Hello.
This is a resend of v2 patch series, which didn't include explanation of
changes from previous version nor proper Cc to devicetree maintainers.
Changes from previous version (v1)
- devicetree binding and documentation is in a separate patch
- watchdog pinging is done differently. In previous version only
one counter (ID 1) was used to implement watchdog, and this counter
had to be restarted when pinging, leaving the system in a vulnerable
state for a brief period of time.
Now pinging is done by forcing immediate expiration event on counter
ID 0, and counter ID 1 is set to restart counting from the initial value
when this event happens. Thus counter ID 1, which is set to be watchdog
counter, is never disabled, and the system is never in this brief
vulnerable state
- reading current value of the counter now reads low 32 bits first, because
when this is done, the hardware puts the high 32 bits in flip-flops so as
to avoid software debouncing. This way get_counter_value always returns
correct result
Marek
