Re: [PATCH v3 1/2] drivers/watchdog: Add optional ASPEED device tree properties

[Guenter Roeck <linux@roeck-us.net>]

On Thu, Jul 06, 2017 at 02:27:18PM -0500, Christopher Bostic wrote:
> 
> 
> On 7/6/17 9:35 AM, Rob Herring wrote:
> >On Thu, Jun 29, 2017 at 08:04:17AM -0700, Guenter Roeck wrote:
> >>On Thu, Jun 29, 2017 at 08:39:59AM -0500, Christopher Bostic wrote:
> >>>
> >>>On 6/28/17 10:33 PM, Guenter Roeck wrote:
> >>>>On 06/28/2017 05:28 PM, Christopher Bostic wrote:
> >>>>>Describe device tree optional properties:
> >>>>>
> >>>>>   * aspeed,arm-reet - ARM CPU reset on signal
> >>>>>   * aspeed,no-soc-reset - SOC reset on signal
> >>>>>   * aspeed,no-sys-reset - System reset on signal
> >>>>>   * aspeed,interrupt - Interrupt CPU on signal
> >>>>>   * aspeed,external-signal - Generate external signal (WDT1 and WDT2
> >>>>>only)
> >>>>>   * aspeed,alt-boot - Boot from alternate block on signal
> >>>>>
> >>>>>Signed-off-by: Christopher Bostic <cbostic@linux.vnet.ibm.com>
> >>>>>---
> >>>>>v3 - Invert soc and sys reset to 'no' to preserve backwards
> >>>>>      compatibility.  SOC and SYS reset will be set by default
> >>>>>      without any optional parameters set
> >>>>>v2 - Add 'aspeed,' prefix to all optional properties
> >>>>>    - Add arm-reset, soc-reset, interrupt, alt-boot properties
> >>>>>---
> >>>>>  .../devicetree/bindings/watchdog/aspeed-wdt.txt    | 24
> >>>>>++++++++++++++++++++++
> >>>>>  1 file changed, 24 insertions(+)
> >>>>>
> >>>>>diff --git a/Documentation/devicetree/bindings/watchdog/aspeed-wdt.txt
> >>>>>b/Documentation/devicetree/bindings/watchdog/aspeed-wdt.txt
> >>>>>index c5e74d7..6f18005 100644
> >>>>>--- a/Documentation/devicetree/bindings/watchdog/aspeed-wdt.txt
> >>>>>+++ b/Documentation/devicetree/bindings/watchdog/aspeed-wdt.txt
> >>>>>@@ -8,9 +8,33 @@ Required properties:
> >>>>>   - reg: physical base address of the controller and length of memory
> >>>>>mapped
> >>>>>     region
> >>>>>  +Optional properties:
> >>>>>+   Signal behavior - Whenever a timeout occurs the watchdog can be
> >>>>>programmed
> >>>>>+   to generate/not generate 6 types of signals:
> >>>>>+
> >>>>>+ - aspeed,arm-reset:    If property is present then reset ARM CPU only.
> >>>>>+            If not specified no ARM CPU reset is done.
> >>>>>+
> >>>>>+ - aspeed,no-soc-reset: If property is present then do not reset SOC.
> >>>>>+            If not specified then SOC reset is done.
> >>>>>+
> >>>>>+ - aspeed,no-sys-reset: If property is present then do not reset
> >>>>>system.
> >>>>>+            Typcally used in tandem with 'aspeed-external-signal'
> >>>>Is this correct ? As I understand the datasheet, it could also used in
> >>>>tandem with
> >>>>aspeed,interrupt.
> >>>True, that should be documented.  Will add that.
> >>>>>+            If not specified then system reset is done.
> >>>>>+
> >>>>I'll leave it up to Rob to decide, but for my part I don't understand
> >>>>no-soc-reset.
> >>>As the aspeed watchdog driver exists prior to this change an SOC reset is
> >>>done by
> >>>default.   In order to preserve backwards compatibility a missing optional
> >>>property
> >>>should result in default behavior.  I however need to be able to specify
> >>>that SOC
> >>>reset be disabled in some way.  This goes back to our discussion about why
> >>>we'd
> >>>ever want to disable SYSTEM reset in the first place.  Same reasoning
> >>>applies for
> >>>SOC reset.
> >>>
> >>>>I would instead use four properties.
> >>>>
> >>>>    aspeed,arm-reset
> >>>>    aspeed,soc-reset
> >>>Per my response above I think it should remain as aspeed,no-soc-reset due to
> >>>backwards compatibility requirements.
> >>The same can be accomplished with "aspeed,no-reset", which would avoid the, in
> >>my opinion, awkward "no-{sys,soc}-reset" poperties.
> >>
> >>>>    aspeed,sys-reset (which is the default)
> >>>Again as per our discussion yesterday I need some way to specify how system
> >>>reset is to  be done.  For backwards compatibility, a lack of parameter here
> >>>would
> >>>result in a system reset being configured.  Only way to indicate to the
> >>>driver
> >>>that no system reset is to be done is to indicate 'no' system reset in the
> >>>optional
> >>>parameter.
> >>Or "aspeed,no-reset".
> >>
> >>>>    aspeed,no-reset
> >>>This parameter seems ambiguous as we could be doing a 'no system reset'
> >>>or a 'no SOC reset' in theory.
> >>{arm,soc,sys}-reset are mutually exclusive per datasheet, and there is a
> >>separate configuration bit which enables the reset in the first place.
> >>I don't see how that is ambiguous.
> >>
> >>Anyway, I don't think we are making any progress here. Let's wait for
> >>guidance from Rob.
> >I tend to agree with Guenter.
> >
> >Maybe using the form 'aspeed,reset-type = "cpu|soc|system"' would be
> >more aligned to the type of reset being mutually exclusive.
> 
> OK I'll update to this method.
> 

Maybe add 'aspeed,reset-type = "none"' ? The default (no property) would
then be "system".

Thanks,
Guenter