Re: [Xen-users] Guest domain with PCI frontend disabled IRQ, backend device dies

[Itai Tavor <itai@xxxxxxxxx>]

> Hi,
>
> I'm having a problem with a guest domain which is using PCI cards.  
> The symptom is that as soon as one of the cards is accessed in the  
> guest, the guest's console shows "Disabled IRQ #16" and the whole  
> system loses networking. Networking is restored when I shut down  
> the guest.
> In the guest IRQ 16 is assigned to the PCI card, but in the host  
> it's assigned to the ethernet controller.
> Not sure what information would be useful for diagnosing this...  
> here are a few details:
> AMD Athlon 64 on ASUS A8N-SLI Premium
> Xen unstable from 5/19/06
> Dom0 and domU systems are both Debian 3.1
>
> Dom0 kernel commaind line:
> /boot/vmlinuz-2.6-xen0 root=/dev/sdc1 ro console=tty0 max_loop=64  
> pciback.hide=(05:06.0)(05:06.2)(05:07.0)(05:07.2)(05:08.0)(05:08.1) 
> (00:02.0)(00:02.1) noapic
> DomU config (relevant part):
>
> pci=[
>       '00:02.0',                        # USB
>       '05:06.0', '05:06.2',             # DVB-T tuner 1
>       '05:07.0', '05:07.2',             # DVB-T tuner 2
>       '05:08.0', '05:08.1',             # Video capture
>     ]
> extra = "noapic acpi=off"
>
> DomU lspci:
> 0000:00:02.0 USB Controller: nVidia Corporation CK804 USB  
> Controller (rev a2)
> 0000:05:06.0 Multimedia video controller: Conexant CX23880/1/2/3  
> PCI Video and Audio Decoder (rev 05)
> 0000:05:06.2 Multimedia controller: Conexant CX23880/1/2/3 PCI  
> Video and Audio Decoder [MPEG Port] (rev 05)
> 0000:05:07.0 Multimedia video controller: Conexant CX23880/1/2/3  
> PCI Video and Audio Decoder (rev 05)
> 0000:05:07.2 Multimedia controller: Conexant CX23880/1/2/3 PCI  
> Video and Audio Decoder [MPEG Port] (rev 05)
> 0000:05:08.0 Multimedia video controller: Brooktree Corporation  
> Bt878 Video Capture (rev 11)
> 0000:05:08.1 Multimedia controller: Brooktree Corporation Bt878  
> Audio Capture (rev 11)
> Dom0 /proc/interrupts:
>            CPU0
>   1:        269        Phys-irq  i8042
>   8:          0        Phys-irq  rtc
>   9:          0        Phys-irq  acpi
> 14:         37        Phys-irq  ide0
> 15:        237        Phys-irq  ide1
> 16:   14779398        Phys-irq  sk98lin
> 21:          0        Phys-irq  libata
> 22:      76699        Phys-irq  libata
> 23:    2746987        Phys-irq  libata
> ...
>
> DomU /proc/interrupts:
>            CPU0
> 16:      41475        Phys-irq  cx88[1], cx88[1]
> 17:          2        Phys-irq  bttv0
> 18:          0        Phys-irq  cx88[0], cx88[0]
> 20:         14        Phys-irq  ohci_hcd:usb1
> 256:      24879     Dynamic-irq  timer0
> 257:        287     Dynamic-irq  xenbus
> 258:        436     Dynamic-irq  xencons
> 259:       3547     Dynamic-irq  blkif
> 260:         68     Dynamic-irq  blkif
> 261:          1     Dynamic-irq  blkif
> 262:       3170     Dynamic-irq  eth0
> NMI:          0
> ERR:          0
>
> Any ideas why this is happening or what I can do about it?
Hm... no idea, huh

Ok, let me try this another way: Is there any way to control the IRQ  
numbers assigned by each kernel, to prevent the host and guest  
kernels from selecting the same IRQ?

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users