From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from smtp.nokia.com ([192.100.105.134] helo=mgw-mx09.nokia.com) by bombadil.infradead.org with esmtps (Exim 4.69 #1 (Red Hat Linux)) id 1MRLTJ-0002sl-2d for linux-mtd@lists.infradead.org; Thu, 16 Jul 2009 07:33:44 +0000 Subject: RE: UBIFS Corrupt during power failure From: Artem Bityutskiy To: Eric Holmberg In-Reply-To: References: <1246627771.20721.191.camel@localhost.localdomain> <7207AAC68CE347458026863515A07DA102901F3C@usw-am-xch-02.am.trimblecorp.net> <1246629940.20721.219.camel@localhost.localdomain> <7207AAC68CE347458026863515A07DA102901F9C@usw-am-xch-02.am.trimblecorp.net> <1246633131.20721.224.camel@localhost.localdomain> <1246854654.20721.271.camel@localhost.localdomain> <1246855913.20721.287.camel@localhost.localdomain> <1246862635.20721.291.camel@localhost.localdomain> <1246949184.20721.302.camel@localhost.localdomain> <20090715205217.GH3056@shareable.org> Content-Type: text/plain; charset="UTF-8" Date: Thu, 16 Jul 2009 10:33:07 +0300 Message-Id: <1247729587.11353.80.camel@localhost.localdomain> Mime-Version: 1.0 Content-Transfer-Encoding: 8bit Cc: Jamie Lokier , linux-mtd@lists.infradead.org, Urs Muff , Stefan Roese , Nicolas Pitre , Adrian Hunter Reply-To: dedekind@infradead.org List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wed, 2009-07-15 at 15:35 -0600, Eric Holmberg wrote: > Test Results - PASS > ------------------- > I just finished a test run of 3 boards with a total of 8104 power cycles > during read/modify/write operations on UBIFS. The system cuts power > randomly after mounting the UBIFS image while performing read/write file > operations on the UBIFS root file system. > > I experienced 0 failures. Typically, I was getting less than 100 cycles > before failures. > > Software: > * 2.6.27-rc9 kernel > * UBIFS patches up to July 10 > * CFI Write buffer size limited to 8 bytes (NOR chip supports 32 bytes) > * No power cycling during mount > > Patch: > http://git.infradead.org/users/dedekind/ubifs-v2.6.27.git?a=commit;h=201 > fce21a67d74e9703127af0d6c6084003796bb I'm also running a power cut tests. But I do not have any buffer size limiting patch. I'm running the 'fsstress' program from the LTP test suite with the following parameters: fsstress -d /mnt/ubifs -l0 -r -n1000 -p2 And I cut power randomly after 0-300 seconds after the start. So far so good. > Remaining tasks: > * Anybody: Add support to UBI/UBIFS to allow for block-write sizes > greater than 8 bytes Yeah, this is in my TODO. I just wanted to first reproduce this issue, then trying to fix it. Cannot reproduce it so far. > * Test with power failures during deferred recovery Good thing to test. I did not test this. Just in case: I assume by "deferred" recovery you mean the situation when you mount the FS which needs recovery first R/O, so the recovery is deferred, and it happens when you re-mount it R/W, right? > Thanks for everybody's help so far! I will hopefully get a chance to > setup the power-failures during deferred recovery this weekend. Thanks for cooperation. We are interested to make UBIFS robust. -- Best regards, Artem Bityutskiy (Битюцкий Артём)