maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/memcpy-struct-by-value.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

191 lines
6.8 KiB
LLVM
Raw Normal View History

[NFC] Update memcpy tests Summary: Runs utils/update_llc_test_checks.py on a few memcpy files Reviewers: courbet Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D61507 Remove cfi noise by adding nounwind llvm-svn: 360023
2019-05-06 17:46:50 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=i686-linux-gnu -mattr=-ermsb < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=NOFAST32
; RUN: llc -mtriple=i686-linux-gnu -mattr=+ermsb < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=FAST32
Rename FastString flag. llvm-svn: 300959
2017-04-21 17:20:50 +08:00
; RUN: llc -mtriple=x86_64-linux-gnu -mattr=-ermsb < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=NOFAST
; RUN: llc -mtriple=x86_64-linux-gnu -mattr=+ermsb < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=FAST
add 32 bit tests llvm-svn: 300961
2017-04-21 17:20:58 +08:00
; RUN: llc -mtriple=x86_64-linux-gnu -mcpu=generic < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=NOFAST
; RUN: llc -mtriple=x86_64-linux-gnu -mcpu=haswell < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=FAST
add skylake llvm-svn: 300962
2017-04-21 17:21:01 +08:00
; RUN: llc -mtriple=x86_64-linux-gnu -mcpu=skylake < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=FAST
; FIXME: The documentation states that ivybridge has ermsb, but this is not
; enabled right now since I could not confirm by testing.
add 32 bit tests llvm-svn: 300961
2017-04-21 17:20:58 +08:00
; RUN: llc -mtriple=x86_64-linux-gnu -mcpu=ivybridge < %s -o - | FileCheck %s --check-prefix=ALL --check-prefix=NOFAST
X86 memcpy: use REPMOVSB instead of REPMOVS{Q,D,W} for inline copies when the subtarget has fast strings. This has two advantages: - Speed is improved. For example, on Haswell thoughput improvements increase linearly with size from 256 to 512 bytes, after which they plateau: (e.g. 1% for 260 bytes, 25% for 400 bytes, 40% for 508 bytes). - Code is much smaller (no need to handle boundaries). llvm-svn: 300957
2017-04-21 17:20:39 +08:00
%struct.large = type { [4096 x i8] }
declare void @foo(%struct.large* align 8 byval) nounwind
define void @test1(%struct.large* nocapture %x) nounwind {
[NFC] Update memcpy tests Summary: Runs utils/update_llc_test_checks.py on a few memcpy files Reviewers: courbet Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D61507 Remove cfi noise by adding nounwind llvm-svn: 360023
2019-05-06 17:46:50 +08:00
; NOFAST32-LABEL: test1:
; NOFAST32: # %bb.0:
; NOFAST32-NEXT: pushl %edi
; NOFAST32-NEXT: pushl %esi
; NOFAST32-NEXT: subl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; NOFAST32-NEXT: movl $1024, %ecx # imm = 0x400
; NOFAST32-NEXT: movl %esp, %edi
; NOFAST32-NEXT: rep;movsl (%esi), %es:(%edi)
; NOFAST32-NEXT: calll foo
; NOFAST32-NEXT: addl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: popl %esi
; NOFAST32-NEXT: popl %edi
