[LITMUS^RT] Question about the RT_F_XXX flags

[Glenn Elliott]

On Oct 3, 2012, at 6:41 PM, Manohar Vanga <mvanga at mpi-sws.org> wrote:

> Hey Glenn,
> 
> Any suggestions on how to replace this bit of code? I'm not entirely sure how to check for this condition (ie. task was suspended due to semaphore wait).
> 
> Thanks!
> /manohar
> 
> On Oct 4, 2012, at 12:02 AM, Glenn Elliott wrote:
> 
>> 
>> On Oct 3, 2012, at 5:40 PM, Manohar Vanga <mvanga at mpi-sws.org> wrote:
>> 
>>> Hey all,
>>> 
>>>>> We could just replace it with a single bit field in the rt_param struct.  There are already many "int myvar:1" fields in rt_param.
>>>> 
>>>> Indeed. Patches welcome (after RTAS).
>>> 
>>> I am patching this up at the moment and I notice that another flag RT_F_EXIT_SEM exists to indicate that a task was suspended due to a semaphore being acquired. There are checks for this flag in sched_cedf.c and sched_gsn_edf.c however the flag itself if never set anywhere. Can I assume this is also an artifact from an older version and get rid of it? Or does it serve a purpose I am unable to see (with my grepping eyes :D)?
>>> 
>>> Thanks!
>>> /manohar
>> 
>> Looks like you can nuke it.  There was a more complicated version of FMLP that Björn replaced two or three years ago.  This might be left over from back then.


I'll need to take a closer look, but my guess is that we can just remove all code that acts on RT_F_EXIT_SEM; it's all dead code.  Just assume RT_F_EXIT_SEM is always false and remove all code paths where RT_F_EXIT_SEM is true.

-Glenn