Release/2.22

Contents

Current status
Planning
1. Release blockers?
2. Desirable this release?
Known Issues
1. Build and test issues
2. Packaging Changes

1. Current status

The release branch of glibc-2.22 is maintained by Carlos O'Donell and was released on 2015-08-14. There are no immediate plans for the next release.

The ref structure of this branch is:

release/2.22/master: main branch
glibc-2.22.N: revision releases tagged out of release/2.22/master

These people are interested in contents and further revisions tagged on the branch:

PERSON1
PERSON2
...

The general policies for release branches apply to this branch. Do you think a certain bugfix should be included in this branch?

Is the fix committed in master? It has to be, unless it's not applicable to master (e.g. code has been rewritten meantime).
Do you have commit permissions? If so, go ahead if you think it's reasonably safe. break;
Can you handle Git yourself? Then you can clone the glibc repository, cherry-pick the appropriate fixes, push your branch out and send a pull request at libc-alpha. break;
Add the glibc_2.22 keyword to the appropriate bug report.
If there is no appropriate bug report, send a request for the fix to be included to libc-alpha.

A revision release is tagged either when some critical bug-fix appears, or after some period of real-world testing, usually mainly in some SUSE distribution branch (but other distributions are welcome to run latest release/2.22/master as well, more so if they tell me about it!).

2. Planning

What things do we want to accomplish this release?

2.1. Release blockers?

Fix thread_local deadlocks: https://sourceware.org/ml/libc-alpha/2015-07/msg00314.html
- Waiting on Siddhesh for updated test case.
- Version 7 accepted.
- Fixed by commit 90b37cac8b5a3e1548c29d91e3e0bff1014d2e5c
Fix sparc32 sigaction ABI: https://sourceware.org/ml/libc-alpha/2015-07/msg00592.html
- Fixed by commit 7fde904c73c57faea48c9679bbdc0932d81b3a2f
Revert cancellation changes: https://sourceware.org/ml/libc-alpha/2015-05/msg00883.html
- Compiler can now reorder asm syscall wrappers (with cancellation code) out of cancel regions and fail to call cleanup handlers.
- Fixed by commit 203c1a898dd2e281eae30f3c57ceb8d4a50b00f4
Mark tst-makecontext XFAIL for i686: https://sourceware.org/ml/libc-alpha/2015-07/msg00241.html
- i686 tst-makecontext failure due to cfi problems. Bug in gas?
- Fixed by commit f6267fd880a3c93a05e731c4ae6d661bc91aaf25
POWER string ops worst case O(n^2)?
- Currently under discussion.
- May not be O(n^2), but rather O(mn) with large m.
- Still an issue?
- Discuss with IBM and interested people.
- Ignored: Not considered a release blocker for 2.22 because the algorithm is O(mn) instead of O(n^2).
  - Future discussions must be had over acceptability of O(mn) in glibc.
Review UB in elf/elf.h for DT_EXTRATAGIDX.
- https://sourceware.org/ml/libc-alpha/2015-07/msg00673.html
- https://sourceware.org/ml/libc-alpha/2015-08/msg00088.html
- Ignored: Not considered a release blocker for 2.22 because there is no final patch for the fix, and only impacts gcc 6.0.
Review intermittent failures for sort-test.sh
- https://sourceware.org/ml/libc-alpha/2015-07/msg00944.html
- Ignored: This is indeed a regression in the sorting of Danish and possibly other languages, but is not sufficiently serious to block the release. Noting this issue in packaging notes.
Review i686, x32, and x86_64 failures with locale generation.
- https://www.sourceware.org/ml/libc-alpha/2015-08/msg00013.html
- Ignored: On RHEL7 builds for 32-bit x86 the problems are not seen. This indicates the failure is likely a compiler interaction between glibc and the F21 compiler, or an environment issue. The fix that H.J. proposes is a good idea, but doesn't solve the "memory clobbered past end of allocated block" errors seen during testing.

2.2. Desirable this release?

3. Known Issues

This release has known issues with sorting of Danish characters, see https://sourceware.org/bugzilla/show_bug.cgi?id=18589#c1. The glibc community fixed this issue with commit 87701a58e291bd7ac3b407d10a829dac52c9c16e on the 2.22 branch, e.g. 2.22.1, after release on 2015-10-08. The 2.22 branch is a rolling release branch, no release will be made from the branch. Distributions should rebase to the latest 2.22.1.

