maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/MemCpyOpt/memcpy.ll

173 lines
5.4 KiB
LLVM
Raw Normal View History

remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; RUN: opt < %s -basicaa -memcpyopt -dse -S | FileCheck %s
Allow GVN to hack on memcpy's, making them open to further optimization. llvm-svn: 46693
2008-02-04 10:59:58 +08:00
target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
target triple = "i686-apple-darwin9"
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
%0 = type { x86_fp80, x86_fp80 }
%1 = type { i32, i32 }
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-10 01:41:24 +08:00
define void @test1(%0* sret %agg.result, x86_fp80 %z.0, x86_fp80 %z.1) nounwind {
Allow GVN to hack on memcpy's, making them open to further optimization. llvm-svn: 46693
2008-02-04 10:59:58 +08:00
entry:
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
%tmp2 = alloca %0
%memtmp = alloca %0, align 16
%tmp5 = fsub x86_fp80 0xK80000000000000000000, %z.1
call void @ccoshl(%0* sret %memtmp, x86_fp80 %tmp5, x86_fp80 %z.0) nounwind
%tmp219 = bitcast %0* %tmp2 to i8*
%memtmp20 = bitcast %0* %memtmp to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %tmp219, i8* %memtmp20, i32 32, i32 16, i1 false)
%agg.result21 = bitcast %0* %agg.result to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %agg.result21, i8* %tmp219, i32 32, i32 16, i1 false)
ret void
filecheckize, this is still not optimal, see PR8643 llvm-svn: 119693
2010-11-18 15:49:32 +08:00
; Check that one of the memcpy's are removed.
;; FIXME: PR 8643 We should be able to eliminate the last memcpy here.
; CHECK: @test1
; CHECK: call void @ccoshl
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; CHECK: call void @llvm.memcpy
filecheckize, this is still not optimal, see PR8643 llvm-svn: 119693
2010-11-18 15:49:32 +08:00
; CHECK-NOT: llvm.memcpy
; CHECK: ret void
Allow GVN to hack on memcpy's, making them open to further optimization. llvm-svn: 46693
2008-02-04 10:59:58 +08:00
}
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-10 01:41:24 +08:00
declare void @ccoshl(%0* sret , x86_fp80, x86_fp80) nounwind
Allow GVN to hack on memcpy's, making them open to further optimization. llvm-svn: 46693
2008-02-04 10:59:58 +08:00
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; The intermediate alloca and one of the memcpy's should be eliminated, the
; other should be related with a memmove.
define void @test2(i8* %P, i8* %Q) nounwind {
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
%memtmp = alloca %0, align 16
%R = bitcast %0* %memtmp to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %R, i8* %P, i32 32, i32 16, i1 false)
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %Q, i8* %R, i32 32, i32 16, i1 false)
ret void
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; CHECK: @test2
; CHECK-NEXT: call void @llvm.memmove{{.*}}(i8* %Q, i8* %P
; CHECK-NEXT: ret void
}
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-10 01:41:24 +08:00
@x = external global %0
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-10 01:41:24 +08:00
define void @test3(%0* noalias sret %agg.result) nounwind {
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
%x.0 = alloca %0
%x.01 = bitcast %0* %x.0 to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %x.01, i8* bitcast (%0* @x to i8*), i32 32, i32 16, i1 false)
%agg.result2 = bitcast %0* %agg.result to i8*
call void @llvm.memcpy.p0i8.p0i8.i32(i8* %agg.result2, i8* %x.01, i32 32, i32 16, i1 false)
ret void
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; CHECK: @test3
When looking for dependencies on the src pointer, scan the src pointer. Scanning on the memcpy call will pull up other unrelated stuff. Fixes PR11142. llvm-svn: 142150
2011-10-17 04:13:32 +08:00
; CHECK-NEXT: %agg.result1 = bitcast
