[PATCH 1/2] ARM: make virt_to_idmap() return unsigned long

[santosh.shilimkar at oracle.com]

On 2/1/16 9:31 AM, Russell King - ARM Linux wrote:
> On Mon, Feb 01, 2016 at 09:10:23AM -0800, santosh shilimkar wrote:
>> Based on the log, I think the patch seems to work fine since the boot
>> reached upto rootfs. The crash seems to be coming from mostky NetCP
>> related compents.
>
> I read the boot log slightly differently - I'm assuming that the
> boot log is from the kernel which _will_ be kexec'd from, so we
> don't yet know whether the patches work or not.
>
I didn't see your [PATCH 2/2] which updates the kexec code with
idmap API. I see your point now but IIUC, this log isn't from
kexec kernel based on command line.

> I looked at the oops, and it's in tasklet_action, but for some
> reason the unwinder is unable to unwind and provide a backtrace,
> so it's not very debuggable.
>
> However, this is a useful backtrace for me in a different way:
> to provide to people who want to deprecate the frame-pointer
> based backtracing to prove why the unwinder based solution still
> isn't up to the job.
>
Indeed. Not having stack trace doesn't help.

Regards,
Santosh