Hi Simon,

On Fri, May 18, 2012 at 10:44 AM, Simon Horman [off-list ref] wrote:

Hi Magnus,

I have tested these patches on top of the soc-new branch of
rafael's renesas tree, revision 5658c94, which is based on
v3.4-rc6.

Excellent!

I also have the cherry picked following patches before applying your patches.

serial8250-em: clk_get() IS_ERR() error handling fix
serial8250-em: Emma Mobile UART driver V2
serial8250: Introduce serial8250_register_8250_port()
serial8250: Clean up default map and dl code
serial8250: Use dl_read()/dl_write() on RM9K
serial8250: Use dl_read()/dl_write() on Alchemy
serial8250: Add dl_read()/dl_write() callbacks
8250.c: less than 2400 baud fix.
serial/8250_pci: Clear FIFOs for Intel ME Serial Over Lan device on BI
tegra, serial8250: add ->handle_break() uart_port op
clocksource: em_sti: Add DT support
clocksource: em_sti: Emma Mobile STI driver V2
clockevents: Make clockevents_config() a global symbol

Thanks, those are indeed needed for runtime UART and STI support, but
there shouldn't be any build time dependencies.

I have added the follwing patch of yours

gpio: Emma Mobile GPIO driver V2

And one of my own, which just fixes a build problem
in rafael's tree.

ARM: mach-types: Manually add KZM-A9-GT

Right, that's a bit of a mess.

On Wed, May 16, 2012 at 03:44:49PM +0900, Magnus Damm wrote:

quoted

mach-shmobile: Emma Mobile EV2 support V3

[PATCH 01/08] mach-shmobile: Emma Mobile EV2 SoC base support V3

Tested-by: Simon Horman <horms@verge.net.au>

quoted

[PATCH 02/08] ARM: Undelete KZM9D mach-type V3

I did not need this patch with the base I am using,
so I did not test it.

Ok!

Reviewed-by: Simon Horman <horms@verge.net.au>

quoted

[PATCH 03/08] mach-shmobile: KZM9D board support V3

Tested-by: Simon Horman <horms@verge.net.au>

quoted

[PATCH 04/08] mach-shmobile: Emma Mobile EV2 SMP support V3

Tested-by: Simon Horman <horms@verge.net.au>

quoted

[PATCH 05/08] mach-shmobile: Emma Mobile EV2 GPIO support V3

I am only seeing interripts for em_gio.0, is there a
way to exercise em_gio.1 .. 4 ?

When we start using other GPIO pins with interrupts then that will
happen. Right now there is no code.

Duplicate names for interrupts, that is
each of em_gio.0..4 have two entries in /proc/interripts.
Is this intentional?

Yes, it is intentional. There are two interrupt sources per 32 GPIO pins.

quoted

[PATCH 06/08] mach-shmobile: KZM9D board Ethernet support V3

Tested-by: Simon Horman <horms@verge.net.au>

quoted

[PATCH 07/08] mach-shmobile: Emma Mobile EV2 DT support V3

I applied this and the resulting kernel booted.

That's good! Thanks for testing!

quoted

[PATCH 08/08] mach-shmobile: Use DT_MACHINE for KZM9D V3

I applied this and the resulting kernel did not boot.
I saw nothing after the uboot prompt..

Yeah, you need to select CONFIG_ARM_APPENDED_DTB=y and append the DTB
to the zImage, then convert that to uImage.

Lastly, could you consider suppling a patch which provides a defconfig?

I've been asked to provide a defconfig that we can share between
multiple platforms. So it's on my TODO.

Thank you!

/ magnus