From: Tejun Heo Subject: Re: [PATCH] libata: Whitelist SSDs that are known to properly return zeroes after TRIM Date: Fri, 5 Dec 2014 09:51:48 -0500 Message-ID: <20141205145148.GI4080@htj.dyndns.org> References: <20141204170611.GB2995@htj.dyndns.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from mail-qc0-f173.google.com ([209.85.216.173]:40565 "EHLO mail-qc0-f173.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751119AbaLEOv7 (ORCPT ); Fri, 5 Dec 2014 09:51:59 -0500 Received: by mail-qc0-f173.google.com with SMTP id i17so540848qcy.32 for ; Fri, 05 Dec 2014 06:51:59 -0800 (PST) Content-Disposition: inline In-Reply-To: Sender: linux-ide-owner@vger.kernel.org List-Id: linux-ide@vger.kernel.org To: "Martin K. Petersen" Cc: linux-ide@vger.kernel.org Hello, On Thu, Dec 04, 2014 at 09:13:22PM -0500, Martin K. Petersen wrote: > Tejun> If there's something on the horizon which would solve the > Tejun> identification problem and we only have to worry about the > Tejun> current batch of devices, whitelisting can be useful but > Tejun> otherwise I'm not sure this is a good idea. > > There isn't :( The only saving grace is that SSDs are gravitating > towards NVMe and other non-ATA interfaces. Yeap, that's true. This whole thing is on the way out. > I've already tightened up things in SCSI so we now prefer WRITE SAME > which does give hard guarantees unlike UNMAP. But because we use WRITE > SAME in the libata SATL it is imperative that we change our internal > flagging to be equally restrictive. Can you please elaborate this in the changelog and add comments explaning why ATA_HORKAGE_ZERO_AFTER_TRIM is necessary and how it's used? > I just feel bad about disabling the feature for the many existing users > (and there are quite a few) that are using well-behaved drives in their > RAID deployments. And the filesystem folks have been begging for the > zeroout discard variant that I posted a few weeks ago. So the users are > there. I'm just trying to accommodate them the best I can given the lame > spec. Can you please explain further the practical gains of using trims which guarantee zeroing? Thanks. -- tejun