Commit 75320f567d "configs/imx6ulz_bsh_smm_m2: bump U-Boot to 2024.10
and Linux to 6.1.120" updated the kernel from 6.1.68 to 6.1.120 but
forgot to update its per-config kernel hash file.
The defconfig fails with the error:
ERROR: No hash found for linux-6.1.120.tar.xz
This failure happen since commit 07149e5f30 "{linux, linux-headers}:
bump 5.{4, 10, 15}.x / 6.{1, 6, 12}.x series" which updated the
Buildroot 6.1 kernel series from 6.1.120 to 6.1.122.
This commit fixes the issue by updating the hash.
While at it, it also changes the "Locally computed" comment
by the upstream link to hashes.
Fixes:
https://gitlab.com/buildroot.org/buildroot/-/jobs/8751983243
Cc: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Cc: Michael Trimarchi <michael@amarulasolutions.com>
Reviewed-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Usually, ubxtool, a python-based tool to drive u-blox devices, connects
to a gpsd instance and delegates to it the responsibility to write to
and read from the actual device. This is sane, because a serial device
can only be opened once, and if gpsd is running, it has that device
open.
However, in some cases, ubxtool can be used to directly talk to the
device, to pre-configure it before gpsd runs, or even in the absence of
gpsd altogether. This is not used very often, except when setting up an
RTK base, where gpsd is not needed.
In that case, ubxtool will directly talk to the serial device. It uses
the pyserial python module. Since this is not the traditional way to
talk to the device, failure to import the module is ignored, and the
error reporting is deferred until it is actually needed, which is why we
did not catch the issue earlier. See [1] and [2].
Fixes: f3ef0723cf (package/gpsd: enable python support and modules)
[1] https://gitlab.com/gpsd/gpsd/-/blob/release-3.25/clients/ubxtool.py.in#L47
[2] https://gitlab.com/gpsd/gpsd/-/blob/release-3.25/gps/gps.py.in#L36
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Bernd Kuhls <bernd@kuhls.net>
[Julien: add link to described code portion]
Signed-off-by: Julien Olivain <ju.o@free.fr>
libxcrypt has been added as a replacement for the libcrypt
implementation that was part of glibc, but dropped from glibc starting
from version 2.39.
However, libxcrypt was made available for all C libraries, and this is
unfortunately causing some problems as it can clash with the libcrypt
implementation provided by the C library.
In particular, linux-pam has been consistently failing with uclibc, in
BR2_PER_PACKAGE_DIRECTORIES=y builds, with the following build
failure:
opasswd.c: In function 'compare_password':
opasswd.c:133:27: error: invalid application of 'sizeof' to incomplete type 'struct crypt_data'
What happens is relatively tricky, but let's try to break it down:
- uclibc-ng install a stub libcrypt.a (no shared variant, as for
shared libraries, everything is in libc.so), and crypt.h
- libxcrypt installs libcrypt.so.* and crypt.h
So there is no "clash" on the library itself, but there is a clash on
the header file.
Since we're using BR2_PER_PACKAGE_DIRECTORIES=y, when building
linux-pam, we are creating the per-package STAGING_DIR by copying the
STAGING_DIR of linux-pam dependencies, i.e both the libxcrypt
STAGING_DIR and the uclibc-ng STAGING_DIR. But the latter ends up
being copied last, which means that at the end of the day, we have in
the per-package STAGING_DIR of linux-pam:
- The libcrypt.so from libxcrypt
- The crypt.h header from uclibc-ng
- The libcrypt.a from uclibc-ng
When the ./configure script of linux-pam tests whether the library has
crypt_r(), it concludes that yes it's available: and indeed
libcrypt.so from libxcrypt has it.
So it tries to use 'struct crypt_data' and 'crypt_r()', but those are
not supported in uClibc-ng, and so cannot be found in the <crypt.h>
header. So even if the ./configure script and the linux-pam code has
some logic to fallback to crypt() if crypt_r() isn't available, this
fallback doesn't trigger because the installed libcrypt.so does have
crypt_r().
Basically what happens is that uclibc-ng + libxcrypt is a combo that
violates a golden rule of our BR2_PER_PACKAGE_DIRECTORIES=y
implementation: packages shouldn't overwrite files from each other.
To avoid this situation, we make libxcrypt only installable on
glibc. This isn't a problem because as of today, BR2_PACKAGE_LIBXCRYPT
is always selected "if BR2_TOOLCHAIN_USES_GLIBC".
It should be noted though that the case of an older glibc (which still
had its own internal libcrypt) + libxcrypt continues to exist. It's
less likely to cause trouble though, as the libcrypt implementations
are much more similar.
