[6.0.0 Release] Release Candidate 2 tagged

Dear testers,

There's been a lot of merges since rc1, and hopefully the tests are in
a better state now.

6.0.0-rc2 was just tagged, after r324506.

Please test, let me know how it goes, and upload binaries.

Thanks,
Hans

All zig tests passed with 6.0.0-rc2.

I noticed that my ELF debug-info loading code no longer finds source information, but this was also true for 6.0.0-rc1, and it’s probably a dwarf feature that my code should be updated to support. I’ll investigate and report back.

Windows binaries are ready:
$ sha1sum LLVM-6.0.0-rc2-win*.exe
53440640d7218497e2df43be4f31c08ace953bb3 LLVM-6.0.0-rc2-win32.exe
cd8332062e25e03d2a4ed9a2c10613569a62633a LLVM-6.0.0-rc2-win64.exe

They were built with the attached batch file.

build_llvm_600.bat|attachment (4.64 KB)

Hi,

If i were to thest them, where would i find them?

Thanks Franz

FYI, unofficial CentOS6 RPMs are there.
https://github.com/llvm-project/llvm-project-20170507/releases/tag/release_600_rc2

Hi Franz,

I assume you mean the Windows binaries? They will be published here in
a few days: LLVM 6.0.0 Release Candidates There will be a "RC2
binaries available" email send to the lists.

Thanks,
Hans

Built, tested and uploaded:

SHA256 (clang+llvm-6.0.0-rc2-amd64-unknown-freebsd10.tar.xz) = 1334db5949ec0c78cd6e9b798f0b397fe9e99240d98bcfc834e4410258b98372
SHA256 (clang+llvm-6.0.0-rc2-i386-unknown-freebsd10.tar.xz) = a7c0b1e4cfe3d608ee77472c7f17fe6b9493c7259aca80fcdd3b8a4fe49ad92d

On amd64-freebsd10 there were 896 unexpected test failures:

  Expected Passes : 45004
  Expected Failures : 186
  Unsupported Tests : 2939
  Unexpected Failures: 896

On i386-freebsd10 there were 619 unexpected test failures:

  Expected Passes : 43849
  Expected Failures : 195
  Unsupported Tests : 1954
  Unexpected Failures: 619

At least the i386 version looks quite a bit better, down from 1773 to 619 test failures. :slight_smile:

-Dimitry

Uploaded ARM & AArch64:

929794f931a075b9981ac0b3828b74ff0a78c541
clang+llvm-6.0.0-rc2-aarch64-linux-gnu.tar.xz
e1bebfc9c6dbbbdac3be0d629850442a0418b242
clang+llvm-6.0.0-rc2-armv7a-linux-gnueabihf.tar.xz

The good news is that the previous issues have been fixed, the bad
news is there are two new ones:
PR36313 on AArch64 (failures in openmp tests)
PR34182 on ARM (libcxxabi, we've seen this before in 5.0.0 but it was
reverted on the release branch; I'm really sorry that I missed this in
rc1, it got lost among all the tautological compare failures)

Cheers,
Diana

Hi,

What means "uploaded" (uploaded where?), nothing available at
LLVM 6.0.0 Release Candidates yet?

Thx

Franz

Hello,

Dear testers,

There's been a lot of merges since rc1, and hopefully the tests are in
a better state now.

6.0.0-rc2 was just tagged, after r324506.

Please test, let me know how it goes, and upload binaries.

Looks great on Debian. Bug 36051 (arm64 build issue) has been fixed.

We did a rebuild of the Debian archive using 6.0rc1. I need to investigate the results
but for now, I haven't seen anything crazy except a bug with qmake [1] (but not a regression
on the clang side as I have the same issue with 5.0)

Cheers,
Sylvestre

[1] [QTBUG-62531] Configure fails to parse default search paths from compiler output while using ICC - Qt Bug Tracker

What are all these test failures? Does it seems like they have a
common root cause and do we have a bug for it?

It means uploaded to the staging area used by the release team.

I've published the binaries that we've got on the web page now.

