| |
Greetings,
This is a follow up to earlier postings discussing the development status of
BrickOS. While the Sourceforge project has not had much activity of late,
I've come across various modifications that, unfortunately, have never made
it upstream. Even more unfortunate is that as time passes, it seems a few
more RCX sites go offline.
After finding BrickOS modifications or patches, there are at least two key
issues: 1) the changes are not always against the latest version of BrickOS
and 2) different sets of changes might modify the same section of code.
I have assembled a collection of patches that I have used to update my
version of BrickOS. Some may be more familiar than others, but I have
listed them all here for completeness. The numbers correspond to the
numbers in the diff file name in the archive file.
0. There is a slight difference between the 0.9.0 source release and the
current Sourceforge CVS
1 - 5. Dr. Hoenicke's patches ( http://hoenicke.ath.cx/rcx/brickOS.html ),
including gcc33, highmem, performance, Makefile, and tcpcomm
6. Carl Troein's signedness patch (
http://news.lugnet.com/robotics/rcx/legos/?n=3987&t=i&v=a )
7. Outstanding Sourceforge patch - enable Makelx to handle longer file names
8. Outstanding Sourceforge patch - update entrypoint specification to the S9
record (Lego.NET)
9. Outstanding Sourceforge patch - address compatibility issues in
non-Linux/Cygwin environments
10. Outstanding Sourceforge patch - serial port init fix for Mac OS X
11. Mark Riley ( http://news.lugnet.com/org/ca/rtltoronto/?n=14996 ) - LDCC
12. Brown.edu ( http://www.cs.brown.edu/courses/cs148/old/2004fall/brickOS/
and
http://cs.brown.edu/courses/csci1480/old/2005/brickOS/quickstart.html ) -
LNP printf capabilities
13. Mike LeSauvage (
http://tarpit.rmc.ca/mcgaughey/eee243/labs/resources/legoshrink.html ) - LNP
communication utility; made modifications to reuse existing code and added
support for TCP communication (such as used by BrickEmu). Please note that
I have not been able to test these changes for all platforms.
14. Taiichi Yuasa ( http://xslisp.com/ ) - miscellaneous patches to BrickOS
that are packaged along with xsLisp
I have uploaded this collection of patches to
http://sourceforge.net/tracker2/index.php?func=detail&aid=2601672&group_id=58151&atid=486699
This is by no means an exhaustive inclusion of everything that is out there,
though what remains might require more effort to incorporate.
* Brown.edu also includes "reliable LNP" code
* Powolny had several postings in 2008 on serial communication
* ECU.edu (via archive.org -
http://web.archive.org/web/20070624025341/http://www.cs.ecu.edu/~hochberg/fall2002/brickOS/index.html )
- task prioritization, task scheduling, memory allocation, designate sensor
as real-time
* NCSU.edu ( http://moss.csc.ncsu.edu/~mueller/rt/rt05/readings/g1/ ) -
task scheduling, prioritization, etc.
While probably too large to include in a regular distribution, Olaf Christ's
TCP/IP-enabled RCX is interesting (via archive.org -
http://web.archive.org/web/20040324131317/http://users.informatik.haw-hamburg.de/~christ_o/)
As a side note, several programming languages are available for developing
programs that run on BrickOS.
* XS [Lisp] ( http://www.xslisp.com/ )
* Esterel ( http://www.emn.fr/x-info/lego/ and
http://www.emn.fr/x-info/lego/esterel-tools/ )
* Lustre ( http://www.emn.fr/x-info/lego/ ) - The Lustre compiler is no
longer available at the link given on the Lustre page; it appears you now
have to go through
http://www-verimag.imag.fr/~raymond/tools/lv4-distrib.html to get it.
* Grafcet ( http://www.emn.fr/x-info/lego/ ) In French; doesn't appear to
be any software to download
* Lego.NET ( http://www.dcl.hpi.uni-potsdam.de/research/lego.NET/ ) -
Supports a subset of the Microsoft .NET Framework 2.0
Colibri ( http://opensource.se/colibri/ ) has some API's that could
potentially be incorporated. "Version 0.1.1 has a reentrant scheduler,
timer functions, advanced semaphores (enhanced POSIX style), interrupt
handling code, some h8 uart code and a small collection of generic data
types."
Then, there is Dr. Hoenick's Bibo ( http://hoenicke.ath.cx/rcx/bibo.html ).
WIth the only real difference from BrickOS being in the task scheduler,
would future development be farther ahead by using Bibo as the starting
point rather then the current BrickOS?
Cheers,
Matthew
|
|
| |
Matthew Sheets wrote:
> This is a follow up to earlier postings discussing the development status of
> BrickOS. While the Sourceforge project has not had much activity of late,
> I've come across various modifications that, unfortunately, have never made
> it upstream. Even more unfortunate is that as time passes, it seems a few
> more RCX sites go offline.
Case in point: when I first started on this post some weeks(!) ago,
hoenicke.ath.cx wasn't responding.
> After finding BrickOS modifications or patches, there are at least two key
> issues: 1) the changes are not always against the latest version of BrickOS
> and 2) different sets of changes might modify the same section of code.
And it's not getting easier if we've collected different sets of patches
and then make further changes based on those... So yeah, we need to bring
everything that already exists together and then go from there.
> I have assembled a collection of patches that I have used to update my
> version of BrickOS. Some may be more familiar than others, but I have
> listed them all here for completeness. The numbers correspond to the
> numbers in the diff file name in the archive file.
