[PATCH] D154628: [1/8][RISCV] Add rounding mode control variant for vfsub, vfrsub

[Yueh-Ting (eop) Chen via Phabricator via cfe-commits]

eopXD marked an inline comment as done.
eopXD added inline comments.


================
Comment at: clang/lib/Sema/SemaChecking.cpp:4811
     return SemaBuiltinConstantArgRange(TheCall, 4, 0, 3);
   case RISCVVector::BI__builtin_rvv_vfadd_vv_rm_ta:
   case RISCVVector::BI__builtin_rvv_vfadd_vf_rm_ta:
----------------
craig.topper wrote:
> Not related to this patch, but why do the builtins end in _ta when the C intrinsics don't?
Yes this is a redundancy here, the `_ta` suffix should be removed. I will do this is a separate patch.


================
Comment at: llvm/lib/Target/RISCV/RISCVInstrInfoVPseudos.td:5916
 //===----------------------------------------------------------------------===//
-let mayRaiseFPException = true in {
+let mayRaiseFPException = true, hasSideEffects = 0 in {
 defm PseudoVFADD  : VPseudoVALU_VV_VF_RM;
----------------
craig.topper wrote:
> Why do we need `hasSideEffects = 0` now? We didn't have it before.
This is redundant here. I was assuming that the table gen can't correctly infer here. Removed.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D154628/new/

https://reviews.llvm.org/D154628