From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from fed1rmmtao06.cox.net ([68.230.241.33]) by canuck.infradead.org with esmtp (Exim 4.43 #1 (Red Hat Linux)) id 1Cy0Zz-00022r-42 for linux-mtd@lists.infradead.org; Sun, 06 Feb 2005 23:32:56 -0500 Received: from laptop ([70.179.62.151]) by fed1rmmtao06.cox.net (InterMail vM.6.01.04.00 201-2131-117-20041022) with ESMTP id <20050207043249.EQAI5874.fed1rmmtao06.cox.net@laptop> for ; Sun, 6 Feb 2005 23:32:49 -0500 Content-Type: text/plain; charset="iso-8859-1" From: Rick Adams To: linux-mtd@lists.infradead.org Date: Sun, 6 Feb 2005 20:32:05 -0800 References: In-Reply-To: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Message-Id: <200502062032.05112.rgadams@cox.net> Subject: Re: ST flashes List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Thursday 03 February 2005 02:56 am, linux-mtd-request@lists.infradead.= org=20 wrote: >Im using 2.4.19-rmk7-pxa1 ... i know its really old and all=20 >that ... but i have no options right now... >but anyway.. I have two pxa 255 boards , one with ST flash and one with AMD. >I configured them to use the lubbock flash driver , and made one=20 >partition for jffs2. >But on the ST , the moment you power down the board ! , the next time ,=20 >it only mounts , but doesnt >let me perform any fops. I keep getting I/O errors..for only write=20 >operations... read works , coz i could ls in the partition. >It also doesnt let me format the partition. Hi Ashwin What's the part number of the Intel flash, is it a J or K version? The pa= rt=20 number will look like 28FnnnJ3 where nnn is the size or 28FnnnKvv where n= nn=20 is size an vv is voltage. These are the part numbers for the TSOP package= ,=20 for the BGA package the first part will be different but will have the J = or K=20 designation. The J or K designation is the only thing that matters for th= is=20 discussion the rest of the number isn't important. Intel is obsoleting th= e J=20 part and replacing it with the K part. The version of the kernel and the=20 Lubbock board were built with J parts (before the K was around). The K is= =20 "almost" identical to the J. One important difference is the the K part h= as=20 all it's sectors locked at power up and the J has all it's sectors unlock= ed.=20 If you have a driver written for the J and a board with a K part then thi= s=20 could lead to the exact problem you talked about. If this is your situation there are two possible fixes. One is to get a d= river=20 that understands how to talk to a K, or modify the boot or initialization= =20 code to unlock all the sector of the K at power up, and then it will func= tion=20 exactly like a J. The other main feature the K has is the ability to burs= t=20 read. This will improve some flash operations if the processor supports D= MA=20 from flash and you modify the boot or initialization code to enable this=20 mode. I know the ARM9 core supports it, I don't know offhand if the xScal= e=20 does. Anyhow the point is that the J driver lack of burst mode support on= ly=20 has a performance implication and no functional one. The easiest thing is= to=20 just unlock all flash sectors of the K at power up and use the J driver. If this isn't your situation (you don't have K parts) then never mind. --=20 Rick Adams rick@theptrgroup.com