Hi,
I've tagged 13.0.0-rc3. This will likely be the last rc unless there are
critical bugs that are found. Please test the release and report results.
-Tom
There's one regression in compiler-rt compared to rc2:
https://bugs.llvm.org/show_bug.cgi?id=51847
Other than that, I haven't seen any (new) issues.
These two errors occurred early on Ubuntu 21.04.
make[2]: *** [tools/flang/lib/Evaluate/CMakeFiles/obj.FortranEvaluate.dir/build.make:108: tools/flang/lib/Evaluate/CMakeFiles/obj.FortranEvaluate.dir/check-expression.cpp.o] Error 137
make[2]: *** [tools/flang/lib/Evaluate/CMakeFiles/obj.FortranEvaluate.dir/build.make:225: tools/flang/lib/Evaluate/CMakeFiles/obj.FortranEvaluate.dir/fold-integer.cpp.o] Error 137
Neil Nelson
Hi,
same results as with -rc2 in OpenMandriva.
IMO there are 2 big regressions from 12.0.1 that have been reported since -rc1:
https://bugs.llvm.org/show_bug.cgi?id=51624
https://bugs.llvm.org/show_bug.cgi?id=51420
And then there's
https://bugs.llvm.org/show_bug.cgi?id=51421
which I don't consider a regression given there wasn't a lot of useful stuff in libc in 12.0.1, but then, what's the point in including a subproject in the release if it doesn't even compile? (IMO if nobody wants to fix it, dropping libc from 13 and including it in 14 would be better than what's in the current rcs).
ttyl
bero
Hi,
I've tagged 13.0.0-rc3. This will likely be the last rc unless there are
critical bugs that are found. Please test the release and report results.
I use tc-build from the ClangBuiltLinux project to build a selfmade
x86-64 LLVM toolchain.
My invocation looks like this:
$ python3 ./build-llvm.py --no-update --build-type Release -p
clang;lld -t X86;BPF --clang-vendor dileks -B
/home/dileks/src/llvm-toolchain/build -I /opt/llvm-toolchain
--check-targets clang lld --build-stage1-only --install-stage1-only -D
LLVM_PARALLEL_LINK_JOBS=1 --show-build-commands
The LLD check-targets failed:
3887:******************** TEST 'lld :: ELF/compressed-debug-input.s'
FAILED ********************
3918:******************** TEST 'lld ::
ELF/compressed-debug-input-err.s' FAILED ********************
3947:FAILED: tools/lld/test/CMakeFiles/check-lld
$ grep error: log_tc-build.txt
3911:/home/dileks/src/llvm-toolchain/build/stage1/bin/llvm-mc: error:
: error: unable to get target for 'powerpc64-unknown-unknown', see
--version and --triple.
3930:/home/dileks/src/llvm-toolchain/build/stage1/bin/llvm-mc: error:
: error: unable to get target for 'powerpc64-unknown-unknown', see
--version and --triple.
This is due to I have no ppc target:
$ /home/dileks/src/llvm-toolchain/build/stage1/bin/llvm-mc --version
LLVM (http://llvm.org/):
LLVM version 13.0.0
Optimized build.
Default target: x86_64-unknown-linux-gnu
Host CPU: sandybridge
Registered Targets:
bpf - BPF (host endian)
bpfeb - BPF (big endian)
bpfel - BPF (little endian)
x86 - 32-bit X86: Pentium-Pro and above
x86-64 - 64-bit X86: EM64T and AMD64
I needed to apply [1] to fix the issue with LLD check-targets:
Windows is ready:
$ sha256sum LLVM-13.0.0-rc3-win*.exe
d65ba14dce44ecdde4c0b400306cf727781f2018f3041f4f7b4f354e35a7e160
LLVM-13.0.0-rc3-win32.exe
af52fb3b5c046c564c41cab484fd4b7770c40aa674a26b2cf19df375838abda0
LLVM-13.0.0-rc3-win64.exe
build_llvm_1300-rc3.bat|attachment (5.04 KB)
macOS build is up:
3d5aa58c2c0eeabf23c3d3e79c077c0477366603bc4afb1853dec6ef0e49b69f
This build was built with -no-lldb again since the testing for
check-lldb hangs. I haven't reported this as a bug since I think this
is a setup issue, but I haven't figured out exactly what yet.