3.1. Build and test issues

Describe build and test issues for each architecture, or confirm a clean build with no testsuite failures. The list below is not a complete list of ABI variants; testing should try to cover the different ABI variants as far as possible.

Build system: UNAME -a, GCC?, Binutils?, Kernel ?

TRIMMED LIST OF FAILURES.

3.1.1. Architecture-independent

On systems with limited memory, stdio-common/test-vfprintf and stdio-common/bug22 may fail (bug 14231).
On some systems, a race condition may cause nptl/tst-cancel7 and nptl/tst-cancelx7 to fail (bug 14232).
If cross-testing, and the physical path to the build directory is different on the build system and the host used for testing (if it involves a symlink on one system but not the other), io/ftwtest fails.
On systems with Linux kernels before September 2011, rt/tst-cpuclock2 fails (https://sourceware.org/ml/libc-alpha/2012-02/msg00468.html). But this failure has also been reported with more recent kernels, so that kernel fix seems not to have been sufficient.
rt/tst-shm fails if /dev/shm is not mounted or does not have write permissions for the user running the tests.
If the test system does not have suitable copies of libgcc_s.so and libstdc++.so installed in system library directories, it is necessary to copy or symlink them into the build directory before testing (see https://sourceware.org/ml/libc-alpha/2012-04/msg01014.html regarding the use of system library directories here).
nss/tst-nss-getpwent may time out with large numbers of users on slow systems.
Targets with copy relocation should verify if tests for PR 17711, elf/tst-protected1a and elf/tst-protected1b, pass. If not, they can be fixed similar to commit for x86. On x86, these 2 tests may not link or may fail with older linkers. Linkers on master and 2.25 branch with bug fix for PR 17709 work correctly. Linkers for other targets may need the similar fix if these 2 tests fail.
A number of tests in the localedata directory such as tst_mbrlen and tst_wctomb have been observed to fail in parallel invocations of make check due to bug 17617.

3.1.2. AArch64

Build system: GCC 6.0.0 20150724, binutils 2.25.51.20150727, Linux 3.18.5 (Szabolcs Nagy)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: debug/tst-backtrace5
FAIL: debug/tst-backtrace6
FAIL: nptl/tst-stack4

debug/tst-backtrace[56] fail because backtrace_symbol does not find the symbol name of syscalls that are cancellation points. Will be fixed by the cancellation design changes.
nptl/tst-stack4 fails (rarely) because of a dynamic-linker race (arch independent, not a new regression), Glibc bug 17918.

Build system: RHEL 7 Build box, gcc version 4.8.5 20150623 (Red Hat 4.8.5-4) (GCC), GNU assembler version 2.23.52.0.1 (aarch64-redhat-linux) using BFD version version 2.23.52.0.1-50.el7 20130226, kernel 4.1.0-0.12.el7.aarch64 (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: debug/tst-backtrace5
FAIL: debug/tst-backtrace6
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: math/basic-test
FAIL: math/test-double
FAIL: math/test-float
FAIL: rt/tst-mqueue6
Summary of test results:
      8 FAIL
   2247 PASS
     88 XFAIL
      3 XPASS

3.1.3. Alpha

3.1.4. ARM

Build system: (soft-float) GCC 5.2.1 20150721, binutils 2.25.51.20150721, Linux 2.6.34.1 (Joseph Myers)

FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: io/ftwtest
FAIL: localedata/sort-test
FAIL: math/basic-test
FAIL: nptl/tst-cancel24-static
FAIL: nptl/tst-cancel7
FAIL: nptl/tst-cancelx7
FAIL: nptl/tst-mutexpi9
FAIL: nptl/tst-stack4
FAIL: nss/tst-nss-getpwent
FAIL: rt/tst-shm
FAIL: stdio-common/bug22
FAIL: stdio-common/test-vfprintf
FAIL: stdio-common/tst-sprintf
FAIL: stdlib/tst-makecontext
FAIL: timezone/tst-tzset
FAIL: wcsmbs/test-wcsncmp

On soft-float systems, math/basic-test fails because of GCC bug 59833.
Cause of nptl/tst-mutexpi9 and nptl/tst-stack4 failures unknown.
stdio-common/tst-sprintf failure possibly arising from limited memory available.
Cause of localedata/sort-test failure unknown; also seen on other architectures. Works when re-run
Cause of nptl/tst-cancel24-static failure unknown.
Cause of stdlib/tst-makecontext failure unknown.
Cause of timezone/tst-tzset failure unknown, but may result from limited memory or slow system.
wcsmbs/test-wcsncmp failure may be alignment issue (patch).

