[5.0.0 Release] Release Candidate 2 tagged

Dear testers,

5.0.0-rc2 was just tagged.

I know we still have a bunch of open release blockers, but there has
been a lot of merged patches and I'd like to find out what the status
is.

Please build, test and upload binaries to the sftp. When uploading,
make sure to use the /data/testers-uploads/ directory on the server.
Let me know if there are any issues.

I'll upload sources, docs and your binaries to the pre-release website
once they're ready.

Cheers,
Hans

Much or all of “Bitcode/simd_ops/simd_ops_*.test” (254 failures) seem to be failing for me with SIGILL. I’m guessing that my host CPU doesn’t support the instructions we’re testing?

Is there already a bug on this failure? Are there any CPUID-style feature detection flags I can easily add to mark these tests UNSUPPORTED for, or would I have to write this detection feature?

I'm not aware of a bug filed for this, so please file one. I don't
know much about the workings of test-suite myself, though.

+CC Alina

Maybe the bitcode is generated too aggressively and we should find ways to disable the Bitcode tests if the host CPU isn't new enough (or alternatively disable the bitcode tests by default and only enable them when someone asks for them via TEST_SUITE_SUBDIRS?)

- Matthias

Opened https://bugs.llvm.org/show_bug.cgi?id=34168

rc2 uploaded to /home/testers/uploads/clang+llvm-5.0.0-rc2-linux-x86_64-sles11.3.tar.xz – I tried to use /data/testers-uploads/ but that failed.

292e39fe52d7404fcd5cd962cf8917600d45f507 rc2/clang+llvm-5.0.0-rc2-linux-x86_64-sles11.3.tar.xz

Hi,

Uploaded ARM and AArch64:

b0a84bc679946b733803b8a687e4f0fb84d2f643
clang+llvm-5.0.0-rc2-aarch64-linux-gnu.tar.xz
d5a275f470687beaf2aabc279926c4430da1c88e
clang+llvm-5.0.0-rc2-armv7a-linux-gnueabihf.tar.xz

AArch64 has the same failures as before (PR33858, libunwind tests, and
PR33972, allocator_oom_test).
ARM has the same failures as before (PR33858, libunwind tests) and
also a new one in libcxxabi, which I've reported as PR34182.

Cheers,
Diana

5.0.0-rc2 was just tagged.

I know we still have a bunch of open release blockers, but there has
been a lot of merged patches and I'd like to find out what the status
is.

As in the last rc, most of the ASan tests failed with either:

[ DEATH ] ==5754==AddressSanitizer CHECK failed: /home/dim/llvm-5.0.0/rc2/llvm.src/projects/compiler-rt/lib/asan/asan_errors.h:99 "((second_free_stack->size)) > ((0))" (0x0, 0x0)

or:

[ DEATH ] ==7514==AddressSanitizer CHECK failed: /home/dim/llvm-5.0.0/rc2/llvm.src/projects/compiler-rt/lib/asan/asan_descriptions.cc:176 "((id)) != (0)" (0x0, 0x0)

This is likely due to r305058, as mentioned in my report for rc1.

I could not upload my tarballs, since the upload disk appears to be full:

df -h

    Size Used Avail (root) %Capacity
   7.7GB 7.3GB 192KB 425MB 94%

So I have uploaded them to my own server, here:

http://www.andric.com/freebsd/clang/clang+llvm-5.0.0-rc2-amd64-unknown-freebsd10.tar.xz
http://www.andric.com/freebsd/clang/clang+llvm-5.0.0-rc2-i386-unknown-freebsd10.tar.xz

SHA256 (clang+llvm-5.0.0-rc2-amd64-unknown-freebsd10.tar.xz) = 8e27a2fe43785eff1927100c59bee3d662c3e75f28adf87cb8ce31253a5dd67f
SHA256 (clang+llvm-5.0.0-rc2-i386-unknown-freebsd10.tar.xz) = 421ccd60c22c751711fca0169f5c7461ef537a110e86c91d619e7d2f5b0452c5

-Dimitry

Windows is ready:

$ sha1sum *5.0.0-rc2*
7376bfd5d33fa4d4d8102d1469ad01f62686b8fb LLVM-5.0.0-rc2-win32.exe
66b92d054ed36caf96e0a584a43816f3ad79b5f7 LLVM-5.0.0-rc2-win64.exe

All green in Debian:

Except that I had to force the usage of gcc 6 instead gcc 7 because of

(other versions are also affected)

S

W dniu czw, 10.08.2017 o godzinie 19∶00 -0700, użytkownik Hans Wennborg
via cfe-dev napisał:

Dear testers,

5.0.0-rc2 was just tagged.

