[Gnash-dev] Re: 1-millisecond sleeps

[John Gilmore]

> >> But whatever, using usleep(1000) should be fine in any case.
> 
> MG> It depends what you are trying to achieve. It's ok if you want to
> MG> suspend program execution for at least 1 millisecond (where "at least"
> MG> may mean "far more than"!)
> 
> What I wanted was: allow a task switch, but don't exceed 1000 HZ on
> systems that would be able to.

Usleep(1000) is a terrible way to allow a task switch.  The Linux
kernel will do a task switch whenever it's appropriate.  The buzzword
is "fully preemptible".  It's not like a 1980's Macintosh, where
applications had to explicitly task switch or the kernel would never
regain control.  You don't need to tell Unix or Linux; it's automatic.

So what are you *really* trying to do here?  Is there some other task
that you want the kernel to run?  If so, send it a message; write a byte
to a socket or FIFO that that task is listening on; etc.

What you *do* need to tell the Linux kernel when you sleep is "when is
the next time my current task will have some useful work to do?".  You
tell it that by the duration of your sleep request.  If you don't know
when there'll be more work to do, then don't sleep at all.  Instead,
do a system call like read() or select() or poll(), which will put you
to sleep until something interesting happens on one of your file
descriptors.

The whole Linux community is going around cleaning up old code that
does things like wake up 1000 times a second for no reason.  Besides
the inefficiency and ugliness of that approach, it also burns power
for no reason.  Many CPUs will slow their clocks and save power when
lightly loaded.  The OLPC will actually power off its CPU chip if it
knows you'll be sleeping for at least a few seconds.  It can't do that
for a thousandth of a second -- and it has no way to know that your
code will go right back to sleep when awakened.  

Here are some examples of other people cleaning up such code recently:

  http://www.linuxpowertop.org/
  http://www.linuxpowertop.org/known.php
  https://ols2006.108.redhat.com/reprints/jones-reprint.pdf
     "Why userspace sucks -- Or 101 Really Dumb Things Your App Shouldn't Do"
  
http://blogs.gnome.org/desrt/2006/07/27/burning-cpu-and-battery-on-the-gnome-desktop/
  http://dev.laptop.org/ticket/82
     "Look for misbehaving userspace apps."

Let's write gnash cleanly and efficiently, rather than using kludges
that others will just have to clean up after us.

        John