[PATCH] D85392: [WebAssembly][ConstantFolding] Fold fp-to-int truncation intrinsics

Mon Aug 10 13:20:01 PDT 2020

tlively added a comment.

Yes, most of the SIMD intrinsics could be folded as well. It’s certainly on my todo list.

================
Comment at: llvm/test/Analysis/ConstantFolding/WebAssembly/trunc.ll:160
+  store volatile i32 %t14, i32* %p
+  %t15 = call i32 @llvm.wasm.trunc.unsigned.i32.f32(float 0x7ff0000000000000); inf
+  store volatile i32 %t15, i32* %p
----------------
aheejin wrote:
> It'd probably be easier to read if there is a comment dividing the instructions that are supposed to be folded and ones that are not, such as
> ```
> ; Instructions below this line are not supposed to folded because they are nan or overflow/underflow
> ```
> or something. The same for other functions
Good idea! I will add these comments in a follow up. 

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D85392/new/

https://reviews.llvm.org/D85392