Re: [PATCH v4 00/16] OMAP2+ OneNAND driver update

[Ladislav Michl <ladis@linux-mips.org>]

On Tue, Nov 14, 2017 at 01:48:53PM -0800, Tony Lindgren wrote:
> Hi,
> 
> * Ladislav Michl <ladis@linux-mips.org> [171111 21:13]:
> > Hi there,
> > 
> > this patch serie updates OMAP2+ OneNAND driver to the present times, making
> > it fully DT configurable, using libgpio and dmaengine apis.
> > 
> > Please note that unlike previous driver version, which basically ignored
> > DT specified timings, this one relies on it, so it is important to get
> > it right in your DT (dumping it from previous kernel version).
> > 
> > In case synchronous timings is requested, it is okay to specify timings
> > for the slowest chip ever used for you board as it is evetually optimized
> > after chip probe.
> > 
> > Original driver used DMA only if user specified r/b gpio in platform
> > data, now DMA is used unconditionally and PIO mode is used as fallback.
> > 
> > In case anyone wants to give it a try, few DT related changes are needed:
> > - onenand node needs 'ti,omap2-onenand' compatible (for mailine boards this is
> >   done in patch 2)
> > - to use R/B pin, rb-gpios needs to be specified (for n900 and n8x0 this is
> >   done in patch 16, however patch lacks pinmux configuration.
> > 
> > Most notable changes from previous version:
> > - added dmaengine patches by Peter Ujfalusi
> > - added dt bindings documentation
> > - added cleanup patches
> > - DMA enabled by default
> > 
> > Also note that driver will fail probe OneNAND chip after patch 13 and start
> > working again after patch 14.
> > 
> > Testing and benchmarking very welcome.
> > 
> > Depends on "memory: omap-gpmc: Make 'bank-width' property optional"
> > https://patchwork.kernel.org/patch/10043259/
> 
> Thanks for doing this. I'm adding Aaro to Cc here, maybe he can test
> n8x0 as my n800 in my rack is not currently booting.
> 
> I've tested this series briefly on n900 by writing a 10MB file to
> onenand and comparing it to the original, and it seems to work so
> for patches 1 to 15 you can add:
> 
> Tested-by: Tony Lindgren <tony@atomide.com>
> 
> The last patch has a build error for n8x0, and in general we really
> should test this series on n8x0 before merging as the configuration
> is a little bit different. So I'd say we need an ack from Aaro there.

Just sent a quick update on that one.

> I've also acked the mach-omap2 touching patches so when others are
> happy this can be merged as a single set via the MTD tree.

While it is running for a few days on IGEPv2 boards in DMA mode without
issues, I'm still unsure whenever we should leave DMA enabled by default.
On the other side, it _was_ used on n900, but disabled later with device
tree introducion. So n8x0 is to be verified. I'll happily leave that
decision to others :-)

Best regards,
	ladis