Tulip Bug ?

(I can't tell if it made its way to the list......)

I also have some more information. This ethdev is connected to a Efficient Networks Speedstream 5260, which may be the cause of the problems (the modem came w/ a D-Link RTL8139C ethdev, but I replaced both of my ethdevs w/ tulips.) Maybe I need to find out how to program the EEPROM for 10Mbps half duplex ?????

I am currently running 2.4.18-pre9-mjc2, but this also happens w/ 2.4.18-pre9 (and 2.4.18-pre7-mjc, maybe others)

I am getting NETDEV Watchdog timeouts

Mar 13 02:12:54 tabriel kernel: NETDEV WATCHDOG: eth1: transmit timed out
Mar 13 02:12:54 tabriel kernel: eth1: 21140 transmit timed out, status 
fc665410, SIA fffffeff ffffffff 1c09fdc0 fffffec8, resetting...
Mar 13 02:12:54 tabriel kernel: eth1: transmit timed out, switching to 
100baseTx-FDX media.
Mar 13 02:13:02 tabriel kernel: NETDEV WATCHDOG: eth1: transmit timed out
Mar 13 02:13:02 tabriel kernel: eth1: 21140 transmit timed out, status 
fc665410, SIA fffffe1b ffffffff 1c09fdc0 fffffec8, resetting...
Mar 13 02:13:02 tabriel kernel: eth1: transmit timed out, switching to 
10baseT media.
Mar 13 02:13:34 tabriel kernel: NETDEV WATCHDOG: eth1: transmit timed out
Mar 13 02:13:34 tabriel kernel: eth1: 21140 transmit timed out, status 
fc665410, SIA fffffe0b ffffffff 1c09fdc0 fffffec8, resetting...
Mar 13 02:13:34 tabriel kernel: eth1: transmit timed out, switching to 

100baseTx-FDX media.

I have been getting this at largely unpredictable intervals. The solutions seems to be to ifdown the ethdevs, and rmmod the driver. then ifup the ethdevs (which also means I have to stop my PPPoE-DSL connection).

eth1, the dev that is timing out is not sharing its interrupt. These boards are/were marketed under the brand name SVEC. I cannot find the box they came in at the moment, so don't have the model number.

Just for reference, here's the relevant lines from lspci -vv

00:09.0 Ethernet controller: Macronix, Inc. [MXIC] MX98713
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr-
Stepping+ SERR- FastB2B-
Status: Cap- 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-

	Latency: 32
	Interrupt: pin A routed to IRQ 9
	Region 0: I/O ports at e400 [size=256]
	Region 1: Memory at e0100000 (32-bit, non-prefetchable) [size=256]
	Expansion ROM at de000000 [disabled] [size=64K]

00:0c.0 Ethernet controller: Macronix, Inc. [MXIC] MX98713
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr-
Stepping+ SERR- FastB2B+
Status: Cap- 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-

	Latency: 32
	Interrupt: pin A routed to IRQ 11
	Region 0: I/O ports at ec00 [size=256]
	Region 1: Memory at e0102000 (32-bit, non-prefetchable) [size=256]
	Expansion ROM at df000000 [disabled] [size=64K]

TIA
--
tabris.net

It made it both times :)

I see it a lot on a Netgear FA310TX
( detected as Lite-On 82c168 PNIC rev 33) With both v 1.1.8(June 16,2001) and
v 0.9.15-pre9(Nov 6, 2001)
kernel is 2.4.19-pre2-ac2 but it's been a problem for months.

I only get the NETDEV messages.

Using xfs top serve fonts to a Solaris8 box was impossible. too many timeouts. Same for NFS mounted $HOME (from Solaris) making it impossible to use the box.

-Thomas