6-CUR: panic in ipsec processing w/o ipsec in use

[Bjoern A. Zeeb]

On Sun, 17 Oct 2004 gnn at freebsd.org wrote:

Hi,

> At Sat, 16 Oct 2004 19:39:00 +0000 (UTC),
> Bjoern A. Zeeb wrote:
> >
> > Hi,
> >
> > on my MP machine running 6-CURRENT from around late 2004-09-26
> > I got a panic while editing a Makefile of a port; though compiled in
> > I have no ipsec in use on that machine atm.
> >
> > hand transcribed:
> > 	panic
> > 	free +0xaa
> > 	_key_delsp +0x1b
> > 	key_delsp +0xab
> > 	_key_freesp +0x8a
> > 	ipsec4_hdrsiz +0x9b
> > 	ipsec4_hdrsiz_tc +0xa9
> > 	tcp_output +0x9c1
> > 	tcp_usr_send +0x18b
> > 	sosend +0x5e7
> > 	soo_write +0x46
> > 	dofilewrite +0xa8
> > 	write +0x39
> > 	syscall
> > 	Xint0x80_syscall
> >
> >
> > the bt looks strange to me, so I decided to post though
> > this may already be corrected ?
>
> This might be related to a fix, that needs testing, with respect to
> the mbuf data size.  The thing I'm thinking about is this:
>
> There are reports that racoon is unable to complete IKE negotiation
> due to a send to the pfkey socket returning ENOBUFS. This appears to
> be a result of an incorrect assumption about mbuf data size due to a
> change resulting from mbuma. A fix for this is in RELENG_5 now.

I know but as said: "no ipsec in use on that machine".
The questions are:

a) why does anything go into the ipsec processing path if I do not
   use ipsec (no policies, no SAs, no raccon, ...)?
b) why does this happen from dofilewrite?


> You might want to update your system.

yeah I might want to do that.

> From what you say in your email this looks like an intermittent
> error. Have you been able to reproduce this?

no. it just happened this one time in ~14 days.

-- 
Bjoern A. Zeeb				bzeeb at Zabbadoz dot NeT