Excellent. I've gone through your set of patches, making sure they compile
without warnings with gcc 4.3.2. The binutils 2.16.1 assembler (on x86_64)
prints lots of annoying warning messages (like "Warning: operand
#0xfffffffffffffffa out of range."), but they appear to be completely bogus.
> 1 - 5. Dr. Hoenicke's patches ( http://hoenicke.ath.cx/rcx/brickOS.html ),
> including gcc33, highmem, performance, Makefile, and tcpcomm
The second, "experimental" version of the performance patch (which adds
CONF_DSENSOR_FAST) also works well with the following patches.
> 6. Carl Troein's signedness patch (
> http://news.lugnet.com/robotics/rcx/legos/?n=3987&t=i&v=a )
I've made some minor (I think) update to this one. Also, it now comments out
the nasty "find /" in the configure script, as I got a bit annoyed when it
printed hundreds of error messages when scanning /proc, /etc and other silly
paths (and even more annoyed when various noisy disks were spun up).
> 7. Outstanding Sourceforge patch - enable Makelx to handle longer file names
> 8. Outstanding Sourceforge patch - update entrypoint specification to the S9
> record (Lego.NET)
> 9. Outstanding Sourceforge patch - address compatibility issues in
> non-Linux/Cygwin environments
> 10. Outstanding Sourceforge patch - serial port init fix for Mac OS X
A line break in the middle of a comment util/firmdl/rcx_comm.c breaks
this patch. My version of the patch is almost identical (but working), and
as a bonus the indentation is consistent with the surrounding code.
[Update: ah, I see the bug is fixed in patch 13.]
> 11. Mark Riley ( http://news.lugnet.com/org/ca/rtltoronto/?n=14996 ) - LDCC
> 12. Brown.edu ( http://www.cs.brown.edu/courses/cs148/old/2004fall/brickOS/
> and
> http://cs.brown.edu/courses/csci1480/old/2005/brickOS/quickstart.html ) -
> LNP printf capabilities
At least with gcc 4.3.2 there's a warning about vsnprintf being implicitly
declared. I've created include/stdio.h with declarations of [v]snprintf and
made a new version of the patch.
> 13. Mike LeSauvage (
> http://tarpit.rmc.ca/mcgaughey/eee243/labs/resources/legoshrink.html ) - LNP
> communication utility; made modifications to reuse existing code and added
> support for TCP communication (such as used by BrickEmu). Please note that
> I have not been able to test these changes for all platforms.
Neat. Good to see that you're reusing existing code. That reminds me that
there's some fairly trivial code duplication between dll and firmdl. It's
not very important, but I'll try to do something about it soon. As for
not tested on all platforms: the code doesn't compile on *ix because of
some calls to open() with the undefined mode flags O_BINARY and O_TEXT.
I've added a test for _WIN32.
> 14. Taiichi Yuasa ( http://xslisp.com/ ) - miscellaneous patches to BrickOS
> that are packaged along with xsLisp
I've added the doxygen update and my edgecount stuff as patches 15-16.
> I have uploaded this collection of patches to
> http://sourceforge.net/tracker2/index.php?func=detail&aid=2601672&group_id=58151&atid=486699
Revised version at
http://sourceforge.net/tracker/?func=detail&aid=2722649&group_id=58151&atid=486699
> This is by no means an exhaustive inclusion of everything that is out there,
> though what remains might require more effort to incorporate.
> [...]
> * ECU.edu (via archive.org -
> http://web.archive.org/web/20070624025341/http://www.cs.ecu.edu/~hochberg/fall2002/brickOS/index.html )
> - task prioritization, task scheduling, memory allocation, designate sensor
> as real-time
I like the idea of being able to reliably get every sensor value, and
to do so without a long and unpredictable delay. A while back I wanted
something similar, but I want for a simpler solution: callback functions
in the sensor handler (one per sensor, called every time the sensor is
sampled), but that's a very fragile and error-prone solution (for one
thing, the user would have to disable interrupts to interact with
whatever struct the callback function uses to store the results).
> * NCSU.edu ( http://moss.csc.ncsu.edu/~mueller/rt/rt05/readings/g1/ ) -
> task scheduling, prioritization, etc.
>
> While probably too large to include in a regular distribution, Olaf Christ's
> TCP/IP-enabled RCX is interesting (via archive.org -
> http://web.archive.org/web/20040324131317/http://users.informatik.haw-hamburg.de/~christ_o/)
It's way cool, but I'm less sure that it'd actually be useful. :-)
> Then, there is Dr. Hoenick's Bibo ( http://hoenicke.ath.cx/rcx/bibo.html ).
> WIth the only real difference from BrickOS being in the task scheduler,
> would future development be farther ahead by using Bibo as the starting
> point rather then the current BrickOS?
That's a very good question. From what I gather the bibo kernel is
quite a bit smaller and simpler than what we have today, but is that
reason enough to switch? Would we miss the features that we'd lose?
And what exactly _are_ the differences anyway, from a practical point
of view?
Another thing: We should have a good asm implementation of the Power
Functions protocol. I've ordered the Emerald Night collection so soon
I'll have some power functions parts and a reason to control them
with BrickOS. Hopefully that will lead to something.
//Carl
|
|
| |
Hello,
2009/3/31 Carl Troein <carl@thep.lu.se>:
> Matthew Sheets wrote:
>
> > This is a follow up to earlier postings discussing the development status of
> > BrickOS. While the Sourceforge project has not had much activity of late,
> > I've come across various modifications that, unfortunately, have never made
> > it upstream. Even more unfortunate is that as time passes, it seems a few
> > more RCX sites go offline.
