UBI(FS) issues: how to debug?

[Josh Cartwright]

On Thu, Mar 28, 2013 at 08:17:40PM +0000, Matthieu Castet wrote:
> Hi,

Hey Matthieu,

Thanks for the response.

> I saw that you have :    - NAND chip - Micron 8GB, on-die 4-bit ECC
>
> Could you post the patch you are using to use internal ecc ?

It was linked in my original post.  At this point, the vendor-provided
NAND driver has this hacked in.

> Did you try to add bitflip in nand (disable internal ecc, read block, add bitflip to data, erase et write) to check if bitflip are detected.
> Idem if you add more that 4 bit flip, do you have an incorrectable error ?

I like this idea.  I'll figure out how to give this a test.

> >  Loading file '.safe/linux_safemode.itb' to addr 0x08000000 with size 29085407 (0x01bbcedf)...
> >  UBIFS error (pid 0): ubifs_check_node: bad CRC: calculated 0x60c5bc7f, read 0x26c2f675
> >  UBIFS error (pid 0): ubifs_check_node: bad node at LEB 237:49728
> >  UBIFS error (pid 0): ubifs_read_node: expected node type 1
> >  UBIFS error (pid 0): do_readpage: cannot read page 7097 of inode 66, error -117
> >  Error reading file '.safe/linux_safemode.itb'
> Does it happen on readonly file ?
>
> If yes can't you compare the data/node between a working target and a bad target ?
>
>
> When doing your test in room temp, did you try to enable SLAB_DEBUG or other ram test ?
>
> RAM corruption can produce weird result, but I expect I cause some oops that you will notice.

Good idea, the extent of kernel debugging features enabled for the test
runs in the chamber were fairly limited, I'll see to adding more checks
in place to eliminate other non-NAND related issues.

Thanks again,

  Josh
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-mtd/attachments/20130328/e9f164b2/attachment.sig>