From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from verein.lst.de ([213.95.11.211]:48763 "EHLO newverein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727383AbeLST30 (ORCPT ); Wed, 19 Dec 2018 14:29:26 -0500 Date: Wed, 19 Dec 2018 20:29:24 +0100 From: Christoph Hellwig Subject: Re: [PATCH 06/11] xfs: don't use delalloc extents for COW on files with extsize hints Message-ID: <20181219192924.GA28624@lst.de> References: <20181203222503.30649-1-hch@lst.de> <20181203222503.30649-7-hch@lst.de> <20181218214425.GL27208@magnolia> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20181218214425.GL27208@magnolia> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: "Darrick J. Wong" Cc: Christoph Hellwig , linux-xfs@vger.kernel.org On Tue, Dec 18, 2018 at 01:44:25PM -0800, Darrick J. Wong wrote: > > + uint *lockmode, > > + unsigned flags) > > I'm not thrilled with passing iomap flags into the reflink code here... So what is the alternative? A non-descriptive bool argument? > ...because I feel that it's easy to miss the subtlety here that for > buffered writes we don't care if the cow extent is unwritten or written, > but for directio we very /much/ care that the cow extent is written, > because we're writing to it immediately. Can this grow a comment to > reinforce why we skip the conversion? Sure. > Also, can we call this 'iomap_flags' to make it clearer which flags > we're talking about? Sure.