[RFC v5] UBI: Fastmap support (aka checkpointing)

[Artem Bityutskiy]

On Wed, 2012-05-16 at 22:51 +0200, Richard Weinberger wrote:
> This is a bug-fix release. v4 handled static volumes wrong.
> v6 will address everything that Artem pointed out.

Hi, I'll try to review this further, but few points I'd like to
highlight.

1. We'll need to re-structure the code a bit and rename scan.[ch] to
attach.[ch], all data structures like 'struct ubi_scan_volume' which you
re-use we will also need to rename and remove the "scan" word, etc. I am
willing to do this preparation.

2. You need to re-use the scanning code we have for scanning the fastmap
volumes, I think. May be you need some amendments there, but currently
you have more duplication than needed. You'll just plug more of your
code to the attach.c file: check if there fastmap, if yes, read it and
scan the internal fastmap volumes by re-using functions in attach.c. So
all the fastmap reading/checking/manipulating stuff is in fastmap.c, the
scanning and interpreting is in attach.c.

Again, I am willing to do corresponding preparations for you, you'll
need to amend your code then.

Also, I think you probably do not have to split your patches. For me it
looks like there is a lot of work needed anyway, so you could just send
it as one patch so far, to make things easier. Because reviewing this
stuff by reading patches is too difficult anyway, and the reviewer has
to apply it and review / navigate the real code, and just look sometimes
to the patch.

-- 
Best Regards,
Artem Bityutskiy
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://lists.infradead.org/pipermail/linux-mtd/attachments/20120517/36abe605/attachment.sig>