> On Thu, Oct 17, 2002 at 03:07:07AM +0900, JINMEI Tatuya / ?$B?@L@C#:H?(B
wrote:
> ...
> > (I re-read the thread) perhaps the example in my previous message
> > wasn't good (and it was at least incorrect). According to the
> > discussion on the thread, we'll probably keep m_tag_cookie being 0 and
> > use m_tag_id in (e.g.) ip6_output.c. So, we'll be happy if this
> > convention is kept (and will be kept) at least under netinet6.
>
> unfortunately this will not prevent code from having to check that
> the m_tag_cookie actually corresponds to the value you want, to
> make sure that your code does not misinterpret as own tags
> generated/destined to other clients.
>
> Am I correct, Sam ?
>
Correct. If you explicitly look inside the m_tag structure instead of using
one of the m_tag_* routines to search then you will need to validate
m_tag_cookie to avoid interpreting tags created by other modules. The
comments I wrote in mbuf.h for this stuff describe this and say that when
writing code that is to be compatible with other systems one should always
use MTAG_ABI_COMPAT and the openbsd-compatible m_tag_get and m_tag_find
routines.
m_tag_cookie was mainly added so that netgraph could piggyback on top of the
facility and remove it's private code. At some point it may be worthwhile
to talk with the openbsd+netbsd folks about adopting this new m_tag
facility.
Sam
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-net" in the body of the message
