Cryptsetup 2.4.1 Release Notes
==============================
Stable bug-fix release with minor extensions.
All users of cryptsetup 2.4.0 should upgrade to this version.
Changes since version 2.4.0
~~~~~~~~~~~~~~~~~~~~~~~~~~~
* Fix compilation for libc implementations without dlvsym().
Some alternative libc implementations (like musl) do not provide
versioned symbols dlvsym function. Code now fallbacks to dlsym
operation for dynamic LUKS2 token load.
It is up to maintainers to ensure that LUKS2 token plugins are
compiled for the supported version.
* Fix compilation and tests on systems with non-standard libraries
(standalone argp library, external gettext library, BusyBox
implementations of standard tools).
* Try to workaround some issues on systems without udev support.
NOTE: non-udev systems cannot provide all functionality for kernel
device-mapper, and some operations can fail.
* Fixes for OpenSSL3 crypto backend (including FIPS mode).
Because cryptsetup still requires some hash functions implemented
in OpenSSL3 legacy provider, crypto backend now uses its library
context and tries to load both default and legacy OpenSSL3 providers.
If FIPS mode is detected, no library context is used, and it is up
to the OpenSSL system-wide policy to load proper providers.
NOTE: We still use some deprecated API in the OpenSSL3 backend,
and there are some known problems in OpenSSL 3.0.0.
* Print error message when assigning a token to an inactive keyslot.
* Fix offset bug in LUKS2 encryption code if --offset option was used.
* Do not allow LUKS2 decryption for devices with data offset.
Such devices cannot be used after decryption.
* Fix LUKS1 cryptsetup repair command for some specific problems.
Repair code can now fix wrongly used initialization vector
specification in ECB mode (that is insecure anyway!) and repair
the upper-case hash specification in the LUKS1 header.
Signed-off-by: Wang Mingyu <wangmy@fujitsu.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Without the udevrules cryptsetup luksOpen will be hanging with "Udev
cookie 0xd4de0f6 (semid 5) waiting for zero".
Signed-off-by: Khem Raj <raj.khem@gmail.com>
This is the result of automated script (0.9.1) conversion:
oe-core/scripts/contrib/convert-overrides.py .
converting the metadata to use ":" as the override character instead of "_".
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Otherwise cryptsetup-native depends on the target kernel and thus the
target compiler, as can be seen by:
$ bitbake -g cryptsetup-native
$ grep 'cryptsetup.*linux-yocto' task-depends.dot
"cryptsetup-native.do_build" -> "linux-yocto.do_deploy"
"cryptsetup-native.do_build" -> "linux-yocto.do_package_write_rpm"
"cryptsetup-native.do_populate_sysroot" -> "linux-yocto.do_populate_sysroot"
$ grep 'linux-yocto.*gcc-cross' task-depends.dot
"linux-yocto.do_kernel_configme" -> "gcc-cross-x86_64.do_populate_sysroot"
"linux-yocto.do_prepare_recipe_sysroot" -> "gcc-cross-x86_64.do_populate_sysroot"
This also moves the runtime dependencies to near the end of the recipe,
which is more customary.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The util-linux recipe in Poky has been split, and libuuid is separate now:
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=d42487bf52310d095178e480b7ddec2666471009
This allows to build util-linux with cryptsetup, for native dm-verity
support.
The main cryptsetup build needs libuuid and not the full util-linux, so
switch the build-dependency over, thus allowing users to enable the
cryptsetup util-linux's PACKAGECONFIG.
The libblkid dependency is handled individually by the crypsetup's
PACKAGECONFIG option.
Signed-off-by: Luca Boccassi <luca.boccassi@microsoft.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Since commit [libdevmapper/lvm2: force recipe libdevmapper to populate
sysroot only] applied, if recipe DEPENDS on libdevmapper, we have to
add it to packages RDEPENDS to fix [file-rdeps] QA issue
Set PREFERRED_RPROVIDER_libdevmapper = "lvm2" in layer.conf to explicit
RDEPENDS on libdevmapper in cryptsetup
Suggested-by : peter.kjellerstedt@axis.com
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Since commit [libdevmapper/lvm2: force recipe libdevmapper to populate
sysroot only] applied, if recipe DEPENDS on libdevmapper, we have to
add it to packages RDEPENDS to fix [file-rdeps] QA issue
Skip [build-deps] and [file-rdeps] QA checking, add lvm2 to DEPENDS will
triger circular dependencies, use recipe libdevmapper to replace,
ignore the QA checking is fine.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
The cryptsetup 2.1 uses LUKS2 format as the default LUKS format. This
change introduced the following issues:
* LUKS2 requires kernel userspace crypto API to be available
(CONFIG_CRYPTO_USER_API and CONFIG_CRYPTO_USER_API_SKCIPHER). But
linux-yocto doesn't enable these options by default. If missing these
kernel modules, the cryptsetup will fall back to using dmcrypt-device
for keyslot processing.
$ cryptsetup --debug --type luks luksFormat /dev/sda3
[snip]
Checking if cipher aes-xts-plain64 is usable.
Userspace crypto wrapper cannot use aes-xts-plain64 (-95).
Using dmcrypt to access keyslot area.
[snip]
* The grub can not decrypt a LUKS2 encrypted boot partition because it
doesn't support LUKS2 now.
See grub bug: https://savannah.gnu.org/bugs/?55093
Add a PACKAGCONFIG for luks format and set the default LUKS format to
LUKS1. The users can specify '--type luks2' in cryptsetup command line
if they want to use LUKS2.
Signed-off-by: Yi Zhao <yi.zhao@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
When building cryptsetup-native, don't enable udev support since there
is not a udev-native recipe.
When udev is enabled, change the dependency from a DEPENDS to an
RDEPENDS. The --enable-udev option adds a runtime check for udev before
using udev features of libdevmapper. Because of this, udev isn't needed
at build time, just runtime.
Signed-off-by: Robert Joslyn <robert.joslyn@redrectangle.org>
Reviewed-by: Mikko Rapeli <mikko.rapeli@bmw.de>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Instead of hardcoding the version in SRC_URI,
get the version for the folder of the source
dynamically in SRC_URI.
Signed-off-by: Mingli Yu <mingli.yu@windriver.com>
Signed-off-by: Khem Raj <raj.khem@gmail.com>
