Re: [RFC][PATCH v3 1/3] runtime interpreted power sequences

On Tue, Jul 31, 2012 at 03:26:07PM +0100, Mark Brown wrote:

On Tue, Jul 31, 2012 at 04:22:17PM +0200, Thierry Reding wrote:

quoted

On Tue, Jul 31, 2012 at 03:13:29PM +0100, Mark Brown wrote:

quoted

quoted

__devinit can be discarded if you disable enough kernel features,
HOTPLUG is the main one IIRC, modules might also need to go - drivers
really ought to take a copy of platform data they plan to use at
runtime, though practically speaking you have to try to trigger any
problems.

quoted

HOTPLUG is marked EXPERT and explicitly states that it should only be
disabled if you're not using modules or dynamic device discovery. I
think if you've ignored all of that you're no longer entitled to
complain.

This is framework code - it doesn't have much option.  Disabling HOTPLUG
is totally reasonable on space constrained systems, there's no reason
for the code to break things for people.

Still if you use this code and disable HOTPLUG, then you shouldn't be
using modules either. I mean there is no way you can write a driver that
can gracefully handle its platform data being discarded.

Thierry