[libclc] [libclc] Produce subnormal results for fmod on FP32 (PR #215768)

[via cfe-commits]

Lurie97 wrote:

Gentle ping.

This patch fixes FP32 fmod incorrectly flushing subnormal results to zero
when CL_FP_DENORM is supported.

The change has been validated against OpenCL CTS math_brute_force tests
and has also become one of the prerequisites for enabling CL_FP_DENORM
without CTS failures in Mesa/OpenCL.

@arsenm @wenju-he Could one of you take a look when you have a chance?

Thanks!

https://github.com/llvm/llvm-project/pull/215768