> > Are there kernel crash dumps written to /var/crash/`hostname` ?
> I checked after several reboots, and nothing at all.
Too bad.
> >What kind of hardware are you using? cpu? mainboard? chipset? hdd? etc..
> Here's what I've got:
> * ASUS M2N32-SLI Delux motherboard
> * CORSAIR XMS2 2GB (2 X 1GB Sticks) RAM
> * AMD Athlon 64 X2 4600+ Windsor 2.4GHz Socket AM2
> * Western Digital Caviar SE16 320 GB SATA 3.0Gb/s HD
> * XFX GeForce 7800GT
Ok, so a system with nvidia chipset and a s-ata hdd.
This might be the nv_sata sgpio issue,
bug 6748159 / 6748574.
A new feature (SGPIO) was added to the nv_sata driver, but it doesn't work
correctly with quite a few desktop nvidia chipsets (that don't have the
SGPIO hardware). See also:
http://www.opensolaris.org/os/community/on/flag-days/pages/2008091101/
It results in kernel memory corruption or other kernel panics during / after
boot. It is fixed in build 101.
Try to boot the kernel like this, maybe that avoids the panics that you're
getting:
- boot the kernel with options "-kd"
- in kmdb:
::bp nv_sata`_init
::cont
nv_sgp_led_init?w c3
::cont
(this disables the code in the nv_sgp_led_init() function, so that
SGPIO support won't get enabled. The patch should work for both 32-
and the 64-bit kernel)
See also this opensolaris bug entry:
http://defect.opensolaris.org/bz/show_bug.cgi?id=4121
--
This message posted from opensolaris.org
