Re: port-i386/42966: kernel built from today's netbsd-5 crashes

[Anne Bennett <anne%porcupine.montreal.qc.ca@localhost>]

> unfortunately, none of the ideas changes are "right".  this is partly
> fixed in -current already, but there are two places to fix.  thanks to
> matthias drochner for helping me understand all the parts..
> 
> this patch should work for netbsd-5.  i'll work on getting the 2nd
> part fixed up for -current as well, and submit a pullup.

I have backed out of the earlier patches to drmP.h and drm_pci.c,
and have applied your new patch to drm_pci.c.  The resulting kernel
boots; here I am using it.

Boot message oddities (modulo the USB devices dancing to different
numbers again):

  - I still get this one, and I wish I knew what it meant:

    raid0: cooldown 1168701999 out of range

  - These changes make me wonder whether there's a wrong type
    definition somewhere:

    < cpu0 at mainbus0 apid 0: Intel 686-class, 3000MHz, id 0xf65
    < cpu1 at mainbus0 apid 1: Intel 686-class, 3000MHz, id 0xf65

    > cpu0 at mainbus0 apid 0: Intel 686-class, 129638MHz, id 0xf65
    > cpu1 at mainbus0 apid 1: Intel 686-class, 129638MHz, id 0xf65

    and:

    < timecounter: Timecounter "TSC" frequency 3000980550 Hz quality
    > ERROR: 2362 cycle TSC drift observed

Hmm, I did this as an update, but when I restored the original
drmP.h, I got the old file timestamp.  I suspect I should "touch" 
both files and recompile again.

Will do ASAP; supper time now!


Anne.
-- 
Ms. Anne Bennett, as a private citizen:  anne%porcupine.montreal.qc.ca@localhost
Also reachable more officially at work:  anne%encs.concordia.ca@localhost