Page MenuHomePhabricator

[Local] Do not introduce a new `llvm.trap` before `unreachable`
AcceptedPublic

Authored by jdoerfert on Mon, Jul 19, 1:40 PM.

Details

Summary

This is the second attempt to remove the llvm.trap insertion after
https://reviews.llvm.org/rGe14e7bc4b889dfaffb7180d176a03311df2d4ae6
reverted the first one. It is not clear what the exact issue was back
then and it might already be gone by now, it has been >5 years after
all.

Replaces D106299.

Diff Detail

Event Timeline

jdoerfert created this revision.Mon, Jul 19, 1:40 PM
jdoerfert requested review of this revision.Mon, Jul 19, 1:40 PM
Herald added a reviewer: baziotis. · View Herald Transcript
Herald added a project: Restricted Project. · View Herald Transcript
Herald added a subscriber: bbn. · View Herald Transcript
lebedev.ri accepted this revision.Mon, Jul 19, 1:45 PM
lebedev.ri edited the summary of this revision. (Show Details)

LG

This revision is now accepted and ready to land.Mon, Jul 19, 1:45 PM

I like this lots more than the abandoned patch. Thanks!

I like this lots more than the abandoned patch. Thanks!

Me too!

jdoerfert updated this revision to Diff 359987.Mon, Jul 19, 7:12 PM

Adjust tests, including 2 XFAILS.

I will XFAIL llvm/test/CodeGen/Thumb2/ifcvt-rescan-diamonds.ll as part of this. The test is full of UB and runs O2, it is not surprising it will eventually break. I tried to make the pointers arguments but that changed the output. The author (@iteratee) seem to be inactive by now and there was no review on phab, I think.
I made a similar comment here: https://reviews.llvm.org/D53450#2889172

You can probably repair the ARM/Thumb tests using -arm-atomic-cfg-tidy=0... not a great solution, but if it works, better than XFAILing the test.

jdoerfert updated this revision to Diff 360280.Tue, Jul 20, 2:38 PM

Replace XFAIL