New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
GCC 13: EQ/Filters/Waveforms broken #11483
Comments
Did you post two issues in GitHub issue? This is hard to handle. Can you spit them? |
It's a single issue. Probably a duplicate of #11409, which only mentions the audible errors. But the errors are also visible in waveforms. Same root cause. The preconditions for reproducing it are described here, i.e. GCC 13 + |
Ah OK, so the visual issue in the waveforms happens due to the broken filter. In #11409 the command line of fildlib.c has some extra flags. It would be nice to see the original compiler invocation in your case as well, to check if there are any differences. Even though @hapst3r has compiled with OPTIMIZE=portable he experienced the filter issues. So there must be a subtle difference. |
|
|
Let's burn the C/C++ toolchain. After multiple rebuilds with previously wiping the With Clang it still works as expected. |
I have suggested to close #11409 cause your issue is more comprehensive and you seem to know your way around debugging much better than I do. Thus, I will try to contribute to this issue to the extent that I can and it makes sense. For now, the only thing I can meaningfully contribute is to echo that when using I did The steps I took
Afterwards, running If there is anything I can contribute to the resolution even given my non-existent c++ skills, feel free to tell me what to do :) |
The build with GCC 13 is broken. See also: <mixxxdj/mixxx#11483>
@uklotzde @hapst3r
|
The build process terminates prematurely: 230418-analyzerwavefromtest_2-build.log. Nonetheless, here's the ctest: 230418-analyzerwavefromtest_2-ctest.log After the |
It is pointless trying to run the tests if the build already failed. The huge test log only contains messages that the test executable was not found. |
I am not planning to spend more time and resources on analyzing these issues. Switching to Clang works for me for the time being. Keeping this ancient C code working is a huge technical risk for the future of Mixxx. |
@hapst3r I guess your 2.3 branch was not clean before
|
good news I guess, https://github.com/daschuer/mixxx/tree/analyzerwavefromtest_2 fails
|
I assume you'll want to conduct some tests on the results? |
It looks like I have found a GCC 13 bug. The -HUGE_VAL is optimized away which makes fidlib go wild. |
I have reported it upstream: |
And it is not a GCC 13 bug, especially when you use -ffast-math/-ffinite-math-only . |
This can be closed, right? |
Bug Description
Tested with both the old and new fidlib code from 2.3 (#11475 + #11478). Probably undefined behavior.
Built with
-DOPTIMIZE=native
and-DOPTIMIZE=portable
on Intel Skylake.All tests pass.
Updates
-DOPTIMIZE=native
works (but causes another issue: Clang 15/16 on Linux: MathUtilTest.Denormal (Failed) #11484)The text was updated successfully, but these errors were encountered: