[Xenomai] Request for support of SIOCSHWTSTAMP

[Xenomai] Request for support of SIOCSHWTSTAMP

[Varun Deshpande]

Hello Folks,

As Xenomai-RTnet has combined, it is the solution for many hard realtime
systems but the main feature which is missing is SIOCSHWTSTAMP.
As i210 supports harware timestamping, it is crucial for performance
analysis. But it is not supported by rt_igb driver as SIOCSHWTSTAMP returns
"OPERATION NOT SUPPORTED"

I kindly request you guys to please incorporate this feature so that user
can get hardware timestamp in cmsg header by enabling SIOCSHWTSTAMP option
for rt_igb driver as many of the ptp, avb programs evaluation is done by
hardware timestamp only.

Best Regards

Varun Deshpande
Exchange Program Student,
ESIEE Paris
+33 760 349 859

Alma Mater
Electronics & Communication Engineering,
National Institute of Technology, Bhopal (INDIA)
+91 8965 080 507

Re: [Xenomai] Request for support of SIOCSHWTSTAMP

[Gilles Chanteperdrix]

On Fri, Apr 01, 2016 at 06:17:38PM +0200, Varun Deshpande wrote:
> Hello Folks,
> 
> As Xenomai-RTnet has combined, it is the solution for many hard realtime
> systems but the main feature which is missing is SIOCSHWTSTAMP.
> As i210 supports harware timestamping, it is crucial for performance
> analysis. But it is not supported by rt_igb driver as SIOCSHWTSTAMP returns
> "OPERATION NOT SUPPORTED"
> 
> I kindly request you guys to please incorporate this feature so that user
> can get hardware timestamp in cmsg header by enabling SIOCSHWTSTAMP option
> for rt_igb driver as many of the ptp, avb programs evaluation is done by
> hardware timestamp only.

Correct me if I am wrong, but to get meaningful hardware timestamps,
you need the machines to be synchronized with PTP, so, there is much
more work to get this working than just implement this ioctl. I
think we should start small, with software timestamps, as with
Xenomai low latencies, and the fact that determinism requires to not
fill the hardware queues, you should have some good precision already.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Request for support of SIOCSHWTSTAMP

[Varun Deshpande]

Hello Gilles Sir,

This plan is great ! I wish Xenomai Project a great Success !

Best Regards

Varun Deshpande
Exchange Program Student,
ESIEE Paris
+33 760 349 859

Alma Mater
Electronics & Communication Engineering,
National Institute of Technology, Bhopal (INDIA)
+91 8965 080 507

On Fri, Apr 1, 2016 at 8:50 PM, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Fri, Apr 01, 2016 at 06:17:38PM +0200, Varun Deshpande wrote:
> > Hello Folks,
> >
> > As Xenomai-RTnet has combined, it is the solution for many hard realtime
> > systems but the main feature which is missing is SIOCSHWTSTAMP.
> > As i210 supports harware timestamping, it is crucial for performance
> > analysis. But it is not supported by rt_igb driver as SIOCSHWTSTAMP
> returns
> > "OPERATION NOT SUPPORTED"
> >
> > I kindly request you guys to please incorporate this feature so that user
> > can get hardware timestamp in cmsg header by enabling SIOCSHWTSTAMP
> option
> > for rt_igb driver as many of the ptp, avb programs evaluation is done by
> > hardware timestamp only.
>
> Correct me if I am wrong, but to get meaningful hardware timestamps,
> you need the machines to be synchronized with PTP, so, there is much
> more work to get this working than just implement this ioctl. I
> think we should start small, with software timestamps, as with
> Xenomai low latencies, and the fact that determinism requires to not
> fill the hardware queues, you should have some good precision already.
>
> --
>                                             Gilles.
> https://click-hack.org
>