michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.1.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
abdu has joined #ffmpeg-devel
kasper93 has quit [Ping timeout: 276 seconds]
minimal has joined #ffmpeg-devel
kasper93 has joined #ffmpeg-devel
abdu has quit [Quit: Client closed]
mkver has quit [Ping timeout: 265 seconds]
_whitelogger has joined #ffmpeg-devel
minimal has quit [Quit: Leaving]
Guest40 has quit [Ping timeout: 240 seconds]
user23 has quit [Remote host closed the connection]
realies9 has quit [Quit: Ping timeout (120 seconds)]
realies9 has joined #ffmpeg-devel
cone-729 has quit [Quit: transmission timeout]
Martchus_ has joined #ffmpeg-devel
Martchus has quit [Ping timeout: 268 seconds]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
tufei__ has joined #ffmpeg-devel
tufei_ has quit [Ping timeout: 264 seconds]
jamrial has quit []
Chagalle has joined #ffmpeg-devel
Chagall has quit [Ping timeout: 252 seconds]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
tufei__ has quit [Remote host closed the connection]
tufei__ has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
j45 has quit [Read error: Connection reset by peer]
abdu has joined #ffmpeg-devel
j45 has joined #ffmpeg-devel
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
abdu has joined #ffmpeg-devel
j45_ has joined #ffmpeg-devel
j45 has quit [Ping timeout: 276 seconds]
j45_ is now known as j45
j45 has quit [Changing host]
j45 has joined #ffmpeg-devel
abdu47 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
abdu47 has quit [Ping timeout: 240 seconds]
ngaullier has joined #ffmpeg-devel
<thardin>
pal: I see
<thardin>
I might be working with a client that wants j2k support in autumn. for now my plate is full
<ramiro>
haasn: is op_copy_flags() still needed? it seems they're always regenerated with ff_sws_op_list_update_comps(). I get no changes in output if I remove op_copy_flags().
<haasn>
oh, yes, it's unneeded now
<haasn>
btw, I changed the signature slightly again, I kept SwsOpExec immutable but just moved the x/y coords to the SwsOpFunc signature
<haasn>
more convenient that way anyway
<haasn>
and I changed num_blocks and num_lines to block_x_end and y_end
<ramiro>
haasn: I noticed the signature change. I'm battling with that commit right now :)
<ramiro>
btw do you run your tests with valgrind? sometimes I get leaked dither matrix, even before your latest change to add a new malloc.
<haasn>
I'm battling the tablegen infrastructure right now..
abdo35 has joined #ffmpeg-devel
abdo35 has quit [Client Quit]
<haasn>
is tablegen not intended for use outside avcodec?
<haasn>
at least the placement of tableprint.h would seem to suggest so
<haasn>
well it's header only so no harm in just including it
psykose has quit [Remote host closed the connection]
psykose has joined #ffmpeg-devel
<Lynne>
tablegen is evil
System_Error has quit [Remote host closed the connection]
<fflogger>
[editedticket] MasterQuestionable: Ticket #11587 ([avcodec] Problem decoding APV (Advanced Professional Video): "pbu_size too large") updated https://trac.ffmpeg.org/ticket/11587#comment:3
<pal>
JEEB: yes, the framework is the same for all J2K tests
<haasn>
do we have a PRNG besides avutil/lfg.h?
<haasn>
seems kinda outdated
<haasn>
now I want to microbench av_lfg vs xxh3
abdu has joined #ffmpeg-devel
abdu10 has joined #ffmpeg-devel
abdu has quit [Ping timeout: 240 seconds]
abdu10 has quit [Client Quit]
<jamrial>
jkqxz: re #11587, the last packet is apparently truncated, but there are decoding errors in previous complete packets
<Lynne>
haasn: its the most proper one we have
<Lynne>
we have some LFSRs for codecs
<jkqxz>
It works and has bit-identical output to the reference decoder on the multisymbol version.
<jkqxz>
Not sure I want to debug further.
