Re: Card Locking Up

On Mon, 7 Apr 2003, James Jeffrey wrote:

> We are using a 4 port D-Link server card which has 4 TULIP network
> interfaces on board. They are detected as follows:
..
> tulip.c:v0.95f 11/17/2002  Written by Donald Becker <[email protected]>
>   http://www.scyld.com/network/tulip.html
> eth0: Digital DS21143-xD Tulip rev 65 at 0xe0002c00, 00:80:C8:CD:42:21, IRQ
> 3.
> eth0:  Index #0 - Media MII (#11) described by a 21142 MII PHY (3) block.
...
> After a while the card locks up, stops transmitting and/or receiving and
> shows this in syslog, over and over again:
> 
> Apr  7 06:55:55 pacman kernel: eth1: Restarted Rx at 10764297 / 10764297.
> Apr  7 06:57:10 pacman kernel: eth1: Too much work during an interrupt,
> csr5=0xf
> 06980c0.
> Apr  7 06:57:10 pacman kernel: eth1: Restarted Rx at 10764329 / 10764329.
> Apr  7 06:57:54 pacman kernel: eth1: Too much work during an interrupt,
> csr5=0xf
> 06980c0.

This appears to show that the card is operating, but that the system is
somehow overloaded.  Either it is running short of kernel memory, or
some other device is very busy and blocks interrupts for a long time.

You can get rid of the message by increasing max_interrupt_work, but
that only masks the system problem, it doesn't solve the problem.

If the interface actually stop receiving traffic completely, please post
the output of
   tulip-diag -a
while the interface is up but not working.

-- 
Donald Becker				[email protected]
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Scyld Beowulf cluster system
Annapolis MD 21403			410-990-9993

_______________________________________________
tulip mailing list, [email protected]
To change to digest mode or unsubscribe visit
http://www.scyld.com/mailman/listinfo/tulip