Differences between revisions 54 and 55
Revision 54 as of 2013-07-17 14:57:23
Size: 20693
Comment:
Revision 55 as of 2013-07-17 15:08:29
Size: 20765
Editor: tschwinge
Comment: x86 GNU Hurd status.
Deletions are marked like this. Additions are marked like this.
Line 87: Line 87:
 * x86 GNU/Hurd, ???  * x86 GNU/Hurd, [[http://www.gnu.org/software/hurd/open_issues/glibc.html|needs patching]].

Release/2.18

1. Planning

1.1. Release Blockers?

  • PowerPC POWER8 Hardware Enablement
    • Re: [PATCH V2] PowerPC: Merge ports/ dl-procinfo.[hc] with base and remove ports versions of these files http://sourceware.org/ml/libc-alpha/2013-05/msg00395.html

      • Subsumed in latest patch on rsa/power8
      • Preparing to commit

                commit b90832c35a08a29ecfbe3a35331771ac1a448c2b
                Author: Ryan S. Arnold <rsa@linux.vnet.ibm.com>
                Date:   Mon May 13 15:38:13 2013 -0500
        
                    PowerPC: Remove redundant ports/sysdeps/powerpc/dl-procinfo.[hc].
        
                commit c796dcbe0df1184991f332f2eb955becff7e5770
                Author: Ryan S. Arnold <rsa@linux.vnet.ibm.com>
                Date:   Mon May 13 15:31:55 2013 -0500
        
                    PowerPC: Merge ports/ dl-procinfo.[hc] with base.
    • [PATCH rsa/power8] PowerPC: Add support for POWER8. http://sourceware.org/ml/libc-alpha/2013-05/msg00601.html

      • This has been acked by Roland.. just waiting on HWCAP2 and dl-procinfo to push.
                commit 7ec34a51099d54a38517d6856e32282d2f1187ae
                Author: Ryan S. Arnold <rsa@linux.vnet.ibm.com>
                Date:   Thu May 16 12:12:16 2013 -0500
        
                    PowerPC: Add support for POWER8.
    • Power Architecture Event-Based Branching facility ABI

1.2. Desirable this release?

  • Open bug count below 400
  • Some cross-localdef support
  • Testsuite generating logs with PASS/FAIL statuses
  • Substantial progress on merging distribution patches
  • Merging in changes from gnulib
  • Unify wiki todo lists and items from miscellaneous text files (Carlos)
  • libm multiprecision code cleanup and performance improvement (Siddhesh)
  • Review and fix getaddrinfo (particularly AI_ADDRCONFIG issues) issues (Carlos)
    • There are 24 getaddrinfo issues in the bug tracker.
    • There are 15 getaddrinfo issues in the Red Hat bugzilla.
    • There needs to be a review of the conformance of the implementation.
    • We need to compare against other vendors e.g. Solaris, FreeBSD, Windows, etc.
    • We need to fix things with an eye to not breaking applications.
  • BZ#11787 - Programs with large TLS segment fail (Carlos)

    • A workaround here is going to be to use Siddhesh's new LIBC_PTHREAD_DEFAULT_STACK_SIZE env var to bump up default stack sizes.
  • BZ#13690 - pthread_mutex_unlock potentially cause invalid access (Carlos)

    • Review code and potential race.
  • PowerPC POWER8 Hardware Enablement

2. Current Status

The release branch of glibc-2.18 is maintained by David Miller.

The 2.18 release branch is release/2.18/master.

Machine status:

  • x86, builds, no testsuite failures (By: Allan McRae on Arch Linux on Linux 3.9.8, gcc version 4.8.1, BFD assembler version 2.23.2)

  • x86 GNU/Hurd, needs patching.

  • x86-64, builds, no testsuite failures (By: Allan McRae on Arch Linux on Linux 3.9.8, gcc version 4.8.1, BFD assembler version 2.23.2)

