gpi-hash.c segfaults_(re)

Thu, 31 Jul 1997 15:08:38 +0300 (EET DST)

On Thu, 31 Jul 1997, Peter Gerwinski wrote:

> According to Bill Currie:
> > 
> > I'VE FIXED IT!!  The problem was caused by a negative index into the 
> > hash table.
> 
> Congratulations and thank you!
> 
> > To fix it change the line (gci-hash.c:122):
> >
> > (((((char *)(NODE))-(char *)0) >> 2) % MAX_HASH_TABLE)
> > 
> > to:
> > 
> > (((((unsigned long)(NODE))-(unsigned long)0) >> 2) % MAX_HASH_TABLE)
> 
> Since pointers and `long' integers don't have the same size
> on some sytems (I know it for the Alpha), I think this should
> be changed to something like `(unsigned size_t)'.

   Peter,

	In the alpha:

	sizeof(long) == sizeof(void *)
	sizeof(int)  != sizeof(void *)

	Quickly thinking, in all Unix machines it might be
	that sizeof(long) == sizeof (void *). But this just
	might not always be the case, so it can not be trusted.

	In ANSI C the result of (ptr - ptr) is of type ptrdiff_t,
	which in the alpha is (signed long).

	Using cast (unsigned size_t) might not be a good idea, at least
	I do not know if Ansi C quarantees that size_t always
	is as large as ptrdiff_t.

	I suggest that this problem is fixed by casting
	the result of the pointer subtraction to
	(unsigned ptrdiff_t) which is always large enough,
	something like:

	(((unsigned ptrdiff_t)(((char *)(NODE))-(char *)0) >> 2) 
		% MAX_HASH_TABLE)

					Juki
					jtv@hut.fi

Jukka Virtanen (jtv@cc.hut.fi)