On Fri, May 06, 2005 at 11:17:00AM +0000, juan@xtrarom.org wrote:
> >Description:
> 
> 	One domU just crashed when running a pkgsrc bulk build
> 	(60% completed):
> 
> panic: HYPERVISOR_mmu_update failed
> 
> Stopped in pid 27247.1 (gcc) at netbsd:cpu_Debugger+0x4:        leave
> cpu_Debugger(c033a025,caecfa78,cbedceec,6,1) at netbsd:cpu_Debugger+0x4
> panic(c033a940,c04d6cc0,0,0,0) at netbsd:panic+0x12e
> xpq_flush_queue(0,3,22,c08c0900,2) at netbsd:xpq_flush_queue+0x4a
> xennet_rx_push_buffer(c099e000,17,c08c0500,c08c0500,c08c0900) at netbsd:xennet_r
> x_push_buffer+0x9f
> xennet_rx_mbuf_free(c08c0900,caa1a000,42,c099ef80,0) at netbsd:xennet_rx_mbuf_fr
> ee+0x2e
> m_freem(c08c0900,1d0,caa1a022,14,20) at netbsd:m_freem+0x10c
> tcp_input(c08c0900,14,6,ca1be7b0,caecfc94) at netbsd:tcp_input+0x10cf

I wonder if this could be caused by TCP bugs that were recently fixed.
I've seen panics in pool_put() with a similar stack trace. A double-free,
or mbuf corruption in the TCP code could lead to to both panics, depending
on from where the mbuf comes. Now that these fixes have been pulled up to
netbsd-3, could you update your kernels, and tell me if this is still
occuring ?

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--