[AG-TECH] AccessGrid 3: What information is available

[Robert Olson]

The discussions I recall had to do with preemptive scheduling patches  
for linux.

On May 24, 2005, at 12:30 PM, shudo at ni.aist.go.jp wrote:

> From: Colin Perkins <csp at csperkins.org>
>
>
>>>> And assuming that the OS was a realtime OS and thus could actually
>>>> provide timing assurances to applications.
>>>>
>>>
>>> I know as of a couple years ago the hardcore linux audio folks were
>>> doing exactly this; I don't know the current state of the realtime
>>> mods for linux.
>>>
>
>
>> but for rat we intentionally ignored this support, since it
>> exposes a number of security issues (you have to make the application
>> suid root and carefully drop privileges after enabling real-time
>> scheduling) and didn't seem to offer much benefit.
>>
>
> I guess the above mentioned one is POSIX.1b scheduling interface
> (i.e. sched_setscheduler(2) syscall).  AFAIK, it is not really related
> to real-time scheduling and it is just a meaning to give higher
> priority (1-99) to special processes than usual processes (priority  
> 0).
> Real "real-time scheduling" involves timing deadlines, but the POSIX
> scheduling interface does not concern the concept of deadline at all.
>
> Yes, it improved stability of MP3 playing on my Linux box several
> years ago as you mentioned and it requires root priviledges.
>
>   Kazuyuki Shudo
>   Grid Technology Research Center
>   National Institute of Advanced Industrial Science and Technology  
> (AIST)
>
>