Re: Dom0 kernel panic when porting xen to new arm soc

[Julien Grall <julien.grall@citrix.com>]

On 25/06/2015 16:09, Peng Fan wrote:
> Hi Julien,

Hi,

> On 6/23/2015 9:56 PM, Peng Fan wrote:
>> Hi,
>>
>> On 6/23/2015 9:36 PM, Julien Grall wrote:
>>> Hi,
>>>
>>> On 23/06/15 14:03, Peng Fan wrote:
>>>> I did not enable LPAE for DOM0 kernel, use shor page table.
>>>> Following is the full log from uboot to kernel with DOM0 512M:
>>>
>>> Which CONFIG_VMSPLIT_* do you use? Can you try to use another one? I
>>> remembered it had some effect on the offset between physical and virtual
>>> address.
>>   CONFIG_VMSPLIT_2G=y
>>
>> CONFIG_PAGE_OFFSET=0x80000000
>>
>> ok. Will try 3G:1G split.
> Later I'll reply with log info about this(Do not have log at hand).
> Still panic, maybe something wrong with gnutab configration in my side,
> I use default gnutab address/size.

Did you check that the gnttab doesn't overlap a device/RAM region of 
your hardware?

I've posted a patch a week ago to find automatically a region for the 
grant table [1] in DOM0 memory. It will avoid you to go through the 
datasheet.

> Current I met a DomU boot issue, if do not use blk backend, DomU can
> boot with ramfs as rootfs. If use an image file as rootfs, DomU can not
> boot.
>
> I am not sure why this happends:"
> libxl: error: libxl_create.c:1195:domcreate_launch_dm: unable to add
> disk devices
> libxl: error: libxl_device.c:799:libxl__initiate_device_remove: unable
> to get my domid
> "
> By using gdb, I found domcreate_launch_dm fails to get domid, then it
> reports unable to add disk devices. I am not familiar with xenstore and
> etc. Did I miss some configuation?

The "unable to get my domid" looks like an issue with xenstore. Is 
xenstored running?

Also, did you built you DOM0 kernel with CONFIG_XEN_BLKDEV_BACKEND?

Regards,

[1] http://lists.xen.org/archives/html/xen-devel/2015-06/msg02831.html

-- 
Julien Grall