[PATCH 1/2] Decode setsockopt() multicast arguments

[Dmitry V. Levin]

On Sat, Feb 07, 2015 at 06:05:56PM +0100, Ben Noordhuis wrote:
> On Sat, Feb 7, 2015 at 12:55 AM, Dmitry V. Levin <ldv at altlinux.org> wrote:
> > On Thu, Feb 05, 2015 at 07:28:45PM +0100, Ben Noordhuis wrote:
> >> +#ifdef IP_ADD_MEMBERSHIP
> >> +static void
> >> +print_mreq(struct tcb *tcp, long addr, int len)
> >> +{
> >> +     struct ip_mreq mreq;
> >> +     if (len == sizeof(mreq) && umove(tcp, addr, &mreq) == 0) {
> >> +             tprints("{imr_multiaddr=inet_addr(");
> >> +             print_quoted_string(inet_ntoa(mreq.imr_multiaddr),
> >> +                                 16, QUOTE_0_TERMINATED);
> >> +             tprints("), imr_interface=inet_addr(");
> >> +             print_quoted_string(inet_ntoa(mreq.imr_interface),
> >> +                                 16, QUOTE_0_TERMINATED);
> >> +             tprints(")}");
> >> +     }
> >> +     else {
> >> +             printstr(tcp, addr, len);
> >> +     }
> >> +}
> >
> > Is there any use to print the address with printstr if length is not
> > sizeof(ip_mreq) or umove has failed?  Other sockopt parsers in such
> > situations just print the address in hex.
> 
> Printing the raw data helps troubleshooting more than printing a
> memory address does, IMO.  printstr() is also what print_getsockopt()
> and print_setsockopt() seem to fall back to.  Let me know what you
> think is preferable.

Agreed, lets fall back to printstr() in case of length mismatch.

> I agree that there is no point calling printstr() when umove() fails,
> I'll update that.

-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20150207/d1eb6a78/attachment.bin>