From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [195.209.228.254] (helo=shelob.oktetlabs.ru) by canuck.infradead.org with esmtps (Exim 4.54 #1 (Red Hat Linux)) id 1EgMSd-0008JA-RQ for linux-mtd@lists.infradead.org; Sun, 27 Nov 2005 08:20:58 -0500 Message-ID: <4389B293.3020507@yandex.ru> Date: Sun, 27 Nov 2005 16:20:19 +0300 From: "Artem B. Bityutskiy" MIME-Version: 1.0 To: Ferenc Havasi References: <8126E4F969BA254AB43EA03C59F44E840C3738@pdsmsx404> <4386FC5F.9050305@yandex.ru> <4389AE3D.8000807@inf.u-szeged.hu> In-Reply-To: <4389AE3D.8000807@inf.u-szeged.hu> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Cc: linux-mtd@lists.infradead.org, "Zhao, Forrest" Subject: Re: The problem that I didn't think out List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Ferenc, Greetings! > We had some time, so we read the plan of JFFS3 (with RaiserFS > documentation). Oh, what a delightful news! :-) > The key compression is the only one in the plan that I think that is > better if we don't use. I think to make key management as simple and > fast as possible is more important than some percent in flash usage. (if > I am right the bottleneck of real products is not the flash usage but > the speed) But it is small techniqual question. Well, I would not agree, compression Guru Ferenc though you be :-) When you start thinking about GC, you'll notice that Indexing nodes are to be rewritten *a great deal* of times. So, the smaller is the index, the faster is JFFS3. The main idea of key compression is *not* to save flash space, but to lessen the (index size)/(data size) ratio. Nonetheless, tests should show the worthiness of keys compression. There is obviously a (CPU time) vs. (amount of flash IO) trade-off. Thus, I offer to wait for a JFFS3 prototype and evaluate this. Let's mark this stuff as "to be evaluated". > The big questions are that questions you already thinging on: garbage > collection and wear leveling. Without solving them JFFS3 have no nice > future. I only would like to say that now we are also thinking on these > important problems... and we will write if anything usable found. (I may > be better feeling to thinking on someting not alone... :) ) Great! :-) Thank you for this feedback. -- Best Regards, Artem B. Bityutskiy, St.-Petersburg, Russia.