; NOFAST32-NEXT: retl
;
; FAST32-LABEL: test1:
; FAST32: # %bb.0:
; FAST32-NEXT: pushl %edi
; FAST32-NEXT: pushl %esi
; FAST32-NEXT: subl $4100, %esp # imm = 0x1004
; FAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; FAST32-NEXT: movl $4096, %ecx # imm = 0x1000
; FAST32-NEXT: movl %esp, %edi
; FAST32-NEXT: rep;movsb (%esi), %es:(%edi)
; FAST32-NEXT: calll foo
; FAST32-NEXT: addl $4100, %esp # imm = 0x1004
; FAST32-NEXT: popl %esi
; FAST32-NEXT: popl %edi
; FAST32-NEXT: retl
;
; NOFAST-LABEL: test1:
; NOFAST: # %bb.0:
; NOFAST-NEXT: subq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: movq %rdi, %rsi
; NOFAST-NEXT: movl $512, %ecx # imm = 0x200
; NOFAST-NEXT: movq %rsp, %rdi
; NOFAST-NEXT: rep;movsq (%rsi), %es:(%rdi)
; NOFAST-NEXT: callq foo
; NOFAST-NEXT: addq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: retq
;
; FAST-LABEL: test1:
; FAST: # %bb.0:
; FAST-NEXT: subq $4104, %rsp # imm = 0x1008
; FAST-NEXT: movq %rdi, %rsi
; FAST-NEXT: movl $4096, %ecx # imm = 0x1000
; FAST-NEXT: movq %rsp, %rdi
; FAST-NEXT: rep;movsb (%rsi), %es:(%rdi)
; FAST-NEXT: callq foo
; FAST-NEXT: addq $4104, %rsp # imm = 0x1008
; FAST-NEXT: retq
X86 memcpy: use REPMOVSB instead of REPMOVS{Q,D,W} for inline copies when the subtarget has fast strings. This has two advantages: - Speed is improved. For example, on Haswell thoughput improvements increase linearly with size from 256 to 512 bytes, after which they plateau: (e.g. 1% for 260 bytes, 25% for 400 bytes, 40% for 508 bytes). - Code is much smaller (no need to handle boundaries). llvm-svn: 300957
2017-04-21 17:20:39 +08:00
call void @foo(%struct.large* align 8 byval %x)
ret void
}
use repmovsb when optimizing forminsize llvm-svn: 300960
2017-04-21 17:20:55 +08:00
define void @test2(%struct.large* nocapture %x) nounwind minsize {
[NFC] Update memcpy tests Summary: Runs utils/update_llc_test_checks.py on a few memcpy files Reviewers: courbet Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D61507 Remove cfi noise by adding nounwind llvm-svn: 360023
2019-05-06 17:46:50 +08:00
; NOFAST32-LABEL: test2:
; NOFAST32: # %bb.0:
; NOFAST32-NEXT: pushl %edi
; NOFAST32-NEXT: pushl %esi
; NOFAST32-NEXT: subl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; NOFAST32-NEXT: movl $1024, %ecx # imm = 0x400
; NOFAST32-NEXT: movl %esp, %edi
; NOFAST32-NEXT: rep;movsl (%esi), %es:(%edi)
; NOFAST32-NEXT: calll foo
; NOFAST32-NEXT: addl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: popl %esi
; NOFAST32-NEXT: popl %edi
; NOFAST32-NEXT: retl
;
; FAST32-LABEL: test2:
; FAST32: # %bb.0:
; FAST32-NEXT: pushl %edi
; FAST32-NEXT: pushl %esi
; FAST32-NEXT: subl $4100, %esp # imm = 0x1004
; FAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; FAST32-NEXT: movl $4096, %ecx # imm = 0x1000
; FAST32-NEXT: movl %esp, %edi
; FAST32-NEXT: rep;movsb (%esi), %es:(%edi)
; FAST32-NEXT: calll foo
; FAST32-NEXT: addl $4100, %esp # imm = 0x1004
; FAST32-NEXT: popl %esi
; FAST32-NEXT: popl %edi
; FAST32-NEXT: retl
;
; NOFAST-LABEL: test2:
