On Fri, 3 May 2013 21:26:57 +0200
Mike Belopuhov <[email protected]> wrote:

[skip]
 
> Thanks for figuring this out.  Unfortunately, I think your diff
> is slightly wrong.  Since skw and sks are distinct pointers you
> need to keep them in sync when they're equal, otherwise sks will
> point to the old memory.  The diff below resets state keys after
> pf_state_key_attach.

Yes, you are right. As this branch (*sks == *skw) wasn't the case for
my problem I didn't pay enough attention to it. My bad.

> Would you be so kind to verify that it still fixes the original
> problem?

Just tested the kernel with the modified patch applied. It works for me
as well.

> diff --git sys/net/if_pfsync.c sys/net/if_pfsync.c
> index fdbc665..2003325 100644
> --- sys/net/if_pfsync.c
> +++ sys/net/if_pfsync.c

[offtopic]
I see you are using git and wonder if there is a well known convenient
way to use git and cvs together.  Could you share your way of doing
this?
[/offtopic]

-- 
Vitaly Sinilin <[email protected]>

Reply via email to