[PATCH v2] watchdog: imx2: fix hang-up on boot for i.MX21, i.MX27 and i.MX31 SoCs

[Guenter Roeck]

On Wed, Feb 22, 2017 at 08:44:00PM +0100, Magnus Lilja wrote:
> Hi
> 
> On 17 January 2017 at 09:38, Uwe Kleine-König
> <u.kleine-koenig at pengutronix.de> wrote:
> > Hello Vladimir,
> >
> > On Tue, Jan 17, 2017 at 03:23:53AM +0200, Vladimir Zapolskiy wrote:
> >> Power down counter enable/disable bit switch is located in WMCR register,
> >> but watchdog controllers found on legacy i.MX21, i.MX27 and i.MX31 SoCs
> >> don't have this register. As a result of writing data to the non-existing
> >> register on driver probe the SoC hangs up, to fix the problem add more OF
> >> compatible strings and on this basis get information about availability of
> >> the WMCR register.
> >>
> >> Fixes: 5fe65ce7ccbb ("watchdog: imx2_wdt: Disable power down counter on boot")
> >> Reviewed-by: Guenter Roeck <linux at roeck-us.net>
> >> Tested-by: Magnus Lilja <lilja.magnus at gmail.com>
> >> Reviewed-by: Fabio Estevam <fabio.estevam at nxp.com>
> >> Cc: <stable at vger.kernel.org>
> >> Signed-off-by: Vladimir Zapolskiy <vz at mleia.com>
> >> ---
> >> Changes from v1 to v2:
> >> * removed i.MX27 and i.MX31 compatibles from the list, since it is expected
> >>   that their DTB files has a reference to the listed i.MX21 compatible,
> >>   many thanks to Uwe Kleine-König for the hint,
> >> * start the list of compatibles from an i.MX35 specific one, since it is
> >>   agreed that i.MX35 precedes i.MX25, thanks to Uwe Kleine-König for review,
> >> * added a comment describing any potential changes in future over the list
> >>   of watchdog compatible values,
> >> * added all collected tags to the commit message.
> >>
> >> Changes from RFC to v1:
> >> * replaced private data struct with a macro as suggested by Guenter,
> >
> > I didn't see the RFC, but I like a data struct better. It has some
> > overhead, but allows for easier expansion later, needs less casting and
> > is easier to understand for a reader (see below).
> >
> >> * updated the comment in the source code to reflect the change,
> >> * rearranged and simplified the logic of detecting WMCR presence,
> >> * pulled the fix out from the series with optional proposed DTS changes.
> 
> So, did we come to an agreement on a patch to be submitted? Perhaps I

Not to my knowledge and understanding.

Guenter

> missed the info on it being accepted.
>