Re: Thoshiba Tecra 8000 with 3com 3CXFE575CT

On Monday 17 October 2005 20:06, M. Warner Losh wrote:
> : When I start it outside and then connect the card it will detect
> : and configure it correctly.
>
> OK.  This is a HUGE clue.  One that shouldn't be overlooked :-). 
> What it is telling me is that FUNCTION interrupts aren't getting
> through, but STATUS interrupts ARE getting through.  I'll have to go
> re-read my two TOPIC data sheets to make sure, but I think that the
> TOPIC chips, especially the older ones, have a number of knobs for
> function interrupts that are not in the standard bridge chipset.
>
> Please refresh my recollection: Did this used to work with 5.4? 
> IIRC, I did some work to fix o2 micro hangs that disabled the
> function interrupts during card insertion by routing them via IRQ 1
> (which the o2micro actually doesn't route, so they just disappear). 
> Maybe that work around had bad effects for the TOPIC95B that you
> have?

I get this (xl0 watchdog etc) with my little Sony picturebook, trying to 
run current. I'm getting around it using polling for now. I've attached 
a verbose dmesg from 7.0 here. This combination works great with 5.4 
but getting a dmesg for that configuration would be a bit trickier.

As far as I can see from vmstat, some interrupts are getting through to 
irq9 but not the ones that were intended for xl0.

$ vmstat -i
interrupt                          total       rate
irq0: clk                        1145039        999
irq1: atkbd0                           2          0
irq8: rtc                         146591        127
irq9: cbb0 fwohci*                   194          0
irq13: npx0                            1          0
irq14: ata0                         3980          3
Total                            1295807       1130