From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from a.ns.miles-group.at ([95.130.255.143] helo=radon.swed.at) by bombadil.infradead.org with esmtps (Exim 4.80.1 #2 (Red Hat Linux)) id 1aAvCb-0007DH-16 for linux-mtd@lists.infradead.org; Mon, 21 Dec 2015 07:48:18 +0000 Subject: Re: Corrupt Empty Space Error at Runtime To: Sheng Yong , Adam References: <56775906.2040506@huawei.com> Cc: "linux-mtd@lists.infradead.org" From: Richard Weinberger Message-ID: <5677AEA3.7010208@nod.at> Date: Mon, 21 Dec 2015 08:47:47 +0100 MIME-Version: 1.0 In-Reply-To: <56775906.2040506@huawei.com> Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Hi! Am 21.12.2015 um 02:42 schrieb Sheng Yong: > We have met several empty space corruptions these days, since the ECC > functionality of the NAND controller driver seems not work correctly. > But we are still considering if there is any workaroud to let UBIFS > check if the corruption occurs really in empty space. If it is, UBIFS > should recover the LEB. > > There are 2 conditions we may check: > 1. the left space size is less than the min size of a node, it must be > empty space; > 2. how many bits are fliped in left space, if they are less than 4 bits > (many NAND support 1~4 bits ECC), it should be in empty space; Please consider fixing the driver first. We also have patches for gpmi. Maybe we can find a common way to fix this clearly within MTD without weakening UBIFS. Thanks, //richard