There have been some buildbot failures in the release/13.x branch, so
I'm going to delay -rc2 until Thursday to investigate.
-Tom
_______________________________________________
LLVM Developers mailing list
llvm...@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
Overall, aside from the issues mentioned for rc1 (new-ish components like MLIR, flang, libc failing to build) mostly looks good. We found one more regression in 13.0-rc2 (compared to 12.x - this was also in rc1, just went unnoticed) affecting PGO.
Doing something along the lines of
export LLVM_PROFILE_FILE=xyz-%p.profile.d
export CFLAGS="-O2 -fprofile-instr-generate"
export CXXFLAGS="-O2 -fprofile-instr-generate"
./configure
make
# Run the generated binaries in some expected ways, e.g. "make check"
llvm-profdata merge --output=xyz.profile xyz-*.profile.d
consistently (obviously with PIDs varying) results in
warning: xyz-670250.profile.d: malformed instrumentation profile data
warning: xyz-670257.profile.d: malformed instrumentation profile data
error: no profile can be merged
The *.profile.d files look ok at a first glance, and "file" recognizes them as "LLVM raw profile data, version 7".
Looks like only "llvm-profdata merge" is broken, using -fprofile-instr-use=<any of the individual files> seems to be ok.
Will file a bug report...
ttyl
bero
On Wednesday, August 25, 2021 08:12 CEST, Tom Stellard via Release-testers <release...@lists.llvm.org> wrote:
> Hi,
>
> There have been some buildbot failures in the release/13.x branch, so
> I'm going to delay -rc2 until Thursday to investigate.
>
> -Tom
>
> _______________________________________________
> Release-testers mailing list
> Release...@lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/release-testers