2016-11-16 22:12:11 +08:00
|
|
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
|
|
|
; RUN: llc < %s -mtriple=i686-unknown | FileCheck %s --check-prefix=X32
|
|
|
|
|
; RUN: llc < %s -mtriple=x86_64-unknown | FileCheck %s --check-prefix=X64
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
|
|
|
|
|
define zeroext i8 @test_udivrem_zext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_udivrem_zext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: divb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movzbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: movb %al, z
|
|
|
|
|
; X32-NEXT: movl %ecx, %eax
|
|
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_udivrem_zext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movzbl %dil, %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: divb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movzbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: movb %al, {{.*}}(%rip)
|
|
|
|
|
; X64-NEXT: movl %ecx, %eax
|
|
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%div = udiv i8 %x, %y
|
|
|
|
|
store i8 %div, i8* @z
|
|
|
|
|
%1 = urem i8 %x, %y
|
|
|
|
|
ret i8 %1
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define zeroext i8 @test_urem_zext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_urem_zext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: divb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movzbl %ah, %eax # NOREX
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_urem_zext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movzbl %dil, %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: divb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movzbl %ah, %eax # NOREX
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = urem i8 %x, %y
|
|
|
|
|
ret i8 %1
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define i8 @test_urem_noext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_urem_noext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %cl
|
|
|
|
|
; X32-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: divb %cl
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movzbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: addb %cl, %al
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_urem_noext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movzbl %dil, %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: divb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movzbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: addb %sil, %al
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = urem i8 %x, %y
|
|
|
|
|
%2 = add i8 %1, %y
|
|
|
|
|
ret i8 %2
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define i64 @test_urem_zext64_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_urem_zext64_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: divb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movzbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: xorl %edx, %edx
|
|
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_urem_zext64_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movzbl %dil, %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: divb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movzbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = urem i8 %x, %y
|
|
|
|
|
%2 = zext i8 %1 to i64
|
|
|
|
|
ret i64 %2
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define signext i8 @test_sdivrem_sext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_sdivrem_sext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %al
|
|
|
|
|
; X32-NEXT: cbtw
|
|
|
|
|
; X32-NEXT: idivb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movsbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: movb %al, z
|
|
|
|
|
; X32-NEXT: movl %ecx, %eax
|
|
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_sdivrem_sext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movl %edi, %eax
|
|
|
|
|
; X64-NEXT: cbtw
|
|
|
|
|
; X64-NEXT: idivb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movsbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: movb %al, {{.*}}(%rip)
|
|
|
|
|
; X64-NEXT: movl %ecx, %eax
|
|
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%div = sdiv i8 %x, %y
|
|
|
|
|
store i8 %div, i8* @z
|
|
|
|
|
%1 = srem i8 %x, %y
|
|
|
|
|
ret i8 %1
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define signext i8 @test_srem_sext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_srem_sext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %al
|
|
|
|
|
; X32-NEXT: cbtw
|
|
|
|
|
