[U-Boot] [PATCH RFC] nand: remove spurious read cycle in OOB first page reads

[Scott Wood]

On Mon, Nov 23, 2009 at 10:38:02AM +0000, Nick Thompson wrote:
> > Ideally a high-level driver like fsl_elbc_nand wouldn't have to implement
> > cmdfunc at all.
> 
> I'm not sure I know what you mean here. I'll have a look at that code.

The eLBC NAND controller presents an interface that runs entire NAND
operations at once.  In order to implement cmdfunc, we have to interpret
what the higher levels are trying to do and then turn it into something the
hardware will accept.

It would be much more straightforward to implement high-level "erase block",
"progam page", "read page", etc. operations.

> cmdfunc should probably be simplified to only send command sequences and
> not invoke any nand wait functions. This gets in the way of some useful
> optimisation. If some drivers are replacing it, it will be more difficult
> to coordinate changes.

For the eLBC, we have to issue commands and wait as one operation, or else
the bus controller can change chipselects in the meantime, causing problems
with shared pins.  More reason to not make cmdfunc the only way to hook in a
driver...

-Scott