Re: [PATCH v3] mtd: rawnand: ams-delta: Drop board specific partition info

[Ladislav Michl]

On Thu, Apr 25, 2019 at 08:42:22PM +0200, Janusz Krzysztofik wrote:
> Hi Ladislav,
> 
> On Thursday, April 25, 2019 12:14:28 AM CEST Ladislav Michl wrote:
> > Hi Janusz,
> > 
> > On Wed, Apr 24, 2019 at 08:02:12PM +0200, Janusz Krzysztofik wrote:
> > > After recent modifications, only a hardcoded partition info makes
> > > the driver device specific.  Other than that, the driver uses GPIO
> > > exclusively and can be used on any hardware.
> > > 
> > > Drop the partition info and use MTD partition parser with default list
> > > of parser names instead.  For the OF parser to work correctly, pass
> > > device of_node to mtd.
> > > 
> > > Amstrad Delta users should append the following partition info to their
> > > kernel command line, possibly embedding it in CONFIG_CMDLINE:
> > > 
> > > mtdparts=ams-delta-nand:3584k(Kernel),256k(u-boot),256k(u-boot_params),\
> > > 256k(Amstrad_LDR),27m(File_system),768k(PBL_reserved)
> > 
> > now, when driver is no longer Amstrad Delta specific, why would you want
> > to have ams-delta-nand hardcoded on kernel cmdline? I'm assuming at some
> > point this driver will become gpio-nand [*] or something like that and
> > asking users to change their kernel cmdline twice is just unwise :)
> 
> Hmm, I have no idea of a good name for the driver if not "gpio-nand". Can you 
> suggest one?

gpio-nand is so good name that it should be worth merging gpio.c and
ams-delta.c into gpio_nand.c :)

> As a workaround, I can add a platform device id table to the driver with "ams-
> delta-nand" as a supported device name in hope that survives possible future 
> driver renaming.
> 
> > [*] btw, it is really shame gpio-nand name is already taken by driver
> > living in drivers/mtd/nand/raw/gpio.c which is more likely gpio-mem-nand
> > used by at least CompuLab CM-X255 and Picochip picoXcell.
> 
> I think the best approach would be to expose NAND data ports of those machines 
> as GPIO ports, possibly reusing the "gpio-nand" driver code while creating a 
> new GPIO driver for them if "basic-mmio-gpio" occurs inappropriate, and use 
> the pure GPIO NAND driver on top.

What about adding two fields into struct ams_delta_nand holding pointers to
either ams_delta_{read,write}_buf (renamed to gpio_nand_...) or mmio r/w
functions depending on driver configuration?

> > Otherwise your work on this driver is so amazing that I just spent
> > couple of hours finding that phone and compiling some decent userspace
> > for it :) Thank you!
> 
> I'm glad you like it :-)
> Janusz

Best regards,
	ladis