Fix PR2341 - when the length is 4 use an i32 not

an i16!  Cleaned up trailing whitespace while there.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@51240 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Duncan Sands 2008-05-19 09:27:24 +00:00
parent e46924bdf7
commit ec00fcb33d
2 changed files with 24 additions and 8 deletions

View File

@ -731,17 +731,17 @@ struct VISIBILITY_HIDDEN MemCmpOpt : public LibCallOptimization {
!isa<PointerType>(FT->getParamType(1)) ||
FT->getReturnType() != Type::Int32Ty)
return 0;
Value *LHS = CI->getOperand(1), *RHS = CI->getOperand(2);
if (LHS == RHS) // memcmp(s,s,x) -> 0
return Constant::getNullValue(CI->getType());
// Make sure we have a constant length.
ConstantInt *LenC = dyn_cast<ConstantInt>(CI->getOperand(3));
if (!LenC) return 0;
uint64_t Len = LenC->getZExtValue();
if (Len == 0) // memcmp(s1,s2,0) -> 0
return Constant::getNullValue(CI->getType());
@ -750,18 +750,20 @@ struct VISIBILITY_HIDDEN MemCmpOpt : public LibCallOptimization {
Value *RHSV = B.CreateLoad(CastToCStr(RHS, B), "rhsv");
return B.CreateZExt(B.CreateSub(LHSV, RHSV, "chardiff"), CI->getType());
}
// memcmp(S1,S2,2) != 0 -> (*(short*)LHS ^ *(short*)RHS) != 0
// memcmp(S1,S2,4) != 0 -> (*(int*)LHS ^ *(int*)RHS) != 0
if ((Len == 2 || Len == 4) && IsOnlyUsedInZeroEqualityComparison(CI)) {
LHS = B.CreateBitCast(LHS, PointerType::getUnqual(Type::Int16Ty), "tmp");
RHS = B.CreateBitCast(RHS, LHS->getType(), "tmp");
const Type *PTy = PointerType::getUnqual(Len == 2 ?
Type::Int16Ty : Type::Int32Ty);
LHS = B.CreateBitCast(LHS, PTy, "tmp");
RHS = B.CreateBitCast(RHS, PTy, "tmp");
LoadInst *LHSV = B.CreateLoad(LHS, "lhsv");
LoadInst *RHSV = B.CreateLoad(RHS, "rhsv");
LHSV->setAlignment(1); RHSV->setAlignment(1); // Unaligned loads.
return B.CreateZExt(B.CreateXor(LHSV, RHSV, "shortdiff"), CI->getType());
}
return 0;
}
};

View File

@ -0,0 +1,14 @@
; RUN: llvm-as < %s | opt -simplify-libcalls | llvm-dis | grep i32
; PR2341
@_2E_str = external constant [5 x i8] ; <[5 x i8]*> [#uses=1]
declare i32 @memcmp(i8*, i8*, i32) nounwind readonly
define i1 @f(i8** %start_addr) {
entry:
%tmp4 = load i8** %start_addr, align 4 ; <i8*> [#uses=1]
%tmp5 = call i32 @memcmp( i8* %tmp4, i8* getelementptr ([5 x i8]* @_2E_str, i32 0, i32 0), i32 4 ) nounwind readonly ; <i32> [#uses=1]
%tmp6 = icmp eq i32 %tmp5, 0 ; <i1> [#uses=1]
ret i1 %tmp6
}