rpcs3
rpcs3 copied to clipboard
RPCS3
[Regression] No audio when buffering is enabled
Quick summary
Recent attempts to compile builds result in binaries where audio no longer plays at all when buffering is enabled.
Details
Issue occurs whether audio backend is Cubeb or FAudio.
An attempt to bisect failed, as all builds present this issue. However, the last build with successful audio was from commit feff2ba09e3b4bba230e90e968304f87590fac43. It was presumed that the update to SDL2-2.30.4 (commit 7367f72f4a817806a57bebcc54d5a94c80becd2d) may have been the cause, however the test system was updated to SDL2-2.30.5 with no positive results.
Build with regression
See above
Attach two log files
Working log: RPCS3_working.log
Not working log: RPCS3_not_working.log
Attach capture files for visual issues
No response
System configuration
- OS: Linux Mint 21.3 / kernel 6.5
- CPU: Intel i7-10870H
- GPU: Nvidia GTX 3070 / driver 545.29
Other details
No response
Does the used compiler (clang vs gcc) change your results? We've had the same issue in the past with gcc and switched the AppImage to clang because of it.
Thanks, that's interesting. I've been using Ninja, though I also tried with make but that too produced a build with no audio (the gcc compiler version is gcc-13.1.0).
I'll give clang a shot and post results, ~~though I've never tried that and the readme only shows how to build with gcc. Are the build instructions posted somewhere and I'm just missing them?~~ Nevermind, I was using -DCMAKE_C_COMPILER=/usr/bin/clang17 etc. which is standard and it wasn't working, but it turns out I just mistyped something.
So yeah, using clang-17 produces a build with working audio. I wonder if this issue is confined to specific gcc versions, would be good for someone with a later version like gcc-14.1.0 to confirm seeing this too. I also don't know if the root cause is with the compiler or RPCS3. In case of the former I'll close this, or someone else can, though it should be noted somewhere that clang is recommended for reasons like this.
I believe this was reported before. Thats why I'm always building using clang. I'll do a build with gcc 14 tomorrow and report back.
The older report: https://github.com/RPCS3/rpcs3/issues/14741 I switched my builds from clang to gcc and can't reproduce this on gcc 14.1 (Arch toolchain, 14.1.1+r58+gfc9fb69ad62-1).
The older report: #14741 I switched my builds from clang to gcc and can't reproduce this on gcc 14.1 (Arch toolchain, 14.1.1+r58+gfc9fb69ad62-1).
NixOS 24.05's package also suffers from this regression. Its GCC version is currently 13.2.0.
Does this mean we just have to wait it out before re-enabling the setting ? Nice.
Do you still have this issue? Still no repro for me, using clang19, and our AppImage builds are also using clang19.
Ah right, thanks for the reminder. I don't have the issue anymore,. building with clang and gcc-14 and I think that should be good enough at this point. Closing.