From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S965271AbXDMWnj (ORCPT ); Fri, 13 Apr 2007 18:43:39 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S965277AbXDMWnj (ORCPT ); Fri, 13 Apr 2007 18:43:39 -0400 Received: from nigel.suspend2.net ([203.171.70.205]:36209 "EHLO nigel.suspend2.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1030873AbXDMWnf (ORCPT ); Fri, 13 Apr 2007 18:43:35 -0400 Subject: Re: [linux-pm] [RFD] swsusp problem: Drivers allocate much memory during suspend (was: Re: 2.6.21-rc5: swsusp: Not enough free memory) From: Nigel Cunningham Reply-To: nigel@nigel.suspend2.net To: Pavel Machek Cc: linux-pm@lists.osdl.org, Jiri Slaby , Linux kernel mailing list In-Reply-To: <20070413223841.GD28269@elf.ucw.cz> References: <460B6E4B.3060004@gmail.com> <200704131400.12021.rjw@sisk.pl> <1176466896.7112.159.camel@nigel.suspend2.net> <200704132241.52725.rjw@sisk.pl> <20070413221016.GQ28264@elf.ucw.cz> <1176503692.7112.189.camel@nigel.suspend2.net> <20070413223841.GD28269@elf.ucw.cz> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="=-sytiPjWJdglYn/ivsfE6" Date: Sat, 14 Apr 2007 08:43:30 +1000 Message-Id: <1176504210.7112.204.camel@nigel.suspend2.net> Mime-Version: 1.0 X-Mailer: Evolution 2.10.0 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org --=-sytiPjWJdglYn/ivsfE6 Content-Type: text/plain Content-Transfer-Encoding: quoted-printable Hi. On Sat, 2007-04-14 at 00:38 +0200, Pavel Machek wrote: > Hi! >=20 > > > > > > Well, it looks like someone allocated about 6000 pages after we= had freed > > > > > > enough memory for suspending. > > > > >=20 > > > > > 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= . > > > >=20 > > > > I wasn't aware of that, thanks for the information. > > > >=20 > > > > I think this means we'll probably need to add a tunable, similar to= image_size, > > > > that will allow the users to specify how much spare memory they wan= t to reserve > > > > for suspending (instead of the constant PAGES_FOR_IO). IMO we can = call it > > > > 'spare_memory'. > > >=20 > > > Just increase PAGES_FOR_IO. This should not be tunable. > >=20 > > If we don't have a means for drivers to pre-allocate or say how much > > memory they need, it should be tunable. Frankly, I'm startled that you > > guys haven't heard of this issue before now. I can't believe everyone > > who has ever wanted to hibernate with DRM enabled has been using > > Suspend2. Maybe this is one of the sources of complaints that swsusp > > isn't reliable? >=20 > We do not support closed-source drivers, and open-source drivers are > well behaved. I didn't say fglrx was the only example. Any system using DRI (not DRM, sorry), would, I think, be expected. I just mention fglrx because I have a Radeon 200M that can only use fglrx for Beryl etc at the mo - it's the one I'm familiar with. > > > > IMO to really fix the problem, we should let the drivers that need = much memory > > > > for suspending allocate it _before_ the memory shrinker is called. = For this > > > > purpose we can use notifiers that will be called before we start th= e shrinking > > > > of memory. Namely, if a driver needs to allocate substantial amoun= t > > > > of memory > > >=20 > > > Yes please. Using that notifier without leaking the memory will be > > > "interesting" but if someone needs so much memory during suspend, let > > > them eat their own complexity. > >=20 > > It doesn't need to be that complex. Add another (optional) function to > > the driver model to let drivers say how much they want and it becomes > > trivial. Maybe this idea should be preferred over the notifier chain. >=20 > Actually, it is trivial to prealocate during boot ;-). As the notifier > chain can be useful for other stuff, too, I'd go that way. Pavel! Talk sense! You're not seriously suggesting squirreling away 35 megabytes of a user's memory at boot just because they might want to hibernate with DRI enabled later? Yes, 35 megabytes is a realistic amount. Regards, Nigel --=-sytiPjWJdglYn/ivsfE6 Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQBGIAeSN0y+n1M3mo0RAp1dAKDhopOaaMNcRBGYC0FTpeDtSIu4vwCgxkxN cLeOLCFZSA64VEa67l6fBsc= =nFM4 -----END PGP SIGNATURE----- --=-sytiPjWJdglYn/ivsfE6--