>
> Case in point: when I first started on this post some weeks(!) ago,
> hoenicke.ath.cx wasn't responding.
For some unknown reason the address suddenly pointed to the old server
again which was shutdown in December last year. I noticed it myself
only by chance. It seems that the dyndns record has changed back to
the old address on March 12th, and I still have no idea how this has
happened.
> > I have assembled a collection of patches that I have used to update my
> > version of BrickOS. Some may be more familiar than others, but I have
> > listed them all here for completeness. The numbers correspond to the
> > numbers in the diff file name in the archive file.
>
> Excellent. I've gone through your set of patches, making sure they compile
> without warnings with gcc 4.3.2. The binutils 2.16.1 assembler (on x86_64)
> prints lots of annoying warning messages (like "Warning: operand
> #0xfffffffffffffffa out of range."), but they appear to be completely bogus.
Sounds like a 64bit bug: -6 is converted to a 64 bit number and then
it complains that it doesn't fit in an 8 or 16 bit operand :) But if
it is only a warning one can ignore it.
It is really hard to find a working binutils version. The newer ones
do not officially support the architecture any more and with many of
the older versions I was hit by some bug that produced buggy code
without even issueing a warning.
binutils-2.16.1 has problems with local labels ("bne 1f ;....; 1:"),
so make sure they're not used in inline assembler code. I removed
them from my second performance patch, but they are still in bibo at
several places. I'm not sure if they will lead to crashes there.
> > Then, there is Dr. Hoenick's Bibo ( http://hoenicke.ath.cx/rcx/bibo.html ).
> > WIth the only real difference from BrickOS being in the task scheduler,
> > would future development be farther ahead by using Bibo as the starting
> > point rather then the current BrickOS?
>
> That's a very good question. From what I gather the bibo kernel is
> quite a bit smaller and simpler than what we have today, but is that
> reason enough to switch? Would we miss the features that we'd lose?
> And what exactly _are_ the differences anyway, from a practical point
> of view?
The most important difference is that the handlers are more robust, as
they are running in the idle thread with its separate stack, they can
invoke malloc/free, and they can be pre-empted by interrupts. Another
point is the event-based scheduler: Processes can register themselves
to wait on events and are scheduled when the event is triggered, e.g.
by a handler routine. My plan was to make everything triggered by
events. However, to have backwards compatibility I still support the
old wait_event(&cond), which just runs the function cond every 5 ms
(in brickos cond is called, whenever the scheduler is run, which is
done every 100ms or when the system is idle or yield is called
explicitly). The right way would be to run the function cond when
sensors are polled or lnp message is received, depending on the type
of event we are waiting for. However, waiting for a sensor poll is
not (yet) supported. I would also like to see handlers for sensors in
bibo that get called after the sensor was polled.
Bibo is mostly backwards compatible to brickos, but you have to
recompile your sources against the right include files. At least that
is the theory, I do not have so many programs to test against. And I
didn't test if bibo compiles fine under cygwin. Two of the wakeup_t
functions have been replaced by real event-based waiting functions:
use dkey_wait(KEY_PRESSED,x) instead of wait_event(dkey_pressed,x) and
dsound_wait() instead of wait_event(dsound_finished, 0). A known
missing feature is the time-slice scheduling of processes with the
same priority. It would not be too difficult to add it. However,
adding a yield at the right position to the user code gives much
better performance in most cases.
Regards,
Jochen
|
|
| |
Hi,
> > 1 - 5. Dr. Hoenicke's patches (
> > http://hoenicke.ath.cx/rcx/brickOS.html ),
> > including gcc33, highmem, performance, Makefile, and tcpcomm
> The second, "experimental" version of the performance patch (which adds
> CONF_DSENSOR_FAST) also works well with the following patches.
I had worked to assimilate those patches prior to the second version of the
performance patch being posted, so I'm glad to hear that it also works well.
> > That reminds me that there's some fairly trivial code duplication between
> > dll and firmdl. It's not very important, but I'll try to do something
> > about it soon.
When I went back through to add TCP support to other things, that "similar
but different code" made things a little more difficult. There are more
things that could use the addition of TCP support, including xsLisp. I
think a nice enhancement to the TCP ir-server would be to give it the same
"--tty" interface options as dll and firmdl--i.e., directly with a serial
tower, USB tower, or another TCP ir-server. This, for example, would enable
a "real" brick and a BrickEmu instance to communicate with each other.
Anyway, the short of it is that, yes, I think a merged communication library
would be very useful.
By the way, I don't know if you monitor the NQC newsgroup, but I submitted a
patch to add TCP support to NQC....
- RE: BrickOS and Bibo
Personally, my feeling is that Bibo improves upon BrickOS and will provide a
better foundation moving forward, and I say that as one who has spent some
time incorporating a number of patches into the latest BrickOS. I'm not
sure yet how much effort will be involved in working back through them to
apply them to Bibo (I'm thinking in particular of patches 7 - 16), but I
think it would ultimately be worth the effort. Of course, a few of the
patches to BrickOS may not even be relevant in Bibo, but I would at least be
interested in having the LDCC, LNP printf, and LNP communication utility
functionality in Bibo.
