Re: [PATCH 4/10] (Was: [PATCH 4/4]) MIPS: bmips: Convert bcm63xx_wdt to use WATCHDOG_CORE

[Simon Arlott <simon@fire.lp0.eu>]

On 22/11/15 02:32, Guenter Roeck wrote:
> On 11/21/2015 01:44 PM, Simon Arlott wrote:
>> On 21/11/15 21:32, Guenter Roeck wrote:
>>> this is really doing a bit too much in a single patch.
>>> Conversion to the watchdog infrastructure should probably be
>>> the first step, followed by further optimizations and improvements.
>>

I've split patch 4 up into 7 patches, which will be patches 4/10..10/10
in this thread.

Instead of using bcm63xx_timer_register in #ifdefs, I'll remove most of
bcm63xx_timer because it's only used by the watchdog.

>>> We have some infrastructure changes in the works which will move
>>> the need for soft-timers from individual drivers into the watchdog core.
>>> Would this possibly be helpful here ? The timer-driven watchdog ping
>>> seems to accomplish pretty much the same.
>>
>> There is no need for a software timer. This is not a timer-driven
>> watchdog ping, there is an unmaskable timer interrupt when the watchdog
>> timer has less than 50% remaining.
>>
> Ok. Maybe I got confused by the interrupt-triggered watchdog ping.
> I'll have to look into that much more closely; it is quite unusual
> and complex. The explanation is also not easy to understand.
> What does "The only way to stop this interrupt" mean ? Repeatedly

The interrupt is level triggered at less than 50% of the time remaining.
Unless the watchdog is stopped or restarted, the interrupt will not stop
occurring.

> triggering the interrupt in 1/2, 1/4, 1/8 of the remaining time is
> really odd.

It's a bit odd but there's no way to ack the interrupt. This seemed like
the best approach without adding the complexity of a software timer or
trying to mask and unmask the timer interrupt. I don't  want to ignore
the interrupt entirely because I'd like to know if the watchdog is going
to cause a reboot.

> On side note, the subject tag should be "watchdog:", not "MIPS:".

Fixed.

-- 
Simon Arlott