Re: [Xen-devel] Problems booting latest xen pv-ops linux kernel

[Sander Eikelenboom <linux@xxxxxxxxxxxxxx>]

It seems with the bios option "ACPI APIC" on disabled:
(BTW the description of this bios option is: "Allows you to enable or disable 
the ACPI support in the APIC, When enabled the ACPI APIC table pointer is 
included in the RSDT pointer list.")

- xen 3.5 tip && 2.6.18.8-xen                            boots fine
- 2.6.31.6 jeremy's tree (without hypervisor)            boots fine
- xen 3.5 tip && 2.6.31.6 jeremy's tree                  doesn't boot 
completely at gives the problems reported.

With the option on enabled, it also boots xen 3.5 tip && 2.6.31.6 fine, but for 
what i recall i have seen the same problem occur rarely too.

-- 
Sander















Friday, November 20, 2009, 11:26:26 PM, you wrote:

> Hello Konrad,

> With the bios option "ACPI APIC" on disabled, the bug comes up with
> - xen-3.5 tip && 2.6.31.6
> - xen-3.5 #20436 && 2.6.31.6
> - xen 3.5 #20365 && 2.6.31.6


> - xen 3.5 #20365 && 2.6.18.8-xen boots fine.


> Anything else i could try or give additional info about ?

> BTW the description of this bios option is:
> "Allows you to enable or disable the ACPI support in the APIC, When enabled 
> the ACPI APIC table pointer is included in the RSDT pointer list."

> -- 
> Sander




> Friday, November 20, 2009, 3:50:20 PM, you wrote:

>> On Fri, Nov 20, 2009 at 12:25:16PM +0100, Sander Eikelenboom wrote:
>>> I occasionally have the same problem (same messages spit out by the kernel 
>>> on boot.
>>> System keeps running, but everything is terribly slow it seems (these 
>>> messages appear one by one with sometimes several seconds in between)
>>> In the end it can not mount the root filesystem and bails out.


>> This is most curious. Based on your Xen hypervisor logs:

>> ..snip..
>>> (XEN)    IRQ:   0, IRQ affinity:0xffffffff, Vec:240 type=IO-APIC-edge    
>>> status=
>>> 00000000 mapped, unbound
>>> (XEN)    IRQ:   1, IRQ affinity:0x00000001, Vec: 40 type=IO-APIC-edge    
>>> status=
>>> 00000006 mapped, unbound
>>> (XEN)    IRQ:   2, IRQ affinity:0xffffffff, Vec:226 type=XT-PIC          
>>> status=
>>> 00000000 mapped, unbound
>>> (XEN)    IRQ:   3, IRQ affinity:0xffffffff, Vec:227 type=XT-PIC          
>>> status=
>>> 00000002 mapped, unbound
>>> (XEN)    IRQ:   4, IRQ affinity:0xffffffff, Vec:241 type=IO-APIC-edge    
>>> status=
>>> 00000000 mapped, unbound
>>> (XEN)    IRQ:   5, IRQ affinity:0xffffffff, Vec:229 type=XT-PIC          
>>> status=
>>> 00000010 in-flight=0 domain-list=0:  5(----),
>>> (XEN)    IRQ:   6, IRQ affinity:0x00000001, Vec: 48 type=IO-APIC-edge    
>>> status=
>>> 00000002 mapped, unbound
>>> (XEN)    IRQ:   7, IRQ affinity:0xffffffff, Vec:231 type=XT-PIC          
>>> status=
>>> 00000010 in-flight=0 domain-list=0:  7(----),
>>> (XEN)    IRQ:   8, IRQ affinity:0x00000001, Vec: 56 type=IO-APIC-edge    
>>> status=
>>> 00000002 mapped, unbound
>> .. snip..

>> That are not mapped to the Dom0. Meaning that well, ..., they are not being 
>> received by
>> Dom0, which definilty is a problem.

>> <side-track>
>> Mr Goswin,

>> Can you try getting the hypervisor output? That can be done by pressing 
>> Ctrl-A three
>> times and one of the keys for the IRQ to domains mapping. I am just looking 
>> to see
>> if you have the same issue.
>> </side-track>

>> There were some patches (cs/ 20437) that changed the IRQ mapping and I 
>> wonder if
>> it inadvertly triggered this failure.

>> Can you revert back before cs 20437 and compile Xen.gz? Here is one way I 
>> know
>> of how to do this (there is probably a better one):

>> mkdir ../test
>> cd ../test
>> hg clone ../xen-unstable.hg/ -r20436
>> cd xen-unstable.hg
>> hg tip
>> (and it should show c/s 20436 as the latest).







-- 
Best regards,
 Sander                            mailto:linux@xxxxxxxxxxxxxx


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