Build system: (Angstrom/OpenEmbedded cortex-a15,hard-float) GCC 4.9.3, binutils 2.25.0, Linux 3.10.40 (Khem Raj)

FAIL: math/test-double
FAIL: math/test-float
FAIL: resolv/tst-leaks
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: stdlib/tst-makecontext

for both float and double GCC bug 66731.
resolv/tst-leaks exits with Timed out: killed the child process.
for both tst-protected1a and tst-protected1b Glibc bug 17711.
Cause of stdlib/tst-makecontext Didn't expect signal from child: got `Aborted'.

3.1.5. HPPA

Build system: parisc PA8600 (PCX-W+) 9000/785/C3600, gcc (Gentoo 4.8.4 p1.6, pie-0.6.1), GNU assembler (Gentoo 2.24 p1.4), 3.18.7-gentoo. (Mike Frysinger)

FAIL: conform/POSIX/sys/mman.h/conform
FAIL: conform/POSIX2008/sys/mman.h/conform
FAIL: conform/UNIX98/sys/mman.h/conform
FAIL: conform/XOPEN2K/sys/mman.h/conform
FAIL: conform/XOPEN2K8/sys/mman.h/conform
FAIL: conform/XPG4/sys/mman.h/conform
FAIL: debug/tst-backtrace2
FAIL: debug/tst-backtrace3
FAIL: debug/tst-backtrace4
FAIL: debug/tst-backtrace5
FAIL: debug/tst-backtrace6
FAIL: debug/tst-longjmp_chk2
FAIL: elf/check-abi-libc
FAIL: elf/check-execstack
FAIL: elf/check-localplt
FAIL: elf/check-textrel
FAIL: elf/tst-addr1
FAIL: elf/tst-audit2
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-idouble
FAIL: math/test-ifloat
FAIL: nptl/tst-attr3
FAIL: nptl/tst-cancel2
FAIL: nptl/tst-cancel21-static
FAIL: nptl/tst-cancel23
FAIL: nptl/tst-cancel24-static
FAIL: nptl/tst-cancel3
FAIL: nptl/tst-cancel4
FAIL: nptl/tst-cancel5
FAIL: nptl/tst-cancelx11
FAIL: nptl/tst-cancelx2
FAIL: nptl/tst-cancelx20
FAIL: nptl/tst-cancelx21
FAIL: nptl/tst-cancelx3
FAIL: nptl/tst-cancelx4
FAIL: nptl/tst-cancelx5
FAIL: nptl/tst-cleanup4
FAIL: nptl/tst-cleanupx4
FAIL: nptl/tst-default-attr
FAIL: nptl/tst-join6
FAIL: nptl/tst-stack1
FAIL: nptl/tst-stack4
FAIL: rt/tst-cputimer1
FAIL: rt/tst-cputimer2
FAIL: rt/tst-cputimer3
FAIL: rt/tst-mqueue8
FAIL: rt/tst-mqueue8x
FAIL: rt/tst-timer4
FAIL: rt/tst-timer5
FAIL: stdio-common/test-vfprintf
FAIL: stdlib/tst-makecontext
FAIL: stdlib/tst-setcontext2
FAIL: stdlib/tst-setcontext3

Note: Results required a few pthread patches related to stack direction.

3.1.6. IA64

Note: Building with gcc-4.8+ and --hash-style=gnu will create an ldso that segfaults immediately. See https://gcc.gnu.org/PR60465 for more details. You can workaround this in three ways:

build glibc with <=gcc-4.7
build glibc with -Wl,--hash-style=sysv
build glibc with this hack

Build system: ia64 Dual-Core Intel(R) Itanium(R) Processor 9040, gcc (Gentoo 4.7.4 p1.1, pie-0.5.5), GNU assembler (Gentoo 2.24 p1.4), Linux 3.14.14-gentoo. (Mike Frysinger)

