[PATCH 0/9] AT24 EEPROM MTD Support

[Andy Shevchenko]

On Mon, Aug 26, 2024 at 09:51:10AM +0200, Marco Felsch wrote:
> On 24-08-23, Andy Shevchenko wrote:
> > On Mon, Jul 01, 2024 at 03:53:39PM +0200, Marco Felsch wrote:
> > > This series adds the intial support to handle EEPROMs via the MTD layer
> > > as well. This allow the user-space to have separate paritions since
> > > EEPROMs can become quite large nowadays.
> > > 
> > > With this patchset applied EEPROMs can be accessed via:
> > >   - legacy 'eeprom' device
> > >   - nvmem device
> > >   - mtd device(s)
> > > 
> > > The patchset targets only the AT24 (I2C) EEPROMs since I have no access
> > > to AT25 (SPI) EEPROMs nor to one of the other misc/eeprom/* devices.
> > > 
> > > Note: I'm not familiar with Kconfig symbol migration so I don't know if
> > > the last patch is required at the moment. Please be notified that the
> > > list of recipients is quite large due to the defconfig changes.
> > 
> > FWIW, I think that MTD is *not* the place for EEPROMs.
> > 
> > Yeah, we have the driver spread over the kernel for EEPROMs (mostly due to
> > historical reasons and absence an umbrella subsystem for them), but it's not
> > the reason to hack them into something which is not quite suitable.
> 
> Thank you for you input. There are two things to mention:
>  1st) I send a RFC patch and asked for feedback and all I got was: looks
>       okay, please send a proper patch [1] which I did.

I was on a long vacation, I haven't had time or even wishes to look at the
patches or patch series. Sorry for that.

Second point, RFC means "request for comments", here is mine. It's up to the
maintainers and you on how to proceed it.

>  2nd) I don't see the hacky part in this patchset.

I haven't talked about patchset, I have talked about architectural / design
point of view. I read the discussion and to me it seems like it solves the
issue with a quite big hammer. If you can prove that on embedded systems with
limited resources it is not a problem, just mention that in the cover letter.

> Anyway the customer doesn't need the nvmem-partitions anymore and
> therefore this patchset can be seen as obsolote.
> 
> [1] https://lore.kernel.org/lkml/20231201144441.imk7rrjnv2dugo7p@pengutronix.de/T/#m1e0e5778448971b50a883f62bd95622f6422b9a2
> 
> > If NVMEM needs to be updated and may cover these cases after all (and do not
> > forget about *small* size EEPROMs that most likely appear on the devices with
> > limited amount of resources!) in a reasonable size and performance, why not?

-- 
With Best Regards,
Andy Shevchenko