Previously unconditionally stored to the return value. This is
incorrect, we should only return if user value is non-null.
Details
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Can you please separate this patch out into two so that the null return value can be go in separately? I can then also add comments for the other changes in their own focused review.
Sorry for more work, but this change is required for proper functioning of https://reviews.llvm.org/D148291, which has already landed. So, can you just prepare patch for the nullptr handling and not make it depend on the attributes patch? Or, make the tests in the patch not use the detach functionality. Either way, we should land this patch with some urgency.
libc/test/integration/src/pthread/pthread_join_test.cpp | ||
---|---|---|
58 | We should not use randomized inputs/data in unit tests because we want them to be reproducible. |
I am accepting but have also requested changes to reduce the scope of testing even further. I will review your other patch before the end of this week and we can include all kinds of testing in that patch.
libc/test/integration/src/pthread/pthread_join_test.cpp | ||
---|---|---|
30 | Type names not specified by the C/C++ standard should be in UpperCase style: https://libc.llvm.org/dev/code_style.html#naming-style | |
72 | Since we do not currently honor attrs, can you move this part to your other patch? | |
112 | Just keep this in this patch. |
@sivachandra this appears to be causing: https://lab.llvm.org/buildbot/#/builders/257/builds/630
want me to revert this and the key commit? Unable to reproduce locally so not really sure how to debug.
It is not because of your change. Accidentally, we have not been running tests on the builders for some time. I fixed it today so it started showing up regressions. I have already fixed that particular failure you have pointed out: https://reviews.llvm.org/D148844.
We should not use randomized inputs/data in unit tests because we want them to be reproducible.