From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: [PATCH 1/5][RFC][CFT] percpu fixes, part 1 Date: Thu, 6 Mar 2014 15:47:26 -0500 Message-ID: <20140306204726.GE14033@htj.dyndns.org> References: <20140305034751.GW18016@ZenIV.linux.org.uk> <20140305034919.GA26528@ZenIV.linux.org.uk> <20140306192026.GA14033@htj.dyndns.org> <20140306203030.GA18016@ZenIV.linux.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: linux-fsdevel@vger.kernel.org, Linus Torvalds , linux-kernel@vger.kernel.org, Stephen Tweedie , Jeremy Eder To: Al Viro Return-path: Received: from mail-yh0-f45.google.com ([209.85.213.45]:39953 "EHLO mail-yh0-f45.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751986AbaCFUra (ORCPT ); Thu, 6 Mar 2014 15:47:30 -0500 Content-Disposition: inline In-Reply-To: <20140306203030.GA18016@ZenIV.linux.org.uk> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: Hello, Al. On Thu, Mar 06, 2014 at 08:30:30PM +0000, Al Viro wrote: > > Also, I think it'd be better to split addition of first_free hint to a > > separate patch. > > OK, but I'm not sure how much does it simplify things, actually. Not much, but it should at least help bisection if something goes wrong, I think. > > A comment explaining why ignoring the free bit during bin search is > > okay would be nice? > > Huh? We are not ignoring it - we are searching for exact value, including > the lower bit being set. It might be worth adding a comment next to > "freeme |= 1;" before the loop, but that's it. These two BUG_ON() fold Oh yeah, nothing too much. I was just thinking about noting that looking for the exact match is enough as we're looking for the matching busy slot and the bit itself can't change ordering between entries. > > > + if (unlikely(align < 2)) > > > + align = 2; > > > > Please add a comment explaining why the above min alignment is > > necessary. > > Umm... Will "we want the lowest bit of offset available for free/in_use > indicator" do? Yeap. Thanks! -- tejun