[PATCH] ARM: shmobile: Include all 2 GiB of memory on APE6EVM

[Magnus Damm]

Hi Simon, Arnd,

On Wed, Apr 10, 2013 at 9:35 AM, Simon Horman <horms at verge.net.au> wrote:
> On Wed, Apr 10, 2013 at 07:27:41AM +0900, Magnus Damm wrote:
>> Hi Simon,
>>
>> On Wed, Apr 10, 2013 at 1:14 AM, Simon Horman <horms at verge.net.au> wrote:
>> > On Tue, Apr 09, 2013 at 09:34:34PM +0900, Simon Horman wrote:
>> >> On Mon, Apr 08, 2013 at 10:44:44AM +0900, Magnus Damm wrote:
>> >> > From: Takashi Yoshii <takashi.yoshii.zj at renesas.com>
>> >> >
>> >> > Add 1GiB of DRAM at 0x2_0000_0000
>> >> >
>> >> > Signed-off-by: Takashi Yoshii <takashi.yoshii.zj at renesas.com>
>> >> > Signed-off-by: Magnus Damm <damm at opensource.se>
>> >>
>> >> Thanks, queued up in the boards-ape6evm branch.
>> >
>> > I have dropped this patch as it appears to prevent
>> > the board from booting.
>>
>> Is it possible that LPAE=n and HIGHMEM=y in your particular kernel
>> configuration?
>
> Yes, I have confirmed that is what using the ape6evm_defconfig results in.
>
>> >From what I can tell HIGHMEM is broken in cases of large amounts of
>> memory. Try disabling that as a short-term workaround. Then of course
>> someone needs to track down why the generic code doesn't work as
>> expected.
>
> I have confirmed that disabling HIGHMEM allows the board to boot
> with this patch applied.

Thanks for checking.

So HIGHMEM without LPAE seems busted on boards using more than 32-bits
for memory.

I'd like to help out fixing this issue, but I'd like to make sure that
no one else is working on this.

Arnd, are you aware on any work related to making HIGHMEM work without LPAE?

> Should I prepare a patch to the defconfig to accompany this patch?

I believe that would affect a bunch of boards so I don't think that is
necessary at this point. Thanks anyway.

Would it be possible for you to queue up this change in a APE6EVM topic branch?

Thanks,

/ magnus