Thanks,
Hans

Understood thx

Hi Franz,

We have a server where the release testers upload the binaries. This
is what we're all reporting on this thread. It doesn't mean the
binaries are public yet. As Hans explained earlier, he will send an
email when the binaries are made public on
http://prereleases.llvm.org/6.0.0/

I hope that explains things.

Regards,
Diana

Yes it does, thx.

W dniu śro, 07.02.2018 o godzinie 21∶51 +0100, użytkownik Hans Wennborg
via Release-testers napisał:

Dear testers,

There's been a lot of merges since rc1, and hopefully the tests are in
a better state now.

6.0.0-rc2 was just tagged, after r324506.

Please test, let me know how it goes, and upload binaries.

RC2 looks much better on Gentoo. The only remaining failures are LLDB
(as usual) and compiler-rt [1]. I've pushed the ebuilds without keywords
for brave users who want to play with them.

[1]:https://bugs.llvm.org/show_bug.cgi?id=36065

There are multiple issues, some of which might be easily fixable, others have been there since a long time, and might be harder to fix.

A full overview of the failures on amd64:

...

What are all these test failures? Does it seems like they have a
common root cause and do we have a bug for it?

...

The Clang Tools and Extra Tools Unit tests all appear to crash with:

   exception_ptr not yet implemented

This turns out to be caused by libc++ being compiled without -DLIBCXXRT. (In the FreeBSD base system build, we always add this option, so libc++ knows how to handle exceptions.)

In the libc++ CMakeFiles, it appears to be governed by LIBCXX_CXX_ABI_LIBNAME, but it isn't being set to the correct value of "cxxrt" on FreeBSD. I am going to try the following diff:

--- llvm.src/projects/libcxx/CMakeLists.txt
+++ llvm.src/projects/libcxx/CMakeLists.txt
@@ -135,6 +135,8 @@
   elseif (APPLE)
     set(LIBCXX_CXX_ABI_LIBNAME "libcxxabi")
     set(LIBCXX_CXX_ABI_SYSTEM 1)
+ elseif (CMAKE_SYSTEM_NAME MATCHES "FreeBSD")
+ set(LIBCXX_CXX_ABI_LIBNAME "libcxxrt")
   else()
     set(LIBCXX_CXX_ABI_LIBNAME "default")
   endif()

-Dimitry

... and unfortunately that didn't work, since the CMakeFiles are unable to find the libcxxrt headers:

CMake Warning at projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:67 (message):
  Failed to find cxxabi.h
Call Stack (most recent call first):
  projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:112 (setup_abi_lib)
  projects/libcxx/CMakeLists.txt:428 (include)

CMake Warning at projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:67 (message):
  Failed to find unwind.h
Call Stack (most recent call first):
  projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:112 (setup_abi_lib)
  projects/libcxx/CMakeLists.txt:428 (include)

CMake Warning at projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:67 (message):
  Failed to find unwind-arm.h
Call Stack (most recent call first):
  projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:112 (setup_abi_lib)
  projects/libcxx/CMakeLists.txt:428 (include)

CMake Warning at projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:67 (message):
  Failed to find unwind-itanium.h
Call Stack (most recent call first):
  projects/libcxx/cmake/Modules/HandleLibCXXABI.cmake:112 (setup_abi_lib)
  projects/libcxx/CMakeLists.txt:428 (include)

The problem is that on FreeBSD, these libcxxrt headers are installed into the same location as the base system's libc++ headers, which is /usr/include/c++/v1, and if I add that to the include path of libc++'s build, it will certainly cause conflicts.

Does anybody have a suggestion on how this could be avoided?

-Dimitry

Ok, this turned out to be easier than I thought. After applying https://reviews.llvm.org/D43166, the number of failed tests drops roughly by half (from 896 to 512):

  Expected Passes : 45381
  Expected Failures : 185
  Unsupported Tests : 2937
  Unexpected Passes : 1
  Unexpected Failures: 521

I am going to have a look at some other low hanging fruit, and I have also created a few PRs to merge test changes into 6.0.

-Dimitry