From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Subject: Re: Windows and TRIM support - files, not devices References: <99e2bdcb-3067-cb48-90c8-c141191465d5@cran.org.uk> <40890e19-4484-2265-8b24-ef98b6f61717@bluestop.org> From: Jens Axboe Message-ID: <9495a910-beb6-76dd-2c1e-3f67c23ef59f@kernel.dk> Date: Mon, 30 Apr 2018 14:24:21 -0600 MIME-Version: 1.0 In-Reply-To: <40890e19-4484-2265-8b24-ef98b6f61717@bluestop.org> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit To: Rebecca Cran , Sitsofe Wheeler Cc: fio List-ID: On 4/20/18 9:46 AM, Rebecca Cran wrote: > On 04/20/18 01:15, Sitsofe Wheeler wrote: > >> On 28 January 2017 at 01:05, Rebecca Cran wrote: >>> I've been looking at the TRIM support in Windows 8 and newer, and have come >>> across a potential problem. Other OSes use TRIM on a physical/block device, >>> but Windows only provides support for TRIM on files. >> >> How far did you get with this? > > I didn't do any more work on it after realizing it only worked on files. Missed this originally - files is totally fine, for what it's worth, on Linux we could map this to hole punching too. So I don't think that's a show stopper at all, it'd be perfectly possible to define trim based workloads just on files. In fact, that's more flexible that what we currently support on Linux. -- Jens Axboe