Thanks,
Matthew
|
|
| |
Hi all
Lately I've noticed a lot of old-school Lego Mindstorm resources are
starting to disappear from teh webz. I've blagged about wanting to
preserve this stuff here: http://rubyredbricks.com/2009/4/12/back-to-
the-brick
If anyone knows of any resources you think should be maintained
please let me know. I'm going to do my utmost to make sure stuff
doesn't just "blow away". TIA
Cheers
Tom
|
|
| |
After surfing through many sites, and (sadly) even more 404s I have been
unable to find a definitive list of gcc, binutils, BrickOS etc. versions and
patches that I should and could be using.
Presumably everybody here has up to date versions of all the necessary programs
working with BrickOS.
What would be very useful would be to have a solid list of patches, version
numbers, problems as of now.
I spotted the patch list on the first post of "BrickOS Patches and Development"
and found it very handy.
The main problem in getting everything together is that every site I visit seems
to have been left unattended for many years, and are all recommending old and
outdated versions of compilers and patches.
A list of all the current versions and patches, along with links and perhaps
backups would be very useful.
|
|
| |
> A list of all the current versions and patches, along with links and
> perhaps backups would be very useful.
Nice to see a few of us share the interest. :-) The best I can offer for a
list of patches is what I listed in the post here [
http://news.lugnet.com/robotics/rcx/legos/?n=4032&t=i&v=a ], along with
Carl's follow-up post. That list was culled from links I've collected over
the years. As was noted, there are different patches out there for
different versions of BrickOS, and those two posts are an effort to merge
together appropriate patches against the latest SourceForge version of
BrickOS. From Carl's post, the link to download the latest version of this
collection of patches is
http://sourceforge.net/tracker/?func=detail&aid=2722649&group_id=58151&atid=486699 .
The patches are numbered and are intended to be applied in numerical order.
As was also noted, patches yet to be merged deal with "reliable LNP," task
prioritization, task scheduling, memory allocation, real-time sensors, etc.
To the tools versions--
* binutils: 2.16.1 - h8300 COFF support was dropped from subsequent
releases, so unless BrickOS is updated to h8300 ELF, we will have to stick
with this version
* gcc: 3.4.6 - this is the version that I use; I think Carl has had some
success with some releases in the version 4 series
Dr. Jochen Hoenicke's Bibo is also worth considering as a "next version" of
BrickOS. I have started going through that list of patches to carry over
any relevant changes to Bibo. If I can get them ported over to Bibo, I
expect I will base any of my future "legOS" development on Bibo.
For some 404 pages, http://www.archive.org/ might be your friend; no
guarantees, though.
Hope this helps,
Matthew
|
|
| |
Now for Bibo.... I have merged these patch rollup collections with Bibo
(where appropriate) and have posted the resulting diff files to
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
Most of this follows the pattern of the previous patch rollup collection for
BrickOS, with the following notes:
* 00 - Update configuration and make slight makefile modifications to work
with Cygwin
* 10 - As Carl noted, there are similarities between rcxtty.c and
rcx_comm.c. I updated this patch to add code to rcx_comm.c that previously
was only in rcxtty.c. There are still some differences (such as in return
types, use of exit(), and FileDescriptor handling), but this should reduce
the gap.
* 13 - I originally added LegOShrink as lnp_shrink (due to the OS naming
issue) but have added it here as lnpmsg.
* 16 - [@Carl]: I have not included the edgecount patch. Most of the
edgecount code appeared to be fairly straightfoward to incorporate into
Bibo; however, the assembly code in ds_interrupt in dsensor.c is a little
different, which could impact the edgecount patch. If you have the time to
take a look at it, that would be great, as it would be nice to have it
included.
Thanks,
Matthew
|
|
| |
> A very useful site is the cs university page here:
> http://www.cs.brown.edu/courses/cs148/old/2004fall/brickos.shtml
Agreed. There is also a set of pages from the following school year at
http://cs.brown.edu/courses/csci1480/old/2005/brickOS/quickstart.html .
Unfortunately (at least from a code merge perspective), their "version 1.2"
is based on BrickOS 0.2.5. The lnp_printf functionality has been adapted
and included in the recently-posted rollup patches, but incorporating other
modifications, such as "Reliable LNP," is more involved due to the
difference in BrickOS versions.
> A page of patches: http://carl.troein.com/
[@Carl]: Please correct me if I'm wrong, but I believe these patches are
already included in the rollup posted to SourceForge?
If the SourceForge project wiki for BrickOS could be enabled....
I've noticed another thing that is disappearing is the custom sensors and
multiplexor boards that were created for the RCX....
Thanks,
Matthew
|
|
| |
> Revised version [BrickOS patches] at
> http://sourceforge.net/tracker/?func=detail&aid=2722649&group_id=58151&atid=486699
> Now for Bibo.... I have merged these patch rollup collections with Bibo
> (where appropriate) and have posted the resulting diff files to
> http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
I have found two problems with the patches. I narrowed down the problems to
specific patches from the Bibo set. However the symptoms are identical under
BrickOS so I assume they would be linked to the same patches there.
Patch 10: Under Ubuntu with the usb tower, this causes firmdl3 to exit with
error "read: resource temporarily available". However the light does light up on
the tower for a small moment.
Patch 11: This seems to do something funny with the motors. Power does not seem
to be supplied to the motors. when running the motor, the arrows appear on the
lcd however there is no power being outputted.
|
|
| |
Patch 10: This works for me under Cygwin with a USB tower, but some of the
files such as rcx_comm.c have a bit of platform-specific code.
Patch 11: I've noticed that the linker command file is a little different in
Bibo than it was in BrickOS. If _motor_controller is 0x00 in bibo.ld and
dm_mask is set to 0x00 in dmotor.c dm_init(), there is "regular" motor
output; if _motor_controller is the BrickOS value of 0x80, DCC motor output
functions as expected.
|
|
| |
I have updated the Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
The primary changes are in Patch 11, which provides DCC support, and the key
change in that patch was the modification of some assembly code that
performed motor controller bitmasking.
