[SYCL] Support __builtin_printf for SYCL device

[srividya-sundaram]

No description provided.

[premanandrao]

Could you also a CodeGen test for this to check if we do the right thing after we allow this through in Sema?

[smanna12]

Could you also a CodeGen test for this to check if we do the right thing after we allow this through in Sema?

I agree with Prem.

[srividya-sundaram]

I see the ESIMD and CUDA LLVM Test Suite failures in other PRs too. They don't seem to be caused by changes in this PR.

[srividya-sundaram]

@intel/llvm-gatekeepers Hi, this is ready for merging. Thanks!

[steffenlarsen]

Failures in CI are seen on other PRs (for example #7563) and have been reported in #6524 (comment).

[pvchupin]

@srividya-sundaram, can you please follow up on Windows failure in post-commit for the new test you've added: https://github.com/intel/llvm/actions/runs/3576029259/jobs/6013328053

[srividya-sundaram]

@srividya-sundaram, can you please follow up on Windows failure in post-commit for the new test you've added: https://github.com/intel/llvm/actions/runs/3576029259/jobs/6013328053

Hi! Is there a way I can reproduce the post-commit failure locally on a Windows machine?
I do have a fix looking at the log file, but I want to make sure it does really fix the failure.

[premanandrao]

Is there a way I can reproduce the post-commit failure locally on a Windows machine?

I think you should be able to run the command-line on a Windows local build.

[srividya-sundaram]

@srividya-sundaram, can you please follow up on Windows failure in post-commit for the new test you've added: https://github.com/intel/llvm/actions/runs/3576029259/jobs/6013328053

@pvchupin Just curious. Would it have been possible to capture this failure before it got merged?
I am not aware of the details of the post-commit checks, but wondering if it would help to have such failures be captured sooner than later in the pipeline.

[premanandrao]

Would it have been possible to capture this failure before it got merged? I am not aware of the details of the post-commit checks, but wondering if it would help to have such failures be captured sooner than later in the pipeline.

Does it not show up in a check-clang test run on Windows?

[pvchupin]

@pvchupin Just curious. Would it have been possible to capture this failure before it got merged? I am not aware of the details of the post-commit checks, but wondering if it would help to have such failures be captured sooner than later in the pipeline.

We don't run windows in pre-commit (due to longer pre-commit time). So you could detect this only if run locally.
It's ok to address windows specific issues post-commit.

[elizabethandrews]

Would it have been possible to capture this failure before it got merged? I am not aware of the details of the post-commit checks, but wondering if it would help to have such failures be captured sooner than later in the pipeline.

Does it not show up in a check-clang test run on Windows?

Yes it would but I very rarely test my patch on Windows because I work on Linux and I suspect its the same for most developers who work on Linux. I don't think we're expected to do this anyway.

[premanandrao]

Yes it would but I very rarely test my patch on Windows because I work on Linux and I suspect its the same for most developers who work on Linux. I don't think we're expected to do this anyway.

No, I understand. I am that way too. I was worried initially that local runs on Windows were not showing the failure; but then I understood the question was more about the expected development/testing process.