This is an archive of the discontinued LLVM Phabricator instance.

[AArch64] Implement floating point subreg copies
AbandonedPublic

Authored by jroelofs on Jul 12 2021, 5:16 PM.

Details

Diff Detail

Event Timeline

jroelofs created this revision.Jul 12 2021, 5:16 PM
jroelofs requested review of this revision.Jul 12 2021, 5:16 PM
Herald added a project: Restricted Project. · View Herald TranscriptJul 12 2021, 5:16 PM
paquette added inline comments.Jul 13 2021, 10:39 AM
llvm/test/CodeGen/AArch64/GlobalISel/postrapseudos-fmov-fp-subreg.mir
1 ↗(On Diff #358107)

This test should probably be under llvm/test/CodeGen/AArch64, not under GlobalISel.

jroelofs updated this revision to Diff 358400.Jul 13 2021, 1:00 PM

move test. support all of {hsdq}x{hsdq}

jroelofs abandoned this revision.Jul 14 2021, 10:06 AM

The motivating case for this turned out to be a bug in selection of G_UNMERGE_VALUES.

See: https://reviews.llvm.org/D105935