bt878 based dvr card vs. pci pass-thru

bt878 based dvr card vs. pci pass-thru

[Sipos Ferenc]

Hi All,

I'm trying to set up a surveillance system in a domU using
pciback/pcifront to pass through the 4x bt878 chips, however this what I
get in the kernel logs:

"""
Dec 15 16:28:18 hive kernel: pciback 0000:07:0c.0: Driver tried to write
to a read-only configuration space field at offset 0x50, size 2. This
may be harmless, but if you have problems with your device:
Dec 15 16:28:18 hive kernel: 1) see permissive attribute in sysfs
Dec 15 16:28:18 hive kernel: 2) report problems to the xen-devel mailing
list along with details of your device obtained from lspci.
Dec 15 16:28:18 hive kernel: PCI: Enabling device 0000:07:0c.0 (0000 ->
0002)
Dec 15 16:28:18 hive kernel: ACPI: PCI Interrupt 0000:07:0c.0[A] -> GSI
16 (level, low) -> IRQ 16
"""
(Such message for every bttv chip.)

According to option 1) above here is what's in my
xend-pci-permissive.sxp file:
"""
(unconstrained_dev_ids
    ('0000:07:0c.0'
    '0000:07:0c.1'
    '0000:07:0d.0'
    '0000:07:0d.1'
    '0000:07:0e.0'
    '0000:07:0e.1'
    '0000:07:0f.0'
    '0000:07:0f.1')
)
"""

lspci says (relevant part only):
"""
0000:07:0c.0 Multimedia video controller: Brooktree Corporation Bt878
Video Capture (rev 11)
0000:07:0c.1 Multimedia controller: Brooktree Corporation Bt878 Audio
Capture (rev 11)
0000:07:0d.0 Multimedia video controller: Brooktree Corporation Bt878
Video Capture (rev 11)
0000:07:0d.1 Multimedia controller: Brooktree Corporation Bt878 Audio
Capture (rev 11)
0000:07:0e.0 Multimedia video controller: Brooktree Corporation Bt878
Video Capture (rev 11)
0000:07:0e.1 Multimedia controller: Brooktree Corporation Bt878 Audio
Capture (rev 11)
0000:07:0f.0 Multimedia video controller: Brooktree Corporation Bt878
Video Capture (rev 11)
0000:07:0f.1 Multimedia controller: Brooktree Corporation Bt878 Audio
Capture (rev 11)
"""

For some reason my system froze the other day (dom0) and I'm not sure if
this has anything to do with that as I was not able to gather any
information on what might have caused the freeze.

If anyone has any thoughts on how to make those warnings go away or has
experience with similar setup, please put me on the right track.

Thanks,
Frank

PS: Devguys, please note I'm not subscribed on the -devel list, so
either reply to me directly or the -users list which is also copied on
this mail. 

Re: bt878 based dvr card vs. pci pass-thru

[Keir Fraser]

On 16/12/06 9:05 pm, "Sipos Ferenc" <frank@sudo.hu> wrote:

> According to option 1) above here is what's in my
> xend-pci-permissive.sxp file:
> """
> (unconstrained_dev_ids
>     ('0000:07:0c.0'
>     '0000:07:0c.1'

This file should contain your device id, not slot address. Find out the
numeric device id from 'lspci -n'.

 -- Keir

Re: Re: [Xen-devel] bt878 based dvr card vs. pci pass-thru

[Sipos Ferenc]

Hi All.

> This file should contain your device id, not slot address. Find out the
> numeric device id from 'lspci -n'.
It was clearly stated in the docs that I should have used the output of
'lspci -n' but somehow it supressed my attention. Sorry for wasting your
time on that.

On an other note, I've did what you and the manual suggested and here is
the result:

