Subject:
|
Re: RFC: smaller task scheduler
|
Newsgroups:
|
lugnet.robotics.rcx.legos
|
Date:
|
Tue, 26 Mar 2002 06:42:04 GMT
|
Viewed:
|
1724 times
|
| |
|
|
In lugnet.robotics.rcx.legos, Joseph Woolley writes:
<snip>
> Overall: Cut the kernel size by about 300 bytes. Improved performance by
> reducing complexity.
This is exciting...
<snip>
> I did this by adding the file stm.c (simple task manager) I modified tm.h
> and tm.c to add #ifdef and added a CONF_SIMPLE_TM to the config.h So,
> the old tm and the new tm can be easily swapped.
>
> I modified the latest versions of the files from CVS, so I could just check
> them in. Or I could post the actual code for review. Or I could generate a
> diff and post for review. How should this be done?
Good work on this!
Since you have the config and simple swap we'll need to prove that
both versions are still working and I'd like a 2nd opinion (review)
to ensure that we still have what our tasker was intended to be,
but after that, there should be no harm in versioning the
changes in CVS. Sound OK?
> After I finish testing, I will have some time to work on TODOs.... like
> setting the RCX ID for lnp during download to the RCX or whatever is needed
> most.
I like your offer to work on firmdl and the RCX id stuff. However,
can I divert you for a brief 2nd opinion I need? I've read the
7 page writeup of the priority patch the university students did.
They are telling us, correctly, that we have opportunity in how
we handle reading of sensors and the ossociated prioritized
handling of same. They were able to dramatically improve the sensor
response and have shown that legOS can miss many fewer events. This
excites me as I am sure that most of us have experienced missed events
(line crossings, touches, etc.) Would you be able to read their writeup and
then render an opinion on wether we should consider their changes
or we should investigate another approach? Their code is at the site too
so you can review it if you would like. I'm studying our
kernel over the next couple of days to see if there is any other
obvious approach... theirs has some downsides. I would like to
hear what you think. (I put a link to their dat4 project web
on our DRAFT page, in the list of patches.)
No matter what the outcome you certainly could work on the RCX id
stuff if, as we identify tasks, that's where you want to work.
Regards,
Stephen
--
|
|
Message has 1 Reply:
 | | Re: RFC: smaller task scheduler
|
| "Stephen M Moraco" <stephen_moraco@agilent.com> wrote in message news:GtKIM4.Gws@lugnet.com... (...) <snip> (...) Sounds good. Now, who would like to provide a second opinion? BTW, as far as intended purpose of the task scheduler... I can see how (...) (23 years ago, 26-Mar-02, to lugnet.robotics.rcx.legos)
|
Message is in Reply To:
| | RFC: smaller task scheduler
|
| I have been working on some changes to the task scheduler (and therefore execi, kill, killall) and have reduced the size/complexity a bit. I think it is ready for use (although I am going to continue to test it). Here is a bit of info. about it: (...) (23 years ago, 25-Mar-02, to lugnet.robotics.rcx.legos)
|
4 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
