[clang] [compiler-rt] [clang-tools-extra] [llvm] [AMDGPU] Avoid hitting AMDGPUAsmPrinter related asserts for local functions at O0 (PR #72129)

[Joseph Huber via cfe-commits]

jhuber6 wrote:

> As a somewhat naive question, what would it take to turn off requiring codegen to be in SCC order? We seem to be the only target doing that. The comments on that line say something about function calls and noinline

I believe this is also the reason parallel codegen via `--lto-partitions` creates incorrect code, so if there were a way to avoid that it would be beneficial in other ways. I'm by no means an expert, but as far as I'm aware the SCC order is used for some resource scheduling.

https://github.com/llvm/llvm-project/pull/72129