[llvm] [AMDGPU] Fix isCanonicalized to require canonical input for NaN-propa… (PR #208733)

[Wooseok Lee via llvm-commits]

wooseoklee wrote:

> > v_frexp_mant, v_rcp, v_rsq, v_sqrt, v_log, v_exp, v_fract, and their intrinsic counterparts
> 
> How did you come up with this list?

They are provided by the bug report from the Semi Analysis. Sorry. It seems that I was lazy at verifying all instructions have such issue. As Matt mentioned, some of them might be false alarm. I only confirmed v_frexp_mant. I will review them again and update the code.

https://github.com/llvm/llvm-project/pull/208733