508 lines
16 KiB
LLVM
508 lines
16 KiB
LLVM
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
||
|
|
; RUN: llc < %s -mtriple=x86_64-linux | FileCheck %s --check-prefix=X64
|
||
|
|
; RUN: llc < %s -mtriple=i686 -mattr=cmov | FileCheck %s --check-prefix=X86
|
||
|
|
|
||
|
|
declare i4 @llvm.udiv.fix.sat.i4 (i4, i4, i32)
|
||
|
|
declare i15 @llvm.udiv.fix.sat.i15 (i15, i15, i32)
|
||
|
|
declare i16 @llvm.udiv.fix.sat.i16 (i16, i16, i32)
|
||
|
|
declare i18 @llvm.udiv.fix.sat.i18 (i18, i18, i32)
|
||
|
|
declare i64 @llvm.udiv.fix.sat.i64 (i64, i64, i32)
|
||
|
|
declare <4 x i32> @llvm.udiv.fix.sat.v4i32(<4 x i32>, <4 x i32>, i32)
|
||
|
|
|
||
|
|
define i16 @func(i16 %x, i16 %y) nounwind {
|
||
|
|
; X64-LABEL: func:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: movzwl %si, %ecx
|
||
|
|
; X64-NEXT: movzwl %di, %eax
|
||
|
|
; X64-NEXT: shll $8, %eax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divl %ecx
|
||
|
|
; X64-NEXT: cmpl $131071, %eax # imm = 0x1FFFF
|
||
|
|
; X64-NEXT: movl $131071, %ecx # imm = 0x1FFFF
|
||
|
|
; X64-NEXT: cmovael %ecx, %eax
|
||
|
|
; X64-NEXT: shrl %eax
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: movzwl %ax, %eax
|
||
|
|
; X86-NEXT: shll $8, %eax
|
||
|
|
; X86-NEXT: xorl %edx, %edx
|
||
|
|
; X86-NEXT: divl %ecx
|
||
|
|
; X86-NEXT: cmpl $131071, %eax # imm = 0x1FFFF
|
||
|
|
; X86-NEXT: movl $131071, %ecx # imm = 0x1FFFF
|
||
|
|
; X86-NEXT: cmovael %ecx, %eax
|
||
|
|
; X86-NEXT: shrl %eax
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%tmp = call i16 @llvm.udiv.fix.sat.i16(i16 %x, i16 %y, i32 7)
|
||
|
|
ret i16 %tmp
|
||
|
|
}
|
||
|
|
|
||
|
|
define i16 @func2(i8 %x, i8 %y) nounwind {
|
||
|
|
; X64-LABEL: func2:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: movsbl %dil, %eax
|
||
|
|
; X64-NEXT: andl $32767, %eax # imm = 0x7FFF
|
||
|
|
; X64-NEXT: movsbl %sil, %ecx
|
||
|
|
; X64-NEXT: andl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X64-NEXT: shll $14, %eax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divl %ecx
|
||
|
|
; X64-NEXT: cmpl $32767, %eax # imm = 0x7FFF
|
||
|
|
; X64-NEXT: movl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X64-NEXT: cmovbl %eax, %ecx
|
||
|
|
; X64-NEXT: addl %ecx, %ecx
|
||
|
|
; X64-NEXT: movswl %cx, %eax
|
||
|
|
; X64-NEXT: shrl %eax
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func2:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movsbl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: andl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X86-NEXT: movsbl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: andl $32767, %eax # imm = 0x7FFF
|
||
|
|
; X86-NEXT: shll $14, %eax
|
||
|
|
; X86-NEXT: xorl %edx, %edx
|
||
|
|
; X86-NEXT: divl %ecx
|
||
|
|
; X86-NEXT: cmpl $32767, %eax # imm = 0x7FFF
|
||
|
|
; X86-NEXT: movl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X86-NEXT: cmovbl %eax, %ecx
|
||
|
|
; X86-NEXT: addl %ecx, %ecx
|
||
|
|
; X86-NEXT: movswl %cx, %eax
|
||
|
|
; X86-NEXT: shrl %eax
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%x2 = sext i8 %x to i15
|
||
|
|
%y2 = sext i8 %y to i15
|
||
|
|
%tmp = call i15 @llvm.udiv.fix.sat.i15(i15 %x2, i15 %y2, i32 14)
|
||
|
|
%tmp2 = sext i15 %tmp to i16
|
||
|
|
ret i16 %tmp2
|
||
|
|
}
|
||
|
|
|
||
|
|
define i16 @func3(i15 %x, i8 %y) nounwind {
|
||
|
|
; X64-LABEL: func3:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: # kill: def $edi killed $edi def $rdi
|
||
|
|
; X64-NEXT: leal (%rdi,%rdi), %eax
|
||
|
|
; X64-NEXT: movzbl %sil, %ecx
|
||
|
|
; X64-NEXT: shll $4, %ecx
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divw %cx
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax def $eax
|
||
|
|
; X64-NEXT: movzwl %ax, %ecx
|
||
|
|
; X64-NEXT: cmpl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X64-NEXT: movl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X64-NEXT: cmovbl %eax, %ecx
|
||
|
|
; X64-NEXT: addl %ecx, %ecx
|
||
|
|
; X64-NEXT: movswl %cx, %eax
|
||
|
|
; X64-NEXT: shrl %eax
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func3:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: addl %eax, %eax
|
||
|
|
; X86-NEXT: movzbl %cl, %ecx
|
||
|
|
; X86-NEXT: shll $4, %ecx
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X86-NEXT: xorl %edx, %edx
|
||
|
|
; X86-NEXT: divw %cx
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax def $eax
|
||
|
|
; X86-NEXT: movzwl %ax, %ecx
|
||
|
|
; X86-NEXT: cmpl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X86-NEXT: movl $32767, %ecx # imm = 0x7FFF
|
||
|
|
; X86-NEXT: cmovbl %eax, %ecx
|
||
|
|
; X86-NEXT: addl %ecx, %ecx
|
||
|
|
; X86-NEXT: movswl %cx, %eax
|
||
|
|
; X86-NEXT: shrl %eax
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%y2 = sext i8 %y to i15
|
||
|
|
%y3 = shl i15 %y2, 7
|
||
|
|
%tmp = call i15 @llvm.udiv.fix.sat.i15(i15 %x, i15 %y3, i32 4)
|
||
|
|
%tmp2 = sext i15 %tmp to i16
|
||
|
|
ret i16 %tmp2
|
||
|
|
}
|
||
|
|
|
||
|
|
define i4 @func4(i4 %x, i4 %y) nounwind {
|
||
|
|
; X64-LABEL: func4:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: andb $15, %sil
|
||
|
|
; X64-NEXT: andb $15, %dil
|
||
|
|
; X64-NEXT: shlb $2, %dil
|
||
|
|
; X64-NEXT: movzbl %dil, %eax
|
||
|
|
; X64-NEXT: divb %sil
|
||
|
|
; X64-NEXT: movzbl %al, %ecx
|
||
|
|
; X64-NEXT: cmpb $15, %cl
|
||
|
|
; X64-NEXT: movl $15, %eax
|
||
|
|
; X64-NEXT: cmovbl %ecx, %eax
|
||
|
|
; X64-NEXT: # kill: def $al killed $al killed $eax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func4:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movb {{[0-9]+}}(%esp), %cl
|
||
|
|
; X86-NEXT: andb $15, %cl
|
||
|
|
; X86-NEXT: movb {{[0-9]+}}(%esp), %al
|
||
|
|
; X86-NEXT: andb $15, %al
|
||
|
|
; X86-NEXT: shlb $2, %al
|
||
|
|
; X86-NEXT: movzbl %al, %eax
|
||
|
|
; X86-NEXT: divb %cl
|
||
|
|
; X86-NEXT: movzbl %al, %ecx
|
||
|
|
; X86-NEXT: cmpb $15, %al
|
||
|
|
; X86-NEXT: movl $15, %eax
|
||
|
|
; X86-NEXT: cmovbl %ecx, %eax
|
||
|
|
; X86-NEXT: # kill: def $al killed $al killed $eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%tmp = call i4 @llvm.udiv.fix.sat.i4(i4 %x, i4 %y, i32 2)
|
||
|
|
ret i4 %tmp
|
||
|
|
}
|
||
|
|
|
||
|
|
define i64 @func5(i64 %x, i64 %y) nounwind {
|
||
|
|
; X64-LABEL: func5:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: pushq %rbx
|
||
|
|
; X64-NEXT: movq %rsi, %rdx
|
||
|
|
; X64-NEXT: leaq (%rdi,%rdi), %rsi
|
||
|
|
; X64-NEXT: movq %rdi, %rax
|
||
|
|
; X64-NEXT: shrq $63, %rax
|
||
|
|
; X64-NEXT: shrdq $33, %rax, %rsi
|
||
|
|
; X64-NEXT: shlq $32, %rdi
|
||
|
|
; X64-NEXT: xorl %ebx, %ebx
|
||
|
|
; X64-NEXT: xorl %ecx, %ecx
|
||
|
|
; X64-NEXT: callq __udivti3@PLT
|
||
|
|
; X64-NEXT: cmpq $1, %rdx
|
||
|
|
; X64-NEXT: sbbq %rbx, %rbx
|
||
|
|
; X64-NEXT: notq %rbx
|
||
|
|
; X64-NEXT: orq %rax, %rbx
|
||
|
|
; X64-NEXT: cmpq $1, %rdx
|
||
|
|
; X64-NEXT: movl $1, %ecx
|
||
|
|
; X64-NEXT: cmovbq %rdx, %rcx
|
||
|
|
; X64-NEXT: cmoveq %rax, %rbx
|
||
|
|
; X64-NEXT: shrdq $1, %rcx, %rbx
|
||
|
|
; X64-NEXT: movq %rbx, %rax
|
||
|
|
; X64-NEXT: popq %rbx
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func5:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: pushl %ebp
|
||
|
|
; X86-NEXT: movl %esp, %ebp
|
||
|
|
; X86-NEXT: pushl %esi
|
||
|
|
; X86-NEXT: andl $-8, %esp
|
||
|
|
; X86-NEXT: subl $24, %esp
|
||
|
|
; X86-NEXT: movl 8(%ebp), %eax
|
||
|
|
; X86-NEXT: movl 12(%ebp), %ecx
|
||
|
|
; X86-NEXT: movl %ecx, %edx
|
||
|
|
; X86-NEXT: shrl %edx
|
||
|
|
; X86-NEXT: shldl $31, %eax, %ecx
|
||
|
|
; X86-NEXT: shll $31, %eax
|
||
|
|
; X86-NEXT: movl %esp, %esi
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl 20(%ebp)
|
||
|
|
; X86-NEXT: pushl 16(%ebp)
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl %edx
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: pushl %eax
|
||
|
|
; X86-NEXT: pushl %esi
|
||
|
|
; X86-NEXT: calll __udivti3
|
||
|
|
; X86-NEXT: addl $32, %esp
|
||
|
|
; X86-NEXT: movl (%esp), %ecx
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edx
|
||
|
|
; X86-NEXT: cmpl $-1, %edx
|
||
|
|
; X86-NEXT: movl %ecx, %eax
|
||
|
|
; X86-NEXT: cmovel %edx, %eax
|
||
|
|
; X86-NEXT: cmovel %ecx, %eax
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: orl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: movl $-1, %ecx
|
||
|
|
; X86-NEXT: cmovnel %ecx, %edx
|
||
|
|
; X86-NEXT: cmovnel %ecx, %eax
|
||
|
|
; X86-NEXT: leal -4(%ebp), %esp
|
||
|
|
; X86-NEXT: popl %esi
|
||
|
|
; X86-NEXT: popl %ebp
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%tmp = call i64 @llvm.udiv.fix.sat.i64(i64 %x, i64 %y, i32 31)
|
||
|
|
ret i64 %tmp
|
||
|
|
}
|
||
|
|
|
||
|
|
define i18 @func6(i16 %x, i16 %y) nounwind {
|
||
|
|
; X64-LABEL: func6:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: movswl %di, %eax
|
||
|
|
; X64-NEXT: andl $262143, %eax # imm = 0x3FFFF
|
||
|
|
; X64-NEXT: movswl %si, %ecx
|
||
|
|
; X64-NEXT: andl $262143, %ecx # imm = 0x3FFFF
|
||
|
|
; X64-NEXT: shll $7, %eax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divl %ecx
|
||
|
|
; X64-NEXT: cmpl $262143, %eax # imm = 0x3FFFF
|
||
|
|
; X64-NEXT: movl $262143, %ecx # imm = 0x3FFFF
|
||
|
|
; X64-NEXT: cmovael %ecx, %eax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func6:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movswl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: andl $262143, %ecx # imm = 0x3FFFF
|
||
|
|
; X86-NEXT: movswl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: andl $262143, %eax # imm = 0x3FFFF
|
||
|
|
; X86-NEXT: shll $7, %eax
|
||
|
|
; X86-NEXT: xorl %edx, %edx
|
||
|
|
; X86-NEXT: divl %ecx
|
||
|
|
; X86-NEXT: cmpl $262143, %eax # imm = 0x3FFFF
|
||
|
|
; X86-NEXT: movl $262143, %ecx # imm = 0x3FFFF
|
||
|
|
; X86-NEXT: cmovael %ecx, %eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%x2 = sext i16 %x to i18
|
||
|
|
%y2 = sext i16 %y to i18
|
||
|
|
%tmp = call i18 @llvm.udiv.fix.sat.i18(i18 %x2, i18 %y2, i32 7)
|
||
|
|
ret i18 %tmp
|
||
|
|
}
|
||
|
|
|
||
|
|
define i16 @func7(i16 %x, i16 %y) nounwind {
|
||
|
|
; X64-LABEL: func7:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: movzwl %si, %ecx
|
||
|
|
; X64-NEXT: movzwl %di, %eax
|
||
|
|
; X64-NEXT: addl %eax, %eax
|
||
|
|
; X64-NEXT: shlq $16, %rax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divq %rcx
|
||
|
|
; X64-NEXT: cmpq $131071, %rax # imm = 0x1FFFF
|
||
|
|
; X64-NEXT: movl $131071, %ecx # imm = 0x1FFFF
|
||
|
|
; X64-NEXT: cmovaeq %rcx, %rax
|
||
|
|
; X64-NEXT: shrl %eax
|
||
|
|
; X64-NEXT: # kill: def $ax killed $ax killed $rax
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: func7:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: movzwl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: movzwl %cx, %ecx
|
||
|
|
; X86-NEXT: addl %ecx, %ecx
|
||
|
|
; X86-NEXT: movl %ecx, %edx
|
||
|
|
; X86-NEXT: shrl $16, %edx
|
||
|
|
; X86-NEXT: shll $16, %ecx
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl %eax
|
||
|
|
; X86-NEXT: pushl %edx
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: calll __udivdi3
|
||
|
|
; X86-NEXT: addl $16, %esp
|
||
|
|
; X86-NEXT: cmpl $131071, %eax # imm = 0x1FFFF
|
||
|
|
; X86-NEXT: movl $131071, %ecx # imm = 0x1FFFF
|
||
|
|
; X86-NEXT: cmovael %ecx, %eax
|
||
|
|
; X86-NEXT: testl %edx, %edx
|
||
|
|
; X86-NEXT: cmovnel %ecx, %eax
|
||
|
|
; X86-NEXT: shrl %eax
|
||
|
|
; X86-NEXT: # kill: def $ax killed $ax killed $eax
|
||
|
|
; X86-NEXT: retl
|
||
|
|
%tmp = call i16 @llvm.udiv.fix.sat.i16(i16 %x, i16 %y, i32 16)
|
||
|
|
ret i16 %tmp
|
||
|
|
}
|
||
|
|
|
||
|
|
define <4 x i32> @vec(<4 x i32> %x, <4 x i32> %y) nounwind {
|
||
|
|
; X64-LABEL: vec:
|
||
|
|
; X64: # %bb.0:
|
||
|
|
; X64-NEXT: pxor %xmm3, %xmm3
|
||
|
|
; X64-NEXT: movdqa %xmm1, %xmm2
|
||
|
|
; X64-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
|
||
|
|
; X64-NEXT: movq %xmm2, %rcx
|
||
|
|
; X64-NEXT: movdqa %xmm0, %xmm2
|
||
|
|
; X64-NEXT: punpckhdq {{.*#+}} xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
|
||
|
|
; X64-NEXT: paddq %xmm2, %xmm2
|
||
|
|
; X64-NEXT: psllq $31, %xmm2
|
||
|
|
; X64-NEXT: movq %xmm2, %rax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divq %rcx
|
||
|
|
; X64-NEXT: movq %rax, %xmm7
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm2 = xmm2[2,3,2,3]
|
||
|
|
; X64-NEXT: movq %xmm2, %rax
|
||
|
|
; X64-NEXT: movdqa %xmm1, %xmm2
|
||
|
|
; X64-NEXT: psrldq {{.*#+}} xmm2 = xmm2[12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
|
||
|
|
; X64-NEXT: movq %xmm2, %rcx
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divq %rcx
|
||
|
|
; X64-NEXT: movq %rax, %xmm2
|
||
|
|
; X64-NEXT: punpcklqdq {{.*#+}} xmm7 = xmm7[0],xmm2[0]
|
||
|
|
; X64-NEXT: movdqa {{.*#+}} xmm4 = [9223372039002259456,9223372039002259456]
|
||
|
|
; X64-NEXT: movdqa %xmm7, %xmm2
|
||
|
|
; X64-NEXT: pxor %xmm4, %xmm2
|
||
|
|
; X64-NEXT: movdqa {{.*#+}} xmm8 = [9223372043297226751,9223372043297226751]
|
||
|
|
; X64-NEXT: movdqa %xmm8, %xmm6
|
||
|
|
; X64-NEXT: pcmpgtd %xmm2, %xmm6
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm9 = xmm6[0,0,2,2]
|
||
|
|
; X64-NEXT: pcmpeqd %xmm8, %xmm2
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm5 = xmm2[1,1,3,3]
|
||
|
|
; X64-NEXT: pand %xmm9, %xmm5
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm2 = xmm6[1,1,3,3]
|
||
|
|
; X64-NEXT: por %xmm5, %xmm2
|
||
|
|
; X64-NEXT: movdqa {{.*#+}} xmm6 = [8589934591,8589934591]
|
||
|
|
; X64-NEXT: pand %xmm2, %xmm7
|
||
|
|
; X64-NEXT: pandn %xmm6, %xmm2
|
||
|
|
; X64-NEXT: por %xmm7, %xmm2
|
||
|
|
; X64-NEXT: psrlq $1, %xmm2
|
||
|
|
; X64-NEXT: punpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
|
||
|
|
; X64-NEXT: movss {{.*#+}} xmm3 = xmm1[0],xmm3[1,2,3]
|
||
|
|
; X64-NEXT: movq %xmm3, %rcx
|
||
|
|
; X64-NEXT: paddq %xmm0, %xmm0
|
||
|
|
; X64-NEXT: psllq $31, %xmm0
|
||
|
|
; X64-NEXT: movq %xmm0, %rax
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divq %rcx
|
||
|
|
; X64-NEXT: movq %rax, %xmm3
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
|
||
|
|
; X64-NEXT: movq %xmm0, %rax
|
||
|
|
; X64-NEXT: psrlq $32, %xmm1
|
||
|
|
; X64-NEXT: movq %xmm1, %rcx
|
||
|
|
; X64-NEXT: xorl %edx, %edx
|
||
|
|
; X64-NEXT: divq %rcx
|
||
|
|
; X64-NEXT: movq %rax, %xmm0
|
||
|
|
; X64-NEXT: punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm0[0]
|
||
|
|
; X64-NEXT: pxor %xmm3, %xmm4
|
||
|
|
; X64-NEXT: movdqa %xmm8, %xmm0
|
||
|
|
; X64-NEXT: pcmpgtd %xmm4, %xmm0
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm1 = xmm0[0,0,2,2]
|
||
|
|
; X64-NEXT: pcmpeqd %xmm8, %xmm4
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
|
||
|
|
; X64-NEXT: pand %xmm1, %xmm4
|
||
|
|
; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
|
||
|
|
; X64-NEXT: por %xmm4, %xmm0
|
||
|
|
; X64-NEXT: pand %xmm0, %xmm3
|
||
|
|
; X64-NEXT: pandn %xmm6, %xmm0
|
||
|
|
; X64-NEXT: por %xmm3, %xmm0
|
||
|
|
; X64-NEXT: psrlq $1, %xmm0
|
||
|
|
; X64-NEXT: shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[0,2]
|
||
|
|
; X64-NEXT: retq
|
||
|
|
;
|
||
|
|
; X86-LABEL: vec:
|
||
|
|
; X86: # %bb.0:
|
||
|
|
; X86-NEXT: pushl %ebp
|
||
|
|
; X86-NEXT: pushl %ebx
|
||
|
|
; X86-NEXT: pushl %edi
|
||
|
|
; X86-NEXT: pushl %esi
|
||
|
|
; X86-NEXT: subl $12, %esp
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %edi
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ecx
|
||
|
|
; X86-NEXT: xorl %eax, %eax
|
||
|
|
; X86-NEXT: addl %ecx, %ecx
|
||
|
|
; X86-NEXT: setb %al
|
||
|
|
; X86-NEXT: shldl $31, %ecx, %eax
|
||
|
|
; X86-NEXT: shll $31, %ecx
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl {{[0-9]+}}(%esp)
|
||
|
|
; X86-NEXT: pushl %eax
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: calll __udivdi3
|
||
|
|
; X86-NEXT: addl $16, %esp
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $0, %ecx
|
||
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
||
|
|
; X86-NEXT: notl %ecx
|
||
|
|
; X86-NEXT: orl %eax, %ecx
|
||
|
|
; X86-NEXT: movl %ecx, %esi
|
||
|
|
; X86-NEXT: xorl %ecx, %ecx
|
||
|
|
; X86-NEXT: addl %edi, %edi
|
||
|
|
; X86-NEXT: setb %cl
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %ebx
|
||
|
|
; X86-NEXT: movl $1, %ebp
|
||
|
|
; X86-NEXT: cmovael %ebp, %edx
|
||
|
|
; X86-NEXT: movl %edx, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
|
||
|
|
; X86-NEXT: movl $1, %ebp
|
||
|
|
; X86-NEXT: cmovel %eax, %esi
|
||
|
|
; X86-NEXT: movl %esi, {{[-0-9]+}}(%e{{[sb]}}p) # 4-byte Spill
|
||
|
|
; X86-NEXT: shldl $31, %edi, %ecx
|
||
|
|
; X86-NEXT: shll $31, %edi
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl {{[0-9]+}}(%esp)
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: pushl %edi
|
||
|
|
