Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Don't modify existing MDNodes in SIMDloop pass. #31691

Merged
merged 1 commit into from
Apr 12, 2019
Merged

Conversation

maleadt
Copy link
Member

@maleadt maleadt commented Apr 11, 2019

This has been fixed on master, but might be interesting to backport to 1.1 or even 1.0 depending on our plans for those branches. Fixes debug code emission with CUDAnative.jl, but the underlying problem is significant.

Fixes JuliaGPU/CUDAnative.jl#344

@maleadt maleadt added compiler:codegen Generation of LLVM IR and native code bugfix This change fixes an existing bug labels Apr 11, 2019
@maleadt maleadt requested a review from vchuravy April 11, 2019 16:07
src/llvm-simdloop.cpp Show resolved Hide resolved
@vchuravy
Copy link
Member

Thanks! Any good ideas for a test? How does this materialize?

@maleadt
Copy link
Member Author

maleadt commented Apr 12, 2019

I added a ref to a CUDAnative issue. It's hard to reproduce faithfully, e.g., running codegen for that function a second time resulted in valid debug info.

@maleadt
Copy link
Member Author

maleadt commented Apr 12, 2019

CI looks good, only BSD fails due to what's probably an issue with release-1.1

@maleadt maleadt merged commit 7132c63 into release-1.1 Apr 12, 2019
@maleadt maleadt deleted the tb/temp_mdnode branch April 12, 2019 11:45
@KristofferC KristofferC mentioned this pull request Apr 19, 2019
39 tasks
@KristofferC KristofferC mentioned this pull request Aug 26, 2019
55 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bugfix This change fixes an existing bug compiler:codegen Generation of LLVM IR and native code
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants