From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from ptb-relay02.plus.net ([212.159.14.213]) by canuck.infradead.org with esmtps (Exim 4.42 #1 (Red Hat Linux)) id 1Cf3T2-0006WL-EA for linux-mtd@lists.infradead.org; Thu, 16 Dec 2004 16:47:26 -0500 From: "Gareth Bult (Encryptec)" To: Josh Boyer In-Reply-To: <41C1FDFF.3020308@us.ibm.com> References: <1103152743.15913.33.camel@squizzey.bult.co.uk> <41C0D3AC.3050801@us.ibm.com> <1103158974.15929.37.camel@squizzey.bult.co.uk> <1103201631.25765.7.camel@weaponx.rchland.ibm.com> <1103232128.15927.70.camel@squizzey.bult.co.uk> <41C1FDFF.3020308@us.ibm.com> Content-Type: text/plain Date: Thu, 16 Dec 2004 21:47:49 +0000 Message-Id: <1103233669.15929.81.camel@squizzey.bult.co.uk> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Cc: Linux MTD Subject: Re: JFFS2 mount time List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , If you have any ideas or patches, feel free to submit them to the list. > That's what JFFS3 was cloned for, and there isn't too much concern > with backwards compatibility. Mmm, blkmtd.c is pretty much a rewrite, less than half the size and uses the system buffer cache instead of reading / writing directly to the device. This boosts performance x 3 on read x lots on write. [I get x 100 faster] (albeit you lose bad block detection on write) If anyone's interested in the code, let me know and I'll clean it up and post it .. however it'll be a new file and not a patch. I've only tested it on USB key flash, but I'm assuming it'll be similar for any block device / hard disk / whatever .. Gareth. On Thu, 2004-12-16 at 15:28 -0600, Josh Boyer wrote: > Gareth Bult (Encryptec) wrote: > > Hi, > > > >>There is one more feature pending that I know of. Ferenc Havasi has > >>some eraseblock summary changes that should help. I believe the > >>intention was to commit that to JFFS3, but I haven't seen it go in yet. > > > > > > Ok, I have these and will be trying them presently .. > > > > just wondered if there was any way I could avoid running "patch" .. :) > > Bug Ferenc to commit ;). I'm not sure if he subscribes to this list or not. > > Also, Artem Bityuckiy is working on an inode checkpoint feature for > JFFS3. Once he gets it working, it should increase overall performance. > > If you have any ideas or patches, feel free to submit them to the list. > That's what JFFS3 was cloned for, and there isn't too much concern > with backwards compatibility. > > josh