[Mlir-commits] [clang] [llvm] [mlir] [IR] Make semantics of strictfp consistent v2 (PR #211769)

[Ömer Sinan Ağacan]

================
@@ -26653,9 +26654,6 @@ point intrinsics must have the `strictfp` attribute either on the
 calling instruction or on the declaration or definition of the function
 being called.
 
-All function *definitions* that use constrained floating point intrinsics
-must have the `strictfp` attribute.
----------------
osa1 wrote:

Sorry I don't quite fully follow the question.. but to elaborate on my point, what this sentence was clarifying (which is not already clear from the rest of the `strictfp` section) is that `strictfp` is all about **constrainted fp intrinsic usage**, not about calling other `strictfp` functions.

You can call `strictfp` functions from non-`strictfp` one.

You can't call "constrained floating point intrinsics" from non-`strictfp` functions. (I mean you can, but you shouldn't, and maybe we should be checking this)

That's my understanding at least.

https://github.com/llvm/llvm-project/pull/211769