Re: [Xen-devel] Xen 3.2.0 crash: Pagetable walk

[Keir Fraser <keir.fraser@xxxxxxxxxxxxx>]

On 4/3/08 21:11, "Christopher S. Aker" <caker@xxxxxxxxxxxx> wrote:

> Can someone shed some light on what just happened here, and if this is a
> known/already fixed issue?
> 
> (XEN) Pagetable walk from 00000000c13a7f30:
> (XEN)  L4[0x000] = 0000000395f1b027 0000000000015dc6

Crashed while returning to compat-mode guest. Probably was trying to push an
exception or interrupt frame onto the compat guest's kernel stack at the
time. The faulting guest address was c13a7f30.

> (XEN) ----[ Xen-3.2.0  x86_64  debug=y  Not tainted ]----
> (XEN) CPU:    7
> (XEN) RIP:    e008:[<ffff828c801b3506>] show_page_walk+0xd1/0x1dc

Then the show_page_walk() diagnostic function itself crashes. It had just
printed out the L4 entry, so was probably reading the L3 entry...

> (XEN) ****************************************
> (XEN) Panic on CPU 7:
> (XEN) FATAL PAGE FAULT
> (XEN) [error_code=0000]
> (XEN) Faulting linear address: ffff87aec2ad1088
> (XEN) ****************************************

...except that the faulting linear address is unexpected. The L4 entry of
the original pagetable walk was 0000000395f1b027, hence the physical address
of the L3 entry should be somewhere around 395f1b000, which should have a
virtual address (in Xen's mapping of all physical memory) of
ffff830395f1b000. The actual faulting linear address is literally in the
middle of nowhere: it's an unused virtual address.

How much memory does this machine have? Do you have the xen-syms file
associated with this crash dump? It may shed some light. Mail it to me
privately or stick it somewhere for download if possible.

 -- Keir



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