Still known to be outstanding is an issue using firmdl3 on Linux when using
a USB IR tower.
Thanks,
Matthew
|
|
| |
After all the amazing post across lugnet the past few weeks, it feels
a bit lame putting this one up:
http://rubyredbricks.com/2009/4/27/simple-traffic-light
I guess once it's running up on an RCX it'll be a bit more interesting.
On the topic of RCX - and the whole old-school Mindstorms
preservation idea - I haven't forgotten about it. I've registered a
domain and will aim to get a wiki up and running over the weekend.
Cheers
Tom
|
|
| |
Hey folks
I've created a very basic site here: http://www.rcxzone.com/
When I say very basic I mean VERY basic.. it's just an empty instance
of mediawiki, but I'll look to adding content when I get a chance.
Any help appreciated.
Cheers
Tom
On 20/04/2009, at 6:29 PM, Joe Bradley wrote:
> > Revised version [BrickOS patches] at
> > http://sourceforge.net/tracker/?
> > func=detail&aid=2722649&group_id=58151&atid=486699
>
> > Now for Bibo.... I have merged these patch rollup collections
> > with Bibo
> > (where appropriate) and have posted the resulting diff files to
> > http://sourceforge.net/tracker/?
> > func=detail&aid=2773502&group_id=58151&atid=486699
>
> I have found two problems with the patches. I narrowed down the
> problems to
> specific patches from the Bibo set. However the symptoms are
> identical under
> BrickOS so I assume they would be linked to the same patches there.
>
> Patch 10: Under Ubuntu with the usb tower, this causes firmdl3 to
> exit with
> error "read: resource temporarily available". However the light
> does light up on
> the tower for a small moment.
>
> Patch 11: This seems to do something funny with the motors. Power
> does not seem
> to be supplied to the motors. when running the motor, the arrows
> appear on the
> lcd however there is no power being outputted.
|
|
| |
I have again updated the Bibo Patch Rollup Collection posted to SourceForge
at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
A big change this time is the consolidation of the code bases of util/firmdl
and util/dll-src. Working with all the similar-but-different code was
getting a bit cumbersome, so I finally decided to try to clean it up. This
should make enhancing or creating other host utilities much easier.
Point by point, updates since the last lugnet posting are as follows:
* Kudos to Joe for identifying the source of the IR tower problem in Linux,
as he tracked it down to the O_NDELAY flag. For now, this flag has been
removed.
* Made NCD tty types a runtime option instead of a compile-time #define.
* Merged patches for BrickOS with Bibo, including LDCC, lnp_printf, and
lnpmsg (formerly LegOShrink)
* Reworked firmdl3, dll, and lnpmsg under utils as well as lnp and
lnp-logic in the kernel to facilitate better code reuse, remove essentially
duplicate code, and establish a set of common functions for rcx_comm and
LNP. Merged the resulting code into util/host, deprecating util/firmld,
util/dll-src, and util/fontdesign.c.
* Util dll--made fast mode a runtime option instead of a compile-time
#define.
* Added ability to send LNP messages directly from the lnpmsg command line.
* Added ir-server from Jochen Hoenike's BrickEmu package and used the code
reworked from above to add support for tty downlinking (e.g. serial, USB,
TCP) or TCP uplinking.
Notes and issues:
* If lnp_integrity_byte() is processing firmware download communication on
the host (e.g. if lnpmsg is listening on an ir-server repeating the firmware
download), a segmentation fault will eventually result.
* Command-line argument format and command-line argument processing - the
various programs use different methods of processing command-line arguments,
resulting in similar argument being handled differently by different
programs. Ideally, this would be addressed in another round of code
cleanup.
* I am not currently setup to test either serial IR towers or NCD serial
connections, so any reports for these devices would be welcome.
Using the updated ir-server, I have actually been able to send firmdl3 to
ir-server (instance A) to ir-server (instance B) to USB IR tower to physical
RCX. I then used
dll to send helloworld_lnp.lx up the same route. Finally with one instance
of lnpmsg listening on ir-server (A) and another instance of lnpmsg
listening on ir-server (B), I ran the helloworld_lnp program on the physical
RCX. I did have to increase RCX_TIMEOUT to 500, but everything otherwise
works as expected.
I also tried the same sequence using BrickEmu (with a ROM image from my RCX)
instead of the USB IR tower. The only anomaly was that "unlock firmware"
fails, but everything else seems to work fine.
Thanks,
Matthew
|
|
| |
I've again updated the Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
This update adds two new patches--18 and 19.
18: For the host utilities, this patch attempts to standardize the command
line arguments and their processing. There is also some additional cleanup
in lnpmsg, plus the addressing of a gcc 4 warning that was brought to my
attention. The original lnpmsg (actually LegOShrink) supported logging to
file, sending from file, and filtering LNP addressing messages based on
destination host and/or port. To ease the process of standardizing the
common host utility code, some of those features were trimmed out; however,
some of the supporting functions remain in place, including SendFile() and
LogString in lnpmsg_comm.c
19: The assembly file dccasm.S with functions necessary for DCC operation
was moved from libs/dcc to alongside kernel/dcc.c. As has been the case
dcc.c, the code in dccasm.S now builds based on whether or not CONF_DCC has
been defined. The folder lib/dcc may be deleted.
