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

Make jlm/Makefile.sub build system agnostic. #6

Merged
merged 1 commit into from
Dec 12, 2024

Conversation

Felix-Rm
Copy link
Collaborator

Until now jlm/Makefile.sub assumed that Unix Makefiles was configured as the default generator for Cmake. On systems where this was not the case, the build step failed. This PR switches over to using cmake --build for the build step instead of the previously hardcoded make. All targets are now also defined in a uniform way without the manual mkdir and cd calls.

Behavior on systems with CMAKE_GENERATOR='Unix Makefiles' should be unchanged. On other systems, the script will no longer fail but use the configured generator.

The changes were only tested for a subset of targets (llvm-build-opt, llvm-build-aa, llvm-build-andersen-agnostic and llvm-build-andersen-region-aware) as compilation times are quite long on my current system.

Note: A second PR / change on phate/jlm will be necessary to bump the commit hash in the scripts. A full test via the CI runner might also be possible at that point.

Until now 'jlm/Makefile.sub' assumed that Make was configured
as the default generator for Cmake. On systems where this was
not the case, the build step failed. This commit switches over
to using 'cmake --build' for the build step instead of the
previously hardcoded 'make'.
@Felix-Rm Felix-Rm requested a review from phate November 25, 2024 13:08
@Felix-Rm Felix-Rm self-assigned this Nov 25, 2024
@phate phate requested review from sjalander and haved November 25, 2024 18:11
@phate
Copy link
Owner

phate commented Nov 25, 2024

@haved It seems that one of the unit tests is failing for the andersen region-aware encoding. I am tempted to approve this PR though as the changes have nothing to do with the failing test.

@phate
Copy link
Owner

phate commented Nov 26, 2024

I just had a look, and it is the same test that fails for andersen region-aware on the jlm side: phate/jlm#526

@haved
Copy link
Collaborator

haved commented Nov 26, 2024

@phate
Ah, so it is non-deterministic, that explains why it got into the master branch in the first place. It should probably be removed until we actually fix it

@phate
Copy link
Owner

phate commented Nov 27, 2024

@haved Yes, it is non-deterministic. I would rather not remove it though and fix it instead. It is not mandatory, so that is all fine by now. Anyway, do you have any objections to the PR though?

Copy link
Owner

@phate phate left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@haved
Copy link
Collaborator

haved commented Nov 27, 2024

@phate @Felix-Rm LGTM, too

@phate
Copy link
Owner

phate commented Dec 12, 2024

@Felix-Rm Would you like to merge it?

@Felix-Rm Felix-Rm merged commit 5c3c0f6 into phate:jlm-with-llvm-18 Dec 12, 2024
3 of 5 checks passed
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.

4 participants