Joshua Watt [Fri, 12 Oct 2018 16:24:09 +0000 (11:24 -0500)]
initscripts: populate-volatiles: Speed up processing
Checking the requirements for each volatiles file in the
populate-volatiles script can be very slow when there are a large number
of volatiles files, easily consuming over 80% of the processing time.
These checks don't usually uncover any problems so concatenate all the
volatiles files together and process them as one large file for a "fast
path" option. This ensures that the penalty for checking the
requirements is only incurred once. In the event that checking the
requirements for the unified file fails, fall back to the slow process
of checking each one individually so that the offending one can be
skipped.
The core file is handled separately because it is responsible for
creating the temp directory used by check_requirements and thus must
always run first and without having its requirements checked.
[YOCTO #12949]
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This ensures that the nativesdk python functions correctly without needing to
set PYTHONHOME in the sdk environment setup script.
]
it's also needed for python3.
Signed-off-by: Cheuk Wing Leung <cwleung@kth.se> Signed-off-by: Ming Liu <liu.ming50@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ross Burton [Fri, 12 Oct 2018 13:44:03 +0000 (14:44 +0100)]
python: don't use runtime checks to identify float endianism
Python uses AC_RUN_IFELSE to determine the byte order for floats and doubles,
and falls back onto "I don't know" if it can't run code. This results in
crippled floating point numbers in Python, and the regression tests fail.
Instead of running code, take a macro from autoconf-archive which compiles C
with a special double in which has an ASCII representation, and then greps the
binary to identify the format.
This is essentially a backport of the Python 3 patch in oe-core 1781b87.
Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Hongxu Jia [Fri, 12 Oct 2018 16:28:48 +0000 (09:28 -0700)]
openssl: use deterministic perl Text::Template module bundled by openssl source
1. The building openssl requires to install perl Text::Template module(>=1.46),
but Text::Template is a non core Perl module, openssl chooses to bundle
Text::Template 1.46 into the source, for convenience.
https://github.com/openssl/openssl/commit/8ff2af548303d311ce3591406111f77862875a60
2. While Text::Template < 1.46, the produced build files are gravely faulty.
https://github.com/openssl/openssl/pull/6682
3. If host has installed Text::Template < 1.46 (such as CentOS-7.5 has Text::
Template 1.45). The mismatched old module was used although the right one in
openssl source.
So set PERL5LIB to use deterministic perl Text::Template module bundled
by openssl source and ignore the one of host
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rui Wang [Fri, 12 Oct 2018 02:51:36 +0000 (10:51 +0800)]
unfs3: Fix utime for symlink in attr.c
unfs3 has an old defect that it can not change the timestamps of a
symlink file because it only uses utime(), which will follow the
symlink. This will not cause an error if the symlink points to an
existent file but can be triggered with rpm workflows.
Making unfs3 support lutimes() addresses this problem.
Signed-off-by: Rui Wang <rui.wang@windriver.com> Signed-off-by: Joe Slater <joe.slater@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
nativesdk-flex is needed for kernel development inside traidtional
SDK environment, more particularly, `make scripts' under /usr/src/kernel.
So add it to nativesdk-packagegroup-sdk-host.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Chen Qi [Fri, 12 Oct 2018 02:24:51 +0000 (10:24 +0800)]
packagegroup-core-lsb/-full-cmdline: add bzip2
We are having the following error when using 'tar' from tar
recipe to decompress .tar.bz2 files.
tar (child): bzip2: Cannot exec: No such file or directory
tar (child): Error is not recoverable: exiting now
tar: Child returned status 2
tar: Error is not recoverable: exiting now
The tar package is introduced by these two packagegroups into image.
>From the README file from tar's source codes:
"""
** gzip and bzip2.
GNU tar uses the gzip and bzip2 programs to read and write compressed
archives. If you don't have these programs already, you need to
install them.
"""
So we'd better cluster gzip and bzip2 with tar. These two packagegroups
already get 'gzip', so we also add 'bzip2'.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changqing Li [Fri, 12 Oct 2018 02:21:12 +0000 (10:21 +0800)]
e2fsprogs: fix test f_detect_junk print error msg
This test need to check config.h, so install it
to fix below err msg:
grep: ../lib/config.h: No such file or directory
./test/f_detect_junk/script: line 3: [: : integer expression expected
f_detect_junk: detect non-fs file data: ok
Signed-off-by: Changqing Li <changqing.li@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Geoff Parker [Thu, 11 Oct 2018 16:31:26 +0000 (09:31 -0700)]
wic: make engine.py:get_partitions() resilient to parted/dmidecode stderr output
Running wic commands on Debian 10 systems fail in
scripts/lib/wic/engine.py:get_partitions() due to new stderr output captured
when trying to parse the output from /sbin/parted as a non-root user.
The parted command calls the dmidecode utility, which produces this error
as a non-root user:
/sys/firmware/dmi/tables/smbios_entry_point: Permission denied
/dev/mem: Permission denied
scripts/lib/wic/engine.py:get_partitions() calls misc.py:exec_cmd(),
a subprocess wrapper which returns a combined stderr and sdtdout.
These messages to stderr confuse the partition table parser in
get_partitions().
This patch has the partition table parser ignore lines before the expected
"BYT;" header string.
Running wic in Debian 9 does not have this issue.
Signed-off-by: Geoff Parker <geoffhp@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Joshua Watt [Thu, 11 Oct 2018 18:09:04 +0000 (13:09 -0500)]
initscripts: read-only-rootfs-hook: Use overlay if available
Copying files from the read-only /var/lib to tmpfs can be slow and waste
memory. If the kernel supports the overlay file system, use it to mount
a writable tmpfs on top of the read-only /var/lib and avoid the file
copy.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Mark Asselstine [Wed, 10 Oct 2018 19:35:44 +0000 (15:35 -0400)]
meson.bbclass: Fix build issues with /tmp mounted with noexec
Since commit d297f7ebf3f6 [fribidi: use Meson instead of autotools]
build failures have been observed with this package. The immediate
issue was related to improperly named #defines per
https://github.com/fribidi/fribidi/commit/46f52d588ab5, however, the
root cause was FRIBIDI_SIZEOF_INT getting a value of "-1".
After searching the meson logs the following was found:
Could not run: /tmp/tmp2fxe6ha1/output.exe (error: [Errno 13] Permission denied)
Checking for size of "int": -1
Which pointed to the real root cause being /tmp mounted with noexec, a
common configuration on Redhat and other distros. This issues has been
raised in the meson community:
https://github.com/mesonbuild/meson/issues/2972 but is yet to be
addressed.
Using the discussion from issue#2972 and the fact that the underlying
code makes use of python 'tempfile' we can simply create a 'tmp'
directory and make use of TMPDIR to avoid this issue.
Signed-off-by: Mark Asselstine <mark.asselstine@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Changqing Li [Thu, 11 Oct 2018 10:16:18 +0000 (18:16 +0800)]
e2fsprogs: fix test i_bitmaps failed
without diffutils, diff from busybox will be used, diff of
busybox cause diff result different with expect file under test.
The diff result generated by 2 /usr/bin/diff is different.
change to use diff under diffutils to fix it
Signed-off-by: Changqing Li <changqing.li@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Chen Qi [Thu, 11 Oct 2018 09:04:08 +0000 (17:04 +0800)]
util-linux: fix alternatives setting for unshare
The alternative setting for unshare belongs to the util-linux-unshare
package instead of the util-linux package. Fix this problem to avoid
a dead symlink /usr/bin/unshare -> /usr/bin/unshare.util-linux on target.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 11 Oct 2018 04:06:53 +0000 (12:06 +0800)]
prelink: Fix Segmentation fault error when prelink qemu
The info->resolvetls might be NULL according to src/prelink.c:
[snip]
/* Dynamic linker does not depend on any other library,
all symbols resolve to themselves with the exception
of SHN_UNDEF symbols which resolve to 0. */
if (info->symtab[r_sym].st_shndx == SHN_UNDEF)
{
info->resolveent = NULL;
info->resolvetls = NULL;
[snip]
So we must check it before use its members, otherwise, there might be
Segmentation fault error.
Yi Zhao [Thu, 11 Oct 2018 00:30:15 +0000 (08:30 +0800)]
ltp: use 'ping -6' if ping6 is not avaliable in netns_helper.sh
The iputils-ping6 was dropped since the 'ping6' command had been merged
into ping command. Backport patch from upstream to let both 'ping6' and
'ping -6' work.
Signed-off-by: Yi Zhao <yi.zhao@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fix multilib isntall file conflicts for gobject-introspection.
