[PATCH] D24623: AMDGPU: Implement memory model

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: lib/Target/AMDGPU/SIMemoryLegalizer.cpp:194-195
+    const MachineBasicBlock::iterator &MI) const {
+  if (!MI->hasOneMemOperand())
+    return false;
+
----------------
kzhuravl wrote:
> arsenm wrote:
> > If there are no memory operands, this should still work and be handled as conservatively as possible
> Do you see an issue with this approach?
Yes. If there are no memory operands, this could skip an atomic load.


https://reviews.llvm.org/D24623