tech-kern archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: pcmaudio(4), for fun and games.

>>>>> "Jared" == Jared D McNeill <> writes:


    Jared>   1. pcmaudio_attach: change aprint_naive to "\n" 

Not sure I understand, it already has a "\n" in there ?

    Jared> 2. please implement pmf :) 

Sure, patch updated. Unfortunately, the parent devices don't do much
with the hooks other than register and deregister themselves. So I've
just followed suit. Patch updated, same url as before.

    Jared> 3. how does this interact with the i8254 timecounter?

I've assumed that the i8254 timecounter code just reads out the
counter values, and does not depend on the rate of interrupting of it
due to counter0. This means that the timecounter code is not affected
by re-programming any of the counters on the chip. It seems to work
fine. I've run xclock with kern.timecounter.hardware = i8254 and the
system clock seems to be running fine. Note that x86/isa/clock.c has
been tweaked for the case where this assumption is not true, ie; when
the system clock actually depends on this timecounter, in which case,
hardclock() is arranged to be called at hz, whatever the sampling
frequency ( == i8254 interrupting frequency ).

    Jared> 4. AUDIO_PROP_INDEPENDENT means you can playback and
    Jared> capture in different formats simultaneously; i don't think
    Jared> that's what you want to advertise :)

Thanks, fixed :-)

>>>>> "Antti" == Antti Kantee <> writes:


    Antti> If it a) works b) doesn't break anything c) could be
    Antti> remotely useful for someone, why not develop it in the
    Antti> tree?

Ok, as soon as someone other than me tries this on a regular setup and
nothing bad happens :-)

What I'm worried about is mucking with a reasonably well used port,
albeit on -current.

>>>>> "Michael" == Michael Lorenz <> writes:

    Michael> Yeah, I'm sure it will have more users than some of my
    Michael> drivers ;) ( Fujitsu AG-10e anyone? )

>>>>> "Brian" == Brian Buhrow <> writes:

    Brian>      I'm all for it going in the tree if it works.  That
    Brian> way, it will still work as the code moves along. And, I can
    Brian> think of several uses for it already.  -Brian

Cheers guys. Ok, lets see how this thing goes. As long as its some use
to someone. :-)


Home | Main Index | Thread Index | Old Index