+ %64 = call noalias i8* @malloc(i64 %62) nounwind
+...
+ %67 = call i64 @llvm.objectsize.i64(i8* %64, i1 false) nounwind
+ %68 = call i8* @__memset_chk(i8* %64, i32 0, i64 %62, i64 %67) nounwind
+
+llvm.objectsize.i64 should be taught about malloc/calloc, allowing it to
+fold to %62. This is a security win (overflows of malloc will get caught)
+and also a performance win by exposing more memsets to the optimizer.
+
+This occurs several times in viterbi.
+
+Note that this would change the semantics of @llvm.objectsize which by its
+current definition always folds to a constant. We also should make sure that
+we remove checking in code like
+
+ char *p = malloc(strlen(s)+1);
+ __strcpy_chk(p, s, __builtin_objectsize(p, 0));
+
+//===---------------------------------------------------------------------===//
+
+This code (from Benchmarks/Dhrystone/dry.c):
+
+define i32 @Func1(i32, i32) nounwind readnone optsize ssp {
+entry:
+ %sext = shl i32 %0, 24
+ %conv = ashr i32 %sext, 24
+ %sext6 = shl i32 %1, 24
+ %conv4 = ashr i32 %sext6, 24
+ %cmp = icmp eq i32 %conv, %conv4
+ %. = select i1 %cmp, i32 10000, i32 0
+ ret i32 %.
+}
+
+Should be simplified into something like:
+
+define i32 @Func1(i32, i32) nounwind readnone optsize ssp {
+entry:
+ %sext = shl i32 %0, 24
+ %conv = and i32 %sext, 0xFF000000
+ %sext6 = shl i32 %1, 24
+ %conv4 = and i32 %sext6, 0xFF000000
+ %cmp = icmp eq i32 %conv, %conv4
+ %. = select i1 %cmp, i32 10000, i32 0
+ ret i32 %.
+}
+
+and then to:
+
+define i32 @Func1(i32, i32) nounwind readnone optsize ssp {
+entry:
+ %conv = and i32 %0, 0xFF
+ %conv4 = and i32 %1, 0xFF
+ %cmp = icmp eq i32 %conv, %conv4
+ %. = select i1 %cmp, i32 10000, i32 0
+ ret i32 %.
+}
+//===---------------------------------------------------------------------===//
+
+clang -O3 currently compiles this code
+
+int g(unsigned int a) {
+ unsigned int c[100];
+ c[10] = a;
+ c[11] = a;
+ unsigned int b = c[10] + c[11];
+ if(b > a*2) a = 4;
+ else a = 8;
+ return a + 7;
+}
+
+into
+
+define i32 @g(i32 a) nounwind readnone {
+ %add = shl i32 %a, 1
+ %mul = shl i32 %a, 1
+ %cmp = icmp ugt i32 %add, %mul
+ %a.addr.0 = select i1 %cmp, i32 11, i32 15
+ ret i32 %a.addr.0
+}
+
+The icmp should fold to false. This CSE opportunity is only available
+after GVN and InstCombine have run.
+
+//===---------------------------------------------------------------------===//
+
+memcpyopt should turn this:
+
+define i8* @test10(i32 %x) {
+ %alloc = call noalias i8* @malloc(i32 %x) nounwind
+ call void @llvm.memset.p0i8.i32(i8* %alloc, i8 0, i32 %x, i32 1, i1 false)
+ ret i8* %alloc
+}
+
+into a call to calloc. We should make sure that we analyze calloc as
+aggressively as malloc though.
+
+//===---------------------------------------------------------------------===//
+
+clang -O3 doesn't optimize this:
+
+void f1(int* begin, int* end) {
+ std::fill(begin, end, 0);
+}
+
+into a memset. This is PR8942.
