Re: ubifs_decompress: cannot decompress ...

[Artem Bityutskiy <dedekind1@gmail.com>]

On Mon, 2011-06-06 at 12:04 -0400, Matthew L. Creech wrote:
> On Mon, Jun 6, 2011 at 5:58 AM, Artem Bityutskiy <dedekind1@gmail.com> wrote:
> >
> > So the corruption starts exactly at the NAND page boundary. This makes
> > me believe that the reason is most probably power cut recovery. But you
> > say your client ensures there were none...
> >
> > Yes, you are right that info from past is gone... What I'd like to see
> > is a dump of the whole LEB. Could you please add 'dbg_dump_leb()' -
> > basically I want to look if this LEB was passed through GC.
> >
> 
> [Repost, gmail mangled my node dump]
> 
> The dbg_dump_leb() call fails due to a bad CRC; I've posted its output
> plus the corresponding raw LEB dump (obtained via mtd->read_oob())
> here:

I have to go home now - could you please improve dbg_dump_leb().
Currently it calls ubifs_scan(), which scans, finds corrupted node,
prints corruption information and returns -EUCLEAN and destroys the
scanned data.

Instead, we could do:

1. Add another parameter to ubifs_scan() which makes it to _not_ free
scanned data on corruption.

2. In dbg_dump_leb() when you get -EUCLEAN - just go ahead and print the
scanned information instead of exiting. Should not be too difficult to
do.

-- 
Best Regards,
Artem Bityutskiy (Артём Битюцкий)