[flashrom] operation details and future changes (was: flashrom ignoring ROM layout region)

[Stefan Tauner]

On Fri, 8 Feb 2013 11:08:21 +0100
Mark Marshall <markmarshall14 at gmail.com> wrote:

Hi Mark,

> I too would love for these changes to go in.  I'm currently working with a
> 64MB flash, and
> I've got a number of problems, one of which is that flashrom tries to read
> the entire flash
> before doing anything (which will never be what I want...).

It does so to be able to skip blocks that have the wanted content
already and to be able to tell you if anything has changed in the case
of an error (not so important on external programmes, but very
interesting for people writing their live BIOSes and seeing errors).

Usually reading does take way less time than erasing/writing or might
even speed up the whole process. ATM there is no way to disable this
behavior but we plan to do so and IIRC a part of the mentioned patch set
is a first step into that direction.

> The other problems are:
> - That most of flashrom seems to be limited to 2^24 address bits.
>   (this is relatively easy to fix)

We will see about that :) We are of course aware that this is needed
quite soon. Do you want to help us with this (mostly testing, but if
you like you can also try to create the needed patch)?

> - That flashrom thinks a page is always 256 byes (I think I've found
>   most cases of this, but I'm not sure.  What's the granularity flag
>   for?

The flag is there to fix exactly this :)
See also http://patchwork.coreboot.org/patch/3757/
Not many chips need this and so we never moved into this direction...
do you really need it?

> I've also found that flashrom doesn't always do what I expect, and I'm not
> sure if this is
> by design or not.  for instance, if I want to erase my entire flash chip,
> it ends up calling
> the sector erase on each sector, not by doing a chip-erase?  Are these
> types of things
> written down anywhere?

We try to erase the smallest possible blocks first (in order of the
eraser array of the chip definition(s) in flashchips.c. This potentially
allows to skip writing of block as explained above. It is of course
rather useless for the -E/--erase operation, but the code is shared and
no one complained yet IIRC (or sent a patch). Most of these details are
not documented I think, but some are (in the manpage). Patches for the
manpage are welcomed. Everyone is invited to edit the wiki too :)
-- 
Kind regards/Mit freundlichen Grüßen, Stefan Tauner