[RESUBMIT] [PATCH] [MTD] NAND nand_ecc.c: rewrite for improved performance

[Frans Meulenbroeks]

2008/8/15, David Woodhouse <dwmw2 at infradead.org>:
> Frans, thank you very much for your work. One question -- has this been
>  tested on 64-bit machines?

Good question, and more difficult than I thought.
Yes, It has been tested on 64 bit hardware. More specific a DELL 9150
with a D920 (presler) dual core processor.
However, not too sure if this system runs 64bit linux. Basically it
runs an out-of-the-box install of Ubuntu 7.10.

Note also that I tested this on this PC by running a test application
and comparing the results with the original code as my PC does not
have (user accessible) NAND flash.

If desired I can install and test on a 64 bit kernel.
Then again, I do not really expect issues (it is only an algorithm
with simple operations).
Also 64 bit is not that interesting. I doubt if we will ever see
systems 64 bit processors that use NAND flash without hardware ECC
calculations...

>
>  The patch still doesn't apply, unfortunately -- something seems to be
>  mangling it. In general, it's a good idea to send patches to yourself in
>  email first, then save them off from your own INBOX and try to apply
>  them to a clean kernel tree.

Yikes. Can you email me the error log?
I know about sending patches to myself and testing them, but again
things are more complicated here.
I've developed this on a fresh kernel from kernel.org. The previous
version failed because the mtd git has a change from beginning of june
where all cvs comments have been removed. This change has not
propagated to mainline yet (2.6.26.2 still contains the cvs comment).
As I have no git experience, what I did was pull a copy of nand_ecc.c
(the only file I change) from the mtd git using my browser
(http://git.infradead.org/mtd-2.6.git?a=blob_plain;f=drivers/mtd/nand/nand_ecc.c;hb=HEAD)
and overwrite the one in 2.6.26.2. The only thing I can imagine is
that something went wrong with that. (e.g. tabs and spaces).

>
>  Please could you let me have a copy as an attachment, or just a straight
>  copy of both files?

I'll send you both files later today (they are on a different system).

Best regards, Frans. (and apologies for the trouble I caused with this).