RE: [PATCH 6/6] ptp_ocelot: support 4 programmable pins

[Y.b. Lu]

Hi Vladimir and Richard,

> -----Original Message-----
> From: Vladimir Oltean <olteanv@xxxxxxxxx>
> Sent: Friday, March 20, 2020 9:21 PM
> To: Y.b. Lu <yangbo.lu@xxxxxxx>
> Cc: lkml <linux-kernel@xxxxxxxxxxxxxxx>; netdev <netdev@xxxxxxxxxxxxxxx>;
> David S . Miller <davem@xxxxxxxxxxxxx>; Richard Cochran
> <richardcochran@xxxxxxxxx>; Vladimir Oltean <vladimir.oltean@xxxxxxx>;
> Claudiu Manoil <claudiu.manoil@xxxxxxx>; Andrew Lunn <andrew@xxxxxxx>;
> Vivien Didelot <vivien.didelot@xxxxxxxxx>; Florian Fainelli
> <f.fainelli@xxxxxxxxx>; Alexandre Belloni <alexandre.belloni@xxxxxxxxxxx>;
> Microchip Linux Driver Support <UNGLinuxDriver@xxxxxxxxxxxxx>
> Subject: Re: [PATCH 6/6] ptp_ocelot: support 4 programmable pins
> 
> Hi Yangbo,
> 
> On Fri, 20 Mar 2020 at 12:42, Yangbo Lu <yangbo.lu@xxxxxxx> wrote:
> >
> > Support 4 programmable pins for only one function periodic
> > signal for now. Since the hardware is not able to support
> > absolute start time, driver starts periodic signal immediately.
> >
> 
> Are you absolutely sure it doesn't support absolute start time?
> Because that would mean it's pretty useless if the phase of the PTP
> clock signal is out of control.

I'm absolutely sure that absolute start time is not supported for periodic clock unless reference manual is wrong.
And I donât think we need to consider phase for periodic clock which is with a specified period.

But PPS is different. Pulse should be generated must after seconds increased.
The waveform_high/low should be configurable for phase and pulse width if supported.
This is supported by hardware but was not implemented by this patch. I was considering to add later.

In my one previous patch, I was suggested to implement PPS with programmable pin periodic clock function.
But I didnât find how should PPS be implemented with periodic clock function after checking ptp driver.
https://patchwork.ozlabs.org/patch/1215464/

Vladimir talked with me, for the special PPS case, we may consider,
if (req.perout.period.sec ==1 && req.perout.period.nsec == 0) and configure WAVEFORM_LOW to be equal to req_perout.start.nsec.

Richard, do you think is it ok?

And another problem I am facing is, in .enable() callback (PTP_CLK_REQ_PEROUT request) I defined.
                /*
                 * TODO: support disabling function
                 * When ptp_disable_pinfunc() is to disable function,
                 * it has already held pincfg_mux.
                 * However ptp_find_pin() in .enable() called also needs
                 * to hold pincfg_mux.
                 * This causes dead lock. So, just return for function
                 * disabling, and this needs fix-up.
                 */
Hope some suggestions here.
Thanks a lot.

> 
> I tested your patch on the LS1028A-RDB board using the following commands:
> 
> # Select PEROUT function and assign a channel to each of pins
> SWITCH_1588_DAT0 and SWITCH_1588_DAT1
> echo '2 0' > /sys/class/ptp/ptp1/pins/switch_1588_dat0
> echo '2 1' > /sys/class/ptp/ptp1/pins/switch_1588_dat1
> # Generate pulses with 1 second period on channel 0
> echo '0 0 0 1 0' > /sys/class/ptp/ptp1/period
> # Generate pulses with 1 second period on channel 1
> echo '1 0 0 1 0' > /sys/class/ptp/ptp1/period
> 
> And here is what I get:
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdrive.g
> oogle.com%2Fopen%3Fid%3D1ErWufJL0TWv6hKDQdF1pRL5gn4hn4X-r&amp;
> data=02%7C01%7Cyangbo.lu%40nxp.com%7Cbd3e65bdaabb4999737d08d7c
> cd17eee%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63720307
> 2457124468&amp;sdata=4D97D9ZoA%2FDJeSAN%2Fha4zNuZL6GwRLNxpNY
> QiLsOsyM%3D&amp;reserved=0
> 
> So the periodic output really starts 'now' just like the print says,
> so the output from DAT0 is not even in sync with DAT1.
> 
> > Signed-off-by: Yangbo Lu <yangbo.lu@xxxxxxx>
> > ---
> 
> Thanks,
> -Vladimir