Nivedita Singhvi wrote: > > Thanks for the tip, Andrew. However, it didnt seem > to work for me. (Although it may have caused it to > happen less frequently). > > Found that I'm still running into the ethernet card > time out and hang problem when I booted with noapic. > Additionally found: > > - SMP / non-SMP made no difference > - IO_APIC support compile option made no difference > - uniproc / multiproc makes no difference > - load / size of packet etc also didnt matter > - still happens on both tulip, natsemi drivers. > > I did go back to the 2.2-16 kernel and did not run > into the problem - so it is at least a 2.4 kernel > issue. Also, I do have 2 network cards in each > machine. It does seem to look like an interrupt lost issue (??). > Dunno. You did all the right things. You do seem to be losing interrupts, from my reading of the natsemi status you reported. It doesn't look like an APIC problem. Do you have much interrupt sharing going on? I doubt it... I gues you should watch the values in /proc/interrupts when it starts happening. Do they stop increasing? - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org
