Re: Xen dom0 crash: "d0:v0: unhandled page fault (ec=0000)"

[sven <ml@infinitumb.de>]

On 10/19/2010 08:20 PM, Gianni Tedesco wrote:
>> Like disabling "Optimize for size" in the config?
>> Just tried, didn't make any difference.
>
> Well, I mean I still have"Optimize for size" disabled and it's still
> broken.

Ok.
I meant - regardless how it is set, it fails :)


> I am doing a little more digging, what I have found is that
> xen_build_dynamic_phys_to_machine() is building a valid looking p2m
> table, but later on we get some calls to get_phys_to_machine() which are
> outside the range of max_pfn (which Jeremy tells me is acceptable) but
> at that point max_pfn is zero.

I have to admit that I do not understand (due to lack of any knowledge 
about the internals) what these functions are for..


> In other words, it looks like some important data structures are getting
> over-written with nonsense.

.. but this (even) I do understand.


> I will have a think and get back to debugging this when I have a bit
> more time.
>
> In the mean time, can you provide me with some info about the machines
> you're reproing this on? Mines a brand new dell power-edge xeon thing.

Sure, I'll do all I can to help pest-controlling it.

I'm already at home, off mind it is:

#1 Athlon X2 5600+ on Asus AM2-VM [NForce 430 Chipset]
    (I'm not 100% sure about the Mainboard right now)
    - no addon-cards
    4x1G RAM, also tried with 1x1G

#2 Core i7 920 on Asus P6T-Deluxe [Intel X58 Chipset]
    - a (prehistoric) PCI VGA-Card
    3x2G RAM

So you think the problem only affects specific platforms?


Regards,
  Sven


>> p.s.: someone let me know if i should keep everybody on cc!
>
> Please, that's the convention!

Oh. This isn't nntp, isn't it?
So, Alan & Jeremy, blame Gianni :-)