On Wed, 2011-08-24 at 14:09 +0200, Michael Stapelberg wrote: > Hi Dan, > > Excerpts from Dan Williams's message of 2011-08-20 00:47:59 +0200: > > Thanks! Though we do risk buffer overflows because the buffer being > > passed into the function is only INET6_ADDRSTRLEN long, so we need to be > > more careful here when appending the interface name. I've reworked that > > a bit, can you test the attached patch and make sure it does the same > > thing yours does? It's compile-tested but not run tested. When I > > commit you'll still get credit, of course. Thanks for looking into > > this. > Your version of the patch has a little flaw: It only handles the case where > inet_ntop() already returns an address with %interface in it (and replaces the > % with an @ sign). However, on Linux, inet_ntop() does not return a scope > % identifier at all. In this case, we append @interface. > > I reworked the code and added a better comment, so hopefully the two cases are > clear now.
Pushed to 0.9 and 0.8 with a small fix to avoid the extra allocation when the interface is not returned. If you can re-verify my g_strdup_printf -> strncat conversion that'd be great. Thanks! Dan _______________________________________________ networkmanager-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/networkmanager-list
