Re: [RFC PATCH] mtd: add OTP (one-time-programmable) erase ioctl

[Michael Walle]

Hi,

Am 2021-03-02 13:46, schrieb Miquel Raynal:
> Michael Walle <michael@xxxxxxxx> wrote on Tue,  2 Mar 2021 12:09:27
> +0100:
>
> > This may sound like a contradiction but some SPI-NOR flashes really
> > support erasing their OTP region until it is finally locked. Having 
> > the
> > possibility to erase an OTP region might come in handy during
> > development.
> >
> > The ioctl argument follows the OTPLOCK style.
> >
> > Signed-off-by: Michael Walle <michael@xxxxxxxx>
> > ---
> > OTP support for SPI-NOR flashes may be merged soon:
> > https://lore.kernel.org/linux-mtd/20210216162807.13509-1-michael@xxxxxxxx/
> >
> > Tudor suggested to add support for the OTP erase operation most 
> > SPI-NOR
> > flashes have:
> > https://lore.kernel.org/linux-mtd/d4f74b1b-fa1b-97ec-858c-d807fe1f9e57@xxxxxxxxxxxxx/
> >
> > Therefore, this is an RFC to get some feedback on the MTD side, once 
> > this
> > is finished, I can post a patch for mtd-utils. Then we'll have a 
> > foundation
> > to add the support to SPI-NOR.

[..]

> > +int mtd_erase_user_prot_reg(struct mtd_info *mtd, loff_t from, size_t 
> > len)
> > +{
> > +	struct mtd_info *master = mtd_get_master(mtd);
> > +
> > +	if (!master->_erase_user_prot_reg)
> > +		return -EOPNOTSUPP;
> > +	if (!len)
> > +		return 0;
>
> Should we add a sanity check enforcing that we don't try to access
> beyond the end of the OTP area?

This is checked in the op itself, as it is done for all the
other OTP read/write/lock ops.

Right at the moment, I don't see how this could be achieved in
an elegant way. Without additional changes, you'd have to call
mtd_get_user_prot_info() and iterate over the returned values
and figure out whether from and len are valid.

[..]

-michael