From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nigel Cunningham Subject: Re: [RFD] swsusp problem: Drivers allocate much memory during suspend (was: Re: 2.6.21-rc5: swsusp: Not enough free memory) Date: Sat, 14 Apr 2007 08:36:00 +1000 Message-ID: <1176503760.7112.194.camel@nigel.suspend2.net> References: <460B6E4B.3060004@gmail.com> <200704132241.52725.rjw@sisk.pl> <20070413221016.GQ28264@elf.ucw.cz> <200704140035.36355.rjw@sisk.pl> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Return-path: In-Reply-To: <200704140035.36355.rjw@sisk.pl> 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: "Rafael J. Wysocki" Cc: linux-pm@lists.osdl.org, Jiri Slaby , Pavel Machek , Linux kernel mailing list List-Id: linux-pm@vger.kernel.org Hi. On Sat, 2007-04-14 at 00:35 +0200, Rafael J. Wysocki wrote: > On Saturday, 14 April 2007 00:10, Pavel Machek wrote: > > Hi! > > = > > > > > Well, it looks like someone allocated about 6000 pages after we h= ad freed > > > > > enough memory for suspending. > > > > = > > > > We have a tunable allowance in Suspend2 for this, because fglrx > > > > allocates a lot of pages in its suspend routine if DRI is enabled. I > > > > think some other drivers do too, but fglrx is the main one I know. > > > = > > > I wasn't aware of that, thanks for the information. > > > = > > > I think this means we'll probably need to add a tunable, similar to i= mage_size, > > > that will allow the users to specify how much spare memory they want = to reserve > > > for suspending (instead of the constant PAGES_FOR_IO). IMO we can ca= ll it > > > 'spare_memory'. > > = > > Just increase PAGES_FOR_IO. This should not be tunable. > = > Well, I'm not sure. First, we don't really know what the value of it sho= uld be > and this alone is a good enough reason for making it tunable, IMHO. Seco= nd, I > think different systems may need different PAGES_FOR_IO and taking just t= he > maximum (even if we learn how much that actually is) seems to be wasteful= in > the vast majority of cases. Finally, I think it may be possible to speed= up > image saving by increasing PAGES_FOR_IO without playing with the > image size and we can let the user try it (think of distro kernels that a= re > compiled for many different users). It does vary according to the amount of video memory used for DRM, if I understand correctly. Nigel