[alsa-devel] [PATCH 1/3 v3] alsa-lib: fixed coverity reported issues under "FORWARD_NULL" checker.

[Sudarshan Bisht]

On Mon, 2011-04-04 at 10:18 +0100, ext Mark Brown wrote:
> On Mon, Apr 04, 2011 at 11:53:57AM +0300, sudarshan.bisht at nokia.com wrote:
> 
> > This patch has fix for situations where variable can be NULL 
> > but not been checked beforehand.
> 
> This description doesn't seem to match the patch...
> 
> > -	if (h == NULL)
> > +	if (h)
> >  		snd_dlclose(h);
> 
> This is a coding style change, the two conditions are equivalent.
> 
> >  			SNDERR("Invalid type for func %s definition", str);
> > +			err = -EINVAL;
> >  			goto _err;
> 
> This appears to be fixing an uninitialised return value error.
> 
> >  			return err;
> >  		if (periods == 1)
> >  			return -EINVAL;
> > -		if (*period_time == 0) {
> > +		if (period_time) {
> >  			err = INTERNAL(snd_pcm_hw_params_get_period_time)(hw_params, period_time, NULL);
> >  			if (err < 0)
> 
> This check alters the semantics - rather than checking if *period_time is
> non-zero we now check to see if period_time is non-NULL, paying no
> attention to the value if the pointer is set.
 If  "period_time == NULL" ( in preceding if condition) is true then "if
(*period_time == 0)" has no meaning ( causes segmentation fault ) and 
if  "*period_time == 0" ( in preceding if condition) is true then
obviously period_time will be a valid pointer and satisfy the condition.
Thats what original source
codes demands.      

> 
> There was at least one relevant fix in there but lots of the changes are
> as above.
> _______________________________________________
> Alsa-devel mailing list
> Alsa-devel at alsa-project.org
> http://mailman.alsa-project.org/mailman/listinfo/alsa-devel