Re: [Emc-developers] uspace

[Valerio Bellizzomi]

On Sat, 2020-06-13 at 23:40 -0400, Gene Heskett wrote:
> On Saturday 13 June 2020 20:53:36 andy pugh wrote:
> 
> > Is there an expectation that a uspace package would work on whatever
> > realtime system it found itself on, or would we need separate builds
> > for preempt-rt, xenomai and rtai-lxrt?
> 
> What diff would it make if you are spinning an iso?  Spin it for whatever 
> flavor runs best.
> 
> Preempt-rt, particularly with helper cards, has had little or no problems 
> getting the job done to my satisfaction on the two copies of the intel 
> D525-MW board that I own, one even doing software stepping for several 
> years, until it unscrewed a switch protected ball nut and scattered 
> balls all over.  And my bag of balls I had restuffed those nuts with a 
> decade back cannot now be found.  So that D525MW and all the motor 
> drivers that go with it is running the 6040, rather nicely but it grew 
> some mesa help in getting moved from the old hf mill.  With the mesa 
> help it can do rapids at 200 ipm.
> 
> I built a zenomai system to replace a BDI but before I could formulate an 
> opinion, a wheezy based iso appeared and thats still running on 3 of my 
> 4 machines. So I don't know a lot about zenomai.
> 
> My 4th machine is an armhf running buster with a preempt-rt kernel that 
> gets 16 u-secs latency but that dissolves to 200 u-secs while starting 
> firefox. But I don't run FF on that and cut metal. Nothing else seems to 
> bother it. Running with a 4.19.71-rt24-v7l+ #1 SMP PREEMPT RT I built, 
> getting long in the tooth now, but the rest of the install is uptodate 
> buster and the currently running LinuxCNC is the buildbots master.  All 
> have been updated in the last 2 days.


FireFox is historically slow to fire up, I could not even start it up on
armhf raspberrypi with Gnome. I tried Xfce but there is a login loop,
ended up installing LXDE which is not too bad.

> > I did try the experiment of running the current buildbot 2.8-preemtrt
> > package on an RTAI system and it chose to use POSIX non-realtime.
> >
> > I am trying to write some explanatory notes for the 2.8 release, and
> > so would like to know if what I observed is what would be expected.
> 
> So, do what you think is the best for the "normal" user. Keep in mind 
> that everything today is 64 bit, a place few of us has jumped into on 
> our work machines. I expect some teething problems we'll have to work 
> out.
> 
> But I have confidence that won't be as big a problem as our imagination 
> is telling us. :)
> 
> Cheers, Gene Heskett




___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] uspace

[Gene Heskett]

On Saturday 13 June 2020 20:53:36 andy pugh wrote:

> Is there an expectation that a uspace package would work on whatever
> realtime system it found itself on, or would we need separate builds
> for preempt-rt, xenomai and rtai-lxrt?

What diff would it make if you are spinning an iso?  Spin it for whatever 
flavor runs best.

Preempt-rt, particularly with helper cards, has had little or no problems 
getting the job done to my satisfaction on the two copies of the intel 
D525-MW board that I own, one even doing software stepping for several 
years, until it unscrewed a switch protected ball nut and scattered 
balls all over.  And my bag of balls I had restuffed those nuts with a 
decade back cannot now be found.  So that D525MW and all the motor 
drivers that go with it is running the 6040, rather nicely but it grew 
some mesa help in getting moved from the old hf mill.  With the mesa 
help it can do rapids at 200 ipm.

I built a zenomai system to replace a BDI but before I could formulate an 
opinion, a wheezy based iso appeared and thats still running on 3 of my 
4 machines. So I don't know a lot about zenomai.

My 4th machine is an armhf running buster with a preempt-rt kernel that 
gets 16 u-secs latency but that dissolves to 200 u-secs while starting 
firefox. But I don't run FF on that and cut metal. Nothing else seems to 
bother it. Running with a 4.19.71-rt24-v7l+ #1 SMP PREEMPT RT I built, 
getting long in the tooth now, but the rest of the install is uptodate 
buster and the currently running LinuxCNC is the buildbots master.  All 
have been updated in the last 2 days.

> I did try the experiment of running the current buildbot 2.8-preemtrt
> package on an RTAI system and it chose to use POSIX non-realtime.
>
> I am trying to write some explanatory notes for the 2.8 release, and
> so would like to know if what I observed is what would be expected.

So, do what you think is the best for the "normal" user. Keep in mind 
that everything today is 64 bit, a place few of us has jumped into on 
our work machines. I expect some teething problems we'll have to work 
out.

