Re: [PATCH] spi-nor: Verify written data in paranoid mode

[Csókás Bence]

Hi,

On 2025. 04. 16. 13:59, Michael Walle wrote:
> Hi,
>
> > Add MTD_SPI_NOR_PARANOID config option for verifying all written data to
> > prevent silent bit errors to be undetected, at the cost of halving SPI
> > bandwidth.
>
> What is the use case for this? Why is it specific to SPI-NOR
> flashes? Or should it rather be an MTD "feature". I'm not sure
> whether this is the right way to do it, thus I'd love to hear more
> about the background story to this.

Well, our case is quite specific, but we wanted to provide a general 
solution for upstream. In our case we have a component in the data path 
that can cause a burst bit error, on average after about a hundred 
megabytes written.

We _could_ make it MTD-wide, in our case we only have a NOR Flash 
onboard so this is where we added it. If it were in the MTD core, where 
would it make sense?

* mtd_write()
* mtd_write_oob()
* mtd_write_oob_std()
* or somewhere else entirely?

> > Co-developed-by: Szentendrei, Tamás <szentendrei.tamas@xxxxxxxxx>
> > Signed-off-by: Szentendrei, Tamás <szentendrei.tamas@xxxxxxxxx>
> > Signed-off-by: Csókás, Bence <csokas.bence@xxxxxxxxx>
> > ---
> >   drivers/mtd/spi-nor/Kconfig | 10 ++++++++++
> >   drivers/mtd/spi-nor/core.c  | 33 +++++++++++++++++++++++++++++++++
> >   2 files changed, 43 insertions(+)
> >
> > diff --git a/drivers/mtd/spi-nor/Kconfig b/drivers/mtd/spi-nor/Kconfig
> > index 24cd25de2b8b..425ea9a22424 100644
> > --- a/drivers/mtd/spi-nor/Kconfig
> > +++ b/drivers/mtd/spi-nor/Kconfig
> > @@ -68,6 +68,16 @@ config MTD_SPI_NOR_SWP_KEEP
> >   
> >   endchoice
> >   
> > +config MTD_SPI_NOR_PARANOID
> > +	bool "Read back written data (paranoid mode)"
>
> No kernel configs please. This doesn't scale. What if you have two
> flashes and one should have this and one does not?

Yes, we have thought about this, but concluded that "paranoid mode" is 
not device-specific, you either have a requirement to be extra sure 
about data integrity, or you can be fairly sure your system is sane, in 
both cases it is a judgement about the entire system. I also thought 
that maybe some devices could be exempt from this, contingent on a 
`no-paranoia` Device Tree property, to selectively sacrifice integrity 
for performance even in paranoid mode, but we only have one Flash 
anyway, so I didn't implement it.

> -michael

Bence