FAIL: debug/tst-backtrace4
FAIL: math/test-double
FAIL: math/test-fenv-return
FAIL: math/test-float
FAIL: math/test-idouble
FAIL: math/test-ifloat
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
FAIL: nptl/tst-cancel24-static
FAIL: stdio-common/tst-long-dbl-fphex
FAIL: stdlib/tst-makecontext3

tst-backtrace4 is known to fail on IA64 due to limited backtrace output; mailing list thread
all the math/ tests probably match some open bugs on the topic, or need some baselines updated, or are limitations in the IA64 optimized libm
tst-cancel24-static crash is unknown
tst-cpuclock2 seems to be flaky as it passes normally, but sometimes fails under load
tst-makecontext3 fails as makecontext is known to be limited on IA64; bug 14250

3.1.7. M68K

Build system: ARAnyM 1.0.2, openSUSE Tumbleweed, gcc5-5.1.1+r224716, binutils-2.25.0, Linux 3.11.6 (Andreas Schwab)

FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: malloc/tst-malloc-backtrace
FAIL: math/basic-test
FAIL: math/test-double
FAIL: math/test-fenv
FAIL: math/test-float
FAIL: math/test-idouble
FAIL: math/test-ifloat
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
FAIL: math/test-snan
FAIL: nptl/tst-cancel24-static
FAIL: nptl/tst-cleanupx4
FAIL: stdio-common/bug22
FAIL: stdio-common/tstdiomisc
FAIL: stdlib/tst-makecontext

3.1.8. MicroBlaze

3.1.9. MIPS

Build system: GCC 5.2.1 20150721, binutils 2.25.51.20150721, Linux Linux 2.6.35.9 (Joseph Myers). n64 requires a subsequent binutils fix when binutils is built for a 32-bit host (commit d5cff5df74b18e1e5ed94de8f4c9adee3ffd95c6 (23 July), if commit a5499fa4649e4325cf46edfff2f24dae2fe2afef (11 June) is present in the binutils version used; not applicable for older binutils). The following failures apply to both BE and LE, o32, n32 and n64, both hard-float and soft-float.

FAIL: elf/check-execstack
FAIL: elf/check-localplt
FAIL: io/ftwtest
FAIL: nptl/tst-cancel24-static
FAIL: nss/tst-nss-getpwent
FAIL: stdio-common/bug22
FAIL: stdio-common/test-vfprintf
FAIL: timezone/tst-tzset

For o32 and n32 (but not n64):

FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b

Seen for o32 hard float:

FAIL: math/test-double
FAIL: math/test-float

Seen for soft float:

FAIL: math/test-fenv
FAIL: math/test-fenvinline

Seen for n32 soft float (BE):

FAIL: nptl/tst-setuid3

Seen for o32 soft float (BE):

FAIL: nptl/tst-mutexpi5

On soft-float MIPS systems, there is no support for floating-point exceptions and rounding modes, causing math/test-fenv to fail.
For hard-float MIPS, o32 ABI (maybe only some architecture variants), math/test-float and math/test-double fail with missing "invalid" exceptions; see bug 16399.
MIPS GCC does not use PT_GNU_STACK markers (this is a GCC issue), causing elf/check-execstack to fail (the relevant kernel support went in around February 2010).
Peculiarities of MIPS ELF cause elf/check-localplt to fail.
The cause of the nptl/ failures is unknown, some likely intermittent.

3.1.10. Nios II

Build system: GCC 5.1.1 20150716, binutils 2.25.51, Linux 4.0 (Chung-Lin Tang)

FAIL: elf/tst-rtld-load-self
FAIL: elf/tst-tls13
FAIL: iconvdata/tst-tables
FAIL: io/ftwtest
FAIL: nptl/tst-cancel24-static
FAIL: nptl/tst-cancel7
FAIL: nptl/tst-cancelx7
FAIL: nptl/tst-cond13
FAIL: nptl/tst-rwlock6
FAIL: nptl/tst-stack4
FAIL: nss/tst-nss-getpwent
FAIL: stdio-common/bug22
FAIL: stdio-common/test-vfprintf
FAIL: stdio-common/tst-sprintf
FAIL: string/test-strcasecmp
FAIL: string/test-strncasecmp
FAIL: timezone/tst-tzset
FAIL: wcsmbs/test-wcsncmp
FAIL: build stdlib/bug-getcontext.o
FAIL: build sunrpc/tst-xdrmem
FAIL: build sunrpc/tst-xdrmem2

Most are known from 2.21, only new regressions are described below.
Cause of nptl/tst-cancel24-static failure unknown.
Cause of nptl/tst-cond13 failure unknown.
Cause of nptl/tst-rwlock6 failure unknown.
nss/tst-nss-getpwent failed due to timeout. May be due to test environment speed issues, manually verified to be working slowly when timed out.
Cause of timezone/tst-tzset unknown.
wcsmbs/test-wcsncmp failure may be alignment issue (patch).

