globaledf

Gedare Bloom gedare at rtems.org
Tue Jun 25 18:49:08 UTC 2013


Should be fine to reuse some of the same scheduler functions from the
EDF scheduler, at least initially. There may need to be some more
thought put into locking/synchronization for SMP. However, I would not
be opposed to unifying the same functions with some #ifdef SMP to
control the use of SMP features, since the EDF and global EDF use
similar logic.
-Gedare

On Tue, Jun 25, 2013 at 2:39 PM, Sree Harsha Konduri
<sreehars at buffalo.edu> wrote:
> Hello,
>
> I've been going through the entry points of the simplesmp and edf scheduler
> and was thinking,
> for the globaledf scheduler entry points
> 1) if it is possible to re-use the enqueue for the rbtree
> 2) modify the schedule, initialize and extract based on the functionality of
> smp scheduler,
>
> Since the enqueue and update in globaledf uses only inserts into the rbtree,
> where as the schedule and initialize needs to be modified since i will use a
> chain of executing tasks.
>
> Your comments will be helpful.
>
> Thanks,
> Sree Harsha
>
>
>
> On Mon, Jun 24, 2013 at 8:59 AM, Gedare Bloom <gedare at rtems.org> wrote:
>>
>> Good I think you are on track.
>>
>> On Mon, Jun 24, 2013 at 11:38 AM, Sree Harsha Konduri
>> <sreehars at buffalo.edu> wrote:
>> > Hello,
>> >
>> > I have the globalEDF template up, so this week i will try to examine the
>> > current smp scheduler, in terms of scheduler and thread data structures
>> > and
>> > also functions, and come up with an algorithm or flow chart describing
>> > the
>> > entry points to the new scheduler.
>> >
>> > I will update the tasks document and also post the link to the blog at
>> > the
>> > earliest.
>> >
>> > Thanks,
>> > Sree
>> >
>> >
>> > On Sat, Jun 22, 2013 at 6:47 AM, Sree Harsha Konduri
>> > <sreehars at buffalo.edu>
>> > wrote:
>> >>
>> >> Hello,
>> >>
>> >> As disscussed earlier in the week, i have a GlobalEDF scheduler
>> >> template
>> >> with the same functionality of the simplesmp scheduler.
>> >> I've tested it with some of the smp tests using rtems-testing.
>> >>
>> >> In order to test with rtems-schedsim do i have to configure
>> >> rtems-schedsim
>> >> again from scratch? since there are changes to the cpukit with added
>> >> header
>> >> files.
>> >>
>> >> Thanks,
>> >> Harsha
>> >>
>> >>
>> >>
>> >> On Fri, Jun 21, 2013 at 3:38 PM, Joel Sherrill
>> >> <Joel.Sherrill at oarcorp.com>
>> >> wrote:
>> >>>
>> >>> You have added a .h file. You need to run bootstrap -p and bootstrap
>> >>> in
>> >>> cpukit.
>> >>>
>> >>> Do that followed by a build from scratch and see if it helps
>> >>>
>> >>> Sree Harsha Konduri <sreehars at buffalo.edu> wrote:
>> >>>
>> >>> I do have auto tools in my Path. After i do configure with
>> >>> --enable-maintainer-mode i cant see the Makefile.in refreshing, or new
>> >>> headers being included.
>> >>> So i ran bootstrap with -c to clear all the auto tools generated
>> >>> files.
>> >>> Now when i re run bootstrap with -p only preinstall.am gets generated
>> >>> and
>> >>> after configure with --enable-maintainer-mode there is no Makefile.in
>> >>> generated?
>> >>>
>> >>> Thanks,
>> >>> Harsha
>> >>>
>> >>>
>> >>> On Fri, Jun 21, 2013 at 6:05 AM, Ralf Corsepius
>> >>> <ralf.corsepius at rtems.org> wrote:
>> >>>>
>> >>>> On 06/21/2013 03:07 AM, Gedare Bloom wrote:
>> >>>>>
>> >>>>> Cc:rtems-devel
>> >>>>>
>> >>>>> When modifying the build system i.e. Makefile.am files, you need to
>> >>>>> rerun the bootstrap script and maybe with the -p flag if you add new
>> >>>>> header files. Run bootstrap in the nearest directory to the modified
>> >>>>> Makefile.am that contains a configure.ac <http://configure.ac> file.
>> >>>>> Then recompile rtems.
>> >>>>
>> >>>>
>> >>>> Make sure to have RTEMS autoconf and automake in $PATH,
>> >>>> then pass --enable-maintainer-mode to configure
>> >>>> then running make will automatically take care about this in most
>> >>>> cases.
>> >>>>
>> >>>> Ralf
>> >>>>
>> >>>
>> >>
>> >
>
>



More information about the devel mailing list