[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 04 Apr 2007 16:50:11 +0200
From: Patrick McHardy <kaber@...sh.net>
To: hadi@...erus.ca
CC: Denys <denys@...p.net.lb>,
Stephen Hemminger <shemminger@...ux-foundation.org>,
netdev@...r.kernel.org
Subject: Re: HTB/act_mirred problem [was: one more... iproute commands lockup
whole system]
jamal wrote:
> On Wed, 2007-04-04 at 16:13 +0200, Patrick McHardy wrote:
>
>>This seems to be due to be caused by act_mirred returning TC_ACT_STOLEN,
>>which is translated to NET_XMIT_SUCCESS within prio, causing HTB to
>>increase the q.qlen counter and activating the class despite no packet
>>beeing queued.
>>
>>Jamal, we can't return NET_XMIT_SUCCESS unless we've really queued the
>>packet. I can't remeber the reason why this is done, could you remind
>>me?
>
>
> IIRC, It had to do with not confusing TCP to try and retransmit.
No, that was the default return code which applies to TC_ACT_SHOT
and unclassified packets (29f1df6cc1c3ee3530939f0e38d80a9b50645ba5).
Returning NET_XMIT_SUCCESS for TC_ACT_STOLEN/TC_ACT_QUEUED has always
been done.
Anyway, we can't return NET_XMIT_SUCCESS, so how about just returning
NET_XMIT_BYPASS in all cases where the packet was stolen/dropped/...
by TC actions?
> I can
> go back and look at my notes to be certain. At one point i posted those
> notes, it maybe time to add them to the kernel code or doc somewhere.
In case they're still up to date, adding them somewhere under
Documentation/ sounds good.
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists