headers
Anders Wallin | 19 Jun 2013 08:52
Picon

Xenomai on dual-core ARM?

Hi all,

Olimex has just released a Debian image for their latest single-board
computer based on the Allwinner A20 cpu:
http://olimex.wordpress.com/2013/06/19/a20-olinuxino-preliminary-debian-linux-image/
http://en.wikipedia.org/wiki/Allwinner_A20

has anyone tried Xenomai on this, or another, dual-core ARM chip?
The latency numbers are supposed to be pretty good (mostly <10us with rare
occurances of 10-20us) for the single-core A13.

I have an A13 board on order and time permitting will be trying to get
Xenomai running on that, based on the posts on the olimex forum. The newer
A10S and A20 boards are nicer in that they have HDMI and native Ethernet.

Anders
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 1 comment |
headers
Alec Ari | 18 Jun 2013 21:45
Favicon
Gravatar

Question about a header file

Hello all! Since the 3.8 kernel releases, lxrt requires the following header, commit here:

https://github.com/ShabbyX/RTAI/commit/5da690cd3cf05fbe2c018184689b7881b682d835

But while RTAI compiles fine, LinuxCNC doesn't after this commit:

Compiling rtapi/rtapi_common.c
In file included from /usr/realtime/include/rtai_shm.h:227:0,
                 from ./rtapi/rtai-kernel.h:14,
                 from rtapi/rtapi_common.h:77,
                 from rtapi/rtapi_common.c:4:
/usr/realtime/include/rtai_lxrt.h:73:30: fatal error: uapi/linux/sched.h: No such file or directory
compilation terminated.
make: *** [objects/rtapi/rtapi_common.o] Error 1
make: Leaving directory `/home/ntu/devel/emc2-dev/src'

Not sure how to fix it.. Any help is greatly appreciated!

Alec

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 0 comments |
headers
Stuart Stevenson | 18 Jun 2013 18:58
Picon

Re: shipping to mpm

Sure - not a problem.
On Jun 18, 2013 11:36 AM, <emc-developers-request@...>
wrote:

> Send Emc-developers mailing list submissions to
>         emc-developers@...
>
> To subscribe or unsubscribe via the World Wide Web, visit
>         https://lists.sourceforge.net/lists/listinfo/emc-developers
> or, via email, send a message with subject or body 'help' to
>         emc-developers-request@...
>
> You can reach the person managing the list at
>         emc-developers-owner@...
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Emc-developers digest..."
>
>
> Today's Topics:
>
>    1. Re: linuxcnc.org website (Viesturs L?cis)
>    2. Re: 7I39HV question (andy pugh)
>    3. Re: 7I39HV question (John Kasunich)
>    4. Re: Wichita (TJoseph Powderly)
>    5. Re: 7I39HV question (Jon Elson)
>    6. Re: 7I39HV question (Jon Elson)
>    7. Re: 7I39HV question (Peter C. Wallace)
>    8. Re: 7I39HV question (John Kasunich)
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Viesturs Lācis | 18 Jun 2013 11:40
Picon

linuxcnc.org website

Hello!

I just discovered that main website of the project seems strange:
if I type address www.linuxcnc.org in address bar, I see this:
http://picpaste.com/ok-sHrM9Epp.png
which seems ok

if I type linuxcnc.org in address bar, without www in front, I see this:
http://picpaste.com/not-ok-CqbIAJzk.png
which does not seem ok as there is a text in russian, which is fine for me
as I understand it, but that is not the case for vast majority of other
users.

Whom should I report this to?

--

-- 
Viesturs

If you can't fix it, you don't own it.
http://www.ifixit.com/Manifesto
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 3 comments |
headers
Alec Ari | 18 Jun 2013 09:07
Favicon
Gravatar

Re: perf record pagefault stats for PREEMPT_RT

http://dpaste.com/hold/1249843/

----- Original Message -----
From: Lars Segerlund <lars.segerlund@...>
To: Alec Ari <neotheuser@...>
Cc: 
Sent: Tuesday, June 18, 2013 1:05 AM
Subject: Re: perf record pagefault stats for PREEMPT_RT

It's actually excellent !

The only thing I'd like is the -g flag in the build .... we still
have some unresolved symbols ... could you send me a pmap -x of the
process(es) ...

But looking at this were using _setjmp ! ! ! so that's probably a
nono ... it saves current state of the C process .... I don't know the
implementatioin details in libc, but it's a good candidate for causing
faults.

There is an article thats called 'Getting rid of SIGSEGV for fun, but
not for profit' I think it's on phrack, but I can't get it while at
work, ( firewall & webfilter ), I will look if we can modify the
signal handling of the process ( rtapi ) to do out own page fault
tracking .... then we could print the exact location .
I have to think a bit and test it first but I think something like
that might work.

Thanks !
(Continue reading)

Permalink | Reply | 0 comments |
headers
Alec Ari | 18 Jun 2013 09:02
Favicon
Gravatar

Re: perf record pagefault stats for PREEMPT_RT

Compiled with -g3, hope this helps!

