[LinuxPPS] [RFC] time_pps_createbyname()

[Rodolfo Giometti]

On Mon, Aug 06, 2007 at 01:00:05PM +0200, Frank Kardel wrote:

> Yes - you got the idea (which I hope will save us all much work).
> 
> I'd like the idea to have the PPS source by their name in sysfs.
> It would be easier for the user to pick the right one instead of
> having to figure out whether pps1 is ttyS0 or is it pps0 ? That would
> again depend on configured devices and be prone to changes in the
> environment. Maybe /dev/pps-<name><number> could work.

Mmm...	so you prefere having /sys/class/pps/ttyS0/... rather than
/sys/class/pps/01/...?

> I don't know about the policies within Linux wrt/ sysfs, but maybe
> a working PPS device is fine with them. After all, it would be the real
> device anyway.

I think that to get Linux inclusion we should implement one char
device per PPS source with a proper ioctl() interface to exachange
data with the userland. Files under /sys/ can be kept and used for
other non-kernel tasks as looking for PPS source connected with
/dev/ttyS1, for instance.

On Mon, Aug 06, 2007 at 01:30:48PM +0200, Frank Kardel wrote:

> on an afterthought I seem to recall the /dev/pps0 is the name expected by
> ATOM. Thus it would make sense to name the PPS devices differently to
> /dev/pps* so that ATOM can also use symlinks to get pointed to the correct
> PPS device.

This is not a kernel problem since files under /dev/ are created into
userland by the sysadmin or udev. :)

Ok, with this assumptions I think I should change LinuxPPS again in
order to implement char devices... I hope that this last change may
allow NTP and kernel inclusion! ;)

Thanks a lot,

Rodolfo

-- 

GNU/Linux Solutions                  e-mail:    giometti at enneenne.com
Linux Device Driver                             giometti at gnudd.com
Embedded Systems                     		giometti at linux.it
UNIX programming                     phone:     +39 349 2432127