Re: Nand read/write

Re: Nand read/write

[Vitaly Wool]

M Sudharshana-A18475 wrote:
> It is a flash partition where no file system is defined for that
> mount-able partition. Read/write operations are performed on /dev/mtd0
> (place where memory partition is mounted
Ok, then the answer is yes, unless your board driver selects 
NAND_ECC_NONE for ECC calculation.

I'd also expect to hear a lection on top-posting if I were you.

Vitaly

Re: Nand read/write

[Thomas Gleixner]

At Thu, 13 Jul 2006 13:48:32 +0400,
Vitaly Wool wrote:
> 
> M Sudharshana-A18475 wrote:
> > It is a flash partition where no file system is defined for that
> > mount-able partition. Read/write operations are performed on /dev/mtd0
> > (place where memory partition is mounted
> Ok, then the answer is yes, unless your board driver selects 
> NAND_ECC_NONE for ECC calculation.
> 
> I'd also expect to hear a lection on top-posting if I were you.

And you get a lesson for the wrong answer. :)

The question was if bad blocks are handled when he accesses
the FLASH via /dev/mtdX. The answer is:

It's not possible to write / erase bad blocks.

But there is no bad block replacement functionality when you
access the FLASH through the raw MTD interface. Neither MTD
nor NAND provide anything like this. Raw access has to take
care of this itself. Therefor we require that images are
written with nandwrite instead of dd or cp. nandwrite has
the simplest form of bad block handling (skip it).

	tglx

Nand read/write

[M Sudharshana-A18475]

I have the following scenario on an embedded Linux application.

There is a partition in the NAND flash memory (with RAW type, no file
system), which will be mounted and used for read/write. As an
application which performs "read"/"write" on this partition after
mounting, should it be assumed that "bad block algorithm" is covered? Or
application has to handle the same?

Thanks for your help

Sudharshana

Re: Nand read/write

[Vitaly Wool]

M Sudharshana-A18475 wrote:
> I have the following scenario on an embedded Linux application.
>
> There is a partition in the NAND flash memory (with RAW type, no file
> system), which will be mounted and used for read/write. As an
> application which performs "read"/"write" on this partition after
> mounting, should it be assumed that "bad block algorithm" is covered? Or
> application has to handle the same?
>   
Can you please explain the term "mounted" for raw flash case you're 
talking about?

Thx,
   Vitaly

Re: Nand read/write

[dedekind]

>I have the following scenario on an embedded Linux application.
>
>There is a partition in the NAND flash memory (with RAW type, no file
>system), which will be mounted and used for read/write. As an
>application which performs "read"/"write" on this partition after
>mounting, should it be assumed that "bad block algorithm" is covered? Or
>application has to handle the same?

Nope. But UBI can do a lot. Try to look at it.