1 files changed, 109 insertions, 28 deletions

diff --git a/contrib/llvm-project/llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp b/contrib/llvm-project/llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp
index bca3b0538c5d..03087d8370d5 100644
--- a/contrib/llvm-project/llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp
+++ b/contrib/llvm-project/llvm/lib/Transforms/Utils/SimplifyLibCalls.cpp
@@ -75,39 +75,109 @@ static bool callHasFP128Argument(const CallInst *CI) {
   });
 }
 
-static Value *convertStrToNumber(CallInst *CI, StringRef &Str, Value *EndPtr,
-                                 int64_t Base, IRBuilderBase &B) {
+// Convert the entire string Str representing an integer in Base, up to
+// the terminating nul if present, to a constant according to the rules
+// of strtoul[l] or, when AsSigned is set, of strtol[l].  On success
+// return the result, otherwise null.
+// The function assumes the string is encoded in ASCII and carefully
+// avoids converting sequences (including "") that the corresponding
+// library call might fail and set errno for.
+static Value *convertStrToInt(CallInst *CI, StringRef &Str, Value *EndPtr,
+                              uint64_t Base, bool AsSigned, IRBuilderBase &B) {
   if (Base < 2 || Base > 36)
-    // handle special zero base
     if (Base != 0)
+      // Fail for an invalid base (required by POSIX).
       return nullptr;
 
-  char *End;
-  std::string nptr = Str.str();
-  errno = 0;
-  long long int Result = strtoll(nptr.c_str(), &End, Base);
-  if (errno)
-    return nullptr;
+  // Strip leading whitespace.
+  for (unsigned i = 0; i != Str.size(); ++i)
+    if (!isSpace((unsigned char)Str[i])) {
+      Str = Str.substr(i);
+      break;
+    }
 
-  // if we assume all possible target locales are ASCII supersets,
-  // then if strtoll successfully parses a number on the host,
-  // it will also successfully parse the same way on the target
-  if (*End != '\0')
+  if (Str.empty())
+    // Fail for empty subject sequences (POSIX allows but doesn't require
+    // strtol[l]/strtoul[l] to fail with EINVAL).
     return nullptr;
 
-  if (!isIntN(CI->getType()->getPrimitiveSizeInBits(), Result))
-    return nullptr;
+  // Strip but remember the sign.
+  bool Negate = Str[0] == '-';
+  if (Str[0] == '-' || Str[0] == '+') {
+    Str = Str.drop_front();
+    if (Str.empty())
+      // Fail for a sign with nothing after it.
+      return nullptr;
+  }
+
+  // Set Max to the absolute value of the minimum (for signed), or
+  // to the maximum (for unsigned) value representable in the type.
+  Type *RetTy = CI->getType();
+  unsigned NBits = RetTy->getPrimitiveSizeInBits();
+  uint64_t Max = AsSigned && Negate ? 1 : 0;
+  Max += AsSigned ? maxIntN(NBits) : maxUIntN(NBits);
+
+  // Autodetect Base if it's zero and consume the "0x" prefix.
+  if (Str.size() > 1) {
+    if (Str[0] == '0') {
+      if (toUpper((unsigned char)Str[1]) == 'X') {
+        if (Str.size() == 2 || (Base && Base != 16))
+          // Fail if Base doesn't allow the "0x" prefix or for the prefix
+          // alone that implementations like BSD set errno to EINVAL for.
+          return nullptr;
+
+        Str = Str.drop_front(2);
+        Base = 16;
+      }
+      else if (Base == 0)
+        Base = 8;
+    } else if (Base == 0)
+      Base = 10;
+  }
+  else if (Base == 0)
+    Base = 10;
+
+  // Convert the rest of the subject sequence, not including the sign,
+  // to its uint64_t representation (this assumes the source character
+  // set is ASCII).
+  uint64_t Result = 0;
+  for (unsigned i = 0; i != Str.size(); ++i) {
+    unsigned char DigVal = Str[i];
+    if (isDigit(DigVal))
+      DigVal = DigVal - '0';
+    else {
+      DigVal = toUpper(DigVal);
+      if (isAlpha(DigVal))
+        DigVal = DigVal - 'A' + 10;
+      else
+        return nullptr;
+    }
+
+    if (DigVal >= Base)
+      // Fail if the digit is not valid in the Base.
+      return nullptr;
+
+    // Add the digit and fail if the result is not representable in
+    // the (unsigned form of the) destination type.
+    bool VFlow;
+    Result = SaturatingMultiplyAdd(Result, Base, (uint64_t)DigVal, &VFlow);
+    if (VFlow || Result > Max)
+      return nullptr;
+  }
 
   if (EndPtr) {
     // Store the pointer to the end.
-    uint64_t ILen = End - nptr.c_str();
-    Value *Off = B.getInt64(ILen);
+    Value *Off = B.getInt64(Str.size());
     Value *StrBeg = CI->getArgOperand(0);
     Value *StrEnd = B.CreateInBoundsGEP(B.getInt8Ty(), StrBeg, Off, "endptr");
     B.CreateStore(StrEnd, EndPtr);
   }
 
-  return ConstantInt::get(CI->getType(), Result);
+  if (Negate)
+    // Unsigned negation doesn't overflow.
+    Result = -Result;
+
+  return ConstantInt::get(RetTy, Result);
 }
 
 static bool isOnlyUsedInComparisonWithZero(Value *V) {
@@ -2531,27 +2601,35 @@ Value *LibCallSimplifier::optimizeToAscii(CallInst *CI, IRBuilderBase &B) {
                      ConstantInt::get(CI->getType(), 0x7F));
 }
 
+// Fold calls to atoi, atol, and atoll.
 Value *LibCallSimplifier::optimizeAtoi(CallInst *CI, IRBuilderBase &B) {
+  CI->addParamAttr(0, Attribute::NoCapture);
+
   StringRef Str;
   if (!getConstantStringInfo(CI->getArgOperand(0), Str))
     return nullptr;
 
-  return convertStrToNumber(CI, Str, nullptr, 10, B);
+  return convertStrToInt(CI, Str, nullptr, 10, /*AsSigned=*/true, B);
 }
 
-Value *LibCallSimplifier::optimizeStrtol(CallInst *CI, IRBuilderBase &B) {
-  StringRef Str;
-  if (!getConstantStringInfo(CI->getArgOperand(0), Str))
-    return nullptr;
-
+// Fold calls to strtol, strtoll, strtoul, and strtoull.
+Value *LibCallSimplifier::optimizeStrToInt(CallInst *CI, IRBuilderBase &B,
+                                           bool AsSigned) {
   Value *EndPtr = CI->getArgOperand(1);
-  if (isa<ConstantPointerNull>(EndPtr))
+  if (isa<ConstantPointerNull>(EndPtr)) {
+    // With a null EndPtr, this function won't capture the main argument.
+    // It would be readonly too, except that it still may write to errno.
+    CI->addParamAttr(0, Attribute::NoCapture);
     EndPtr = nullptr;
-  else if (!isKnownNonZero(EndPtr, DL))
+  } else if (!isKnownNonZero(EndPtr, DL))
+    return nullptr;
+
+  StringRef Str;
+  if (!getConstantStringInfo(CI->getArgOperand(0), Str))
     return nullptr;
 
   if (ConstantInt *CInt = dyn_cast<ConstantInt>(CI->getArgOperand(2))) {
-    return convertStrToNumber(CI, Str, EndPtr, CInt->getSExtValue(), B);
+    return convertStrToInt(CI, Str, EndPtr, CInt->getSExtValue(), AsSigned, B);
   }
 
   return nullptr;
@@ -3390,7 +3468,10 @@ Value *LibCallSimplifier::optimizeCall(CallInst *CI, IRBuilderBase &Builder) {
       return optimizeAtoi(CI, Builder);
     case LibFunc_strtol:
     case LibFunc_strtoll:
-      return optimizeStrtol(CI, Builder);
+      return optimizeStrToInt(CI, Builder, /*AsSigned=*/true);
+    case LibFunc_strtoul:
+    case LibFunc_strtoull:
+      return optimizeStrToInt(CI, Builder, /*AsSigned=*/false);
     case LibFunc_printf:
       return optimizePrintF(CI, Builder);
     case LibFunc_sprintf: