--On Friday, January 16, 2009 05:07:03 PM -0600 Nicolas Williams
wrote:
>> - we need to be able to tell when there are no remaining references to a
>> PAG, so we can free the corresponding kernel data structures
>
> Must this be done in kernel? It wouldn't be hard, of course.
Freeing the kern
--On Thursday, January 15, 2009 04:31:11 PM -0800 Garrett D'Amore
wrote:
> I *need* something for audio, but the use case is more like what I've
> described above -- I just need an ID that can be set in a parent process
> and inherited across children. I don't even need system calls for this,
>
--On Thursday, January 15, 2009 05:23:27 PM -0600 Nicolas Williams
wrote:
> I think the right design is this:
That looks reasonable, except that most of these interfaces also need to be
available from within kernel. OpenAFS, for example, would absolutely
depend on this:
- we need to be able
On Fri, Jan 16, 2009 at 06:31:21PM -0500, Jeffrey Hutzelman wrote:
> --On Friday, January 16, 2009 05:07:03 PM -0600 Nicolas Williams
> wrote:
>
> >>- we need to be able to tell when there are no remaining references to a
> >> PAG, so we can free the corresponding kernel data structures
> >
> >M
On Thu, Jan 15, 2009 at 04:31:11PM -0800, Garrett D'Amore wrote:
> Nicolas Williams wrote:
> >What needs this:
> >
> > - audio (so we can associate a default audio output device with
> > sessions -- the equivalent for audio of /dev/tty, but without
> > terminal session IDs and SVR3 process grou
David Powell wrote:
> Tony Nguyen wrote:
>> Dave,
>>
>> See my responses inline and the updated webrev at:
>>
>> http://cr.opensolaris.org/~tonyn/firewall13Jan2009-inc/
>> http://cr.opensolaris.org/~tonyn/firewall13Jan2009/
>
> Just responding to your mail; I'm still reviewing the code:
>
>>>
On Fri, Jan 16, 2009 at 06:09:05PM -0500, Jeffrey Hutzelman wrote:
> --On Thursday, January 15, 2009 04:31:11 PM -0800 Garrett D'Amore
> wrote:
>
> >I *need* something for audio, but the use case is more like what I've
> >described above -- I just need an ID that can be set in a parent process
>
On Fri, Jan 16, 2009 at 06:00:11PM -0500, Jeffrey Hutzelman wrote:
> --On Thursday, January 15, 2009 05:23:27 PM -0600 Nicolas Williams
> wrote:
> >I think the right design is this:
>
> That looks reasonable, except that most of these interfaces also need to be
> available from within kernel.
Tony Nguyen wrote:
> Dave,
>
> See my responses inline and the updated webrev at:
>
> http://cr.opensolaris.org/~tonyn/firewall13Jan2009-inc/
> http://cr.opensolaris.org/~tonyn/firewall13Jan2009/
Just responding to your mail; I'm still reviewing the code:
>> servinfo.c`uaddr2port:
>>
>>