3.1.11. PowerPC (32-bit soft-float)

Build system: (soft-float) GCC 5.2.1 20150721, binutils 2.25.51.20150721, Linux 2.6.34.1 (Joseph Myers)

FAIL: io/ftwtest
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
FAIL: rt/tst-cpuclock2
FAIL: rt/tst-shm
FAIL: stdio-common/bug22
FAIL: stdlib/tst-setcontext2

test-double, test-float fail because of GCC bug 64811 causing __gcc_qadd in libgcc to generate spurious floating-point exceptions.
test-ildoubl and test-ldouble fail because of IBM long double issues.
Cause of stdlib/tst-setcontext2 failure unknown.

3.1.12. PowerPC (32-bit hard-float)

Build system: Debian GNU/Linux sid, gcc version 4.9.3, GNU Binutils version 2.25, Linux 4.0.0 ppc64 (Tulio Magno)

Had to use -Wno-error=memset-transposed-args to avoid the following warning in string/tester.c.

tester.c: In function ‘test_memset’:
tester.c:1313:10: warning: ‘memset’ used with constant zero length parameter; this could be due to transposed parameters [-Wmemset-transposed-args]
   (void) memset(one+2, 'y', 0);

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-idouble
FAIL: math/test-ifloat
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
Summary of test results:
      6 FAIL
   2243 PASS
     88 XFAIL
      3 XPASS

3.1.13. PowerPC64 (64-bit hard-float)

Build system: ppc64 Fedora 21, gcc version 4.9.2, GNU Binutils 2.24, Linux 3.19.3 (Tulio Magno)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
FAIL: rt/tst-mqueue5
Summary of test results:
      3 FAIL
   2286 PASS
     88 XFAIL
      3 XPASS

Build System: RHEL 7 Test box, gcc version 4.8.5 20150623 (Red Hat 4.8.5-4) (GCC), GNU assembler version 2.23.52.0.1 (ppc64-redhat-linux) using BFD version version 2.23.52.0.1-54.el7 20130226, Kernel 3.10.0-300.el7.ppc64 (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
Summary of test results:
      4 FAIL
   2285 PASS
     88 XFAIL
      3 XPASS

3.1.14. PowerPC64LE (64-bit hard-float)

Build system 1: ppc64le Ubuntu 14.04, gcc version 4.8.4, GNU Binutils version 2.24, Linux 3.13.0 (Tulio Magno)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
Summary of test results:
      2 FAIL
   2287 PASS
     88 XFAIL
      3 XPASS

Build system 2: ppc64le running Red Hat Enterprise Linux 7.2 Beta, GCC version 4.8.5 20150623 (Red Hat 4.8.5-4), Binutils 2.23.52.0.1-50, Linux kernel version 3.10.0-296 (???)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
FAIL: posix/tst-getaddrinfo5
FAIL: rt/tst-cpuclock2
FAIL: rt/tst-mqueue5
Summary of test results:
      7 FAIL
   2273 PASS
     88 XFAIL
      3 XPASS

The failures in the tests in the math and posix subdirectories were discussed in the thread Re: glibc 2.21 -- ppc64le test results.
The failures in the rt tests have been observed with other versions of GLIBC on RHEL 7 and are believed to be due to problems elsewhere. See bugs 1064059 and 1064063 for more.

Build System: RHEL 7 Build Box, gcc version 4.8.5 20150623 (Red Hat 4.8.5-4), GNU assembler version 2.23.52.0.1 (ppc64le-redhat-linux) using BFD version version 2.23.52.0.1-54.el7 20130226, Kernel 3.10.0-300.el7.ppc64le (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: localedata/sort-test
FAIL: math/test-double
FAIL: math/test-float
FAIL: math/test-ildoubl
FAIL: math/test-ldouble
Summary of test results:
      5 FAIL
   2284 PASS
     88 XFAIL
      3 XPASS

3.1.15. S/390 (32-bit)

Build system: s390 GNU/Linux, gcc version 4.9.2, GNU Binutils version 2.25, Linux 3.19.2 (Stefan Liebler)

Test-Results for tests.sum:
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
Summary of tests.sum results:
      2 FAIL
   2279 PASS
     84 XFAIL
      7 XPASS


Test-Results for xtests.sum:
FAIL: nptl/tst-setuid2
Summary of xtests.sum results:
      1 FAIL
     15 PASS

nptl/tst-setuid2 passes with root privileges.
See architecture independent test issues for elf/tst-protected1a|b.

