Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1758961AbZDRPYY (ORCPT ); Sat, 18 Apr 2009 11:24:24 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755414AbZDRPYP (ORCPT ); Sat, 18 Apr 2009 11:24:15 -0400 Received: from ogre.sisk.pl ([217.79.144.158]:45059 "EHLO ogre.sisk.pl" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754801AbZDRPYO (ORCPT ); Sat, 18 Apr 2009 11:24:14 -0400 From: "Rafael J. Wysocki" To: Alan Jenkins Subject: [PATCH] PM/Hibernate: Fix memory shrinking (Re: [Bug #13058] First hibernation attempt fails) Date: Sat, 18 Apr 2009 17:23:41 +0200 User-Agent: KMail/1.11.2 (Linux/2.6.30-rc2-rjw; KDE/4.2.2; x86_64; ; ) Cc: Linus Torvalds , Jens Axboe , Linux Kernel Mailing List , Kernel Testers List References: <200904181438.18166.rjw@sisk.pl> <49E9CE51.8000803@tuffmail.co.uk> In-Reply-To: <49E9CE51.8000803@tuffmail.co.uk> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-2" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200904181723.42311.rjw@sisk.pl> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 7077 Lines: 164 On Saturday 18 April 2009, Alan Jenkins wrote: > Rafael J. Wysocki wrote: > > On Saturday 18 April 2009, Alan Jenkins wrote: > > > >> Linus Torvalds wrote: > >> > >>> On Fri, 17 Apr 2009, Rafael J. Wysocki wrote: > >>> > >>> > >>>> Can you please try to reproduce the problem with the appended debug patch > >>>> applied and send the output of dmesg to me? > >>>> > >>>> > >>> Maybe something like this instead (or in addition to). > >>> > >>> It does "show_mem()" when memory shrinking fails. It will show a _lot_ of > >>> data. > >>> > >>> Untested, but trivial. > >>> > >>> Linus > >>> --- > >>> > >>> > >> Ok, I applied both your and Rafael's debug patches. dmesg attached. > >> > >> After the failed hibernation, I noticed my touchpad wasn't working. But > >> I think that's something else. I had another go and couldn't reproduce > >> that. It's happened to me once before while testing 2.6.30-; I've also > >> had the keyboard stop working at least once. I'm hoping it's the same > >> bug as "20 ACPI interrupts per second on EEEPC" bug. It could be > >> overloading my bug-ridden EC, which also acts as the keyboard controller. > >> > > > > Thanks for testing! > > > > Clearly, sc.nr_reclaimed is reset in each iteration of the loop in > > shrink_all_memory(): > > > > [ 61.135207] PM: Shrinking memory... <6>before: sc.nr_reclaimed = 0 > > [ 61.180993] pass = 0, prio = 12, sc.nr_reclaimed = 0 > > [ 61.181004] pass = 0, prio = 11, sc.nr_reclaimed = 0 > > [ 61.181014] pass = 0, prio = 10, sc.nr_reclaimed = 0 > > [ 61.181024] pass = 0, prio = 9, sc.nr_reclaimed = 0 > > [ 61.181033] pass = 0, prio = 8, sc.nr_reclaimed = 0 > > [ 61.181043] pass = 0, prio = 7, sc.nr_reclaimed = 0 > > [ 61.186509] pass = 0, prio = 6, sc.nr_reclaimed = 0 > > [ 61.186525] pass = 0, prio = 5, sc.nr_reclaimed = 0 > > [ 61.186534] pass = 0, prio = 4, sc.nr_reclaimed = 0 > > [ 61.186544] pass = 0, prio = 3, sc.nr_reclaimed = 0 > > [ 61.333383] pass = 0, prio = 2, sc.nr_reclaimed = 7746 > > [ 61.436711] pass = 0, prio = 1, sc.nr_reclaimed = 1957 > > [ 61.556712] pass = 0, prio = 0, sc.nr_reclaimed = 4528 > > [ 61.556729] pass = 1, prio = 12, sc.nr_reclaimed = 0 > > [ 61.556739] pass = 1, prio = 11, sc.nr_reclaimed = 0 > > [ 61.556749] pass = 1, prio = 10, sc.nr_reclaimed = 0 > > [ 61.556759] pass = 1, prio = 9, sc.nr_reclaimed = 0 > > [ 61.556768] pass = 1, prio = 8, sc.nr_reclaimed = 0 > > [ 61.556778] pass = 1, prio = 7, sc.nr_reclaimed = 0 > > [ 61.556787] pass = 1, prio = 6, sc.nr_reclaimed = 0 > > [ 61.556797] pass = 1, prio = 5, sc.nr_reclaimed = 0 > > [ 61.556806] pass = 1, prio = 4, sc.nr_reclaimed = 0 > > [ 61.556816] pass = 1, prio = 3, sc.nr_reclaimed = 0 > > [ 61.556825] pass = 1, prio = 2, sc.nr_reclaimed = 0 > > [ 61.556835] pass = 