[PATCH] UBI: optimize erase-header IO read checks

[Gupta, Pekon]

Hi Artem,

>From: Artem Bityutskiy [mailto:dedekind1 at gmail.com]
>Subject: Re: [PATCH] UBI: optimize erase-header IO read checks
>
>
>On Mon, 2013-04-29 at 10:37 +0530, Gupta, Pekon wrote:
>...
>> Signed-off-by: Gupta, Pekon <pekon at ti.com>
>
>I do generally understand what you are trying to do in this patch, but I
>cannot carefully review it, because it is too big, just like the second
>patch.
>
>It really should be split. I think you should start with a preparation
>patch which does not change the current logic, but just transforms the
>code to using a switch statement.
>
>Then you should start modifying the big switch statement, preferably one
>step per patch, explaining each step. I guess the first step would be
>removing the magic numbers check. Another step would be removing the
>0xFF comparison. Etc.
>
>Then your work could be reviewed, and we could be somewhat confident
>that it does not break anything.
>

Sorry for late reply. I'm bit stuck in other omap2 NAND driver
updates. I'll take this update independently from other activities.
As this needs rigorous testing.
(but this is always in my todo list marked in red :-))

Also, I purposely tweaked the functionality a bit, (like removal of 0xff
comparison) to speed-up parsing of erase-header and volume-headers.
These checks were not explicitely required, as some of these are
covered implicitly in other checks like CRC match.
I'll re-visit these tweaks with proper comments in code when I resend.
However, I needs some help in performance testing this to check
whether these patches could help in speeding up mount time abit.
Though I agree these may not match 'fast-scan'.


with regards, pekon