From mboxrd@z Thu Jan 1 00:00:00 1970 From: Shaohua Li Subject: Re: [PATCH v3 5/9] raid5-ppl: Partial Parity Log write logging implementation Date: Thu, 9 Feb 2017 09:13:14 -0800 Message-ID: <20170209171314.qrmdltnk2tzaooex@kernel.org> References: <20170130185953.30428-1-artur.paszkiewicz@intel.com> <20170130185953.30428-6-artur.paszkiewicz@intel.com> <20170207214225.3jenvleiz3tldzlx@kernel.org> <43b52cb7-b9d6-f615-d7c0-d78fdbb9a0cd@intel.com> <20170208053444.b4yfmahmlrjgvv7x@kernel.org> <589C9368.1020602@youngman.org.uk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <589C9368.1020602@youngman.org.uk> Sender: linux-raid-owner@vger.kernel.org To: Wols Lists Cc: Artur Paszkiewicz , linux-raid@vger.kernel.org, shli@fb.com, neilb@suse.com, jes.sorensen@gmail.com List-Id: linux-raid.ids On Thu, Feb 09, 2017 at 04:06:00PM +0000, Wols Lists wrote: > On 08/02/17 05:34, Shaohua Li wrote: > > so it's the implementation which doesn't flush disk cache before writting new > > data to disks, right? That makes sense then. I think we should fix this soon. > > Don't think people will (or remember to) disable write-back cache before using > > ppl. > > Dunno whether I'm making a fool of myself :-) but might a quick-n-dirty > workaround be to make enabling PPL also disable write-back cache? With > maybe a warning not to switch it back on? We don't have API for this in kernel side. This doesn't mean we can't add it but since the limitation is in the implementation, we'd better fix the implementation. Thanks, Shaohua