From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751791AbdBBTUV (ORCPT ); Thu, 2 Feb 2017 14:20:21 -0500 Received: from gum.cmpxchg.org ([85.214.110.215]:41128 "EHLO gum.cmpxchg.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751672AbdBBTUU (ORCPT ); Thu, 2 Feb 2017 14:20:20 -0500 From: Johannes Weiner To: Andrew Morton Cc: Mel Gorman , Michal Hocko , Minchan Kim , Rik van Riel , Hillf Danton , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: [PATCH 0/7] mm: vmscan: fix kswapd writeback regression v2 Date: Thu, 2 Feb 2017 14:19:50 -0500 Message-Id: <20170202191957.22872-1-hannes@cmpxchg.org> X-Mailer: git-send-email 2.11.0 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi Andrew, here are some minor updates to the series. It's nothing functional, just code comments and updates to the changelogs from the mailing list discussions. Since we don't have a good delta system for changelogs I'm resending the entire thing as a drop-in replacement for -mm. These are the changes: 1. mm: vmscan: scan dirty pages even in laptop mode Mel tested the entire series, not just one patch. Move his test conclusions from 'mm: vmscan: remove old flusher wakeup from direct reclaim' into the series header in patch 1. Also, reflect the fact that these test results are indeed Mel's, not mine. 2. mm: vmscan: kick flushers when we encounter dirty pages on the LRU Mention the trade-off between flush-the-world/flush-the-scanwindow type wakeups in the changelog, as per the mailing list discussion. 3. mm: vmscan: move dirty pages out of the way until they're flushed Correct the last paragraph in the changelog. We're not activating dirty/writeback pages after they have rotated twice; they are being activated straight away to get them out of the reclaimer's face. This was a vestige from an earlier version of the patch. 4. mm: vmscan: move dirty pages out of the way until they're flushed fix Code comment fixlet to explain why we activate dirty/writeback pages. Thanks! include/linux/mm_inline.h | 7 ++++ include/linux/mmzone.h | 2 - include/linux/writeback.h | 2 +- include/trace/events/writeback.h | 2 +- mm/swap.c | 9 +++-- mm/vmscan.c | 77 ++++++++++++++++++-------------------- 6 files changed, 50 insertions(+), 49 deletions(-)