I know we still have a bunch of open release blockers, but there has
been a lot of merged patches and I'd like to find out what the status
is.

Please build, test and upload binaries to the sftp. When uploading,
make sure to use the /data/testers-uploads/ directory on the server.
Let me know if there are any issues.

I'm sorry for not replying earlier but I have much less time than I wish
I'd have. Nevertheless, I'm trying to regularly test the release_50
branch and fix the regressions Gentoo is seeing.

It seems that we've been able to fix most of the apparent issues.
The two things that still remain are:

a. lldb fails to build with (more) linking errors -- there are already
two related fixes in the trunk, and I'm currently working on a third one
and should have a complete set of patches to backport on the weekend,

b. all DFSan tests segfault for me -- I have tried to bisect it, and it
seems that all old versions started failing just the same all
of a sudden. I suspect this might be related to system upgrade but
haven't been able to find time to properly investigate. Hopefully I'd be
able to at least provide more data soon.

Thanks for all the hard work.

Hi,

There are no regressions from RC1 on Darwin.

I've uploaded clang+llvm-5.0.0-rc2-x86_64-apple-darwin.tar.xz to /data/testers-uploads.

SHA 1: 8ffaf7ab44e7e7fe582dc9573aad9745ee63e186
SHA 256: 578c18ceb6485fd2fcac329a7a5fe7e55d1c70669a942651d49cdc3ea626a5af

I've also resurrected the Darwin bot which tests the release branch (thanks to Mehdi for creating the original!):

  http://lab.llvm.org:8080/green/view/All/job/clang-stage1-configure-RA-release-5/

thanks,
vedant

Hi,

There are no regressions from RC1 on Darwin.

I've uploaded clang+llvm-5.0.0-rc2-x86_64-apple-darwin.tar.xz to /data/testers-uploads.

SHA 1: 8ffaf7ab44e7e7fe582dc9573aad9745ee63e186
SHA 256: 578c18ceb6485fd2fcac329a7a5fe7e55d1c70669a942651d49cdc3ea626a5af

Thanks! I've added them to the pre-release page.

I've also resurrected the Darwin bot which tests the release branch (thanks to Mehdi for creating the original!):

  http://lab.llvm.org:8080/green/view/All/job/clang-stage1-configure-RA-release-5/

That's awesome! Many thanks.

Was there ever any progress on this, or at least a bug report filed?
Do you think it's just the tests that are failing, or does it mean
ASan is broken on FreeBSD? How severe is this issue from your
perspective?

Thanks,
Hans

5.0.0-rc2 was just tagged.

I know we still have a bunch of open release blockers, but there has
been a lot of merged patches and I'd like to find out what the status
is.

As in the last rc, most of the ASan tests failed with either:

[ DEATH ] ==5754==AddressSanitizer CHECK failed: /home/dim/llvm-5.0.0/rc2/llvm.src/projects/compiler-rt/lib/asan/asan_errors.h:99 "((second_free_stack->size)) > ((0))" (0x0, 0x0)

or:

[ DEATH ] ==7514==AddressSanitizer CHECK failed: /home/dim/llvm-5.0.0/rc2/llvm.src/projects/compiler-rt/lib/asan/asan_descriptions.cc:176 "((id)) != (0)" (0x0, 0x0)

This is likely due to r305058, as mentioned in my report for rc1.

Was there ever any progress on this, or at least a bug report filed?

No progress, but I just filed https://bugs.llvm.org/show_bug.cgi?id=34324 so it hopefully won't get lost.

Do you think it's just the tests that are failing, or does it mean
ASan is broken on FreeBSD? How severe is this issue from your
perspective?

It looks like the tests are failing, but simple programs compiled with -fsanitize=address seem to work. I haven't tested it extensively, though.

I'm not completely happy with the situation, as we also have some major problems with AddressSanitizer on FreeBSD 12.0 (though this isn't the fault of compiler-rt). But there's no need to block the release for this.

-Dimitry