Age | Commit message (Collapse) | Author |
|
This recipe is intended to be run manually or via oeqa.
[YOCTO #13737]
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This task is not intended to be run as this recipe does not produce any
packages or install any sysroot content. Additionally running this task
behaves differently from a normal glibc build, and can result in issues.
[YOCTO #13737]
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Two things: The -Wl,-rpath-link comment is stale (due to per-recipe
staging), so we no longer need to set our own LDFLAGS to avoid
-Wl,-rpath-link being in there. Second, overriding LDFLAGS should be
done at the recipe level so "bitbake -e" can show what is going on.
Otherwise debugging why one's global LDFLAGS tweaks are being ignored
is needlessly painful.
So pull out the LDFLAGS setting from do_compile, and change it to an
append instead of assignment.
For the benefit of future git blame: the -fuse-ld=bfd setting was
added by ac64c3b96b (glibc: always use bfd linker).
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There's only one user of libc-common now that we don't ship both glibc and
eglibc, so copy the contents directly into the recipe.
[ YOCTO #12135 ]
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If documemtation generation is disabled, the recipe throws warnings about
unpackaged files. Avoid this.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This helps clang to do a better job with fortify on but it is better
suited for clang layer
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Drop ppc 8xx clear cache optimization, which we have been carrying
through eglibc days, it has been redone in glibc now
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
License-Update: Prefer https to http for gnu.org and fsf.org URLs [1]
[1] https://sourceware.org/git/?p=glibc.git;a=commit;h=5a82c74822d3272df2f5929133680478c0cfb4bd
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Patches are currently, crossing across repo boundaries, between glibc
and localedef, therefore it is better to divide the patches
accordiningly, with this patchset makery patch is spun out since that
applies to localedef alone.
There are no other code changes
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Cc: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
ldconfig should be prior to glibc-utils in PACKAGES variable, or else
ldconfig binary would not be split to its own package, hence will lead
to runtime issues for the packages that depending on ldconfig, like
systemd.
Signed-off-by: Ming Liu <liu.ming50@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Remove the generation of the testglibc script which could be used to run
the glibc test suite with a remote target. The same functionality can
now be achieved with the 'do_check' task of glibc-testsuite or with
oe-selftest (for automation of execution against qemu-user/qemu-system
targets).
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Backport from 2.30 stable branch and drop NEWS section.
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
rename
Besides checking DISTRO_FEATURES for required or conflicting features,
being able to check MACHINE_FEATURES and/or COMBINED_FEATURES may also
be useful at times.
Temporarily support the old class name with a warning about future
deprecation.
Signed-off-by: Denys Dmytriyenko <denys@ti.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Only recommend its installation, if it's enabled in distro features.
Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The removal of the supposedly empty /etc when ldconfig is not in
DISTRO_FEATURES seems to be a remnant from a long time ago when nothing
else was installed in /etc. However, that is no longer the case as,
e.g., nscd.conf is always installed to /etc now.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
nativesdk-glibc should be spared of recompile when the distro features
are changed e.g. ldconfig is not in
DISTRO_FEATURES, this happens when sdk with musl and another one with
glibc is built
Fixes
Variable do_install value changed:
...
-DISTRO_FEATURES{ldconfig} = Set
+DISTRO_FEATURES{ldconfig} = Unset
Signed-off-by: Khem Raj <raj.khem@gmail.com>
s
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
To prevent issues with parsing or dependencies, limit this recipe to use
only when the libc is glibc (and libc-locale is glibc-locale).
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
These recipes don't need to generate packages but did contain the packaging tasks which
would be triggered by a world build. They showed warnings or errors. Simplest fix
is to remove the unneeded tasks with the nopackages class.
Also don't attempt stash_locales task
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
A recipe needs to be created for the test suite due to the dependency
chain between libgcc -> glibc -> libgcc-initial, and the requirements of
the test suite to have libgcc for compilation and execution.
The glibc test suite does not use dejagnu like the gcc test suites do.
