optimization without intrinsic instructions?


i run my LLVM bitcode through some optimization, and this introduces some intrinsic functions into the resulted bitcode.this is not what i want, because in the post-analysis step i dont want to deal with instructions like @llvm.uadd.with.overflow.

so the question is: how can i tell the optimization process to not use these intrinsic instructions, or rather no intrinsic at all? which optimization pass i should not enable to achieve this?

thanks so much, and happy holidays to everybody :slight_smile: