[llvm] [AMDGPU] Capping max number of registers for indirect calls (PR #199765)

[Janek van Oirschot via llvm-commits]

https://github.com/JanekvO commented:

Sorry, I know it's still in draft/wip mode but thought I'd have a peek
Quick question: does this never happen for sgpr (and I guess, for any other propagative resource usage item like named barriers)?

https://github.com/llvm/llvm-project/pull/199765