** Description changed:
[ Impact ]
gettext 1.0-5 fails to build from source on Ubuntu stonking with glibc >=
2.44.
During dh_auto_test, tests calling posix_spawn with chdir action
(test-posix_spawn-chdir and test-execute.sh) fail:
- On i386: crash with SIGSEGV (exit code 139) from infinite recursion.
- On amd64, arm64, ppc64el, s390x: child processes hang until the 150-minute
builder timeout.
This blocks gettext in stonking-proposed.
[ Fix ]
In glibc 2.44, <spawn.h> redirects posix_spawn_file_actions_addchdir to
posix_spawn_file_actions_addchdir_np via __REDIRECT_NTH.
gnulib compiles spawn_faction_addchdir.c to call
posix_spawn_file_actions_addchdir_np.
Because asm redirection cannot be undone by #undef, gcc emits
posix_spawn_file_actions_addchdir_np calling itself.
The patch guards spawn_faction_addchdir.c under defined __GLIBC__ &&
defined __USE_XOPEN2K24XSI. On glibc >= 2.44, libc already provides the
function, so gnulib skips redefining it and callers link directly to
libc.
[ Test Plan ]
1. On Ubuntu stonking with glibc 2.44-1ubuntu1:
dpkg-buildpackage -b -uc -us
2. Verify gnulib tests pass:
PASS: test-execute.sh
PASS: test-posix_spawn-chdir
3. Verify binary deb packages are built.
[ Where problems could occur ]
The guard is restricted to '#if defined __GLIBC__ && defined
__USE_XOPEN2K24XSI'. Older glibc versions or non-glibc systems use existing
code.
glibc 2.44 provides posix_spawn_file_actions_addchdir directly in libc.so.6.
[ Other Info ]
- Debian BTS report: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1149910
- Development series: stonking (26.10)
- Target release: Ubuntu stonking
- Bug reference: LP: #2167365
+ - PPA: https://launchpad.net/~steelf/+archive/ubuntu/ppa (gettext
1.0-5ubuntu1)
--- [ Original Report ]
gettext 1.0-5 fails to build from source in stonking.
Regressing architectures (built before, fail now): amd64, amd64v3, arm64,
armhf, i386, ppc64el, s390x.
Build log(s):
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595016/+files/buildlog_ubuntu-stonking-amd64.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595017/+files/buildlog_ubuntu-stonking-amd64v3.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595018/+files/buildlog_ubuntu-stonking-arm64.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595019/+files/buildlog_ubuntu-stonking-armhf.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595020/+files/buildlog_ubuntu-stonking-i386.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595021/+files/buildlog_ubuntu-stonking-ppc64el.gettext_1.0-5_BUILDING.txt.gz
-
https://launchpad.net/ubuntu/+source/gettext/1.0-5/+build/33595023/+files/buildlog_ubuntu-stonking-s390x.gettext_1.0-5_BUILDING.txt.gz
Diagnosis (deterministic matcher): testsuite-failure -> dh_auto_test:
error: make -j1 check TESTSUITEFLAGS="-j1 --verbose" VERBOSE=1 returned
exit code 2.
Root Cause Analysis:
gnulib tests (test-posix_spawn-chdir, test-execute.sh) segfault (i386) or
hang (amd64) across most architectures due to a known incompatibility with
glibc 2.44. The failure is caused by a known gnulib posix_spawn wrapper
conflict with glibc 2.44 headers that leads to infinite recursion. Fixing it
requires updating the gnulib submodule or backporting specific gnulib/glibc
patches, which is too complex for an automated patch.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2167365
Title:
gettext 1.0-5 FTBFS on amd64, amd64v3, arm64, armhf, i386, ppc64el,
s390x
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/gettext/+bug/2167365/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
