Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 01 Jun 2009 16:30:41 +0100
From:      Bruce Simpson <bms@incunabulum.com>
To:        deeptech71@gmail.com
Cc:        freebsd-current@freebsd.org
Subject:   Re: panic: igmp_v3_dispatch_general_query: called when version 2
Message-ID:  <4A23F421.8070809@incunabulum.com>
In-Reply-To: <4A2325F1.8010300@gmail.com>
References:  <4A1DD57A.7010704@gmail.com>	<4A1E9831.4010606@incunabulum.net>	<4A1EC8FB.6090206@gmail.com>	<4A1FBFF5.6090103@incunabulum.net> <4A20791D.5070209@gmail.com>	<4A22E2F1.6070404@incunabulum.net> <4A2325F1.8010300@gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
deeptech71@gmail.com wrote:
>>
>> Thanks. Can you please test this patch and let me know if it works 
>> for you?
>
> OK, applied, but what now? If you are sure that you have fixed the 
> bug, but just want me to run a "crash test" before commiting, then all 
> I can say is there's nothing wrong yet, I'll keep running the patch 
> until something comes up, like a panic (and report it). Otherwise I 
> can't test wether the patch does avoid the non-reproducable panic.

I believe (without reproducing it) that the problem was 
igmp_v3_cancel_link_timers() not cancelling the v3 link timer in all 
situations. The panic you saw was due to a v3 timer firing even though 
the timer should have been cancelled by reception of the v2 query from 
your university's LAN router.

The RFC could be worded better about how the 'Older Querier' timer is 
heeded -- on re-reading it makes sense not to flap between IGMP versions 
-- the oldest version in use on the link persists for up to 260s with 
the default protocol timers, only switching version after the timer 
expires is best as it provides some built-in hysteresis.

It sounds like the fix is OK. Obviously, let me know if you see problems 
again, I've checked the patch into HEAD now.

thanks,
BMS




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4A23F421.8070809>