3.1.16. S/390 (64-bit)

Build system: s390x GNU/Linux, gcc version 4.9.2, GNU Binutils version 2.25, Linux 3.19.2 (Stefan Liebler)

Test-Results for tests.sum:
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: nptl/tst-cancelx16
FAIL: nptl/tst-cancelx4
FAIL: nptl/tst-cancelx5
Summary of tests.sum results:
      5 FAIL
   2277 PASS
     84 XFAIL
      7 XPASS


Test-Results for xtests.sum:
FAIL: nptl/tst-setuid2
Summary of xtests.sum results:
      1 FAIL
     15 PASS

nptl/tst-setuid2 passes with root privileges.
See architecture independent test issues for elf/tst-protected1a|b.
nptl/tst-cancelx4|5 fails due to "cleanup handler not called for 'sigwait'". Gcc 4.9 and higher saves r15 in a floating-point register in do_sigwait() and emits false cfi information. If a thread is cancelled while sigwait blocks, unwinding has already stopped before the function with the cleanup-handler is reached. See gcc patch "[PATCH] S/390: Fix cfi for GPR 2 FPR saves" (https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01843.html).
nptl/tst-cancelx16 fails due to not called cleanup handler around lockf - same issue as above.

Build system: RHEL 7 Build box, gcc version 4.8.5 20150623 (Red Hat 4.8.5-4) (GCC), GNU assembler version 2.25.51 (s390x-ibm-linux-gnu) using BFD version (GNU Binutils) 2.25.51.20150801, Kernel 3.10.0-300.el7.s390x (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: nptl/tst-cancelx17
Summary of test results:
      3 FAIL
   2280 PASS
     88 XFAIL
      3 XPASS

3.1.17. SH

3.1.18. SPARC (32-bit)

3.1.19. SPARC (64-bit)

Build system: GCC 5.3.1 20151207, binutils 2.25.90.20151209, Linux 3.16.7-ckt11 (Aurelien Jarno)

FAIL: conform/ISO/setjmp.h/conform
XPASS: conform/ISO11/complex.h/conform
FAIL: conform/ISO11/setjmp.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: conform/ISO99/setjmp.h/conform
FAIL: conform/POSIX/pthread.h/conform
FAIL: conform/POSIX/setjmp.h/conform
FAIL: conform/POSIX2008/pthread.h/conform
FAIL: conform/POSIX2008/setjmp.h/conform
FAIL: conform/UNIX98/pthread.h/conform
FAIL: conform/UNIX98/setjmp.h/conform
FAIL: conform/XOPEN2K/pthread.h/conform
FAIL: conform/XOPEN2K/setjmp.h/conform
FAIL: conform/XOPEN2K8/pthread.h/conform
FAIL: conform/XOPEN2K8/setjmp.h/conform
FAIL: conform/XPG3/setjmp.h/conform
FAIL: conform/XPG4/setjmp.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: wcsmbs/test-wcsncmp
Summary of test results:
     18 FAIL
   2236 PASS
     88 XFAIL
      3 XPASS

conform tests are due to the definition of struct sparc64_jmp_buf in <bits/setjmp.h>
see architecture independent test issues for elf/tst-protected1a|b.
wcsmbs/test-wcsncmp are due to unaligned accesses in the tests: https://sourceware.org/ml/libc-alpha/2015-07/msg00698.html

3.1.20. TILE-Gx (64-bit and 32-bit)

Build system: CentOS 6.3, but with gcc 4.8.2, binutils 2.25, kernel 3.10.55. (Chris Metcalf)

FAIL: nptl/tst-cancel24
FAIL: nptl/tst-cancelx16
FAIL: nptl/tst-cancelx17
FAIL: nptl/tst-cancelx18
FAIL: nptl/tst-cancelx20
FAIL: nptl/tst-cancelx21
FAIL: nptl/tst-cancelx4
FAIL: nptl/tst-cancelx5
FAIL: nptl/tst-cleanupx0
FAIL: nptl/tst-cleanupx1
FAIL: nptl/tst-cleanupx3
FAIL: nptl/tst-cleanupx4
FAIL: nptl/tst-oncex3
FAIL: nptl/tst-oncex4
FAIL: rt/tst-mqueue8x