+
+//===---------------------------------------------------------------------===//
+
+clang -O3 -fno-exceptions currently compiles this code:
+
+void f(int N) {
+ std::vector<int> v(N);
+
+ extern void sink(void*); sink(&v);
+}
+
+into
+
+define void @_Z1fi(i32 %N) nounwind {
+entry:
+ %v2 = alloca [3 x i32*], align 8
+ %v2.sub = getelementptr inbounds [3 x i32*]* %v2, i64 0, i64 0
+ %tmpcast = bitcast [3 x i32*]* %v2 to %"class.std::vector"*
+ %conv = sext i32 %N to i64
+ store i32* null, i32** %v2.sub, align 8, !tbaa !0
+ %tmp3.i.i.i.i.i = getelementptr inbounds [3 x i32*]* %v2, i64 0, i64 1
+ store i32* null, i32** %tmp3.i.i.i.i.i, align 8, !tbaa !0
+ %tmp4.i.i.i.i.i = getelementptr inbounds [3 x i32*]* %v2, i64 0, i64 2
+ store i32* null, i32** %tmp4.i.i.i.i.i, align 8, !tbaa !0
+ %cmp.i.i.i.i = icmp eq i32 %N, 0
+ br i1 %cmp.i.i.i.i, label %_ZNSt12_Vector_baseIiSaIiEEC2EmRKS0_.exit.thread.i.i, label %cond.true.i.i.i.i
+
+_ZNSt12_Vector_baseIiSaIiEEC2EmRKS0_.exit.thread.i.i: ; preds = %entry
+ store i32* null, i32** %v2.sub, align 8, !tbaa !0
+ store i32* null, i32** %tmp3.i.i.i.i.i, align 8, !tbaa !0
+ %add.ptr.i5.i.i = getelementptr inbounds i32* null, i64 %conv
+ store i32* %add.ptr.i5.i.i, i32** %tmp4.i.i.i.i.i, align 8, !tbaa !0
+ br label %_ZNSt6vectorIiSaIiEEC1EmRKiRKS0_.exit
+
+cond.true.i.i.i.i: ; preds = %entry
+ %cmp.i.i.i.i.i = icmp slt i32 %N, 0
+ br i1 %cmp.i.i.i.i.i, label %if.then.i.i.i.i.i, label %_ZNSt12_Vector_baseIiSaIiEEC2EmRKS0_.exit.i.i
+
+if.then.i.i.i.i.i: ; preds = %cond.true.i.i.i.i
+ call void @_ZSt17__throw_bad_allocv() noreturn nounwind
+ unreachable
+
+_ZNSt12_Vector_baseIiSaIiEEC2EmRKS0_.exit.i.i: ; preds = %cond.true.i.i.i.i
+ %mul.i.i.i.i.i = shl i64 %conv, 2
+ %call3.i.i.i.i.i = call noalias i8* @_Znwm(i64 %mul.i.i.i.i.i) nounwind
+ %0 = bitcast i8* %call3.i.i.i.i.i to i32*
+ store i32* %0, i32** %v2.sub, align 8, !tbaa !0
+ store i32* %0, i32** %tmp3.i.i.i.i.i, align 8, !tbaa !0
+ %add.ptr.i.i.i = getelementptr inbounds i32* %0, i64 %conv
+ store i32* %add.ptr.i.i.i, i32** %tmp4.i.i.i.i.i, align 8, !tbaa !0
+ call void @llvm.memset.p0i8.i64(i8* %call3.i.i.i.i.i, i8 0, i64 %mul.i.i.i.i.i, i32 4, i1 false)
+ br label %_ZNSt6vectorIiSaIiEEC1EmRKiRKS0_.exit
+
+This is just the handling the construction of the vector. Most surprising here
+is the fact that all three null stores in %entry are dead (because we do no
+cross-block DSE).
+
+Also surprising is that %conv isn't simplified to 0 in %....exit.thread.i.i.
+This is a because the client of LazyValueInfo doesn't simplify all instruction
+operands, just selected ones.
+
+//===---------------------------------------------------------------------===//
+
+clang -O3 -fno-exceptions currently compiles this code:
+
+void f(char* a, int n) {
+ __builtin_memset(a, 0, n);
+ for (int i = 0; i < n; ++i)
+ a[i] = 0;
+}
+
+into:
+
+define void @_Z1fPci(i8* nocapture %a, i32 %n) nounwind {
+entry:
+ %conv = sext i32 %n to i64
+ tail call void @llvm.memset.p0i8.i64(i8* %a, i8 0, i64 %conv, i32 1, i1 false)
+ %cmp8 = icmp sgt i32 %n, 0
+ br i1 %cmp8, label %for.body.lr.ph, label %for.end
+
+for.body.lr.ph: ; preds = %entry
+ %tmp10 = add i32 %n, -1
+ %tmp11 = zext i32 %tmp10 to i64
+ %tmp12 = add i64 %tmp11, 1
+ call void @llvm.memset.p0i8.i64(i8* %a, i8 0, i64 %tmp12, i32 1, i1 false)
+ ret void
+
+for.end: ; preds = %entry
+ ret void
+}
+
+This shouldn't need the ((zext (%n - 1)) + 1) game, and it should ideally fold
+the two memset's together.
+
+The issue with the addition only occurs in 64-bit mode, and appears to be at
+least partially caused by Scalar Evolution not keeping its cache updated: it
+returns the "wrong" result immediately after indvars runs, but figures out the
+expected result if it is run from scratch on IR resulting from running indvars.
+
+//===---------------------------------------------------------------------===//
+
+clang -O3 -fno-exceptions currently compiles this code:
+
+struct S {
+ unsigned short m1, m2;
+ unsigned char m3, m4;