Thank you,
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
This update adds a handful of patch files--20 through 24.
20: Nice group of Linux error/warning fixes from Carl and posted to
SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2793018&group_id=58151&atid=486699
Thanks, Carl!
21: This is a group of bibo paches by Dr. Hoenicke, of which the major
changes relate to makefile, assembly branching, and util man page updates
(plus a documentation typo on my part--sorry! :-). Thank you!
The above patches were incorporated into the rollup to minimize merge
differences with subsequent patches.
22: This patch changes kexeci's argv argument from type char** to type
void** and is based on some (now old) feedback from the Lego.NET folks (
http://www.dcl.hpi.uni-potsdam.de/research/lego.NET/ ). Some delegate-type
functions aren't (yet) implemented for Lego.NET, but the capability should
be there for wait_event-type delegates. For the case of kexeci, the
Lego.NET delegate's argv would not be of type char**. From what I can tell
by reviewing the code, the type of argv does not matter to the BrickOS/Bibo
kernel, so it should be safe for a Lego.NET delegate to pass a non-char-type
argv. If anyone has any deeper insight, I would welcome the feedback.
23: Back to work on LNP. Implemented some code and ideas from
http://lnphost.sf.net/ , which was designed as a replacement for lnpd.
While there are a few drawbacks due to implementation differences (no
tty_type_t support, no sharing/reuse of lnp and lnp-logical with the kernel,
a small documented problem that "in some rare cases lego packets might not
be detected correctly"), it does have some nice features, including handlers
for standard Lego messages and a nicer hexdump routine (shows both hex and
ASCII-printable characters. That hexdump routine has been incorporated,
replacing the existing one.
LNP was updated to include a handler for standard Lego messages (this is in
addition to existing functionality), and the ability to send "Lego remote
codes" was added. The lnpmsg program also received several updates.
Support was added for both displaying and sending both standard firmware
messages and Lego remote codes. The LNP message output format was changed
to use the updated hexdump routine (instead of just a text string), which
should make it easier to monitor non-text LNP messages.
24: Improved the keepalive capabilities to remove some of the keepalive
burden from programs that might use rcx_comm. Unfortunately, I'm not setup
right now to test either a serial IR tower or an NCD port, but I did try to
setup a dummy NCD-like tty for testing. Keepalive works of a signal and
alarm, with code in place to prevent a keepalive byte from being sent while
either writing (rcx_nbwrite) or actively reading (rcx_nbread). I would be
interested in feedback from anyone using a device that needs a keepalive.
Thank you again to those who have responded or provided feedback. If you
ever catch a few more free moments, Carl, I would be interested in a Bibo
version of your edgecount patch. :-)
Thanks,
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
New for this update are patches 25 through 35.
25: Cleanup in the util subdirectory
- Updated lnpmsg to more cleanly handle and process command-line arguments
- Removed in mkimg that is a duplicate of code in srecload
- Updated some Makefile cross targets
26: Updated makefiles
- Changed the existing TOOLPREFIX to CROSSTOOLPREFIX, then used TOOLPREFIX
for the host tools (needed when creating packages that must be targeted to,
for example, a 486 processor)
- Moved some target definitions to Makefile.user, since it is shared by
both Makefile.dist and Bibo's make process. This also reduced Makefile
dependencies across SUBDIRS (util, lib, include, kernel, demo, doc). Some
individual subdirs can now be removed from the SUBDIRS assignment statement
in the main Makefile and the rest of the project will still build
successfully.
- Updates related to util/host, as it seems to be stabilizing. Some items
in util/Makefile.sub have been moved to util/host/Makefile.sub, since the
code now resides there
- Removed the need to set PROGRAMS when building *.lx files. Building %.c
/ %.cpp / %.s / %.S will now look for a make variable by the name of "%_SRC"
to determine if additional object files need to be built.
- Changed the cross-compiler optimization from -O2 to -Os (space)
27: Configuration #defines
- Added a CONF_VIEW_BUTTON define based on Taiichi Yuasa's modification,
allowing exclusion of the information display normally provided when
pressing the RCX View button. The primary reason for adding this
configuration define is enable building a smaller kernel.
- Added a configuration define needed in one of the demo programs in case
CONF_RCX_MESSAGE is not defined.
28: Merged XS Lisp - http://www.xslisp.com/
- A few of the above patches were in preparation for this update
- Was originally based on BrickOS 0.2.6.10
- Is not an addition to the kernel but instead can be used to build *.lx
files that can be downloaded to the RCX. The generated *.lx programs are
large, so a smaller kernel is needed.
- Supports both an IR-link mode and an autonomous mode (IR features were
updated to use rcx_comm)
- Added better support for building from Makefiles
- For the full detail on XS, please see the file xs/README and or visit
the website. Note that information on the website refers to files such as
xs, xs0, xs1, xs2, etc. When merged with Bibo, the numeric suffixes were
replace with more descriptive text, and xs/README reflects this
modification.
- A list of suggested configurations to disable is also included in
xs/README. Any other recommendations or suggestions here would be
appreciated.
29: RCX Communication
- Created a tty struct to replace the existing FILEDESCR and tty_type_t
that get passed around in host-based programs.
- Replaced the fast/slow options in the host utilities with support for
specifying a baud rate at run-time.
30: RCX Communication Timeout for Host Utilities
- Timeout value in milliseconds can now be specified at run-time instead
of being a build-time #define. This is useful when switching between a
low-latency, local IR tower and a TCP connection over the Internet (e.g.
using the uplink and downlink features added to ir-server) with a higher
latency.