The failures above are due to bug 14266

FAIL: math/test-float
FAIL: math/test-ifloat

The failures above are due to bug 18699

FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b

The failures above are due to bug 17711 (fixed for x86 but not yet tilegx)

FAIL: nptl/tst-cancel24-static

Cause of nptl/tst-cancel24-static failure unknown

3.1.21. TILEPro

Cross "make" on x86_64 successful with binutils-2.23.1, linux-3.9.2 headers, and gcc 4.7.3. "make check" was not run. (Chris Metcalf)

3.1.22. x86 (32-bit, Linux)

Build system: Fedora 21 Build Box, gcc version 4.9.2 20150212 (Red Hat 4.9.2-6) (GCC), GNU assembler version 2.24 (x86_64-redhat-linux) using BFD version version 2.24, Kernel 4.0.6-200.fc21.x86_64. (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: math/test-float
Summary of test results:
      3 FAIL
   2283 PASS
     89 XFAIL
      3 XPASS

Build system: RHEL 7 Build Box, gcc version 4.8.5 20150623 (Red Hat 4.8.5-4) (GCC), GNU assembler version 2.23.52.0.1 (x86_64-redhat-linux) using BFD version version 2.23.52.0.1-54.el7 20130226, kernel 3.10.0-300.el7.x86_64.

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: localedata/sort-test
FAIL: rt/tst-cputimer1
Summary of test results:
      4 FAIL
   2285 PASS
     89 XFAIL
      3 XPASS

3.1.23. x86_64 (64-bit, Linux)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
Summary of test results:
      2 FAIL
   2304 PASS
     88 XFAIL
      3 XPASS

Build system: Fedora 21 Build Box, gcc version 4.9.2 20150212 (Red Hat 4.9.2-6) (GCC), GNU assembler version 2.25.51 (x86_64-unknown-linux-gnu) using BFD version (GNU Binutils) 2.25.51.2015072, Kernel 4.0.7-200.fc21.x86_64. (Carlos O'Donell)

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
Summary of test results:
   2311 PASS
     88 XFAIL
      3 XPASS

Newer binutils fixes protected test failures.

XPASS: conform/ISO11/complex.h/conform
XPASS: conform/ISO11/stdalign.h/conform
XPASS: conform/ISO11/stdnoreturn.h/conform
FAIL: elf/tst-protected1a
FAIL: elf/tst-protected1b
FAIL: nptl/tst-robust8
Summary of test results:
      3 FAIL
   2306 PASS
     88 XFAIL
      3 XPASS

3.1.24. x86_64 (x32, Linux)

Build system: Fedora 22, gcc version 5.1.1 20150707 (Red Hat 5.1.1-5) (GCC), GNU assembler (Linux/GNU Binutils) 2.25.51.0.4.20150713, Kernel 4.0.8-300.fc22.x86_64 (H.J. Lu)

XPASS: conform/UNIX98/ndbm.h/linknamespace
XPASS: conform/XOPEN2K/ndbm.h/linknamespace
XPASS: conform/XOPEN2K8/ndbm.h/linknamespace
XPASS: conform/XPG4/ndbm.h/linknamespace
FAIL: elf/check-localplt
FAIL: elf/tst-leaks1-mem
Summary of test results:
      2 FAIL
   1870 PASS
      5 XFAIL 
      4 XPASS 
FAIL: nptl/tst-setuid2
Summary of test results for extra tests:
      1 FAIL
     15 PASS

3.1.25. x86 (32-bit, Hurd)

3.2. Packaging Changes

3.2.1. regexp.h

The header file regexp.h (not to be confused with regex.h) is deprecated, and will be removed from GNU libc in version 2.23 (leaving behind a stub, containing only an #error directive). In 2.22, any use of this header provokes an unconditional compile-time warning.

This header and the API it defines were formerly part of SUS, but were deprecated in 1994 and removed from the standard in 2001. Moreover, the glibc implementation suffered from bugs that had gone unnoticed from 1996 through 2009, including memory leaks which were impractical to fix. See bug 18681 for more information.

Programs that use this header are expected to be rare. They should be updated to use regex.h, and the somewhat different API defined there, instead. regcomp is the replacement for the compile function and its associated macros, and regexec is the replacement for the step and advance functions and their associated global variables.