Fixes:
http://autobuild.buildroot.net/results/560f66b0311d02dc884732221d6870ae3c38067c/
Note: we do not add a Config.in comment for this glibc dependency,
because libxcrypt really is a "replacement" library to fill in the
void left by libcrypt's removal from glibc. There isn't realy a point
showing "libxcrypt needs a toolchain w/ glibc", because with musl or
uclibc-ng, the libcrypt functionality is directly part of the C
library.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
We are seeing build issues with linux-pam in the autobuilders such as:
md5_crypt.c: In function 'Goodcrypt_md5':
md5_crypt.c:145:13: error: implicit declaration of function 'asprintf'; did you mean 'vsprintf'? [-Wimplicit-function-declaration]
145 | if (asprintf(&passwd, "%s%.*s$%s", magic, sl, sp, buf) < 0)
| ^~~~~~~~
| vsprintf
This is due to the fact that <stdio.h> gets included without
_GNU_SOURCE being defined, and so the prototype of asprintf() is not
accessible, at least with uclibc-ng.
The _GNU_SOURCE definition is properly in linux-pam's config.h, but
config.h doesn't get properly included first everywhere. This issue
has been fixed upstream in the mean time, so we simply backport the
upstream patch.
Fixes:
http://autobuild.buildroot.net/results/49b190b3fbae3cdca4c7a08b3ab5100a937ede9e/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Add custom version option, to enable easily specifying an official OP-TEE
version to use.
This is similar to what is done in other packages, such as linux, uboot or
arm-trusted-firmware.
Signed-off-by: Vincent Stehlé <vincent.stehle@arm.com>
Cc: Étienne Carrière <etienne.carriere@foss.st.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
We can embed a TEE in ELF format into U-Boot, but versions of OP-TEE since
3.8.0 must be embedded in binary format, as the tee.bin contains important
meta-data. [1]
Update the configuration menu and the Makefile to allow selecting to
embed either tee.elf or tee.bin.
By default, embed the TEE in ELF format to stay compatible with existing
configurations.
[1] https://github.com/OP-TEE/optee_os/issues/4542
Signed-off-by: Vincent Stehlé <vincent.stehle@arm.com>
Cc: Christoph Muellner <christoph.muellner@theobroma-systems.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Buildroot commit
1f4b4ccde7 ("package/opensc: security
bump to version 0.26.0") bumped opensc from 0.24 to 0.26, and the
build started failing with:
pkcs11-tool.c:7854:45: warning: implicit declaration of function 'EVP_bf_cbc'; did you mean 'EVP_sm4_cbc'? [-Wimplicit-function-declaration]
on configurations that have BR2_PACKAGE_LIBOPENSSL_ENABLE_BLOWFISH
disabled (it is not explicitly selected by this package).
Our initial fix was to simply select
BR2_PACKAGE_LIBOPENSSL_ENABLE_BLOWFISH, but when investigating when
EVP_bf_cbc() started being used in OpenSC, we discovered it has been
in use for a while... but in code that kept being disabled from
version to version as it was broken (upstream bug
https://github.com/OpenSC/OpenSC/issues/1796), but it was apparently
forgotten to be disabled again for 0.26 (the issue is still
open). Therefore, we opted to continue disabling this known broken
part of the code, and submit an upstream PR for that
https://github.com/OpenSC/OpenSC/pull/3303, which ultimately will
clarify what is the right fix.
In the mean time, this allows to fix the build issue.
Fixes:
http://autobuild.buildroot.net/results/ca51b3e8e3ac83e2a69814caa84d9862385b956f/
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
qt6quicktimeline has qt6declative in its dependencies, but does not
select BR2_PACKAGE_QT6DECLARATIVE. It does select
BR2_PACKAGE_QT6DECLARATIVE_QUICK, but this option "depends on"
BR2_PACKAGE_QT6DECLARATIVE, and does not select it.
Fixes:
WARNING: unmet direct dependencies detected for BR2_PACKAGE_QT6DECLARATIVE_QUICK
Depends on [n]: BR2_PACKAGE_QT6 [=y] && BR2_PACKAGE_QT6DECLARATIVE [=n]
Selected by [y]:
- BR2_PACKAGE_QT6QUICKTIMELINE [=y] && BR2_PACKAGE_QT6 [=y]
reported by kconfig.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
qt6quick3d has qt6declative in its dependencies, but does not select
BR2_PACKAGE_QT6DECLARATIVE. It does select
BR2_PACKAGE_QT6DECLARATIVE_QUICK, but this option "depends on"
BR2_PACKAGE_QT6DECLARATIVE, and does not select it.
Fixes:
WARNING: unmet direct dependencies detected for BR2_PACKAGE_QT6DECLARATIVE_QUICK
Depends on [n]: BR2_PACKAGE_QT6 [=y] && BR2_PACKAGE_QT6DECLARATIVE [=n]
Selected by [y]:
- BR2_PACKAGE_QT6QUICK3D [=y] && BR2_PACKAGE_QT6 [=y]
reported by kconfig.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since commit 9c0c7846cd (support/dependencies: don't check for python
on the host), we no longer check for a host python interpreter installed
on the system.
Drop the comment in support/dependencies/check-host-python3.sh, as it is
now confusing.
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The _SUPPORTS option should be blind: it represents a restriction that
we certainly do not want to allow the user to lift by manually enabling
that option...
Fixes: 65f268e21c (package/arm-gnu-toolchain: add
BR2_PACKAGE_HOST_ARM_GNU_TOOLCHAIN_SUPPORTS)
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Bryce Johnson <bryce@redpinelabs.com>
Cc: Suniel Mahesh <sunil@amarulasolutions.com>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Commit ed12e2fbed (package/libvirt: add lxc and qemu options)
introduced the definition of the 'qemu' user when the libvirt daemon
is enabled, but unconditionally uses that user in its permissions
table.
When enabling libvirt without its qemu support, for example with the
commands:
cat <<EOF >.config
BR2_aarch64=y
BR2_PACKAGE_LIBVIRT=y
BR2_ROOTFS_DEVICE_CREATION_DYNAMIC_EUDEV=y
BR2_TARGET_ROOTFS_EXT2=y
BR2_TOOLCHAIN_EXTERNAL=y
EOF
make olddefconfig
make
The build fails with output:
>>> Generating filesystem image rootfs.ext2
...
makedevs: unknown user name: qemu
Move the permissions needing the 'qemu' user under the same condition
the 'qemu' user is defined under. It means that a few permissions
needing root must also be moved, as they belong under a directory
needing the 'qemu' user. It also moves a few qemu-related permissions
introduced in that same commit. The list of qemu permissions is
reordered alphabetically (the others are left unchanged).
Of course, it also requires that the qemu-related directory and symlink
be moved under the same condition as well.
Reported-by: Alessandro <alex@0x65c.net>
Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Jared Bents <jared.bents@rockwellcollins.com>
[Julien: add the commands to reproduce the issue]
Signed-off-by: Julien Olivain <ju.o@free.fr>
It is very common to use the output of get-developers to add cc: lines
in the commit log.
Add an option so that get-developers reports Cc: lines ready to be
pasted in a commit log. That new option behaves similarly to the
existing -e option: it only affects the output when parsing a patch.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The list of reported developers is not ordered: that may leave the
impression (when receiving a patch) that a Cc is more important than
another, by virtue of being earlier in the list.
Also, the ordering changes on every call.
Report the developers in an alphabetically order, so that there is no
confusion anymore, and so the ordering is reproducible across calls.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
It is very common that get-developers be used with its stdin a pipe from
git-show:
git show |./utils-get-developers -
In this case, the '-' is superfluous: we can very easily deduce that the
user wants to read stdin as the patch.
So, if no other action was requested, and stdin is not a tty, use it as
the source of the patch, and thus '-' is then no longer required.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
parser.error() reports a nice error message, that also displays a short
reminder of the available options.
Adapt the test-suite accordingly: previously, the error string was an
exact string in the stdout list, while it now is a substring in one of
the strings in stderr. The exit code changes, too.
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Offloading parser.parse_args() to a helper function does not bring much,
if at all; it even is restrictive: indeed, we can't use parser.error()
to report errors and thus have to resort to a canned print+return
sequence...
Signed-off-by: Yann E. MORIN <yann.morin@orange.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
The variable <pkg>_LINUX_CONFIG_FIXUPS defined in the
cryptodev-linux... has no effect. Indeed, the variable is only named
CRYPTODEV_LINUX_CONFIG_FIXUPS.
But the variable name being <pkg>_LINUX_CONFIG_FIXUPS and the package
name being CRYPTODEV_LINUX, the correct variable name is
CRYPTODEV_LINUX_LINUX_CONFIG_FIXUPS.
Prior to this commit, a configuration with cryptodev-linux enabled
would result in:
$ make VARS=PACKAGES_LINUX_CONFIG_FIXUPS printvars
$
Aka, empty, while PACKAGES_LINUX_CONFIG_FIXUPS collects in
package/pkg-generic.mk the value of the <pkg>_LINUX_CONFIG_FIXUPS
variables from all enabled packages.
With this patch applied:
$ make VARS=PACKAGES_LINUX_CONFIG_FIXUPS printvars
PACKAGES_LINUX_CONFIG_FIXUPS= @if ! grep -q '^CONFIG_CRYPTO=[my]' /; then /usr/bin/sed -i -e '/^\(# \)\?CONFIG_CRYPTO\>/d' / && echo 'CONFIG_CRYPTO=y' >> /; fi
@if ! grep -q '^CONFIG_CRYPTO_USER_API_AEAD=[my]' /; then /usr/bin/sed -i -e '/^\(# \)\?CONFIG_CRYPTO_USER_API_AEAD\>/d' / && echo 'CONFIG_CRYPTO_USER_API_AEAD=y' >> /; fi
$
As one would expect.