31: TTY Device Names
- Added support for interpreting NQC-style port name strings of the format
"<device type>:<device name>". This format was already being used by the
TCP and NQC Bibo additions and was extended to include "serial:" and "usb:".
Backwards compatibility with existing conventions has been maintained.
32: Read Default TTY from a Config File
- Added support for reading the default tty device name from a
configuration file ("~/.rcx/tty.conf"). This essentially fulfills the same
purpose as the RCXTTY environment variable, but without some of the
limitations of environment variables. I am hoping to be able to add this
support to NQC, too.
33: IR Communication and Hacked-on (e.g. Bluetooth) Communication
- Attempted to provide an easier starting point for those using
communication hacks such as the bluetooth hack
~ Ability to set the baud rate
~ Ability to turn of echo handling
- Created configuration #defines to allow setting defaults for
~ Baud rate
~ IR Carrier enabled/disabled
~ Transmit echo/no echo
- Replaced slow/fast (2400/4800 bps) options with the ability to select a
baud rate
- Firmware downloads at 2400 bps can now also be completed without sending
the complement bytes (previously, this option was only available for "fast"
4800 baud)
- Enabled setting the communication timeout value at run-time. This is
useful, for example, when switching between a low-latency serial connection
and a higher-latency TCP-based connection.
- Improved the keepalive functionality for serial IR towers
- Moved help text for standard rcx_comm options to the rcx_comm header
file
** IMPORTANT CHANGE NOTE: Parity for all baud rates is now "odd"
34: Config File Cleanup
- Removed #define entries for configurations that no longer exist
- Removed RCX-only #define entries from the host configuration file
35: Remove CONF_ASCII Dependencies
- Currently, if CONF_PROGRAM is defined CONF_ASCII must also be defined.
This patch reduces the CONF_PROGRAM dependency level from CONF_ASCII to
CONF_CONIO, reducing the size of the kernel needed for program support.
- Created #defines for each of the entries in CONF_ASCII's
ASCII-to-CONIO-mask lookup table
- With those #defines, CONF_ASCII is no longer needed for displaying
static strings. (The small hex lookup table included by CONF_CONIO is still
present there.) Functions that are variants of cputc_native*, cpuc_hex*, or
cputw may be used to write to the display. For example, cputs("Lego") may
now be written as cputc_native_user(CHAR_L, CHAR_e, CHAR_g, CHAR_o). While
it's a little longer statement to type, it does help toward reducing kernel
size. ;-)
- All demo programs were able updated to eliminate the need for
CONF_ASCII.
- For dynamically generated strings, CONF_ASCII will still be required.
XS utilizes CONF_ASCII features for the XS functions that mirror the cputs
andcputc C functions (puts and putc), but due to the large size of XS, this
patch just limits the cputs/cputc functionality if CONF_ASCII is disabled
(puts will output "ASCII" and putc will display a '-' in the specified
position).
Thanks,
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
New for this update are patches 36 through 39.
36: RCX tty config file - Extends patch 32 so that /etc/rcx/device.conf will
also be checked for a default tty. The precedence order is first, the
RCXTTY environment variable; second, ~/.rcx/device.conf; third,
/etc/rcx/device.conf.
37: Automatic Shutoff Timer - Reset the automatic shutoff timer if the Lego
remote is used in direct mode (when no programs are running).
38: Firmware Fastloader - Added code to initialize the IR carrier frequency.
39: Linux Patches - As some of you may already know, I have been using
Cygwin. I am currently working to get a Linux-based BrickOS/Bibo
environment up and running, and though that Linux environment is not yet
fully functional, this patch is needed to help some of the earlier patches
run under Linux.
A word of thanks goes to Dr. Jochen Hoenicke for his assistance in answering
questions; any errors are mine. :-)
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
There is just one minor Bibo-specific update this time, patch 40.
40: H8/300 TOOLPREFIX - This patch comes from Debian, where
"h8300-hitachi-coff-" is added to the list of potential H8/300 toolprefixes.
The larger update this time is a set of patches for NQC running on Linux.
Included in these patches is a patch that corresponds with Bibo patches 32
and 36, which enables Bibo to read a default tty device name from a
configuration file. This same capability has now been extended to NQC.
More details on the NQC patches are available here -
http://news.lugnet.com/robotics/rcx/nqc/?n=1900
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
New for this update are patches 41 and 42.
41: Makefiles and the Esterel Synchronous Programming Language - The
makefiles were updated to support building Esterel source files for the RCX.
More information and links to the Esterel software required for building is
available at http://www.emn.fr/x-info/lego/. Five sample/demo programs
listed on this site have been included in an "esterel" folder in the bibo
distribution (make will only attempt to build these programs if the Esterel
compiler is found).
Also, when doing a "make && make install" on the bibo source, the installed
makefile may be used to build program source files by running "make -f
/path/to/installed/makefile SOURCES=demo.strl" (or demo.c, demo.cpp, demo.s,
demo.S); alternatively, you may run "make -f /path/to/installed/makefile
PROGRAMS=demo.lx" and make will attempt to build the corresponding code
file.
NOTE: While Esterel supports the CPUTS command, this requires that ASCII
support be enabled in Bibo. ASCII support is now disabled by default due to
(1) the size required for its support and (2) the fact that CONIO was
updated to make it easier to output characters to the display without the
size bloat of ASCII support. Some Esterel-generated programs can be quite
large, so space can be at a particular premium when using Esterel. Ideally,
Esterel could be updated to support some of the CONIO functions.