  • PowerPC (soft-float), builds, some testsuite failures listed below.
  • PowerPC (hard-float), builds, some testsuite failutes listed below (By: Adhemerval Zanella on RHEL 6.4 on Linux 2.6.32-279.2.1.el6.ppc64, gcc version 4.8.0 (GCC), GNU assembler version 2.23.52 (powerpc64-linux) using BFD version (GNU Binutils) 2.23.52.20130429)
  • PowerPC64, builds, some testsuite failutes listed below (By: Adhemerval Zanella on RHEL 6.4 on Linux 2.6.32-279.2.1.el6.ppc64, gcc version 4.8.0 (GCC), GNU assembler version 2.23.52 (powerpc64-linux) using BFD version (GNU Binutils) 2.23.52.20130429)
  • s390x, builds, some testsuite failures listed below (By: Carlos on RHEL6.4 on Linux 2.6.32-358.el6.s390x, gcc version 4.4.7 20120313 (Red Hat 4.4.7-3) (GCC), GNU assembler version 2.23.52 (s390x-ibm-linux-gnu) using BFD version (GNU Binutils) 2.23.52.20130425)
  • ARM, builds, some testsuite failures listed below.
  • MIPS, builds, some testsuite failures listed below.
  • Alpha, builds, some testsuite failures listed below (By: rth on qemu 1.5.50 alpha-linux-user, gcc version 4.8.2 20130702 (prerelease), GNU assembler 2.22.52.20120604)
  • x32, ???
  • s390, ???
  • AArch64, ???
  • Microblaze, ???
  • sparc, ???
  • sparc64, ???
  • IA64, ???
  • HP-PARISC, ???
  • SH, ???
  • m68k, ???
  • tilepro/tilegx, ???

The ref structure of this branch is:

  • release/2.18/master: main branch
  • glibc-2.18.1: revision releases tagged out of release/2.18/master

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

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

  1. Is the fix committed in master? It has to be, unless it's not applicable to master (e.g. code has been rewritten meantime).
  2. Do you have commit permissions? If so, go ahead if you think it's reasonably safe. break;
  3. 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;
  4. Add the glibc_2.17 keyword to the appropriate bug report.
  5. 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/2.17/master as well, more so if they tell me about it!).

3. LSB Tests Status

No regressions (relative to 2.17).

Fixed problems: bug543(hypotl), bug512(tgamma), bug511_1(initstate) and bug803(getdate). Fixed requirements: math.exp.hypotl.08, math.gamma.tgamma.11.01, math.rand.initstate.11 and util.format.time.getdate.{01, 03, 04, 05, 07, 13, 18.01}.

More info: Testing/LSB_Core#Current_results

4. ABI Compatibility Status

ABI checker result (relative to 2.17):

CPU \ Kernel

GNU/Linux

x86

OK (2013/06/24)

x86_64

OK (2013/06/24)

arm

???

Added field si_addr_lsb to anon-struct-siginfo.h-108 which is a part of siginfo_t::anon-union-siginfo.h-70. The size of the siginfo_t struct is not changed. This change is safe. See related discussion on the mailing list from 07 May 2013.

4 changed constants: E_MIPS_ARCH_32, E_MIPS_ARCH_64, EF_MIPS_ARCH_32 and EF_MIPS_ARCH_64 (elf.h). This change is safe. See related discussion on the mailing list from 10 Jan 2013.

3 removed constants: EPOLL_NONBLOCK (epoll.h), GLIBC_HAVE_LONG_LONG and USE_ANSI (features.h). This is safe change too. See related discussions on the mailing list: 1, 2 and 3.

See also changes in private ABI.

5. Known Issues

5.1. Testsuite Failures

5.1.1. 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 (http://sourceware.org/ml/libc-alpha/2012-02/msg00468.html).

  • 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 http://sourceware.org/ml/libc-alpha/2012-04/msg01014.html regarding the use of system library directories here).

5.1.2. AArch64

make[2]: *** [/work/build-glibc/math/test-ldouble.out] Error 1

testing long double (without inline functions)
Failure: Real part of: casin (0x1.fp-16385 + 0x1p-63 i): Exception "Underflow" not set
Failure: Real part of: casin (0x1.fp-16385 - 0x1p-63 i): Exception "Underflow" not set
Failure: Real part of: casin (-0x1.fp-16385 + 0x1p-63 i): Exception "Underflow" not set
Failure: Real part of: casin (-0x1.fp-16385 - 0x1p-63 i): Exception "Underflow" not set
Failure: Real part of: casin (0x1.fp-16385 + 0x1p-112 i): Exception "Underflow" not set
Failure: Real part of: casin (0x1.fp-16385 - 0x1p-112 i): Exception "Underflow" not set
Failure: Real part of: casin (-0x1.fp-16385 + 0x1p-112 i): Exception "Underflow" not set
Failure: Real part of: casin (-0x1.fp-16385 - 0x1p-112 i): Exception "Underflow" not set
Failure: Real part of: casinh (0x1p-63 + 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (-0x1p-63 + 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (0x1p-63 - 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (-0x1p-63 - 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (0x1p-112 + 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (-0x1p-112 + 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (0x1p-112 - 0x1.fp-16385 i): Exception "Underflow" not set
Failure: Real part of: casinh (-0x1p-112 - 0x1.fp-16385 i): Exception "Underflow" not set

