++ 194, /* 148 */ ++ 194, /* 149 */ ++ 195, /* 150 */ ++ 196, /* 151 */ ++ 196, /* 152 */ ++ 197, /* 153 */ ++ 198, /* 154 */ ++ 198, /* 155 */ ++ 199, /* 156 */ ++ 200, /* 157 */ ++ 200, /* 158 */ ++ 201, /* 159 */ ++ 201, /* 160 */ ++ 202, /* 161 */ ++ 203, /* 162 */ ++ 203, /* 163 */ ++ 204, /* 164 */ ++ 205, /* 165 */ ++ 205, /* 166 */ ++ 206, /* 167 */ ++ 206, /* 168 */ ++ 207, /* 169 */ ++ 208, /* 170 */ ++ 208, /* 171 */ ++ 209, /* 172 */ ++ 210, /* 173 */ ++ 210, /* 174 */ ++ 211, /* 175 */ ++ 211, /* 176 */ ++ 212, /* 177 */ ++ 213, /* 178 */ ++ 213, /* 179 */ ++ 214, /* 180 */ ++ 214, /* 181 */ ++ 215, /* 182 */ ++ 216, /* 183 */ ++ 216, /* 184 */ ++ 217, /* 185 */ ++ 217, /* 186 */ ++ 218, /* 187 */ ++ 218, /* 188 */ ++ 219, /* 189 */ ++ 220, /* 190 */ ++ 220, /* 191 */ ++ 221, /* 192 */ ++ 221, /* 193 */ ++ 222, /* 194 */ ++ 222, /* 195 */ ++ 223, /* 196 */ ++ 224, /* 197 */ ++ 224, /* 198 */ ++ 225, /* 199 */ ++ 225, /* 200 */ ++ 226, /* 201 */ ++ 226, /* 202 */ ++ 227, /* 203 */ ++ 228, /* 204 */ ++ 228, /* 205 */ ++ 229, /* 206 */ ++ 229, /* 207 */ ++ 230, /* 208 */ ++ 230, /* 209 */ ++ 231, /* 210 */ ++ 231, /* 211 */ ++ 232, /* 212 */ ++ 233, /* 213 */ ++ 233, /* 214 */ ++ 234, /* 215 */ ++ 234, /* 216 */ ++ 235, /* 217 */ ++ 235, /* 218 */ ++ 236, /* 219 */ ++ 236, /* 220 */ ++ 237, /* 221 */ ++ 237, /* 222 */ ++ 238, /* 223 */ ++ 238, /* 224 */ ++ 239, /* 225 */ ++ 240, /* 226 */ ++ 240, /* 227 */ ++ 241, /* 228 */ ++ 241, /* 229 */ ++ 242, /* 230 */ ++ 242, /* 231 */ ++ 243, /* 232 */ ++ 243, /* 233 */ ++ 244, /* 234 */ ++ 244, /* 235 */ ++ 245, /* 236 */ ++ 245, /* 237 */ ++ 246, /* 238 */ ++ 246, /* 239 */ ++ 247, /* 240 */ ++ 247, /* 241 */ ++ 248, /* 242 */ ++ 248, /* 243 */ ++ 249, /* 244 */ ++ 249, /* 245 */ ++ 250, /* 246 */ ++ 250, /* 247 */ ++ 251, /* 248 */ ++ 251, /* 249 */ ++ 252, /* 250 */ ++ 252, /* 251 */ ++ 253, /* 252 */ ++ 253, /* 253 */ ++ 254, /* 254 */ ++ 255 /* 255 */ ++}; ++ + /* + In Visual C, VC6 has mmintrin.h in the "Processor Pack" add-on. + Checking if _mm_free is #defined in malloc.h is is the only way to +@@ -2679,6 +2985,7 @@ static void BlitNtoNPixelAlpha(SDL_BlitInfo *info) + int dstskip = info->d_skip; + SDL_PixelFormat *srcfmt = info->src; + SDL_PixelFormat *dstfmt = info->dst; ++ uint8_t alpha_multiply = srcfmt->alpha; + + int srcbpp; + int dstbpp; +@@ -2705,6 +3012,8 @@ static void BlitNtoNPixelAlpha(SDL_BlitInfo *info) + unsigned sA; + unsigned dA; + DISEMBLE_RGBA(src, srcbpp, srcfmt, Pixel, sR, sG, sB, sA); ++ sA=(sA*alpha_multiply)>>8; ++ sA=GFX_ALPHA_ADJUST_ARRAY[sA & 255]; + if(sA) { + DISEMBLE_RGBA(dst, dstbpp, dstfmt, Pixel, dR, dG, dB, dA); + ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB); diff --git a/SDK/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch b/SDK/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch new file mode 100644 index 0000000..46e9524 --- /dev/null +++ b/SDK/board/funkey/patches/sdl/003-alsa-fix-excessiveio.patch @@ -0,0 +1,26 @@ +Merge this bug as it can affect performance : +https://github.com/OpenDingux/SDL/commit/e51100dce8da9099278dac9f5affbecf6396378b + +--- a/src/audio/alsa/SDL_alsa_audio.c ++++ b/src/audio/alsa/SDL_alsa_audio.c +@@ -479,6 +479,10 @@ + return(-1); + } + ++ /* Switch to blocking mode for playback */ ++ /* Note: this must happen before hw/sw params are set. */ ++ SDL_NAME(snd_pcm_nonblock)(pcm_handle, 0); ++ + /* Figure out what the hardware is capable of */ + snd_pcm_hw_params_alloca(&hwparams); + status = SDL_NAME(snd_pcm_hw_params_any)(pcm_handle, hwparams); +@@ -611,9 +615,6 @@ + } + SDL_memset(mixbuf, spec->silence, spec->size); + +- /* Switch to blocking mode for playback */ +- SDL_NAME(snd_pcm_nonblock)(pcm_handle, 0); +- + /* We're ready to rock and roll. :-) */ + return(0); + } diff --git a/SDK/board/funkey/patches/sdl/sdl-fix-kb-input.patch b/SDK/board/funkey/patches/sdl/sdl-fix-kb-input.patch new file mode 100644 index 0000000..8f7db83 --- /dev/null +++ b/SDK/board/funkey/patches/sdl/sdl-fix-kb-input.patch @@ -0,0 +1,22 @@ +diff --git a/src/video/fbcon/SDL_fbevents.c b/src/video/fbcon/SDL_fbevents.c +index 5e369a4..549a7ad 100644 +--- a/src/video/fbcon/SDL_fbevents.c ++++ b/src/video/fbcon/SDL_fbevents.c +@@ -270,17 +270,6 @@ int FB_OpenKeyboard(_THIS) + fprintf(stderr, "vtpath = %s, fd = %d\n", + vtpath, keyboard_fd); + #endif /* DEBUG_KEYBOARD */ +- +- /* This needs to be our controlling tty +- so that the kernel ioctl() calls work +- */ +- if ( keyboard_fd >= 0 ) { +- tty0_fd = open("/dev/tty", O_RDWR, 0); +- if ( tty0_fd >= 0 ) { +- ioctl(tty0_fd, TIOCNOTTY, 0); +- close(tty0_fd); +- } +- } + } + } + if ( keyboard_fd < 0 ) { diff --git a/SDK/configs/funkey_defconfig b/SDK/configs/funkey_defconfig new file mode 100644 index 0000000..56e64a0 --- /dev/null +++ b/SDK/configs/funkey_defconfig @@ -0,0 +1,82 @@ +BR2_arm=y +BR2_cortex_a7=y +BR2_ARM_FPU_VFPV4=y +BR2_DL_DIR="../download" +BR2_CCACHE=y +BR2_OPTIMIZE_FAST=y +BR2_SHARED_STATIC_LIBS=y +BR2_GLOBAL_PATCH_DIR="$(BR2_EXTERNAL_SDK_PATH)/board/funkey/patches" +BR2_TOOLCHAIN_BUILDROOT_VENDOR="funkey" +BR2_TOOLCHAIN_BUILDROOT_MUSL=y +BR2_KERNEL_HEADERS_VERSION=y +BR2_DEFAULT_KERNEL_VERSION="4.14.14" +BR2_PACKAGE_HOST_LINUX_HEADERS_CUSTOM_4_14=y +BR2_BINUTILS_VERSION_2_35_X=y +BR2_GCC_VERSION_10_X=y +BR2_TOOLCHAIN_BUILDROOT_CXX=y +BR2_GCC_ENABLE_LTO=y +BR2_PACKAGE_HOST_GDB=y +BR2_PACKAGE_HOST_GDB_TUI=y +BR2_GDB_VERSION_9_2=y +BR2_TARGET_OPTIMIZATION="-fno-PIC -march=armv7-a+neon-vfpv4 -mtune=cortex-a7 -mfpu=neon-vfpv4 -mvectorize-with-neon-quad" +BR2_INIT_NONE=y +BR2_PACKAGE_BUSYBOX_SHOW_OTHERS=y +BR2_PACKAGE_ALSA_UTILS=y +BR2_PACKAGE_GSTREAMER1=y +BR2_PACKAGE_GST1_PLUGINS_BASE=y +BR2_PACKAGE_MPG123=y +BR2_PACKAGE_LZOP=y +BR2_PACKAGE_E2FSPROGS=y +BR2_PACKAGE_E2FSPROGS_RESIZE2FS=y +BR2_PACKAGE_SDL_GFX=y +BR2_PACKAGE_SDL_IMAGE=y +BR2_PACKAGE_SDL_IMAGE_GIF=y +BR2_PACKAGE_SDL_IMAGE_JPEG=y +BR2_PACKAGE_SDL_IMAGE_PNG=y +BR2_PACKAGE_SDL_MIXER=y +BR2_PACKAGE_SDL_NET=y +BR2_PACKAGE_SDL_SOUND=y +BR2_PACKAGE_SDL_TTF=y +BR2_PACKAGE_PARTED=y +BR2_PACKAGE_UBOOT_TOOLS=y +BR2_PACKAGE_UBOOT_TOOLS_MKIMAGE=y +BR2_PACKAGE_UBOOT_TOOLS_MKENVIMAGE=y +BR2_PACKAGE_LUA=y +BR2_PACKAGE_LIBSAMPLERATE=y +BR2_PACKAGE_LIBSNDFILE=y +BR2_PACKAGE_OPENAL=y +BR2_PACKAGE_TINYALSA=y +BR2_PACKAGE_TREMOR=y +BR2_PACKAGE_LIBARCHIVE=y +BR2_PACKAGE_LIBARCHIVE_BSDTAR=y +BR2_PACKAGE_LIBARCHIVE_BSDCPIO=y +BR2_PACKAGE_LIBARCHIVE_BSDCAT=y +BR2_PACKAGE_SQLITE=y +BR2_PACKAGE_LIBCONFIG=y +BR2_PACKAGE_GIFLIB=y +BR2_PACKAGE_LIBQRENCODE=y +BR2_PACKAGE_LIBQRENCODE_TOOLS=y +BR2_PACKAGE_PIXMAN=y +BR2_PACKAGE_TINYXML2=y +BR2_PACKAGE_LIBNL=y +BR2_PACKAGE_LIBRSYNC=y +BR2_PACKAGE_FMT=y +BR2_PACKAGE_ICU=y +BR2_PACKAGE_ACL=y +BR2_PACKAGE_PROCPS_NG=y +# BR2_TARGET_ROOTFS_TAR is not set +BR2_PACKAGE_HOST_DOSFSTOOLS=y +BR2_PACKAGE_HOST_DTC=y +BR2_PACKAGE_HOST_E2FSPROGS=y +BR2_PACKAGE_HOST_ENVIRONMENT_SETUP=y +BR2_PACKAGE_HOST_GENIMAGE=y +BR2_PACKAGE_HOST_KMOD=y +BR2_PACKAGE_HOST_MKPASSWD=y +BR2_PACKAGE_HOST_MTOOLS=y +BR2_PACKAGE_HOST_SQUASHFS=y +BR2_PACKAGE_HOST_UBOOT_TOOLS=y +BR2_PACKAGE_LIBOPK=y +BR2_PACKAGE_LIBXDGMIME=y +BR2_PACKAGE_AGG=y +BR2_PACKAGE_FLUIDLITE=y +BR2_PACKAGE_LIBMIKMOD=y diff --git a/SDK/external.desc b/SDK/external.desc new file mode 100644 index 0000000..1623a12 --- /dev/null +++ b/SDK/external.desc @@ -0,0 +1,2 @@ +name: SDK +desc: FunKey SDK diff --git a/SDK/external.mk b/SDK/external.mk new file mode 100644 index 0000000..7433fb4 --- /dev/null +++ b/SDK/external.mk @@ -0,0 +1 @@ +include $(sort $(wildcard $(BR2_EXTERNAL_SDK_PATH)/package/*/*.mk)) diff --git a/SDK/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch b/SDK/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch new file mode 100644 index 0000000..eaf0467 --- /dev/null +++ b/SDK/package/agg/0001-Fix-non-terminating-loop-conditions-when-len-1.patch @@ -0,0 +1,81 @@ +From efd33aad5e69f36ab343b1f28839a55db4538104 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 10:55:37 +0100 +Subject: [PATCH 01/15] Fix non-terminating loop conditions when len=1 + +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; + } +--- + include/agg_renderer_outline_aa.h | 8 ++++---- + include/agg_renderer_outline_image.h | 4 ++-- + 2 files changed, 6 insertions(+), 6 deletions(-) + +diff --git a/include/agg_renderer_outline_aa.h b/include/agg_renderer_outline_aa.h +index ce25a2e..cb2aa00 100644 +--- a/include/agg_renderer_outline_aa.h ++++ b/include/agg_renderer_outline_aa.h +@@ -1659,7 +1659,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -1726,7 +1726,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +@@ -1798,7 +1798,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -1811,7 +1811,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +diff --git a/include/agg_renderer_outline_image.h b/include/agg_renderer_outline_image.h +index fbfac10..66d2b9a 100644 +--- a/include/agg_renderer_outline_image.h ++++ b/include/agg_renderer_outline_image.h +@@ -969,7 +969,7 @@ namespace agg + } + else + { +- while(abs(sx - lp.x1) + abs(sy - lp.y1) > lp2.len) ++ while(abs(sx - lp.x1) + abs(sy - lp.y1) > 1 + lp2.len) + { + sx = (lp.x1 + sx) >> 1; + sy = (lp.y1 + sy) >> 1; +@@ -982,7 +982,7 @@ namespace agg + } + else + { +- while(abs(ex - lp.x2) + abs(ey - lp.y2) > lp2.len) ++ while(abs(ex - lp.x2) + abs(ey - lp.y2) > 1 + lp2.len) + { + ex = (lp.x2 + ex) >> 1; + ey = (lp.y2 + ey) >> 1; +-- +1.8.1.4 + diff --git a/SDK/package/agg/0001-autogen.patch b/SDK/package/agg/0001-autogen.patch new file mode 100644 index 0000000..b773f12 --- /dev/null +++ b/SDK/package/agg/0001-autogen.patch @@ -0,0 +1,15 @@ +Author: Andrea Veri +Description: Disable configure's run from the autogen file. + +Index: agg-2.5+dfsg1/autogen.sh +=================================================================== +--- a/autogen.sh 2007-10-11 00:06:16.000000000 +0200 ++++ b/autogen.sh 2012-05-01 16:57:37.916862783 +0200 +@@ -18,6 +18,6 @@ + automake --foreign --add-missing --ignore-deps + + # and finally invoke our new configure +-./configure $* ++[ -n "$NOCONFIGURE" ] || ./configure $* + + # end diff --git a/SDK/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch b/SDK/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch new file mode 100644 index 0000000..4fe7434 --- /dev/null +++ b/SDK/package/agg/0002-Cure-recursion-by-aborting-if-the-co-ordinates-are-t.patch @@ -0,0 +1,40 @@ +From e269fe9b62af6fe314cebe0ee7a6d6d1a4a84d1c Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 11:03:26 +0100 +Subject: [PATCH 02/15] Cure recursion by aborting if the co-ordinates are to + big to handle + +--- + include/agg_rasterizer_cells_aa.h | 9 ++++++++- + 1 file changed, 8 insertions(+), 1 deletion(-) + +diff --git a/include/agg_rasterizer_cells_aa.h b/include/agg_rasterizer_cells_aa.h +index d3bb138..3a616d9 100644 +--- a/include/agg_rasterizer_cells_aa.h ++++ b/include/agg_rasterizer_cells_aa.h +@@ -40,7 +40,8 @@ + #define AGG_RASTERIZER_CELLS_AA_INCLUDED + + #include +-#include ++#include ++#include + #include "agg_math.h" + #include "agg_array.h" + +@@ -333,6 +334,12 @@ namespace agg + { + int cx = (x1 + x2) >> 1; + int cy = (y1 + y2) >> 1; ++ ++ // Bail if values are so large they are likely to wrap ++ if ((std::abs(x1) >= std::numeric_limits::max()/2) || (std::abs(y1) >= std::numeric_limits::max()/2) || ++ (std::abs(x2) >= std::numeric_limits::max()/2) || (std::abs(y2) >= std::numeric_limits::max()/2)) ++ return; ++ + line(x1, y1, cx, cy); + line(cx, cy, x2, y2); + } +-- +1.8.1.4 + diff --git a/SDK/package/agg/0002-caca.patch b/SDK/package/agg/0002-caca.patch new file mode 100644 index 0000000..f98a573 --- /dev/null +++ b/SDK/package/agg/0002-caca.patch @@ -0,0 +1,34 @@ +--- a/configure.in ++++ b/configure.in +@@ -39,7 +39,7 @@ + # used as platform library in examples: + # todo, make the PREFERED_PLATFORM selectable, after the set of possible + # Platforms to link the examples have been evaluated. +-PREFERED_PLATFORM=X11 ++PREFERED_PLATFORM=sdl + case "$host" in + *darwin* ) + OSX_LIBS="-framework Carbon -framework QuickTime" +@@ -120,9 +120,7 @@ + if test "$no_x" = "yes"; then + AC_MSG_WARN([*** X11 not found! Omitting X11 layer.]) + fi +-AM_CONDITIONAL(ENABLE_X11,[test x$no_x = x -a xno != x$enable_platform -a x$win32_host != xyes]) +-AC_SUBST(x_includes) +-AC_SUBST(x_libraries) ++AM_CONDITIONAL(ENABLE_X11,0) + dnl ############################################### + + dnl Settung up library version + +--- a/include/agg_renderer_outline_aa.h ++++ b/include/agg_renderer_outline_aa.h +@@ -1375,7 +1375,7 @@ + //--------------------------------------------------------------------- + void profile(const line_profile_aa& prof) { m_profile = &prof; } + const line_profile_aa& profile() const { return *m_profile; } +- line_profile_aa& profile() { return *m_profile; } ++// line_profile_aa& profile() { return *m_profile; } + + //--------------------------------------------------------------------- + int subpixel_width() const { return m_profile->subpixel_width(); } diff --git a/SDK/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch b/SDK/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch new file mode 100644 index 0000000..b12684d --- /dev/null +++ b/SDK/package/agg/0003-Get-coordinates-from-previous-vertex-if-last-command.patch @@ -0,0 +1,30 @@ +From 032d5342430f4c5dfbc34a2817d67386a14fd51b Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 11:40:49 +0100 +Subject: [PATCH 03/15] Get coordinates from previous vertex if last command is + path_cmd_end_poly + +--- + include/agg_path_storage.h | 6 ++++++ + 1 file changed, 6 insertions(+) + +diff --git a/include/agg_path_storage.h b/include/agg_path_storage.h +index 7be7393..8922fc8 100644 +--- a/include/agg_path_storage.h ++++ b/include/agg_path_storage.h +@@ -878,6 +878,12 @@ namespace agg + *x += x2; + *y += y2; + } ++ else if (!is_stop(m_vertices.last_command()) && ++ is_vertex(m_vertices.prev_vertex(&x2, &y2))) ++ { ++ *x += x2; ++ *y += y2; ++ } + } + } + +-- +1.8.1.4 + diff --git a/SDK/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch b/SDK/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch new file mode 100644 index 0000000..0cecaf7 --- /dev/null +++ b/SDK/package/agg/0004-Make-rasterizer_outline_aa-ignore-close_polygon-when.patch @@ -0,0 +1,138 @@ +From b9c4b1c72b4ad6b24c37f402d3eec39ef393b0eb Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 14:17:43 +0100 +Subject: [PATCH 04/15] Make rasterizer_outline_aa ignore close_polygon when + vertex count < 3 + +--- + include/agg_rasterizer_outline_aa.h | 107 ++++++++++++++++++------------------ + 1 file changed, 52 insertions(+), 55 deletions(-) + +diff --git a/include/agg_rasterizer_outline_aa.h b/include/agg_rasterizer_outline_aa.h +index 4d6dd57..24301d5 100644 +--- a/include/agg_rasterizer_outline_aa.h ++++ b/include/agg_rasterizer_outline_aa.h +@@ -333,68 +333,65 @@ namespace agg + int y2; + int lprev; + +- if(close_polygon) ++ if(close_polygon && (m_src_vertices.size() >= 3)) + { +- if(m_src_vertices.size() >= 3) ++ dv.idx = 2; ++ ++ v = &m_src_vertices[m_src_vertices.size() - 1]; ++ x1 = v->x; ++ y1 = v->y; ++ lprev = v->len; ++ ++ v = &m_src_vertices[0]; ++ x2 = v->x; ++ y2 = v->y; ++ dv.lcurr = v->len; ++ line_parameters prev(x1, y1, x2, y2, lprev); ++ ++ v = &m_src_vertices[1]; ++ dv.x1 = v->x; ++ dv.y1 = v->y; ++ dv.lnext = v->len; ++ dv.curr = line_parameters(x2, y2, dv.x1, dv.y1, dv.lcurr); ++ ++ v = &m_src_vertices[dv.idx]; ++ dv.x2 = v->x; ++ dv.y2 = v->y; ++ dv.next = line_parameters(dv.x1, dv.y1, dv.x2, dv.y2, dv.lnext); ++ ++ dv.xb1 = 0; ++ dv.yb1 = 0; ++ dv.xb2 = 0; ++ dv.yb2 = 0; ++ ++ switch(m_line_join) + { +- dv.idx = 2; +- +- v = &m_src_vertices[m_src_vertices.size() - 1]; +- x1 = v->x; +- y1 = v->y; +- lprev = v->len; +- +- v = &m_src_vertices[0]; +- x2 = v->x; +- y2 = v->y; +- dv.lcurr = v->len; +- line_parameters prev(x1, y1, x2, y2, lprev); +- +- v = &m_src_vertices[1]; +- dv.x1 = v->x; +- dv.y1 = v->y; +- dv.lnext = v->len; +- dv.curr = line_parameters(x2, y2, dv.x1, dv.y1, dv.lcurr); +- +- v = &m_src_vertices[dv.idx]; +- dv.x2 = v->x; +- dv.y2 = v->y; +- dv.next = line_parameters(dv.x1, dv.y1, dv.x2, dv.y2, dv.lnext); +- +- dv.xb1 = 0; +- dv.yb1 = 0; +- dv.xb2 = 0; +- dv.yb2 = 0; +- +- switch(m_line_join) +- { +- case outline_no_join: +- dv.flags = 3; +- break; ++ case outline_no_join: ++ dv.flags = 3; ++ break; + +- case outline_miter_join: +- case outline_round_join: +- dv.flags = +- (prev.diagonal_quadrant() == dv.curr.diagonal_quadrant()) | +- ((dv.curr.diagonal_quadrant() == dv.next.diagonal_quadrant()) << 1); +- break; ++ case outline_miter_join: ++ case outline_round_join: ++ dv.flags = ++ (prev.diagonal_quadrant() == dv.curr.diagonal_quadrant()) | ++ ((dv.curr.diagonal_quadrant() == dv.next.diagonal_quadrant()) << 1); ++ break; + +- case outline_miter_accurate_join: +- dv.flags = 0; +- break; +- } ++ case outline_miter_accurate_join: ++ dv.flags = 0; ++ break; ++ } + +- if((dv.flags & 1) == 0 && m_line_join != outline_round_join) +- { +- bisectrix(prev, dv.curr, &dv.xb1, &dv.yb1); +- } ++ if((dv.flags & 1) == 0 && m_line_join != outline_round_join) ++ { ++ bisectrix(prev, dv.curr, &dv.xb1, &dv.yb1); ++ } + +- if((dv.flags & 2) == 0 && m_line_join != outline_round_join) +- { +- bisectrix(dv.curr, dv.next, &dv.xb2, &dv.yb2); +- } +- draw(dv, 0, m_src_vertices.size()); ++ if((dv.flags & 2) == 0 && m_line_join != outline_round_join) ++ { ++ bisectrix(dv.curr, dv.next, &dv.xb2, &dv.yb2); + } ++ draw(dv, 0, m_src_vertices.size()); + } + else + { +-- +1.8.1.4 + diff --git a/SDK/package/agg/0005-Remove-VC-6-workaround.patch b/SDK/package/agg/0005-Remove-VC-6-workaround.patch new file mode 100644 index 0000000..f38f7c4 --- /dev/null +++ b/SDK/package/agg/0005-Remove-VC-6-workaround.patch @@ -0,0 +1,52 @@ +From b8c43fb0ba13af0cc2b1050f48f81d76d2fdf0c7 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:04:05 +0100 +Subject: [PATCH 05/15] Remove VC++ 6 workaround + +--- + include/agg_renderer_scanline.h | 29 +---------------------------- + 1 file changed, 1 insertion(+), 28 deletions(-) + +diff --git a/include/agg_renderer_scanline.h b/include/agg_renderer_scanline.h +index c3bb6f0..c27ca60 100644 +--- a/include/agg_renderer_scanline.h ++++ b/include/agg_renderer_scanline.h +@@ -79,34 +79,7 @@ namespace agg + sl.reset(ras.min_x(), ras.max_x()); + while(ras.sweep_scanline(sl)) + { +- //render_scanline_aa_solid(sl, ren, ren_color); +- +- // This code is equivalent to the above call (copy/paste). +- // It's just a "manual" optimization for old compilers, +- // like Microsoft Visual C++ v6.0 +- //------------------------------- +- int y = sl.y(); +- unsigned num_spans = sl.num_spans(); +- typename Scanline::const_iterator span = sl.begin(); +- +- for(;;) +- { +- int x = span->x; +- if(span->len > 0) +- { +- ren.blend_solid_hspan(x, y, (unsigned)span->len, +- ren_color, +- span->covers); +- } +- else +- { +- ren.blend_hline(x, y, (unsigned)(x - span->len - 1), +- ren_color, +- *(span->covers)); +- } +- if(--num_spans == 0) break; +- ++span; +- } ++ render_scanline_aa_solid(sl, ren, ren_color); + } + } + } +-- +1.8.1.4 + diff --git a/SDK/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch b/SDK/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch new file mode 100644 index 0000000..f1e465b --- /dev/null +++ b/SDK/package/agg/0006-Implement-grain-merge-blending-mode-GIMP.patch @@ -0,0 +1,85 @@ +From 9422570f4e099a834fc43619f7b2a7eb6b442e25 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:31:01 +0100 +Subject: [PATCH 06/15] Implement grain-merge blending mode (GIMP) + +--- + include/agg_pixfmt_rgba.h | 42 ++++++++++++++++++++++++++++++++++++++++-- + 1 file changed, 40 insertions(+), 2 deletions(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 79d10dc..f576ce4 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1401,9 +1401,46 @@ namespace agg + } + }; + ++ //================================================comp_op_rgba_grain_merge ++ template struct comp_op_rgba_grain_merge ++ { ++ typedef ColorT color_type; ++ typedef Order order_type; ++ typedef typename color_type::value_type value_type; ++ typedef typename color_type::calc_type calc_type; ++ typedef typename color_type::long_type long_type; ++ enum base_scale_e ++ { ++ base_shift = color_type::base_shift, ++ base_mask = color_type::base_mask ++ }; + ++ // E = I + M - 128 ++ static AGG_INLINE void blend_pix(value_type* p, ++ unsigned sr, unsigned sg, unsigned sb, ++ unsigned sa, unsigned cover) ++ { + +- ++ if(cover < 255) ++ { ++ sr = (sr * cover + 255) >> 8; ++ sg = (sg * cover + 255) >> 8; ++ sb = (sb * cover + 255) >> 8; ++ sa = (sa * cover + 255) >> 8; ++ } ++ if(sa) ++ { ++ calc_type da = p[Order::A]; ++ int dr = sr + p[Order::R] - 128; ++ int dg = sg + p[Order::G] - 128; ++ int db = sb + p[Order::B] - 128; ++ p[Order::R] = (value_type)(dr < 0 ? 0 : (dr > 255 ? 255 : dr)); ++ p[Order::G] = (value_type)(dg < 0 ? 0 : (dg > 255 ? 255 : dg)); ++ p[Order::B] = (value_type)(db < 0 ? 0 : (db > 255 ? 