[PATCH] D122918: [RISCV][CodeGen] Support float-arith in Zfinx

[Hugh Perkins via Phabricator via llvm-commits]

hughperkins added a comment.

Ok, I've since learned that this PR doesnt handle load/store. Only arithmetic.

However note that immediate float arithmetic also seems to be missing, e.g. the following crashes

  ; ModuleID = 'test.cpp'
  source_filename = "test.cpp"
  target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
  target triple = "x86_64-apple-macosx11.0.0"
  
  ; Function Attrs: mustprogress noinline nounwind optnone ssp uwtable
  define noundef float @_Z4funcf(float noundef %0) #0 {
    %2 = fadd float %0, 3.000000e+00
    ret float %2
  }
  
  attributes #0 = { mustprogress noinline nounwind optnone ssp uwtable "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="penryn" "target-features"="+cx16,+cx8,+fxsr,+mmx,+sahf,+sse,+sse2,+sse3,+sse4.1,+ssse3,+x87" "tune-cpu"="generic" }
  
  !llvm.module.flags = !{!0, !1, !2, !3}
  !llvm.ident = !{!4}
  
  !0 = !{i32 1, !"wchar_size", i32 4}
  !1 = !{i32 7, !"PIC Level", i32 2}
  !2 = !{i32 7, !"uwtable", i32 1}
  !3 = !{i32 7, !"frame-pointer", i32 2}
  !4 = !{!"clang version 14.0.0 (https://github.com/tru/llvm-release-build fc075d7c96fe7c992dde351695a5d25fe084794a)"}

Compiled with:

  bin/llc --march riscv32 -mattr +zfinx test_fadd_imm.ll

Gives:

  LLVM ERROR: Cannot select: t15: f32,ch = load<(load (s32) from constant-pool)> t0, t19, undef:i32
    t19: i32 = ADDI t18, TargetConstantPool:i32<float 3.000000e+00> 0 [TF=3]
      t18: i32 = LUI TargetConstantPool:i32<float 3.000000e+00> 0 [TF=4]
        t16: i32 = TargetConstantPool<float 3.000000e+00> 0 [TF=4]
      t17: i32 = TargetConstantPool<float 3.000000e+00> 0 [TF=3]
    t14: i32 = undef
  In function: _Z4funcf

(but compiles ok with `+f`)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D122918/new/

https://reviews.llvm.org/D122918