From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mf2.realtek.com.tw ([220.128.56.22]) by canuck.infradead.org with esmtp (Exim 4.54 #1 (Red Hat Linux)) id 1EZPwV-0007nD-Lf for linux-mtd@lists.infradead.org; Tue, 08 Nov 2005 04:39:08 -0500 Received: from msx.realtek.com.tw (unverified [172.21.1.77]) by mf2.realtek.com.tw (Clearswift SMTPRS 5.1.4) with ESMTP id for ; Tue, 8 Nov 2005 16:38:33 +0800 Message-ID: <010201c5e43f$796bd5f0$135415ac@realtek.com.tw> From: =?big5?B?s6+s/bN4IFwoQ2h1Y2tcKQ==?= To: Date: Tue, 8 Nov 2005 16:36:12 +0800 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_00FF_01C5E482.8785A010" Subject: CRC error of jffs2_get_inode_nodes() for JFFS2 on NAND List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , This is a multi-part message in MIME format. ------=_NextPart_000_00FF_01C5E482.8785A010 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="big5" Hello all, Our system use JFFS2 on NAND flash, and code base is linux 2.6.10-rc3. A few days ago, we dived into the JFFS2 for the CRC error after a system = reboot. We found that it is caused by an incomplete writing and it always = happens in the end of file.=20 If we write some data into flash which size is not the multiple of = page_size (512-byte),=20 there is a remainder (under a page-size) in 'wbuf' of function = jffs2_flash_writev(). Then, at this time, we reset the system, JFFS2 will not yet write the = remainder into flash and it will cause the inode incomplete. Is it normal ? What time will JFFS2 flush wbuf out to sync the data if = there is a remainder in wbuf ? I appreciate if anyone can give us some suggestion. best regards, Chuck Chen = ------=_NextPart_000_00FF_01C5E482.8785A010 Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset="big5"

Hello all,

Our system use JFFS2 on NAND flash, and code base is linux=20 2.6.10-rc3.

A few days ago, we dived into the JFFS2 for the CRC error = after a=20 system reboot.

We found that it is caused by an incomplete writing = and it=20 always happens in the end of file.

If we write some data into flash which size is not the multiple of=20 page_size (512-byte),

there is a remainder (under a page-size) in 'wbuf' of function=20 jffs2_flash_writev().

Then, at this time, we reset the system, JFFS2 will not yet = write the=20 remainder into flash and it will cause the inode incomplete.

Is it normal ? What time will JFFS2 flush wbuf out to = sync the=20 data if there is a remainder in wbuf ?

I appreciate if anyone can give us some suggestion.

best regards,

Chuck=20 Chen &nb= sp; &nbs= p; = ; = &= nbsp; &n= bsp; &nb= sp; &nbs= p; = ; =20

------=_NextPart_000_00FF_01C5E482.8785A010--