[FFmpeg-devel] FATE review

[Måns Rullgård]

Here's a quick summary of the failures we're currently seeing on FATE.

- ARM, armcc 4.1
  Several compiler bugs reported.  Awaiting fixes.

- PPC, gcc 4.0.4
  All VP8 tests fail, compiler bug likely.

- PPC, gcc 4.2.4
  GSM test fails with same output as sparc64/gcc-4.2.4

- Sparc64, gcc 4.2.4
  GSM test fails with same output as ppc/gcc-4.2.4

- Sparc64, gcc 3.4.6
  MPC8 demuxer is disabled with one test failing as result.
  Shall we make fate exclude tests depending on disabled features?

- Sparc64, OpenBSD
  Segmentation fault in lavfi-pixdesc_be.
  Random failures in mpeg2thread and mpeg4thread tests.

- x86_32, gcc 4.5.1
  NUT muxer gives different but valid output.

- x86_32, suncc
  Various tests using RGB fail.

- x86_64, icc 11.1
  mpeg4-als and sipr tests fail.

- x86_64, pathscale
  Lots of tests fail.  Compiler devs notified.

- win64
  Lots of mpegvideo related tests fail, probably due to broken inline asm.

- valgrind
  Invalid read in lavf-pixfmt, introduced after r25054.
  Memory leak in wmv8-drm: http://article.gmane.org/gmane.comp.video.ffmpeg.cvs/32899

- ARMv5 and AVR32 are offline for the rest of the month.

- Blackfin fails tests randomly due to memory fragmentation.

- Cygwin is down.

- Sparc64/solaris/gcc-4.4.0 is down.

- Darwin systems are down.

The situation overall is pretty good.  Most of the failures have known
causes, and the remainder affect only obscure or obsolete systems.
Nevertheless, I think the unexplained failures should be investigated,
whether or not we choose to do anything about them.

-- 
M?ns Rullg?rd
mans at mansr.com