Re: [Lksctp-developers] sctp_getpaddrs, sctp_getladdrs, 8.3 - 8.5

[Gaurav N. <gn...@at...>]

Thanks for the amazing response, both functions work just fine now.

gn


On Mon, 19 May 2003, Sridhar Samudrala wrote:

> On Wed, 21 May 2003, Jon Grimm wrote:
>
> > Gaurav Naik wrote:
> > >
> > > I am having trouble using the two new sockets draft calls for getting a
> > > list of local/peer addrs in a association.
> > >
> > > I am calling sctp_getpaddrs() with the following parameters,
> > > 1] socket desc of the association for int sd
> > > 2] 0 (using TCP style sockets) for sctp_assoc_t id (should be ignored)
> > > 3] ptr to a ptr of sockaddr_storage
> > >
> > > The calls always returns -1.  I decided to dig into the user library to
> > > see where the problem was.  The call was returing from line 109 in
> > > addrs.c, at which the condition was checking to see if *addrs was NULL.
> > > This is where, my confusion became apparent.  I figured since I was
> > > passing in a ptr that was to be filled by the function, why did it matter
> > > if the incoming parameter was NULL or not?   i decided that this was incorrect,
> >
> > Yes.  Its probably overly protective.
> >
> > > and commented it out, but the getsockopt() call on line 112 of addrs.c was
> > > returning -1.  I did some more digging, and determined that the
> > > getsockopt() call on line 92 was returning 2 (which is correct).  But that
> > > is where I stopped.
> > >
> > > I believe I am using this calls incorrectly.  Is anyone using these
> > > functions?  This one is a real show stopper for me right now.
> > >
> >
> > Hmmm... addrs.c::sctp_getpaddrs doesn't look right.  A couple things
> > wrong...
> >
> >
> > L48 says
> > 	len = ret * sizeof(struct sctp_getaddrs);
> >
> > but len value of malloced mem should be ret * sizeof(struct
> > sockaddr_storage)
> >
> > and the len value passed to getsockopt() should be sizeof(struct
> > sctp_getaddrs).
> >
> > Sridhar and I just chatted on this... he'll take a look to see if there
> > are any other issues.  Right now this looks fairly easy to fix just
> > reading the code.
>
> Gaurav,
>
> Please try out this patch and let me know if it works for you. It basically
> implements the fixes Jon pointed out.
>
> Thanks
> Sridhar
>
> diff -Nru a/src/lib/addrs.c b/src/lib/addrs.c
> --- a/src/lib/addrs.c	Mon May 19 14:47:36 2003
> +++ b/src/lib/addrs.c	Mon May 19 14:47:36 2003
> @@ -45,16 +45,13 @@
>  		return 0;
>  	}
>
> -	len = ret * sizeof(struct sctp_getaddrs);
> +	len = ret * sizeof(struct sockaddr_storage);
>
>  	getaddrs.assoc_id = id;
>  	getaddrs.addr_num = ret;
>  	getaddrs.addrs = (struct sockaddr_storage *)malloc(len);
>
> -	if (NULL == *addrs) {
> -		return -1;
> -	}
> -
> +	len = sizeof(getaddrs);
>  	err = getsockopt(sd, SOL_SCTP, SCTP_GET_PEER_ADDRS, (void *)&getaddrs,
>  			 &len);
>  	if (err < 0) {
> @@ -99,16 +96,13 @@
>  		return 0;
>  	}
>
> -	len = ret * sizeof(struct sctp_getaddrs);
> +	len = ret * sizeof(struct sockaddr_storage);
>
>  	getaddrs.assoc_id = id;
>  	getaddrs.addr_num = ret;
>  	getaddrs.addrs = (struct sockaddr_storage *)malloc(len);
>
> -	if (NULL == *addrs) {
> -		return -1;
> -	}
> -
> +	len = sizeof(getaddrs);
>  	err = getsockopt(sd, SOL_SCTP, SCTP_GET_LOCAL_ADDRS, (void *)&getaddrs,
>  			 &len);
>  	if (err < 0) {
>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: ObjectStore.
> If flattening out C++ or Java code to make your application fit in a
> relational database is painful, don't do it! Check out ObjectStore.
> Now part of Progress Software. http://www.objectstore.net/sourceforge
> _______________________________________________
> Lksctp-developers mailing list
> Lks...@li...
> https://lists.sourceforge.net/lists/listinfo/lksctp-developers
>

-- 
--
Gaurav Naik
Lockheed Martin - Advanced Technology Labs

email: 	gnaik(at)atl.lmco.com
snail: 	6th Floor East; 3 Executive Campus; Cherry Hill, NJ 08002
voice: 	856.792.9863
fax:	856.792.9930