[llvm] r323591 - [X86][SSE] Regenerate fp2int/int2fp tests
Simon Pilgrim via llvm-commits
llvm-commits at lists.llvm.org
Sat Jan 27 08:39:12 PST 2018
Author: rksimon
Date: Sat Jan 27 08:39:12 2018
New Revision: 323591
URL: http://llvm.org/viewvc/llvm-project?rev=323591&view=rev
Log:
[X86][SSE] Regenerate fp2int/int2fp tests
Cleanup check prefixes and check full codegen
Modified:
llvm/trunk/test/CodeGen/X86/scalar-fp-to-i64.ll
llvm/trunk/test/CodeGen/X86/scalar-int-to-fp.ll
Modified: llvm/trunk/test/CodeGen/X86/scalar-fp-to-i64.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/scalar-fp-to-i64.ll?rev=323591&r1=323590&r2=323591&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/scalar-fp-to-i64.ll (original)
+++ llvm/trunk/test/CodeGen/X86/scalar-fp-to-i64.ll Sat Jan 27 08:39:12 2018
@@ -1,3 +1,19 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X32,AVX512_32,AVX512_32_WIN
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X32,AVX512_32,AVX512_32_LIN
+; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X64,AVX512_64,AVX512_64_WIN
+; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,X32,AVX512_64,AVX512_64_LIN
+; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+sse3 | FileCheck %s --check-prefixes=CHECK,X32,SSE3_32,SSE3_32_WIN
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+sse3 | FileCheck %s --check-prefixes=CHECK,X32,SSE3_32,SSE3_32_LIN
+; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+sse3 | FileCheck %s --check-prefixes=CHECK,X64,SSE3_64,SSE3_64_WIN
+; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+sse3 | FileCheck %s --check-prefixes=CHECK,X64,SSE3_64,SSE3_64_LIN
+; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,X32,SSE2_32,SSE2_32_WIN
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,X32,SSE2_32,SSE2_32_LIN
+; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,X64,SSE2_64,SSE2_64_WIN
+; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,X64,SSE2_64,SSE2_64_LIN
+; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=-sse | FileCheck %s --check-prefixes=CHECK,X32,X87,X87_WIN
+; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=-sse | FileCheck %s --check-prefixes=CHECK,X32,X87,X87_LIN
+
; Check that scalar FP conversions to signed and unsigned int64 are using
; reasonable sequences, across platforms and target switches.
;
@@ -13,139 +29,1643 @@
;
; The interesting subtargets are AVX512F (vcvtts[ds]2usi), SSE3 (fisttp),
; SSE2 (cvtts[ds]2si) and vanilla X87 (fnstcw+fist, 32-bit only).
-;
-; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_32
-; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_32
-; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_64
-; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=AVX512_64
-; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+sse3 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE3_32
-; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+sse3 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE3_32
-; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+sse3 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE3_64
-; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+sse3 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE3_64
-; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE2_32
-; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE2_32
-; RUN: llc < %s -mtriple=x86_64-pc-windows-msvc -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE2_64
-; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE2_64
-; RUN: llc < %s -mtriple=i386-pc-windows-msvc -mattr=-sse | FileCheck %s --check-prefix=CHECK --check-prefix=X87
-; RUN: llc < %s -mtriple=i386-unknown-linux-gnu -mattr=-sse | FileCheck %s --check-prefix=CHECK --check-prefix=X87
-
-; CHECK-LABEL: f_to_u64
-; X87-DAG: fsub
-; X87-DAG: fnstcw
-; X87: fist
-; SSE2_32-DAG: {{subss|fsub}}
-; SSE2_32-DAG: fnstcw
-; SSE2_32: fist
-; SSE2_64: subss
-; SSE2_64: cvttss2si
-; SSE3_32: {{subss|fsub}}
-; SSE3_32: fistt
-; SSE3_64: subss
-; SSE3_64: cvttss2si
-; AVX512_32: {{subss|fsub}}
-; AVX512_32: fistt
-; AVX512_64: vcvttss2usi
-; CHECK: ret
+
define i64 @f_to_u64(float %a) nounwind {
+; AVX512_32_WIN-LABEL: f_to_u64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512_32_WIN-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512_32_WIN-NEXT: vcmpltss %xmm1, %xmm0, %k1
+; AVX512_32_WIN-NEXT: vsubss %xmm1, %xmm0, %xmm2
+; AVX512_32_WIN-NEXT: vmovss %xmm0, %xmm0, %xmm2 {%k1}
+; AVX512_32_WIN-NEXT: vmovss %xmm2, {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: xorl %edx, %edx
+; AVX512_32_WIN-NEXT: vucomiss %xmm0, %xmm1
+; AVX512_32_WIN-NEXT: setbe %dl
+; AVX512_32_WIN-NEXT: shll $31, %edx
+; AVX512_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: f_to_u64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $20, %esp
+; AVX512_32_LIN-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512_32_LIN-NEXT: vmovss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512_32_LIN-NEXT: vcmpltss %xmm1, %xmm0, %k1
+; AVX512_32_LIN-NEXT: vsubss %xmm1, %xmm0, %xmm2
+; AVX512_32_LIN-NEXT: vmovss %xmm0, %xmm0, %xmm2 {%k1}
+; AVX512_32_LIN-NEXT: vmovss %xmm2, {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: xorl %edx, %edx
+; AVX512_32_LIN-NEXT: vucomiss %xmm0, %xmm1
+; AVX512_32_LIN-NEXT: setbe %dl
+; AVX512_32_LIN-NEXT: shll $31, %edx
+; AVX512_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: addl $20, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64-LABEL: f_to_u64:
+; AVX512_64: # %bb.0:
+; AVX512_64-NEXT: vcvttss2usi %xmm0, %rax
+; AVX512_64-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: f_to_u64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $16, %esp
+; SSE3_32_WIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3_32_WIN-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE3_32_WIN-NEXT: movaps %xmm0, %xmm2
+; SSE3_32_WIN-NEXT: cmpltss %xmm1, %xmm2
+; SSE3_32_WIN-NEXT: movaps %xmm2, %xmm3
+; SSE3_32_WIN-NEXT: andps %xmm0, %xmm2
+; SSE3_32_WIN-NEXT: xorl %edx, %edx
+; SSE3_32_WIN-NEXT: ucomiss %xmm0, %xmm1
+; SSE3_32_WIN-NEXT: subss %xmm1, %xmm0
+; SSE3_32_WIN-NEXT: andnps %xmm0, %xmm3
+; SSE3_32_WIN-NEXT: orps %xmm3, %xmm2
+; SSE3_32_WIN-NEXT: movss %xmm2, {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: setbe %dl
+; SSE3_32_WIN-NEXT: shll $31, %edx
+; SSE3_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: f_to_u64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $20, %esp
+; SSE3_32_LIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3_32_LIN-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE3_32_LIN-NEXT: movaps %xmm0, %xmm2
+; SSE3_32_LIN-NEXT: cmpltss %xmm1, %xmm2
+; SSE3_32_LIN-NEXT: movaps %xmm2, %xmm3
+; SSE3_32_LIN-NEXT: andps %xmm0, %xmm2
+; SSE3_32_LIN-NEXT: xorl %edx, %edx
+; SSE3_32_LIN-NEXT: ucomiss %xmm0, %xmm1
+; SSE3_32_LIN-NEXT: subss %xmm1, %xmm0
+; SSE3_32_LIN-NEXT: andnps %xmm0, %xmm3
+; SSE3_32_LIN-NEXT: orps %xmm3, %xmm2
+; SSE3_32_LIN-NEXT: movss %xmm2, {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: setbe %dl
+; SSE3_32_LIN-NEXT: shll $31, %edx
+; SSE3_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: addl $20, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64-LABEL: f_to_u64:
+; SSE3_64: # %bb.0:
+; SSE3_64-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE3_64-NEXT: movaps %xmm0, %xmm2
+; SSE3_64-NEXT: subss %xmm1, %xmm2
+; SSE3_64-NEXT: cvttss2si %xmm2, %rax
+; SSE3_64-NEXT: movabsq $-9223372036854775808, %rcx # imm = 0x8000000000000000
+; SSE3_64-NEXT: xorq %rax, %rcx
+; SSE3_64-NEXT: cvttss2si %xmm0, %rax
+; SSE3_64-NEXT: ucomiss %xmm1, %xmm0
+; SSE3_64-NEXT: cmovaeq %rcx, %rax
+; SSE3_64-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: f_to_u64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $24, %esp
+; SSE2_32_WIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2_32_WIN-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2_32_WIN-NEXT: movaps %xmm0, %xmm2
+; SSE2_32_WIN-NEXT: cmpltss %xmm1, %xmm2
+; SSE2_32_WIN-NEXT: movaps %xmm2, %xmm3
+; SSE2_32_WIN-NEXT: andps %xmm0, %xmm2
+; SSE2_32_WIN-NEXT: xorl %edx, %edx
+; SSE2_32_WIN-NEXT: ucomiss %xmm0, %xmm1
+; SSE2_32_WIN-NEXT: subss %xmm1, %xmm0
+; SSE2_32_WIN-NEXT: andnps %xmm0, %xmm3
+; SSE2_32_WIN-NEXT: orps %xmm3, %xmm2
+; SSE2_32_WIN-NEXT: movss %xmm2, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: setbe %dl
+; SSE2_32_WIN-NEXT: shll $31, %edx
+; SSE2_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: f_to_u64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $28, %esp
+; SSE2_32_LIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2_32_LIN-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2_32_LIN-NEXT: movaps %xmm0, %xmm2
+; SSE2_32_LIN-NEXT: cmpltss %xmm1, %xmm2
+; SSE2_32_LIN-NEXT: movaps %xmm2, %xmm3
+; SSE2_32_LIN-NEXT: andps %xmm0, %xmm2
+; SSE2_32_LIN-NEXT: xorl %edx, %edx
+; SSE2_32_LIN-NEXT: ucomiss %xmm0, %xmm1
+; SSE2_32_LIN-NEXT: subss %xmm1, %xmm0
+; SSE2_32_LIN-NEXT: andnps %xmm0, %xmm3
+; SSE2_32_LIN-NEXT: orps %xmm3, %xmm2
+; SSE2_32_LIN-NEXT: movss %xmm2, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: setbe %dl
+; SSE2_32_LIN-NEXT: shll $31, %edx
+; SSE2_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64-LABEL: f_to_u64:
+; SSE2_64: # %bb.0:
+; SSE2_64-NEXT: movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2_64-NEXT: movaps %xmm0, %xmm2
+; SSE2_64-NEXT: subss %xmm1, %xmm2
+; SSE2_64-NEXT: cvttss2si %xmm2, %rax
+; SSE2_64-NEXT: movabsq $-9223372036854775808, %rcx # imm = 0x8000000000000000
+; SSE2_64-NEXT: xorq %rax, %rcx
+; SSE2_64-NEXT: cvttss2si %xmm0, %rax
+; SSE2_64-NEXT: ucomiss %xmm1, %xmm0
+; SSE2_64-NEXT: cmovaeq %rcx, %rax
+; SSE2_64-NEXT: retq
+;
+; X87_WIN-LABEL: f_to_u64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: flds 8(%ebp)
+; X87_WIN-NEXT: flds __real at 5f000000
+; X87_WIN-NEXT: fld %st(1)
+; X87_WIN-NEXT: fsub %st(1)
+; X87_WIN-NEXT: fxch %st(1)
+; X87_WIN-NEXT: fucomp %st(2)
+; X87_WIN-NEXT: fnstsw %ax
+; X87_WIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_WIN-NEXT: sahf
+; X87_WIN-NEXT: ja LBB0_2
+; X87_WIN-NEXT: # %bb.1:
+; X87_WIN-NEXT: fstp %st(1)
+; X87_WIN-NEXT: fldz
+; X87_WIN-NEXT: LBB0_2:
+; X87_WIN-NEXT: fstp %st(0)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: setbe %al
+; X87_WIN-NEXT: movzbl %al, %edx
+; X87_WIN-NEXT: shll $31, %edx
+; X87_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: f_to_u64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: flds {{\.LCPI.*}}
+; X87_LIN-NEXT: fld %st(1)
+; X87_LIN-NEXT: fsub %st(1)
+; X87_LIN-NEXT: fxch %st(1)
+; X87_LIN-NEXT: fucomp %st(2)
+; X87_LIN-NEXT: fnstsw %ax
+; X87_LIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_LIN-NEXT: sahf
+; X87_LIN-NEXT: ja .LBB0_2
+; X87_LIN-NEXT: # %bb.1:
+; X87_LIN-NEXT: fstp %st(1)
+; X87_LIN-NEXT: fldz
+; X87_LIN-NEXT: .LBB0_2:
+; X87_LIN-NEXT: fstp %st(0)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: setbe %al
+; X87_LIN-NEXT: movzbl %al, %edx
+; X87_LIN-NEXT: shll $31, %edx
+; X87_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptoui float %a to i64
ret i64 %r
}
-; CHECK-LABEL: f_to_s64
-; X87: fnstcw
-; X87: fist
-; SSE2_32: fnstcw
-; SSE2_32: fist
-; SSE2_64: cvttss2si
-; SSE3_32: fistt
-; SSE3_64: cvttss2si
-; AVX512_32: fistt
-; AVX512_64: vcvttss2si
-; CHECK: ret
define i64 @f_to_s64(float %a) nounwind {
+; AVX512_32_WIN-LABEL: f_to_s64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512_32_WIN-NEXT: vmovss %xmm0, {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: f_to_s64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $20, %esp
+; AVX512_32_LIN-NEXT: vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512_32_LIN-NEXT: vmovss %xmm0, {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: addl $20, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64-LABEL: f_to_s64:
+; AVX512_64: # %bb.0:
+; AVX512_64-NEXT: vcvttss2si %xmm0, %rax
+; AVX512_64-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: f_to_s64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $16, %esp
