thx for fix.

i am seeing the same assert on the amdgpu target

Hi Peter

@saiislam for awareness

i forwarded a couple buildbot failures to Amit to look into

i pushed it in. will watch for build bot issues

https://github.com/llvm/llvm-project/issues/54618

thanks for suggestion: https://reviews.llvm.org/D122658

having looked at the generated code for g++ and clang++ , i agree this does not and will not fix the underlying issue.
abandoning

seeing some buildbot failures ...
https://lab.llvm.org/buildbot/#/builders/165/builds/18307
https://lab.llvm.org/buildbot/#/builders/193/builds/9008

The patch passed internal CI.

i applied it locally and ran bot equiv, looks good to me

Hi Johannes,
assert in two of the tests

Hi Alexey
with this patch, i noticed an assert building one of our runtime files , the test case .c produced is around 24000 lines
would you like it as is? or reduced ?

Hi Sam
i think this breaks our amdgpu buildbot
https://lab.llvm.org/buildbot/#/builders/193/builds/8513

breaking HIP and AMDGPU buildbots
https://lab.llvm.org/buildbot/#/builders/193/builds/8433
https://lab.llvm.org/buildbot/#/builders/165/builds/17691

FWIW, only the top commit causes the hangs, the 3 predecessor commits seem ok.
noticed it while i started looking in more detail.

This set of commits seems to have broken the amdgpu buildbot, test(s) are hanging.
see

https://lab.llvm.org/buildbot/#/builders/193/builds/8002

and a force run to see if the previous hang was spurious, seems reproducible.

https://lab.llvm.org/buildbot/#/builders/193/builds/8003

broke our hip buildbot as well.

lgtm

Hi Joseph,
seems to have broken amdgpu buildbit,
please revert and fix ?
https://lab.llvm.org/buildbot/#/builders/193/builds/6939

Hi
Looks like hip buildbot breaks from this
https://lab.llvm.org/buildbot/#/builders/165/builds/15363

Thx
Much needed!

seeing same breakage on https://lab.llvm.org/buildbot/#/builders/165/builds/14870
and https://lab.llvm.org/buildbot/#/builders/193/builds/5889

i dont see the failures locally, so i am now getting on the buildbot system to see what happens

i will give it a try locally ...

Hi Joseph,
3 commits came in, one bot exited with a fail
https://lab.llvm.org/buildbot/#/builders/193/builds/4828

seeing a buildbot failure on ;: https://lab.llvm.org/buildbot/#/builders/193/builds/4748

LGTM,

thx for quick fix

Hi Shilei
seems like this broke our amdgpu openmp buildbot
https://lab.llvm.org/buildbot/#/builders/193/builds/4072

Hi Philip
This seems to break our amdgpu buildbot https://lab.llvm.org/buildbot/#/builders/193/builds/4050

[AMD Official Use Only]

hi Ram, landed patch breaks https://lab.llvm.org/buildbot/#/builders/193/builds/3514

Our conclusion is that the /opt/rocm weak default needs to stay in place. Thus this patch should be abandoned in my opinion. MarkS and I are requesting the folks that requestedt his change within AMD, revisit it.

This needs to be revisited.

@estewart08 thoughts on a good CMAKE variable to allow users to define equivalent of /opt/rocm ? and not use environment variable inside the cmake file.

i see the same failures in my local build, and when i revert this commit, and rebuild/test, failures go away

LGTM, but needs Matt or Stas or Brian ...

fyi: if i remove the now 'dead' use of the CMAKE variable from my cmake , then i am able to build. so i see you reverted, thx. i guess some coordination amongst buildbot maintainers who might use this option

Hi Simon,
i am seeing a failure in our amdgpu buildbot after this patch . https://lab.llvm.org/staging/#/builders/200/builds/1407
we do depend on the cmake flag you removed.
we specify this
-DCLANG_DEFAULT_LINKER=lld

yes, thanks, sounds good

thx for trying it out, please revert so we can look into it more on the AMDGPU target

based on Shilei's last comment, do it in the morning ?

works for me, i think Greg is ok with it too, we chatted internally an hour or so ago

perhaps we can try this patch as is, and if it passes buildbot, let the new DeviceRTL be the default upstream for all targets.
if it fails the AMDGPU buildbot, then perhaps apply the above suggested change of leaving old runtime default for now for AMD.
or consider some other soultion.

Both of our amdgpu buildbots are broken by this commit, when I revert it locally my build will pass.
https://lab.llvm.org/buildbot/#/builders/193
https://lab.llvm.org/buildbot/#/builders/165

simpler approach like flang per MKruse does seem more attractive for this test. agree.

i backed up to your reverted patch, and applied this one.
I see some new errors

[AMD Official Use Only]

[AMD Official Use Only]

looks like subsequent builders passed. we might have an issue where the runtime was not fully built when the test ran.
please disregard , your patch is fine.

lld: error: undefined symbol: __kmpc_distribute_static_init_4

referenced by /tmp/private_mapping-6f3f07-gfx906-d001f0.o:(omp_outlined)
referenced by /tmp/private_mapping-6f3f07-gfx906-d001f0.o:(omp_outlined)

lld: error: undefined symbol: __kmpc_distribute_static_fini

referenced by /tmp/private_mapping-6f3f07-gfx906-d001f0.o:(omp_outlined)
referenced by /tmp/private_mapping-6f3f07-gfx906-d001f0.o:(omp_outlined)
referenced by /tmp/private_mapping-6f3f07-gfx906-d001f0.o:(__omp_outlined__.1)
referenced 1 more times

clang-14: error: amdgcn-link command failed with exit code 1 (use -v to see invocation)

apologies, that pasted very poorly ...

hi, we seem to have a link time error after this patch in our buildbot

@pdhaliwal
i will pass the problem over to Pushpinder Singh who should be waking up soon.