RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-01 08:23:21 +01:00
Code Issues Projects Releases Wiki Activity
48961d3ca6
llvm-mirror/test/CodeGen/X86/memcmp.ll

111 lines
3.5 KiB
LLVM
Raw Normal View History

move an optimization for memcmp out of simplifylibcalls and into SDISel. This optimization was causing simplifylibcalls to introduce type-unsafe nastiness. This is the first step, I'll be expanding the memcmp optimizations shortly, covering things that we really really wouldn't want simplifylibcalls to do. llvm-svn: 92098
2009-12-24 01:37:38 +01:00
; RUN: llc %s -o - -march=x86-64 | FileCheck %s
; This tests codegen time inlining/optimization of memcmp
; rdar://6480398
handle equality memcmp of 8 bytes on x86-64 with two unaligned loads and a compare. On other targets we end up with a call to memcmp because we don't want 16 individual byte loads. We should be able to use movups as well, but we're failing to select the generated icmp. llvm-svn: 92107
2009-12-24 02:07:17 +01:00
@.str = private constant [23 x i8] c"fooooooooooooooooooooo\00", align 1 ; <[23 x i8]*> [#uses=1]
move an optimization for memcmp out of simplifylibcalls and into SDISel. This optimization was causing simplifylibcalls to introduce type-unsafe nastiness. This is the first step, I'll be expanding the memcmp optimizations shortly, covering things that we really really wouldn't want simplifylibcalls to do. llvm-svn: 92098
2009-12-24 01:37:38 +01:00
declare i32 @memcmp(...)
define void @memcmp2(i8* %X, i8* %Y, i32* nocapture %P) nounwind {
entry:
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* %Y, i32 2) nounwind ; <i32> [#uses=1]
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp2:
; CHECK: movw (%rsi), %ax
; CHECK: cmpw %ax, (%rdi)
}
define void @memcmp2a(i8* %X, i32* nocapture %P) nounwind {
entry:
handle equality memcmp of 8 bytes on x86-64 with two unaligned loads and a compare. On other targets we end up with a call to memcmp because we don't want 16 individual byte loads. We should be able to use movups as well, but we're failing to select the generated icmp. llvm-svn: 92107
2009-12-24 02:07:17 +01:00
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* getelementptr inbounds ([23 x i8]* @.str, i32 0, i32 1), i32 2) nounwind ; <i32> [#uses=1]
move an optimization for memcmp out of simplifylibcalls and into SDISel. This optimization was causing simplifylibcalls to introduce type-unsafe nastiness. This is the first step, I'll be expanding the memcmp optimizations shortly, covering things that we really really wouldn't want simplifylibcalls to do. llvm-svn: 92098
2009-12-24 01:37:38 +01:00
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp2a:
; CHECK: cmpw $28527, (%rdi)
}
define void @memcmp4(i8* %X, i8* %Y, i32* nocapture %P) nounwind {
entry:
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* %Y, i32 4) nounwind ; <i32> [#uses=1]
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp4:
; CHECK: movl (%rsi), %eax
; CHECK: cmpl %eax, (%rdi)
}
define void @memcmp4a(i8* %X, i32* nocapture %P) nounwind {
entry:
handle equality memcmp of 8 bytes on x86-64 with two unaligned loads and a compare. On other targets we end up with a call to memcmp because we don't want 16 individual byte loads. We should be able to use movups as well, but we're failing to select the generated icmp. llvm-svn: 92107
2009-12-24 02:07:17 +01:00
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* getelementptr inbounds ([23 x i8]* @.str, i32 0, i32 1), i32 4) nounwind ; <i32> [#uses=1]
move an optimization for memcmp out of simplifylibcalls and into SDISel. This optimization was causing simplifylibcalls to introduce type-unsafe nastiness. This is the first step, I'll be expanding the memcmp optimizations shortly, covering things that we really really wouldn't want simplifylibcalls to do. llvm-svn: 92098
2009-12-24 01:37:38 +01:00
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp4a:
handle equality memcmp of 8 bytes on x86-64 with two unaligned loads and a compare. On other targets we end up with a call to memcmp because we don't want 16 individual byte loads. We should be able to use movups as well, but we're failing to select the generated icmp. llvm-svn: 92107
2009-12-24 02:07:17 +01:00
; CHECK: cmpl $1869573999, (%rdi)
}
define void @memcmp8(i8* %X, i8* %Y, i32* nocapture %P) nounwind {
entry:
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* %Y, i32 8) nounwind ; <i32> [#uses=1]
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp8:
; CHECK: movq (%rsi), %rax
; CHECK: cmpq %rax, (%rdi)
}
define void @memcmp8a(i8* %X, i32* nocapture %P) nounwind {
entry:
%0 = tail call i32 (...)* @memcmp(i8* %X, i8* getelementptr inbounds ([23 x i8]* @.str, i32 0, i32 0), i32 8) nounwind ; <i32> [#uses=1]
%1 = icmp eq i32 %0, 0 ; <i1> [#uses=1]
br i1 %1, label %return, label %bb
bb: ; preds = %entry
store i32 4, i32* %P, align 4
ret void
return: ; preds = %entry
ret void
; CHECK: memcmp8a:
; CHECK: movabsq $8029759185026510694, %rax
; CHECK: cmpq %rax, (%rdi)
move an optimization for memcmp out of simplifylibcalls and into SDISel. This optimization was causing simplifylibcalls to introduce type-unsafe nastiness. This is the first step, I'll be expanding the memcmp optimizations shortly, covering things that we really really wouldn't want simplifylibcalls to do. llvm-svn: 92098
2009-12-24 01:37:38 +01:00
}
Reference in New Issue Copy Permalink