This is an archive of the discontinued LLVM Phabricator instance.

AMDGPU/R600: Use machine operands instead of ints to track literals
ClosedPublic

Authored by jvesely on May 1 2016, 9:51 PM.

Diff Detail

Event Timeline

jvesely updated this revision to Diff 55777.May 1 2016, 9:51 PM
jvesely retitled this revision from to AMDGPU/R600: Use machine operands instead of ints to track literals.
jvesely updated this object.
jvesely set the repository for this revision to rL LLVM.
tstellarAMD added inline comments.May 6 2016, 5:13 PM
lib/Target/AMDGPU/R600ControlFlowFinalizer.cpp
432

What is this value for?

jvesely added inline comments.May 6 2016, 6:12 PM
lib/Target/AMDGPU/R600ControlFlowFinalizer.cpp
432

Just a forgotten magic value during development (I wanted to make it different from 0 used below). It's replaced in the next patch (D19792).

jvesely added inline comments.May 13 2016, 5:22 AM
lib/Target/AMDGPU/R600ControlFlowFinalizer.cpp
432

should I change it to zero?

tstellarAMD accepted this revision.May 13 2016, 12:53 PM
tstellarAMD edited edge metadata.

LGTM.

lib/Target/AMDGPU/R600ControlFlowFinalizer.cpp
432

Sure.

This revision is now accepted and ready to land.May 13 2016, 12:53 PM
jvesely updated this revision to Diff 57234.May 13 2016, 1:10 PM
jvesely edited edge metadata.
jvesely removed rL LLVM as the repository for this revision.
jvesely added a subscriber: llvm-commits.

remove debugging magic value

jvesely marked 4 inline comments as done.May 13 2016, 1:11 PM
This revision was automatically updated to reflect the committed changes.