From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail190.messagelabs.com (mail190.messagelabs.com [216.82.249.51]) by kanga.kvack.org (Postfix) with SMTP id 107CF6B0047 for ; Sat, 30 Jan 2010 15:53:25 -0500 (EST) From: "Rafael J. Wysocki" Subject: Re: [Update][PATCH] MM / PM: Force GFP_NOIO during suspend/hibernation and resume Date: Sat, 30 Jan 2010 21:53:52 +0100 References: <201001212121.50272.rjw@sisk.pl> <201001301956.41372.rjw@sisk.pl> <1264884140.13861.7.camel@maxim-laptop> In-Reply-To: <1264884140.13861.7.camel@maxim-laptop> MIME-Version: 1.0 Content-Type: Text/Plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Message-Id: <201001302153.53016.rjw@sisk.pl> Sender: owner-linux-mm@kvack.org To: Maxim Levitsky Cc: Alexey Starikovskiy , KOSAKI Motohiro , Benjamin Herrenschmidt , linux-pm@lists.linux-foundation.org, LKML , linux-mm , Andrew Morton List-ID: On Saturday 30 January 2010, Maxim Levitsky wrote: > On Sat, 2010-01-30 at 19:56 +0100, Rafael J. Wysocki wrote:=20 > > On Monday 25 January 2010, Alexey Starikovskiy wrote: > > > Rafael J. Wysocki =D0=BF=D0=B8=D1=88=D0=B5=D1=82: > > > > On Saturday 23 January 2010, Maxim Levitsky wrote: > > > >> On Fri, 2010-01-22 at 22:19 +0100, Rafael J. Wysocki wrote:=20 > > > >>> On Friday 22 January 2010, Maxim Levitsky wrote: > > > >>>> On Fri, 2010-01-22 at 10:42 +0900, KOSAKI Motohiro wrote:=20 > > > >>>>>>>> Probably we have multiple option. but I don't think GFP_NOIO= is good > > > >>>>>>>> option. It assume the system have lots non-dirty cache memor= y and it isn't > > > >>>>>>>> guranteed. > > > >>>>>>> Basically nothing is guaranteed in this case. However, does = it actually make > > > >>>>>>> things _worse_? =20 > > > >>>>>> Hmm.. > > > >>>>>> Do you mean we don't need to prevent accidental suspend failur= e? > > > >>>>>> Perhaps, I did misunderstand your intention. If you think your= patch solve > > > >>>>>> this this issue, I still disagree. but If you think your patch= mitigate > > > >>>>>> the pain of this issue, I agree it. I don't have any reason to= oppose your > > > >>>>>> first patch. > > > >>>>> One question. Have anyone tested Rafael's $subject patch?=20 > > > >>>>> Please post test result. if the issue disapper by the patch, we= can > > > >>>>> suppose the slowness is caused by i/o layer. > > > >>>> I did. > > > >>>> > > > >>>> As far as I could see, patch does solve the problem I described. > > > >>>> > > > >>>> Does it affect speed of suspend? I can't say for sure. It seems = to be > > > >>>> the same. > > > >>> Thanks for testing. > > > >> I'll test that too, soon. > > > >> Just to note that I left my hibernate loop run overnight, and now = I am > > > >> posting from my notebook after it did 590 hibernate cycles. > > > >=20 > > > > Did you have a chance to test it? > > > >=20 > > > >> Offtopic, but Note that to achieve that I had to stop using global= acpi > > > >> hardware lock. I tried all kinds of things, but for now it just ha= nds > > > >> from time to time. > > > >> See http://bugzilla.kernel.org/show_bug.cgi?id=3D14668 > > > >=20 > > > > I'm going to look at that later this week, although I'm not sure I = can do more > > > > than Alex about that. > > > >=20 > > > > Rafael > > > Rafael, > > > If you can point to where one may insert callback to be called just b= efore handing control to resume kernel, > > > it may help... > >=20 > > Generally speaking, I'd do that in a .suspend() callback of one of devi= ces. > >=20 > > If that's inconvenient, you can also place it in the .pre_restore() pla= tform > > hibernate callback (drivers/acpi/sleep.c). It only disables GPEs right= now, > > it might release the global lock as well. > >=20 > > The .pre_restore() callback is executed after all devices have been sus= pended, > > so there's no danger any driver would re-acquire the global lock after = that. >=20 >=20 > Well, I did that very late, very close to image restore. > Still, it didn't work (It hung after the resume, in the kernel that was > just restored, on access to the hardware lock, or in other words in same > way) >=20 > Here is what I did: I saw the patch in the bug entry (http://bugzilla.kernel.org/show_bug.cgi?id=3D14668). Please see the comments in there. Please also test the patch I attached and let's use the bug entry for the tracking of this issue from now on. Rafael -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org