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

[Robert Imschweiler via llvm-branch-commits]

================
@@ -16476,12 +16476,12 @@ SITargetLowering::getRegForInlineAsmConstraint(const TargetRegisterInfo *TRI_,
 
   const TargetRegisterClass *RC = nullptr;
----------------
ro-i wrote:

No, this is actually important to support the asm label constraint "!i". This should fall through to the default case (imho) without causing `VT.getSizeInBits()` to crash.

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