On Mon, Apr 22, 2002 at 02:14:15PM +0200, Stefan Eletzhofer wrote:
> Well, Russell, I got your hint now :)
> 
> Is a PM callback in assabet.c The Right Thing To Do, or where
> should the platform-dependent GPIO/BCR/whatever PM code live?

I've been trying to avoid both.  The cleaner solution out of both
would be to put a callback into the platform dependent code.

However, if you're working on 2.5, you should really look towards
using the new device tree stuff - Mochel should be updating the
docs today which includes how the power management of the device
tree works.

> And about the UCB reset issue in the l3 code: IMHO only the first line
> should be removed, thus:
> 
> --- l3-bit-sa1100.c.org       Mon Apr 22 14:10:39 2002
> +++ l3-bit-sa1100.c   Mon Apr 22 14:10:48 2002
> @@ -250,8 +250,6 @@
>                * need to do this here so that we can communicate on
>                * the I2C/L3 buses.
>                */
> -             ASSABET_BCR_set(ASSABET_BCR_CODEC_RST);
> -             mdelay(1);
>               ASSABET_BCR_clear(ASSABET_BCR_CODEC_RST);
>               mdelay(1);
>               ASSABET_BCR_set(ASSABET_BCR_CODEC_RST);
> 
> The UCB would then get a reset when the l3 module is inserted, though.
> Maybe we need some sort of resource management there?

It's probably fine to get rid of that - I only put it in because nothing
else appeared to guarantee it was reset.

Really, we want to release the reset when anything that needs it is
loaded, and not before.

_______________________________________________
http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm
http://www.arm.linux.org.uk/armlinux/mailinglists.php
Please visit the above addresses for information on this list.

Reply via email to