[nas] isolate input/output?

[Jon Trulson]

On Mon, 12 Mar 2007, Paul Fox wrote:

> erik wrote:
> > On Sun, Mar 11, 2007 at 01:33:33PM -0400, Paul Fox wrote:
> > > i'll let you know how it goes.
> >
> > Thanks, I'd really like to know if this works.
>
> success!  at least, lircd and nasd are running together.
>
> the last (tiny) hurdle was that i'd never noticed the "readwrite"
> parameter in nasd.conf.  as distributed, it's set to "no" for
> input devices, and "yes" for output devices.  changing the output
> device's value to "no" makes "cat /dev/dsp >/dev/null" succeed
> while nasd is running, where it gave EBUSY before.  and, as i
> said, lircd no longer fails to open the device either.
>
> anyone have any thoughts on why the defaults for the open() mode
> are the way they are?  they're as i wrote above in the config
> file (i.e., different for input and output).  worse, in the code,
> both devices are opened O_RDWR unless specified otherwise via the
> config.  i'm not sure i can imagine why one wouldn't default to
> O_RDONLY for input, and O_WRONLY for output.  (i note that this
> is the case for CYGWIN.)  at the very least, i think we should
> change the default nasd.conf values (which come from
> nasd.conf.eg, i assume).
>

<grumpyoldman>
         Because in the old days sonny, we used to open full-duplex
         audio devices read/write for no particular reason, and we
         LIKED IT! 
</grumpyoldman>

         :)

         I guess if you've never seen the grumpy old man on SNL, the
         above reference will be meaningless, sorry.

         Since we deal with input and output using seperate fd's, I
         can't really think of a good reason why we cannot simply
         always open output devices write-only and input devices read-only.

         I tried it hear, and it seems to work fine, at least for my
         basic output-only duties.  I am using an SB PCI (basic).  I do
         not recording via NAS.

         Think it's a safe default?

> erik -- thanks for all your help.  jon -- i'd go ahead and apply
> the patches from my and erik's mail.  they're no worse than
> what we had before, and almost certainly better.  :-)

   It is done.

-- 
Jon Trulson
mailto:jon at radscan.com 
#include <std/disclaimer.h>
"No Kill I" -Horta