On 14.05.2022 11:18:08, Matej Vasilevski wrote:
> > > +	timecounter_init(&priv->tc, &priv->cc, 0);
> > 
> > You here set the offset of the HW clock to 1.1.1970. The mcp driver sets
> > the offset to current time. I think it's convenient to have the current
> > time here....What do you think.
> 
> I actually searched in the mailing list and read your conversation with
> Vincent on timestamps starting from 0 or synced to current time.
> https://lore.kernel.org/linux-can/CAMZ6RqL+n4tRy-B-W+fzW5B3QV6Bedrko57pU_0TE023Oxw_5w@mail.gmail.com/

Thanks for looking up that discussion. Back than I was arguing for the
start from 0, but in the mean time I added TS support for the mcp251xfd,
which starts with the current time :)

> Then I discussed it with Pavel Pisa and he requested to start from 0.
> Reasons are that system time can change (NTP, daylight saving time,
> user settings etc.), so when it starts from 0 it is clear that it is
> "timestamp time".
> 
> Are there a lot of CAN drivers synced to system time? I think this would
> be a good argument for syncing, to keep things nice and cohesive in
> the CAN subsystem.
> 
> Overall I wouldn't want to block this patch over such a minutiae.

ACK

regards,
Marc

-- 
Pengutronix e.K.                 | Marc Kleine-Budde           |
Embedded Linux                   | https://www.pengutronix.de  |
Vertretung West/Dortmund         | Phone: +49-231-2826-924     |
Amtsgericht Hildesheim, HRA 2686 | Fax:   +49-5121-206917-5555 |