BUG: Compiler Generates Incorrect Code with Unsigned Integer Operand Greater Than 0x80000000 (263715)


The information in this article applies to:

  • Microsoft Visual C++, 32-bit Enterprise Edition 6.0
  • Microsoft Visual C++, 32-bit Professional Edition 6.0
  • Microsoft Visual C++, 32-bit Learning Edition 6.0
This article was previously published under Q263715

SYMPTOMS

The compiler may generate incorrect code for floating point binary expressions (multiply, add, divide and sub) where the second argument is an unsigned integer in the range 0x80000000 to 0xFFFFFFFF.

RESOLUTION

There are two ways to resolve this problem:
  • Assign the argument to a float before you use it in the expression.
  • Cast the argument to a double instead of a float.

STATUS

Microsoft has confirmed that this is a bug in the Microsoft products that are listed at the beginning of this article.

MORE INFORMATION

The following sample code demonstrates the bug, as well as the ways to work around it: 
//test.cpp
// compiler option needed: /GX
//test.cpp
#include <iostream>
unsigned a=1,b=-2;
int main()
{
//PROBLEM: May produce an incorrect result.
std::cout << (float)a /(float)b <<std::endl;

//WORKAROUND 1: Create a temporary float.
float b1=(float)b;
std::cout << (float)a /(float)b1 <<std::endl; 

//WORKAROUND 2: Cast to a double.
std::cout << (float)a /(double)b <<std::endl;

return 0;
}
 
/* Program Output: 
-0.5
2.32831e-010
2.32831e-010 
*/
Modification Type:MinorLast Reviewed:7/5/2005
Keywords:kbBug kbCodeGen kbCompiler KB263715
©2004 Microsoft Corporation. All rights reserved.