I'm not sure how to directly observe this invisible cache for a test.
Diff Detail
Diff Detail
Event Timeline
Comment Actions
I'm not sure how to directly observe this invisible cache for a test.
Only thing I can think of is to have the MachineVerifier check.
Comparing this with MIRParserImpl::setupRegisterInfo there also needs to be some handling for landing pads (getCustomEHPadPreservedMask(MF))...
Either way this change only makes things better. LGTM
Comment Actions
Which on the other hand doesn't really work either, since this information seems to only exist after register rewriting; and I don't think the verifier has a notion of running before/after a pass like the rewriter...