Instead a test wrapper script is used along with the assumed dependency
of having the same filesystem available on build host and target. For
qemu linux-user the same filesystem is inherently available, for remote
targets NFS is used. Separate test wrapper scripts are created for qemu
linux-user or ssh targets, with the same TOOLCHAIN_TEST_* variables used for
configuration.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Remove remnants of the glibc-initial recipe.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The hard link resolver that is built into localedef cannot be run in
parallel. It will search sibling directories (which are be processed
in parallel) and perform a creation of a .tmp file and remove the
original and move the .tmp file in. The problem is that if a probe
occurs a hard link can be requested to the file that is being removed.
This will lead to a stray copy or potentially, on a loaded system
cause race condition which pseudo cannot deal with, where it is left
with a hard link request to a file that no longer exists. In this
situation psuedo will inherit the permissions of what ever the target
inode had to offer.
In short, there are two problems:
1) You will be left with stray copies when using the hard link
resolution that is built in while running in parallel with
localedef.
2) When running under pseudo the possibility exists for uid/gid
leakage when the source file is removed before the hard link can
be completed.
The solution is to call localedef with --no-hard-links and separately
process the hardlinks at a later point. To do this requires the
inclusion of the hardlink utility found in modern versions of
util-linux. Most host systems do not have this, so it will be
included with the cross-localedef binary.
There are two patches here. The first imports the raw version of
hardlink.c and a couple of header files directly from util-linux. The
second patch applies the fix-ups to make it compile, along with a
change to recipe to package the new binary.
[YOCTO #11299]
[YOCTO #12434]
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Without linux-libc-headers-dev being added to the libc6-dev as a RDEPENDS,
the system may fail to install the necessary libc headers.
This can happen when NO_RECOMMENDATIONS = "1" is defined.
During the 'testsdk' this results in failures that look like:
fatal error: linux/errno.h: No such file or directory
# include <linux/errno.h>
^~~~~~~~~~~~~~~
This also matches the behavior of musl, which does not suffer from this
problem.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
- Drop backported patches
- Move common pieces between cross-localedef and glibc into a common file
- Move latest checksums to glibc-common.inc and remove duplicates from glibc recipe
- Detailed release notes [1]
[1] https://sourceware.org/ml/libc-alpha/2019-08/msg00029.html
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
See:
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=34141
https://www.securityfocus.com/bid/107160
Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
As glibc will be scanned for CVEs, we don't need to scan glibc-locale,
glibc-mtrace, and glibc-scripts which are all separate recipes for technical
reasons.
Exclude the recipes by setting CVE_PRODUCT in the recipe, instead of using the
global whitelist.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Using bitbake variable syntax (i.e., ${FOO}) for shell variables is
bad practice. First of all it is confusing, but more importantly it
can lead to weird problems if someone actually defines a bitbake
variable with the same name as the shell variable.
Also correct the indentation in stash_locale_cleanup().
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When using sstate, or performing an incremental build any change to
the do_stash_locale() will cause a build failure because
do_stash_locale() was destroying the results obtained from the
do_install() with several mv operations. A recent change to
do_stash_locale() for a different problem illustrated a number of
build failures for users in the community.
To fix the problem, do_stash_locale() must use copy operations instead
of the mv operations. Because this is changed to a copy, the sysroot
and package stage need to remove the files that would have been
previously removed. The correct "fixup" code to deal with the removal
already existed in the previous do_poststash_install_cleanup(). All
that needed change was the path to where to remove the files
from the sysroot and package stages.
