From: "Rafael J. Wysocki" Subject: Re: [RFC][PATCH 0/3] PM: Asynchronous suspend and resume Date: Fri, 14 Aug 2009 13:58:54 +0200 Message-ID: <200908141358.54129.rjw@sisk.pl> References: <200908132008.26508.rjw@sisk.pl> <1250220244.1804.9.camel@rzhang-dt> Mime-Version: 1.0 Content-Type: Text/Plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from ogre.sisk.pl ([217.79.144.158]:39398 "EHLO ogre.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755553AbZHNL62 convert rfc822-to-8bit (ORCPT ); Fri, 14 Aug 2009 07:58:28 -0400 In-Reply-To: <1250220244.1804.9.camel@rzhang-dt> Sender: linux-acpi-owner@vger.kernel.org List-Id: linux-acpi@vger.kernel.org To: Zhang Rui Cc: Alan Stern , linux-pm , linux-acpi , Linux Kernel Mailing List , Len Brown , Arjan van de Ven On Friday 14 August 2009, Zhang Rui wrote: > On Fri, 2009-08-14 at 02:08 +0800, Rafael J. Wysocki wrote: > > On Thursday 13 August 2009, Zhang Rui wrote: > > > On Thu, 2009-08-13 at 05:43 +0800, Rafael J. Wysocki wrote: > > > > On Wednesday 12 August 2009, Alan Stern wrote: > > > > > On Wed, 12 Aug 2009, Rafael J. Wysocki wrote: > > > > >=20 > > > > > > Hi, > > > > > >=20 > > > > > > The following patches introduce a mechanism allowing us to = execute device > > > > > > drivers' suspend and resume callbacks asynchronously during= system sleep > > > > > > transitions, such as suspend to RAM. The idea is explained= in the [1/1] patch > > > > > > message. > > > > > >=20 > > > > > > Comments welcome. > > > > >=20 > > > > > I get the idea. Not bad. > > > >=20 > > > > Thanks! > > > >=20 > > > > > Have you tried it in a serious way? For example, turning on = the > > > > > async_suspend flag for every device? > > > >=20 > > > > No, I've only tested it with a few selected drivers. I'm going= to try the > > > > "async everyone" scenario, though. > > > >=20 > > > > > In one way it isn't as efficient as it could be. You fire of= f a bunch > > > > > of async threads and then make many of them wait for parent o= r child > > > > > devices. They could be doing useful work instead. > > > >=20 > > > =EF=BB=BF=EF=BB=BFare you talking about this scenario, or I find = another problem of this > > > approach: > > > there is a part of dpm_list, dev1->dev_aaa->...->dev_bbb->dev2 > > >=20 > > > dev2 is dev1's first child. > > > dev1 resume takes 1s > > > dev_aaa~dev_bbb resume takes 0.1s. > > >=20 > > > if we call =EF=BB=BFdevice_enable_async_suspend(dev1, true) in or= der to resume > > > device1 asynchronously, the real asynchronous resume only happens > > > between dev1 and dev_aaa to dev_bbb because dev2 needs to wait un= til > > > dev1 resume finished. > >=20 > > Yes, that's how it works, but I would call it a limitation rather t= han a > > problem. It partially stems from the fact that __async_schedule() = executes > > ptr() synchronously in some circumstances (e.g. async_enabled unset= ), so the > > suspend and resume callbacks have to be scheduled in the same order= , in which > > they would have been called synchronously. > >=20 > > > so kernel schedules dev1 resume in an async thread first, and the= n takes > > > 0.1s to finish the dev_aaa to dev_bbb resume, and then sleep 0.9s > > >=20 > > > > I kind of agree, but then the patches would be more complicated= =2E > > > >=20 > > > The problem is that we need to invoke device_resume for every dev= ice > > > synchronously. > >=20 > > Yes, we do. > >=20 > > > I wonder if we can make the child devices inherit the > > > parent's =EF=BB=BFdev->power.async_suspend flag, so that devices = that need to > > > wait are resumed asynchronously, i.e. we never wait/sleep when pa= rsing > > > the dpm_list. > > >=20 > > > this doesn't bring too much benefit in suspend case but it can sp= eed up > > > the resume process a lot. > >=20 > > We can do that at the core level, >=20 > I think you mean we can't do that at the core level. That's correct, sorry for the mistake. > > because there may be dependencies between > > the children the core doesn't know about. Subsystems are free to s= et > > async_suspend for the entire branches of device hierarchy if they a= re known > > not to contain any off-tree dependencies, but the core has no infor= mation > > about that. > >=20 > hmm, but the current patch doesn't handle the off-tree dependencies > neither. > e.g. > dev1, dev2, dev3 > dev2 depends on dev1, dev3 is dev1's first child, > we only promise that dev1 has been resumed before resuming dev3 in th= e > current proposal. Sure, and that's why I said "if they are known not to contain any off-t= ree dependencies". IOW, if a subsystem (e.g. platform) knows that none of = its devices have any off-tree dependencies, it can safely set async_suspend= for all of them. > anyway, this is not a problem after the pm_connection stuff is > implemented. :) Well, that's going to take some time I guess. Thanks, Rafael -- To unsubscribe from this list: send the line "unsubscribe linux-acpi" i= n the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html