Re: [PATCH V4 1/2] mtd: spi-nor: Bindings for Cadence Quad SPI Flash Controller driver.

[Marek Vasut <marex@denx.de>]

On Friday, July 24, 2015 at 07:16:48 PM, Graham Moore wrote:
> On 07/24/2015 11:25 AM, Marek Vasut wrote:
> > On Friday, July 24, 2015 at 06:12:01 PM, Graham Moore wrote:
> >> On 07/24/2015 07:45 AM, Marek Vasut wrote:
> >>> On Monday, March 23, 2015 at 02:36:21 PM, Graham Moore wrote:
> >>>> Signed-off-by: Graham Moore <grmoore@opensource.altera.com>
> >>>> ---
> >>>> V2: Add cdns prefix to driver-specific bindings.
> >>>> V3: Use existing property "is-decoded-cs" instead of creating a
> >>>> duplicate, "ext-decoder".  Timing parameters are in nanoseconds, not
> >>>> master reference clocks.  Remove bus-num completely.
> >>> 
> >>> Hi!
> >>> 
> >>> do you plan to continue on this driver any soon? If not, I'd like to
> >>> take over the mainlining if you're not opposed. I'd like to see this
> >>> in mainline soon.
> >> 
> >> I don't have a problem with you taking over the mainlining.  I've been
> >> stalled on the micron quad protocol issue that Cyrille Pitchen has
> >> solved.
> > 
> > Do you mean the protocol change notification please?
> 
> Yes, that's what I mean, the protocol change notification.

OK

> >>> btw you'll need some kind of a variation on the attached two patches in
> >>> the next iteration.
> >> 
> >> I have a V5 with fixes for sparse and smatch issues.  Do you want me to
> >> post it and you add those two patches?  Alternatively I can incorporate
> >> them into my V5.
> > 
> > Please post those patches, yup, that'd be nice :)
> > 
> > Also, just squash those two micropatches into yours, no problem.
> > 
> > Be careful about the 0002 patch, see the commit message. I do NOT think
> > it is correct to pass the pdev->dev which represents the controller, not
> > the subnode, into spi_nor_scan() in cqspi_probe() . I might be wrong here
> > though, I am not sure. I think Brian or David would be much better people
> > to comment on this part.
> 
> I'm doing it the way fsl_quadspi.c does it.  It works on my Cyclone5
> test system with and without that 0002 patch.  So maybe I'll leave it
> out for now.

It indeed works, but that doesn't mean it is correct. The most simple test
to show you why I needed that 0002 patch would be for you to check if the
m25p,fast-read prop, which you are actually adding in this patchset to the
QSPI flash device, is set from within the spi-nor driver. It will never be
set because you're not passing the DT of_node pointer into the spi-nor driver.

> V5 patches on the way...

Thanks!

Best regards,
Marek Vasut