Task Switching in Linux

Task Switching in Linux

[Jaswinder Singh]

In Linux , If we assume that there are only 2 tasks A and B and both are
equal , this is correct or not :-

TASK A -> schedule -> switch_to -> TASK B -> schedule -> switch_to ->
schedule -> switch_to -> TASK A.

Thank you,

Best Regards,

Jaswinder.
--
These are my opinions not 3Di.

Re: Task Switching in Linux

[george anzinger]

Jaswinder Singh wrote:
> 
> In Linux , If we assume that there are only 2 tasks A and B and both are
> equal , this is correct or not :-
> 
> TASK A -> schedule -> switch_to -> TASK B -> schedule -> switch_to ->
> schedule -> switch_to -> TASK A.
> 
Heck no.  TASK A will be run until it either blocks or its time quantum
drops below TASK B's.  It does not matter how many times it calls
schedule(), 'cept its a darn waste of time.

Now, it you call sched_yield() your chances are higher, but still no
guarantee.  If you get the tasks out of SCHED_OTHER sched_yield() will
do the above (if they have the same priority).

George

Re: Task Switching in Linux

[Rick Hohensee]

Jaswinder Singh wrote:
>
> In Linux , If we assume that there are only 2 tasks A and B and both are
> equal , this is correct or not :-
>
> TASK A -> schedule -> switch_to -> TASK B -> schedule -> switch_to ->
> schedule -> switch_to -> TASK A.
>

If you mean "->" as "specifically calls" then that looks like cooperative
multi-tasking, which is what kernel threads AND the Linux userland
scheduler do. If an in-kernel thread doesn't call schedule, it keeps the
CPU. See the H3rL stuff in ftp://linux01.gwdg.de/pub/cLIeNUX/interim

Rick Hohensee
:; cLIeNUX /dev/tty11  21:00:45   /
:;d -d */
Cintpos/     boot/        device/      incoming/    owner/       temp/
Debian/      command/     floppy/      log/         source/
Linux/       configure/   guest/       lost+found/  subroutine/
NetBSD/      dev/         help/        mounts/      suite/