[Mer-discuss] SmartQ mainline kernel

Maurus Cuelenaere mcuelenaere at gmail.com
Wed Oct 14 21:39:05 EEST 2009


Hi all,

as you may or may not know, I'm working on porting the SmartQ 5/7 to 
mainline Linux kernel.

Currently I have LCD, touchscreen, wifi, GPIO buttons, internal NAND and 
external SD card working.
USB host and device are partly working (they init fine, but give USB 
errors, probably due to clocking issues).

I've requested a new machine number on [1] for the SmartQ 7, one for the 
SmartQ 5 should be requested too but I've been waiting for that until 
some actual code is running on it (and driving the LCD).

My idea is to have 2 bootloaders, one compiled for the SmartQ 5 and one 
for the 7, having their respective mach id.

That way one kernel could be used and I *think* this is the cleanest way 
to do (and will get accepted into mainline).


An issue I've been running into is that the SmartQ board doesn't have 
the 48M clock, so I needed to change some things in order to get USB 
host/device to initialize.

Currently I'm doing these clock changes in the platform driver 
(arch/arm/mach-s3c6410/mach-smartq.c), but I don't know whether this 
should be done in the bootloader or in the kernel itself.

Linux doesn't care, and happily takes whatever clocks it is fed (granted 
it can actually boot it :) ), so this could be done from within the 
bootloader.

Another thing that definitely should be done in the bootloader, is 
running the ARM clock at full speed; as Linux currently assumes the 
clock it is booted from is the maximum it can use (when not having a 
regulator for VDDARM, which we (sadly) don't).


Any comments/suggestions/.. on this?


P.S.: my work can be found at [2]

[1] - http://www.arm.linux.org.uk/developer/machines/list.php?id=2479
[2] - 
http://gitorious.org/mcuelenaere-smartq-kernel/mcuelenaere-smartq-kernel
[2] - 
git://gitorious.org/mcuelenaere-smartq-kernel/mcuelenaere-smartq-kernel.git


Regards,
Maurus Cuelenaere


More information about the Mer-discuss mailing list