255 : db)); ++ p[Order::A] = (value_type)(sa + da - ((sa * da + base_mask) >> base_shift)); ++ } ++ } ++ }; + + //======================================================comp_op_table_rgba + template struct comp_op_table_rgba +@@ -1451,6 +1488,7 @@ namespace agg + comp_op_rgba_contrast ::blend_pix, + comp_op_rgba_invert ::blend_pix, + comp_op_rgba_invert_rgb ::blend_pix, ++ comp_op_rgba_grain_merge::blend_pix, + 0 + }; + +@@ -1486,6 +1524,7 @@ namespace agg + comp_op_contrast, //----comp_op_contrast + comp_op_invert, //----comp_op_invert + comp_op_invert_rgb, //----comp_op_invert_rgb ++ comp_op_grain_merge, //----comp_op_grain_merge + + end_of_comp_op_e + }; +@@ -2908,4 +2947,3 @@ namespace agg + } + + #endif +- +-- +1.8.1.4 + diff --git a/SDK/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch b/SDK/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch new file mode 100644 index 0000000..cafb36e --- /dev/null +++ b/SDK/package/agg/0007-Implement-grain-extract-blending-mode-GIMP.patch @@ -0,0 +1,85 @@ +From abd440342e166a90d08610bf5b31d2a8357eafbe Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 15:43:18 +0100 +Subject: [PATCH 07/15] Implement grain-extract blending mode (GIMP) + +--- + include/agg_pixfmt_rgba.h | 48 +++++++++++++++++++++++++++++++++++++++++++++++ + 1 file changed, 48 insertions(+) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index f576ce4..42f0a05 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1442,6 +1442,52 @@ namespace agg + } + }; + ++ //==============================================comp_op_rgba_grain_extract ++ template struct comp_op_rgba_grain_extract ++ { ++ typedef ColorT color_type; ++ typedef Order order_type; ++ typedef typename color_type::value_type value_type; ++ typedef typename color_type::calc_type calc_type; ++ typedef typename color_type::long_type long_type; ++ enum base_scale_e ++ { ++ base_shift = color_type::base_shift, ++ base_mask = color_type::base_mask ++ }; ++ ++ // E = I - M + 128 ++ static AGG_INLINE void blend_pix(value_type* p, ++ unsigned sr, unsigned sg, unsigned sb, ++ unsigned sa, unsigned cover) ++ { ++ calc_type da = (p[Order::A] * sa + 255) >> 8; ++ ++ int dr = p[Order::R] - sr + 128; ++ int dg = p[Order::G] - sg + 128; ++ int db = p[Order::B] - sb + 128; ++ ++ dr = dr < 0 ? 0 : (dr > 255 ? 255 : dr); ++ dg = dg < 0 ? 0 : (dg > 255 ? 255 : dg); ++ db = db < 0 ? 0 : (db > 255 ? 255 : db); ++ ++ p[Order::A] = da; ++ ++ if(da < 255) ++ { ++ p[Order::R] = (dr * da + 255) >> 8; ++ p[Order::G] = (dg * da + 255) >> 8; ++ p[Order::B] = (db * da + 255) >> 8; ++ } ++ else ++ { ++ p[Order::R] = dr; ++ p[Order::G] = dg; ++ p[Order::B] = db; ++ } ++ } ++ }; ++ + //======================================================comp_op_table_rgba + template struct comp_op_table_rgba + { +@@ -1489,6 +1535,7 @@ namespace agg + comp_op_rgba_invert ::blend_pix, + comp_op_rgba_invert_rgb ::blend_pix, + comp_op_rgba_grain_merge::blend_pix, ++ comp_op_rgba_grain_extract::blend_pix, + 0 + }; + +@@ -1525,6 +1572,7 @@ namespace agg + comp_op_invert, //----comp_op_invert + comp_op_invert_rgb, //----comp_op_invert_rgb + comp_op_grain_merge, //----comp_op_grain_merge ++ comp_op_grain_extract, //----comp_op_grain_extract + + end_of_comp_op_e + }; +-- +1.8.1.4 + diff --git a/SDK/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch b/SDK/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch new file mode 100644 index 0000000..0ed92ee --- /dev/null +++ b/SDK/package/agg/0008-Declare-multiplication-and-division-operators-as-con.patch @@ -0,0 +1,36 @@ +From 2688af280836b95908d3cfd6915510d55de673b8 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:15:01 +0100 +Subject: [PATCH 08/15] Declare multiplication and division operators as const + +--- + include/agg_trans_affine.h | 8 ++++---- + 1 file changed, 4 insertions(+), 4 deletions(-) + +diff --git a/include/agg_trans_affine.h b/include/agg_trans_affine.h +index a662099..2f602a0 100644 +--- a/include/agg_trans_affine.h ++++ b/include/agg_trans_affine.h +@@ -216,15 +216,15 @@ namespace agg + } + + // Multiply the matrix by another one and return +- // the result in a separete matrix. +- trans_affine operator * (const trans_affine& m) ++ // the result in a separate matrix. ++ trans_affine operator * (const trans_affine& m) const + { + return trans_affine(*this).multiply(m); + } + + // Multiply the matrix by inverse of another one +- // and return the result in a separete matrix. +- trans_affine operator / (const trans_affine& m) ++ // and return the result in a separate matrix. ++ trans_affine operator / (const trans_affine& m) const + { + return trans_affine(*this).multiply_inv(m); + } +-- +1.8.1.4 + diff --git a/SDK/package/agg/0009-Add-a-static-identity-transformation.patch b/SDK/package/agg/0009-Add-a-static-identity-transformation.patch new file mode 100644 index 0000000..01555cb --- /dev/null +++ b/SDK/package/agg/0009-Add-a-static-identity-transformation.patch @@ -0,0 +1,37 @@ +From be9ed90897bc43b4547a3a1f8046827caaf13b4c Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:15:36 +0100 +Subject: [PATCH 09/15] Add a static identity transformation + +--- + include/agg_trans_affine.h | 1 + + src/agg_trans_affine.cpp | 1 + + 2 files changed, 2 insertions(+) + +diff --git a/include/agg_trans_affine.h b/include/agg_trans_affine.h +index 2f602a0..67fe5ca 100644 +--- a/include/agg_trans_affine.h ++++ b/include/agg_trans_affine.h +@@ -92,6 +92,7 @@ namespace agg + //---------------------------------------------------------------------- + struct trans_affine + { ++ static const trans_affine identity; + double sx, shy, shx, sy, tx, ty; + + //------------------------------------------ Construction +diff --git a/src/agg_trans_affine.cpp b/src/agg_trans_affine.cpp +index aca18c2..b3d9bc0 100644 +--- a/src/agg_trans_affine.cpp ++++ b/src/agg_trans_affine.cpp +@@ -28,6 +28,7 @@ + + namespace agg + { ++ const trans_affine trans_affine::identity; + + //------------------------------------------------------------------------ + const trans_affine& trans_affine::parl_to_parl(const double* src, +-- +1.8.1.4 + diff --git a/SDK/package/agg/0010-Add-renderer_scanline_aa_alpha.patch b/SDK/package/agg/0010-Add-renderer_scanline_aa_alpha.patch new file mode 100644 index 0000000..b0be258 --- /dev/null +++ b/SDK/package/agg/0010-Add-renderer_scanline_aa_alpha.patch @@ -0,0 +1,193 @@ +From 749c8cd11e9e6f81e93ae5ce19258431722b6bdf Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:43:25 +0100 +Subject: [PATCH 10/15] Add renderer_scanline_aa_alpha + +--- + include/agg_pixfmt_rgba.h | 24 +++++++++++++- + include/agg_renderer_base.h | 28 ++++++++++++++++ + include/agg_renderer_scanline.h | 71 +++++++++++++++++++++++++++++++++++++++++ + 3 files changed, 122 insertions(+), 1 deletion(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 42f0a05..6c4bc37 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -2247,7 +2247,6 @@ namespace agg + } + + +- + //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, + unsigned len, +@@ -2751,6 +2750,29 @@ namespace agg + } + + //-------------------------------------------------------------------- ++ void blend_color_hspan_alpha(int x, int y, unsigned len, ++ const color_type* colors, ++ value_type alpha, ++ const int8u* covers, ++ int8u cover) ++ { ++ value_type* p = (value_type*)m_rbuf->row_ptr(x, y, len) + (x << 2); ++ do ++ { ++ blender_type::blend_pix(m_comp_op, ++ p, ++ (colors->r * alpha + 255) >> 8, ++ (colors->g * alpha + 255) >> 8, ++ (colors->b * alpha + 255) >> 8, ++ (colors->a * alpha + 255) >> 8, ++ covers ? *covers++ : cover); ++ p += 4; ++ ++colors; ++ } ++ while(--len); ++ } ++ ++ //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, unsigned len, + const color_type* colors, + const int8u* covers, +diff --git a/include/agg_renderer_base.h b/include/agg_renderer_base.h +index 1808944..25f07c3 100644 +--- a/include/agg_renderer_base.h ++++ b/include/agg_renderer_base.h +@@ -37,6 +37,7 @@ namespace agg + public: + typedef PixelFormat pixfmt_type; + typedef typename pixfmt_type::color_type