[realtek-bug] realtek 8139C - ongoing problems under kernel 2.2.19

Dan Miller dmiller@Kromos.COM
Tue Aug 27 16:31:04 2002

We are continuing to have ongoing "Abnormal interrupt" problems with the Realtek 8139C chip.  We're using the V1.20 driver from the Scyld website.  These problems do not occur when an Intel eepro100 card is used in place of the Realtek chip.  They appear to happen during or after we've run Webstone tests thru the driver.  In most cases the Ethernet card stops passing data once this error occurs; sometimes we can restore it by downing and uping the interface, but last night that didn't even work.

Mr. Becker has previously made comments about these errors indicating that the errors indicated that the switch was using flow control, or some such; I don't know how to get from that to a solution to this problem, but we *really* need for the Realtek driver to be reliable and to keep running; it does not currently do so.  

Do I need to modify the driver to not use flow control??  Is that even what the problem is?  We've noticed that register dumps from the driver get pushed into dmesg when these errors occur, though for some reason they don't end up in /var/log/messages; should I post one or more of those dumps here?  Would that help get this driver fixed??

I would be grateful for any assistance that we could get with this driver.  It will be a serious problem for us if we have to disqualify the chip because we can't get a reliable driver for it, but I'm being pushed in that direction at this point...

	Dan Miller