From mboxrd@z Thu Jan 1 00:00:00 1970 From: NeilBrown Subject: Re: [PATCH 1/9] MD: add a new disk role to present cache device Date: Wed, 5 Aug 2015 10:25:20 +1000 Message-ID: <20150805102520.78222c47@noble> References: <20150804142824.GA8204@infradead.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: Sender: linux-raid-owner@vger.kernel.org To: Song Liu Cc: Christoph Hellwig , Shaohua Li , "linux-raid@vger.kernel.org" , Kernel Team , "dan.j.williams@intel.com" List-Id: linux-raid.ids On Tue, 4 Aug 2015 18:17:25 +0000 Song Liu wrote: > > -----Original Message----- > > From: Christoph Hellwig [mailto:hch@infradead.org] > > Sent: Tuesday, August 4, 2015 7:28 AM > > To: Shaohua Li > > Cc: linux-raid@vger.kernel.org; Kernel Team; Song Liu; hch@infradead.org; > > dan.j.williams@intel.com; neilb@suse.de > > Subject: Re: [PATCH 1/9] MD: add a new disk role to present cache device > > > > > case 0xfffe: /* faulty */ > > > set_bit(Faulty, &rdev->flags); > > > break; > > > + case 0xfffd: /* cache device */ > > > > Any chance to get constants for these magic numbers as an additional prep > > patch? > I will add patch for special roles (spare, faulty, cache, etc.). > > > > > > Also I don't really think that adding the role without the actual implementation > > is that useful. > > Currently, we are planning to use 0xfffd for both "cache device" and > "journal device" (fix write hole only). Would you prefer to separate these > two scenarios with two different roles (0xfffd and 0xfffc)? No, definitely not. It is a log device. whether it is being used just to close the write hole or more aggressively as a write-ahead cache to reduce latency doesn't change the nature of the data on the device. So just one new device type: "log" or "journal" or something describing what is on the device. Thanks, NeilBrown