DragonFly BSD
DragonFly submit List (threaded) for 2004-09
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: PF test for inclusion


From: Max Laier <max@xxxxxxxxxxxxxx>
Date: Thu, 16 Sep 2004 02:42:43 +0200

On Thursday 16 September 2004 02:16, Simon 'corecode' Schubert wrote:
> general opinion (sometimes after thinking a bit longer) about this was
> to get rid of the needed m_tag allocation each time. pkthdr can be
> easily extended as m_buf is much larger anyways and this space is not
> used most of the time.

And how many packets really use tags (in the non-ALTQ case)? ... anyway

> > Anyway, nice to see this happening.
>
> yea, finally :)
>
> >> Please test it and provide me with any feedback.
> >
> > Take a look at:
> > http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/contrib/pf/net/
> > pf.c.diff?r1=1.18&r2=1.19&f=h
>
> this doesn't apply to us as we're using freebsd native byte order for
> the headers, not the openbsd one. this is why we can use ip->ip_len
> directly. arguable if using len is a worthwile speedup :)

ip->ip_len is not the right length for in_pseudo() in this piece of code. In 
tcp_input() (where this piece of code is derived from) ip->ip_len is already 
subtrackted with the ip header length ...

> > and maybe
> > http://www.freebsd.org/cgi/cvsweb.cgi/src/sys/contrib/pf/net/
> > pf_if.c.diff?r1=1.5&r2=1.6&f=h
>
> what's that exactly? do you think that's also a problem with
> dragonfly/freebsd-stable's ppp? It seems that this is more a problem of
> ppp. Do you think that this can be fixed in ppp? But as a preliminary
> solution definitively worth consideration.

The problem is that there seem to be things out there in the wild, that rely 
on the IPCP address sticking around ... seems to be broken by design, but if 
you can fix ppp w/o breaking something else - please let me know. In any case 
this (tries to) fix the symptom ...

> thanks for your support!
>
> >> Time to commit: Sunday
>
> oh btw: joerg, how do you intend pf to be updated later? without a
> vendor tag, i can imagine that creating + merging diffs from upstream
> could be a bitch.

Yes ... good point.

-- 
/"\  Best regards,                      | mlaier@xxxxxxxxxxx
\ /  Max Laier                          | ICQ #67774661
 X   http://pf4freebsd.love2party.net/  | mlaier@EFnet
/ \  ASCII Ribbon Campaign              | Against HTML Mail and News

Attachment: pgp00004.pgp
Description: PGP signature



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]