On Wed, 8 Sep 1999, Craig Small wrote:
> Tomi Manninen said:
> > It seems the interface to getname has changed a bit. From that code I
> > deduce that &len is now supposed be _filled_in_ by sock->ops->getname().
> > At least inet_getname() does that. And it's now move_addr_to_user() that
> > handles truncating the address struct if needed.
> >
> > Now it should only be a matter of a small adjustment to ax25_getname(),
> > nr_getname() and rose_getname() and we are back to business...
>
> Did this change from 2.2.11 to 2.2.12? What the hell were they thinking
> changing this in a stable kernel? Isn't this the whole point of 2.3.x
> kernels.
No, I think it was changed quite some time ago, in the 2.1.x series. The
reason we haven't noticed it is that as len is uninitialized, everything
works if it just happens to be greater than the size of the struct and it
seems that happens surprisingly often...
--
--- Tomi Manninen / [EMAIL PROTECTED] / OH2BNS @ OH2RBI.FIN.EU ---
