Re: [RFC 1/1][PATCH] POSIX SCHED_SPORADIC implementation for tasksand groups

[Randy Dunlap]

On Mon, 11 Aug 2008 16:45:06 +0200 Dario Faggioli wrote:

> diff --git a/init/Kconfig b/init/Kconfig
> index b678803..97fbdfc 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -323,6 +323,44 @@ config CPUSETS
>  config HAVE_UNSTABLE_SCHED_CLOCK
>  	bool
>  
> +config POSIX_SCHED_SPORADIC
> +	bool "SCHED_SPORADIC POSIX scheduling policy support"
> +	default n
> +	help
> +	  Enable the SCHED_SPORADIC POSIX scheduling policy, which mimics
> +	  the behaviour of the Sporadic Server, a well known mechanism
> +	  to schedule aperiodic or sporadic tasks in periodic-based fixed
> +	  priorit real-time systems.

	  priority

> +
> +	  To get details about the SCHED_SPORADIC scheduling policy refer to
> +	  "The Open Group Base Specifications Issue 6", IEEE Std 1003.1,
> +	  2004 Edition.
> +
> +	  [http://www.opengroup.org/onlinepubs/009695399/nfindex.html]
> +
> +config SCHED_SPORADIC_REPL_MAX
> +	int "Maximum number of pending recharges"
> +	range 1 1000
> +	depends on POSIX_SCHED_SPORADIC
> +	default 10
> +	help
> +	  Let you chose the maximum number of recharges that can be

	  Lets you choose

> +	  pending at a given time instant for a SCHED_SPORADIC task
> +	  (or task group if group scheduling is configured).
> +
> +	  The kernel data structure handling the SCHED_SPORADIC scheduling
> +	  policy for tasks and groups contains a buffer of pending
> +	  replenishment this size large.
> +	  For tasks it is dynamically allocated when SCHED_SPORADIC is set
> +	  as the scheduling policy of the task, while for group it is created
> +	  as soon as the group itself is allocated.
> +
> +	  Anyway, since this is going to cause some memory overhead, it is
> +	  wise to not set this to a too much high value.

	  wise not to set this to a very high value.	// or some other rewording

> +	  The best value depends on the characteristics of the application
> +	  you are going to use, but probably a value from 10 to 20 is
> +	  suitable for the most of the situations.
> +
>  config GROUP_SCHED
>  	bool "Group CPU scheduler"
>  	depends on EXPERIMENTAL


---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/