linux-kernel - Re: [PATCH] mtd: denali: Disable sub-page writes in Denali NAND driver

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 5 Feb 2015 10:18:16 -0600
From:	Graham Moore <grmoore@...nsource.altera.com>
To:	Richard Weinberger <richard.weinberger@...il.com>
CC:	Dinh Nguyen <dinh.linux@...il.com>,
	Dinh Nguyen <dinguyen@...nsource.altera.com>,
	Brian Norris <computersforpeace@...il.com>,
	David Woodhouse <dwmw2@...radead.org>,
	<yamada.m@...panasonic.com>, Josh Triplett <josh@...htriplett.org>,
	"linux-mtd@...ts.infradead.org" <linux-mtd@...ts.infradead.org>,
	Linux List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] mtd: denali: Disable sub-page writes in Denali NAND driver

On 02/03/2015 04:23 PM, Richard Weinberger wrote:
...

>
> So this driver never worked?
> If it worked for some users we have to make sure that your change does not break
> existing setups.
> /me thinks of UBI.
>

Actually, we made this change to make UBIFS work.  So, yes, the driver 
never worked for UBI.  Worked fine for JFFS2, raw data.

A customer reported an issue with ECC errors when using UBIFS on NAND 
flash with Altera SoC.

We debugged it and found the ECC errors occur because the UBI subsystem 
is trying to write sub-pages in the NAND, but neither the NAND chip 
itself nor the Denali NAND controller support sub-page writes.

In more detail, the UBIFS tries to write a sub-page, but the controller 
writes a whole page instead.  The controller's buffer is initialized to 
FF, so all the data outside the subpage is FF.

But in this case, part of the page in the device is already non-FF.  And 
that data does not change when FF is written to it.

The Denali controller calculates ECC on the data written, but that data 
will not match the data in the device, and so the ECC is incorrect.

When a subsequent read occurs, the ECC will not match and an error will 
occur.

Thanks,
Graham Moore
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/