Re: [PATCH 2/2] Add support for > 2GiB MTD devices

[Trent Piepho <tpiepho@freescale.com>]

On Wed, 27 Aug 2008, Carl-Daniel Hailfinger wrote:
> On 27.08.2008 20:51, Jamie Lokier wrote:
>> Bruce_Leonard@selinc.com wrote:
>>
>>> I'm still reluctant to change size to a 64-bit value.  There's a vague
>>> recolection of early conversations on the list that there would be little
>>> acceptance for that.  And that probably has to do with the ongoing
>>> conversation about ABI changes.  What I could do to eliminate the
>>> multiplication is introduce the same concept that the NAND layer uses,
>>> shift values.  After all, erasesize should always be a power of 2, making
>>> that a power of 2 multiplication which can be done via shifts.  By
>>> changing erasesize to erasesize_shift, I'd get something like this:
>>>
>>> return a->num_eraseblocks == 0 ? a->size : a->num_eraseblocks <<
>>> a->erasesize_shift
>>>
>>> How would that suit you?
>>>
>>
>> Are you sure it's always going to be a power of 2?
>>
>> What if someone targets a board with 3 chips wired to shared address
>> and parallel data buses?
>>
>> Or if someone makes a weird chip?  Or if you can format it in
>> different ways according to desired ECC level (like you can with CDs)?
>>
>
> IIRC I saw a datasheet for such a chip (selectable erasesize with
> non-power-of-2 default) some weeks ago and it had entered production a
> few months ago. The erasesize was alwas a multiple of 16, though. Sorry
> for not remembering more details.

It seems like the device size is always going to have some zeros in the least
significant bits.  Maybe a 32-bit size plus a shift is enough?  That could be
a lot more efficient than a 64-bit size and avoid penalizing most users who
don't need >32-bit size chips.