color_type; ++ typedef typename pixfmt_type::color_type::value_type value_type; + typedef typename pixfmt_type::row_data row_data; + + //-------------------------------------------------------------------- +@@ -383,6 +384,33 @@ namespace agg + } + + //-------------------------------------------------------------------- ++ void blend_color_hspan_alpha(int x, int y, int len, ++ const color_type* colors, ++ value_type alpha, ++ const cover_type* covers, ++ cover_type cover = agg::cover_full) ++ { ++ if(y > ymax()) return; ++ if(y < ymin()) return; ++ ++ if(x < xmin()) ++ { ++ int d = xmin() - x; ++ len -= d; ++ if(len <= 0) return; ++ if(covers) covers += d; ++ colors += d; ++ x = xmin(); ++ } ++ if(x + len > xmax()) ++ { ++ len = xmax() - x + 1; ++ if(len <= 0) return; ++ } ++ m_ren->blend_color_hspan_alpha(x, y, len, colors, alpha, covers, cover); ++ } ++ ++ //-------------------------------------------------------------------- + void blend_color_vspan(int x, int y, int len, + const color_type* colors, + const cover_type* covers, +diff --git a/include/agg_renderer_scanline.h b/include/agg_renderer_scanline.h +index c27ca60..4fcb557 100644 +--- a/include/agg_renderer_scanline.h ++++ b/include/agg_renderer_scanline.h +@@ -156,6 +156,35 @@ namespace agg + } + } + ++ //================================================render_scanline_aa_alpha ++ template ++ void render_scanline_aa_alpha(const Scanline& sl, BaseRenderer& ren, ++ SpanAllocator& alloc, SpanGenerator& span_gen, ++ unsigned alpha) ++ { ++ int y = sl.y(); ++ ++ unsigned num_spans = sl.num_spans(); ++ typename Scanline::const_iterator span = sl.begin(); ++ for(;;) ++ { ++ int x = span->x; ++ int len = span->len; ++ const typename Scanline::cover_type* covers = span->covers; ++ ++ if(len < 0) len = -len; ++ typename BaseRenderer::color_type* colors = alloc.allocate(len); ++ span_gen.generate(colors, x, y, len); ++ ren.blend_color_hspan_alpha(x, y, len, colors, alpha, ++ (span->len < 0) ? 0 : covers, *covers); ++ ++ if(--num_spans == 0) break; ++ ++span; ++ } ++ } ++ ++ + //=====================================================render_scanlines_aa + template +@@ -216,8 +245,50 @@ namespace agg + }; + + ++ //==============================================renderer_scanline_aa_alpha ++ template ++ class renderer_scanline_aa_alpha ++ { ++ public: ++ typedef BaseRenderer base_ren_type; ++ typedef SpanAllocator alloc_type; ++ typedef SpanGenerator span_gen_type; + ++ //-------------------------------------------------------------------- ++ renderer_scanline_aa_alpha() : m_ren(0), m_alloc(0), m_span_gen(0), m_alpha(1.0) {} ++ renderer_scanline_aa_alpha(base_ren_type& ren, ++ alloc_type& alloc, ++ span_gen_type& span_gen, ++ unsigned alpha) : ++ m_ren(&ren), ++ m_alloc(&alloc), ++ m_span_gen(&span_gen), ++ m_alpha(alpha) ++ {} ++ void attach(base_ren_type& ren, ++ alloc_type& alloc, ++ span_gen_type& span_gen) ++ { ++ m_ren = &ren; ++ m_alloc = &alloc; ++ m_span_gen = &span_gen; ++ } + ++ //-------------------------------------------------------------------- ++ void prepare() { m_span_gen->prepare(); } ++ ++ //-------------------------------------------------------------------- ++ template void render(const Scanline& sl) ++ { ++ render_scanline_aa_alpha(sl, *m_ren, *m_alloc, *m_span_gen, m_alpha); ++ } ++ ++ private: ++ base_ren_type* m_ren; ++ alloc_type* m_alloc; ++ span_gen_type* m_span_gen; ++ unsigned m_alpha; ++ }; + + + //===============================================render_scanline_bin_solid +-- +1.8.1.4 + diff --git a/SDK/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch b/SDK/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch new file mode 100644 index 0000000..2a0d198 --- /dev/null +++ b/SDK/package/agg/0011-Avoid-division-by-zero-in-color-burn-mode.patch @@ -0,0 +1,58 @@ +From 0ec68d7f5695403eccac75025ba7f6f7ecf1814e Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sun, 19 May 2013 16:49:08 +0100 +Subject: [PATCH 11/15] Avoid division by zero in color-burn mode + +FIXME: re-work using latest math from http://www.w3.org/TR/SVGCompositing/ +--- + include/agg_pixfmt_rgba.h | 21 ++++++++++++++++++--- + 1 file changed, 18 insertions(+), 3 deletions(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 6c4bc37..5d6b511 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -1027,6 +1027,21 @@ namespace agg + // Dca' = Sa.(Sca.Da + Dca.Sa - Sa.Da)/Sca + Sca.(1 - Da) + Dca.(1 - Sa) + // + // Da' = Sa + Da - Sa.Da ++ ++ ++ // http://www.w3.org/TR/SVGCompositing/ ++ // if Sca == 0 and Dca == Da ++ // Dca' = Sa × Da + Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Sa × Da + Dca × (1 - Sa) ++ // = Da = Dca ++ // otherwise if Sca == 0 ++ // Dca' = Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Dca × (1 - Sa) ++ // otherwise if Sca > 0 ++ // Dca' = Sa × Da - Sa × Da × min(1, (1 - Dca/Da) × Sa/Sca) + Sca × (1 - Da) + Dca × (1 - Sa) ++ // = Sa × Da × (1 - min(1, (1 - Dca/Da) × Sa/Sca)) + Sca × (1 - Da) + Dca × (1 - Sa) ++ ++ // sa * da * (255 - std::min(255, (255 - p[0]/da)*(sa/(sc*sa)) + + static AGG_INLINE void blend_pix(value_type* p, + unsigned sr, unsigned sg, unsigned sb, + unsigned sa, unsigned cover) +@@ -1056,15 +1071,15 @@ namespace agg + + p[Order::R] = (value_type)(((srda + drsa <= sada) ? + sr * d1a + dr * s1a : +- sa * (srda + drsa - sada) / sr + sr * d1a + dr * s1a + base_mask) >> base_shift); ++ (sr > 0 ? sa * (srda + drsa - sada) / sr + sr * d1a + dr * s1a + base_mask : 0)) >> base_shift); + + p[Order::G] = (value_type)(((sgda + dgsa <= sada) ? + sg * d1a + dg * s1a : +- sa * (sgda + dgsa - sada) / sg + sg * d1a + dg * s1a + base_mask) >> base_shift); ++ (sg > 0 ? sa * (sgda + dgsa - sada) / sg + sg * d1a + dg * s1a + base_mask : 0)) >> base_shift); + + p[Order::B] = (value_type)(((sbda + dbsa <= sada) ? + sb * d1a + db * s1a : +- sa * (sbda + dbsa - sada) / sb + sb * d1a + db * s1a + base_mask) >> base_shift); ++ (sb > 0 ? sa * (sbda + dbsa - sada) / sb + sb * d1a + db * s1a + base_mask : 0)) >> base_shift); + + p[Order::A] = (value_type)(sa + da - ((sa * da + base_mask) >> base_shift)); + } +-- +1.8.1.4 + diff --git a/SDK/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch b/SDK/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch new file mode 100644 index 0000000..b3e641e --- /dev/null +++ b/SDK/package/agg/0012-Avoid-pixel-artifacts-when-compositing.patch @@ -0,0 +1,26 @@ +From bf0e0b71360cfbc690a29f4abe15d7b9b61b8479 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:11:54 +0100 +Subject: [PATCH 12/15] Avoid pixel artifacts when compositing + +Change src_over alpha to avoid pixel artifacts by reordering computations. +--- + include/agg_pixfmt_rgba.h | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +diff --git a/include/agg_pixfmt_rgba.h b/include/agg_pixfmt_rgba.h +index 5d6b511..bb255cd 100644 +--- a/include/agg_pixfmt_rgba.h ++++ b/include/agg_pixfmt_rgba.h +@@ -346,7 +346,7 @@ namespace agg + p[Order::R] = (value_type)(sr + ((p[Order::R] * s1a + base_mask) >> base_shift)); + p[Order::G] = (value_type)(sg + ((p[Order::G] * s1a + base_mask) >> base_shift)); + p[Order::B] = (value_type)(sb + ((p[Order::B] * s1a + base_mask) >> base_shift)); +- p[Order::A] = (value_type)(sa + p[Order::A] - ((sa * p[Order::A] + base_mask) >> base_shift)); ++ p[Order::A] = (value_type)(sa + ((p[Order::A] * s1a + base_mask) >> base_shift)); + } + }; + +-- +1.8.1.4 + diff --git a/SDK/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch b/SDK/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch new file mode 100644 index 0000000..9deb904 --- /dev/null +++ b/SDK/package/agg/0013-Modify-agg-conv-classes-to-allow-access-to-the-origi.patch @@ -0,0 +1,93 @@ +From 6f1ab5f4b470bcf4e7e72aac6e2f7f6ee3e7b424 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:16:42 +0100 +Subject: [PATCH 13/15] Modify agg conv classes to allow access to the original + geometry type + +--- + include/agg_conv_adaptor_vcgen.h | 2 ++ + include/agg_conv_adaptor_vpgen.h | 1 + + include/agg_conv_clip_polygon.h | 1 + + include/agg_conv_clip_polyline.h | 1 + + include/agg_conv_smooth_poly1.h | 2 ++ + 5 files changed, 7 insertions(+) + +diff --git a/include/agg_conv_adaptor_vcgen.h b/include/agg_conv_adaptor_vcgen.h +index 7bd9b07..fef4579 100644 +--- a/include/agg_conv_adaptor_vcgen.h ++++ b/include/agg_conv_adaptor_vcgen.h +@@ -38,6 +38,7 @@ namespace agg + + void rewind(unsigned) {} + unsigned vertex(double*, double*) { return path_cmd_stop; } ++ unsigned type() const { return 0; } + }; + + +@@ -73,6 +74,7 @@ namespace agg + } + + unsigned vertex(double* x, double* y); ++ unsigned type() const { return m_source->type(); } + + private: + // Prohibit copying +diff --git a/include/agg_conv_adaptor_vpgen.h b/include/agg_conv_adaptor_vpgen.h +index dca9415..a39102d 100644 +--- a/include/agg_conv_adaptor_vpgen.h ++++ b/include/agg_conv_adaptor_vpgen.h +@@ -42,6 +42,7 @@ namespace agg + + void rewind(unsigned path_id); + unsigned vertex(double* x, double* y); ++ unsigned type() const { return m_source->type(); } + + private: + conv_adaptor_vpgen(const conv_adaptor_vpgen&); +diff --git a/include/agg_conv_clip_polygon.h b/include/agg_conv_clip_polygon.h +index 3c34590..e417a7d 100644 +--- a/include/agg_conv_clip_polygon.h ++++ b/include/agg_conv_clip_polygon.h +@@ -60,6 +60,7 @@ namespace agg + double y1() const { return base_type::vpgen().y1(); } + double x2() const { return base_type::vpgen().x2(); } + double y2() const { return base_type::vpgen().y2(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_clip_polygon(const conv_clip_polygon&); +diff --git a/include/agg_conv_clip_polyline.h b/include/agg_conv_clip_polyline.h +index d45067f..0de4b57 100644 +--- a/include/agg_conv_clip_polyline.h ++++ b/include/agg_conv_clip_polyline.h +@@ -60,6 +60,7 @@ namespace agg + double y1() const { return base_type::vpgen().y1(); } + double x2() const { return base_type::vpgen().x2(); } + double y2() const { return base_type::vpgen().y2(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_clip_polyline(const conv_clip_polyline&); +diff --git a/include/agg_conv_smooth_poly1.h b/include/agg_conv_smooth_poly1.h +index 15f7f8d..0956c4e 100644 +--- a/include/agg_conv_smooth_poly1.h ++++ b/include/agg_conv_smooth_poly1.h +@@ -48,6 +48,7 @@ namespace agg + + void smooth_value(double v) { base_type::generator().smooth_value(v); } + double smooth_value() const { return base_type::generator().smooth_value(); } ++ unsigned type() const { return base_type::type(); } + + private: + conv_smooth_poly1(const conv_smooth_poly1&); +@@ -70,6 +71,7 @@ namespace agg + + void smooth_value(double v) { m_smooth.generator().smooth_value(v); } + double smooth_value() const { return m_smooth.generator().smooth_value(); } ++ unsigned type() const { return m_smooth.type(); } + + private: + conv_smooth_poly1_curve(const conv_smooth_poly1_curve&); +-- +1.8.1.4 + diff --git a/SDK/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch b/SDK/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch new file mode 100644 index 0000000..547b0d2 --- /dev/null +++ b/SDK/package/agg/0014-Avoid-potential-zero-division-resulting-in-nan-in-ag.patch @@ -0,0 +1,30 @@ +From 6433a64f4cd41e88499386b0b7c7ae05d30683b8 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:33:32 +0100 +Subject: [PATCH 14/15] Avoid potential zero division resulting in nan in + agg::gamma_linear + +--- + include/agg_gamma_functions.h | 6 +++++- + 1 file changed, 5 insertions(+), 1 deletion(-) + +diff --git a/include/agg_gamma_functions.h b/include/agg_gamma_functions.h +index fa38a45..beb0c04 100644 +--- a/include/agg_gamma_functions.h ++++ b/include/agg_gamma_functions.h +@@ -94,7 +94,11 @@ namespace agg + { + if(x < m_start) return 0.0; + if(x > m_end) return 1.0; +- return (x - m_start) / (m_end - m_start); ++ double delta = m_end - m_start; ++ // avoid nan from potential zero division ++ // https://github.com/mapnik/mapnik/issues/761 ++ if (delta <= 0.0) return 0.0; ++ return (x - m_start) / delta; + } + + private: +-- +1.8.1.4 + diff --git a/SDK/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch b/SDK/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch new file mode 100644 index 0000000..6214bd6 --- /dev/null +++ b/SDK/package/agg/0015-Ensure-first-value-in-the-gamma-table-is-always-zero.patch @@ -0,0 +1,24 @@ +From ca818d4dcd428c5560fc3c341fbaf427a7485e32 Mon Sep 17 00:00:00 2001 +From: Tom Hughes +Date: Sat, 22 Jun 2013 12:34:37 +0100 +Subject: [PATCH 15/15] Ensure first value in the gamma table is always zero + +--- + include/agg_gamma_functions.h | 1 + + 1 file changed, 1 insertion(+) + +diff --git a/include/agg_gamma_functions.h b/include/agg_gamma_functions.h +index beb0c04..b8eda52 100644 +--- a/include/agg_gamma_functions.h ++++ b/include/agg_gamma_functions.h +@@ -49,6 +49,7 @@ namespace agg + + double operator() (double x) const + { ++ if (x == 0.0) return 0.0; + return pow(x, m_gamma); + } + +-- +1.8.1.4 + diff --git a/SDK/package/agg/CREATE_FILES.patch b/SDK/package/agg/CREATE_FILES.patch new file mode 100644 index 0000000..1a78125 --- /dev/null +++ b/SDK/package/agg/CREATE_FILES.patch @@ -0,0 +1,14 @@ +--- a/README.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/README 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac + +--- a/NEWS.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/NEWS 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac + +--- a/AUTHORS.orig 2007-01-07 13:58:28.000000000 +0000 ++++ b/AUTHORS 2007-01-07 14:02:40.000000000 +0000 +@@ -0,0 +1 @@ ++cacac diff --git a/SDK/package/agg/Config.in b/SDK/package/agg/Config.in new file mode 100644 index 0000000..a842098 --- /dev/null +++ b/SDK/package/agg/Config.in @@ -0,0 +1,13 @@ +config BR2_PACKAGE_AGG + bool "agg" + depends on BR2_INSTALL_LIBSTDCPP + select BR2_PACKAGE_SDL + help + The Anti-Grain Geometry project. A High Quality 2D Graphics Rendering + Engine for C++. + We select the SDL backend by default. + + http://www.antigrain.com/index.html + +comment "agg needs a toolchain with C++ support" + depends on !BR2_INSTALL_LIBSTDCPP diff --git a/SDK/package/agg/agg-2.4-depends.patch b/SDK/package/agg/agg-2.4-depends.patch new file mode 100644 index 0000000..f5506e2 --- /dev/null +++ b/SDK/package/agg/agg-2.4-depends.patch @@ -0,0 +1,36 @@ +--- agg-2.4.orig/font_freetype/Makefile.am 2005-10-18 11:45:40.000000000 +0100 ++++ agg-2.4/font_freetype/Makefile.am 2006-07-10 15:11:55.000000000 +0100 +@@ -4,8 +4,9 @@ + agginclude_HEADERS = agg_font_freetype.h + lib_LTLIBRARIES = libaggfontfreetype.la + +-libaggfontfreetype_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ @FREETYPE_LIBS@ ++libaggfontfreetype_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ + libaggfontfreetype_la_SOURCES = agg_font_freetype.cpp + libaggfontfreetype_la_CXXFLAGS = -I$(top_srcdir)/include @FREETYPE_CFLAGS@ ++libaggfontfreetype_la_LIBADD = ../src/libagg.la @FREETYPE_LIBS@ + endif + +--- agg-2.4.orig/src/platform/sdl/Makefile.am 2005-10-17 23:49:35.000000000 +0100 ++++ agg-2.4/src/platform/sdl/Makefile.am 2006-07-10 15:11:55.000000000 +0100 +@@ -5,6 +5,6 @@ + libaggplatformsdl_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ + libaggplatformsdl_la_SOURCES = agg_platform_support.cpp + libaggplatformsdl_la_CXXFLAGS = -I$(top_srcdir)/include @SDL_CFLAGS@ +-libaggplatformsdl_la_LIBADD = @SDL_LIBS@ ++libaggplatformsdl_la_LIBADD = ../../libagg.la @SDL_LIBS@ + endif + +--- agg-2.5.orig/src/platform/X11/Makefile.am 2006-12-11 00:59:45.000000000 +0000 ++++ agg-2.5/src/platform/X11/Makefile.am 2007-01-07 14:07:39.000000000 +0000 +@@ -1,8 +1,8 @@ + if ENABLE_X11 + lib_LTLIBRARIES = libaggplatformX11.la + +-libaggplatformX11_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ -L@x_libraries@ ++libaggplatformX11_la_LDFLAGS = -version-info @AGG_LIB_VERSION@ @X_LDFLAGS@ + libaggplatformX11_la_SOURCES = agg_platform_support.cpp + libaggplatformX11_la_CXXFLAGS = -I$(top_srcdir)/include -I@x_includes@ +-libaggplatformX11_la_LIBADD = -lX11 ++libaggplatformX11_la_LIBADD = ../../libagg.la -lX11 + endif diff --git a/SDK/package/agg/agg-2.5-autotools.patch b/SDK/package/agg/agg-2.5-autotools.patch new file mode 100644 index 0000000..1272b65 --- /dev/null +++ b/SDK/package/agg/agg-2.5-autotools.patch @@ -0,0 +1,11 @@ +--- a/configure.in 2013-02-22 09:30:00.000000000 -0600 ++++ b/configure.in 2013-02-22 09:30:49.030777571 -0600 +@@ -8,7 +8,7 @@ + AC_PROG_CC + AC_PROG_CXX + AC_ISC_POSIX +-AM_C_PROTOTYPES ++#AM_C_PROTOTYPES + if test "x$U" != "x"; then + AC_MSG_ERROR(Compiler not ANSI compliant) + fi diff --git a/SDK/package/agg/agg-2.5-pkgconfig.patch b/SDK/package/agg/agg-2.5-pkgconfig.patch new file mode 100644 index 0000000..a303bfb --- /dev/null +++ b/SDK/package/agg/agg-2.5-pkgconfig.patch @@ -0,0 +1,10 @@ +--- agg-2.5/libagg.pc.in.orig 2007-01-07 13:58:28.000000000 +0000 ++++ agg-2.5/libagg.pc.in 2007-01-07 14:02:40.000000000 +0000 +@@ -6,5 +6,6 @@ + Name: libagg + Description: Anti Grain Geometry - A High Quality Rendering Engine for C++ + Version: @VERSION@ +-Libs: -L${libdir} -Wl,-rpath,${exec_prefix}/lib -lagg ++Requires.private: freetype2 ++Libs: -L${libdir} -lagg + Cflags: -I${includedir} diff --git a/SDK/package/agg/agg.mk b/SDK/package/agg/agg.mk new file mode 100644 index 0000000..ecf5749 --- /dev/null +++ b/SDK/package/agg/agg.mk @@ -0,0 +1,32 @@ +############################################################################### +# +# agg +# +############################################################################### + +AGG_VERSION = 2.5 +AGG_SOURCE = agg-$(AGG_VERSION).tar.gz +AGG_SITE = https://ftp.osuosl.org/pub/blfs/8.0/a +AGG_LICENSE = GPLv3+ +AGG_LICENSE_FILES = COPYING +AGG_INSTALL_STAGING = YES +AGG_AUTORECONF = YES + +AGG_DEPENDENCIES = host-pkgconf sdl + +AGG_CONF_OPTS = \ + --with-sdl-prefix=$(STAGING_DIR)/usr \ + --disable-sdltest + +AGG_CONF_OPTS += \ + --with-x=NO \ + --disable-examples --disable-gpc + +ifeq ($(BR2_PACKAGE_FREETYPE),y) +AGG_DEPENDENCIES += freetype +AGG_CONF_OPTS += --enable-freetype +else +AGG_CONF_OPTS += --disable-freetype +endif + +$(eval $(autotools-package)) diff --git a/SDK/package/dmtx-utils/0001-no-static-debug.patch b/SDK/package/dmtx-utils/0001-no-static-debug.patch new file mode 100644 index 0000000..7fbfa01 --- /dev/null +++ b/SDK/package/dmtx-utils/0001-no-static-debug.patch @@ -0,0 +1,35 @@ + dmtxquery/Makefile.am | 2 +- + dmtxread/Makefile.am | 2 +- + dmtxwrite/Makefile.am | 2 +- + 3 files changed, 3 insertions(+), 3 deletions(-) + +diff -Naur a/dmtxquery/Makefile.am b/dmtxquery/Makefile.am +--- a/dmtxquery/Makefile.am ++++ b/dmtxquery/Makefile.am +@@ -9,5 +9,5 @@ + dmtxquery_LDADD = $(LIBOBJS) + + dmtxquery_debug_SOURCES = dmtxquery.c dmtxquery.h ../common/dmtxutil.c ../common/dmtxutil.h +-dmtxquery_debug_LDFLAGS = -static $(DMTX_LIBS) ++dmtxquery_debug_LDFLAGS = $(DMTX_LIBS) + dmtxquery_debug_LDADD = $(LIBOBJS) +diff -Naur a/dmtxread/Makefile.am b/dmtxread/Makefile.am +--- a/dmtxread/Makefile.am ++++ b/dmtxread/Makefile.am +@@ -11,5 +11,5 @@ + + dmtxread_debug_SOURCES = dmtxread.c dmtxread.h ../common/dmtxutil.c ../common/dmtxutil.h + dmtxread_debug_CFLAGS = $(DMTX_CFLAGS) $(MAGICK_CFLAGS) -D_MAGICK_CONFIG_H +-dmtxread_debug_LDFLAGS = -static $(DMTX_LIBS) $(MAGICK_LIBS) ++dmtxread_debug_LDFLAGS = $(DMTX_LIBS) $(MAGICK_LIBS) + dmtxread_debug_LDADD = $(LIBOBJS) +diff -Naur a/dmtxwrite/Makefile.am b/dmtxwrite/Makefile.am +--- a/dmtxwrite/Makefile.am ++++ b/dmtxwrite/Makefile.am +@@ -11,5 +11,5 @@ + + dmtxwrite_debug_SOURCES = dmtxwrite.c dmtxwrite.h ../common/dmtxutil.c ../common/dmtxutil.h + dmtxwrite_debug_CFLAGS = $(DMTX_FLAGS) $(MAGICK_CFLAGS) -D_MAGICK_CONFIG_H +-dmtxwrite_debug_LDFLAGS = -static $(DMTX_LIBS) $(MAGICK_LIBS) ++dmtxwrite_debug_LDFLAGS = $(DMTX_LIBS) $(MAGICK_LIBS) + dmtxwrite_debug_LDADD = $(LIBOBJS) diff --git a/SDK/package/dmtx-utils/Config.in b/SDK/package/dmtx-utils/Config.in new file mode 100644 index 0000000..0d22f43 --- /dev/null +++ b/SDK/package/dmtx-utils/Config.in @@ -0,0 +1,12 @@ +config BR2_PACKAGE_DMTX_UTILS + bool "dmtx utils" + depends on BR2_PACKAGE_LIBDMTX + select BR2_PACKAGE_IMAGEMAGICK + help + libdmtx is a software library that enables programs to read + and write Data Matrix barcodes of the modern ECC200 + variety. This package, dmtx-utils, provides command line + utilities that allow scripts and command line users to use + libdmtx functionality. + + https://github.com/dmtx/dmtx-utils diff --git a/SDK/package/dmtx-utils/dmtx-utils.hash b/SDK/package/dmtx-utils/dmtx-utils.hash new file mode 100644 index 0000000..0b9bf1c --- /dev/null +++ b/SDK/package/dmtx-utils/dmtx-utils.hash @@ -0,0 +1,3 @@ +# Locally computed: +sha256 0d396ec14f32a8cf9e08369a4122a16aa2e5fa1675e02218f16f1ab777ea2a28 dmtx-utils-0.7.6.tar.gz +sha256 d8c320ffc0030d1b096ae4732b50d2b811cf95e9a9b7377c1127b2563e0a0388 COPYING diff --git a/SDK/package/dmtx-utils/dmtx-utils.mk b/SDK/package/dmtx-utils/dmtx-utils.mk new file mode 100644 index 0000000..4560b59 --- /dev/null +++ b/SDK/package/dmtx-utils/dmtx-utils.mk @@ -0,0 +1,20 @@ +################################################################################ +# +# dmtx-utils +# +################################################################################ + +DMTX_UTILS_VERSION = 0.7.6 +DMTX_UTILS_SITE = $(call github,dmtx,dmtx-utils,v$(DMTX_UTILS_VERSION)) +DMTX_UTILS_DEPENDENCIES = libdmtx imagemagick +DMTX_UTILS_LICENSE = LGPL-2.1+ +DMTX_UTILS_LICENSE_FILES = COPYING +# github tarball does not include configure +DMTX_UTILS_AUTORECONF = YES + +define DMTX_UTILS_RUN_AUTOGEN + cd $(@D) && PATH=$(BR_PATH) ./autogen.sh +endef +DMTX_UTILS_PRE_CONFIGURE_HOOKS += DMTX_UTILS_RUN_AUTOGEN + +$(eval $(autotools-package)) diff --git a/SDK/package/fluidlite/0001-fluidlite.patch