5.1.3. alpha

make[2]: *** [/home/rth/work/libc/bld-axp/math/test-float.out] Error 1
make[2]: *** [/home/rth/work/libc/bld-axp/math/test-double.out] Error 1
make[2]: *** [/home/rth/work/libc/bld-axp/math/test-ildoubl.out] Error 1
make[2]: *** [/home/rth/work/libc/bld-axp/math/test-ldouble.out] Error 1
make[2]: *** [/home/rth/work/libc/bld-axp/math/test-ifloat.out] Error 1
make[2]: *** [/home/rth/work/libc/bld-axp/math/test-idouble.out] Error 1

5.1.4. ARM

  • On systems with VFPv3 or VFPv4 (not VFPv2, VFPv3U or VFPv4U), math/test-fenv fails. fesetenv (FE_NOMASK_ENV) fails on such systems for lack of support for trapping exceptions, but the test does not handle that failure; feenableexcept should also fail, but does not, and the test does not handle failure of feenableexcept either. Similarly, math/test-fenv fails on systems without VFP hardware.

5.1.5. MIPS

  • On soft-float MIPS systems, there is no support for floating-point exceptions and rounding modes, causing math/test-fenv to fail.
  • 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/tst-audit1, elf/tst-audit2, elf/tst-audit8 and elf/check-localplt to fail. For the first two, see bug 2981.

5.1.6. s390x

Build system: RHEL6.4 on Linux 2.6.32-358.el6.s390x, gcc version 4.4.7 20120313 (Red Hat 4.4.7-3) (GCC), GNU assembler version 2.23.52 (s390x-ibm-linux-gnu) using BFD version (GNU Binutils) 2.23.52.20130425

make[2]: *** [/root/build/glibc/math/test-float.out] Error 1
make[2]: *** [/root/build/glibc/math/test-double.out] Error 1
make[2]: *** [/root/build/glibc/math/test-ldouble.out] Error 1
make[2]: *** [/root/build/glibc/math/test-ildoubl.out] Error 1
make[2]: *** [/root/build/glibc/math/test-ifloat.out] Error 1
make[2]: *** [/root/build/glibc/math/test-idouble.out] Error 1
make[1]: *** [math/tests] Error 2
make[2]: *** [/root/build/glibc/wcsmbs/wcsatcliff.out] Error 1
make[1]: *** [wcsmbs/tests] Error 2
make[2]: [/root/build/glibc/posix/annexc.out] Error 1 (ignored)
make[2]: [/root/build/glibc/conform/run-conformtest.out] Error 1 (ignored)
make[2]: *** [/root/build/glibc/elf/tst-unique3lib.so] Error 1
make[2]: *** [/root/build/glibc/elf/tst-unique3lib2.so] Error 1
make[2]: *** [/root/build/glibc/elf/tst-unique4lib.so] Error 1
make[1]: *** [elf/tests] Error 2
make: *** [check] Error 2

5.1.7. PowerPC (hard-float)

Build System: RHEL 6.4 on Linux 2.6.32-279.2.1.el6.ppc64, gcc version 4.8.0 (GCC), GNU assembler version 2.23.52 (powerpc64-linux) using BFD version (GNU Binutils) 2.23.52.20130429. The -mcpu=power7 flag was used.