; X32-NEXT: idivb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movsbl %ah, %eax # NOREX
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_srem_sext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movl %edi, %eax
|
|
|
|
|
; X64-NEXT: cbtw
|
|
|
|
|
; X64-NEXT: idivb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movsbl %ah, %eax # NOREX
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = srem i8 %x, %y
|
|
|
|
|
ret i8 %1
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define i8 @test_srem_noext_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_srem_noext_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %al
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %cl
|
|
|
|
|
; X32-NEXT: cbtw
|
|
|
|
|
; X32-NEXT: idivb %cl
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movsbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: addb %cl, %al
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_srem_noext_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movl %edi, %eax
|
|
|
|
|
; X64-NEXT: cbtw
|
|
|
|
|
; X64-NEXT: idivb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movsbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: addb %sil, %al
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %al<def> %al<kill> %eax<kill>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = srem i8 %x, %y
|
|
|
|
|
%2 = add i8 %1, %y
|
|
|
|
|
ret i8 %2
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
define i64 @test_srem_sext64_ah(i8 %x, i8 %y) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: test_srem_sext64_ah:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movb {{[0-9]+}}(%esp), %al
|
|
|
|
|
; X32-NEXT: cbtw
|
|
|
|
|
; X32-NEXT: idivb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movsbl %ah, %eax # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: movl %eax, %edx
|
|
|
|
|
; X32-NEXT: sarl $31, %edx
|
|
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: test_srem_sext64_ah:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movl %edi, %eax
|
|
|
|
|
; X64-NEXT: cbtw
|
|
|
|
|
; X64-NEXT: idivb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movsbl %ah, %eax # NOREX
|
2017-10-27 05:12:03 +08:00
|
|
|
; X64-NEXT: cltq
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: retq
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
%1 = srem i8 %x, %y
|
|
|
|
|
%2 = sext i8 %1 to i64
|
|
|
|
|
ret i64 %2
|
|
|
|
|
}
|
|
|
|
|
|
2016-01-06 17:41:10 +08:00
|
|
|
define i64 @pr25754(i8 %a, i8 %c) {
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-LABEL: pr25754:
|
|
|
|
|
; X32: # BB#0:
|
|
|
|
|
; X32-NEXT: movzbl {{[0-9]+}}(%esp), %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X32-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: divb {{[0-9]+}}(%esp)
|
2017-09-28 04:34:17 +08:00
|
|
|
; X32-NEXT: movzbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: movzbl %al, %eax
|
|
|
|
|
; X32-NEXT: addl %ecx, %eax
|
2017-02-06 22:54:49 +08:00
|
|
|
; X32-NEXT: xorl %edx, %edx
|
2016-11-16 22:12:11 +08:00
|
|
|
; X32-NEXT: retl
|
|
|
|
|
;
|
|
|
|
|
; X64-LABEL: pr25754:
|
|
|
|
|
; X64: # BB#0:
|
|
|
|
|
; X64-NEXT: movzbl %dil, %eax
|
2017-11-29 01:15:09 +08:00
|
|
|
; X64-NEXT: # kill: %eax<def> %eax<kill> %ax<def>
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: divb %sil
|
2017-09-28 04:34:17 +08:00
|
|
|
; X64-NEXT: movzbl %ah, %ecx # NOREX
|
2016-11-16 22:12:11 +08:00
|
|
|
; X64-NEXT: movzbl %al, %eax
|
|
|
|
|
; X64-NEXT: addq %rcx, %rax
|
|
|
|
|
; X64-NEXT: retq
|
2016-01-06 17:41:10 +08:00
|
|
|
%r1 = urem i8 %a, %c
|
|
|
|
|
%d1 = udiv i8 %a, %c
|
|
|
|
|
%r2 = zext i8 %r1 to i64
|
|
|
|
|
%d2 = zext i8 %d1 to i64
|
|
|
|
|
%ret = add i64 %r2, %d2
|
|
|
|
|
ret i64 %ret
|
|
|
|
|
}
|
|
|
|
|
|
[X86] 8bit divrem: Improve codegen for AH register extraction.
For 8-bit divrems where the remainder is used, we used to generate:
divb %sil
shrw $8, %ax
movzbl %al, %eax
That was to avoid an H-reg access, which is problematic mainly because
it isn't possible in REX-prefixed instructions.
This patch optimizes that to:
divb %sil
movzbl %ah, %eax
To do that, we explicitly extend AH, and extract the L-subreg in the
resulting register. The extension is done using the NOREX variants of
MOVZX. To support signed operations, MOVSX_NOREX is also added.
Further, this introduces a new SDNode type, [us]divrem_ext_hreg, which is
then lowered to a sequence containing a single zext (rather than 2).
Differential Revision: http://reviews.llvm.org/D6064
llvm-svn: 221176
2014-11-04 04:26:35 +08:00
|
|
|
@z = external global i8
|
