From: "Darrick J. Wong" Subject: semi-stable page writes Date: Fri, 26 Oct 2012 03:19:09 -0700 Message-ID: <20121026101909.GB19617@blackbox.djwong.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-ext4 , linux-fsdevel To: "Theodore Ts'o" Return-path: Received: from rcsinet15.oracle.com ([148.87.113.117]:46525 "EHLO rcsinet15.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755406Ab2JZKTO (ORCPT ); Fri, 26 Oct 2012 06:19:14 -0400 Content-Disposition: inline Sender: linux-ext4-owner@vger.kernel.org List-ID: Hi everyone, Are people still annoyed about writes taking unexpectedly long amounts of tme due to the stable page write patchset? I'm guessing yes... I'm close to posting a patchset that (a) gates the wait_on_page_writeback calls on a flag that you can set in the bdi to indicate that you need stable writes (which blk_integrity_register will set); (b) (ab)uses a page flag bit (PG_slab) to indicate that a page is actually being sent out to disk hardware; and (c) changes the three calls to wait_on_page_writeback()s that were in the original patchset to only wait if the bit in (b) is set. Sort of a hack, but it'll fix some of the latency complaints. I guess I should let that run overnight, and clean up/mail out the set tomorrow. --D