[Mlir-commits] [mlir] [mlir] Let GPU ID bounds work on any FunctionOpInterfaces (PR #95166)

[Jakub Kuderski]

kuhar wrote:

I'd be leaning towards making this an inherent attribute over `gpu.func` but also allow this as a discardable attribute over other function ops. This is supported by the existing support over `llvm.func`. I don't have any strong options wrt to the sequencing here.

https://github.com/llvm/llvm-project/pull/95166