Open abique opened 2 weeks ago
This ticket as-is isn't really actionable as a compiler bug. It's vague as to what is broken, and isn't reduced to any specific code or behavior. This would be better treated as an ffmpeg bug at this level of description for reduction
This ticket as-is isn't really actionable as a compiler bug. It's vague as to what is broken, and isn't reduced to any specific code or behavior. This would be better treated as an ffmpeg bug at this level of description for reduction
I'm not sure, but I don't think it is an ffmpeg bug. Maybe we'll be lucky and an ffmpeg developer will investigate it and create a more precise bug report.
In the meantime it is good to have this issue documented here.
I'm not sure, but I don't think it is an ffmpeg bug. Maybe we'll be lucky and an ffmpeg developer will investigate it and create a more precise bug report.
Did you file it as an ffmpeg bug? ffmpeg developers won't be looking here for this
I'm not sure, but I don't think it is an ffmpeg bug. Maybe we'll be lucky and an ffmpeg developer will investigate it and create a more precise bug report.
Did you file it as an ffmpeg bug? ffmpeg developers won't be looking here for this
No because it's likely a compiler bug, not an ffmpeg bug. It is triggered by -fveclib=mvec
.
I've reported it to ffmpeg as well: https://trac.ffmpeg.org/ticket/11176#ticket
Hi,
I've been building ffmeg using both clang 18 and clang 19 and it resulted in the aac codec and the vorbis codec unable to decode correctly the audio.
The issue was triggered by
-fveclib=mvec
.I did install clang using the llvm script in a ubuntu 22.04 docker image. I did build ffmpeg using vcpkg and a custom triplet.
I imagine that building ffmpeg with
-O3 -fveclib=mvec
will be enough to trigger issues in their test suite.Cheers, Alex