Builder ffmpegsos-solaris10-sparc Build #14257

Results:

Failed shell_2 shell_3 shell_4 shell_5

SourceStamp:

BuildSlave:

Reason:

The SingleBranchScheduler scheduler named 'schedule-ffmpegsos-solaris10-sparc' triggered this build

Steps and Logfiles:

1. git update ( 19 secs )
   1. stdio
2. shell 'gsed -i ...' ( 1 secs )
   1. stdio
3. shell_1 'gsed -i ...' ( 2 secs )
   1. stdio
4. shell_2 'gsed -i ...' failed ( 1 secs )
   1. stdio
5. shell_3 './configure --samples="../../../ffmpeg/fate-suite" ...' failed ( 17 secs )
   1. stdio
   2. config.log
6. shell_4 'gmake fate-rsync' failed ( 1 secs )
   1. stdio
7. shell_5 '../../../ffmpeg/fate.sh ../../../ffmpeg/fate_config_sos.sh' failed ( 1 secs )
   1. stdio
   2. configure.log
   3. compile.log
   4. test.log

Build Properties:

Forced Build Properties:

Responsible Users:

1. Charly Morgand-Poyac

   charly.morgand-poyacohnoyoudont@bbright.com

Timing:

All Changes:

:

1. Change #286041

   Category	ffmpeg
   Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
   Changed at	Wed 14 Oct 2026 15:35:57
   Repository	https://git.ffmpeg.org/ffmpeg.git
   Project	ffmpeg
   Branch	master
   Revision	05c6ee22be85faa30dbd08a378e85f3e8ee86093

   Comments

   lavc/hevcdec: skip boundary strength computation for identical MvFields
   deblocking_boundary_strengths() runs on the full 8x8 grid without
   knowing the PU partitioning, so on inter content the two MvField
   entries it compares are identical most of the time. When they are
   identical and both sides use the same RefPicList the result is always
   0: same references, zero mv deltas. Check for this first, and compare
   whole edge lines at once in the TU internal loops; anything that does
   not match goes through the unchanged code path.
   
   Up to 3% faster single threaded decoding on 4K inter streams.
   
   Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

   Changed files

   • libavcodec/hevc/filter.c

2. Change #286044

   Category	ffmpeg
   Changed by	Charly Morgand-Poyac <charly.morgand-poyacohnoyoudont@bbright.com>
   Changed at	Wed 14 Oct 2026 15:38:50
   Repository	https://git.ffmpeg.org/ffmpeg.git
   Project	ffmpeg
   Branch	master
   Revision	21441b3400342e1d4d2273f28bddd2368fd0cca4

   Comments

   avcodec/hevc/filter: fix deblocking offsets taken from the wrong CTB
   Equations 8-348, 8-350 and 8-383 take slice_beta_offset_div2 and
   slice_tc_offset_div2 from the slice that contains the q0,0 sample of the
   edge: the current CTB for a vertical edge, and for a horizontal one the CTB
   that contains the sample group.
   
   deblocking_filter_CTB() reused whatever the previous loop had left in
   tc_offset and beta_offset instead. Vertical edges could take the left CTB's
   offsets, and each horizontal chroma sample group mixed the offsets of both
   CTBs.
   
   This only changes the output when a CTB and its left neighbour come from
   slices with different deblocking offsets.
   
   Fixes: 65746bfbae3d1077656b48820f0c18c5703e34f7 (hevc_filter: run vertical and horizontal together)
   Fixes: fa0506e4136b5b883205c82809a113f5edcd200b (hevc/rext: update boundary strength for chroma deblocking filter)
   Signed-off-by: Charly Morgand-Poyac <charly.morgand-poyac@bbright.com>

   Changed files

   • libavcodec/hevc/filter.c