In order to force a re-compilation of glibc some unused white space
was removed from do_compile() for glibc. I could not find any other
way around this and we don't want to have all the community folks to
have another iteration where they have to remove their tmp directories
or purge some portion of the sstate. It also makes this change
bisectable. If the change to the glibc is not included, it will fail
with the following message:
=====
| DEBUG: Executing shell function do_prep_locale_tree
| tar: i18n: Cannot stat: No such file or directory
| tar: Exiting with failure status due to previous errors
| gzip: /poky/build/tmp/work/core2-64-poky-linux/glibc-locale/2.29-r0/locale-tree//usr/share/i18n/charmaps/*gz.gz: No such file or directory
=====
After this one time change I tested changing only the
do_stash_locale() function and it now works well because it is
deterministically operating off the sstate data or a local build.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The do_stash_locale was not working consistently across the 4 build
configurations and the multilib, usrmerge configuration would fail
entirely with the obscure message:
| DEBUG: Executing shell function do_prep_locale_tree
| tar: i18n: Cannot stat: No such file or directory
| tar: Exiting with failure status due to previous errors
| gzip: /poky/build/tmp/work/core2-64-poky-linux/glibc-locale/2.29-r0/locale-tree//usr/share/i18n/charmaps/*gz.gz: No such file or directory
| WARNING: /poky/build/tmp/work/core2-64-poky-linux/glibc-locale/2.29-r0/temp/run.do_prep_locale_tree.124690:1 exit 1 from 'gunzip $i'
Here is the 4 build configurations without the patch applied:
A) x86-64 no multilibs, no usrmerge
find ./tmp/work/*/glibc/2.29-r0/stashed-locale -type f |grep -v nscd.service |wc -l
909
B) x86-64 no multilibs, usrmerge
find ./tmp/work/*/glibc/2.29-r0/stashed-locale -type f |grep -v nscd.service |wc -l
909
C) x86-64 multilibs, no usrmerge
find ./tmp/work/*/glibc/2.29-r0/stashed-locale -type f |grep -v nscd.service |wc -l
885
D) x86-64 multilibs, usrmerge
find ./tmp/work/*/glibc/2.29-r0/stashed-locale -type f |grep -v nscd.service |wc -l
864
The issue here is that all the moves should be processed first, then a
copy should be made of the lib directories, but only in the case they
are different when using the usrmerge feature. Even though the build
worked for the multilib configuration without usrmerge, the content
was not the same.
After applying the patch the same number of files are in all the
configurations. The list of files was also diffed, after normalizing
the directory names to ensure all the correct files were copied.
Ultimately there are probably additional files that should be pruned
from what is copied to the stated_locale, but the purpose of this
patch is make it 100% consistent between the build types and fix the
builds.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When the PACKAGE_NO_GCONV is set to 1 an empty directory is left behind from the do_install rule:
=====
ERROR: glibc-locale-2.29-r0 do_package: QA Issue: glibc-locale: Files/directories were installed but not shipped in any package:
/usr/lib
/usr/lib/locale
Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install.
glibc-locale: 2 installed and not shipped files. [installed-vs-shipped]
ERROR: glibc-locale-2.29-r0 do_package: Fatal QA errors found, failing task.
=====
The simple fix is to prune the empty directory.
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The build of glibc fails when you have multilibs enabled + the distro
feature usrmerge. Here is an example configuration:
===
MACHINE = "qemux86-64"
VIRTUAL-RUNTIME_init_manager = "systemd"
DISTRO_FEATURES_append = " systemd "
DISTRO_FEATURES_append += " usrmerge"
require conf/multilib.conf
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
===
This will fail with the following error:
NOTE: Executing SetScene Tasks
NOTE: Executing RunQueue Tasks
ERROR: glibc-2.28-r0 do_poststash_install_cleanup: Function failed: do_poststash_install_cleanup (log file is located at /poky/build/tmp/work/core2-64-poky-linux/glibc/2.28-r0/temp/log.do_poststash_install_cleanup.107893)
ERROR: Logfile of failure stored in: /poky/build/tmp/work/core2-64-poky-linux/glibc/2.28-r0/temp/log.do_poststash_install_cleanup.107893
The fix is to not perform the rmdir check when using the multilib + usr/merge, namely:
if [ "${libdir}" != "${exec_prefix}/lib" ] && [ "${root_prefix}/lib" != "${exec_prefix}/lib" ]; then
This will evaluate as follows (collecting the output from bitbake -e glibc)
* no multilibs no usrmerge
if [ "/usr/lib" != "/usr/lib" ] && [ "/lib" != "/usr/lib" ]; then
* no multilibs yes usrmerge
if [ "/usr/lib" != "/usr/lib" ] && [ "/usr/lib" != "/usr/lib" ]; then
* yes multilibs no usrmerge
if [ "/usr/lib64" != "/usr/lib" ] && [ "/lib" != "/usr/lib" ]; then
* yes multilibs yes user merge
if [ "/usr/lib64" != "/usr/lib" ] && [ "/usr/lib" != "/usr/lib" ]; then
Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The restriction against glibc-locale depending on virtual/libc is
removed now that libc-initial no longer exists.
