[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH] --cleanmarkers option in mkfs.jffs2



On Wednesday 11 September 2002 12.28, David Woodhouse wrote:
> johan.adolfsson@xxxxxxx.com said:
> >  Is there really a problem, more than that we waste space in the
> > middle  of a block, or that some blocks misses erasemarkers? Does the
> > fs code relies on that the block really is empty if there is a
> > cleanmarker followed by a few 0xFF's?
>
> As of about two days ago when we sped up the mount by an order of
> magnitude, yes. I haven't yet got round to doing the full check later --
> easy to do though now we've done the same for all the data and dnode crcs.
>
> > in that case, the worst thing that could happen is that the write failes
> > and that we have to erase the block again?
>
> Er, the write will fail and we'll panic at the moment. Or at least go into
> an endless loop without skipping over the offending region.

That doesn't sound like an optimal solution ;-)

> The case I'm worried about is where we get a CLEANMARKER node in the
> _middle_ of an erase block.

How about adding erase size in the clean marker?
That way we could now how much free space to expect when writing, 
and could use that instead of the erase size we really have until the block 
is garbage collected.

BTW: Isn't it really a user error to create an image with the 
wrong erase size and something that seldom happens in real life?

/Johan

To unsubscribe from this list: send the line "unsubscribe jffs-dev" in
the body of a message to majordomo@xxxxxxx.com