[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] crash on boot with 4.6.1 on fedora 24

>>> On 28.03.16 at 19:00, <m.a.young@xxxxxxxxxxxx> wrote:
> I get a crash on boot with my Fedora xen-4.6.1-3.fc24 packages. This seems 
> to be related to how it is compiled because the same code compiled under 
> Fedora 23 works. The boot logs are attached. The address mentioned in the 
> crash has the code
>     0xffff82d08023d3c3 <create_bounce_frame+299>:
>      je     0xffff82d08023e90a <autogen_stubs+4241>
> but I have compared it with the Fedora 23 version of create_bounce_frame 
> and as far as I can see the code is the same, so I am a bit stuck on how 
> to debug this further.

Well, it doesn't look like your problem is with create_bounce_frame(),
but instead this

(XEN) d0v0: unhandled page fault (ec=0000)
(XEN) Pagetable walk from ffffffff81d6b665:
(XEN)  L4[0x1ff] = 000000003a088067 0000000000002088
(XEN)  L3[0x1fe] = 000000003a087067 0000000000002087
(XEN)  L2[0x00e] = 000000003a096067 0000000000002096 
(XEN)  L1[0x16b] = 0010000039d6b067 0000000000001d6b

is pointing at an issue with paging of Dom0. The walk shown doesn't,
to me, indicate any reason why a page fault would have got raised
in the first place (not even a missing TLB flush could account for
that, since any fault condition would result in a hardware re-walk).
Some of the data in the registers and on the stack suggest there
are page table manipulations going on in Dom0 around the time of
the crash, so you may want to check where exactly Dom0 was when
that crash occurred.

And then the question of course is: If the crash occurs reliably
with the F24 built binary (but not the F23 one), perhaps you need
to go and compare more than just the one function?


Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.