mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-12-21 00:32:23 +00:00
956d6f0cf5
This patch fixes the following miscompile:
define void @sqrtsd(<2 x double> %a) nounwind uwtable ssp {
%0 = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %a) nounwind
%a0 = extractelement <2 x double> %0, i32 0
%conv = fptrunc double %a0 to float
%a1 = extractelement <2 x double> %0, i32 1
%conv3 = fptrunc double %a1 to float
tail call void @callee2(float %conv, float %conv3) nounwind
ret void
}
Current codegen:
sqrtsd %xmm0, %xmm1 ## high element of %xmm1 is undef here
xorps %xmm0, %xmm0
cvtsd2ss %xmm1, %xmm0
shufpd $1, %xmm1, %xmm1
cvtsd2ss %xmm1, %xmm1 ## operating on undef value
jmp _callee
This is a continuation of http://llvm.org/viewvc/llvm-project?view=revision&revision=224624 ( http://reviews.llvm.org/D6330 )
which was itself a continuation of r167064 ( http://llvm.org/viewvc/llvm-project?view=revision&revision=167064 ).
All of these patches are partial fixes for PR14221 ( http://llvm.org/bugs/show_bug.cgi?id=14221 );
this should be the final patch needed to resolve that bug.
Differential Revision: http://reviews.llvm.org/D6885
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@227111 91177308-0d34-0410-b5e6-96231b3b80d8
93 lines
3.1 KiB
LLVM
93 lines
3.1 KiB
LLVM
; RUN: llc < %s -mtriple=x86_64-apple-macosx -mattr=+sse2 -mcpu=nehalem | FileCheck %s
|
|
|
|
; rdar: 12558838
|
|
; PR14221
|
|
; There is a mismatch between the intrinsic and the actual instruction.
|
|
; The actual instruction has a partial update of dest, while the intrinsic
|
|
; passes through the upper FP values. Here, we make sure the source and
|
|
; destination of each scalar unary op are the same.
|
|
|
|
define void @rsqrtss(<4 x float> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: rsqrtss:
|
|
; CHECK: rsqrtss %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movshdup
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movap
|
|
; CHECK-NEXT: jmp
|
|
|
|
%0 = tail call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %a) nounwind
|
|
%a.addr.0.extract = extractelement <4 x float> %0, i32 0
|
|
%conv = fpext float %a.addr.0.extract to double
|
|
%a.addr.4.extract = extractelement <4 x float> %0, i32 1
|
|
%conv3 = fpext float %a.addr.4.extract to double
|
|
tail call void @callee(double %conv, double %conv3) nounwind
|
|
ret void
|
|
}
|
|
declare void @callee(double, double)
|
|
declare <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float>) nounwind readnone
|
|
|
|
define void @rcpss(<4 x float> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: rcpss:
|
|
; CHECK: rcpss %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movshdup
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movap
|
|
; CHECK-NEXT: jmp
|
|
|
|
%0 = tail call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %a) nounwind
|
|
%a.addr.0.extract = extractelement <4 x float> %0, i32 0
|
|
%conv = fpext float %a.addr.0.extract to double
|
|
%a.addr.4.extract = extractelement <4 x float> %0, i32 1
|
|
%conv3 = fpext float %a.addr.4.extract to double
|
|
tail call void @callee(double %conv, double %conv3) nounwind
|
|
ret void
|
|
}
|
|
declare <4 x float> @llvm.x86.sse.rcp.ss(<4 x float>) nounwind readnone
|
|
|
|
define void @sqrtss(<4 x float> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: sqrtss:
|
|
; CHECK: sqrtss %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movshdup
|
|
; CHECK-NEXT: cvtss2sd %xmm0
|
|
; CHECK-NEXT: movap
|
|
; CHECK-NEXT: jmp
|
|
|
|
%0 = tail call <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float> %a) nounwind
|
|
%a.addr.0.extract = extractelement <4 x float> %0, i32 0
|
|
%conv = fpext float %a.addr.0.extract to double
|
|
%a.addr.4.extract = extractelement <4 x float> %0, i32 1
|
|
%conv3 = fpext float %a.addr.4.extract to double
|
|
tail call void @callee(double %conv, double %conv3) nounwind
|
|
ret void
|
|
}
|
|
declare <4 x float> @llvm.x86.sse.sqrt.ss(<4 x float>) nounwind readnone
|
|
|
|
define void @sqrtsd(<2 x double> %a) nounwind uwtable ssp {
|
|
entry:
|
|
; CHECK-LABEL: sqrtsd:
|
|
; CHECK: sqrtsd %xmm0, %xmm0
|
|
; CHECK-NEXT: cvtsd2ss %xmm0
|
|
; CHECK-NEXT: shufpd
|
|
; CHECK-NEXT: cvtsd2ss %xmm0
|
|
; CHECK-NEXT: movap
|
|
; CHECK-NEXT: jmp
|
|
|
|
%0 = tail call <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double> %a) nounwind
|
|
%a0 = extractelement <2 x double> %0, i32 0
|
|
%conv = fptrunc double %a0 to float
|
|
%a1 = extractelement <2 x double> %0, i32 1
|
|
%conv3 = fptrunc double %a1 to float
|
|
tail call void @callee2(float %conv, float %conv3) nounwind
|
|
ret void
|
|
}
|
|
|
|
declare void @callee2(float, float)
|
|
declare <2 x double> @llvm.x86.sse2.sqrt.sd(<2 x double>) nounwind readnone
|
|
|