NAND flash

NAND flash

[Eugeny Mints]

Hello,

Could anybody help me to solve such a problem with JFFS2 and NAND flash:
- the NAND flash Toshiba TC58256AFTI
- kernel 2.4.22 with latest MTD and JFFS2 
- handy-written miniport for the paticular flash chip

The device works OK almost in any case, but when I try to execute  while
true; do dd if=/dev/zero of=/dev/mnt/foo count=1000; done after 10-15
minutes the system hangs; I cannot login from console or via telnet, but
it is still ping'able. I tried to use YAFFS, perform dd to device itself
instead of file on the filesystem on top of flash - no hangs were
observed.

Could anybody help me, please? 
--
cheers, dmitry pervushin

Re: NAND flash

[David Woodhouse]

On Thu, 2003-10-09 at 19:04 +0400, Eugeny Mints wrote:
> Hello,
> 
> Could anybody help me to solve such a problem with JFFS2 and NAND flash:
> - the NAND flash Toshiba TC58256AFTI
> - kernel 2.4.22 with latest MTD and JFFS2 
> - handy-written miniport for the paticular flash chip

When you say 'latest' do you really mean that? NAND flash support in
JFFS2 has changed a _lot_ in the last weeks, and even in the last few
days.

> The device works OK almost in any case, but when I try to execute  while
> true; do dd if=/dev/zero of=/dev/mnt/foo count=1000; done after 10-15
> minutes the system hangs; I cannot login from console or via telnet, but
> it is still ping'able. I tried to use YAFFS, perform dd to device itself
> instead of file on the filesystem on top of flash - no hangs were
> observed.

Use SysRq-P and/or SysRq-T to see where the kernel is stuck. IF that
doesn't work, enable CONFIG_JFFS2_FS_DEBUG=1 and 
echo > /proc/sys/kernel/printk then watch the console, and see what the
last thing it says is.

-- 
dwmw2

RE: NAND flash

[Stephan Linke]

Hi Eugeny,

maybe some problem with the flash "ready" detection?

Stephan

> -----Original Message-----
> From: linux-mtd-bounces@lists.infradead.org
> [mailto:linux-mtd-bounces@lists.infradead.org]On Behalf Of Eugeny Mints
> Sent: Donnerstag, 9. Oktober 2003 17:04
> To: linux-mtd@lists.infradead.org
> Cc: Dmitry Pervushin
> Subject: NAND flash
> 
> 
> Hello,
> 
> Could anybody help me to solve such a problem with JFFS2 and NAND flash:
> - the NAND flash Toshiba TC58256AFTI
> - kernel 2.4.22 with latest MTD and JFFS2 
> - handy-written miniport for the paticular flash chip
> 
> The device works OK almost in any case, but when I try to execute  while
> true; do dd if=/dev/zero of=/dev/mnt/foo count=1000; done after 10-15
> minutes the system hangs; I cannot login from console or via telnet, but
> it is still ping'able. I tried to use YAFFS, perform dd to device itself
> instead of file on the filesystem on top of flash - no hangs were
> observed.
> 
> Could anybody help me, please? 
> --
> cheers, dmitry pervushin
> 
> 
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/
> 
> 

Re: NAND flash

[Charles Manning]

On Friday 10 October 2003 04:21, Stephan Linke wrote:
> Hi Eugeny,
>
> maybe some problem with the flash "ready" detection?

I do not think so since this is working fine with YAFFS which makes the same 
mtd calls as JFFS2.

Try using the newly announced JFFS2 stuff - that likely addresses some of the 
issues in previous JFFS2 releases.

-- Charles

>
> Stephan
>
> > -----Original Message-----
> > From: linux-mtd-bounces@lists.infradead.org
> > [mailto:linux-mtd-bounces@lists.infradead.org]On Behalf Of Eugeny Mints
> > Sent: Donnerstag, 9. Oktober 2003 17:04
> > To: linux-mtd@lists.infradead.org
> > Cc: Dmitry Pervushin
> > Subject: NAND flash
> >
> >
> > Hello,
> >
> > Could anybody help me to solve such a problem with JFFS2 and NAND flash:
> > - the NAND flash Toshiba TC58256AFTI
> > - kernel 2.4.22 with latest MTD and JFFS2
> > - handy-written miniport for the paticular flash chip
> >
> > The device works OK almost in any case, but when I try to execute  while
> > true; do dd if=/dev/zero of=/dev/mnt/foo count=1000; done after 10-15
> > minutes the system hangs; I cannot login from console or via telnet, but
> > it is still ping'able. I tried to use YAFFS, perform dd to device itself
> > instead of file on the filesystem on top of flash - no hangs were
> > observed.
> >
> > Could anybody help me, please?
> > --
> > cheers, dmitry pervushin
> >
> >
> > ______________________________________________________
> > Linux MTD discussion mailing list
> > http://lists.infradead.org/mailman/listinfo/linux-mtd/
>
> ______________________________________________________
> Linux MTD discussion mailing list
> http://lists.infradead.org/mailman/listinfo/linux-mtd/

Re: NAND FLASH

[Thomas Gleixner]

On Friday, 21. June 2002 19:07, Steve Tsai wrote:
> Do I use GPIO to control CE? Because I use memory-map-io to connnect to
> nand flash, if CE can not use to address the chip, I have to use other
> design to address the flash chip. Thanks.
Yep you statically apply CE. Then the processing is done, then CE is released.
That's the way, how nand.c handles this. And it handles it correct.

