Ohhhhhhhhhhhkay.............. So it's not about the new driver, it _is_ the b2c2 chipset. Janne, the number (24 days) you have is like what I have with the _old_ driver. Now the following is the interrupt status: [INBT:106]# cat /proc/ieqsat/irqerr dma1_irq : 229074 dma1_timer : 1472746 dma2_irq : 0 dma2_timer : 0 ---------------------- recv_err : 255208 <*********> cc_err : 0 llc_snap_flag : 0 ts_err : 0 timeout : 1 recv_err denotes Receive_error_bit in 0x20c, and note that the number is way too high. No one in this list seems to know when this bit is set _precisely_. If we can get this information I'm sure we can get close to solving this problem or come up with a workaround at least. In the meantime, I wrote a manager which monitors the interrupts and if no interrupts are generated for the certain amount of time, I reload the driver and send a signal to the working process to restart whatever needs to be. This is _ugly_ like hell, but I didn't have choice. On 5/10/05, Janne Liimatainen <jeppe@xxxxxxxxxxxxxxxxxx> wrote: > On Thu, 5 May 2005, Q-ha Park wrote: > > > On 5/5/05, Janne Liimatainen <jeppe@xxxxxxxxxxxxxxxxxx> wrote: > >> Always a hang in 3-6 hours. > > > > Hmm. Was it a whole system lock-up or that the device just didn't > > respond (i.e. no interrupts)? For me, it stops generating interrupts > > in about 10 days with some variance. > > Well, just got the interrupt halt with the new flexcop drivers. > It worked well for 24 days. Had to reload flexcop-pci to make it work > again. > > At least it's a lot better than 3-6 hours. :-) > > -Janne >
