From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from [213.239.205.147] (helo=debian.tglx.de) by canuck.infradead.org with esmtp (Exim 4.33 #1 (Red Hat Linux)) id 1BjF2F-00031R-4P for linux-mtd@lists.infradead.org; Sat, 10 Jul 2004 06:24:47 -0400 From: Thomas Gleixner To: Brad Campbell In-Reply-To: <40EFB906.7060701@wasp.net.au> References: <200407091608.14442.philt@pioneer-pdt.com> <40EFB906.7060701@wasp.net.au> Content-Type: text/plain Message-Id: <1089454689.3106.3.camel@thomas.tec.linutronix.de> Mime-Version: 1.0 Date: Sat, 10 Jul 2004 12:18:10 +0200 Content-Transfer-Encoding: 7bit Cc: Patrick Hilt , linux-mtd@lists.infradead.org Subject: Re: JFFS 2 Question Reply-To: tglx@linutronix.de List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Sat, 2004-07-10 at 11:38, Brad Campbell wrote: > Patrick Hilt wrote: > > Hi list! > > Is there a way to make sure data is being written to physical flash when using > > JFFS2? In other words, After writing a file on a JFFS2 partition, is there > > anything that can be done to make sure that the data is getting written to > > the flash right away? Would "sync" or "fsync" have that effect? > > I believe that jffs2 is a synchronous filesystem in any case. By the time your application has > returned from the write() call, the data is on flash. You're right and wrong. :) On standard NOR FLASH it is synchronous. On NAND/AND FLASH and some strange NOR FLASH types we must write page aligned and therefor were use a write buffer of page size, which is either flushed timed by kupdated or by sync/fsync/ummount. tglx