+; SSE3_32_WIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3_32_WIN-NEXT: movss %xmm0, {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: f_to_s64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $20, %esp
+; SSE3_32_LIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE3_32_LIN-NEXT: movss %xmm0, {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: addl $20, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64-LABEL: f_to_s64:
+; SSE3_64: # %bb.0:
+; SSE3_64-NEXT: cvttss2si %xmm0, %rax
+; SSE3_64-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: f_to_s64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $24, %esp
+; SSE2_32_WIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2_32_WIN-NEXT: movss %xmm0, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: f_to_s64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $28, %esp
+; SSE2_32_LIN-NEXT: movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2_32_LIN-NEXT: movss %xmm0, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64-LABEL: f_to_s64:
+; SSE2_64: # %bb.0:
+; SSE2_64-NEXT: cvttss2si %xmm0, %rax
+; SSE2_64-NEXT: retq
+;
+; X87_WIN-LABEL: f_to_s64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: flds 8(%ebp)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: f_to_s64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: flds {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptosi float %a to i64
ret i64 %r
}
-; CHECK-LABEL: d_to_u64
-; X87-DAG: fsub
-; X87-DAG: fnstcw
-; X87: fist
-; SSE2_32-DAG: {{subsd|fsub}}
-; SSE2_32-DAG: fnstcw
-; SSE2_32: fist
-; SSE2_64: subsd
-; SSE2_64: cvttsd2si
-; SSE3_32: {{subsd|fsub}}
-; SSE3_32: fistt
-; SSE3_64: subsd
-; SSE3_64: cvttsd2si
-; AVX512_32: {{subsd|fsub}}
-; AVX512_32: fistt
-; AVX512_64: vcvttsd2usi
-; CHECK: ret
define i64 @d_to_u64(double %a) nounwind {
+; AVX512_32_WIN-LABEL: d_to_u64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
+; AVX512_32_WIN-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
+; AVX512_32_WIN-NEXT: vcmpltsd %xmm1, %xmm0, %k1
+; AVX512_32_WIN-NEXT: vsubsd %xmm1, %xmm0, %xmm2
+; AVX512_32_WIN-NEXT: vmovsd %xmm0, %xmm0, %xmm2 {%k1}
+; AVX512_32_WIN-NEXT: vmovsd %xmm2, {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: xorl %edx, %edx
+; AVX512_32_WIN-NEXT: vucomisd %xmm0, %xmm1
+; AVX512_32_WIN-NEXT: setbe %dl
+; AVX512_32_WIN-NEXT: shll $31, %edx
+; AVX512_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: d_to_u64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $20, %esp
+; AVX512_32_LIN-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
+; AVX512_32_LIN-NEXT: vmovsd {{.*#+}} xmm1 = mem[0],zero
+; AVX512_32_LIN-NEXT: vcmpltsd %xmm1, %xmm0, %k1
+; AVX512_32_LIN-NEXT: vsubsd %xmm1, %xmm0, %xmm2
+; AVX512_32_LIN-NEXT: vmovsd %xmm0, %xmm0, %xmm2 {%k1}
+; AVX512_32_LIN-NEXT: vmovsd %xmm2, {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: xorl %edx, %edx
+; AVX512_32_LIN-NEXT: vucomisd %xmm0, %xmm1
+; AVX512_32_LIN-NEXT: setbe %dl
+; AVX512_32_LIN-NEXT: shll $31, %edx
+; AVX512_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: addl $20, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64-LABEL: d_to_u64:
+; AVX512_64: # %bb.0:
+; AVX512_64-NEXT: vcvttsd2usi %xmm0, %rax
+; AVX512_64-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: d_to_u64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $16, %esp
+; SSE3_32_WIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3_32_WIN-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE3_32_WIN-NEXT: movapd %xmm0, %xmm2
+; SSE3_32_WIN-NEXT: cmpltsd %xmm1, %xmm2
+; SSE3_32_WIN-NEXT: movapd %xmm2, %xmm3
+; SSE3_32_WIN-NEXT: andpd %xmm0, %xmm2
+; SSE3_32_WIN-NEXT: xorl %edx, %edx
+; SSE3_32_WIN-NEXT: ucomisd %xmm0, %xmm1
+; SSE3_32_WIN-NEXT: subsd %xmm1, %xmm0
+; SSE3_32_WIN-NEXT: andnpd %xmm0, %xmm3
+; SSE3_32_WIN-NEXT: orpd %xmm3, %xmm2
+; SSE3_32_WIN-NEXT: movsd %xmm2, {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: setbe %dl
+; SSE3_32_WIN-NEXT: shll $31, %edx
+; SSE3_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: d_to_u64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $20, %esp
+; SSE3_32_LIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3_32_LIN-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE3_32_LIN-NEXT: movapd %xmm0, %xmm2
+; SSE3_32_LIN-NEXT: cmpltsd %xmm1, %xmm2
+; SSE3_32_LIN-NEXT: movapd %xmm2, %xmm3
+; SSE3_32_LIN-NEXT: andpd %xmm0, %xmm2
+; SSE3_32_LIN-NEXT: xorl %edx, %edx
+; SSE3_32_LIN-NEXT: ucomisd %xmm0, %xmm1
+; SSE3_32_LIN-NEXT: subsd %xmm1, %xmm0
+; SSE3_32_LIN-NEXT: andnpd %xmm0, %xmm3
+; SSE3_32_LIN-NEXT: orpd %xmm3, %xmm2
+; SSE3_32_LIN-NEXT: movsd %xmm2, {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: setbe %dl
+; SSE3_32_LIN-NEXT: shll $31, %edx
+; SSE3_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: addl $20, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64-LABEL: d_to_u64:
+; SSE3_64: # %bb.0:
+; SSE3_64-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE3_64-NEXT: movapd %xmm0, %xmm2
+; SSE3_64-NEXT: subsd %xmm1, %xmm2
+; SSE3_64-NEXT: cvttsd2si %xmm2, %rax
+; SSE3_64-NEXT: movabsq $-9223372036854775808, %rcx # imm = 0x8000000000000000
+; SSE3_64-NEXT: xorq %rax, %rcx
+; SSE3_64-NEXT: cvttsd2si %xmm0, %rax
+; SSE3_64-NEXT: ucomisd %xmm1, %xmm0
+; SSE3_64-NEXT: cmovaeq %rcx, %rax
+; SSE3_64-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: d_to_u64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $24, %esp
+; SSE2_32_WIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE2_32_WIN-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE2_32_WIN-NEXT: movapd %xmm0, %xmm2
+; SSE2_32_WIN-NEXT: cmpltsd %xmm1, %xmm2
+; SSE2_32_WIN-NEXT: movapd %xmm2, %xmm3
+; SSE2_32_WIN-NEXT: andpd %xmm0, %xmm2
+; SSE2_32_WIN-NEXT: xorl %edx, %edx
+; SSE2_32_WIN-NEXT: ucomisd %xmm0, %xmm1
+; SSE2_32_WIN-NEXT: subsd %xmm1, %xmm0
+; SSE2_32_WIN-NEXT: andnpd %xmm0, %xmm3
+; SSE2_32_WIN-NEXT: orpd %xmm3, %xmm2
+; SSE2_32_WIN-NEXT: movsd %xmm2, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: setbe %dl