Adding the DEPENDS on virtual/libc fixes a race (and reproducibility
issue) where the packages created by this recipe would have different
runtime dependencies depending on if libc had generated its packages yet
or not.
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* Work around broken ld-2.29.so when gold is used
causing qemu-arm to segfault during e.g. gobject-introspection
or postinst at do_rootfs time, more details in:
http://lists.openembedded.org/pipermail/openembedded-devel/2019-March/198937.html
https://sourceware.org/bugzilla/show_bug.cgi?id=24148
https://sourceware.org/bugzilla/show_bug.cgi?id=10937
https://sourceware.org/bugzilla/show_bug.cgi?id=18103
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
fix below error:
file /usr/include/bits/procfs-id.h conflicts between attempted installs of lib32-libc6-dev-2.29-r0.armv7vet2hf_vfp and libc6-dev-2.29-r0.aarch64
file /usr/include/bits/procfs.h conflicts between attempted installs of lib32-libc6-dev-2.29-r0.armv7vet2hf_vfp and libc6-dev-2.29-r0.aarch64
file /usr/include/bits/shmlba.h conflicts between attempted installs of lib32-libc6-dev-2.29-r0.armv7vet2hf_vfp and libc6-dev-2.29-r0.aarch64
Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
These patches were applied, hoping that they will eventually be accepted
upstream but they have been rejected, I think its best that they are
dropped so we can avoid novel unintended behaviours that no other
distros will be seeing
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Since compiler does not optimize away a lot of stuff we end up with
Werrors e.g.
./sysdeps/ieee754/flt-32/s_log1pf.c: In function '__log1pf':
../sysdeps/ieee754/flt-32/s_log1pf.c:114:22: error: 'c' may be used uninitialized in this function [-Werror=maybe-uninitialized]
114 | + (k * ln2_lo + c))) - f);
| ~~~~~~~~~~~~^~~~
which otherwise wont happen, so lets build with warnings-as-errors
disabled in debug mode
given we disable werror, now we don't have to restrict user to compile
without -O0
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We have been duplicating few variables in glibc recipes which could
actually be defined once, therefore move them to glibc-common.inc which is
included by all glibc family of recipes
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
After eglibc was merged into glibc, Kconfig support was also dropped so
these libc features therefore are not effective anymore and can be
removed
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The AArch64 little-endian ABI requires that the dynamic loader is always
available at /lib/ld-linux-aarch64.so.1. Similarly, the big-endian ABI
requires that the dynamic loader is always available at
/lib/ld-linux-aarch64_be.so.1.
glibc-package.inc contains code that tries to ensure this, but
unfortunately it is defeated by the combination of multilib and usrmerge
because it does not take into account that /lib is the same as /usr/lib
with usrmerge when it adds the loader path to libc_baselibs and when it
attempts to show that /usr/lib is empty in do_poststash_install_cleanup.
This results in the symlink not being included in the package and a build
failure due to rmdir failing.
Richard Purdie also suggested[1] that ${nonarch_base_libdir} should not be
used as a synonym for /lib in this case.
This hopefully-fixed version always sets ARCH_DYNAMIC_LOADER and then uses
${root_prefix}/lib/${ARCH_DYNAMIC_LOADER} to refer to the dynamic loader
which works with both multilib and usrmerge. Since ARCH_DYNAMIC_LOADER is
only non-empty if the symlink is required, the code to create it can move
to do_install_append. Then do_poststash_install_cleanup needs to be taught
that ${exec_prefix}/lib may not be empty if the dynamic loader symlink is
there.
