Hi, As part of recovering from a failed NIC driver (in this case e1000), we would like to reset the device as quickly as possible to allow almost uninterrupted packet flow.
We are using the following devices and driver: e1000 driver e1000-7.5.5. e1000 NIC we are using is 80003ES2LAN Gigabit Ethernet (Copper) (rev 01). We are using a 1Gb switch. >From preliminary measurements, it looks like maybe bringing the link down then up is taking a long time, possibly because of auto-negotiation. We invoked e1000_down then e1000_up. So we tried forcing the link state by using a cross-over cable to connect 2 e1000 NICs on two different machines. One machine was set to master and the other slave mode. We set both to (1000 Mbs, full duplex, autoneg off). Connection was established and we could ping. However, when we bring the NIC on one side down then up, link was not established until after a long delay (1.8s) Forcing the link state should not cause autoneg to occur, what other activities can delay the link to go from down to up? Does the training phase of a 1Gb card have anything to do with the delay? Thanks, Michael Le ------------------------------------------------------------------------------ Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are powering Web 2.0 with engaging, cross-platform capabilities. Quickly and easily build your RIAs with Flex Builder, the Eclipse(TM)based development software that enables intelligent coding and step-through debugging. Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com _______________________________________________ E1000-devel mailing list E1000-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/e1000-devel
