From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from webbox1416.server-home.net ([77.236.96.61]) by bombadil.infradead.org with esmtp (Exim 4.80.1 #2 (Red Hat Linux)) id 1ab6Q1-0003J9-Hi for linux-mtd@lists.infradead.org; Wed, 02 Mar 2016 13:02:22 +0000 From: Alexander Stein To: Richard Weinberger Cc: David Woodhouse , Brian Norris , "linux-mtd@lists.infradead.org" Subject: Re: [PATCH 1/1] mtd: mtdram: Add parameter for setting writebuf size Date: Wed, 02 Mar 2016 14:01:50 +0100 Message-ID: <6654580.lx3DHdo0Ov@ws-stein> In-Reply-To: <56D6E0F6.8070905@nod.at> References: <1456911954-12929-1-git-send-email-alexander.stein@systec-electronic.com> <2781096.oJXdWOB4df@ws-stein> <56D6E0F6.8070905@nod.at> MIME-Version: 1.0 Content-Transfer-Encoding: 7Bit Content-Type: text/plain; charset="us-ascii" List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Wednesday 02 March 2016 13:47:50, Richard Weinberger wrote: > Am 02.03.2016 um 11:41 schrieb Alexander Stein: > > On Wednesday 02 March 2016 11:23:59, Richard Weinberger wrote: > >> On Wed, Mar 2, 2016 at 10:45 AM, Alexander Stein > >> wrote: > >>> ubifs uses the write buffer size in recovery algorithm. When inspecting > >>> an unclean ubifs recovery fails with writebuf size 64 in mtdram while > >>> recovery on actual mtd device with writebuf size of 1024 succeeds. > >>> So add a parameter for setting this property. > >> > >> Can it be that you've tested an NAND image on mtdram? > > > > Nope. We copied that image within barebox from device /dev/nor0 (the whole 128MiB NOR flash) and used that in mtdram. > > Unfortunately mounting that "broken" ubifs in barebox suffers from essentially the same problem: using a different writebuf size for recovery results in failure. But that's another issue. > > The apparently important changes in linux are the commits: > > 428ff9d2e37d3a82af0f56b476f70c244cf550d1 ("UBIFS: remove dead code") > > 2765df7da540687c4d57ca840182122f074c5b9c ("UBIFS: use max_write_size during recovery") > > With those two the writebuf size gets used by ubifs. > > Makes sense to me! > One minor nitpick, the comment "Mimic CFI NOR flashes" is no longer valid then. > Maybe you can note in Kconfig that the default value 64 mimics CFI NOR... Well, I stumbled also on that because we actually do use a CFI NOR flash. The commit message introducing that states: "Set the 'mtd->writebufsize' field to 64 to mimic modern CFI flashes." I guess 64 Bytes was modern at that time. I'll remove it all together in v2. > Beside of that: > Reviewed-by: Richard Weinberger Thanks, best regards, Alexander -- Dipl.-Inf. Alexander Stein SYS TEC electronic GmbH alexander.stein@systec-electronic.com Legal and Commercial Address: Am Windrad 2 08468 Heinsdorfergrund Germany Office: +49 (0) 3765 38600-0 Fax: +49 (0) 3765 38600-4100 Managing Directors: Director Technology/CEO: Dipl.-Phys. Siegmar Schmidt; Director Commercial Affairs/COO: Dipl. Ing. (FH) Armin von Collrepp Commercial Registry: Amtsgericht Chemnitz, HRB 28082; USt.-Id Nr. DE150534010