It appears not to be possible to specify the name of the loader via a
variable with an override, since the _aarch64 override is applied even for
_aarch64-be, so I've set the loader name using ${TARGET_ARCH} instead.
Build-tested and inspected core-image-minimal rootfs with:
* AArch64 no multilib (real loader in correct place)
MACHINE = "qemuarm64"
* AArch64 multilib (symlink in correct place)
MACHINE = "qemuarm64"
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "armv7at-neon"
require conf/multilib.conf
* AArch64 usrmerge (real loader in correct place)
DISTRO_FEATURES += "usrmerge"
MACHINE = "qemuarm64"
* AArch64 multilib usrmerge (symlink in correct place)
DISTRO_FEATURES += "usrmerge"
MACHINE = "qemuarm64"
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "armv7at-neon"
require conf/multilib.conf
* big-endian versions of all of the above by also setting
DEFAULTTUNE = "aarch64_be".
(building glibc only.)
* x86_64 (real loader in /lib as before)[2]
MACHINE = "qemux86"
* x86_64 multilib (real loader in /lib64 as before)
MACHINE="qemux86-64"
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
require conf/multilib.conf
I also tested leaving an unwanted file in ${exec_prefix}/lib for
do_poststash_install_cleanup to detect, and I believe the detection always
worked correctly.
[1] http://lists.openembedded.org/pipermail/openembedded-core/2018-November/276120.html
Signed-off-by: Mike Crowe <mac@mcrowe.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
To avoid conflicts with other forks of glibc remove the diff from the
ChangeLog.
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Backport the CVE patch from the upstream
commit 3f635fb43389b54f682fc9ed2acc0b2aaf4a923d
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This has been a constant source of trouble for build failures due to host-user-contaminated QA
errors of sort
ERROR: QA Issue: glibc-locale: /glibc-binary-localedata-ca-es+valencia/usr/lib/locale/ca_ES@valencia/LC_MONETARY is owned by uid 3004, which is the same as the user running bitbake. This may be due to host contamination [host-user-contaminated]
So far we have tried to mould cp command into not carrying the build
user permissions into install area but it is never entirely fixed since
the issue keeps popping up in various scenes
This patch replaces use of cp with install utility and specifies install
mode for files explcitly
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Brings in a backport as described in
https://sourceware.org/git/?p=glibc.git;a=commit;h=86013ef5cea322b8f4b9c22f230c22cce369e947
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Jonas Bonn <jonas@norrbonn.se>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We do not have initial phase of bootstrapping toolchains anymore
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This lets linker to use its internal linker script for shared objects
Fixes with when gold is default linker
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We no longer have special "libc" for gcc so we can rely on plain
virtual/libc and reduce the complexity in the dependencies.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We need a libgcc to build glibc. Tranditionally we therefore build
a non-threaded and non-shared compiler (gcc-cross-initial), then use
that to build libgcc-initial which is used to build glibc which we can
then build gcc-cross and libgcc against.
Firstly, we can drop the glibc dependency from gcc-cross, *if* we make
two changes:
a) specify the minimum glibc version to support in a configure option
b) create a dummy limits.h file so that later when glibc creates one,
the headers structure has support for it. We can do this with a simple
empty file
Once gcc-cross is libc independent, we can use it to build both
libgcc-initial and then later libgcc.
libgcc-initial is tricky as we need to imitate the non-threaded and
non-shared case. We can do that by hacking the threading mode back to
"single" even if gcc reports "posix" and disable libc presence for the
libgcc-intial build. We have to create the dummy limits.h to avoid
compiler errors from a missing header.
glibc will fail to link with libgcc-initial due to a missing "exception
handler" capable libgcc (libgcc_eh.a). Since we know glibc doesn't need
any exception handler, we can safely symlink to libgcc.a.
With those changes, gcc-cross can be used in all places and we only need
one build of gcc for each architecture.
For some reason ifunc was being disabled on mips prior to these changes
but afterwards became enabled but caused assertion failures. This is
therefore disabled until we can debug that.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|