make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/math/test-float.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/math/test-double.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/math/test-ldouble.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/math/test-ildoubl.out] Error 1
make[1]: *** [math/tests] Error 2
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/stdlib/tst-setcontext.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/stdlib/tst-makecontext3.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/stdlib/tst-setcontext-fpscr.out] Error 5
make[1]: *** [stdlib/tests] Error 2
make[2]: [/home/azanella/glibc/build/glibc32_power7_4.8/posix/annexc.out] Error 1 (ignored)
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/rt/tst-cpuclock2.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/rt/tst-mqueue5.out] Error 1
make[1]: *** [rt/tests] Error 2
make[2]: [/home/azanella/glibc/build/glibc32_power7_4.8/conform/run-conformtest.out] Error 1 (ignored)
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/debug/tst-backtrace5.out] Error 1
make[1]: *** [debug/tests] Error 2
make[2]: *** [/home/azanella/glibc/build/glibc32_power7_4.8/elf/check-localplt.out] Error 1
make[2]: *** [check-abi-libc] Error 1
make[1]: *** [elf/tests] Error 2
  • The single and double precision math test fail with the same issue: "Failure: nexttoward (qNaN, 1.1): Exception "Inexact" set". It is due a compiler issue regarding IBM log double implementation of addition, where adding NaN and normal number raises inexact exception. Bug already opened against GCC: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57363 .

  • The IBM long double tests shows 155 errors and while some are just ULPs update, some recent fixes for cacos, cacosh, casin, casinh, catan, catanh, tgamma are triggering inacuracies and underflow exceptions.
  • As previous releases, tst-setcontext, tst-makecontext3, tst-setcontext-fpscr are due a kernel bug.
  • The local PLT test fails due a GCC optimization (-ftree-loop-distribute-patterns) enabled at -O3 that transform some loops into memset/memmove calls. Since in some source code, the functions are not defined to use internal glibc symbols, compiler ends up generating PLT calls. The patch http://sourceware.org/ml/libc-alpha/2013-05/msg00878.html (pending review) fixes it.

5.1.8. PowerPC64

Build System: RHEL 6.4 on Linux 2.6.32-279.2.1.el6.ppc64, gcc version 4.8.0 (GCC), GNU assembler version 2.23.52 (powerpc64-linux) using BFD version (GNU Binutils) 2.23.52.20130429. The -mcpu=power7 flag was used.

make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/math/test-float.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/math/test-ldouble.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/math/test-double.out] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/math/test-ildoubl.out] Error 1
make[1]: *** [math/tests] Error 2
make[2]: [/home/azanella/glibc/build/glibc64_power7_4.8/posix/annexc.out] Error 1 (ignored)
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/rt/tst-cpuclock2.out] Error 1
make[1]: *** [rt/tests] Error 2
make[2]: [/home/azanella/glibc/build/glibc64_power7_4.8/conform/run-conformtest.out] Error 1 (ignored)
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/debug/tst-backtrace5.out] Error 1
make[1]: *** [debug/tests] Error 2
make[2]: *** [check-abi-libc] Error 1
make[2]: *** [/home/azanella/glibc/build/glibc64_power7_4.8/elf/check-localplt.out] Error 1
make[1]: *** [elf/tests] Error 2
  • The single and double precision math test fail with the same issue: "Failure: nexttoward (qNaN, 1.1): Exception "Inexact" set". It is due a compiler issue regarding IBM log double implementation of addition, where adding NaN and normal number raises inexact exception. Bug already opened against GCC: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57363 .

  • The IBM long double tests shows 155 errors and while some are just ULPs update, some recent fixes for cacos, cacosh, casin, casinh, catan, catanh, tgamma are triggering inacuracies and underflow exceptions.
  • The local PLT test fails due a GCC optimization (-ftree-loop-distribute-patterns) enabled at -O3 that transform some loops into memset/memmove calls. Since in some source code, the functions are not defined to use internal glibc symbols, compiler ends up generating PLT calls. The patch http://sourceware.org/ml/libc-alpha/2013-05/msg00878.html (pending review) fixes it.

5.1.9. PowerPC (soft-float)

  • math/test-float, math/test-double, math/test-ldouble, math/test-ildoubl and debug/tst-backtrace5 fail similarly to hard-float.
  • There are further failures in math/test-ldouble and math/test-ildoubl beyond those for hard float. Underflowing results of fmal are inaccurate in directed rounding modes, and one test of clog10l and one of clogl have the wrong sign of zero in the imaginary part of the result, probably because the soft-fp code used in libgcc as part of implementing IBM long double does not handle the software rounding mode from libc and does not handle signs of zero results the same as the instruction used in the hardware floating-point case.

5.2. Build Failures

Describe build failures here and how to fix them:

5.3. Packaging Changes

Describe packaging changes and how to fix them:

None: Release/2.18 (last edited 2014-08-08 08:21:21 by tschwinge)