[PATCH 5/8] netlink: add a basic socket diag parser of AF_INET messages

[Dmitry V. Levin]

On Wed, Jun 14, 2017 at 11:26:41PM +0800, JingPiao Chen wrote:
> On Tue, Jun 13, 2017 at 11:31:50PM +0300, Dmitry V. Levin wrote:
> > On Tue, Jun 13, 2017 at 10:13:22PM +0800, JingPiao Chen wrote:
> [...]
> > >  
> > > @@ -210,9 +214,127 @@ decode_packet_diag_msg(struct tcb *const tcp,
> > >  	return true;
> > >  }
> > >  
> > > +static void
> > > +print_inet_diag_sockid(const struct inet_diag_sockid *id, const uint8_t family)
> > > +{
> > > +	tprintf("{idiag_sport=htons(%u), idiag_dport=htons(%u)",
> > > +		ntohs(id->idiag_sport), ntohs(id->idiag_dport));
> > > +	int text_size = family == AF_INET ?
> > > +		INET_ADDRSTRLEN : INET6_ADDRSTRLEN;
> > > +	char buf[text_size];
> > > +
> > > +	inet_ntop(family, id->idiag_src, buf, text_size);
> > > +	tprintf(", inet_pton(%s, \"%s\", idiag_src)",
> > > +		family == AF_INET ? "AF_INET" : "AF_INET6", buf);
> > > +
> > > +	inet_ntop(family, id->idiag_dst, buf, text_size);
> > > +	tprintf(", inet_pton(%s, \"%s\", idiag_dst)",
> > > +		family == AF_INET ? "AF_INET" : "AF_INET6", buf);
> > 
> > As "family" is untrusted input, we cannot assume it's either AF_INET or
> > AF_INET6.  In fact, inet_ntop currently supports AF_INET and AF_INET6
> > only, so I wouldn't bother invoking inet_ntop with any other address
> > family argument.
> 
> Now family only AF_INET or AF_INET6. All the caller
> (decode_inet_diag_req_compat, decode_inet_diag_req_v2,
> decode_inet_diag_msg, decode_smc_diag_req, decode_smc_diag_msg)
> will pass AF_INET or AF_INET6. I update it, but can not check
> the case family neither AF_INET nor AF_INET6.

Do you mean that "family" is guaranteed to be either AF_INET or AF_INET6
by the calling convention?


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20170614/a0846a0a/attachment.bin>