* use multilib_script.bbclass to handle ${bindir}/g-ir-annotation-tool
and ${bindir}/g-ir-scanner
* add configure option to install .gir files to an alternative path and
only set it when multilib is enabled
Signed-off-by: Kai Kang <kai.kang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 10 Oct 2018 13:48:50 +0000 (14:48 +0100)]
testimage: Ensure opkg/apt test suites are run
Currently only the dnf package manager tests are automatically added to the
list of tests to run. Improve the code to handle automatic addition of
the apt and opkg tests too.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Yeoh Ee Peng [Thu, 11 Oct 2018 09:39:01 +0000 (17:39 +0800)]
oeqa/manual: transfer manual test cases from testopia to oeqa
As part of the solution to replace Testopia, manual test cases
need to be migrated to OEQA. These manual test case json files
will serve two use cases.
Use case#1: as input to the future commandline-based manual test
runner script, where this script will display actions and expected
result information in user friendly text, then it will capture
user input for test result and log, finally it will write test
result and log into existing standardize test result json format
from OEQA framework for automated tests.
Use case#2: QA will open and read these json file manually for
planning manual test execution. Any reader interested in
understanding manual test cases will open and read these files.
Signed-off-by: Yeoh Ee Peng <ee.peng.yeoh@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Append '\n' to the non-empty formatted string before return. If you
write it to the (manifest) file, it will ensure file ends with a newline.
Many GNU utilities have problems processing the last line of a file
if it is not '\n' terminated. E.g. if the last line is not terminated
by a newline character, then "read" will read it but return false,
leaving the broken partial line in the read variable(s).
It can also break or adversely affect some text processing tools,
that operate on the file.
Signed-off-by: grygorii tertychnyi <gtertych@cisco.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Fetch the test tar ball to a subdirectory in ${S}. This avoids the
following error after having done `devtool modify libxml2`:
| DEBUG: Executing shell function do_configure
| find: ‘.../build/tmp/work/mips32r2el-nf-poky-linux/libxml2/2.9.4-r0/xmlconf/’: No such file or directory
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Make the unpack task do nothing if externalsrc is in use. This avoids
the following error after having done `devtool modify lsof`:
ERROR: lsof-4.91-r0 do_unpack: Unpack failure for URL:
'file://.../builds/qemux86-64/tmp/work/core2-64-poky-linux/lsof/4.91-r0/lsof_4.91/lsof_4.91_src.tar'.
Unpack command PATH="..." tar x --no-same-owner -f
.../builds/qemux86-64/tmp/work/core2-64-poky-linux/lsof/4.91-r0/lsof_4.91/lsof_4.91_src.tar
failed with return value 2
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Hongxu Jia [Wed, 10 Oct 2018 07:16:48 +0000 (15:16 +0800)]
webkitgtk/epiphany: add opengl to REQUIRED_DISTRO_FEATURES
They can't be built without opengl in DISTRO_FEATURES.
[snip]
|webkitgtk-2.20.3/Source/WebCore/platform/graphics/OpenGLShims.h:23:10:
fatal error: GL/gl.h: No such file or directory
[snip]
Some sort of GL is needed yes. We indicate that with opengl loosely in OE
The epiphany depends on webkitgtk, add opengl to REQUIRED_DISTRO_FEATURES also.
Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Davis, Michael [Mon, 8 Oct 2018 19:55:22 +0000 (19:55 +0000)]
wic: Support for changing the imager.
Signed-off-by: Michael Davis <michael.davis@essvote.com> Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When running `bitbake -c populate_sdk_ext ...` for an image
recipe inheriting populate_sdk_qt5, `nativesdk-qemu-helper` is
deployed via `nativesdk-packagegroup-sdk-host`, which already
contains some of the tools (by name: "oe-find-native-sysroot
runqemu*").
To avoid error like:
Configuring nativesdk-packagegroup-qt5-toolchain-host.
Jacob Kroon [Sun, 7 Oct 2018 17:11:21 +0000 (19:11 +0200)]
gptfdisk: Use PACKAGECONFIG to control ncurses and popt dependencies
gptfdisk uses a simple Makefile with no automatic detection of
external library support. Use PACKAGECONFIG_CONFARGS and pass
the make targets in EXTRA_OEMAKE.
