Re: [PATCH] mtd: spi-nor: fix options for mx66l51235f

[Marek Vasut <marek.vasut@gmail.com>]

On 08/30/2018 04:16 PM, Cyrille Pitchen wrote:
> Hi all,
> 
> Le 29/08/2018 à 14:12, Cédric Le Goater a écrit :
>> [ ... ] 
>>
>>> Ah, so the whole problem is, the ASpeed 2400 SPI NOR controller does
>>> memory-mapped read from the flash and that has some limitations when the
>>> flash is over 16 MiB in size -- it sends 3 byte standard opcodes with 4
>>> byte address ?
>>
>> yes.
>>
>> [ ... ]
>>
>>> To simplify the situation, if the ASpeed 2400 SPI NOR controller tried
>>> reading the SPI flash past 16 MiB, it would also require that the flash
>>> is in EN4B mode, yes ?
>>
>> yes.
>>
>> [ ... ]
>>
>>>> I start thinking about maybe adding an option to "jedec,spi-nor" DTS
>>>> binding like "enable-4byte-mode", so that we could get rid of our
>>>> local patch and just specify that option in the devicetree. The
>>>> option would force EN4B on the chip if it supports that mode. What do
>>>> you think? How do I do it if you approve? Send patches separately in
>>>> here and in the devicetree mailing list or cross-post both patches to
>>>> both lists?
>>>
>>> That's an option. But what I think this is really about is a fact that
>>> we completely lack any way to negotiate limitations between the SPI NOR
>>> and the controller.
>>
>> what about the hwcaps in spi_nor_scan() ? 
>>
> 
> Maybe only a matter of taste but I agree with Cedric: a new hwcaps sounds
> better than a new DT property. There are already different "compatible"
> strings to make the difference between AST2400 and AST2500, so we can
> add the new hwcap only to AST2400 entries. Then no need to update any
> existing device trees, at least in this case.
> 

Good idea :)

-- 
Best regards,
Marek Vasut