42: RCX Comm - Updated debug output. Modified rcx_init to allow specifying
whether or not to start the IR tower Keepalive handler (this setting does
not have any effect if the device does not require Keepalive functionality).
Thanks,
Matthew
|
|
| |
In lugnet.robotics.rcx.legos, Carl Troein wrote:
> Another thing: We should have a good asm implementation of the Power
> Functions protocol. I've ordered the Emerald Night collection so soon
> I'll have some power functions parts and a reason to control them
> with BrickOS. Hopefully that will lead to something.
I've recently mucked about with the Power Functions protocol, and got a proof of
concept running on the RCX using brickOS. The main problem is that it's quite
timing sensitive, so this code only worked reliably with interrupts disabled
(looking at the oscilloscope).
The sample code is at http://www.algonet.se/~yann/powerfunctions.c
The timing requirements are probably more lax in reality (PF protocol spec cites
ranges like 316-526us for a 0 bit), but my code was thrown off by whole periods
(of the 38kHz modulation signal) as other interrupts occurred.
|
|
| |
Hello Yann,
2010/6/15 Yann Vernier <yann@algonet.se>:
>
> I've recently mucked about with the Power Functions protocol, and got a proof of
> concept running on the RCX using brickOS. The main problem is that it's quite
> timing sensitive, so this code only worked reliably with interrupts disabled
> (looking at the oscilloscope).
>
> The sample code is at http://www.algonet.se/~yann/powerfunctions.c
>
> The timing requirements are probably more lax in reality (PF protocol spec cites
> ranges like 316-526us for a 0 bit), but my code was thrown off by whole periods
> (of the 38kHz modulation signal) as other interrupts occurred.
I really would prefer to see the interrupt code running to disabling
all interrupts for several milliseconds, loosing several ticks and
disabling all sensors during this time.
Did you apply one of my performance patches to brickOS? IIRC, the
current release of brickOS still has a busy waiting loop in the sensor
interrupt, which explains why you get corrupt timings. The patches
are at
http://hoenicke.ath.cx/rcx/brickOS.html
You can also use my simulator brickemu to get some statistics about
maximal run-time of interrupts. It shouldn't be too difficult to add
debugging output to the 8-bit timer to get cycle exact information of
when the registers are set. You can also debug your code with
brickemu by adding printf in your code.
It looks hard to get the interrupts fast enough, though. According to
my statistics, the timer interrupt in bibo can take up to 1031 cycles,
i.e 65 us, which is still more than the period of your timer. On
average it runs in only half the time, so you should not loose so many
cycles that your signal is distorted beyond recognition.
Regards,
Jochen
|
|
| |
Re: Power Functions:
There is an older thread on the topic of Power Functions at [
http://news.lugnet.com/general/?n=54443&t=i&v=a ] which includes links to
some BrickOS-based code for Power Functions. Be sure to follow the thread
to the end, as some of the links mentioned earlier on no longer work. The
current working link is [ http://www.bong69.com/pages/software.php ] (scroll
to the bottom of that page). Note that this code also disables/enables
interrupts.
Re: Interrupts
The LDCC patch, which Mark Riley wrote to output DCC waveforms over the
motor outputs, also disables interrupts, as DCC requires very precise
timings. Since creating the rollup release noted earlier [
http://news.lugnet.com/robotics/rcx/legos/?n=4057 ], which is itself based
on Jochen's Bibo release, I haven't tried running LDCC without disabling the
interrupts, but based on Jochen's description, I'll have to give that a try.
As a side note, LDCC has also been integrated into pbForth [
http://news.lugnet.com/robotics/rcx/?n=2297 ].
LDCC Links
* http://news.lugnet.com/org/ca/rtltoronto/?n=14996
* http://home.surewest.net/markril/lego/dcc/
Cheers,
Matthew
|
|
| |
The Bibo Patch Rollup Collection posted to SourceForge at
http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
has been updated.
New for this update are patches 43 and 44
43: Per the report from SourceForge user bollinge, Cygwin no longer has a
problem building printf.c. A Cygwin-specific conditional in
lib/c/Makefile.sub was removed.
44: Kernel size optimization - The values for BASE1 and BASE2 have been
declared as constants; however, calculating these values dynamically enables
greater ease of flexibility in handling changes to the kernel image size.
This is change is based on SourceForge tracker 1120793, but this
implementation is different from what was proposed in that tracker and does
not add a dependency on Perl like that tracker implementation would. This
particular update closes the last currently outstanding patch (apart from
the "(Unofficial) Bibo Release" patch) listed in SourceForge, though I still
have note of a few various and sundry updates posted elsewhere.
Thanks,
Matthew
|
|
| |
Matthew Sheets wrote:
> The Bibo Patch Rollup Collection posted to SourceForge at
> http://sourceforge.net/tracker/?func=detail&aid=2773502&group_id=58151&atid=486699
> has been updated.
Excellent, Matthew. Many thanks for keeping the project alive and moving
forward.
I've uploaded a little patch with the edge counting (which I needed once
upon a time when I tried to build a plotter without any rotation sensors)
re-implemented in a more sensible way. Here:
https://sourceforge.net/tracker/?func=detail&aid=3527703&group_id=58151&atid=486699
I'm currently working on figuring out all the little quirks of the power
functions protocol (such as what happens with the toggle bit state for
commands where the toggle bit is ignored). Once I have it fully working,
I'll release a patch with power functions control based on Bob Kojima's
code but updated to match the latest PF specification.
Cheers,
//Carl
|
|
|