+; SSE2_32_WIN-NEXT: shll $31, %edx
+; SSE2_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: d_to_u64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $28, %esp
+; SSE2_32_LIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE2_32_LIN-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE2_32_LIN-NEXT: movapd %xmm0, %xmm2
+; SSE2_32_LIN-NEXT: cmpltsd %xmm1, %xmm2
+; SSE2_32_LIN-NEXT: movapd %xmm2, %xmm3
+; SSE2_32_LIN-NEXT: andpd %xmm0, %xmm2
+; SSE2_32_LIN-NEXT: xorl %edx, %edx
+; SSE2_32_LIN-NEXT: ucomisd %xmm0, %xmm1
+; SSE2_32_LIN-NEXT: subsd %xmm1, %xmm0
+; SSE2_32_LIN-NEXT: andnpd %xmm0, %xmm3
+; SSE2_32_LIN-NEXT: orpd %xmm3, %xmm2
+; SSE2_32_LIN-NEXT: movsd %xmm2, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: setbe %dl
+; SSE2_32_LIN-NEXT: shll $31, %edx
+; SSE2_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64-LABEL: d_to_u64:
+; SSE2_64: # %bb.0:
+; SSE2_64-NEXT: movsd {{.*#+}} xmm1 = mem[0],zero
+; SSE2_64-NEXT: movapd %xmm0, %xmm2
+; SSE2_64-NEXT: subsd %xmm1, %xmm2
+; SSE2_64-NEXT: cvttsd2si %xmm2, %rax
+; SSE2_64-NEXT: movabsq $-9223372036854775808, %rcx # imm = 0x8000000000000000
+; SSE2_64-NEXT: xorq %rax, %rcx
+; SSE2_64-NEXT: cvttsd2si %xmm0, %rax
+; SSE2_64-NEXT: ucomisd %xmm1, %xmm0
+; SSE2_64-NEXT: cmovaeq %rcx, %rax
+; SSE2_64-NEXT: retq
+;
+; X87_WIN-LABEL: d_to_u64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: fldl 8(%ebp)
+; X87_WIN-NEXT: flds __real at 5f000000
+; X87_WIN-NEXT: fld %st(1)
+; X87_WIN-NEXT: fsub %st(1)
+; X87_WIN-NEXT: fxch %st(1)
+; X87_WIN-NEXT: fucomp %st(2)
+; X87_WIN-NEXT: fnstsw %ax
+; X87_WIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_WIN-NEXT: sahf
+; X87_WIN-NEXT: ja LBB2_2
+; X87_WIN-NEXT: # %bb.1:
+; X87_WIN-NEXT: fstp %st(1)
+; X87_WIN-NEXT: fldz
+; X87_WIN-NEXT: LBB2_2:
+; X87_WIN-NEXT: fstp %st(0)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: setbe %al
+; X87_WIN-NEXT: movzbl %al, %edx
+; X87_WIN-NEXT: shll $31, %edx
+; X87_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: d_to_u64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: flds {{\.LCPI.*}}
+; X87_LIN-NEXT: fld %st(1)
+; X87_LIN-NEXT: fsub %st(1)
+; X87_LIN-NEXT: fxch %st(1)
+; X87_LIN-NEXT: fucomp %st(2)
+; X87_LIN-NEXT: fnstsw %ax
+; X87_LIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_LIN-NEXT: sahf
+; X87_LIN-NEXT: ja .LBB2_2
+; X87_LIN-NEXT: # %bb.1:
+; X87_LIN-NEXT: fstp %st(1)
+; X87_LIN-NEXT: fldz
+; X87_LIN-NEXT: .LBB2_2:
+; X87_LIN-NEXT: fstp %st(0)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: setbe %al
+; X87_LIN-NEXT: movzbl %al, %edx
+; X87_LIN-NEXT: shll $31, %edx
+; X87_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptoui double %a to i64
ret i64 %r
}
-; CHECK-LABEL: d_to_s64
-; X87: fnstcw
-; X87: fist
-; SSE2_32: fnstcw
-; SSE2_32: fist
-; SSE2_64: cvttsd2si
-; SSE3_32: fistt
-; SSE3_64: cvttsd2si
-; AVX512_32: fistt
-; AVX512_64: vcvttsd2si
-; CHECK: ret
define i64 @d_to_s64(double %a) nounwind {
+; AVX512_32_WIN-LABEL: d_to_s64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
+; AVX512_32_WIN-NEXT: vmovsd %xmm0, {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: d_to_s64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $20, %esp
+; AVX512_32_LIN-NEXT: vmovsd {{.*#+}} xmm0 = mem[0],zero
+; AVX512_32_LIN-NEXT: vmovsd %xmm0, {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: addl $20, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64-LABEL: d_to_s64:
+; AVX512_64: # %bb.0:
+; AVX512_64-NEXT: vcvttsd2si %xmm0, %rax
+; AVX512_64-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: d_to_s64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $16, %esp
+; SSE3_32_WIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3_32_WIN-NEXT: movsd %xmm0, {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: d_to_s64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $20, %esp
+; SSE3_32_LIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE3_32_LIN-NEXT: movsd %xmm0, {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: addl $20, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64-LABEL: d_to_s64:
+; SSE3_64: # %bb.0:
+; SSE3_64-NEXT: cvttsd2si %xmm0, %rax
+; SSE3_64-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: d_to_s64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $24, %esp
+; SSE2_32_WIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE2_32_WIN-NEXT: movsd %xmm0, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: d_to_s64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $28, %esp
+; SSE2_32_LIN-NEXT: movsd {{.*#+}} xmm0 = mem[0],zero
+; SSE2_32_LIN-NEXT: movsd %xmm0, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64-LABEL: d_to_s64:
+; SSE2_64: # %bb.0:
+; SSE2_64-NEXT: cvttsd2si %xmm0, %rax
+; SSE2_64-NEXT: retq
+;
+; X87_WIN-LABEL: d_to_s64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: fldl 8(%ebp)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: d_to_s64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: fldl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptosi double %a to i64
ret i64 %r
}
-; CHECK-LABEL: x_to_u64
-; CHECK-DAG: fsub
-; X87-DAG: fnstcw
-; SSE2_32-DAG: fnstcw
-; SSE2_64-DAG: fnstcw
-; CHECK: fist
-; CHECK: ret
define i64 @x_to_u64(x86_fp80 %a) nounwind {
+; AVX512_32_WIN-LABEL: x_to_u64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $8, %esp
+; AVX512_32_WIN-NEXT: fldt 8(%ebp)
+; AVX512_32_WIN-NEXT: flds __real at 5f000000
+; AVX512_32_WIN-NEXT: fld %st(1)
+; AVX512_32_WIN-NEXT: fsub %st(1)
+; AVX512_32_WIN-NEXT: xorl %edx, %edx
+; AVX512_32_WIN-NEXT: fxch %st(1)
+; AVX512_32_WIN-NEXT: fucompi %st(2)
+; AVX512_32_WIN-NEXT: fcmovnbe %st(1), %st(0)
+; AVX512_32_WIN-NEXT: fstp %st(1)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: setbe %dl
+; AVX512_32_WIN-NEXT: shll $31, %edx
+; AVX512_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: x_to_u64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $12, %esp
+; AVX512_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: flds {{\.LCPI.*}}