<jamrial>
are the printed errors expected, then?
<jamrial>
just a faulty sample?
<jkqxz>
The errors are in the version which is currently there.
<jkqxz>
I have no idea why it is different, but my current state work.
<jkqxz>
*works.
<jamrial>
ah ok
ngaullier has quit [Remote host closed the connection]
kasper93 has quit [Ping timeout: 252 seconds]
Traneptora has quit [Quit: Quit]
kasper93 has joined #ffmpeg-devel
kasper93_ has joined #ffmpeg-devel
kasper93 has quit [Killed (tantalum.libera.chat (Nickname regained by services))]
kasper93_ is now known as kasper93
jamrial has quit []
jamrial has joined #ffmpeg-devel
kasper93 has quit [Quit: kasper93]
kasper93 has joined #ffmpeg-devel
kasper93 has quit [Client Quit]
kasper93 has joined #ffmpeg-devel
kasper93 has quit [Client Quit]
kasper93 has joined #ffmpeg-devel
kasper93 has quit [Client Quit]
kasper93 has joined #ffmpeg-devel
kasper93 has quit [Client Quit]
kasper93 has joined #ffmpeg-devel
kasper93 has quit [Client Quit]
kasper93 has joined #ffmpeg-devel
cone-182 has joined #ffmpeg-devel
<cone-182>
ffmpeg Mark Thompson master:1ad57a847e22: lavc: Add unit test for APV entropy decode
<cone-182>
ffmpeg Mark Thompson master:a3c8fba5da69: apv_decode: Multisymbol entropy decode
<cone-182>
ffmpeg Mark Thompson master:1753d41d4eca: fate: Add test for APV 400-10 profile
Traneptora has joined #ffmpeg-devel
<fflogger>
[editedticket] jkqxz: Ticket #11587 ([avcodec] Problem decoding APV (Advanced Professional Video): "pbu_size too large") updated https://trac.ffmpeg.org/ticket/11587#comment:4
<jkqxz>
michaelni: Is there any guidance on how to add a decoder to the fuzzing? I have tested the APV decoder with the noise bsf and fixed everything that found, but maybe the more directed fuzzing process would find other things.
<jamrial>
i think it's fuzzed automatically as long as it's enabled in the build
<jkqxz>
... and obviously I immediately find another edge case with ubsan after saying I've tested it.
<michaelni>
ossfuzz found 3 issues it seems, ill try to add you to them so you can see them or failing that ill mail you the testcases
<michaelni>
the new shit ossfuzz webapp seems to have no way for me to add anyone, will mail you the testcases
<cone-182>
ffmpeg Mark Thompson master:527d5eaec702: apv_decode: Discard invalid run codes earlier
<kierank>
12 seconds to load trac
<jkqxz>
Thank you. The multisymbol entropy change is large so we may need to wait for another iteration, but I will check everything you send.
<michaelni>
mail sent
<michaelni>
2 seconds to load trac here atm
<jkqxz>
Does this have the hash of the version tested?
<jkqxz>
I recognise the first one and think it was fixed by 2aa2095bb4, it would be helpful to have the confirmation that it was from before then.
<jkqxz>
Other two I do not recognise and will investigate further.
<jkqxz>
The thing which looks like a hash isn't valid (though maybe it carries local changes in the test repo).
<michaelni>
ive commanded ossfuzz to recheck all 3 issues, the revissions it lists look look like this: [2025-05-13 07:08:39 UTC] oss-fuzz-tworker2-cb6r: Progression task finished: Still crashes on latest revision r202504290628.
<michaelni>
so it seems to have tested today revission from 2 weeks ago
jarthur has joined #ffmpeg-devel
<jkqxz>
It looks like all three do not fail any more. Second is from the older apv demuxer before jamrial changed it to use the parser, third is probably the same underlying issue as the first.
<jkqxz>
I'll see if I can run some more myself, but I think this wants to wait for another run with the current version.