instcombine: Split out the __stpcpy_chk simplifications from StrCpyChkOpt

r166198 migrated the strcpy optimization to instcombine.  The strcpy
simplifier that was migrated from Transforms/Scalar/SimplifyLibCalls.cpp
was also doing some __strcpy_chk simplifications.  Those fortified
simplifications were migrated as well, but introduced a bug in the
__stpcpy_chk simplifier in the process.  This happened because the
__strcpy_chk and __stpcpy_chk simplifiers were both mapped to StrCpyChkOpt
which was updated with simplifications that worked for __strcpy_chk, but
not __stpcpy_chk.

This patch fixes the problem by adding proper test coverage and creating a
new simplifier for __stpcpy_chk (instead of sharing one with __strcpy_chk).

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@167082 91177308-0d34-0410-b5e6-96231b3b80d8

lib/Transforms/Utils/SimplifyLibCalls.cpp

@@ -185,8 +185,8 @@ struct StrCpyChkOpt : public InstFortifiedLibCallOptimization {
       return Src;
 
     // If a) we don't have any length information, or b) we know this will
-    // fit then just lower to a plain st[rp]cpy. Otherwise we'll keep our
+    // fit then just lower to a plain strcpy. Otherwise we'll keep our
-    // st[rp]cpy_chk call which may fail at runtime if the size is too long.
+    // strcpy_chk call which may fail at runtime if the size is too long.
     // TODO: It might be nice to get a maximum length out of the possible
     // string lengths for varying.
     if (isFoldable(2, 1, true)) {
@@ -210,6 +210,56 @@ struct StrCpyChkOpt : public InstFortifiedLibCallOptimization {
   }
 };
 
+struct StpCpyChkOpt : public InstFortifiedLibCallOptimization {
+  virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) {
+    this->CI = CI;
+    StringRef Name = Callee->getName();
+    FunctionType *FT = Callee->getFunctionType();
+    LLVMContext &Context = CI->getParent()->getContext();
+
+    // Check if this has the right signature.
+    if (FT->getNumParams() != 3 ||
+        FT->getReturnType() != FT->getParamType(0) ||
+        FT->getParamType(0) != FT->getParamType(1) ||
+        FT->getParamType(0) != Type::getInt8PtrTy(Context) ||
+        FT->getParamType(2) != TD->getIntPtrType(FT->getParamType(0)))
+      return 0;
+
+    Value *Dst = CI->getArgOperand(0), *Src = CI->getArgOperand(1);
+    if (Dst == Src) {  // stpcpy(x,x)  -> x+strlen(x)
+      Value *StrLen = EmitStrLen(Src, B, TD, TLI);
+      return StrLen ? B.CreateInBoundsGEP(Dst, StrLen) : 0;
+    }
+
+    // If a) we don't have any length information, or b) we know this will
+    // fit then just lower to a plain stpcpy. Otherwise we'll keep our
+    // stpcpy_chk call which may fail at runtime if the size is too long.
+    // TODO: It might be nice to get a maximum length out of the possible
+    // string lengths for varying.
+    if (isFoldable(2, 1, true)) {
+      Value *Ret = EmitStrCpy(Dst, Src, B, TD, TLI, Name.substr(2, 6));
+      return Ret;
+    } else {
+      // Maybe we can stil fold __stpcpy_chk to __memcpy_chk.
+      uint64_t Len = GetStringLength(Src);
+      if (Len == 0) return 0;
+
+      // This optimization require DataLayout.
+      if (!TD) return 0;
+
+      Type *PT = FT->getParamType(0);
+      Value *LenV = ConstantInt::get(TD->getIntPtrType(PT), Len);
+      Value *DstEnd = B.CreateGEP(Dst,
+                                  ConstantInt::get(TD->getIntPtrType(PT),
+                                                   Len - 1));
+      if (!EmitMemCpyChk(Dst, Src, LenV, CI->getArgOperand(2), B, TD, TLI))
+        return 0;
+      return DstEnd;
+    }
+    return 0;
+  }
+};
+
 struct StrNCpyChkOpt : public InstFortifiedLibCallOptimization {
   virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B) {
     this->CI = CI;
@@ -556,6 +606,7 @@ class LibCallSimplifierImpl {
   MemMoveChkOpt MemMoveChk;
   MemSetChkOpt MemSetChk;
   StrCpyChkOpt StrCpyChk;
+  StpCpyChkOpt StpCpyChk;
   StrNCpyChkOpt StrNCpyChk;
 
