Re: [2.6.10-rc1-bk5] e1000 broken badly on IBM T42

Re: [2.6.10-rc1-bk5] e1000 broken badly on IBM T42

[Shawn Starr]

I should just answer it myself, restarting fixed the interface negotiation 
'blip'. Perhaps the driver somehow did not reset and retry negotiation?

That did look interesting though :)

On October 27, 2004 00:54, Shawn Starr wrote:
> I should also note, there is no gigabit interface on a linksys router ;-)
> it doesn't negotiate as 100Mbit now?
>
> Shawn.
>
> On October 27, 2004 00:33, Shawn Starr wrote:
> > NETDEV WATCHDOG: eth0: transmit timed out
> > e1000: eth0: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex
> > NETDEV WATCHDOG: eth0: transmit timed out
> >
> > eth0      Link encap:Ethernet  HWaddr 00:0D:60:CA:C1:97
> >           inet addr:192.168.10.5  Bcast:192.168.10.255 
> > Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
> >           RX packets:4294883167 errors:4294370903 dropped:4294796898
> > overruns:4294882097 frame:4294711699 TX packets:4294883949
> > errors:4294796898 dropped:0 overruns:0 carrier:4294711699
> > collisions:4294882097 txqueuelen:1000
> >           RX bytes:470309 (459.2 KiB)  TX bytes:108971 (106.4 KiB)
> >           Base address:0x8000 Memory:c0220000-c0240000
> >
> >
> > I can't even use the interface, some counters are going up others going
> > backwards? ;-)
> >
> > Anyone notice this?
> >
> > Shawn.

Re: [2.6.10-rc1-bk5] e1000 broken badly on IBM T42

[Lee Revell]

On Wed, 2004-10-27 at 05:21 -0400, Shawn Starr wrote:
> I should just answer it myself, restarting fixed the interface negotiation 
> 'blip'. Perhaps the driver somehow did not reset and retry negotiation?
> 
> That did look interesting though :)

AIUI it's impossible to do 100% reliable autonegotiation with Ethernet.
The best you can do is try to detect when you might have gotten it wrong
and reset the interface.

Lee

Re: [2.6.10-rc1-bk5] e1000 broken badly on IBM T42

[Shawn Starr]

Actually, there is a bug, when suspending the laptop its not storing the 
negotiated link status and resets to gigabit.

This looks to be a bug :-)

Shawn.

On October 27, 2004 11:51, Lee Revell wrote:
> On Wed, 2004-10-27 at 05:21 -0400, Shawn Starr wrote:
> > I should just answer it myself, restarting fixed the interface
> > negotiation 'blip'. Perhaps the driver somehow did not reset and retry
> > negotiation?
> >
> > That did look interesting though :)
>
> AIUI it's impossible to do 100% reliable autonegotiation with Ethernet.
> The best you can do is try to detect when you might have gotten it wrong
> and reset the interface.
>
> Lee

Re: [2.6.10-rc1-bk5] e1000 broken badly on IBM T42

[James Turner]

>On October 27, 2004 00:33, Shawn Starr wrote:
>
>>NETDEV WATCHDOG: eth0: transmit timed out
>>e1000: eth0: e1000_watchdog: NIC Link is Up 1000 Mbps Full Duplex
>>NETDEV WATCHDOG: eth0: transmit timed out
>>
>>eth0      Link encap:Ethernet  HWaddr 00:0D:60:CA:C1:97
>>          inet addr:192.168.10.5  Bcast:192.168.10.255 
>>Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>>          RX packets:4294883167 errors:4294370903 dropped:4294796898
>>overruns:4294882097 frame:4294711699 TX packets:4294883949
>>errors:4294796898 dropped:0 overruns:0 carrier:4294711699
>>collisions:4294882097 txqueuelen:1000
>>          RX bytes:470309 (459.2 KiB)  TX bytes:108971 (106.4 KiB)
>>          Base address:0x8000 Memory:c0220000-c0240000

I also get 4bn errors reported using e1000 on a T42 (2378-DXU) after
coming out of APM suspend. However, my network interface seems to work
normally otherwise, as far as I can tell. It always reports 100Mbps.

eth0      Link encap:Ethernet  HWaddr 00:0D:60:FF:0D:9A
           inet addr:172.16.22.206  Bcast:172.16.22.255 Mask:255.255.255.0
           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
           RX packets:887386 errors:4294967254 dropped:4294967284
overruns:4294967290 frame:4294967278
           TX packets:452521 errors:4294967284 dropped:0 overruns:0
carrier:4294967278
           collisions:4294967290 txqueuelen:1000
           RX bytes:1319970165 (1.2 GiB)  TX bytes:32622656 (31.1 MiB)
           Base address:0x8000 Memory:c0220000-c0240000

I'm using 2.6.9 (FC3), which includes e1000 version 5.3.19-k2. Looks
like there is a newer version at http://sourceforge.net/projects/e1000/
(maybe the same as in 2.6.10-rc1?).

James.

Please cc: any replies to me - thanks (IANAKH).