2003-03-21 23:20:21

by Alon

[permalink] [raw]
Subject: VIA Rhine timeouts

Hi,

I am suffering from what a quick Google search has shown
to be the infamous netdev watchdog lock in Rhine NICs,a
although in a somwhat different way from most cases I've
seen reported. Instead of failing under heavy load, my card
(in fact, an onboard chip in the ECS L7VTA-L motherboard)
completely refuses to make a connection under 2.4.x kernels.
I tried all releases between 2.4.18 and current 2.4.21-pre5
to no avail. The driver provided by VIA at their site
fared no better.

I include below a snippet of the syslog (debug=7)when trying
to connect to a DHCP server. I also have a TCP dump of the
dialog, if it's of any use (the messages about promiscuous
mode are caused by tcpdump, I believe).

..(lots of snipped stuff)...
Mar 21 23:45:52 taragui kernel: via-rhine.c:v1.10-LK1.1.16 February-15-2003
Written by Donald Becker
Mar 21 23:45:52 taragui kernel: http://www.scyld.com/network/via-rhine.html
Mar 21 23:45:52 taragui kernel: via-rhine: Reset succeeded.
Mar 21 23:45:52 taragui kernel: eth0: VIA VT6102 Rhine-II at 0xe400,
00:0a:e6:31:e8:7b, IRQ 16.
Mar 21 23:45:52 taragui kernel: eth0: MII PHY found at address 1, status 0x786d
advertising 01e1 Link 0021.
Mar 21 23:46:37 taragui dhclient: Internet Software Consortium DHCP Client
V3.0.1rc9
Mar 21 23:46:37 taragui dhclient: Copyright 1995-2001 Internet Software
Consortium.
Mar 21 23:46:37 taragui dhclient: All rights reserved.
Mar 21 23:46:37 taragui dhclient: For info, please visit
http://www.isc.org/products/DHCP
Mar 21 23:46:37 taragui dhclient:
Mar 21 23:46:37 taragui kernel: eth0: via_rhine_open() irq 16.
Mar 21 23:46:37 taragui kernel: eth0: Reset succeeded.
Mar 21 23:46:37 taragui kernel: eth0: Done via_rhine_open(), status 081a MII
status: 786d.
Mar 21 23:46:37 taragui kernel: eth0: VIA Rhine monitor tick, status 0000.
Mar 21 23:46:38 taragui dhclient: Listening on LPF/eth0/00:0a:e6:31:e8:7b
Mar 21 23:46:38 taragui dhclient: Sending on LPF/eth0/00:0a:e6:31:e8:7b
Mar 21 23:46:38 taragui dhclient: Sending on Socket/fallback
Mar 21 23:46:39 taragui kernel: eth0: Promiscuous mode enabled.
Mar 21 23:46:39 taragui kernel: device eth0 entered promiscuous mode
Mar 21 23:46:40 taragui dhclient: DHCPREQUEST on eth0 to 255.255.255.255 port
67
Mar 21 23:46:40 taragui kernel: eth0: Transmit frame #0 queued in slot 0.
Mar 21 23:46:47 taragui dhclient: DHCPREQUEST on eth0 to 255.255.255.255 port
67
Mar 21 23:46:47 taragui kernel: eth0: Transmit frame #1 queued in slot 1.
Mar 21 23:46:47 taragui kernel: eth0: VIA Rhine monitor tick, status 1003.
Mar 21 23:46:57 taragui dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port
67 interval 8
Mar 21 23:46:57 taragui kernel: eth0: Transmit frame #2 queued in slot 2.
Mar 21 23:46:57 taragui kernel: eth0: VIA Rhine monitor tick, status 1003.
Mar 21 23:47:05 taragui dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port
67 interval 14
(same as above for frames #3 to #9)
Mar 21 23:47:58 taragui dhclient: No DHCPOFFERS received.
Mar 21 23:47:58 taragui dhclient: Trying recorded lease 212.183.239.144
Mar 21 23:48:01 taragui kernel: NETDEV WATCHDOG: eth0: transmit timed out
Mar 21 23:48:01 taragui kernel: eth0: Transmit timed out, status 1003, PHY
status 786d, resetting...
Mar 21 23:48:01 taragui kernel: eth0: Reset did not complete in 5 us. Trying
harder.
Mar 21 23:48:01 taragui kernel: eth0: Reset succeeded.
Mar 21 23:48:08 taragui dhclient: No working leases in persistent database -
sleeping.

The weird thing is, it seems to work perfectly with
Debian Woody's stock 2.2.20 idepci kernel. I'd be glad
to stick to it for the time being, but I can't get
support for most of the mobo's functionality (it sports
a KT400/VT8235 combo) under any kernel earlier than
2.4.20.

TIA,

Alon




2003-03-22 05:32:02

by jurriaan

[permalink] [raw]
Subject: Re: VIA Rhine timeouts

From: Alon <[email protected]>
Date: Sat, Mar 22, 2003 at 12:20:09AM +0100
> Hi,
>
> I am suffering from what a quick Google search has shown
> to be the infamous netdev watchdog lock in Rhine NICs,a
> although in a somwhat different way from most cases I've
> seen reported. Instead of failing under heavy load, my card
> (in fact, an onboard chip in the ECS L7VTA-L motherboard)
> completely refuses to make a connection under 2.4.x kernels.
> I tried all releases between 2.4.18 and current 2.4.21-pre5
> to no avail. The driver provided by VIA at their site
> fared no better.
>
This driver has been worked on recently in the 2.5.x kernels - can you
try 2.5.65?

HTH,
Jurriaan
--
Look, I'm about to buy me a double barreled sawed off shotgun and show
Linus what I think about backspace and delete not working.
some anonymous .signature
GNU/Linux 2.5.65-mm3 4276 bogomips load av: 0.31 0.08 0.05

2003-03-22 10:06:45

by Roger Luethi

[permalink] [raw]
Subject: Re: VIA Rhine timeouts

> (in fact, an onboard chip in the ECS L7VTA-L motherboard)
> completely refuses to make a connection under 2.4.x kernels.
> I tried all releases between 2.4.18 and current 2.4.21-pre5
> to no avail. The driver provided by VIA at their site
> fared no better.
> [...]
> The weird thing is, it seems to work perfectly with
> Debian Woody's stock 2.2.20 idepci kernel. I'd be glad

If you have APIC support on, try turning it off.

Roger

2003-03-22 10:13:39

by Denis Perchine

[permalink] [raw]
Subject: Re: VIA Rhine timeouts

On Saturday 22 March 2003 16:17, Roger Luethi wrote:
> > (in fact, an onboard chip in the ECS L7VTA-L motherboard)
> > completely refuses to make a connection under 2.4.x kernels.
> > I tried all releases between 2.4.18 and current 2.4.21-pre5
> > to no avail. The driver provided by VIA at their site
> > fared no better.
> > [...]
> > The weird thing is, it seems to work perfectly with
> > Debian Woody's stock 2.2.20 idepci kernel. I'd be glad
>
> If you have APIC support on, try turning it off.

Is someone working on fixing APIC for VIA motherboards?

--
Denis