Signed-off-by: Jacob Kroon <jacob.kroon@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
He Zhe [Tue, 9 Oct 2018 11:44:31 +0000 (19:44 +0800)]
lttng-modules: Fix build failure by syncing up with kernel 4.19
linux-yocto-dev has been upreved to v4.19. This patch backports a patch from
upstream to make lttng-modules sync up with linux-yocto-dev and later
linux-yocto, and fix the following do_compile error.
| tmp/work/qemux86_64-poky-linux/lttng-modules/2.10.7-r0/lttng-modules-2.10.7/probes/../probes/lttng-tracepoint-event-impl.h:143:6: error: conflict
ing types for 'trace_sock_exceed_buf_limit'
| void trace_##_name(_proto);
| ^~~~~~
Signed-off-by: He Zhe <zhe.he@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Anuj Mittal [Fri, 5 Oct 2018 11:24:01 +0000 (19:24 +0800)]
strace: fix ptests
* Let automake use parallel test harness which ensures that appropriate
environment variables are actually used.
* Copy generated config.h as part of ptest package and define a variable
in tests Makefile to control the name of target that we'd like to
invoke for tests instead of relying on default value.
* configure relies on tools that are not available in release tarballs
resulting in warnings and failed tests. Get the required information
from available files instead.
* Reduces the count of failing tests to 18 from 35. The rest of
tests fail or get skipped mostly because of missing syscalls. Number of
tests getting skipped can probably be reduced further.
Fixes [YOCTO #12948]
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
util-linux: Use update-alternatives for fsfreeze, nsenter
busybox may also provide a fsfreeze and nsenter implementation,
so use u-a to allow installing them side by side.
If one installs both, busybox and util-linux, one gets the following error
(seen on a populate-sdk run):
| ERROR: Postinstall scriptlets of ['busybox'] have failed. If the intention is to defer them to first boot,
| then please place them into pkg_postinst_ontarget_${PN} ().
| Deferring to first boot via 'exit 1' is no longer supported.
And the coresponding do_populate_sdk extract:
| ...
| update-alternatives: Error: not linking ...sdk/image/usr/local/oecore-x86_64/sysroots/armv7at2hf-neon-angstrom-linux-gnueabi/usr/sbin/fsfreeze
| to /bin/busybox.nosuid /bin/busybox.nosuid since ...sdk/image/usr/local/oecore-x86_64/sysroots/armv7at2hf-neon-angstrom-linux-gnueabi/usr/sbin/fsfreeze
| exists and is not a link
| ...
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
busybox may also provide a partprobe implementation, so use u-a to allow
installing them side by side.
If one installs both, busybox and parted, one gets the following error:
| ERROR: image do_rootfs: Postinstall scriptlets of ['busybox'] have failed. If the intention is to defer them to first boot,
| then please place them into pkg_postinst_ontarget_${PN} ().
| Deferring to first boot via 'exit 1' is no longer supported.
And the coresponding log.do_rootfs extract:
| ...
| update-alternatives: Error: not linking ...image/rootfs/usr/sbin/partprobe to /bin/busybox.nosuid since ...image/rootfs/usr/sbin/partprobe exists and is not a link
| ...
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Mike Crowe [Fri, 5 Oct 2018 13:49:13 +0000 (14:49 +0100)]
gcc: Remove unnecessary --with-mpfr and --with-mpc when cross compiling
Passing --with-mpfr and --with-mpc when compiling the cross compiler
appears to be at best unnecessary, and at worst can cause build failures.
Firstly, the paths passed in gcc-cross-canadian are using the undefined
${layout_exec_prefix}. This results in configure passing
-I${STAGING_DIR_HOST}/include twice to the compiler when it's doing its
test builds. This is mostly pointless since that directory doesn't exist
with the default oe-core configuration - the correct path would be
-I${STAGING_DIR_HOST}/usr/include.
The path for mpfr passed in gcc-cross is correct, but unnecessary since it
is just the sysroot default.
I've gone back through the history, and it seems that these lines (or
similar ones) were originally added way back in 8800d8be25295dd7c7d84dde62c3be4df8e43346 for GCC 4.1.1 in 2006! I asked[1]
if anyone knew why this was necessary but received no response, so I can
only assume that no-one knows.
I've successfully built for various targets with this patch applied and
observed no problems.
libgcrypt: Add the LICENSES file to LIC_FILES_CHKSUM
The LICENSES file contains references to source files with other
licenses than GPL/LGPL that stipulate that they shall be mentioned in
any documentation accompanying a product including this library.
License-Update: Add missing LICENSES file Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
glib-2.0: Include the complete license information for pcre
For some reason, the copyright part was left out of the license
information for pcre in LIC_FILES_CHKSUM, preventing it from being
used in, e.g., documentation to satisfy the requirements of the
license.
Also drop pcre/COPYING from LIC_FILES_CHKSUM as it does not really
contain any licensing information, it only refers to a LICENSE file
that does not exist.
License-Update: Include the complete license information for pcre Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For some reason, the copyright part was left out of the license
information included in LIC_FILES_CHKSUM, preventing it from being
used in, e.g., documentation to satisfy the requirements of the
license.
License-Update: Include the complete license information Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For some reason, the copyright part was left out of the license
information included in LIC_FILES_CHKSUM, preventing it from being
used in, e.g., documentation to satisfy the requirements of the
license.
License-Update: Include the complete license information Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For nativesdk, symlinks created by do_install contain ${SDKPATHNATIVE}.
This is a problem with SDKs that include nativesdk-binutils and nativesdk-gcc.
When extracting such an SDK dangling symlinks are created, because
relocate_sdk.py does not adjust symlinks. As a result gcc ends up calling the
host binutils.
Use the os.path.relpath function to obtain shorter relative paths, which do not
contain ${SDKPATHNATIVE}.
Signed-off-by: Achille Fouilleul <achille.fouilleul@gadz.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
libusb: ptest: don't skip debug output and fix failures processing
Current run-ptest script prints nothing, when stress tests fail.
Fix it in new implementation, discarding external dependency on sed.
Also leave in place all stress output, just add standard ptest result.
Ola x Nilsson [Thu, 4 Oct 2018 10:43:28 +0000 (12:43 +0200)]
externalsrc.bbclass: Set BB_DONT_CACHE for non-target recipes
BB_DONT_CACHE was not set for non-virtual recipes where PN != BPN, such
as quilt-native. Recipes that do not set BBCLASSEXTEND should always
have BB_DONT_CACHE set by externalsrc.
Signed-off-by: Ola x Nilsson <olani@axis.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ross Burton [Thu, 4 Oct 2018 09:20:20 +0000 (10:20 +0100)]
libxml2: refresh CVE-2017-8872
The patch associated with the CVE-2017-8872 report was never merged into
libxml2, but a slightly different patch for the same problem was. Cherry-pick
that as a backport, which also fixes the failing test suite.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Anuj Mittal [Wed, 3 Oct 2018 10:27:50 +0000 (18:27 +0800)]
qemux86-directdisk: remove mem= parameter
Remove usage of a specific amount of memory and let it be controlled by
users. This was the default behaviour before it was changed by commit 3b79d9a78 that switched the wks file to be used for qemux86.
Also fixes the bitbake parsing issues seen because of memory starvation
using build appliance images.
Fixes [YOCTO #12894]
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Aníbal Limón [Tue, 2 Oct 2018 21:43:23 +0000 (16:43 -0500)]
scripts/yocto-check-layer: Don't abort when layer fails to get signatures
When execute in multiple layers like meta-openembedded the execution is
aborted when some sublayer fails to get world signatures, so mark it as
an error and continue the execution in the remaining sublayers.
Signed-off-by: Aníbal Limón <anibal.limon@linaro.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
prior to that commit, musl was loading it and silently letting
subsequent TLS accesses via the miscompiled code clobber memory that
didn't belong to them
This was wrong behavior and it relied on additional space reserved by
libc in TLS space to adjust fo such broken libs, but it also fails
with glibc if the reserved space was already used up
Right fix is that mesa should be patched to remove all the
initial-exec hacks and use real TLS, and -mtls-dialect=gnu2 (TLSDESC)
should be used on archs it's supported on (i386, x86_64, and aarch64)
to make up for the lost performance, but mesa hardcodes the initial-exec,
so there must be a reason that probably is better known to mesa devs.
but we 'fixed' it for musl by adding --disable-glx-tls for mesa in OE,
which uses pthread_getspecific instead and makes is lot slower.
this caused additional problems with security flags on, it get textrels
in .text segment. Therefore this is 'second fix' to get us through this
warning.
Cause is some unknown part of mesa's x86 assembly code is broken by
readonly text segments
[ YOCTO #12918 ]
Signed-off-by: Khem Raj <raj.khem@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
