Re: Booting XEN3_DOM0 results into "panic: pci_get_capability"

[Christoph Schug <chris+netbsd-port-xen%schug.net@localhost>]

On Thu, Sep 18, 2008, Christoph Egger wrote:

> > > Can you give us a full register dump of all pci devices, please?
> > > It seems, there's a piece of hw violating the pci specs.
> >
> > Not quite sure how to do this, do you mean something like ...
> >
> > | # pcictl pci0 list | cut -d ' ' -f 1 | tr ':' ' ' |\
> > |   while read b d f; do pcictl pci0 dump -b $b -d $d -f $f; done
> 
> Yeah, exactly.

Fine, so under [1] you'll find the output of ...

| # pcictl pci0 list | cut -d ' ' -f 1 |\
|   sed -e 's;0*\([0-9]\);\1;g' | tr ':' ' ' |\
|   while read b d f; do echo "+ dump bus=$b device=$d function=$f"; \
|   pcictl pci0 dump -b $b -d $d -f $f; done

I actually had to strip of leading zeros as pcictl(8) seems to have
problems with that in a inconsistent manner. But that's a different
story, so just for the records ...

| # pcictl pci0 dump -b 0 -d 5  | head -n 1
|   PCI configuration registers:
| # pcictl pci0 dump -b 0 -d 05 | head -n 1
|   PCI configuration registers:
| # pcictl pci0 dump -b 0 -d 9  | head -n 1
|   PCI configuration registers:
| # pcictl pci0 dump -b 0 -d 09 | head -n 1
|   pcictl: "09" is not a number

According to "pcilist list" the host bridge [2] and the memory
controller [3] seems to be unknown.

| # pcictl pci0 list
|   000:00:0: NVIDIA product 0x02f0 (RAM memory, revision 0xa2)
|   000:00:1: NVIDIA product 0x02fa (RAM memory, revision 0xa2)
|   000:00:2: NVIDIA product 0x02fe (RAM memory, revision 0xa2)
|   000:00:3: NVIDIA product 0x02f8 (RAM memory, revision 0xa2)
|   000:00:4: NVIDIA product 0x02f9 (RAM memory, revision 0xa2)
|   000:00:5: NVIDIA product 0x02ff (RAM memory, revision 0xa2)
|   000:00:6: NVIDIA product 0x027f (RAM memory, revision 0xa2)
|   000:00:7: NVIDIA product 0x027e (RAM memory, revision 0xa2)
|   000:02:0: NVIDIA product 0x02fc (PCI bridge, revision 0xa1)
|   000:03:0: NVIDIA product 0x02fd (PCI bridge, revision 0xa1)
|   000:04:0: NVIDIA product 0x02fb (PCI bridge, revision 0xa1)
|   000:05:0: NVIDIA GeForce 6150 LE (VGA display, revision 0xa2)
|   000:09:0: NVIDIA product 0x0270 (RAM memory, revision 0xa2)
|   000:10:0: NVIDIA nForce430 PCI-ISA bridge (ISA bridge, revision 0xa3)
|   000:10:1: NVIDIA nForce430 SMBus Controller (SMBus serial bus, revision 
0xa3)
|   000:10:2: NVIDIA product 0x0272 (RAM memory, revision 0xa3)
|   000:11:0: NVIDIA nForce430 USB Controller (USB serial bus, interface 0x10, 
revision 0xa3)
|   000:11:1: NVIDIA nForce430 USB2 Controller (USB serial bus, interface 0x20, 
revision 0xa3)
|   000:14:0: NVIDIA nForce430 Serial ATA Controller (IDE mass storage, 
interface 0x85, revision 0xa1)
|   000:15:0: NVIDIA nForce430 Serial ATA Controller (IDE mass storage, 
interface 0x85, revision 0xa1)
|   000:16:0: NVIDIA product 0x026f (PCI bridge, interface 0x01, revision 0xa2)
|   000:24:0: Advanced Micro Devices K8 AMD64 HyperTransport configuration 
(host bridge)
|   000:24:1: Advanced Micro Devices K8 AMD64 Address Map configuration (host 
bridge)
|   000:24:2: Advanced Micro Devices K8 AMD64 DRAM configuration (host bridge)
|   000:24:3: Advanced Micro Devices K8 AMD64 Miscellaneous configuration (host 
bridge)

And just for completeness, the kernel being used (hostname modified):

| # uname -a
|   NetBSD pkg.example.com 4.99.72 NetBSD 4.99.72 (GENERIC) #0: Wed Sep 17 
19:24:54 CEST 2008  
foobar%pkg.example.com@localhost:/usr/obj/sys/arch/amd64/compile/GENERIC amd64

[1] http://schug.net/tmp/pci_dell_optiplex740.txt (74 KByte)
[2] http://pci-ids.ucw.cz/iii/?i=10de0270
[3] http://pci-ids.ucw.cz/iii/?i=10de0272

Thanks (again!),
Christoph