Re: [PATCH v2 0/5] mtd: spi-nor: Add support for Octal 8D-8D-8D mode

[Boris Brezillon]

On Tue, 21 Apr 2020 15:05:08 +0530
Vignesh Raghavendra <vigneshr@xxxxxx> wrote:

> On 21/04/20 12:53 pm, Boris Brezillon wrote:
> > +Pratyush who's working on a similar patchet [1].
> > 
> > Hello Mason,
> > 
> > On Tue, 21 Apr 2020 14:39:42 +0800
> > Mason Yang <masonccyang@xxxxxxxxxxx> wrote:
> >   
> >> Hello,
> >>
> >> This is repost of patchset from Boris Brezillon's
> >> [RFC,00/18] mtd: spi-nor: Proposal for 8-8-8 mode support [1].  
> > 
> > I only quickly went through the patches you sent and saying it's a
> > repost of the RFC is a bit of a lie. You completely ignored the state
> > tracking I was trying to do to avoid leaving the flash in 8D mode when
> > suspending/resetting the board, and I think that part is crucial. If I
> > remember correctly, we already had this discussion so I must say I'm a
> > bit disappointed.
> > 
> > Can you sync with Pratyush? I think his series [1] is better in that it
> > tries to restore the flash in single-SPI mode before suspend (it's
> > missing the shutdown case, but that can be easily added I think). Of
> > course that'd be even better to have proper state tracking at the SPI
> > NOR level.
> >   
> 
> [1] does soft reset on shutdown which should put it to reset default
> state of 1S-1S-1S mode (if thats the POR default)

Oh ok, looks like I didn't read the patch series carefully enough.

> 
> But, there is still one open question now that we are considering
> supporting stateful modes:
> 
> What to do with flashes that power up in 8D mode either due to factory
> defaults or if 8D mode NV bit is set? Do we say SPI NOR framework won't
> support such flashes?
> Auto discovery of such flashes is quite difficult as different flashes
> use different protocols for RDID cmd in 8D mode (address phase may or
> may not be present, dummy cycles vary etc) is almost impossible w/o any
> hint passed to the driver?

I don't know yet. Looks like we'll have to pass the part-id and default
mode for those flashes (part-name being a part-specific compatible, and
boot-up mode being an extra property). But maybe we can ignore that for
now and focus on flashes booting in single SPI mode first :P.