This is an archive of the discontinued LLVM Phabricator instance.

[ARM] Fix PR37382: Don't optimize mul.with.overflow on thumbv6m.
ClosedPublic

Authored by pftbest on Jul 2 2018, 12:03 PM.

Diff Detail

Repository
rL LLVM

Event Timeline

pftbest created this revision.Jul 2 2018, 12:03 PM

Thanks a lot for the patch @pftbest!

Would it make sense to add a regression test with the reproducer in PR37382?

I've added a RUN line in overflow-intrinsic-optimizations.ll test, and it will cause the same exact assertion if tested without the fix. So I think adding another test would be redundant.

rogfer01 accepted this revision.Jul 2 2018, 12:50 PM

Ah, I didn't notice it was exactly the same. You're right no need to add it again.

LGTM!

This revision is now accepted and ready to land.Jul 2 2018, 12:50 PM
efriedma added inline comments.Jul 2 2018, 1:38 PM
test/CodeGen/ARM/overflow-intrinsic-optimizations.ll
85 ↗(On Diff #153764)

Instead of this CHECK-NOT, could you explicitly CHECK for the call to __aeabi_lmul?

pftbest updated this revision to Diff 153782.Jul 2 2018, 1:51 PM

Explicit check for __aeabi_lmul

efriedma accepted this revision.Jul 2 2018, 1:55 PM

LGTM

This revision was automatically updated to reflect the committed changes.