[llvm-branch-commits] [llvm] [AMDGPU][UnifyDivergentExitNodes][StructurizeCFG] Add support for callbr instruction with basic inline-asm (PR #152161)

[Sameer Sahasrabuddhe via llvm-branch-commits]

================
@@ -1284,13 +1283,13 @@ bool StructurizeCFG::makeUniformRegion(Region *R, UniformityInfo &UA) {
 
 /// Run the transformation for each region found
 bool StructurizeCFG::run(Region *R, DominatorTree *DT) {
-  if (R->isTopLevelRegion())
+  // CallBr and its corresponding blocks must not be modified by this pass.
----------------
ssahasra wrote:

I think this is saying that if a region is headed by a `CallBr`, then anything inside it is separately structurized by visiting the nested regions. Is that correct? Are there any assertions we could put here about the current region R, to demonstrate that there is nothing to structurize?

https://github.com/llvm/llvm-project/pull/152161