   // String and memory library call optimizations.
@@ -583,7 +634,7 @@ void LibCallSimplifierImpl::initOptimizations() {
   Optimizations["__memmove_chk"] = &MemMoveChk;
   Optimizations["__memset_chk"] = &MemSetChk;
   Optimizations["__strcpy_chk"] = &StrCpyChk;
-  Optimizations["__stpcpy_chk"] = &StrCpyChk;
+  Optimizations["__stpcpy_chk"] = &StpCpyChk;
   Optimizations["__strncpy_chk"] = &StrNCpyChk;
   Optimizations["__stpncpy_chk"] = &StrNCpyChk;
 

test/Transforms/InstCombine/stpcpy_chk-1.ll

@@ -0,0 +1,96 @@
+; Test lib call simplification of __stpcpy_chk calls with various values
+; for src, dst, and slen.
+;
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+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"
+
+@a = common global [60 x i8] zeroinitializer, align 1
+@b = common global [60 x i8] zeroinitializer, align 1
+@.str = private constant [12 x i8] c"abcdefghijk\00"
+
+; Check cases where slen >= strlen (src).
+
+define void @test_simplify1() {
+; CHECK: @test_simplify1
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [12 x i8]* @.str, i32 0, i32 0
+
+; CHECK-NEXT: call i8* @stpcpy
+  call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 60)
+  ret void
+}
+
+define void @test_simplify2() {
+; CHECK: @test_simplify2
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [12 x i8]* @.str, i32 0, i32 0
+
+; CHECK-NEXT: call i8* @stpcpy
+  call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 12)
+  ret void
+}
+
+define void @test_simplify3() {
+; CHECK: @test_simplify3
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [12 x i8]* @.str, i32 0, i32 0
+
+; CHECK-NEXT: call i8* @stpcpy
+  call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 -1)
+  ret void
+}
+
+; Check cases where there are no string constants.
+
+define void @test_simplify4() {
+; CHECK: @test_simplify4
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [60 x i8]* @b, i32 0, i32 0
+
+; CHECK-NEXT: call i8* @stpcpy
+  call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 -1)
+  ret void
+}
+
+; Check case where the string length is not constant.
+
+define i8* @test_simplify5() {
+; CHECK: @test_simplify5
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [12 x i8]* @.str, i32 0, i32 0
+
+; CHECK: @__memcpy_chk
+  %len = call i32 @llvm.objectsize.i32(i8* %dst, i1 false)
+  %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 %len)
+; CHECK: ret i8* getelementptr inbounds ([60 x i8]* @a, i32 0, i32 11)
+  ret i8* %ret
+}
+
+; Check case where the source and destination are the same.
+
+define i8* @test_simplify6() {
+; CHECK: @test_simplify6
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+
+; CHECK: [[LEN:%[a-z]+]] = call i32 @strlen
+; CHECK-NEXT: getelementptr inbounds [60 x i8]* @a, i32 0, i32 [[LEN]]
+  %len = call i32 @llvm.objectsize.i32(i8* %dst, i1 false)
+  %ret = call i8* @__stpcpy_chk(i8* %dst, i8* %dst, i32 %len)
+  ret i8* %ret
+}
+
+; Check case where slen < strlen (src).
+
+define void @test_no_simplify1() {
+; CHECK: @test_no_simplify1
+  %dst = getelementptr inbounds [60 x i8]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [60 x i8]* @b, i32 0, i32 0
+
+; CHECK-NEXT: call i8* @__stpcpy_chk
+  call i8* @__stpcpy_chk(i8* %dst, i8* %src, i32 8)
+  ret void
+}
+
+declare i8* @__stpcpy_chk(i8*, i8*, i32) nounwind
+declare i32 @llvm.objectsize.i32(i8*, i1) nounwind readonly

test/Transforms/InstCombine/stpcpy_chk-2.ll

@@ -0,0 +1,21 @@
+; Test that lib call simplification doesn't simplify __stpcpy_chk calls
+; with the wrong prototype.
+;
+; RUN: opt < %s -instcombine -S | FileCheck %s
+
+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"
+
+@a = common global [60 x i16] zeroinitializer, align 1
+@.str = private constant [8 x i8] c"abcdefg\00"
+
+define void @test_no_simplify() {
+; CHECK: @test_no_simplify
+  %dst = getelementptr inbounds [60 x i16]* @a, i32 0, i32 0
+  %src = getelementptr inbounds [8 x i8]* @.str, i32 0, i32 0
+
+; CHECK-NEXT: call i16* @__strcpy_chk
+  call i16* @__strcpy_chk(i16* %dst, i8* %src, i32 8)
+  ret void
+}
+
+declare i16* @__strcpy_chk(i16*, i8*, i32)