remove a pointless restriction from memcpyopt. It was refusing to optimize two memcpy's like this: copy A <- B copy C <- A if it couldn't prove that noalias(B,C). We can eliminate the copy by producing a memmove instead of memcpy. llvm-svn: 119694
2010-11-18 16:00:57 +08:00
; CHECK-NEXT: call void @llvm.memcpy
; CHECK-NEXT: ret void
}
Implement PR8644: forwarding a memcpy value to a byval, allowing the memcpy to be eliminated. Unfortunately, the requirements on byval's without explicit alignment are really weak and impossible to predict in the mid-level optimizer, so this doesn't kick in much with current frontends. The fix is to change clang to set alignment on all byval arguments. llvm-svn: 119916
2010-11-21 08:28:59 +08:00
; PR8644
define void @test4(i8 *%P) {
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
%A = alloca %1
%a = bitcast %1* %A to i8*
Implement PR8644: forwarding a memcpy value to a byval, allowing the memcpy to be eliminated. Unfortunately, the requirements on byval's without explicit alignment are really weak and impossible to predict in the mid-level optimizer, so this doesn't kick in much with current frontends. The fix is to change clang to set alignment on all byval arguments. llvm-svn: 119916
2010-11-21 08:28:59 +08:00
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %a, i8* %P, i64 8, i32 4, i1 false)
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
call void @test4a(i8* byval align 1 %a)
Implement PR8644: forwarding a memcpy value to a byval, allowing the memcpy to be eliminated. Unfortunately, the requirements on byval's without explicit alignment are really weak and impossible to predict in the mid-level optimizer, so this doesn't kick in much with current frontends. The fix is to change clang to set alignment on all byval arguments. llvm-svn: 119916
2010-11-21 08:28:59 +08:00
ret void
; CHECK: @test4
; CHECK-NEXT: call void @test4a(
}
declare void @test4a(i8* byval align 1)
declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
fix a bozo bug I introduced in r119930, causing a miscompile of 20040709-1.c from the gcc testsuite. I was using the size of a pointer instead of the pointee. This fixes rdar://8713376 llvm-svn: 120519
2010-12-01 09:24:55 +08:00
%struct.S = type { i128, [4 x i8]}
@sS = external global %struct.S, align 16
declare void @test5a(%struct.S* byval align 16) nounwind ssp
; rdar://8713376 - This memcpy can't be eliminated.
define i32 @test5(i32 %x) nounwind ssp {
entry:
%y = alloca %struct.S, align 16
%tmp = bitcast %struct.S* %y to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* bitcast (%struct.S* @sS to i8*), i64 32, i32 16, i1 false)
%a = getelementptr %struct.S* %y, i64 0, i32 1, i64 0
store i8 4, i8* %a
call void @test5a(%struct.S* byval align 16 %y)
ret i32 0
; CHECK: @test5(
; CHECK: store i8 4
; CHECK: call void @test5a(%struct.S* byval align 16 %y)
}
enhance memcpyopt to zap memcpy's that have the same src/dst. llvm-svn: 121362
2010-12-09 15:45:45 +08:00
;; Noop memcpy should be zapped.
define void @test6(i8 *%P) {
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %P, i8* %P, i64 8, i32 4, i1 false)
ret void
; CHECK: @test6
; CHECK-NEXT: ret void
}
Teach valuetracking that byval arguments with a specified alignment are aligned. Teach memcpyopt to not give up all hope when confonted with an underaligned memcpy feeding an overaligned byval. If the *source* of the memcpy can be determined to be adequeately aligned, or if it can be forced to be, we can eliminate the memcpy. This addresses PR9794. We now compile the example into: define i32 @f(%struct.p* nocapture byval align 8 %q) nounwind ssp { entry: %call = call i32 @g(%struct.p* byval align 8 %q) nounwind ret i32 %call } in both x86-64 and x86-32 mode. We still don't get a tailcall though, because tailcalls apparently can't handle byval. llvm-svn: 131884
2011-05-23 08:03:39 +08:00
; PR9794 - Should forward memcpy into byval argument even though the memcpy
; isn't itself 8 byte aligned.