; NOFAST: # %bb.0:
; NOFAST-NEXT: subq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: movq %rdi, %rsi
; NOFAST-NEXT: movl $512, %ecx # imm = 0x200
; NOFAST-NEXT: movq %rsp, %rdi
; NOFAST-NEXT: rep;movsq (%rsi), %es:(%rdi)
; NOFAST-NEXT: callq foo
; NOFAST-NEXT: addq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: retq
;
; FAST-LABEL: test2:
; FAST: # %bb.0:
; FAST-NEXT: subq $4104, %rsp # imm = 0x1008
; FAST-NEXT: movq %rdi, %rsi
; FAST-NEXT: movl $4096, %ecx # imm = 0x1000
; FAST-NEXT: movq %rsp, %rdi
; FAST-NEXT: rep;movsb (%rsi), %es:(%rdi)
; FAST-NEXT: callq foo
; FAST-NEXT: addq $4104, %rsp # imm = 0x1008
; FAST-NEXT: retq
use repmovsb when optimizing forminsize llvm-svn: 300960
2017-04-21 17:20:55 +08:00
call void @foo(%struct.large* align 8 byval %x)
ret void
}
%struct.large_oddsize = type { [4095 x i8] }
declare void @foo_oddsize(%struct.large_oddsize* align 8 byval) nounwind
define void @test3(%struct.large_oddsize* nocapture %x) nounwind minsize {
[NFC] Update memcpy tests Summary: Runs utils/update_llc_test_checks.py on a few memcpy files Reviewers: courbet Subscribers: llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D61507 Remove cfi noise by adding nounwind llvm-svn: 360023
2019-05-06 17:46:50 +08:00
; NOFAST32-LABEL: test3:
; NOFAST32: # %bb.0:
; NOFAST32-NEXT: pushl %edi
; NOFAST32-NEXT: pushl %esi
; NOFAST32-NEXT: subl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; NOFAST32-NEXT: movl $4095, %ecx # imm = 0xFFF
; NOFAST32-NEXT: movl %esp, %edi
; NOFAST32-NEXT: rep;movsb (%esi), %es:(%edi)
; NOFAST32-NEXT: calll foo_oddsize
; NOFAST32-NEXT: addl $4100, %esp # imm = 0x1004
; NOFAST32-NEXT: popl %esi
; NOFAST32-NEXT: popl %edi
; NOFAST32-NEXT: retl
;
; FAST32-LABEL: test3:
; FAST32: # %bb.0:
; FAST32-NEXT: pushl %edi
; FAST32-NEXT: pushl %esi
; FAST32-NEXT: subl $4100, %esp # imm = 0x1004
; FAST32-NEXT: movl {{[0-9]+}}(%esp), %esi
; FAST32-NEXT: movl $4095, %ecx # imm = 0xFFF
; FAST32-NEXT: movl %esp, %edi
; FAST32-NEXT: rep;movsb (%esi), %es:(%edi)
; FAST32-NEXT: calll foo_oddsize
; FAST32-NEXT: addl $4100, %esp # imm = 0x1004
; FAST32-NEXT: popl %esi
; FAST32-NEXT: popl %edi
; FAST32-NEXT: retl
;
; NOFAST-LABEL: test3:
; NOFAST: # %bb.0:
; NOFAST-NEXT: subq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: movq %rdi, %rsi
; NOFAST-NEXT: movl $4095, %ecx # imm = 0xFFF
; NOFAST-NEXT: movq %rsp, %rdi
; NOFAST-NEXT: rep;movsb (%rsi), %es:(%rdi)
; NOFAST-NEXT: callq foo_oddsize
; NOFAST-NEXT: addq $4104, %rsp # imm = 0x1008
; NOFAST-NEXT: retq
;
; FAST-LABEL: test3:
; FAST: # %bb.0:
; FAST-NEXT: subq $4104, %rsp # imm = 0x1008
; FAST-NEXT: movq %rdi, %rsi
; FAST-NEXT: movl $4095, %ecx # imm = 0xFFF
; FAST-NEXT: movq %rsp, %rdi
; FAST-NEXT: rep;movsb (%rsi), %es:(%rdi)
; FAST-NEXT: callq foo_oddsize
; FAST-NEXT: addq $4104, %rsp # imm = 0x1008
; FAST-NEXT: retq
use repmovsb when optimizing forminsize llvm-svn: 300960
2017-04-21 17:20:55 +08:00
call void @foo_oddsize(%struct.large_oddsize* align 8 byval %x)
ret void
}