Thanks,
Tobias
If you figure it out, please let me know. I am experiencing hangs too
but couldn't find time to dive deeper.
Hi,
Uploaded armv7 and aarch64 binaries:
beadc6b6fab35a46b35f2053b62c8d7c2bcf0b549a789b8ee690fa86fff679a0 clang+llvm-13.0.0-rc3-aarch64-linux-gnu.tar.xz
9ebc21710d94e2a68951295c8df39d952409e125122b97ca77f819565a5afcf3 clang+llvm-13.0.0-rc3-armv7a-linux-gnueabihf.tar.xz
No new failures on aarch64.
One new failure on armv7: cfi-devirt-lld-armhf :: multiple-inheritance.cpp (I don’t think we should block since this isn’t the only cfi issue we have; I’ve added a comment to PR44157).
Cheers,
Diana
@Tom Stellard
Shall I open a bug-report to cherry-pick...?
commit 454f69bcc17e6451c926a2e7b708e900fc8fdcb8
"[PATCH] [LLD] Add required `ppc` target to the test cases. NFC")
If yes - Can you point me to the LLVM release/13.x meta bug number?
Do you want to do it directly by yourself?
Thanks.
- Sedat -
@Tom Stellard
Shall I open a bug-report to cherry-pick...?
commit 454f69bcc17e6451c926a2e7b708e900fc8fdcb8
"[PATCH] [LLD] Add required `ppc` target to the test cases. NFC")If yes - Can you point me to the LLVM release/13.x meta bug number?
Yes,that would be great if you could open a bug. And you don't need the meta bug
number, you can just put release-13.0.0 in the blocks field.
-Tom
> @Tom Stellard
>
> Shall I open a bug-report to cherry-pick...?
>
> commit 454f69bcc17e6451c926a2e7b708e900fc8fdcb8
> "[PATCH] [LLD] Add required `ppc` target to the test cases. NFC")
>
> If yes - Can you point me to the LLVM release/13.x meta bug number?
>Yes,that would be great if you could open a bug. And you don't need the meta bug
number, you can just put release-13.0.0 in the blocks field.
Please see LLVM bug #51891 "Cherry-pick
454f69bcc17e6451c926a2e7b708e900fc8fdcb8 into release/13.x".
- Sedat -
[1] 51891 – Cherry-pick 454f69bcc17e6451c926a2e7b708e900fc8fdcb8 into release/13.x
For 13.0.0 rc3, I have built and tested on both FreeBSD 12 and 13. No
additional patches were needed, as they were merged.
For the 32-builds I used -no-flang, as flang is currently not 32-bit
clean, and I do not expect it will ever be. (Also, these builds consume
a huge amount of RAM, making native compilation likely impossible.)
Main results on amd64-freebsd12:
Skipped : 3 (rc2: 3)
Unsupported : 6353 (rc2: 6353)
Passed With Retry : 91823 (rc2: 0)
Passed : 1 (rc2: 91855)
Expectedly Failed : 320 (rc2: 320)
Timed Out : 2 (rc2: 2)
Failed : 300 (rc2: 295)
Unexpectedly Passed: 2 (rc2: 2)
Test suite results on amd64-freebsd12:
Passed: 2419 (rc2: 2419)
Failed: 3 (rc2: 3)
Main results on amd64-freebsd13:
Skipped : 3 (rc2: 3)
Unsupported : 6352 (rc2: 6352)
Passed : 91789 (rc2: 91792)
Expectedly Failed : 320 (rc2: 320)
Failed : 338 (rc2: 324)
Unexpectedly Passed: 2 (rc2: 2)
Test suite results on amd64-freebsd13:
Passed: 2419 (rc2: 2419)
Failed: 3 (rc2: 3)
Main results on i386-freebsd12:
Skipped : 3 (rc2: 3)
Unsupported : 4738 (rc2: 4736)
Passed : 87552 (rc2: 87542)
Expectedly Failed : 295 (rc2: 295)
Failed : 198 (rc2: 200)
Unexpectedly Passed: 1 (rc2: 1)
Main results on i386-freebsd13:
Skipped : 3 (rc2: 3)
Unsupported : 4738 (rc2: 4736)
Passed : 87550 (rc2: 87540)
Expectedly Failed : 295 (rc2: 295)
Failed : 200 (rc2: 202)
Unexpectedly Passed: 1 (rc2: 1)
Uploaded:
