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

[Cédric Le Goater <clg@xxxxxxxx>]

[ ... ] 

>> I CCed him, so maybe he can clarify.
>>
>> But this is weird. If you reboot the system, isn't the SPI NOR
>> power-cycled and reset into defined default state ? If not, your
>> hardware is severely broken and you should fix your reset routing.
> 
> I guess it was clarified below already but will say it again: EN4B is
> not reset on soft reboot, there is no power cycle. So the system won't
> boot.

I am working on an Aspeed SoC SPI driver for U-Boot and I am seeing the 
exact same problem with a W25Q256 when Linux reboots. These chips are 
even more perfidious because they have a non-volatile bit making the 
chip operate in 4Byte address mode at power-up. 

> There was a quirk for shutdown, which would EX4B. But it was clumsy
> and probably never sent upstream. 

But the problem is supporting older Linux versions not using the  
4B_OPCODES. So adding an initial EX4B in the U-Boot SPI layer seems 
quite safe and not that ugly. Or add 4Byte support. 

This is a discussion for U-Boot.

> SPI_NOR_4B_OPCODES is much more elegant.
I agree that the SPI_NOR_4B_OPCODES are a good practice because they 
don't alter the chip state and let software do the appropriate 
configuration. 

Is removing EN4B/EX4B part of the long-term plan for the Linux SPI-NOR 
layer ? 

> Sorry for not clarifying this in commit message.

Thanks,

C.

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/