Re: [PATCH 3/4] mtd: mchp23k256: add partitioning support

[Brian Norris]

On Thu, Jun 01, 2017 at 11:08:08PM +0000, Chris Packham wrote:
> Do we need a flag to indicate SRAM-like properties? I assume there is a 
> difference between NO_ERASE on ROM devices where there is just no way of 
> erasing the data. For {S,F,M}RAM there is no block erase operation but 

I think we already have that:

#define MTD_CAP_ROM             0
#define MTD_CAP_RAM             (MTD_WRITEABLE | MTD_BIT_WRITEABLE | MTD_NO_ERASE)

The key signifier for ROM would be !MTD_WRITEABLE.

> you can overwrite data to destroy it (which is actually my use-case with 
> this SPI SRAM). I was tempted to set erase_size = 1 at one point which 
> in my mind was technically accurate but would probably upset the mtd 
> layer just as much as 0.

I'm not sure what erasesize should be here. I suppose 0, but really, I
think the MTD_NO_ERASE flag is the clearer indication that erase is not
needed, and that one should ignore the erasesize.

Brian