; X86-NEXT: calll __udivdi3
|
||
|
|
; X86-NEXT: addl $16, %esp
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $0, %edi
|
||
|
|
; X86-NEXT: sbbl %edi, %edi
|
||
|
|
; X86-NEXT: notl %edi
|
||
|
|
; X86-NEXT: orl %eax, %edi
|
||
|
|
; X86-NEXT: xorl %ecx, %ecx
|
||
|
|
; X86-NEXT: addl %ebx, %ebx
|
||
|
|
; X86-NEXT: setb %cl
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %esi
|
||
|
|
; X86-NEXT: cmovael %ebp, %edx
|
||
|
|
; X86-NEXT: movl %edx, (%esp) # 4-byte Spill
|
||
|
|
; X86-NEXT: cmovel %eax, %edi
|
||
|
|
; X86-NEXT: shldl $31, %ebx, %ecx
|
||
|
|
; X86-NEXT: shll $31, %ebx
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl {{[0-9]+}}(%esp)
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: pushl %ebx
|
||
|
|
; X86-NEXT: calll __udivdi3
|
||
|
|
; X86-NEXT: addl $16, %esp
|
||
|
|
; X86-NEXT: movl %edx, %ebx
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $0, %ebp
|
||
|
|
; X86-NEXT: sbbl %ebp, %ebp
|
||
|
|
; X86-NEXT: notl %ebp
|
||
|
|
; X86-NEXT: orl %eax, %ebp
|
||
|
|
; X86-NEXT: xorl %ecx, %ecx
|
||
|
|
; X86-NEXT: addl %esi, %esi
|
||
|
|
; X86-NEXT: setb %cl
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $1, %edx
|
||
|
|
; X86-NEXT: cmovael %edx, %ebx
|
||
|
|
; X86-NEXT: cmovel %eax, %ebp
|
||
|
|
; X86-NEXT: shldl $31, %esi, %ecx
|
||
|
|
; X86-NEXT: shll $31, %esi
|
||
|
|
; X86-NEXT: pushl $0
|
||
|
|
; X86-NEXT: pushl {{[0-9]+}}(%esp)
|
||
|
|
; X86-NEXT: pushl %ecx
|
||
|
|
; X86-NEXT: pushl %esi
|
||
|
|
; X86-NEXT: calll __udivdi3
|
||
|
|
; X86-NEXT: addl $16, %esp
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $0, %ecx
|
||
|
|
; X86-NEXT: sbbl %ecx, %ecx
|
||
|
|
; X86-NEXT: notl %ecx
|
||
|
|
; X86-NEXT: orl %eax, %ecx
|
||
|
|
; X86-NEXT: cmpl $1, %edx
|
||
|
|
; X86-NEXT: movl $1, %esi
|
||
|
|
; X86-NEXT: cmovbl %edx, %esi
|
||
|
|
; X86-NEXT: cmovel %eax, %ecx
|
||
|
|
; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %edx # 4-byte Reload
|
||
|
|
; X86-NEXT: movl {{[-0-9]+}}(%e{{[sb]}}p), %eax # 4-byte Reload
|
||
|
|
; X86-NEXT: shrdl $1, %eax, %edx
|
||
|
|
; X86-NEXT: movl (%esp), %eax # 4-byte Reload
|
||
|
|
; X86-NEXT: shrdl $1, %eax, %edi
|
||
|
|
; X86-NEXT: shrdl $1, %ebx, %ebp
|
||
|
|
; X86-NEXT: shrdl $1, %esi, %ecx
|
||
|
|
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
|
||
|
|
; X86-NEXT: movl %ecx, 12(%eax)
|
||
|
|
; X86-NEXT: movl %ebp, 8(%eax)
|
||
|
|
; X86-NEXT: movl %edi, 4(%eax)
|
||
|
|
; X86-NEXT: movl %edx, (%eax)
|
||
|
|
; X86-NEXT: addl $12, %esp
|
||
|
|
; X86-NEXT: popl %esi
|
||
|
|
; X86-NEXT: popl %edi
|
||
|
|
; X86-NEXT: popl %ebx
|
||
|
|
; X86-NEXT: popl %ebp
|
||
|
|
; X86-NEXT: retl $4
|
||
|
|
%tmp = call <4 x i32> @llvm.udiv.fix.sat.v4i32(<4 x i32> %x, <4 x i32> %y, i32 31)
|
||
|
|
ret <4 x i32> %tmp
|
||
|
|
}
|