Fixes: 4b12336d1f ("package/cryptodev-linux: needs CONFIG_CRYPTO_USER_API_AEAD")
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Remove the download fix introduced by [1] now the edk2 git repository
has been fixed in version stable202411 [2].
[1] 47fc9e5509.
[2] 95d8a1c255
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since edk2-202408, ATF >= 2.12 is required for qemu aarch64 sbsa target
to support CPU topology information provided by edk2 firmware [1].
If ATF <= 2.11 is used, the boot will hand with the following error:
sbsa_sip_smc_handler: unhandled SMC (0xc20000ca) (function id: 202)
In order to update edk2 package to a newer version, update ATF version
used by qemu_aarch64_sbsa_defconfig to 2.12.
[1] c891b4d835
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since edk2-202408, ATF >= 2.12 is required for qemu aarch64 sbsa target
to support CPU topology information provided by edk2 firmware [1].
If ATF <= 2.11 is used, the boot will hand with the following error:
sbsa_sip_smc_handler: unhandled SMC (0xc20000ca) (function id: 202)
In order to update edk2 package to a newer version, update ATF version
used by TestEdk2 to 2.12.
[1] c891b4d835
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since edk2-202408, ATF >= 2.12 is required for qemu aarch64 sbsa target
to support CPU topology information provided by edk2 firmware [1].
If ATF <= 2.11 is used, the boot will hand with the following error:
sbsa_sip_smc_handler: unhandled SMC (0xc20000ca) (function id: 202)
In order to update edk2 package to a newer version, update ATF version
used by TestFwts to 2.12.
[1] c891b4d835
Cc: Julien Olivain <ju.o@free.fr>
Signed-off-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since the bump of xxhash from 0.8.2 to 0.8.3 in Buildroot commit
4b8f2f3a5e, the host of host-php fails
when host-xxhash has been built before.
This is due to the fact that host-xxhash installs
HOST_DIR/include/xxhash.h, and the PHP code has an internal xxhash.h,
but due to the order of CFLAGS, the one from HOST_DIR/include/xxhash.h
gets included instead of the internal xxhash.h from PHP.
We initially tried to fix the order of CFLAGS, but that has proven to
be challenging within the complicated PHP build system. Then, we opted
for renaming the internal xxhash.h header of PHP, which we had
implemented and were about to submit upstream. But while rebasing this
fix on PHP's master, it turned out that PHP has an upstream commit
"Sync headers installation", merged for PHP 8.4.0, which has the side
effect of avoiding the issue. This commit moves xxhash.h into a
subfolder, so it gets included as #include "xxhash/xxhash.h" now,
which avoids the conflict with HOST_DIR/include/xxhash.h. Maybe not
the best fix, but it does fix the issue and it's upstream.
Since this patch is tweaking the PHP build system, we need to do the
equivalent of "autoreconf" but in the PHP world. It was already done
for the target package, we simply add the same logic for the host
package.
Fixes:
http://autobuild.buildroot.net/results/583c054a564f90b8c884565070c5e17b43b89674/
Cc: James Hilliard <james.hilliard1@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
Since the bump of procps-ng to version 4.0.5 in Buildroot commit
f7fe84508a, the build of igt-gpu-tools
fails with:
../lib/igt_aux.c:1358:78: error: macro "PIDS_VAL" passed 4 arguments, but takes just 3
Indeed, the API of procps-ng has slightly changed between 4.0.4 and
4.0.5. Since it looks strange as those are minor releases the API
breakage has been reported to upstream procps-ng.
In the mean time, we still need to fix igt-gpu-tools, which the
patch added in this commit does. This patch has been submitted to
upstream igt-gpu-tools.
There are no autobuilder failures at this point, as the update to
procps-ng is quite recent, and igt-gpu-tools needing quite a lot of
dependencies, the build has a high chance of failing elsewhere
previously. However, the build issue is reproducible with the
following minimal defconfig:
BR2_arm=y
BR2_cortex_a9=y
BR2_ARM_ENABLE_VFP=y
BR2_TOOLCHAIN_EXTERNAL=y
BR2_TOOLCHAIN_EXTERNAL_BOOTLIN=y
BR2_INIT_NONE=y
BR2_ROOTFS_DEVICE_CREATION_DYNAMIC_EUDEV=y
BR2_SYSTEM_BIN_SH_NONE=y
# BR2_PACKAGE_BUSYBOX is not set
BR2_PACKAGE_IGT_GPU_TOOLS=y
# BR2_TARGET_ROOTFS_TAR is not set
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Julien Olivain <ju.o@free.fr>
