Perform more optimization in supercard platform (#107)

arm9/source/patches/platform/supercard/SuperCardCommon.s

@@ -8,6 +8,7 @@
 
 @ void sc_change_mode(uint16_t mode);
 BEGIN_THUMB_FUNCTION sccmn_changeMode
+	push {r0-r3,lr}
     ldr r2,= 0x09FFFFFE
     ldr r3,= 0xA55A
     strh r3, [r2]
@@ -19,24 +20,26 @@ BEGIN_THUMB_FUNCTION sccmn_changeMode
 	@ if mode is 3 (sd enable), reset the sd card
     @ SDResetCard
     @ write 0
-    ldr r2,= sd_resetaddr
+    ldr r2, =sd_resetaddr
     strh r2, [r2]
 1:
-    mov pc, lr
+    pop {r0-r3,pc}
 
-.section "scsd_common", "ax"
-@ this function will trash r4 but leave r0 and r1 untouched
 @ void SDSendClock10(void)
 BEGIN_THUMB_FUNCTION sccmn_sdSendClock10
+	@ here we get lr from the caller
+	push {r0-r3}
     movs r3, #0x10
     @ loads reg_scsd_cmd
     movs r2, #0x98
     lsls r2, r2, #20
 1:
-    ldrh r4, [r2]
+    ldrh r0, [r2]
     subs r3, r3, #1
     bne 1b
-    mov pc, lr
+    pop {r0-r3,pc}
+
+.section "scsd_common", "ax"
 
 @ static uint64_t inline calSingleCRC16(uint64_t crc, uint32_t data_in){
 @ 	// Shift out 8 bits for each line
@@ -73,8 +76,10 @@ BEGIN_THUMB_FUNCTION sccmn_sdSendClock10
 
 @ uint64_t sdio_crc16_4bit_checksum(void*)
 @ r0 and r1 are left untouched, the result is returned in r2-r3 instead
+@ a "continuation function" is provided in r5, and we have to jump to it
 BEGIN_THUMB_FUNCTION sccmn_sdio4BitCrc16
-    push {r0,r1,r4-r5,lr}
+	push {r5,lr}
+    push {r0,r1,r4-r5}
     movs r4, #0 @ r4 = crc_lo
     movs r5, #0 @ r5 = crc_hi
     movs r1, #128
@@ -116,7 +121,7 @@ byteSwap32:
     push {r0,r1,r4-r5,lr}
 byteSwap32_nopush:
     movs r5, #16
-    ldr r4,= 0xFF00FF
+    ldr r4, =0xFF00FF
     rors r2, r5 // ror 16
     ands r4, r2
     bics r2, r4