+; AVX512_32_LIN-NEXT: fld %st(1)
+; AVX512_32_LIN-NEXT: fsub %st(1)
+; AVX512_32_LIN-NEXT: xorl %edx, %edx
+; AVX512_32_LIN-NEXT: fxch %st(1)
+; AVX512_32_LIN-NEXT: fucompi %st(2)
+; AVX512_32_LIN-NEXT: fcmovnbe %st(1), %st(0)
+; AVX512_32_LIN-NEXT: fstp %st(1)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: setbe %dl
+; AVX512_32_LIN-NEXT: shll $31, %edx
+; AVX512_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: addl $12, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64_WIN-LABEL: x_to_u64:
+; AVX512_64_WIN: # %bb.0:
+; AVX512_64_WIN-NEXT: pushq %rax
+; AVX512_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; AVX512_64_WIN-NEXT: flds __real@{{.*}}(%rip)
+; AVX512_64_WIN-NEXT: fld %st(1)
+; AVX512_64_WIN-NEXT: fsub %st(1)
+; AVX512_64_WIN-NEXT: xorl %ecx, %ecx
+; AVX512_64_WIN-NEXT: fxch %st(1)
+; AVX512_64_WIN-NEXT: fucompi %st(2)
+; AVX512_64_WIN-NEXT: fcmovnbe %st(1), %st(0)
+; AVX512_64_WIN-NEXT: fstp %st(1)
+; AVX512_64_WIN-NEXT: fisttpll (%rsp)
+; AVX512_64_WIN-NEXT: setbe %cl
+; AVX512_64_WIN-NEXT: shll $31, %ecx
+; AVX512_64_WIN-NEXT: xorl {{[0-9]+}}(%rsp), %ecx
+; AVX512_64_WIN-NEXT: shlq $32, %rcx
+; AVX512_64_WIN-NEXT: movl (%rsp), %eax
+; AVX512_64_WIN-NEXT: orq %rcx, %rax
+; AVX512_64_WIN-NEXT: popq %rcx
+; AVX512_64_WIN-NEXT: retq
+;
+; AVX512_64_LIN-LABEL: x_to_u64:
+; AVX512_64_LIN: # %bb.0:
+; AVX512_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; AVX512_64_LIN-NEXT: flds {{.*}}(%rip)
+; AVX512_64_LIN-NEXT: fld %st(1)
+; AVX512_64_LIN-NEXT: fsub %st(1)
+; AVX512_64_LIN-NEXT: xorl %ecx, %ecx
+; AVX512_64_LIN-NEXT: fxch %st(1)
+; AVX512_64_LIN-NEXT: fucompi %st(2)
+; AVX512_64_LIN-NEXT: fcmovnbe %st(1), %st(0)
+; AVX512_64_LIN-NEXT: fstp %st(1)
+; AVX512_64_LIN-NEXT: fisttpll -{{[0-9]+}}(%rsp)
+; AVX512_64_LIN-NEXT: setbe %cl
+; AVX512_64_LIN-NEXT: shll $31, %ecx
+; AVX512_64_LIN-NEXT: xorl -{{[0-9]+}}(%rsp), %ecx
+; AVX512_64_LIN-NEXT: shlq $32, %rcx
+; AVX512_64_LIN-NEXT: movl -{{[0-9]+}}(%rsp), %eax
+; AVX512_64_LIN-NEXT: orq %rcx, %rax
+; AVX512_64_LIN-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: x_to_u64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $8, %esp
+; SSE3_32_WIN-NEXT: fldt 8(%ebp)
+; SSE3_32_WIN-NEXT: flds __real at 5f000000
+; SSE3_32_WIN-NEXT: fld %st(1)
+; SSE3_32_WIN-NEXT: fsub %st(1)
+; SSE3_32_WIN-NEXT: xorl %edx, %edx
+; SSE3_32_WIN-NEXT: fxch %st(1)
+; SSE3_32_WIN-NEXT: fucompi %st(2)
+; SSE3_32_WIN-NEXT: fcmovnbe %st(1), %st(0)
+; SSE3_32_WIN-NEXT: fstp %st(1)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: setbe %dl
+; SSE3_32_WIN-NEXT: shll $31, %edx
+; SSE3_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: x_to_u64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $12, %esp
+; SSE3_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: flds {{\.LCPI.*}}
+; SSE3_32_LIN-NEXT: fld %st(1)
+; SSE3_32_LIN-NEXT: fsub %st(1)
+; SSE3_32_LIN-NEXT: xorl %edx, %edx
+; SSE3_32_LIN-NEXT: fxch %st(1)
+; SSE3_32_LIN-NEXT: fucompi %st(2)
+; SSE3_32_LIN-NEXT: fcmovnbe %st(1), %st(0)
+; SSE3_32_LIN-NEXT: fstp %st(1)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: setbe %dl
+; SSE3_32_LIN-NEXT: shll $31, %edx
+; SSE3_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: addl $12, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64_WIN-LABEL: x_to_u64:
+; SSE3_64_WIN: # %bb.0:
+; SSE3_64_WIN-NEXT: subq $24, %rsp
+; SSE3_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE3_64_WIN-NEXT: flds __real@{{.*}}(%rip)
+; SSE3_64_WIN-NEXT: fld %st(1)
+; SSE3_64_WIN-NEXT: fsub %st(1)
+; SSE3_64_WIN-NEXT: fisttpll {{[0-9]+}}(%rsp)
+; SSE3_64_WIN-NEXT: fld %st(1)
+; SSE3_64_WIN-NEXT: fisttpll {{[0-9]+}}(%rsp)
+; SSE3_64_WIN-NEXT: fucompi %st(1)
+; SSE3_64_WIN-NEXT: fstp %st(0)
+; SSE3_64_WIN-NEXT: jbe .LBB4_1
+; SSE3_64_WIN-NEXT: # %bb.2:
+; SSE3_64_WIN-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE3_64_WIN-NEXT: addq $24, %rsp
+; SSE3_64_WIN-NEXT: retq
+; SSE3_64_WIN-NEXT: .LBB4_1:
+; SSE3_64_WIN-NEXT: movabsq $-9223372036854775808, %rax # imm = 0x8000000000000000
+; SSE3_64_WIN-NEXT: xorq {{[0-9]+}}(%rsp), %rax
+; SSE3_64_WIN-NEXT: addq $24, %rsp
+; SSE3_64_WIN-NEXT: retq
+;
+; SSE3_64_LIN-LABEL: x_to_u64:
+; SSE3_64_LIN: # %bb.0:
+; SSE3_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE3_64_LIN-NEXT: flds {{.*}}(%rip)
+; SSE3_64_LIN-NEXT: fld %st(1)
+; SSE3_64_LIN-NEXT: fsub %st(1)
+; SSE3_64_LIN-NEXT: fisttpll -{{[0-9]+}}(%rsp)
+; SSE3_64_LIN-NEXT: fld %st(1)
+; SSE3_64_LIN-NEXT: fisttpll -{{[0-9]+}}(%rsp)
+; SSE3_64_LIN-NEXT: fucompi %st(1)
+; SSE3_64_LIN-NEXT: fstp %st(0)
+; SSE3_64_LIN-NEXT: jbe .LBB4_1
+; SSE3_64_LIN-NEXT: # %bb.2:
+; SSE3_64_LIN-NEXT: movq -{{[0-9]+}}(%rsp), %rax
+; SSE3_64_LIN-NEXT: retq
+; SSE3_64_LIN-NEXT: .LBB4_1:
+; SSE3_64_LIN-NEXT: movabsq $-9223372036854775808, %rax # imm = 0x8000000000000000
+; SSE3_64_LIN-NEXT: xorq -{{[0-9]+}}(%rsp), %rax
+; SSE3_64_LIN-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: x_to_u64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $16, %esp
+; SSE2_32_WIN-NEXT: fldt 8(%ebp)
+; SSE2_32_WIN-NEXT: flds __real at 5f000000
+; SSE2_32_WIN-NEXT: fld %st(1)
+; SSE2_32_WIN-NEXT: fsub %st(1)
+; SSE2_32_WIN-NEXT: xorl %edx, %edx
+; SSE2_32_WIN-NEXT: fxch %st(1)
+; SSE2_32_WIN-NEXT: fucompi %st(2)
+; SSE2_32_WIN-NEXT: fcmovnbe %st(1), %st(0)
+; SSE2_32_WIN-NEXT: fstp %st(1)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: setbe %dl
+; SSE2_32_WIN-NEXT: shll $31, %edx
+; SSE2_32_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: x_to_u64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $20, %esp
+; SSE2_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: flds {{\.LCPI.*}}
+; SSE2_32_LIN-NEXT: fld %st(1)
+; SSE2_32_LIN-NEXT: fsub %st(1)
+; SSE2_32_LIN-NEXT: xorl %edx, %edx
+; SSE2_32_LIN-NEXT: fxch %st(1)
+; SSE2_32_LIN-NEXT: fucompi %st(2)
+; SSE2_32_LIN-NEXT: fcmovnbe %st(1), %st(0)
+; SSE2_32_LIN-NEXT: fstp %st(1)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: setbe %dl
+; SSE2_32_LIN-NEXT: shll $31, %edx
+; SSE2_32_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: addl $20, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64_WIN-LABEL: x_to_u64:
+; SSE2_64_WIN: # %bb.0:
+; SSE2_64_WIN-NEXT: subq $24, %rsp
+; SSE2_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: flds __real@{{.*}}(%rip)
+; SSE2_64_WIN-NEXT: fld %st(1)
+; SSE2_64_WIN-NEXT: fsub %st(1)
+; SSE2_64_WIN-NEXT: fnstcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movzwl {{[0-9]+}}(%rsp), %eax
+; SSE2_64_WIN-NEXT: movw $3199, {{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movw %ax, {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fistpll {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fnstcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movzwl {{[0-9]+}}(%rsp), %eax
+; SSE2_64_WIN-NEXT: movw $3199, {{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movw %ax, {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fld %st(1)
+; SSE2_64_WIN-NEXT: fistpll {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fucompi %st(1)
+; SSE2_64_WIN-NEXT: fstp %st(0)
+; SSE2_64_WIN-NEXT: jbe .LBB4_1
+; SSE2_64_WIN-NEXT: # %bb.2:
+; SSE2_64_WIN-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE2_64_WIN-NEXT: addq $24, %rsp
+; SSE2_64_WIN-NEXT: retq
+; SSE2_64_WIN-NEXT: .LBB4_1:
+; SSE2_64_WIN-NEXT: movabsq $-9223372036854775808, %rax # imm = 0x8000000000000000
+; SSE2_64_WIN-NEXT: xorq {{[0-9]+}}(%rsp), %rax
+; SSE2_64_WIN-NEXT: addq $24, %rsp
+; SSE2_64_WIN-NEXT: retq
+;
+; SSE2_64_LIN-LABEL: x_to_u64:
+; SSE2_64_LIN: # %bb.0:
+; SSE2_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: flds {{.*}}(%rip)
+; SSE2_64_LIN-NEXT: fld %st(1)
+; SSE2_64_LIN-NEXT: fsub %st(1)
+; SSE2_64_LIN-NEXT: fnstcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; SSE2_64_LIN-NEXT: movw $3199, -{{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movw %ax, -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fistpll -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fnstcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; SSE2_64_LIN-NEXT: movw $3199, -{{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movw %ax, -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fld %st(1)
+; SSE2_64_LIN-NEXT: fistpll -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fucompi %st(1)
+; SSE2_64_LIN-NEXT: fstp %st(0)
+; SSE2_64_LIN-NEXT: jbe .LBB4_1
+; SSE2_64_LIN-NEXT: # %bb.2:
+; SSE2_64_LIN-NEXT: movq -{{[0-9]+}}(%rsp), %rax
+; SSE2_64_LIN-NEXT: retq
+; SSE2_64_LIN-NEXT: .LBB4_1:
+; SSE2_64_LIN-NEXT: movabsq $-9223372036854775808, %rax # imm = 0x8000000000000000
+; SSE2_64_LIN-NEXT: xorq -{{[0-9]+}}(%rsp), %rax
+; SSE2_64_LIN-NEXT: retq
+;
+; X87_WIN-LABEL: x_to_u64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: fldt 8(%ebp)
+; X87_WIN-NEXT: flds __real at 5f000000
+; X87_WIN-NEXT: fld %st(1)
+; X87_WIN-NEXT: fsub %st(1)
+; X87_WIN-NEXT: fxch %st(1)
+; X87_WIN-NEXT: fucomp %st(2)
+; X87_WIN-NEXT: fnstsw %ax
+; X87_WIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_WIN-NEXT: sahf
+; X87_WIN-NEXT: ja LBB4_2
+; X87_WIN-NEXT: # %bb.1:
+; X87_WIN-NEXT: fstp %st(1)
+; X87_WIN-NEXT: fldz
+; X87_WIN-NEXT: LBB4_2:
+; X87_WIN-NEXT: fstp %st(0)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: setbe %al
+; X87_WIN-NEXT: movzbl %al, %edx
+; X87_WIN-NEXT: shll $31, %edx
+; X87_WIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: x_to_u64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: flds {{\.LCPI.*}}
+; X87_LIN-NEXT: fld %st(1)
+; X87_LIN-NEXT: fsub %st(1)
+; X87_LIN-NEXT: fxch %st(1)
+; X87_LIN-NEXT: fucomp %st(2)
+; X87_LIN-NEXT: fnstsw %ax
+; X87_LIN-NEXT: # kill: def %ah killed %ah killed %ax
+; X87_LIN-NEXT: sahf
+; X87_LIN-NEXT: ja .LBB4_2
+; X87_LIN-NEXT: # %bb.1:
+; X87_LIN-NEXT: fstp %st(1)
+; X87_LIN-NEXT: fldz
+; X87_LIN-NEXT: .LBB4_2:
+; X87_LIN-NEXT: fstp %st(0)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: setbe %al
+; X87_LIN-NEXT: movzbl %al, %edx
+; X87_LIN-NEXT: shll $31, %edx
+; X87_LIN-NEXT: xorl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptoui x86_fp80 %a to i64
ret i64 %r
}
-; CHECK-LABEL: x_to_s64
-; X87: fnstcw
-; X87: fist
-; SSE2_32: fnstcw
-; SSE2_32: fist
-; SSE2_64: fnstcw
-; SSE2_64: fist
-; SSE3_32: fistt
-; SSE3_64: fistt
-; AVX512_32: fistt
-; AVX512_64: fistt
-; CHECK: ret
define i64 @x_to_s64(x86_fp80 %a) nounwind {
+; AVX512_32_WIN-LABEL: x_to_s64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: pushl %ebp
+; AVX512_32_WIN-NEXT: movl %esp, %ebp
+; AVX512_32_WIN-NEXT: andl $-8, %esp
+; AVX512_32_WIN-NEXT: subl $8, %esp
+; AVX512_32_WIN-NEXT: fldt 8(%ebp)
+; AVX512_32_WIN-NEXT: fisttpll (%esp)
+; AVX512_32_WIN-NEXT: movl (%esp), %eax
+; AVX512_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_WIN-NEXT: movl %ebp, %esp
+; AVX512_32_WIN-NEXT: popl %ebp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: x_to_s64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $12, %esp
+; AVX512_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; AVX512_32_LIN-NEXT: fisttpll (%esp)
+; AVX512_32_LIN-NEXT: movl (%esp), %eax
+; AVX512_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; AVX512_32_LIN-NEXT: addl $12, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64_WIN-LABEL: x_to_s64:
+; AVX512_64_WIN: # %bb.0:
+; AVX512_64_WIN-NEXT: pushq %rax
+; AVX512_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; AVX512_64_WIN-NEXT: fisttpll (%rsp)
+; AVX512_64_WIN-NEXT: movq (%rsp), %rax
+; AVX512_64_WIN-NEXT: popq %rcx
+; AVX512_64_WIN-NEXT: retq
+;
+; AVX512_64_LIN-LABEL: x_to_s64:
+; AVX512_64_LIN: # %bb.0:
+; AVX512_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; AVX512_64_LIN-NEXT: fisttpll -{{[0-9]+}}(%rsp)
+; AVX512_64_LIN-NEXT: movq -{{[0-9]+}}(%rsp), %rax
+; AVX512_64_LIN-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: x_to_s64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl %ebp
+; SSE3_32_WIN-NEXT: movl %esp, %ebp
+; SSE3_32_WIN-NEXT: andl $-8, %esp
+; SSE3_32_WIN-NEXT: subl $8, %esp
