2011/6/18 Måns Rullgård <[email protected]>: > Alex Converse <[email protected]> writes: > >> On Thu, Jun 16, 2011 at 11:37 AM, Anton Khirnov <[email protected]> wrote: >>> Module: libav >>> Branch: master >>> Commit: 9ba38229e5bd0cf6201a8206b2d8be6335f45f46 >>> >>> Author: Anton Khirnov <[email protected]> >>> Committer: Anton Khirnov <[email protected]> >>> Date: Wed May 25 16:42:41 2011 +0200 >>> >>> cmdutils: add opt_default2(). >>> >>> It stores options in a dictionary to be passed to new open calls. >>> >>> It will replace opt_default once all the pieces are in place. >>> >> >> (for real this time) >> >> This seems to have broken: >> >> $ make fate-sipr-6k5 >> TEST sipr-6k5 >> stddev: 0.11 PSNR:115.41 MAXDIFF: 19 bytes: 967680/ 967680 >> >> $ ./ffmpeg -version >> ffmpeg version git-v0.7rc1-27-g9ba3822, Copyright (c) 2000-2011 the >> Libav developers >> built on Jun 17 2011 22:22:00 with clang 2.9 (tags/RELEASE_29/final) >> configuration: --enable-gpl --enable-libvorbis >> --cc=/opt/llvm/bin/clang >> --samples=/home/alex/Projects/libav/fate/fate-suite >> libavutil 51. 6. 0 / 51. 6. 0 >> libavcodec 53. 5. 0 / 53. 5. 0 >> libavformat 53. 1. 3 / 53. 1. 3 >> libavdevice 53. 0. 0 / 53. 0. 0 >> libavfilter 2. 4. 0 / 2. 4. 0 >> libswscale 1. 1. 0 / 1. 1. 0 >> libpostproc 52. 0. 0 / 52. 0. 0 >> ffmpeg git-v0.7rc1-27-g9ba3822 >> libavutil 51. 6. 0 / 51. 6. 0 >> libavcodec 53. 5. 0 / 53. 5. 0 >> libavformat 53. 1. 3 / 53. 1. 3 >> libavdevice 53. 0. 0 / 53. 0. 0 >> libavfilter 2. 4. 0 / 2. 4. 0 >> libswscale 1. 1. 0 / 1. 1. 0 >> libpostproc 52. 0. 0 / 52. 0. 0 >> >> $ /opt/llvm/bin/clang --version >> clang version 2.9 (tags/RELEASE_29/final) >> Target: x86_64-unknown-linux-gnu >> Thread model: posix > > It works fine with clang here. This failure is suspiciously similar to > the intermittent icc12 failure of the same test. I wonder if there's > something weird in the code somewhere, perhaps triggered by random > alignment changes. >
Agreed. "[PATCH 1/3] sipr: Use memmove() to copy overlapped buffers" seems to fix this for me but I'm not sure if I'm just getting a better roll on the rcg. --Alex _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
