[PATCH 09/11] ice: implement dpll interface to control cgu

Vadim Fedorenko vadim.fedorenko at linux.dev
Thu Jul 27 03:28:12 PDT 2023


On 26/07/2023 22:11, Kubalewski, Arkadiusz wrote:
>> From: Jiri Pirko <jiri at resnulli.us>
>> Sent: Wednesday, July 26, 2023 8:38 AM
>>
> 
> [...]
>   
>>>>>>>
>>>>>>> Just to make it clear:
>>>>>>>
>>>>>>> AUTOMATIC:
>>>>>>> - inputs monitored, validated, phase measurements available
>>>>>>> - possible states: unlocked, locked, locked-ho-acq, holdover
>>>>>>>
>>>>>>> FREERUN:
>>>>>>> - inputs not monitored, not validated, no phase measurements available
>>>>>>> - possible states: unlocked
>>>>>>
>>>>>> This is your implementation of DPLL. Others may have it done
>>>>>> differently. But the fact the input is monitored or not, does not make
>>>>>> any difference from user perspective.
>>>>>>
>>>>>> When he has automatic mode and does:
>>>>>> 1) disconnect all pins
>>>>>> 2) reset state    (however you implement it in the driver is totaly up
>>>>>> 		   to the device, you may go to your freerun dpll mode
>>>>>> 		   internally and to automatic back, up to you)
>>>>>> -> state will go to unlocked
>>>>>>
>>>>>> The behaviour is exactly the same, without any special mode.
>>>>>
>>>>> In this case there is special reset button, which doesn't exist in
>>>>> reality, actually your suggestion to go into FREERUN and back to AUTOMATIC
>>>>> to pretend the some kind of reset has happened, where in reality dpll went
>>>>> to
>>>>> FREERUN and AUTOMATIC.
>>>>
>>>> There are 3 pin states:
>>>> disconnected
>>>> connected
>>>> selectable
>>>>
>>>> When the last source disconnects, go to your internal freerun.
>>>> When some source gets selectable or connected, go to your internal
>>>> automatic mode.
>>>>
>>>
>>> This would make the driver to check if all the sources are disconnected
>>> each time someone disconnects a source. Which in first place is not
>>> efficient, but also dpll design already allows different driver instances
>>> to
>>> control separated sources, which in this case would force a driver to
>>> implement
>>> additional communication between the instances just to allow such hidden
>>> FREERUN mode.
>>> Which seems another argument not to do this in the way you are proposing:
>>> inefficient and unnecessarily complicated.
>>>
>>> We know that you could also implement FREERUN mode by disconnecting all
>>> the
>>> sources, even if HW doesn't support it explicitly.
>>>
>>> >From user perspactive, the mode didn't change.
>>>>
>>>
>>> The user didn't change the mode, the mode shall not change.
>>> You wrote to do it silently, so user didn't change the mode but it would
>> have
>>> changed, and we would have pretended the different working mode of DPLL
>> doesn't
>>> exist.
>>>
>>> >From user perepective, this is exacly the behaviour he requested.
>>>>
>>>
>>> IMHO this is wrong and comes from the definition of pin state DISCONNECTED,
>>> which is not sharp, for our HW means that the input will not be considered
>>> as valid input, but is not disconnecting anything, as input is still
>>> monitored and measured.
>>> Shall we have additional mode like PIN_STATE_NOT_SELECTABLE? As it is not
>>> possible to actually disconnect a pin..
>>>
>>>>
>>>>> For me it seems it seems like unnecessary complication of user's life.
>>>>> The idea of FREERUN mode is to run dpll on its system clock, so all the
>>>>> "external" dpll sources shall be disconnected when dpll is in FREERUN.
>>>>
>>>> Yes, that is when you set all pins to disconnect. no mode change needed.
>>>>
>>>
>>> We don't disconnect anything, we used a pin state DISCONNECTED as this
>>> seemed
>>> most appropriate.
>>>
>>>>
>>>>> Let's assume your HW doesn't have a FREERUN, can't you just create it by
>>>>> disconnecting all the sources?
>>>>
>>>> Yep, that's what we do.
>>>>
>>>
>>> No, you were saying that the mode doesn't exist and that your hardware
>>> doesn't
>>> support it. At the same time it can be achieved by manually disconnecting
>>> all
>>> the sources.
>>>
>>>>
>>>>> BTW, what chip are you using on mlx5 for this?
>>>>> I don't understand why the user would have to mangle state of all the pins
>>>>> just
>>>>> to stop dpll's work if he could just go into FREERUN and voila. Also what
>>>>> if
>>>>> user doesn't want change the configuration of the pins at all, and he just
>>>>> want
>>>>> to desynchronize it's dpll for i.e. testing reason.
>>>>
>>>> I tried to explain multiple times. Let the user have clean an abstracted
>>>> api, with clear semantics. Simple as that. Your internal freerun mode is
>>>> just something to abstract out, it is not needed to expose it.
>>>>
>>>
>>> Our hardware can support in total 4 modes, and 2 are now supported in ice.
>>> I don't get the idea for abstraction of hardware switches, modes or
>>> capabilities, and having those somehow achievable through different
>>> functionalities.
>>>
>>> I think we already discussed this long enough to make a decision..
>>> Though I am not convinced by your arguments, and you are not convinced by
>>> mine.
>>>
>>> Perhaps someone else could step in and cut the rope, so we could go further
>>> with this?
>>
>> Or, even better, please drop this for the initial patchset and have this
>> as a follow-up. Thanks!
>>
>>
> 
> On the responses from Jakub and Paolo, they supported the idea of having
> such mode.
> 
> Although Jakub have asked if there could be better name then FREERUN, also
> suggested DETACHED and STANDALONE.
> For me DETACHED seems pretty good, STANDALONE a bit too far..
> I am biased by the FREERUN from chip docs and don't have strong opinion
> on any of those..
> 
> Any suggestions?

It looks like we have a kind of split-brain situation, and my thoughts 
are following:
Even though right now we don't have any hardware supporting 
freerun/standalone mode, I do really like the idea to have it. It will 
be used in monitoring implementations where we refer to internal 
oscillator (Rb/Cs) as a source of truth to compare with the signal on 
the other pins. We can name it DETACHED if it sounds better.

> 
> Thank you!
> Arkadiusz
> 
> [...]




More information about the linux-arm-kernel mailing list