On Mon, Mar 21, 2011 at 2:51 AM, Charles Manning <
manningc2@actrix.gen.nz>wrote:
> On Saturday 19 March 2011 07:48:05 srimugunthan dhandapani wrote:
> > Hi all ,
> > This is my first mail to the yaffs group. I am learning yaffs file
> > system and i have some doubts trying to understand checkpointing and
> > how yaffs copes without having a on-disk format.
> >
> > 1. On mounting how is the starting of the checkpoint block located. I
> > guess it should be at a fixed location on flash as we cant scan for
> > the checkpoint start block. Then if we use the same block for
> > checkpoint wont it hurt the wear of that particular block?
>
> It is not at a fixed location but has to be found by scanning.
>
>
> > 2. Is the checkpoint information stored contiguously on the flash or
> > is it scattered ?
>
> It might be scattered.
>
I am sorry i think i understood wrongly.
I thought that because we have checkpoint info stored on flash , we can
avoid the lengthy scan of the whole flash to reconstruct the RAM structures.
If we also have to scan for the starting of the checkpoint block and the
subsequent checkpoint blocks, what exactly is the advantage of
checkpointing?.
Is it guaranteed like the checkpoint info will be at some first few blocks
of the flash, so scan time is less?
>
> > 3. Where is the super block stored on the flash?
>
> There is no superblock.
>
> > 4. When we open an existing file how do we get the tnode information?
> > When checkpointing does the tnode structure of all files also
> > checkpointed?
>
> Yes the checkpoint info stores the info in tnodes.
>
> > 5. Roughly how much RAM will be needed for a 512 GB flash assuming it
> > stores 1000 files with size of 500MB each.
>
> I have not tested yaffs with that much flash and there might be some
> issues,
>
> It would need a lot of RAM - many hundreds of Mbytes.
>
>
Other than needing more RAM, are there be any other issues you anticipate,
for using 512GB flash?.
Thanks,
Mugunthan
(text/plain)