From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.nokia.com ([192.100.122.230] helo=mgw-mx03.nokia.com) by bombadil.infradead.org with esmtps (Exim 4.69 #1 (Red Hat Linux)) id 1LMf1X-0007EV-Nv for linux-mtd@lists.infradead.org; Tue, 13 Jan 2009 08:53:26 +0000 Subject: Re: Some problem in real application From: Artem Bityutskiy To: Jacky Lam In-Reply-To: <496B05EF.7060401@gsl.com.hk> References: <496B05EF.7060401@gsl.com.hk> Content-Type: text/plain; charset="UTF-8" Date: Tue, 13 Jan 2009 10:53:28 +0200 Message-Id: <1231836808.5973.11.camel@localhost.localdomain> Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Cc: linux-mtd@lists.infradead.org Reply-To: dedekind@infradead.org List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hi, Theoretically, I cannot respond you because of the silly disclamer you added to the end of your e-mail. However, On Mon, 2009-01-12 at 16:57 +0800, Jacky Lam wrote: > I use NFTL before. While in production, we use some self-made IO > programmer to program the NAND flash. We first write the data and ecc > (in spare area), and then read then back. After verify it with ECC, we > can tell whether the data is written correctly or not. Fine. Just do the same with UBI images. > However, in UBI, it doesn't use spare area and embedded the ECC > information in main area. UBI itself does not use OOB. But the NAND driver or NAND controller may use it to store ECC codes. > Will it be difficult to extract the ECC > information from the read back data in order to do verification? The ECC verification is done in the driver level. UBI is above the driver - it asks the driver to read and write NAND pages. And the driver verifies/calculates ECC codes. I'm going to update UBI documentation soon and write a little bit more about what the UBI-aware flasher should do. I'll send you a link later. -- Best regards, Artem Bityutskiy (Битюцкий Артём)