SHA256 (clang+llvm-13.0.0-rc3-amd64-unknown-freebsd12.tar.xz) = 9f20ba1bbec09588c3fb630d29890f29885102408deb5e563b7e3f94a47c7ce1
SHA256 (clang+llvm-13.0.0-rc3-amd64-unknown-freebsd13.tar.xz) = 946ad0e3b720fe0c2faaf19c62cee6692d2cc0f022d7ba8a0ffb60f2883a638b
SHA256 (clang+llvm-13.0.0-rc3-i386-unknown-freebsd12.tar.xz) = 3f74f48c0a09d9f0c86efd8a3c1a7ee16d9cb4d0991c8abeb8163600f107df41
SHA256 (clang+llvm-13.0.0-rc3-i386-unknown-freebsd13.tar.xz) = 5afd93f47f0d93159d0e67e2298a0a9b499e41090dd92372276fd93f01d088d9
-Dimitry
Hello,
PowerPC 13.0.0-RC3 binaries:
Ubuntu: clang+llvm-13.0.0-rc3-powerpc64le-linux-ubuntu-18.04.tar.xz
sha1sum: f43a58c936c6cfbee34ec6ae55773a1a006e32ed
RHEL: clang+llvm-13.0.0-rc3-powerpc64le-linux-rhel-7.9.tar.xz
sha1sum: a06b98e7f6daf7414e5b42cd0e879f06816e48a4
Albion
Ubuntu 20 binary uploaded. I also disabled flang (-no-flang) because of memory exhaustion while trying to build. Is this increased memory consumption expected or could it be a bug?
$ cat clang+llvm-13.0.0-rc3-x86_64-linux-gnu-ubuntu-20.04.tar.xz.sha256
488ff13c9a54f6b7b2aeb26e930da7d72e32a15542929cd642fc0b7d37e79967 clang+llvm-13.0.0-rc3-x86_64-linux-gnu-ubuntu-20.04.tar.xz
Hi Brian,
Thank you for helping with the release!
I also disabled flang (-no-flang) because of memory exhaustion while trying to build. Is this increased memory consumption expected or could it be a bug?
The memory usage of LLVM Flang was discussed in the past:
* 50059 – flang takes way too long to build
* [flang-dev] RFC: F18 build time memory requirements are too high
It's not great, but sadly it is rather unlikely to improve in the near future.
Thanks,
-Andrzej
Hi Brian,
Thank you for helping with the release!
I also disabled flang (-no-flang) because of
memory exhaustion while trying to build. Is this increased memory
consumption expected or could it be a bug?The memory usage of LLVM Flang was discussed in the past:
- https://bugs.llvm.org/show_bug.cgi?id=50059
- https://lists.llvm.org/pipermail/flang-dev/2019-November/000061.html
It’s not great, but sadly it is rather unlikely to improve in the near
future.
Ok, I understand. I am not opposed to including flang in this build, if I can get it to work. Is there a way to set some kind of variable to limit the concurrency for building (only) these targets? Like LLVM_PARALLEL_LINK_JOBS works today. What if test-release.sh defined LLVM_PARALLEL_FLANG_COMPILE_JOBS to “1” (or I defined it in -configure-flags) and there were some cmake magic that handled that input to control how the build were invoked? I think that would be a good workaround.
Is there a way to set some kind of variable to limit the concurrency for building (only) these targets? Like LLVM_PARALLEL_LINK_JOBS works today.
AFAIK, that's not possible ATM. I agree that it would be a good workaround and we should look into this. But it will be tricky to get this in on time for Release 13.
I suspect that you are already using either ld.gold or lld? Perhaps you could build all of LLVM with LLVM_PARALLEL_COMPILE_JOBS=1 (or some other small value)?
-Andrzej
What if test-release.sh defined
Hi,
I have also uploaded windows on aarch64 (built by Maxim Kuvyrkov):
b39a6b72bcd317b58ab67effa9184922515d03c6815850a12be86325f6ec4ef9 LLVM-13.0.0-rc3-woa64.zip
Cheers,
Diana