On Wed, Feb 24, 2016 at 09:20:11AM +0100, Clément Bœsch wrote: > On Tue, Feb 23, 2016 at 10:40:08PM -0300, James Almer wrote: > [...] > > That aside, note that these runtime erros happen with every test using md5 > > and > > don't make ubsan register them as failed. In the link above something else > > made > > it fail. It makes me wonder how many other ubsan errors are hidden this way. > > Ah this was because there were no option to do it differently when the > option was introduced. I can probably use > -fsanitize-undefined-trap-on-error or similar on error nowadays. Maybe I > should use that. >
So I tried to add this but it's simply causing illegal instructions
everywhere.
Starting program: /home/ux/ffbuild/ffmpeg -i
/home/ux/samples/big_buck_bunny_1080p_h264.mov -f null -
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/usr/lib/libthread_db.so.1".
ffmpeg version N-78754-g42c5e1c Copyright (c) 2000-2016 the FFmpeg developers
built with gcc 5.3.0 (GCC)
configuration: --toolchain=gcc-usan --disable-stripping
--extra-cflags=-fsanitize-undefined-trap-on-error
libavutil 55. 19.100 / 55. 19.100
libavcodec 57. 27.100 / 57. 27.100
libavformat 57. 26.100 / 57. 26.100
libavdevice 57. 0.101 / 57. 0.101
libavfilter 6. 37.100 / 6. 37.100
libswscale 4. 0.100 / 4. 0.100
libswresample 2. 0.101 / 2. 0.101
Program received signal SIGILL, Illegal instruction.
ff_h264_decode_nal (h=h@entry=0x7ffff7e61040, sl=sl@entry=0x1fbb820,
src=0x1f85409 "M@)\251\030\017",
dst_length=dst_length@entry=0x7fffffffdaac,
consumed=consumed@entry=0x7fffffffdaa8,
length=<optimized out>) at src/libavcodec/h264.c:261
261 if (!((~AV_RN64A(src + i) &
(gdb)
I don't think FATE will be very useful like this (it's not providing any
backtrace at runtime). Any suggestion is welcome. The box is using gcc 5.3.0.
--
Clément B.
signature.asc
Description: PGP signature
_______________________________________________ ffmpeg-devel mailing list [email protected] http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
