From mboxrd@z Thu Jan  1 00:00:00 1970
From: baruch@tkos.co.il (Baruch Siach)
Date: Thu, 28 Jun 2018 13:47:30 +0300
Subject: [PATCH 1/2] ARM: dts: armada388-clearfog: enable spi flash
In-Reply-To: <20180628103320.GE17271@n2100.armlinux.org.uk>
References: <2bc87e3e365c460943730ded86c8fcb1ff55a533.1530170016.git.baruch@tkos.co.il>
 <20180628095113.GC17271@n2100.armlinux.org.uk>
 <20180628100759.45clx3diuv6tk3e3@tarshish>
 <20180628103320.GE17271@n2100.armlinux.org.uk>
Message-ID: <20180628104730.te6ckfhncftvakjk@tarshish>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Russell,

On Thu, Jun 28, 2018 at 11:33:20AM +0100, Russell King - ARM Linux wrote:
> On Thu, Jun 28, 2018 at 01:07:59PM +0300, Baruch Siach wrote:
> > Thanks for reviewing.
> > 
> > On Thu, Jun 28, 2018 at 10:51:14AM +0100, Russell King - ARM Linux wrote:
> > > On Thu, Jun 28, 2018 at 10:13:35AM +0300, Baruch Siach wrote:
> > > > The SolidRun Armada 388 SOM has the SPI flash populated by default
> > > > unless the customer explicitly asks otherwise. Enable support by
> > > > default.
> > > 
> > > Are you sure about that - at least some of my boards do not have the
> > > SPI flash populated.
> > > 
> > > &spi1 {
> > >         /* The microsom has an optional W25Q32 on board, connected to CS0 */
> > > ...
> > > &spi1 {
> > >         /*
> > >          * Add SPI CS pins for clearfog:
> > >          * CS0: W25Q32 (not populated on uSOM)
> > 
> > I should update this line as well, I guess.
> > 
> > >          * CS1: PIC microcontroller (Pro models)
> > >          * CS2: mikrobus
> > >          */
> > > 
> > > I wouldn't have written "not populated on uSOM" without having first
> > > taken the uSOM off and physically checked.
> > 
> > Are these production SOMs or development/engineering samples?
> 
> I don't remember.
> 
> > Ilya Viten from the SolidRun commercial department told me that all production 
> > SOMs that were shipped to customers have the SPI flash populated.
> 
> Yes, Jon confirms that.
> 
> > This is just the default value for the common case.
> 
> Has it been tested with boards that don't have the SPI flash populated?
> That needs to happen to make sure that this doesn't cause a regression.

I don't have such a SOM handy. I can try to "produce" one next week.

I tested a modified DT with the 'reg' property set to 1 to simulate an 
unpopulated SPI flash. The driver probe seems to fail gracefully:

  m25p80 spi1.1: unrecognized JEDEC id bytes: ff, ff, ff

Is this test sufficient to ensure the safety of this patch?

baruch

-- 
     http://baruch.siach.name/blog/                  ~. .~   Tk Open Systems
=}------------------------------------------------ooO--U--Ooo------------{=
   - baruch at tkos.co.il - tel: +972.52.368.4656, http://www.tkos.co.il -