+; SSE3_32_WIN-NEXT: fldt 8(%ebp)
+; SSE3_32_WIN-NEXT: fisttpll (%esp)
+; SSE3_32_WIN-NEXT: movl (%esp), %eax
+; SSE3_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_WIN-NEXT: movl %ebp, %esp
+; SSE3_32_WIN-NEXT: popl %ebp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: x_to_s64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $12, %esp
+; SSE3_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: fisttpll (%esp)
+; SSE3_32_LIN-NEXT: movl (%esp), %eax
+; SSE3_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE3_32_LIN-NEXT: addl $12, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64_WIN-LABEL: x_to_s64:
+; SSE3_64_WIN: # %bb.0:
+; SSE3_64_WIN-NEXT: pushq %rax
+; SSE3_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE3_64_WIN-NEXT: fisttpll (%rsp)
+; SSE3_64_WIN-NEXT: movq (%rsp), %rax
+; SSE3_64_WIN-NEXT: popq %rcx
+; SSE3_64_WIN-NEXT: retq
+;
+; SSE3_64_LIN-LABEL: x_to_s64:
+; SSE3_64_LIN: # %bb.0:
+; SSE3_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE3_64_LIN-NEXT: fisttpll -{{[0-9]+}}(%rsp)
+; SSE3_64_LIN-NEXT: movq -{{[0-9]+}}(%rsp), %rax
+; SSE3_64_LIN-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: x_to_s64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl %ebp
+; SSE2_32_WIN-NEXT: movl %esp, %ebp
+; SSE2_32_WIN-NEXT: andl $-8, %esp
+; SSE2_32_WIN-NEXT: subl $16, %esp
+; SSE2_32_WIN-NEXT: fldt 8(%ebp)
+; SSE2_32_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_WIN-NEXT: movl %ebp, %esp
+; SSE2_32_WIN-NEXT: popl %ebp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: x_to_s64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $20, %esp
+; SSE2_32_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; SSE2_32_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; SSE2_32_LIN-NEXT: addl $20, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64_WIN-LABEL: x_to_s64:
+; SSE2_64_WIN: # %bb.0:
+; SSE2_64_WIN-NEXT: subq $24, %rsp
+; SSE2_64_WIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fnstcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movzwl {{[0-9]+}}(%rsp), %eax
+; SSE2_64_WIN-NEXT: movw $3199, {{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movw %ax, {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fistpll {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: fldcw {{[0-9]+}}(%rsp)
+; SSE2_64_WIN-NEXT: movq {{[0-9]+}}(%rsp), %rax
+; SSE2_64_WIN-NEXT: addq $24, %rsp
+; SSE2_64_WIN-NEXT: retq
+;
+; SSE2_64_LIN-LABEL: x_to_s64:
+; SSE2_64_LIN: # %bb.0:
+; SSE2_64_LIN-NEXT: fldt {{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fnstcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movzwl -{{[0-9]+}}(%rsp), %eax
+; SSE2_64_LIN-NEXT: movw $3199, -{{[0-9]+}}(%rsp) # imm = 0xC7F
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movw %ax, -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fistpll -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: fldcw -{{[0-9]+}}(%rsp)
+; SSE2_64_LIN-NEXT: movq -{{[0-9]+}}(%rsp), %rax
+; SSE2_64_LIN-NEXT: retq
+;
+; X87_WIN-LABEL: x_to_s64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl %ebp
+; X87_WIN-NEXT: movl %esp, %ebp
+; X87_WIN-NEXT: andl $-8, %esp
+; X87_WIN-NEXT: subl $16, %esp
+; X87_WIN-NEXT: fldt 8(%ebp)
+; X87_WIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_WIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_WIN-NEXT: movl %ebp, %esp
+; X87_WIN-NEXT: popl %ebp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: x_to_s64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $20, %esp
+; X87_LIN-NEXT: fldt {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fnstcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movzwl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movw $3199, {{[0-9]+}}(%esp) # imm = 0xC7F
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movw %ax, {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fistpll {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: fldcw {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %eax
+; X87_LIN-NEXT: movl {{[0-9]+}}(%esp), %edx
+; X87_LIN-NEXT: addl $20, %esp
+; X87_LIN-NEXT: retl
%r = fptosi x86_fp80 %a to i64
ret i64 %r
}
-; CHECK-LABEL: t_to_u64
-; CHECK: __fixunstfdi
-; CHECK: ret
define i64 @t_to_u64(fp128 %a) nounwind {
+; AVX512_32_WIN-LABEL: t_to_u64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovups {{[0-9]+}}(%esp), %xmm0
+; AVX512_32_WIN-NEXT: vmovups %xmm0, (%esp)
+; AVX512_32_WIN-NEXT: calll ___fixunstfdi
+; AVX512_32_WIN-NEXT: addl $16, %esp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: t_to_u64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $28, %esp
+; AVX512_32_LIN-NEXT: vmovaps {{[0-9]+}}(%esp), %xmm0
+; AVX512_32_LIN-NEXT: vmovups %xmm0, (%esp)
+; AVX512_32_LIN-NEXT: calll __fixunstfdi at PLT
+; AVX512_32_LIN-NEXT: addl $28, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64_WIN-LABEL: t_to_u64:
+; AVX512_64_WIN: # %bb.0:
+; AVX512_64_WIN-NEXT: subq $40, %rsp
+; AVX512_64_WIN-NEXT: callq __fixunstfdi
+; AVX512_64_WIN-NEXT: addq $40, %rsp
+; AVX512_64_WIN-NEXT: retq
+;
+; AVX512_64_LIN-LABEL: t_to_u64:
+; AVX512_64_LIN: # %bb.0:
+; AVX512_64_LIN-NEXT: pushq %rax
+; AVX512_64_LIN-NEXT: callq __fixunstfdi at PLT
+; AVX512_64_LIN-NEXT: popq %rcx
+; AVX512_64_LIN-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: t_to_u64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: calll ___fixunstfdi
+; SSE3_32_WIN-NEXT: addl $16, %esp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: t_to_u64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $12, %esp
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: calll __fixunstfdi at PLT
+; SSE3_32_LIN-NEXT: addl $28, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64_WIN-LABEL: t_to_u64:
+; SSE3_64_WIN: # %bb.0:
+; SSE3_64_WIN-NEXT: subq $40, %rsp
+; SSE3_64_WIN-NEXT: callq __fixunstfdi
+; SSE3_64_WIN-NEXT: addq $40, %rsp
+; SSE3_64_WIN-NEXT: retq
+;
+; SSE3_64_LIN-LABEL: t_to_u64:
+; SSE3_64_LIN: # %bb.0:
+; SSE3_64_LIN-NEXT: pushq %rax
+; SSE3_64_LIN-NEXT: callq __fixunstfdi at PLT
