Sorry for long delay. I lost track in this discussion, was busy with other things, but found finally time to test this. Yes, confirmed that patch makes difference and solves Quagga ospfd problem (fro
Author: Thomas Spatzier <thomas.spatzier@xxxxxxxxxx>
Date: Mon, 3 Jan 2005 10:10:56 +0100
jamal <hadi@xxxxxxxxxx> wrote on 22.12.2004 14:48:28: Does Jamal's solution sound good for all? Then I would change my driver to do the following: just call netif_carrier_off() (not netif_stop_queue)
jamal wrote: The change is simple if theres consensus to go this path. Can you resend your patch? My main objection was that any change should be made in the core, not in individual net drivers. Anot
I didnt send any patch - but heres one that looks right - havent tried compiling it. Attached patch resolves that concern Thats a good question and your point of imposing policy in the kernel is vali
On Tue, 4 Jan 2005, Jeff Garzik wrote: Aren't there cases where people would -not- want the queue to be cleared? Why *do* you want the queue to be cleared? (avoiding NIC /dev/null is only reason righ
On Wed, 4 Jan 2005, jamal wrote: Theres possibly people who would want it differently - so for we havent heard from them. Maybe poll far and wide - or push it in and wait for them to whine. We're whi
Ok, Iam confused - I thought you guys _wanted this_ ;-> The issue is about message obsolency more than it is about reliability. Without this the scenario you played you played for us before was: (thi
On Wed, 5 Jan 2005, jamal wrote: Ok, Iam confused - I thought you guys _wanted this_ ;-> I'm confused too now. We dont want "queue packets" - that's the 'newish' policy I was referring to. (newish in
Author: Tommy Christensen <tommy.christensen@xxxxxxxxx>
Date: Wed, 05 Jan 2005 16:35:13 +0100
jamal wrote: On Tue, 2005-01-04 at 18:28, Jeff Garzik wrote: jamal wrote: The change is simple if theres consensus to go this path. Can you resend your patch? I didnt send any patch - but heres one t
If we assume all drivers do: netif_stop then carrier_off then you dont need that extra check. Thats the working assumption i had - maybe a comment is deserving or we could say we dont think that all
Author: Tommy Christensen <tommy.christensen@xxxxxxxxx>
Date: Thu, 06 Jan 2005 16:06:13 +0100
But qdisc_restart() isn't called any more after the queue is stopped. So how do we get to drain the packets? Another approach could be to reset the qdisc (kind of what dev_deactivate does) if the dri
We return -1; so qdisc restart will be called until theres no packets left on the queue. Did i miss something in the state machine there? reseting qdisc could certainly be part of the device watchdog
Yes, qdisc_run does this while (!netif_queue_stopped(dev) && qdisc_restart(dev) < 0) /* NOTHING */; Unfortunately the watchdog won't do this when carrier is off, which is fair enough since the chip i