But I have confidence that won't be as big a problem as our imagination 
is telling us. :)

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

[Emc-developers] uspace

[andy pugh]

Is there an expectation that a uspace package would work on whatever
realtime system it found itself on, or would we need separate builds
for preempt-rt, xenomai and rtai-lxrt?

I did try the experiment of running the current buildbot 2.8-preemtrt
package on an RTAI system and it chose to use POSIX non-realtime.

I am trying to write some explanatory notes for the 2.8 release, and
so would like to know if what I observed is what would be expected.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] "uspace plus": Ready for testing and code-review

[Sebastian Kuzminsky]

On 07/17/2016 11:36 AM, Jeff Epler wrote:
> "Uspace plus", which I originally called "uspace+rtai", is a
> modification of uspace realtime to support a total of 4 modes
> with a single build (depending on the availabiliy of headers for the
> RTAI and Xenomai modes):
> * LXRT realtime with RTAI kernel
> * "Posix Skin" realtime with Xenomai kernel
> * "Posix" realtime with PREEMPT-RT kernel
> * Non-realtime with vanilla kernel

This is exciting, Jeff.  I look forward to reviewing it and updating the 
buildbot to make use of it.


>  22 files changed, 890 insertions(+), 242 deletions(-)
> each new RTOS is under 200 lines of code.

Fantastic.


-- 
Sebastian Kuzminsky

--
What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
patterns at an interface-level. Reveals which users, apps, and protocols are 
consuming the most bandwidth. Provides multi-vendor support for NetFlow, 
J-Flow, sFlow and other flows. Make informed decisions using capacity planning
reports.http://sdm.link/zohodev2dev
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

[Emc-developers] "uspace plus": Ready for testing and code-review

[Jeff Epler]

"Uspace plus", which I originally called "uspace+rtai", is a
modification of uspace realtime to support a total of 4 modes
with a single build (depending on the availabiliy of headers for the
RTAI and Xenomai modes):
* LXRT realtime with RTAI kernel
* "Posix Skin" realtime with Xenomai kernel
* "Posix" realtime with PREEMPT-RT kernel
* Non-realtime with vanilla kernel

This all works great on my test machine, a i7-4790K with Debian Jessie,
and the following kernels:
3.18.0-1-rtai-amd64 (highlab.com)
3.18.20-xenomai-2.6.5 (self-built)
4.1.0-0.bpo.2-rt-amd64 (snapshot.debian.org)
4.4.0-trunk-rt-amd64 (self-built)
4.6.0-0.bpo.1-rt-amd64 (snapshot.debian.org)
so I am ready for testers and for review of the patches.  I *will*
continue rebasing the branch until its inclusion in the master branch,
so take care when fetching new versions.

Known limitations include 
 - hm2_eth and hm2_spi will only work with PREEMPT-RT realtime, since they 
   make Linux syscalls from a realtime context. (this problem is not
   detected, you'll just probably not get good RT performance)
 - until buildbot is reorganized, pre-built linuxcnc-uspace packages
   will not include RTAI support
 - It's not clear what range of RTAI versions can be supported with
   the same binary.
 - Xenomai kernels are "build it yourself", though their directions are
   clear and easy to follow.

The diff relative to our master branch at the moment looks big:
 59 files changed, 18564 insertions(+), 249 deletions(-)
this is because in order to support older Linux distributions, I include
a copy of some newer boost headers.  Ignoring this, the diff is much
smaller:
 22 files changed, 890 insertions(+), 242 deletions(-)
each new RTOS is under 200 lines of code. (is there an alternative to
boost's lockless queue that would be in all Linux distros we are
targeting for master branch, and not RTOS-specific?)

The code is on git.linuxcnc.org and github.com/linuxcnc/linuxcnc in
branch jepler/master/uspace-plus and
https://github.com/LinuxCNC/linuxcnc/pull/110

When you configure linuxcnc, specify --with-relatime=uspace and look for
which RTOSes were detected to support:
checking for realtime API(s) to use... uspace+lxrt+xenomai

If you give this a try, or if you review the changes and have any notes
for me, please put them on the pull request
https://github.com/LinuxCNC/linuxcnc/pull/110 or on this mailing list.

Jeff

--
What NetFlow Analyzer can do for you? Monitors network bandwidth and traffic
patterns at an interface-level. Reveals which users, apps, and protocols are 
consuming the most bandwidth. Provides multi-vendor support for NetFlow, 
J-Flow, sFlow and other flows. Make informed decisions using capacity planning
reports.http://sdm.link/zohodev2dev
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers