No root filesystem

[Mark Atkinson]

John Baldwin wrote:

> On Monday 01 September 2008 03:21:16 am Pascal Hofstee wrote:
>> On Sat, Aug 30, 2008 at 3:05 AM, Christian Weisgerber
>> <naddy at mips.inka.de> wrote:
>> > John Baldwin:
>> >
>> >> So the reports I've seen of this all involve the Nvidia MCP55 ATA
> chipset, and
>> >> only the ata controller loses its marbles (so to speak).
>> >
>> > I also observe that k8temp doesn't attach.  Maybe Pascal can check
>> > this. It's not part of GENERIC, so k8temp.ko needs to be explicitly
>> > loaded by the loader for this.
>> >
>> > k8temp0: <AMD K8 Thermal Sensors> on hostb3
>> 
>> Ok .. this morning i went ahead and collected two boot -v logs. One
>> for the old (working) kernel and one for the new (broken) kernel. For
>> this i made sure not to load the snd_hda module as that tended to
>> bloat the verbose boot output beyond the limit that i could still get
>> to the interesting data at the next boot. The two verbose bootlogs can
>> be found at:
>> 
>> http://shadowrun.homeunix.net/boot.verbose.working for the old kernel
>> http://shadowrun.homeunix.net/boot.verbose.broken for the new kernel
>> http://shadowrun.homeunix.net/boot.verbose.diff for the differences
>> between the two.
>> 
>> The working kernel shows several devices there that do Not show up on
>> the new kernel, according to http://www.pcidatabase.com this concerns
>> the following devices all by"Advanced Micro Devices"
>> -found->        vendor=0x1022, dev=0x1100, revid=0x00
>> 0x1100       HyperTransport Technology Configuration
>> 
>> -found->        vendor=0x1022, dev=0x1101, revid=0x00
>> 0x1101       Address Map
>> 
>> -found->        vendor=0x1022, dev=0x1102, revid=0x00
>> 0x1102       DRAM Controller
>> 
>> -found->        vendor=0x1022, dev=0x1103, revid=0x00
>> 0x1103       Miscellaneous Control
> 
> This explains k8temp.  So my earlier test patch to Christian only checked
> on
> i386 which is why it didn't find an issue before.  Try the updated patch
> at http://www.FreeBSD.org/~jhb/patches/pcie.patch
> 
> This does PCI config reads using both methods and panics if it doesn't get
> the same result.
> 

I have a Tyan 2895 (K8WE) motherboard with a nforce professional chipset and
it also has pci issues with -current.  It finds none of the cards on the
bus.  Reverting the commit also allows it to boot up with all cards
attached.

  SVN rev 181987 on 2008-08-22 02:14:23Z by jhb
  Extend the support for PCI-e memory mapped configuration space access:

The pcie patch also panics in the same way already reported.

-- 
Mark Atkinson
atkin901 at yahoo.com
(!wired)?(coffee++):(wired);