02907f3039
There is logic to track the expected number of instructions produced. It thought in this case an instruction would be necessary to negate the result, but here it folded into a ConstantExpr fneg when the non-undef value operand was cancelled out by the second fsub. I'm not sure why we don't fold constant FP ops with undef currently, but I think that would also avoid this problem. llvm-svn: 301199
68 lines
1.6 KiB
LLVM
68 lines
1.6 KiB
LLVM
; RUN: opt < %s -instcombine -S | FileCheck %s
|
|
|
|
; PR4374
|
|
define float @test1(float %a, float %b) nounwind {
|
|
%t1 = fsub float %a, %b
|
|
%t2 = fsub float -0.000000e+00, %t1
|
|
|
|
; CHECK: %t1 = fsub float %a, %b
|
|
; CHECK-NEXT: %t2 = fsub float -0.000000e+00, %t1
|
|
|
|
ret float %t2
|
|
}
|
|
|
|
; <rdar://problem/7530098>
|
|
define double @test2(double %x, double %y) nounwind {
|
|
%t1 = fadd double %x, %y
|
|
%t2 = fsub double %x, %t1
|
|
|
|
; CHECK: %t1 = fadd double %x, %y
|
|
; CHECK-NEXT: %t2 = fsub double %x, %t1
|
|
|
|
ret double %t2
|
|
}
|
|
|
|
; CHECK-LABEL: @fsub_undef(
|
|
; CHECK: %sub = fsub float %val, undef
|
|
define float @fsub_undef(float %val) {
|
|
bb:
|
|
%sub = fsub float %val, undef
|
|
ret float %sub
|
|
}
|
|
|
|
; XXX - Why doesn't this fold to undef?
|
|
; CHECK-LABEL: @fsub_fast_undef(
|
|
; CHCK: %sub = fsub fast float %val, undef
|
|
define float @fsub_fast_undef(float %val) {
|
|
bb:
|
|
%sub = fsub fast float %val, undef
|
|
ret float %sub
|
|
}
|
|
|
|
; CHECK-LABEL: @fneg_undef(
|
|
; CHECK: ret float fsub (float -0.000000e+00, float undef)
|
|
define float @fneg_undef(float %val) {
|
|
bb:
|
|
%sub = fsub float -0.0, undef
|
|
ret float %sub
|
|
}
|
|
|
|
; CHECK-LABEL: @fneg_fast_undef(
|
|
; CHECK: ret float fsub (float -0.000000e+00, float undef)
|
|
define float @fneg_fast_undef(float %val) {
|
|
bb:
|
|
%sub = fsub fast float -0.0, undef
|
|
ret float %sub
|
|
}
|
|
|
|
; This folds to a constant expression, which produced 0 instructions
|
|
; contrary to the expected one for negation.
|
|
; CHECK-LABEL: @inconsistent_numbers_fsub_undef(
|
|
; CHECK: ret float fsub (float -0.000000e+00, float undef)
|
|
define float @inconsistent_numbers_fsub_undef(float %val) {
|
|
bb:
|
|
%sub0 = fsub fast float %val, undef
|
|
%sub1 = fsub fast float %sub0, %val
|
|
ret float %sub1
|
|
}
|