Re: [Yaffs] Disadvantage of using yaffs checkpointing?

Top Page
Attachments:
Message as email
+ (text/plain)
Delete this message
Reply to this message
Author: Charles Manning
Date:  
To: yaffs
Subject: Re: [Yaffs] Disadvantage of using yaffs checkpointing?
On Wednesday 03 March 2010 01:03:12 Sven Van Asbroeck wrote:
> Hello Shivdas,
>
> Yaffs does not directly concern itself with bad blocks - these are the
> responsibility of the MTD layer. In short: 1. During startup, the MTD layer
> finds bad blocks (usually by scanning or a bad block table on the flash)
> and stores them in RAM.

To clarify: YAFFS does care which blocks are bad, but it does not care how the
mtd knows it is bad. Some mtd drivers will scan the blocks and build up a bad
block table and some do not. What the driver does is hidden from yaffs.

> 2. During file system mount, yaffs uses the MTD
> API's to determine whether a block is bad. 3. If a block becomes bad during
> file system usage, yaffs will again use the MTD APIs to mark the block as
> bad.
>
> Saving a check point does not avoid bad block scanning. Scanning will
> happen when the MTD driver starts, which is well before yaffs comes into
> play.

Partially true. There is no reason why the mtd should do the bad block
scanning at start up. From a yaffs perspective this is just wasted time.

-- Charles