From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-pf0-f199.google.com (mail-pf0-f199.google.com [209.85.192.199]) by kanga.kvack.org (Postfix) with ESMTP id A6BC06B025F for ; Wed, 16 Aug 2017 00:48:03 -0400 (EDT) Received: by mail-pf0-f199.google.com with SMTP id j83so3145456pfe.10 for ; Tue, 15 Aug 2017 21:48:03 -0700 (PDT) Received: from lgeamrelo13.lge.com (LGEAMRELO13.lge.com. [156.147.23.53]) by mx.google.com with ESMTP id d22si3642077pli.530.2017.08.15.21.48.01 for ; Tue, 15 Aug 2017 21:48:01 -0700 (PDT) Date: Wed, 16 Aug 2017 13:48:00 +0900 From: Minchan Kim Subject: Re: [PATCH v1 2/6] fs: use on-stack-bio if backing device has BDI_CAP_SYNC capability Message-ID: <20170816044759.GC24294@blaptop> References: <20170811104615.GA14397@lst.de> <20c5b30a-b787-1f46-f997-7542a87033f8@kernel.dk> <20170814085042.GG26913@bbox> <51f7472a-977b-be69-2688-48f2a0fa6fb3@kernel.dk> <20170814150620.GA12657@bgram> <51893dc5-05a3-629a-3b88-ecd8e25165d0@kernel.dk> <20170814153059.GA13497@bgram> <0c83e7af-10a4-3462-bb4c-4254adcf6f7a@kernel.dk> <058b4ae5-c6e9-ff32-6440-fb1e1b85b6fd@kernel.dk> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <058b4ae5-c6e9-ff32-6440-fb1e1b85b6fd@kernel.dk> Sender: owner-linux-mm@kvack.org List-ID: To: Jens Axboe Cc: Christoph Hellwig , Dan Williams , Matthew Wilcox , Andrew Morton , Linux Kernel Mailing List , linux-mm , Ross Zwisler , "karam . lee" , seungho1.park@lge.com, Dave Chinner , Jan Kara , Vishal Verma , "linux-nvdimm@lists.01.org" , kernel-team Hi Jens, On Mon, Aug 14, 2017 at 10:17:09AM -0600, Jens Axboe wrote: > On 08/14/2017 09:38 AM, Jens Axboe wrote: > > On 08/14/2017 09:31 AM, Minchan Kim wrote: > >>> Secondly, generally you don't have slow devices and fast devices > >>> intermingled when running workloads. That's the rare case. > >> > >> Not true. zRam is really popular swap for embedded devices where > >> one of low cost product has a really poor slow nand compared to > >> lz4/lzo [de]comression. > > > > I guess that's true for some cases. But as I said earlier, the recycling > > really doesn't care about this at all. They can happily coexist, and not > > step on each others toes. > > Dusted it off, result is here against -rc5: > > http://git.kernel.dk/cgit/linux-block/log/?h=cpu-alloc-cache > > I'd like to split the amount of units we cache and the amount of units > we free, right now they are both CPU_ALLOC_CACHE_SIZE. This means that > once we hit that count, we free all of the, and then store the one we > were asked to free. That always keeps 1 local, but maybe it'd make more > sense to cache just free CPU_ALLOC_CACHE_SIZE/2 (or something like that) > so that we retain more than 1 per cpu in case and app preempts when > sleeping for IO and the new task on that CPU then issues IO as well. > Probably minor. > > Ran a quick test on nullb0 with 32 sync readers. The test was O_DIRECT > on the block device, so I disabled the __blkdev_direct_IO_simple() > bypass. With the above branch, we get ~18.0M IOPS, and without we get > ~14M IOPS. Both ran with iostats disabled, to avoid any interference > from that. Looks promising. If recycling bio works well enough, I think we don't need to introduce new split in the path for on-stack bio. I will test your version on zram-swap! Thanks. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org