RE: Intel FLASH

["Doyle, Patrick" <WPD at dtccom dot com>]

OK folks,
Attached is my rework of the strata device driver.  As often happens with
these sorts of things, the scope grew a bit as I worked through this.  The
new driver determines the number of blocks, the size(s) of the blocks, and
the size of the device by querying the CFI.  Unlocking, locking, and erasing
blocks still operate on the single (larger) block size, but handle the
smaller boot/parameter blocks correctly.

I have tried my best to ensure that this new driver is backwards compatible
with platforms mentioned in the tree (as of 2003-09-04 -- I've gotta update
my tree now), but I, very likely, have left something out.  I would
appreciate it if folks would pound on this a little bit and let me know.

Oh yeah, I also added support for CYGNUM_FLASH_SERIES.  One nice side effect
of this (which was totally intentional, since that's the way my hardware is)
is that one can put a bottom boot device in series with a top boot device
and have the locking, unlocking, erasing all handled properly.

I wonder if my changes are significant enough, or the risk of breaking
existing platforms is high enough, that I should change this to an entirely
new device driver, perhaps intel/cfiflash, or something like that; perhaps
even move it up one layer in the tree, since it is nolonger specific to
Intel -- in theory any flash that implents CFI should work with this.

I don't know if you want to apply this to the tree yet.  I would want to see
some folks (Jani?, Gary?, anybody else?) try it for a while and let me know
of problems before it should really be incorporated into the tree.

What do you think?

--wpd

Attachment: st2.diff
Description: Binary data