Messages in this thread | | | Subject | Re: [PATCH 1/3] PM: Introduce new top level suspend and hibernation callbacks (rev. 8) | From | Benjamin Herrenschmidt <> | Date | Mon, 14 Apr 2008 17:23:36 +1000 |
| |
On Mon, 2008-04-14 at 08:43 +0200, Oliver Neukum wrote: > > Which is the problem. Suspension is supposed to be transparent. > We cannot start returning error codes for operations which never > failed in practice (eg. switching configurations in USB), just because > the system is about to be suspended.
Returning errors is better than crashing in any way. So what I meant here is that the problem is not as bad as it sounds.
> If you want to request firmware in a PM callback, which makes a > certain > sense, as we should move to a comprehensive API, if we change the API > at all, we need a model with 3 callbacks.
No. At this pace, we'll find reasons to have 98213674 callbacks and will still not be happy.
Prepare() should be the right place to call request_firmware() and if that is a problem because of bugs in some USB things, then those bugs should be fixed.
Ben.
| |