[PATCH] D137482: [LoongArch] Support parsing target specific flags for MIR

wanglei via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Sat Nov 5 17:47:15 PDT 2022


wangleiat added inline comments.


================
Comment at: llvm/test/CodeGen/LoongArch/mir-target-flags.ll:3
+; RUN:     --relocation-model=pic %s -o %t.mir
+; RUN: llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo \
+; RUN:     %t.mir -o - | FileCheck %s
----------------
wangleiat wrote:
> wangleiat wrote:
> > wangleiat wrote:
> > > MaskRay wrote:
> > > > `llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo ` does not change the output. Is it needed?
> > > > `llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo ` does not change the output. Is it needed?
> > > 
> > > Thanks for your review. Yes, it is necessary. In fact, this step is the real test for parsing target flags.
> > > `llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo ` does not change the output. Is it needed?
> > 
> > 
> > > `llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo ` does not change the output. Is it needed?
> > 
> > Thanks for your review. Yes, it is necessary. In fact, this step is the real test for parsing target flags.
> 
> I seem to have misunderstood your meaning, maybe I don't need that pass? It may not be needed, I will verify and modify it.
> > `llc --mtriple=loongarch64 --run-pass loongarch-prera-expand-pseudo ` does not change the output. Is it needed?
> 
> 

I verified it, and it seems that there is no good way to remove this pass to verify that the correct flags are generated in the mir, unless the previous step adds check. The benefit of this is that both the output and the input are validated.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D137482/new/

https://reviews.llvm.org/D137482



More information about the llvm-commits mailing list