This is an archive of the discontinued LLVM Phabricator instance.

[ARM] Fix FP16 stack loads/stores for Thumb2 with frame pointer
ClosedPublic

Authored by olista01 on Mar 1 2019, 3:41 AM.

Details

Summary

The new addressing mode added for the v8.2A FP16 instructions uses bit 8 of the immediate to encode the sign of the offset, like the other FP loads/stores, so need to be treated the same way.

Diff Detail

Repository
rL LLVM

Event Timeline

olista01 created this revision.Mar 1 2019, 3:41 AM
Herald added a project: Restricted Project. · View Herald TranscriptMar 1 2019, 3:41 AM
SjoerdMeijer accepted this revision.Mar 1 2019, 3:52 AM

Looks like a straightforward fix to me.

test/CodeGen/ARM/fp16-frame-lowering.ll
4 ↗(On Diff #188874)

nit: perhaps omit "that"?

This revision is now accepted and ready to land.Mar 1 2019, 3:52 AM
olista01 marked an inline comment as done.Mar 1 2019, 6:18 AM
This revision was automatically updated to reflect the committed changes.