Alec

----- Original Message -----
From: Lars Segerlund <lars.segerlund@...>
To: Alec Ari <neotheuser@...>
Cc: 
Sent: Tuesday, June 18, 2013 1:36 AM
Subject: Re: perf record pagefault stats for PREEMPT_RT

Hi again,

I think we have to look at the code a bit, the thing is it should set
everything up, frefault and mlock .... before getting going :-D ....
we probably have a problem with that.

I don't know so much about hal, but there is a rtapi_init or
something similar ... I'll see if I get some time later.

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 1 comment |
headers
Alec Ari | 18 Jun 2013 07:26
Favicon
Gravatar

Re: perf record pagefault stats for PREEMPT_RT

http://dpaste.com/hold/1249620/

Is this better?

Alec

----- Original Message -----
From: Lars Segerlund <lars.segerlund@...>
To: Alec Ari <neotheuser@...>
Cc: 
Sent: Tuesday, June 18, 2013 12:07 AM
Subject: Re: perf record pagefault stats for PREEMPT_RT

HI Alec,

I haven't gotten the time to set things up yet, but this is
interesting. I have been working with the perf API on the kernel side
lately, but not so much with the perf program since we have or own :-D
.
It seems that it's rtapi is generating faults, ( there are minor and
major, major are the worst ).

I think it's thime to do a little code revision, since i don't think
there will be 'one' thing that fixes all problems, so I will have to
look for all memory to be properly set up and mlocked, also the stacks
have to be prefaulted ( for threads too ).

There are a number of things to look at, for example rtapi seems to
be faulting in ld ??? We are creating a new sys V ipc shared memory
and that will fault.
(Continue reading)

Permalink | Reply | 0 comments |
headers
John Kasunich | 18 Jun 2013 07:03

7I39HV question

I have some Pittman Elcom ST brushless servos that I'd like
to play with while I'm in Wichita.  They are rated at 19.1V, 
4-6 amps (depending on cooling).

I'm planning to bring an Atom computer, 5i20, and 7I39HV
brushless drive board to Wichita along with a motor.

My question is about power supplies.  I thought I had a
reasonably sized 24V supply here, but I was wrong. 
I have tiny ones (less than n amp), and huge ones
(10A, to heavy for checked baggage and schlepping
thru an airport.

I do have a unregulated supply that puts out about 40V
(no load) and should be good for 5-6 amps, at least
intermittently.  It is fairly compact and travel friendly.
Is 40V too much for a 19V servo?  (The 7i39 is the HV
version, rated for 48V.)

--

-- 
  John Kasunich
  jmkasunich@...

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
(Continue reading)

Permalink | Reply | 14 comments |
headers
Alec Ari | 18 Jun 2013 06:44
Favicon
Gravatar

perf record pagefault stats for PREEMPT_RT

Hello everyone! Lars, how do you do? :)

I've been tackling some pagefault issues which appear to lay within RTAPI. Me along with a few other
LinuxCNC developers are trying to fix PREEMPT_RT for LinuxCNC once and for all.

I personally never used perf before, but here is what I mocked up following Jepler's advice and perf commands:

http://dpaste.com/hold/1249519/

How to make sense out of that, I have no idea. Comparing the differences to the overhead values to certain
symbols, shared objects, or commands? I don't really know what I'm looking for, and could make about 8
different guesses on how to properly read recorded perf data, hopefully someone here can understand what
all this means, and ultimately, fix the underlying issue. Thanks!

Alec Ari

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 0 comments |
headers
Alec Ari | 18 Jun 2013 03:23
Favicon
Gravatar

3.8 kernels broken with LinuxCNC

Hello! In the 3.8 series, version.h moved from include/linux to include/generated/uapi/linux

I think LinuxCNC should detect what kernel version is being used, then adjust itself accordingly.

Exact problem: ./configure --with-kernel-headers=/lib/modules/`uname -r`/build/include

Result:

checking for unistd.h... yes
checking /lib/modules/3.8.13-RTAI/build/include/linux/version.h usability... no
checking /lib/modules/3.8.13-RTAI/build/include/linux/version.h presence... no
checking for /lib/modules/3.8.13-RTAI/build/include/linux/version.h... no
configure: error: version.h not found - Is the kernel headers package installed ?

Thanks!

Alec

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev
Permalink | Reply | 0 comments |
headers
Stuart Stevenson | 16 Jun 2013 22:27
Picon

Wichita

Ok guys,
I think some are arriving this evening (Sunday June 16, 2013).
All others - if you fly in and need a ride.

Call me - 316 258 0953
email - stuart@...

I will pick you up and take you where you need to go.
I am about 15 minutes from the airport - anywhere is Wichita is no more
than 15 minutes from the airport :)

MPM
2100 S West St
Wichita, Kansas
316 945 1227

SFI
1517 N Santa Fe
Wichita, Kansas
316 264 0603

see you soon
Stuart

--

-- 
dos centavos
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.
(Continue reading)

Permalink | Reply | 8 comments |

Gmane