error can shift more than 32 bits.

runtime/third_party/double-conversion/src/strtod.cc

 // Copyright 2010 the V8 project authors. All rights reserved.
 // Redistribution and use in source and binary forms, with or without
 // modification, are permitted provided that the following conditions are
 // met:
 //
 //     * Redistributions of source code must retain the above copyright
 //       notice, this list of conditions and the following disclaimer.
 //     * Redistributions in binary form must reproduce the above
 //       copyright notice, this list of conditions and the following
 //       disclaimer in the documentation and/or other materials provided
@@ skipping 268 matching lines @@
   ReadDiyFp(buffer, &input, &remaining_decimals);
   // Since we may have dropped some digits the input is not accurate.
   // If remaining_decimals is different than 0 than the error is at most
   // .5 ulp (unit in the last place).
   // We don't want to deal with fractions and therefore keep a common
   // denominator.
   const int kDenominatorLog = 3;
   const int kDenominator = 1 << kDenominatorLog;
   // Move the remaining decimals into the exponent.
   exponent += remaining_decimals;
-  int error = (remaining_decimals == 0 ? 0 : kDenominator / 2);
+  uint64_t error = (remaining_decimals == 0 ? 0 : kDenominator / 2);
 
   int old_e = input.e();
   input.Normalize();
   error <<= old_e - input.e();
 
   ASSERT(exponent <= PowersOfTenCache::kMaxDecimalExponent);
   if (exponent < PowersOfTenCache::kMinDecimalExponent) {
     *result = 0.0;
     return true;
   }
@@ skipping 246 matching lines @@
     return next;
   } else if ((Single(guess).Significand() & 1) == 0) {
     // Round towards even.
     return guess;
   } else {
     return next;
   }
 }
 
 }  // namespace double_conversion