From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Rafael J. Wysocki" Subject: Re: [RFC][PATCH -mm] PM: Do not sync from within the freezer during suspend to RAM Date: Thu, 5 Jul 2007 13:28:49 +0200 Message-ID: <200707051328.50845.rjw@sisk.pl> References: <200707041658.59588.rjw@sisk.pl> <20070704224942.GD2491@elf.ucw.cz> <200707050852.15392.nigel@nigel.suspend2.net> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-15" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <200707050852.15392.nigel@nigel.suspend2.net> Content-Disposition: inline List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: linux-pm-bounces@lists.linux-foundation.org Errors-To: linux-pm-bounces@lists.linux-foundation.org To: nigel@suspend2.net Cc: Matthew Garrett , Miklos Szeredi , LKML , Pavel Machek , pm list , Ingo Molnar List-Id: linux-pm@vger.kernel.org On Thursday, 5 July 2007 00:52, Nigel Cunningham wrote: > Hi. > > On Thursday 05 July 2007 08:49:42 Pavel Machek wrote: > > On Thu 2007-07-05 08:48:15, Nigel Cunningham wrote: > > > Hi. > > > > > > On Thursday 05 July 2007 00:58:58 Rafael J. Wysocki wrote: > > > > From: Rafael J. Wysocki > > > > > > > > The syncing of filesystems from within the freezer in not needed for > suspend > > > to > > > > RAM. Change freeze_processes() so that it doesn't execute sys_sync() > and > > > > introduce the "syncing" version of it to be called from the hibernation > code > > > > paths. > > > > > > > > Signed-off-by: Rafael J. Wysocki > > > > --- > > > > include/linux/freezer.h | 14 ++++++++++++-- > > > > kernel/power/disk.c | 2 +- > > > > kernel/power/main.c | 6 ++++++ > > > > kernel/power/process.c | 8 +++++--- > > > > kernel/power/user.c | 2 +- > > > > 5 files changed, 25 insertions(+), 7 deletions(-) > > > > > > Looks ok, except that I wonder if you want the following fragment. It > looks to > > > me (looking at rc6) like with this code, you'll currently call sys_sync > twice > > > when suspending to ram. Maybe I'm misreading it. Also, shouldn't it be > done > > > after taking the mutex? > > > > sys_sync() should be okay to call, mutex or not. > > Yeah. That wasn't my point, sorry. Calling sys_sync is pointless if you're > going to fail to take the mutex. It makes more sense to know you've got it > before you start doing things related to actually suspending. Well, that's a valid point, I'll move it under the mutex. And why do you think it will be called twice? Greetings, Rafael -- "Premature optimization is the root of all evil." - Donald Knuth