> 1. CE should be a static signal
> 2. please read http://www.linux-mtd.infradead.org/tech/nand.html
> 3. Use the latest code from mtd-cvs
> 4. Read one of the board specific nand drivers to setup your own

-- 
Thomas
________________________________
Thomas Gleixner - autronix automation
http://www.autronix.de gleixner@autronix.de

RE: NAND FLASH

[Steve Tsai]

Do I use GPIO to control CE? Because I use memory-map-io to connnect to
nand flash, if CE can not use to address the chip, I have to use other
design to address the flash chip. Thanks.

Steve Tsai

> Hi,
> I try to use nand flash driver to my board, but it seems fail to do 
> some operations. I just can get chip ID, but I can not read data or 
> program data. I use GPIO to control CLE and ALE and connect WE, RE and

> CE to procesor. Is it right? Does anyone have sample code to test NAND

> flash?
1. CE should be a static signal 
2. please read http://www.linux-mtd.infradead.org/tech/nand.html 
3. Use the latest code from mtd-cvs
4. Read one of the board specific nand drivers to setup your own

Re: NAND FLASH

[Thomas Gleixner]

On Friday, 21. June 2002 11:44, Steve Tsai wrote:
> Hi,
> I try to use nand flash driver to my board, but it seems fail to do some
> operations. I just can get chip ID, but I can not read data or program
> data. I use GPIO to control CLE and ALE and connect WE, RE and CE to
> procesor. Is it right? Does anyone have sample code to test NAND flash?
1. CE should be a static signal 
2. please read http://www.linux-mtd.infradead.org/tech/nand.html 
3. Use the latest code from mtd-cvs
4. Read one of the board specific nand drivers to setup your own

-- 
Thomas
___________________________________
autronix automation GmbH
http://www.autronix.de gleixner@autronix.de

NAND FLASH

[Steve Tsai]

Hi,
I try to use nand flash driver to my board, but it seems fail to do some
operations. I just can get chip ID, but I can not read data or program
data. I use GPIO to control CLE and ALE and connect WE, RE and CE to
procesor. Is it right? Does anyone have sample code to test NAND flash?
Thank.

Steve TSai

Re: NAND flash

[David Woodhouse]

ldoolitt@recycle.lbl.gov said:
>  I think you should double check that assertion.  My understanding is
> that the "Smart" in the name is a bug.  These are actually NAND chips,
> consumer-grade packaged, standardized, and marketed, with no smarts at
> all.  So you _could_ use the nand.c driver, with the right interface-
> specific wrapper layer to get at the device.

Correct, with the proviso that some SmartMedia 'adapters' have 
microcontrollers built in, and the host can't actually _get_ at the raw 
flash.

> Unfortunately, the result would not be content-compatible with other
> SmartMedia users, because SmartMedia also has a standard for encoding
> blocks on the nand chips.  I tried to read that standard, but the PDF
> file is encrypted in a way that is incompatible with xpdf-0.92.  The
> URL for that reference material has been posted here before, it's
>  http://www.ssfdc.or.jp/spec/english/

Acroread can manage it. It all looks fairly simple to implement.

--
dwmw2

NAND flash

[Larry Doolittle]

Steven J. Hill wrote:
> First of all, the NAND FLASH driver was written for raw NAND flash
> chips that were IO mapped into the address space.
> ...
> SmartMedia is an entirely different beast in that has a some hardware
> between the CPU and the NAND flash chips inside. It is similar to
> Disk-On-Chip devices which use NAND/NOR flash with wear-leveling,
> error correcting and other things done transparently in the hardware.

I think you should double check that assertion.  My understanding is
that the "Smart" in the name is a bug.  These are actually NAND chips,
consumer-grade packaged, standardized, and marketed, with no smarts at
all.  So you _could_ use the nand.c driver, with the right interface-
specific wrapper layer to get at the device.

Unfortunately, the result would not be content-compatible with other
SmartMedia users, because SmartMedia also has a standard for encoding
blocks on the nand chips.  I tried to read that standard, but the PDF
file is encrypted in a way that is incompatible with xpdf-0.92.  The
URL for that reference material has been posted here before, it's
 http://www.ssfdc.or.jp/spec/english/

      - Larry

Re: NAND flash

[Steven J. Hill]

nagaraj trivedi wrote:
> 
> can I use the MTD NAND FLASH driver for smartmedia along with the jffs2? Does jffs2
> work with the MTD NAND flash driver ?
>
OK...let's break this into small pieces. First of all, the NAND FLASH driver was
written for raw NAND flash chips that were IO mapped into the address space. I
wrote
those drivers originally. Next, I got JFFS working on top of the NAND MTD driver
to the point that it could mount, read, write, delete, remove directories, etc.
I
and another person also did a fair amount of stress testing. That is where
things
currently stand.

SmartMedia is an entirely different beast in that has a some hardware between
the
CPU and the NAND flash chips inside. It is similar to Disk-On-Chip devices which
use NAND/NOR flash with wear-leveling, error correcting and other things done
transparently in the hardware. You would need to get the low-level
specifications
for SmartMedia and then write a SmartMedia driver. Once that is done, you could
then
work on the task of getting JFFS2 working. IIRC though, JFFS2 needs
modifications
to work with NAND flash. Myself and David could work out what changes need to be
in
fairly short order. Hope that helps.

-Steve

-- 
 Steven J. Hill - Embedded SW Engineer

NAND flash

[nagaraj trivedi]

Hello,
   can I use the MTD NAND FLASH driver for smartmedia along with the jffs2? Does jffs2 work with the MTD NAND flash driver ?

 

NAND flash

[nagaraj trivedi]

Hello,
   can I use the MTD NAND FLASH driver for smartmedia along with the jffs2? Does jffs2 work with the MTD NAND flash driver ?