Bug#707118: there really is an incorrect optimization here.

[Zack Weinberg <zackw@panix.com>]

reopen 707118
tags 707118 + upstream
forwarded 707118 http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57230
thanks

The incorrect optimization is not that the comparison was optimized 
away.  It's that the program prints 0 instead of 12.  This simplified 
testcase shows the same problem:

  int main(void)
  {
    char pstring[] = " hello world";
    pstring[0] = (char) (__builtin_strlen(pstring) - 1);
    __builtin_printf("%zd\n", __builtin_strlen(pstring));
    return 0;
  }

The value written to pstring[0] is nonzero, so the length of the string 
remains 12, but tree-ssa-strlen.c replaces the second call to 
__builtin_strlen with a constant zero, which is wrong.

I was able to reproduce the problem with GCC 4.8 as well, so I filed the 
bug upstream: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57230

zw