mirror of
https://github.com/RPCS3/llvm.git
synced 2024-12-13 23:18:58 +00:00
5e74c97ad7
The result type of setcc is dependent on whether or not AVX512 is
present.
We had an X86-specific DAG-combine which assumed that the result type
should be i8 when it could be i1.
This meant that we would generate illegal setccs which LowerSETCC did
not like.
Instead, use an appropriate type and zero extend to i8.
Also, there were some scenarios where the fold should have fired but
didn't because we were overly cautious about the types. This meant that
we generated:
shrl $31, %edi
andl $1, %edi
kmovw %edi, %k0
kxnorw %k0, %k0, %k1
kshiftrw $15, %k1, %k1
kxorw %k1, %k0, %k0
kmovw %k0, %eax
instead of:
testl %edi, %edi
setns %al
This fixes PR27638.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@268609 91177308-0d34-0410-b5e6-96231b3b80d8
81 lines
2.0 KiB
LLVM
81 lines
2.0 KiB
LLVM
; RUN: llc < %s -mtriple=x86_64-apple-darwin | FileCheck %s
|
|
; rdar://7329206
|
|
|
|
; Use sbb x, x to materialize carry bit in a GPR. The value is either
|
|
; all 1's or all 0's.
|
|
|
|
define zeroext i16 @t1(i16 zeroext %x) nounwind readnone ssp {
|
|
entry:
|
|
; CHECK-LABEL: t1:
|
|
; CHECK: seta %al
|
|
; CHECK: movzbl %al, %eax
|
|
; CHECK: shll $5, %eax
|
|
%0 = icmp ugt i16 %x, 26 ; <i1> [#uses=1]
|
|
%iftmp.1.0 = select i1 %0, i16 32, i16 0 ; <i16> [#uses=1]
|
|
ret i16 %iftmp.1.0
|
|
}
|
|
|
|
define zeroext i16 @t2(i16 zeroext %x) nounwind readnone ssp {
|
|
entry:
|
|
; CHECK-LABEL: t2:
|
|
; CHECK: sbbl %eax, %eax
|
|
; CHECK: andl $32, %eax
|
|
%0 = icmp ult i16 %x, 26 ; <i1> [#uses=1]
|
|
%iftmp.0.0 = select i1 %0, i16 32, i16 0 ; <i16> [#uses=1]
|
|
ret i16 %iftmp.0.0
|
|
}
|
|
|
|
define i64 @t3(i64 %x) nounwind readnone ssp {
|
|
entry:
|
|
; CHECK-LABEL: t3:
|
|
; CHECK: sbbq %rax, %rax
|
|
; CHECK: andl $64, %eax
|
|
%0 = icmp ult i64 %x, 18 ; <i1> [#uses=1]
|
|
%iftmp.2.0 = select i1 %0, i64 64, i64 0 ; <i64> [#uses=1]
|
|
ret i64 %iftmp.2.0
|
|
}
|
|
|
|
@v4 = common global i32 0, align 4
|
|
|
|
define i32 @t4(i32 %a) {
|
|
entry:
|
|
; CHECK-LABEL: t4:
|
|
; CHECK: movq _v4@GOTPCREL(%rip), %rax
|
|
; CHECK: cmpl $1, (%rax)
|
|
; CHECK: sbbl %eax, %eax
|
|
; CHECK: andl $32768, %eax
|
|
; CHECK: leal 65536(%rax,%rax), %eax
|
|
%0 = load i32, i32* @v4, align 4
|
|
%not.tobool = icmp eq i32 %0, 0
|
|
%conv.i = sext i1 %not.tobool to i16
|
|
%call.lobit = lshr i16 %conv.i, 15
|
|
%add.i.1 = add nuw nsw i16 %call.lobit, 1
|
|
%conv4.2 = zext i16 %add.i.1 to i32
|
|
%add = shl nuw nsw i32 %conv4.2, 16
|
|
ret i32 %add
|
|
}
|
|
|
|
define i8 @t5(i32 %a) #0 {
|
|
entry:
|
|
; CHECK-LABEL: t5:
|
|
; CHECK: testl %edi, %edi
|
|
; CHECK: setns %al
|
|
%.lobit = lshr i32 %a, 31
|
|
%trunc = trunc i32 %.lobit to i8
|
|
%.not = xor i8 %trunc, 1
|
|
ret i8 %.not
|
|
}
|
|
|
|
define zeroext i1 @t6(i32 %a) #0 {
|
|
entry:
|
|
; CHECK-LABEL: t6:
|
|
; CHECK: testl %edi, %edi
|
|
; CHECK: setns %al
|
|
%.lobit = lshr i32 %a, 31
|
|
%trunc = trunc i32 %.lobit to i1
|
|
%.not = xor i1 %trunc, 1
|
|
ret i1 %.not
|
|
}
|
|
|
|
attributes #0 = { "target-cpu"="skylake-avx512" }
|