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

[Ian Campbell <ian.campbell@citrix.com>]

On Thu, 2015-06-18 at 22:09 +0800, Peng Fan wrote:
> Hi,
> 
> I am porting xen to an Cortex-A7 soc and met Dom0 kernel panic. I have 
> no clear idea about why Dom0 kernel panic.

Have you confirmed that this same kernel runs reliably natively on this
platform?


[...]
> / # Unable to handle kernel NULL pointer dereference at virtual address 
> 00000000
> pgd = 84f1c000
> [00000000] *pgd=8cf15831, *pte=00000000, *ppte=00000000
> Internal error: Oops: 80000007 [#1] PREEMPT SMP ARM
> Modules linked in:
> CPU: 0 PID: 89 Comm: sh Not tainted 3.14.28-01780-g281e5c1-dirty #269
> task: 84c07a80 ti: 84f02000 task.ti: 84f02000
> PC is at 0x0
> LR is at call_timer_fn.isra.33+0x24/0x88
> pc : [<00000000>]    lr : [<80038b38>]    psr: 20000113

Your kernel has jumped to address 0x0 for some reason. You should use
gdb or something to examine you vmlinux file and try and figure out what
LR is doing, which may give you a hint as to why it is apparently
jumping to address zero.

> (XEN) Latest ChangeSet: Mon Jun 15 18:25:34 2015 +0800 git:c01e139-dirty

What changes have you made to your version of Xen? c01e139 is not an
upstream commit.

> I am not sure whether this realted to timer, the timer dts:

I think it looks like a software timer thing in the kernel rather than a
h/w timer thing.

> The Dom0 kernel is also configured with " CONFIG_ARM_ARCH_TIMER=y", will 
> this incur errors? Or should Dom0 kernel not use arm arch timer?

Guests on Xen (including dom0) _should_ be using the arch timer.

Ian.