[Arm-netbook] [advice sought] EOMA68 kernel support

[lkcl luke]

On Sat, Mar 10, 2012 at 1:06 PM, Derek LaHousse <dlahouss at mtu.edu> wrote:
> On 03/10/2012 05:08 AM, arm-netbook-request at lists.phcomp.co.uk wrote:
>> Hold on, hot plug at run time? If you unplug you will lose power, there is
> no space for a battery on the card. Well not much and no designs have one.
>
> J
>
> Hot-swap?  Or Hibernate and swap?

 potentially both.

>  I thought there was already code for
> turning off a bunch of drivers when going in to hibernate, and then
> re-initializing when coming out. (Not that I know where that is).

 i know that the userspace part is uswsusp2, but other than that i
don't know either.

> On the other hand, hibernate would need to write the CPU-card state to
> storage available on the CPU-card.

 yes.  the NAND flash.  which means there needs to be enough spare to
store the RAM state.  hmm... good point.

>  If you're using the SATA connection,
> you can't use a swap file or partition on the external drive.

 no.  ok, well, you _could_... ha, that'd be interesting.  removing
the CPU card from one chassis, booting up on another, doing some work,
then shutting it down, putting the CPU card *back* into the original
chassis and carrying on where you'd left off!

 that's.... really freaky :)

> And a reminder that we talked before about not putting the entire
> device-tree in the EEPROM.  You've got a couple, variable devices on the
> other side of the CPU-card too, so the EEPROM only holds a branch of the
> device tree.

 yes.

 fortunately, the actual devices on each dynamic bus (SATA, USB, ETH)
don't need to be stored in device-tree.  these buses can be
re-scanned, even I2C i think, as well.

 oo this is gonna be fuuuun :)

 l.