b/SDK/package/fluidlite/0001-fluidlite.patch new file mode 100644 index 0000000..9dc01c4 --- /dev/null +++ b/SDK/package/fluidlite/0001-fluidlite.patch @@ -0,0 +1,11 @@ +--- a/CMakeLists.txt ++++ b/CMakeLists.txt +@@ -125,7 +125,7 @@ + endif() + endif() + +-option(FLUIDLITE_BUILD_SHARED "Build shared library" TRUE) ++option(FLUIDLITE_BUILD_SHARED "Build shared library" FALSE) + if(FLUIDLITE_BUILD_SHARED) + add_library(${PROJECT_NAME} SHARED ${SOURCES}) + diff --git a/SDK/package/fluidlite/Config.in b/SDK/package/fluidlite/Config.in new file mode 100644 index 0000000..bdfb7e0 --- /dev/null +++ b/SDK/package/fluidlite/Config.in @@ -0,0 +1,5 @@ +config BR2_PACKAGE_FLUIDLITE + bool "FluidLite" + depends on BR2_USE_MMU + help + FluidLite is a very light version of FluidSynth. diff --git a/SDK/package/fluidlite/fluidsynth.hash b/SDK/package/fluidlite/fluidsynth.hash new file mode 100644 index 0000000..d39b0e8 --- /dev/null +++ b/SDK/package/fluidlite/fluidsynth.hash @@ -0,0 +1,3 @@ +# Locally computed +sha256 ef4d008f9fe2fa9a48135505d42dd7e8e9cc4d7494a4b13d6caa13adb5c61ff8 1.0.tar.gz +sha256 a5564e99fd7f49e1344152a8c5bc1d420d5f973b30e010946764db0b5b9e668c LICENSE diff --git a/SDK/package/fluidlite/fluidsynth.mk b/SDK/package/fluidlite/fluidsynth.mk new file mode 100644 index 0000000..5cd24b7 --- /dev/null +++ b/SDK/package/fluidlite/fluidsynth.mk @@ -0,0 +1,19 @@ +################################################################################ +# +# FLUIDLITE +# +################################################################################ + +FLUIDLITE_VERSION = fdd05bad03cdb24d1f78b5fe3453842890c1b0e8 +FLUIDLITE_SITE = $(call github,gcw0,FluidLite,$(FLUIDLITE_VERSION)) +FLUIDLITE_LICENSE = LGPL-2.1+ +FLUIDLITE_LICENSE_FILES = LICENSE +FLUIDLITE_INSTALL_STAGING = YES +FLUIDLITE_DEPENDENCIES = + +# Disable the shared library for static only build +ifeq ($(BR2_STATIC_LIBS),y) +FLUIDLITE_CONF_OPTS += -DDFLUIDLITE_BUILD_SHARED=FALSE +endif + +$(eval $(cmake-package)) diff --git a/SDK/package/libini/Config.in b/SDK/package/libini/Config.in new file mode 100644 index 0000000..a42967b --- /dev/null +++ b/SDK/package/libini/Config.in @@ -0,0 +1,6 @@ +config BR2_PACKAGE_LIBINI + bool "libini" + help + Tiny library to help reading INI files. + + https://github.com/pcercuei/libini diff --git a/SDK/package/libini/libini.mk b/SDK/package/libini/libini.mk new file mode 100644 index 0000000..7d50b25 --- /dev/null +++ b/SDK/package/libini/libini.mk @@ -0,0 +1,13 @@ +############################################################# +# +# libini +# +############################################################# +LIBINI_VERSION = c3413da +LIBINI_SITE_METHOD = git +LIBINI_SITE = https://github.com/FunKey-Project/libini.git +LIBINI_LICENSE = LGPL-2.1 + +LIBINI_INSTALL_STAGING = YES + +$(eval $(cmake-package)) diff --git a/SDK/package/libmikmod/Config.in b/SDK/package/libmikmod/Config.in new file mode 100644 index 0000000..909cc00 --- /dev/null +++ b/SDK/package/libmikmod/Config.in @@ -0,0 +1,7 @@ +config BR2_PACKAGE_LIBMIKMOD + bool "libmikmod" + help + Mikmod is a module player and library supporting many + tracker formats, including mod, s3m, it, and xm. + + http://mikmod.shlomifish.org/ diff --git a/SDK/package/libmikmod/libmikmod.mk b/SDK/package/libmikmod/libmikmod.mk new file mode 100644 index 0000000..cce5190 --- /dev/null +++ b/SDK/package/libmikmod/libmikmod.mk @@ -0,0 +1,21 @@ +############################################################# +# +# libmikmod +# +############################################################# +LIBMIKMOD_VERSION:=3.3.11.1 +LIBMIKMOD_SITE:=http://sourceforge.net/projects/mikmod/files/libmikmod/$(LIBMIKMOD_VERSION) + +LIBMIKMOD_CONF_OPTS = --localstatedir=/var + +LIBMIKMOD_LIBTOOL_PATCH = NO +LIBMIKMOD_INSTALL_STAGING = YES + +LIBMIKMOD_CONFIG_SCRIPTS = libmikmod-config + +define LIBMIKMOD_REMOVE_LIBMIKMOD_CONFIG +mv $(TARGET_DIR)/usr/bin/libmikmod-config $(HOST_DIR)/bin/ +endef +LIBMIKMOD_POST_INSTALL_TARGET_HOOKS += LIBMIKMOD_REMOVE_LIBMIKMOD_CONFIG + +$(eval $(autotools-package)) diff --git a/SDK/package/libopk/Config.in b/SDK/package/libopk/Config.in new file mode 100644 index 0000000..db3bce9 --- /dev/null +++ b/SDK/package/libopk/Config.in @@ -0,0 +1,8 @@ +config BR2_PACKAGE_LIBOPK + bool "libopk" + select BR2_PACKAGE_ZLIB + select BR2_PACKAGE_LIBINI + help + Library to handle OPK packages. + + https://github.com/pcercuei/libopk diff --git a/SDK/package/libopk/libopk.mk b/SDK/package/libopk/libopk.mk new file mode 100644 index 0000000..8cf840a --- /dev/null +++ b/SDK/package/libopk/libopk.mk @@ -0,0 +1,14 @@ +############################################################# +# +# libopk +# +############################################################# +LIBOPK_VERSION = 3c918c8 +LIBOPK_SITE_METHOD = git +LIBOPK_SITE = https://github.com/FunKey-Project/libopk.git + +LIBOPK_DEPENDENCIES = libini zlib + +LIBOPK_INSTALL_STAGING = YES + +$(eval $(cmake-package)) diff --git a/SDK/package/libxdgmime/Config.in b/SDK/package/libxdgmime/Config.in new file mode 100644 index 0000000..f1d4fc0 --- /dev/null +++ b/SDK/package/libxdgmime/Config.in @@ -0,0 +1,12 @@ +config BR2_PACKAGE_LIBXDGMIME + bool "libxdgmime" + depends on BR2_USE_WCHAR # shared-mime-info + select BR2_PACKAGE_SHARED_MIME_INFO + help + Simple library that parses the proposed MIME spec + listed at http://freedesktop.org/. + + https://github.com/pcercuei/libxdgmime + +comment "libxdgmime requires a toolchain with WCHAR support" + depends on !BR2_USE_WCHAR diff --git a/SDK/package/libxdgmime/libxdgmime.mk b/SDK/package/libxdgmime/libxdgmime.mk new file mode 100644 index 0000000..c6e592f --- /dev/null +++ b/SDK/package/libxdgmime/libxdgmime.mk @@ -0,0 +1,30 @@ +############################################################# +# +# libxdgmime +# +############################################################# +LIBXDGMIME_VERSION = db79e7c +LIBXDGMIME_SITE_METHOD = git +LIBXDGMIME_SITE = https://github.com/FunKey-Project/libxdgmime.git +LIBXDGMIME_DEPENDENCIES = shared-mime-info +LIBXDGMIME_LICENCE = LGPL-2.1+ or AFL-2.1 + +LIBXDGMIME_INSTALL_STAGING = YES + +LIBXDGMIME_MAKE_ENV = CFLAGS="$(TARGET_CFLAGS)" LDFLAGS="$(TARGET_LDFLAGS)" \ + CROSS_COMPILE="$(TARGET_CROSS)" PREFIX=/usr \ + PLATFORM="$(BR2_VENDOR)" + +define LIBXDGMIME_BUILD_CMDS + $(LIBXDGMIME_MAKE_ENV) $(MAKE) -C $(@D) +endef + +define LIBXDGMIME_INSTALL_STAGING_CMDS + $(LIBXDGMIME_MAKE_ENV) DESTDIR="$(STAGING_DIR)" $(MAKE) -C $(@D) install +endef + +define LIBXDGMIME_INSTALL_TARGET_CMDS + $(LIBXDGMIME_MAKE_ENV) DESTDIR="$(TARGET_DIR)" $(MAKE) -C $(@D) install-lib +endef + +$(eval $(generic-package)) diff --git a/buildroot b/buildroot index d8082db..d526838 160000 --- a/buildroot +++ b/buildroot @@ -1 +1 @@ -Subproject commit d8082db677046e004a6537828b3e4f4b9a818a4f +Subproject commit d5268384884357d5f17515ee083939f9596a2e73 diff --git a/genimage.cfg b/genimage.cfg index 5570864..b91bcac 100644 --- a/genimage.cfg +++ b/genimage.cfg @@ -31,7 +31,7 @@ image sdcard.img { partition-type = 0x83 bootable = "yes" image = "FunKey/output/images/rootfs.ext4" - size = 100M # This will be resized to 1G during first boot + size = 160M # This will be resized to 1G during first boot } # These partitions will be created during first boot