Re: Lanfinity keeps breaking under load.

On Mon, 21 Apr 2003, Bernhard Sessner wrote:

> I'm running a Compaq Presario 1211EA with Conexant Lanfinity ethernet.
> I've seen several people on the list with problems similar to mine, but I did not
> find a solution posted.
> 
> My current kernel version is 2.4.18 (Debian),  but I get the same problems 
> with other 2.4.1x kernels, also with unpatched ones.
> The driver in 2.4.20 does work actually without breaking down,
> but the maximum throughput is around 300 KByte/s.
> 
> The problem is always the same: Under a high (mostly) NFS transmit load, 
> the interface stops working completely.

Where is the error message from the driver when it stops?

> #from syslog during boot:
> Apr 21 14:42:13 earnie kernel: pci-scan.c:v1.11 8/31/2002  Donald Becker <[email protected]> http://www.scyld.com/linux/drivers.html
> Apr 21 14:42:13 earnie kernel: tulip.c:v0.95f 11/17/2002  Written by Donald Becker <[email protected]>
> Apr 21 14:42:13 earnie kernel:   http://www.scyld.com/network/tulip.html
> Apr 21 14:42:13 earnie kernel:   The PCI BIOS has not enabled the device at 0/72!  Updating PCI command 0003->0007.
> Apr 21 14:42:13 earnie kernel: eth0: Conexant LANfinity rev 8 at 0xc90e7000, 00:50:8B:FA:D5:4D, IRQ 11.
> Apr 21 14:42:13 earnie kernel: eth0:  MII transceiver #1 config 1000 status 782d advertising 01e1.
> Apr 21 14:42:13 earnie kernel: eth0:  MII transceiver #0 config 1000 status 782d advertising 01e1.

This looks normal. 

> The following output has been generated when the interface was "broken down"

> #./tulip-diag -aaemmf
> tulip-diag.c:v2.15 9/23/2002 Donald Becker ([email protected])
>  http://www.scyld.com/diag/index.html
> Index #1: Found a Conexant LANfinity adapter at 0x1400.
> Conexant LANfinity chip registers at 0x1400:
>  0x00: fff88000 ffffffff ffffffff 0795a800 0795aa00 f4660000 e0042202 7bffebef
>  0x40: fffe0000 fffd80e8 fffe0000 fffe0000 ffffffff ffffffff ffffffff f7f9fec8
>  Extended registers:
>  0x80: cc660000 cbffebef f0000019 ffff00ff ffffffff 0795aaa0 0795a970 00000000
>  0xa0: ffffffff 00000000 00000000 00000000 00000000 00000000 00000000 00000000
>  0xc0: 00000000 00000000 00000000 00000000 00000000 0000ffff 00000000 00000000
>  0xe0: 00000000 00000000 00000000 000000ff ffffffff 00000000 00000000 00000000
>  Port selection is MII, full-duplex.
>  Transmit started, Receive started.
>   The Rx process state is 'Waiting for packets'.
>   The Tx process state is 'Idle'.
>   The transmit threshold is 128.

This looks normal -- the device _appears_ to be operating.

> #ifconfig, note that TX bytes is definitely wrong - at this time i had
> transmitted ~ 120 MB.

I haven't yet tracked down how to get the chip to count Tx packets as
such (rather than Tx errors) when in full duplex mode.

-- 
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