<<< Date Index >>>     <<< Thread Index >>>

Re: gcc 4.1 bug miscompiles pointer range checks, may place you at risk



* Felix von Leitner:

> static inline int range_ptrinbuf(const void* buf,unsigned long len,const 
> void* ptr) {
>   register const char* c=(const char*)buf;      /* no pointer arithmetic on 
> void* */
>   return (c && c+len>c && (const char*)ptr-c<len);
> }

It seems that the problem is that

  c + len > c

is equivalent to

  len != 0.

Either c + len is within the same object c points to, and it's value
is larger than c (provided that len is not zero), or c + len is
undefined (because it's not the same object).  In the latter case, the
outcome is not specified by the C standard (or the GCC documentation),
so it's permissible to choose len != 0 as the value, too.

I wouldn't rule out a compiler bug in this area, but the test case is
invalid.