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

Re: [Xen-devel] [XEN/ARM PATCH v6 1/1] Add OdroidXU board (Exynos 5410)

Hi Suriyan,

On 10/09/14 17:51, Suriyan Ramasami wrote:
On Wed, Sep 10, 2014 at 3:21 PM, Julien Grall <julien.grall@xxxxxxxxxx> wrote:
On 08/09/14 10:26, Suriyan Ramasami wrote:

As can be seen, the secondary CPUs check _hotplug_addr for a non zero
value on first being powered on, or after woken up from a wfe. This
_hotplug_addr happens to be at offset +0x1c from NS_RAM_BASE.
Linux mainline too has this hardcoded +0x1c.

You half read the Linux code... This offset is only add when there is a
secure firmware (detected by "samsung,secure-firmware" node in the device

On the Arndale the node doesn't exist.

Thanks for digging there Julien. Previously, I must have gone through
the linux code with only exynos5410 in mind. Nonetheless, looks like I
have left out the code which handles the arndale and possibly 5420 and
the 5800 (if they do not have "samsung,secure-firmware" defined). But
on the other hand, I do see arndale-octa has the "secure-firmware"
entry which is a 5420 (so does the 5800). This adds to the confusion.
This is from looking at the linux-3.16.y source.

Nonetheless, I think to handle arndale for now, I should add the
"samsung,secure-firmware" logic in the code which will then use
sysram_base_addr instead without any offset.

So, how do I go about this? Should I roll out another patch with these
cumulative changes and also add the BUG_ON change?

I think you could avoid to check the "samsung,secure-firmware" logic by only replicate arch/arm/mach-exynos/platsmp.c to Xen.

I suspect it will also work on the odroid-xu.


Julien Grall

Xen-devel mailing list



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