| From: | Yury Tarasievich <grog@xxxxxxx> |
| Date: | Sun, 3 Jul 2005 16:25:35 +0300 |
On 1 July 2005 19:46, Matthew Dillon wrote: > :On Fri, Jul 01, 2005 at 04:47:24PM +0100, Hiten Pandya wrote: > :> Joerg Sonnenberger wrote: > :> > On Fri, Jul 01, 2005 at 12:25:09PM +0300, Yury Tarasievich wrote: > :> > > :> > Ah, that makes sense. AFAIK we currently don't support native mode > :> > anyway :) > :> > :> Don't we have some form of interrupt storm control that Matt wrote some > :> time ago? > : > :Well, it kind of works, but most devices can't handle it very graceful :) > : > :Joerg > > It works but it sounds like the problem is related to the driver simply > not supporting native mode and doing something weird to the system. If > there is an interrupt storm on irq 15 DFly should report it. So it doesn't, while freebsd-5 manages to deal with it. It's weird, yes, as code in both systems seems to be rather simple. Timer issues perhaps (TIMER_USE_1)? Hang happens after isa_pnp_probe prints its message. Just in case here's the boot verbose printout taken via serial console. This is the printout of May 31 1.3.1-DEVELOPMENT, but June 30 1.3.2-DEVELOPMENT behaves in same way. b.t.w. now I'm trying to make network adapter(s) work with this MB and both integrated Marvell 88E053 (pci id 0x4362) and inserted realtek 8029 do NOT work (do not load), failing with error 6. SMBus module also fails with error 6. I've messed a little with the code previously on freebsd-5 and it seems that contigmalloc fails when asked for about 3.5 megs for jumbo bufs placement. Help/hints/advices, anybody please? --
Attachment:
i915ich6r.bootv1.txt.gz
Description: GNU Zip compressed data