[flang-commits] [flang] [flang][cuda] Flatten memref descriptors in GPU kernel argument packing (PR #193651)

[Zhen Wang via flang-commits]

wangzpgi wrote:

> is there an interface that could be used for that? This feels a bit weird to hardcode it this way. If memref lowering change then this will break as well.

Great point, maybe I can use `mlir::MemRefDescriptor::unpack` for this purpose.

https://github.com/llvm/llvm-project/pull/193651