glibc buildbot status

[Mark Wielaard]

Hi,

(CC Frank, see below for the fedrawhide results
 and Sam, see the [missing] sparc results)

For Cauldron next month we wanted to give a talk on
Sourceware Infrastructure Organization and Automation
https://conf.gnu-tools-cauldron.org/prg26/talk/SVVBRS/

Which includes a discussion about various CI systems, what is
currently being build and whether that is actually useful.

So I looked over the glibc buildbot results to see what the current
status is.

There are two "special" builders:

- glibc-autoregen
  https://builder.sourceware.org/buildbot/#/builders/glibc-autoregen

Which checks that all the generated files that are checked into git
can be regenerated and match what is being checked in. This
occassionally catches issues, all accidents, nothing malicious.
Mostly issues with using the "wrong" version of tools.

- glibc-snapshots-trunk
  https://builder.sourceware.org/buildbot/#/builders/glibc-snapshots-trunk
  https://snapshots.sourceware.org/glibc/trunk/

Creates both a release tarball and regenerates the manual in html and
pdf formats. Makes sure there aren't any surprises at release
time. Occassionally finds issues with (newer) texinfo.

There are a couple of zero-fail (make check) builders:

- glibc-debian-arm64
  https://builder.sourceware.org/buildbot/#/builders/glibc-debian-arm64

   7056 PASS
     83 UNSUPPORTED
     16 XFAIL
      2 XPASS

- glibc-fedora-x86_64
  https://builder.sourceware.org/buildbot/#/builders/glibc-fedora-x86_64

   7238 PASS
    194 UNSUPPORTED
     16 XFAIL

- glibc-debian-ppc64
  https://builder.sourceware.org/buildbot/#/builders/glibc-debian-ppc64

   6630 PASS
    116 UNSUPPORTED
     16 XFAIL
      2 XPASS

- glibc-fedora-s390x
  https://builder.sourceware.org/buildbot/#/builders/glibc-fedora-s390x

   7016 PASS
     61 UNSUPPORTED
     16 XFAIL
      2 XPASS

Any new failure in these normally signals a real bug/issue.

Then there are a couple of nearly zero-fail builders:

- glibc-debian-armhf
  https://builder.sourceware.org/buildbot/#/builders/glibc-debian-armhf

      2 FAIL
   6466 PASS
    133 UNSUPPORTED
     16 XFAIL
      2 XPASS

FAIL: elf/tst-dl-cache-old-format
FAIL: elf/tst-ldconfig-cache

- glibc-fedora-ppc64le
  https://builder.sourceware.org/buildbot/#/builders/glibc-fedora-ppc64le

      1 FAIL
   7279 PASS
     77 UNSUPPORTED
     16 XFAIL
      2 XPASS

FAIL: misc/tst-pidfd-consts

- glibc-debian-i386
  https://builder.sourceware.org/buildbot/#/builders/glibc-debian-i386

      1 FAIL
   7010 PASS
     90 UNSUPPORTED
     16 XFAIL
      2 XPASS

FAIL: nscd/tst-nscd-basic

It would of course be nice to get these to zero fail, either by fixing
the test or ignoring it if it looks like an "environment issue" as
seem the case with the debian FAILs. I am not sure why on ppc64le the
pidfd constants seem to be different than on other arches:

Different values for PIDFD_SELF_PROCESS: -10001 != -20000
Different values for PIDFD_SELF_THREAD_GROUP: -10001 != -20000

Finally there is the fedora rawhide builder:

- gibc-fedrawhide-x86_64
  https://builder.sourceware.org/buildbot/#/builders/glibc-fedrawhide-x86_64

     10 FAIL
   7230 PASS
    242 UNSUPPORTED
     16 XFAIL

FAIL: io/tst-fallocate64
FAIL: io/tst-posix_fallocate64
FAIL: misc/tst-copy_file_range-large
FAIL: misc/tst-preadvwritev64
FAIL: posix/tst-mmap-offset
FAIL: posix/tst-preadwrite64
FAIL: posix/tst-truncate64
FAIL: stdio-common/tst-freopen5
FAIL: stdio-common/tst-freopen5-mem
FAIL: timezone/tst-tzset

Almost all these failures are because of unexpected SIGXFSZ signals or
system calls errors with EFBIG indicating file size limits being
exceeded. I suspect the container this is being run in has a too low
RLIMIT_FSIZE.

Missing are the ubuntu-riscv and gentoo-sparc results.
I just enabled the riscv builder again:
https://sourceware.org/cgit/builder/commit/?id=80d78a9e3e21
And I am hoping Sam can bring back the gentoo-sparc builder.

Cheers,

Mark