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

Re: [Xen-devel] arm: alloc_heap_pages allocates already allocated page





On 07/02/2017 12:47, Julien Grall wrote:


On 07/02/2017 12:41, Vijay Kilari wrote:
On Tue, Feb 7, 2017 at 4:58 PM, Julien Grall <julien.grall@xxxxxxx>
wrote:
On 07/02/2017 11:10, Vijay Kilari wrote:

On Tue, Feb 7, 2017 at 3:37 PM, Julien Grall <julien.grall@xxxxxxx>
wrote:

On 07/02/2017 08:18, Vijay Kilari wrote:

   I am seeing below panic with NUMA during DT mappings in
alloc_heap_pages()
BUG_ON(pg[i].count_info != PGC_state_free);
However, this issue is not there with 4.7 version. The same NUMA
board
boots fine.



I am a bit confused by what you are saying. Xen on ARM does not yet
support
NUMA. I also know you are working on NUMA support. So does the BUG
happen
on
upstream xen or upstream xen + your patches?


I was testing with Andre ITS patches (RFC version 1) + my NUMA patches
+ upstream xen.
However now I tested with upstream xen + Andre ITS patches (staging
branch) on NUMA board.


The RFC v1 is quite an old version. Please give a try using the latest
version [1].

I see panic (similar to what I see with my patches). Log are here.


Well the panic is different now. An ASSERT in list_del is hit this time.
This looks like a memory corruption to me.


http://pastebin.com/QJqUBvD9

The same plain upstream xen + Andre ITS patches boots fine with
non-NUMA
board.


I know that DOM0 cannot boot without ITS on your platform. But as you
don't
reach DOM0, have you tried to boot without the ITS series on NUMA board?

Yes, without ITS patches it is previously (first) reported panic at
"Xen BUG at page_alloc.c:827"

Can you please paste the full log from xen upstream (no debug, no ITS,
no NUMA) and device tree memory node?

Also, please disable CONFIG_DEBUG_DEVICE_TREE it pollutes the logs and I
don't think the option is necessary to solve this problem.


One more thing, if Xen 4.7 was able to go up to booting Dom0 without any patches on a NUMA board. I would recommend to try to bisect and see if you can find an offending commit.

Cheers,

--
Julien Grall

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

 


Rackspace

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