+; SSE3_64_LIN-NEXT: popq %rcx
+; SSE3_64_LIN-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: t_to_u64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: calll ___fixunstfdi
+; SSE2_32_WIN-NEXT: addl $16, %esp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: t_to_u64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $12, %esp
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: calll __fixunstfdi at PLT
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64_WIN-LABEL: t_to_u64:
+; SSE2_64_WIN: # %bb.0:
+; SSE2_64_WIN-NEXT: subq $40, %rsp
+; SSE2_64_WIN-NEXT: callq __fixunstfdi
+; SSE2_64_WIN-NEXT: addq $40, %rsp
+; SSE2_64_WIN-NEXT: retq
+;
+; SSE2_64_LIN-LABEL: t_to_u64:
+; SSE2_64_LIN: # %bb.0:
+; SSE2_64_LIN-NEXT: pushq %rax
+; SSE2_64_LIN-NEXT: callq __fixunstfdi at PLT
+; SSE2_64_LIN-NEXT: popq %rcx
+; SSE2_64_LIN-NEXT: retq
+;
+; X87_WIN-LABEL: t_to_u64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: calll ___fixunstfdi
+; X87_WIN-NEXT: addl $16, %esp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: t_to_u64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $12, %esp
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: calll __fixunstfdi at PLT
+; X87_LIN-NEXT: addl $28, %esp
+; X87_LIN-NEXT: retl
%r = fptoui fp128 %a to i64
ret i64 %r
}
-; CHECK-LABEL: t_to_s64
-; CHECK: __fixtfdi
-; CHECK: ret
define i64 @t_to_s64(fp128 %a) nounwind {
+; AVX512_32_WIN-LABEL: t_to_s64:
+; AVX512_32_WIN: # %bb.0:
+; AVX512_32_WIN-NEXT: subl $16, %esp
+; AVX512_32_WIN-NEXT: vmovups {{[0-9]+}}(%esp), %xmm0
+; AVX512_32_WIN-NEXT: vmovups %xmm0, (%esp)
+; AVX512_32_WIN-NEXT: calll ___fixtfdi
+; AVX512_32_WIN-NEXT: addl $16, %esp
+; AVX512_32_WIN-NEXT: retl
+;
+; AVX512_32_LIN-LABEL: t_to_s64:
+; AVX512_32_LIN: # %bb.0:
+; AVX512_32_LIN-NEXT: subl $28, %esp
+; AVX512_32_LIN-NEXT: vmovaps {{[0-9]+}}(%esp), %xmm0
+; AVX512_32_LIN-NEXT: vmovups %xmm0, (%esp)
+; AVX512_32_LIN-NEXT: calll __fixtfdi at PLT
+; AVX512_32_LIN-NEXT: addl $28, %esp
+; AVX512_32_LIN-NEXT: retl
+;
+; AVX512_64_WIN-LABEL: t_to_s64:
+; AVX512_64_WIN: # %bb.0:
+; AVX512_64_WIN-NEXT: subq $40, %rsp
+; AVX512_64_WIN-NEXT: callq __fixtfdi
+; AVX512_64_WIN-NEXT: addq $40, %rsp
+; AVX512_64_WIN-NEXT: retq
+;
+; AVX512_64_LIN-LABEL: t_to_s64:
+; AVX512_64_LIN: # %bb.0:
+; AVX512_64_LIN-NEXT: pushq %rax
+; AVX512_64_LIN-NEXT: callq __fixtfdi at PLT
+; AVX512_64_LIN-NEXT: popq %rcx
+; AVX512_64_LIN-NEXT: retq
+;
+; SSE3_32_WIN-LABEL: t_to_s64:
+; SSE3_32_WIN: # %bb.0:
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_WIN-NEXT: calll ___fixtfdi
+; SSE3_32_WIN-NEXT: addl $16, %esp
+; SSE3_32_WIN-NEXT: retl
+;
+; SSE3_32_LIN-LABEL: t_to_s64:
+; SSE3_32_LIN: # %bb.0:
+; SSE3_32_LIN-NEXT: subl $12, %esp
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE3_32_LIN-NEXT: calll __fixtfdi at PLT
+; SSE3_32_LIN-NEXT: addl $28, %esp
+; SSE3_32_LIN-NEXT: retl
+;
+; SSE3_64_WIN-LABEL: t_to_s64:
+; SSE3_64_WIN: # %bb.0:
+; SSE3_64_WIN-NEXT: subq $40, %rsp
+; SSE3_64_WIN-NEXT: callq __fixtfdi
+; SSE3_64_WIN-NEXT: addq $40, %rsp
+; SSE3_64_WIN-NEXT: retq
+;
+; SSE3_64_LIN-LABEL: t_to_s64:
+; SSE3_64_LIN: # %bb.0:
+; SSE3_64_LIN-NEXT: pushq %rax
+; SSE3_64_LIN-NEXT: callq __fixtfdi at PLT
+; SSE3_64_LIN-NEXT: popq %rcx
+; SSE3_64_LIN-NEXT: retq
+;
+; SSE2_32_WIN-LABEL: t_to_s64:
+; SSE2_32_WIN: # %bb.0:
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_WIN-NEXT: calll ___fixtfdi
+; SSE2_32_WIN-NEXT: addl $16, %esp
+; SSE2_32_WIN-NEXT: retl
+;
+; SSE2_32_LIN-LABEL: t_to_s64:
+; SSE2_32_LIN: # %bb.0:
+; SSE2_32_LIN-NEXT: subl $12, %esp
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; SSE2_32_LIN-NEXT: calll __fixtfdi at PLT
+; SSE2_32_LIN-NEXT: addl $28, %esp
+; SSE2_32_LIN-NEXT: retl
+;
+; SSE2_64_WIN-LABEL: t_to_s64:
+; SSE2_64_WIN: # %bb.0:
+; SSE2_64_WIN-NEXT: subq $40, %rsp
+; SSE2_64_WIN-NEXT: callq __fixtfdi
+; SSE2_64_WIN-NEXT: addq $40, %rsp
+; SSE2_64_WIN-NEXT: retq
+;
+; SSE2_64_LIN-LABEL: t_to_s64:
+; SSE2_64_LIN: # %bb.0:
+; SSE2_64_LIN-NEXT: pushq %rax
+; SSE2_64_LIN-NEXT: callq __fixtfdi at PLT
+; SSE2_64_LIN-NEXT: popq %rcx
+; SSE2_64_LIN-NEXT: retq
+;
+; X87_WIN-LABEL: t_to_s64:
+; X87_WIN: # %bb.0:
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_WIN-NEXT: calll ___fixtfdi
+; X87_WIN-NEXT: addl $16, %esp
+; X87_WIN-NEXT: retl
+;
+; X87_LIN-LABEL: t_to_s64:
+; X87_LIN: # %bb.0:
+; X87_LIN-NEXT: subl $12, %esp
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: pushl {{[0-9]+}}(%esp)
+; X87_LIN-NEXT: calll __fixtfdi at PLT
+; X87_LIN-NEXT: addl $28, %esp
+; X87_LIN-NEXT: retl
%r = fptosi fp128 %a to i64
ret i64 %r
}
Modified: llvm/trunk/test/CodeGen/X86/scalar-int-to-fp.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/scalar-int-to-fp.ll?rev=323591&r1=323590&r2=323591&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/scalar-int-to-fp.ll (original)
+++ llvm/trunk/test/CodeGen/X86/scalar-int-to-fp.ll Sat Jan 27 08:39:12 2018
@@ -1,9 +1,9 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK32 --check-prefix=AVX512_32
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK64 --check-prefix=AVX512_64
-; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK32 --check-prefix=SSE2_32
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK64 --check-prefix=SSE2_64
-; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=-sse | FileCheck %s --check-prefix=CHECK --check-prefix=CHECK32 --check-prefix=X87
+; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,CHECK32,AVX512_32
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=CHECK,CHECK64,AVX512_64
+; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,CHECK32,SSE2_32
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=CHECK,CHECK64,SSE2_64
+; RUN: llc < %s -mtriple=i386-unknown-unknown -mattr=-sse | FileCheck %s --check-prefixes=CHECK,CHECK32,X87
; Verify that scalar integer conversions to FP compile successfully
; (at one time long double failed with avx512f), and that reasonable
More information about the llvm-commits
mailing list