%struct.p = type { i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32, i32 }
define i32 @test7(%struct.p* nocapture byval align 8 %q) nounwind ssp {
entry:
%agg.tmp = alloca %struct.p, align 4
%tmp = bitcast %struct.p* %agg.tmp to i8*
%tmp1 = bitcast %struct.p* %q to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* %tmp1, i64 48, i32 4, i1 false)
%call = call i32 @g(%struct.p* byval align 8 %agg.tmp) nounwind
ret i32 %call
; CHECK: @test7
; CHECK: call i32 @g(%struct.p* byval align 8 %q) nounwind
}
declare i32 @g(%struct.p* byval align 8)
rip out a ton of intrinsic modernization logic from AutoUpgrade.cpp, which is for pre-2.9 bitcode files. We keep x86 unaligned loads, movnt, crc32, and the target indep prefetch change. As usual, updating the testsuite is a PITA. llvm-svn: 133337
2011-06-18 14:05:24 +08:00
declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture, i8* nocapture, i32, i32, i1) nounwind
Teach valuetracking that byval arguments with a specified alignment are aligned. Teach memcpyopt to not give up all hope when confonted with an underaligned memcpy feeding an overaligned byval. If the *source* of the memcpy can be determined to be adequeately aligned, or if it can be forced to be, we can eliminate the memcpy. This addresses PR9794. We now compile the example into: define i32 @f(%struct.p* nocapture byval align 8 %q) nounwind ssp { entry: %call = call i32 @g(%struct.p* byval align 8 %q) nounwind ret i32 %call } in both x86-64 and x86-32 mode. We still don't get a tailcall though, because tailcalls apparently can't handle byval. llvm-svn: 131884
2011-05-23 08:03:39 +08:00
When looking for dependencies on the src pointer, scan the src pointer. Scanning on the memcpy call will pull up other unrelated stuff. Fixes PR11142. llvm-svn: 142150
2011-10-17 04:13:32 +08:00
; PR11142 - When looking for a memcpy-memcpy dependency, don't get stuck on
; instructions between the memcpy's that only affect the destination pointer.
@test8.str = internal constant [7 x i8] c"ABCDEF\00"
define void @test8() {
; CHECK: test8
Oops! Fix test I forgot to submit as part of r142735. llvm-svn: 142736
2011-10-23 06:07:31 +08:00
; CHECK-NOT: memcpy
When looking for dependencies on the src pointer, scan the src pointer. Scanning on the memcpy call will pull up other unrelated stuff. Fixes PR11142. llvm-svn: 142150
2011-10-17 04:13:32 +08:00
%A = tail call i8* @malloc(i32 10)
%B = getelementptr inbounds i8* %A, i64 2
tail call void @llvm.memcpy.p0i8.p0i8.i32(i8* %B, i8* getelementptr inbounds ([7 x i8]* @test8.str, i64 0, i64 0), i32 7, i32 1, i1 false)
%C = tail call i8* @malloc(i32 10)
%D = getelementptr inbounds i8* %C, i64 2
tail call void @llvm.memcpy.p0i8.p0i8.i32(i8* %D, i8* %B, i32 7, i32 1, i1 false)
ret void
Oops! Fix test I forgot to submit as part of r142735. llvm-svn: 142736
2011-10-23 06:07:31 +08:00
; CHECK: ret void
When looking for dependencies on the src pointer, scan the src pointer. Scanning on the memcpy call will pull up other unrelated stuff. Fixes PR11142. llvm-svn: 142150
2011-10-17 04:13:32 +08:00
}
declare noalias i8* @malloc(i32)
Move the capture analysis from MemoryDependencyAnalysis to a more general place so that it can be reused in MemCpyOptimizer. This analysis is needed to remove an unnecessary memcpy when returning a struct into a local variable. rdar://11341081 PR12686 llvm-svn: 156776
2012-05-15 04:35:04 +08:00
; rdar://11341081
%struct.big = type { [50 x i32] }
define void @test9() nounwind uwtable ssp {
entry:
; CHECK: test9
; CHECK: f1
; CHECK-NOT: memcpy
; CHECK: f2
%b = alloca %struct.big, align 4
%tmp = alloca %struct.big, align 4
call void @f1(%struct.big* sret %tmp)
%0 = bitcast %struct.big* %b to i8*
%1 = bitcast %struct.big* %tmp to i8*
call void @llvm.memcpy.p0i8.p0i8.i64(i8* %0, i8* %1, i64 200, i32 4, i1 false)
call void @f2(%struct.big* %b)
ret void
}
declare void @f1(%struct.big* sret)
declare void @f2(%struct.big*)