Re: [gpsd-dev] gpsd processor load

[Håkan Johansson]

Loose thoughts:

On Thu, 31 Oct 2013, Gary E. Miller wrote:

> > I think select() will return a ready fd with no data bytes from
> > read() when a control line changes.
>
> Thn the PPS thread would wake up on every character as well.  Not good.

It would also mean that the PPS thread might steal real data when read() 
does return something useful.  Also bad.

Question: the reason for having the PPS thread in the first place is that 
not that one cannot mix the select() and TIOCMIWAIT waiting in the main 
thread?

If the BSDs really do tell about line changes by a 'spurious' select() 
event, then perhaps one should not use a separate thread at all and put 
the line change check (i.e. TIOCMGET) in the main thread.

This might not work for the short (microsecond) pulses.  Those perhaps can 
be gotten most of the time by assuming them on really spurious select 
returns, i.e. when there really was no data to get.

Cheers,
Håkan