1, prio = 1, sc.nr_reclaimed = 0 > > [ 61.595841] pass = 1, prio = 0, sc.nr_reclaimed = 0 > > [ 61.595854] pass = 2, prio = 12, sc.nr_reclaimed = 0 > > [ 61.595864] pass = 2, prio = 11, sc.nr_reclaimed = 0 > > [ 61.595873] pass = 2, prio = 10, sc.nr_reclaimed = 0 > > [ 61.595883] pass = 2, prio = 9, sc.nr_reclaimed = 0 > > [ 61.710044] pass = 2, prio = 8, sc.nr_reclaimed = 2895 > > [ 61.710062] pass = 2, prio = 7, sc.nr_reclaimed = 0 > > [ 61.710072] pass = 2, prio = 6, sc.nr_reclaimed = 0 > > [ 61.710081] pass = 2, prio = 5, sc.nr_reclaimed = 0 > > [ 61.710091] pass = 2, prio = 4, sc.nr_reclaimed = 0 > > [ 61.710100] pass = 2, prio = 3, sc.nr_reclaimed = 0 > > [ 61.710110] pass = 2, prio = 2, sc.nr_reclaimed = 0 > > [ 61.710119] pass = 2, prio = 1, sc.nr_reclaimed = 0 > > [ 61.823378] pass = 2, prio = 0, sc.nr_reclaimed = 1802 > > [ 61.823396] pass = 3, prio = 12, sc.nr_reclaimed = 0 > > [ 61.823406] pass = 3, prio = 11, sc.nr_reclaimed = 0 > > [ 61.823416] pass = 3, prio = 10, sc.nr_reclaimed = 0 > > [ 61.823425] pass = 3, prio = 9, sc.nr_reclaimed = 0 > > [ 61.823435] pass = 3, prio = 8, sc.nr_reclaimed = 0 > > [ 61.823444] pass = 3, prio = 7, sc.nr_reclaimed = 0 > > [ 61.823454] pass = 3, prio = 6, sc.nr_reclaimed = 0 > > [ 61.823464] pass = 3, prio = 5, sc.nr_reclaimed = 0 > > [ 61.823473] pass = 3, prio = 4, sc.nr_reclaimed = 0 > > [ 61.823483] pass = 3, prio = 3, sc.nr_reclaimed = 0 > > [ 61.823492] pass = 3, prio = 2, sc.nr_reclaimed = 0 > > [ 61.823502] pass = 3, prio = 1, sc.nr_reclaimed = 0 > > [ 62.586707] pass = 3, prio = 0, sc.nr_reclaimed = 2716 > > [ 62.608037] pass = 4, prio = 12, sc.nr_reclaimed = 3070 > > [ 62.634644] pass = 4, prio = 11, sc.nr_reclaimed = 5048 > > [ 62.638782] pass = 4, prio = 10, sc.nr_reclaimed = 192 > > [ 62.740052] pass = 4, prio = 9, sc.nr_reclaimed = 0 > > [ 62.843385] pass = 4, prio = 8, sc.nr_reclaimed = 640 > > [ 62.946726] pass = 4, prio = 7, sc.nr_reclaimed = 640 > > [ 63.046711] pass = 4, prio = 6, sc.nr_reclaimed = 640 > > [ 63.146717] pass = 4, prio = 5, sc.nr_reclaimed = 608 > > [ 63.246712] pass = 4, prio = 4, sc.nr_reclaimed = 600 > > [ 63.346704] pass = 4, prio = 3, sc.nr_reclaimed = 128 > > [ 63.446698] pass = 4, prio = 2, sc.nr_reclaimed = 0 > > [ 63.546705] pass = 4, prio = 1, sc.nr_reclaimed = 0 > > [ 63.646698] pass = 4, prio = 0, sc.nr_reclaimed = 0 > > [ 63.646708] after: sc.nr_reclaimed = 0 > > [ 63.646715] shrink_all_memory(10000) failed > > > > which obviously is done by shrink_all_zones(). Sigh. > > > > The appended patch should help, please verify. > > > > Yes, that fixes it. --- From: Rafael J. Wysocki Subject: PM/Hibernate: Fix memory shrinking Commit d979677c4c02f0a72db5a03ecd8184bd9d6695c8 (mm: shrink_all_memory(): use sc.nr_reclaimed) broke the memory shrinking used by hibernation, becuse it did not update shrink_all_zones() in accordance with the other changes it made. Fix this by making shrink_all_zones() update sc->nr_reclaimed instead of overwriting its value. Signed-off-by: Rafael J. Wysocki --- mm/vmscan.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) Index: linux-2.6/mm/vmscan.c =================================================================== --- linux-2.6.orig/mm/vmscan.c +++ linux-2.6/mm/vmscan.c @@ -2088,13 +2088,13 @@ static void shrink_all_zones(unsigned lo nr_reclaimed += shrink_list(l, nr_to_scan, zone, sc, prio); if (nr_reclaimed >= nr_pages) { - sc->nr_reclaimed = nr_reclaimed; + sc->nr_reclaimed += nr_reclaimed; return; } } } } - sc->nr_reclaimed = nr_reclaimed; + sc->nr_reclaimed += nr_reclaimed; } /* @@ -2115,6 +2115,7 @@ unsigned long shrink_all_memory(unsigned .may_unmap = 0, .may_writepage = 1, .isolate_pages = isolate_pages_global, + .nr_reclaimed = 0, }; current->reclaim_state = &reclaim_state; -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/