replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[MemorySSA] Consider callsite args for hashing and equality. 

We use a `DenseMap<MemoryLocOrCall, MemlocStackInfo>` to keep track of
prior work when optimizing uses in MemorySSA. Because we weren't
accounting for callsite arguments in either the hash code or equality
tests for `MemoryLocOrCall`s, we optimized uses too aggressively in
some rare cases.

Fix by Daniel Berlin.

Should fix PR36883.

llvm-svn: 328748
This commit is contained in:
George Burgess IV 2018-03-29 00:54:39 +00:00
parent 88a16a0790
commit 3588fd4865
2 changed files with 58 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

29
llvm/lib/Analysis/MemorySSA.cpp
View File

@ -153,9 +153,14 @@ public:
if (IsCall != Other.IsCall)
return false;
if (IsCall)
return CS.getCalledValue() == Other.CS.getCalledValue();
return Loc == Other.Loc;
if (!IsCall)
return Loc == Other.Loc;
if (CS.getCalledValue() != Other.CS.getCalledValue())
return false;
assert(CS.arg_size() == Other.CS.arg_size());
return std::equal(CS.arg_begin(), CS.arg_end(), Other.CS.arg_begin());
}
private:
@ -179,12 +184,18 @@ template <> struct DenseMapInfo<MemoryLocOrCall> {
}
static unsigned getHashValue(const MemoryLocOrCall &MLOC) {
if (MLOC.IsCall)
return hash_combine(MLOC.IsCall,
DenseMapInfo<const Value *>::getHashValue(
MLOC.getCS().getCalledValue()));
return hash_combine(
MLOC.IsCall, DenseMapInfo<MemoryLocation>::getHashValue(MLOC.getLoc()));
if (!MLOC.IsCall)
return hash_combine(
MLOC.IsCall,
DenseMapInfo<MemoryLocation>::getHashValue(MLOC.getLoc()));
hash_code hash =
hash_combine(MLOC.IsCall, DenseMapInfo<const Value *>::getHashValue(
MLOC.getCS().getCalledValue()));
for (const Value *Arg : MLOC.getCS().args())
hash = hash_combine(hash, DenseMapInfo<const Value *>::getHashValue(Arg));
return hash;
}
static bool isEqual(const MemoryLocOrCall &LHS, const MemoryLocOrCall &RHS) {

38
llvm/test/Analysis/MemorySSA/pr36883.ll Normal file
View File

@ -0,0 +1,38 @@
; RUN: opt -basicaa -memoryssa -analyze < %s 2>&1 -S | FileCheck %s
; RUN: opt -aa-pipeline=basic-aa -passes='print<memoryssa>,verify<memoryssa>' -S < %s 2>&1 | FileCheck %s
;
; We weren't properly considering the args in callsites in equality or hashing.
target triple = "armv7-dcg-linux-gnueabi"
; CHECK-LABEL: define <8 x i16> @vpx_idct32_32_neon
define <8 x i16> @vpx_idct32_32_neon(i8* %p, <8 x i16> %v) {
entry:
; CHECK: MemoryUse(liveOnEntry)
%load1 = call <8 x i16> @llvm.arm.neon.vld1.v8i16.p0i8(i8* %p, i32 2) #4 ; load CSE replacement
; CHECK: 1 = MemoryDef(liveOnEntry)
call void @llvm.arm.neon.vst1.p0i8.v8i16(i8* %p, <8 x i16> %v, i32 2) #4 ; clobber
%p_next = getelementptr inbounds i8, i8* %p, i32 16
; CHECK: MemoryUse(liveOnEntry)
%load2 = call <8 x i16> @llvm.arm.neon.vld1.v8i16.p0i8(i8* %p_next, i32 2) #4 ; non-aliasing load needed to trigger bug
; CHECK: MemoryUse(1)
%load3 = call <8 x i16> @llvm.arm.neon.vld1.v8i16.p0i8(i8* %p, i32 2) #4 ; load CSE removed
%add = add <8 x i16> %load1, %load2
%ret = add <8 x i16> %add, %load3
ret <8 x i16> %ret
}
; Function Attrs: argmemonly nounwind readonly
declare <8 x i16> @llvm.arm.neon.vld1.v8i16.p0i8(i8*, i32) #2
; Function Attrs: argmemonly nounwind
declare void @llvm.arm.neon.vst1.p0i8.v8i16(i8*, <8 x i16>, i32) #1
attributes #1 = { argmemonly nounwind }
attributes #2 = { argmemonly nounwind readonly }
attributes #3 = { nounwind readnone }
attributes #4 = { nounwind }