On Tue, 2017-07-04 at 16:31 -0400, Tom Rini wrote:
On Tue, Jul 04, 2017 at 10:48:36PM +0300, Andy Shevchenko wrote:
On Tue, 2017-04-18 at 16:49 +0200, Lukasz Majewski wrote:
The code seems OK, but recently patches to add wdt-uclass has been posted:
http://patchwork.ozlabs.org/patch/751448/ http://patchwork.ozlabs.org/patch/751451/
Maybe it would be better to port this driver to the uclass from the very beginning?
I started looking into this direction and have a question: what the point to move to that class if it's broken from the beginning?
I really do not understand who on the earth would like to have wdt_stop() at ->probe() (the only two users do exactly that!).
It's quite possible I'm just missing something, but I don't see it. If the uclass is totally broken, can you make a suggestion on fixing it?
I need to think about it (I have an idea, though it requires time to try).
I can see a common case being "lets turn off the watchdog now" and not want a wdt until the full OS, but we must also fully support "U-Boot has and pets the Watchdog".
The users of WDT class now just disable watchdog. One of them, though, is using it for reset (not my case for Tangier).
So, it looks to me now as bikeshedding, otherwise where is the documentation which describes how this all stuff should work?
Can we go with the initial patch?
If it looks like some nightmare to fix the DM uclass, I suppose, but, is it really that bad off atm?
Not nightmare, but I really have not much time. Simon approached me couple of weeks ago asking what is the status with my patches for Tangier. That's why I found a slot to look at it.
So, I just have sent a v2 of this driver with old approach to move things forward, otherwise I have no idea when it could be done.