Re: [Xen-devel] Xen 4.4 trunk: device [xxxx:yyyy] has invalid IRQ; check vendor BIOS

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 26/11/2013 21:38, Mark Hurenkamp wrote:
> On 2013-11-26 09:40, Jan Beulich wrote:
>>>>> On 25.11.13 at 22:43, Mark Hurenkamp <mark.hurenkamp@xxxxxxxxx>
>>>>> wrote:
>>> On 2013-11-25 20:16, Mark Hurenkamp wrote:
>>>>
>>>> Looking at the output of dmesg, there is a clear hint of what might be
>>>> wrong; it lists several (>10) of the following messages:
>>>> [    3.957006] pcieport 0000:00:01.0: device [8086:0d01] has invalid
>>>> IRQ; check vendor BIOS
>>>> [    3.957018] pcieport 0000:00:01.0: can't find IRQ for PCI INT A;
>>>> please try using pci=biosirq
>>>
>>> Additionally, i found some more strangeness in the dmesg output:
>>> [    3.725793] pci 0000:06:00.0: BAR 15: can't assign mem pref (size
>>> 0x200000)
>>> [    3.725796] pci 0000:06:04.0: BAR 14: can't assign mem (size
>>> 0x200000)
>>> Also repeating several times for multiple devices.
>>> Could this be related to the problems i see?
>>>
>>>> This while booting the same kernel (3.13-rc1) without xen works fine.
>>
>> With all of the above - posting just snippets is not going to suffice
>> if you want any help with this. Please attach full logs at maximum
>> log level: hypervisor, dom0 kernel, and natively run kernel (either
>> the same binary as used for dom0, or making sure the configs are
>> sufficiently similar).
>>
>> Jan
>
> Thanks for your response, i wasn't sure if this was the right place to
> address this, so i kept my initial mail short. However I would indeed
> appreciate any help that can point me into the direction of a more
> useable xen installation on this system.
> So brace yourself for some information ;-)
>
> xen.cfg:
> ========
> [global]
> default=saucy
>
> [saucy]
> options=loglvl=all guest_loglvl=all sync_console console_to_ring
> kernel=vmlinuz-3.13.0-031300rc1-generic
> root=UUID=f0733825-503b-4c10-b089-adb026767d7e ro libata.force=noncq
> earlyprintk=xen debug loglevel=8
> ramdisk=initrd.img-3.13.0-031300rc1-generic
>
>
> output of xl dmesg:
> Looks like the top part is already gone, and i've no way to capture
> this via serial port.
> I can take some boot camera shots if that is usefull though.

sync_console is quite unnecessary, especially if you are not using a
serial port.

You can use conring_size=<num> to make the Xen console buffer larger.  I
would suggest 128K.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel