This is an archive of the discontinued LLVM Phabricator instance.

[GlobalISel][X86] Support bitwise operations : G_AND, G_OR, G_XOR
ClosedPublic

Authored by igorb on Jun 25 2017, 7:34 AM.

Diff Detail

Repository
rL LLVM

Event Timeline

igorb created this revision.Jun 25 2017, 7:34 AM
igorb retitled this revision from [GlobalISel][X86] Support bitwise operations : G_ADD, G_OR, G_XOR to [GlobalISel][X86] Support bitwise operations : G_AND, G_OR, G_XOR.Jun 25 2017, 7:35 AM
igorb edited the summary of this revision. (Show Details)
aymanmus accepted this revision.Jun 28 2017, 2:08 AM

LGTM.
+ comment on a minor issue.

test/CodeGen/X86/GlobalISel/and-scalar.ll
2 ↗(On Diff #103869)

Add 32-bit run maybe?
remove --check-prefix as you check only ALL

test/CodeGen/X86/GlobalISel/or-scalar.ll
2 ↗(On Diff #103869)

same here.

test/CodeGen/X86/GlobalISel/xor-scalar.ll
2 ↗(On Diff #103869)

same here

This revision is now accepted and ready to land.Jun 28 2017, 2:08 AM
igorb marked 2 inline comments as done.Jun 28 2017, 4:11 AM
igorb added inline comments.
test/CodeGen/X86/GlobalISel/and-scalar.ll
2 ↗(On Diff #103869)
  • to support 64bit operation on 32bit platform require separate patch to implement NarrowScalar action for AND/OR/XOR, not supported yet.
  • going to remove unused check-prefix.

Thanks.

This revision was automatically updated to reflect the committed changes.