[clang] [llvm] [Driver][AMDGPU][HIP][SPIRV] Disable optimizations for AMDGCN SPIR-V (PR #154765)

[Matt Arsenault via llvm-commits]

https://github.com/arsenm commented:

https://godbolt.org/z/Y6vYeYvaW

This shows my main concern. We should not be skipping the mandatory passes on this path. The SPIRV consumer should not be taking on the responsibility of handling the IR-lowered-in-frontend features (mainly always-inline and the coroutine passes).

This is a debug flag, this usecase needs a new one. There is already the similarly named "-disable-llvm-optzns" but it appears to just be an alias right now.

Not sure what loop(loop-idiom-vectorize) or globaldce are doing there at -O0, those seem like bugs 



https://github.com/llvm/llvm-project/pull/154765