Rod.. Whitworth wrote: > On Sat, 10 Jun 2006 13:03:40 +0100, Stuart Henderson wrote: > > Dragging things back slightly more on-topic, wi(4) has proved > > very reliable; there has been some recent mention on soekris-tech > > (http://marc.theaimsgroup.com/?l=soekris-tech&m=114982978006733) > > about ral(4) stopping working after a while, though it's not yet > > clear whether it's related to soekris 48xx, or whether there's > > some interaction with sis(4), or whether it's something else. > > It seems to be the sis that is the problem. Using ral with other > ethernet cards doesn't cause the bug to show. > > There was a recent mail on the Soekris list from a linux user who > noted that the sis still was alive but stops receiving. : > <quote> > When the card stops receiving it does still send data, I had a ping > going from a wireless client, through ral0 to sis0, then to the remote > machine. The remote machine kept seeing the echo requests and sending > back echo replies. A mirror port on the switch the Soekris connects to > confirms the replies where being sent to the network card. > </quote> > > He fels that the sis interrupt handler may be defective and is doing > more research. > > I'd like to have my ral working in a Soekris too but I don't know > where to look for clues as to what is happening and that Soekris is > my main firewall so obviously I don't want to hang the ethernet > ports. 8-(
FWIW, I switched a box of mine that was running wi and fxp to ral and fxp. Since doing this, the box, functioning without issue before, lockups up hard every one to four days now--and this is often reproducible when pushing a large amount of traffic through ral. I also see the following at various times in /var/log/messages and I do not believe they were there when I was running with wi--also, both cards are on irq 11. Jun 9 16:06:24 meth /bsd: fxp0: warning: SCB timed out Jun 9 16:06:37 meth /bsd: fxp0: device timeout