"""
Dec 18 10:47:43 hive kernel: pciback 0000:07:0e.0: enabling permissive
mode configuration space accesses!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0e.0: permissive mode is
potentially unsafe!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0e.1: enabling permissive
mode configuration space accesses!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0e.1: permissive mode is
potentially unsafe!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0f.0: enabling permissive
mode configuration space accesses!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0f.0: permissive mode is
potentially unsafe!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0f.1: enabling permissive
mode configuration space accesses!
Dec 18 10:47:43 hive kernel: pciback 0000:07:0f.1: permissive mode is
potentially unsafe!
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0c.0: assign to
virtual slot 0
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0c.1: assign to
virtual slot 0 func 1
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0d.0: assign to
virtual slot 1
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0d.1: assign to
virtual slot 1 func 1
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0e.0: assign to
virtual slot 2
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0e.1: assign to
virtual slot 2 func 1
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0f.0: assign to
virtual slot 3
Dec 18 10:47:43 hive kernel: pciback: vpci: 0000:07:0f.1: assign to
virtual slot 3 func 1
Dec 18 10:47:43 hive kernel: device vif2.0 entered promiscuous mode
Dec 18 10:47:43 hive kernel: ADDRCONF(NETDEV_UP): vif2.0: link is not
ready
Dec 18 10:47:47 hive kernel: ADDRCONF(NETDEV_CHANGE): vif2.0: link
becomes ready
Dec 18 10:47:47 hive kernel: xenbr0: port 3(vif2.0) entering learning
state
Dec 18 10:47:47 hive kernel: xenbr0: topology change detected,
propagating
Dec 18 10:47:47 hive kernel: xenbr0: port 3(vif2.0) entering forwarding
state
Dec 18 10:47:55 hive kernel: PCI: Enabling device 0000:07:0c.0 (0000 ->
0002)
Dec 18 10:47:55 hive kernel: ACPI: PCI Interrupt 0000:07:0c.0[A] -> GSI
16 (level, low) -> IRQ 16
Dec 18 10:47:55 hive kernel: PCI: Enabling device 0000:07:0d.0 (0000 ->
0002)
Dec 18 10:47:55 hive kernel: ACPI: PCI Interrupt 0000:07:0d.0[A] -> GSI
17 (level, low) -> IRQ 20
Dec 18 10:47:57 hive kernel:  [__report_bad_irq+36/128] __report_bad_irq
+0x24/0x80
Dec 18 10:47:57 hive kernel:  [note_interrupt+162/256] note_interrupt
+0xa2/0x100
Dec 18 10:47:57 hive kernel:  [__do_IRQ+233/256] __do_IRQ+0xe9/0x100
Dec 18 10:47:57 hive kernel:  [do_IRQ+55/112] do_IRQ+0x37/0x70
Dec 18 10:47:57 hive kernel:  [__do_IRQ+179/256] __do_IRQ+0xb3/0x100
Dec 18 10:47:57 hive kernel:  [evtchn_do_upcall+146/256]
evtchn_do_upcall+0x92/0x100
Dec 18 10:47:57 hive kernel:  [hypervisor_callback+61/72]
hypervisor_callback+0x3d/0x48
Dec 18 10:47:57 hive kernel:  [<cd8b5107>] uhci_irq+0x27/0x190
[uhci_hcd]
Dec 18 10:47:57 hive kernel:  [<cd8e03a4>] usb_hcd_irq+0x24/0x60
[usbcore]
Dec 18 10:47:57 hive kernel:  [handle_IRQ_event+89/160] handle_IRQ_event
+0x59/0xa0
Dec 18 10:47:57 hive kernel:  [__do_IRQ+138/256] __do_IRQ+0x8a/0x100
Dec 18 10:47:57 hive kernel:  [do_IRQ+55/112] do_IRQ+0x37/0x70
Dec 18 10:47:57 hive kernel:  [evtchn_do_upcall+146/256]
evtchn_do_upcall+0x92/0x100
Dec 18 10:47:57 hive kernel:  [hypervisor_callback+61/72]
hypervisor_callback+0x3d/0x48
Dec 18 10:47:57 hive kernel:  [safe_halt+32/80] safe_halt+0x20/0x50
Dec 18 10:47:57 hive kernel:  [start_hz_timer+18/32] start_hz_timer
+0x12/0x20
Dec 18 10:47:57 hive kernel:  [xen_idle+44/96] xen_idle+0x2c/0x60
Dec 18 10:47:57 hive kernel:  [cpu_idle+139/224] cpu_idle+0x8b/0xe0
Dec 18 10:47:57 hive kernel:  [start_kernel+427/496] start_kernel
+0x1ab/0x1f0
Dec 18 10:47:57 hive kernel:  [unknown_bootoption+0/480]
unknown_bootoption+0x0/0x1e0
"""

I needed to force my machine to reboot physically. Any suggestions are
very welcome.

(Ubuntu 6.10 - dom0 & domU - and Xen-3.0.3 binary install with Kodicom
4400R-like bt878 based DVR card)
Should you need more info on my setup, I'm happy to provide you with
that.

Frank

Re: [Xen-users] Re: bt878 based dvr card vs. pci pass-thru

[Chris]

I did some searching and found this relevant bugzilla entry:

http://bugzilla.kernel.org/show_bug.cgi?id=6317

The entry makes me think you're experiencing a separate issue that's a
documented cause of instability.  Also, it seems the crash occurred
before the PCI code could be exercised, since at first you were using
the PCI slot numbers instead of the device ID and yet you still observed
the crash.

As for a fix, at present the xenolinux kernel is still based on the
2.6.16 series against which the bug was filed.  However, you might have
success with the -unstable branch on the off chance that the fix was
back-ported.  If you're willing to give it a shot with -unstable please
send us the results.

Cheers
-Chris