Re: [PATCH V2 0/4] introduce device async actions mechanism

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Wednesday 05 August 2009, Zhang Rui wrote:
> On Wed, 2009-08-05 at 01:33 +0800, Alan Stern wrote:
> > On Tue, 4 Aug 2009, Rafael J. Wysocki wrote:
> > 
> > > Not only that.  I'd like to simplify the design, because IMO using one async
> > > domain would be much more straightforward than using multiple ones.
> > 
> > > If I understand the async framework correctly, the domains are only used for
> > > synchronization, ie. if you want to wait for a group of async operations to
> > > complete, you can put them all into one domain and then call
> > > async_synchronize_full_domain() to wait for them all together.
> > > 
> > > You don't need multiple domains to run multiple things in parallel.
> > 
> > There's a basic confusion going on here.
> > 
> > Rui is using "async domain" to mean a collection of devices which 
> > will be suspended or resumed serially.  Different domains run in 
> > parallel.
> > 
> > Rafael is using "async domain" to mean a collection of devices which 
> > will be suspended or resumed in parallel.  Different domains run 
> > serially.
> > 
> cool, thanks for stating the confusion so clearly, Alan. :)
> 
> Hi, Rafael,
> 
> maybe there is still some confusions about my proposal.
> 
> I re-read kernel/async.c file, and notice that Arjan calls the domain as
> *_synchronization_* domain. sorry I use the wrong word before.
> 
> And I use the synchronization domains just to keep devices dependency.
> 
> First, the general idea is to suspend/resume those slow devices in
> parallel. So we don't suspend/resume them synchronously, instead, we
> move these actions to the global domain.
> 
> Then, I found that these actions can not be run asynchronously because
> they depend on other devices.
> For example, sd depends on SATA controller, we should make sure the PM
> callbacks of sd and ahci sata controller are run serially. so a
> synchronization domain is created for them.
> This is how multiple synchronization domains come from in this proposal.

I think I understand now, thanks.

However, I'd like to avoid any naming confusion in future, so let's follow the
convention of async.c, please.

Best,
Rafael