I'm not sure exactly when this problem was introduced, but I'm pretty sure it
didn't exist in the 2.2.x kernels. It does exist in the 2.4.0-test12, 2.4.0,
and 2.4.1 kernels.
Any high bandwidth sustained inward transfer seems to hang up the card after a
little while (< 1 minute with ftp, longer with nfs). Once the card hangs,
the syslog has entries:
NETDEV WATCHDOG: eth0: transmit timed out
cancelling the inward ftp and restarting the network unwedges the problem.
Here is what's in my /proc/pci for the nic:
Bus 0, device 12, function 0:
Ethernet controller: Lite-On Communications Inc LNE100TX (rev 32).
IRQ 11.
There seems to have been some relevant LKML traffic on the topic, but I
couldn't decipher it in terms of what it means to me.
I'm not subscribed to lkml, so please cc me in replies.
Thanks in advance for any help.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
Please read the FAQ at http://www.tux.org/lkml/
