[asterisk-dev] [Code Review] Timer for Mach kernels

[Tilghman Lesher]

> On 2010-05-26 13:23:26, pabelanger wrote:
> > /trunk/res/res_timing_mach.c, lines 214-218
> > <https://reviewboard.asterisk.org/r/665/diff/1/?file=10306#file10306line214>
> >
> >     To be removed?

Those are the lines that I had to add to make the Mach timer test correctly on OS X.  Technically, we're not supposed to ack timers that have not actually fired, so the ack becomes a wait, with that code there.  Really, what should be happening is that the poll() should not return until the event actually fires.


- Tilghman


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/665/#review2087
-----------------------------------------------------------


On 2010-05-19 01:50:38, Tilghman Lesher wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/665/
> -----------------------------------------------------------
> 
> (Updated 2010-05-19 01:50:38)
> 
> 
> Review request for Asterisk Developers.
> 
> 
> Summary
> -------
> 
> While trying to verify the kqueue implementation on OS X, I coded an alternate timing implementation for OS X, since it's based on a Mach kernel.  The end result is that if you uncomment a few lines in the _ack function, you get perfect timing.  Unfortunately, this shows that poll(2) in OS X (or really, select(2), since ast_poll on OS X is coded over select(2)) is complete and utter crap.
> 
> 
> Diffs
> -----
> 
>   /trunk/configure UNKNOWN 
>   /trunk/configure.ac 263949 
>   /trunk/include/asterisk/autoconfig.h.in 263949 
>   /trunk/res/res_timing_mach.c PRE-CREATION 
> 
> Diff: https://reviewboard.asterisk.org/r/665/diff
> 
> 
> Testing
> -------
> 
> Verifies correctly on OS X, if you turn on the wait code in the _ack routine.
> 
> 
> Thanks,
> 
> Tilghman
> 
>