Re: [patch] fix divide by zero in random/taviso()

Dave Jones <[email protected]>
Newsgroups org.kernel.vger.trinity
Message-ID <[email protected]>
On Fri, Jan 17, 2014 at 03:41:58PM -0500, Vince Weaver wrote:

 > I'm not sure if this is the cleanest fix for the issue, but it seems to 
 > work for me.
 
Good enough.

 > In random.c the taviso() routine uses "rand() % rand()" which
 > can cause a floating point exception if rand() returns 0.
 > Add some tests to make sure that doesn't happen.
 > 
 > Signed-off-by: Vince Weaver <[email protected]>
 > 
 > diff --git a/random.c b/random.c
 > index 7def9ff..a71fcea 100644
 > --- a/random.c
 > +++ b/random.c
 > @@ -42,6 +42,7 @@ static unsigned long randbits(int limit)
 >  static unsigned long taviso(void)
 >  {
 >  	unsigned long r = 0;
 > +	unsigned long temp;
 >  
 >  	switch (rand() % 4) {
 >  	case 0:	r = rand() & rand();
 > @@ -51,10 +52,14 @@ static unsigned long taviso(void)
 >  #endif
 >  		break;
 >  
 > -	case 1:	r = rand() % rand();
 > +	case 1:	temp = rand();
 > +		r = rand();
 > +		if (!temp) r %= temp;
 >  #if __WORDSIZE == 64
 >  		r <<= 32;
 > -		r |= rand() % rand();
 > +
 > +		temp = rand();
 > +		if (!temp) r |= rand() % temp;
 >  #endif
 >  		break;

Good catch.  Despite the function name, this is something I introduced,
not something that was in Tavis' original code in iknowthis.

Mea culpa.

	Dave
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.