From: Andrew Perepechko Subject: page eviction from the buddy cache Date: Mon, 25 Mar 2013 16:59:44 +0400 Message-ID: <51504A40.6020604@ya.ru> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: alexey.lyashkov@gmail.com To: linux-ext4@vger.kernel.org Return-path: Received: from forward14.mail.yandex.net ([95.108.130.92]:38313 "EHLO forward14.mail.yandex.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755409Ab3CYNF5 (ORCPT ); Mon, 25 Mar 2013 09:05:57 -0400 Sender: linux-ext4-owner@vger.kernel.org List-ID: Hello! Our recent investigation has found that pages from the buddy cache are evicted too often as compared to the expectation from their usage pattern. This introduces additional reads during large writes under our workload and really hurts overall performance. ext4 uses find_get_page() and find_or_create_page() to look for buddy cache pages, but these pages don't get a chance to become activated until the following lru_add_drain() call, because mark_page_accessed() does not activate pages which are not PageLRU(). As can be found from a kprobe-based test, these pages are often moved on the inactive LRU as a result of shrink_inactive_list()->lru_add_drain() and immediately evicted. From a quick look into linux-2.6.git, the issue seems to exist in the current code as well. A possible and, perhaps, non-optimal solution would be to call lru_add_drain() each time a buddy cache page is used. Any other suggestions? Thank you, Andrew