Skip to content

Conversation

@dkurt
Copy link
Contributor

@dkurt dkurt commented Nov 22, 2022

related: llvm/llvm-project@974e2e6
resolves #7174

@steven-johnson
Copy link
Contributor

LGTM, please push one more commit to trigger the buildbots

@zvookin
Copy link
Member

zvookin commented Nov 28, 2022

Only question here is whether this should be conditional in the LLVM version. It looks like it was kind of always wrong. I guess the way to find out would be to make the change unconditional and see if if it still works with LLVM 15, but that may not be worth it. (The buildbots will not test that and I don't have an LLVM 15 tree handy.) Going to LGTM this.

@steven-johnson
Copy link
Contributor

LGTM once buildbots are green.

@dkurt
Copy link
Contributor Author

dkurt commented Nov 28, 2022

@zvookin, good point. Not sure if that will work but I can check with LLVM 15 as well. The less conditional compilation the better.

Actually, I used LLVM 15 when prepared #7166 so probably there was a different layout string.

@steven-johnson steven-johnson merged commit 3ff9e66 into halide:main Nov 28, 2022
@steven-johnson
Copy link
Contributor

Thanks for the fix!

@dkurt dkurt deleted the patch-1 branch November 29, 2022 07:03
ardier pushed a commit to ardier/Halide-mutation that referenced this pull request Mar 3, 2024
* Use n32:64 in RISC-V data layout

* Remove unused LLVM header
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[RISC-V] Data layout mismatch between module

3 participants