Re: [Emc-developers] Release Platforms

[andy pugh]

On Mon, 27 May 2019 at 21:44, Alec Ari  wrote:

> This bit:
>
>
> https://github.com/LinuxCNC/linuxcnc/blob/master/src/configure.ac#L597
>
> KERNEL_VERS=`uname -r` works, what is there now, doesn't.
>

Ah, yes, I remember now.
Sorry.

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[Alec Ari via Emc-developers]

This bit: 


https://github.com/LinuxCNC/linuxcnc/blob/master/src/configure.ac#L597

KERNEL_VERS=`uname -r` works, what is there now, doesn't.

Alec


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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Mon, 27 May 2019 at 20:10, Alec Ari via Emc-developers <
emc-developers@lists.sourceforge.net> wrote:

Between kernel 3.4.55 (RTAI LinuxCNC live cd) and 3.16.52 (my old heavily
> tested stable tree) -- A lot of changes are made. To my memory, Andy was
> able to build the kernel and RTAI, and even load/unload modules without a
> problem. I believe it would only panic if trying to run the test suite or
> any real-time code. I trust LinuxCNC to start a real-time thread correctly,
> and I've tried the RTAI test suite myself on numerous combinations, so this
> points to hardware.
>

I will try that kernel again with Jessie.


> The biggest problem to me is that LinuxCNC has a line splitting problem
> and has trouble detecting the running kernel (it cuts things off or
> something) I sent the patch to andypugh on IRC -- Can this get looked into?


I don't remember this, either I missed it, or didn't understand the problem
description.
Any chance you can raise it as an issue on the bug tracker?

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[Alec Ari via Emc-developers]

Hi,
I came in a bit late. My RTAI for LinuxCNC repository doesn't have any issues 
on the numerous hardware combos here. If it isn't working, then it's a hardware 
problem.
The reason the LinuxCNC live cd probably works as-is is because of the magic 
combination of kernel version and hardware. Some kernel versions expose 
hardware problems moreso than others. I had a Biostar board once, and only one 
out of 10 kernels would work (kernel code at trampoline bit, crazy early.) It's 
not that the other 9 kernels were necessarily broken, they just didn't work 
with the motherboard. The same kind of thing can happen with IPIPE/RTAI.

Between kernel 3.4.55 (RTAI LinuxCNC live cd) and 3.16.52 (my old heavily 
tested stable tree) -- A lot of changes are made. To my memory, Andy was able 
to build the kernel and RTAI, and even load/unload modules without a problem. I 
believe it would only panic if trying to run the test suite or any real-time 
code. I trust LinuxCNC to start a real-time thread correctly, and I've tried 
the RTAI test suite myself on numerous combinations, so this points to hardware.
The biggest problem to me is that LinuxCNC has a line splitting problem and has 
trouble detecting the running kernel (it cuts things off or something) I sent 
the patch to andypugh on IRC -- Can this get looked into?
Alec

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

Re: [Emc-developers] Release Platforms

[Gene Heskett]

On Sunday 26 May 2019 05:16:23 pm andy pugh wrote:

> On Wed, 22 May 2019 at 18:36, Gene Heskett 
> wrote:
>
> 1: remove any usb-serial devices plugged in so it doesn't setup brltty
>
> > for the braille using blind at install, its broken and will disable
> > all your usb-serial adapters until you sudo apt remove brltty .
>
> Can you add a bug tracker for that, and a guide how to test /
> reproduce the problem?
> https://github.com/LinuxCNC/linuxcnc/issues
>
> https://github.com/LinuxCNC/linuxcnc/issues
>
I would Andy, but github already knows me by my only email address and 
will not let me reuse my old account data to login. No clue what the 
hell they were thinking, if they were thinking at all. One item is that 
they've now made the username I used, now has an illegal character, a @ 
in the middle of it. Paranoid b-rds


> > 2: The hplip package, which includes hpfax, turned out to be a total
> > showstopper crash, intermittent after about a days uptime, and
> > random until I found an error recorded in syslog, and renamed the
> > hpfax binary to hpfox. You can't remove hplip as it takes your
> > desktop with it.  So just disable hpfax. It apparently goes funkity
> > when cron calls it and it has nothing to do, but it kills hid-common
> > locking you out of the machine, with the front panel reset button
> > your only rescue.
>
> Again, please raise this as an issue, and perhaps point to an online
> discussion with more info?

Apparently I'm the only one, but it has also (the hpfax problem) attacked 
my new but wheezy install for the 6040 gantry mill. But its only killed 
me 2 or 3 times in several months, whereas it was taking this machine 
down 1 to 3x a day. BUT! since fixing that, I've lost just the 
keyboard a couple times since on this house machine, getting 3 to 7 days 
uptime, but I still have a working mouse, so I can do a legit reboot.

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)
Genes Web page 



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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Wed, 22 May 2019 at 18:36, Gene Heskett  wrote:

1: remove any usb-serial devices plugged in so it doesn't setup brltty
> for the braille using blind at install, its broken and will disable all
> your usb-serial adapters until you sudo apt remove brltty .
>

Can you add a bug tracker for that, and a guide how to test / reproduce the
problem?
https://github.com/LinuxCNC/linuxcnc/issues

https://github.com/LinuxCNC/linuxcnc/issues


> 2: The hplip package, which includes hpfax, turned out to be a total
> showstopper crash, intermittent after about a days uptime, and random
> until I found an error recorded in syslog, and renamed the hpfax binary
> to hpfox. You can't remove hplip as it takes your desktop with it.  So
> just disable hpfax. It apparently goes funkity when cron calls it and it
> has nothing to do, but it kills hid-common locking you out of the
> machine, with the front panel reset button your only rescue.


Again, please raise this as an issue, and perhaps point to an online
discussion with more info?


-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Thu, 23 May 2019 at 17:03, Chris Radek  wrote:

>
> Would you push your changes to live-build somewhere so I can see
> them (and for posterity?)


Done that too :-)

https://github.com/SebKuzminsky/linuxcnc-wheezy-iso/commit/89631c4523bf772af08fa6660c8e35069d24b95f


-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[Chris Radek]

On Thu, May 23, 2019 at 10:11:26AM +0100, andy pugh wrote:
> 
> Already done, I have the .iso and .iso.zsynch files ready and tested.

Awesome, I wasn't able to get it to work yet.  Got up to isolinux
failing (no such package available), and went to bed.

Would you push your changes to live-build somewhere so I can see
them (and for posterity?)

Thanks for doing it!
Chris


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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Thu, 23 May 2019 at 10:32, andy pugh  wrote:

>
> And I am a bit confused, as the build process (make install) complains
> about not having the kernel headers. But as far as I can see none of the
> kernel building guides shows how to generate the headers.
> Am I missing something? Is this why I failed before?
>

And maybe I am wasting my time? :-)

I just found
http://www.linuxcnc.org/dists/jessie/base/binary-amd64/linux-image-3.16.0-9-rtai-amd64_3.16.7-6linuxcnc_amd64.deb

Is this usable?

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Wed, 22 May 2019 at 23:55, andy pugh  wrote:

>
>
> On Wed, 22 May 2019 at 20:05, Peter C. Wallace  wrote:
>
> As far as RTAI goes, Has anyone tried Bari's suggestion of NTULINUX/RTAI?
>>
>> https://github.com/NTULINUX/RTAI
>
>
> Yes, I tried both this and the 4.9.18 patch from RTAI.org. Both run Linux
> perfectly well, but crash immediately anything realtime is attempted on my
> test machine.
>

I had another go last night, on Jessie rather than stretch.

And I am a bit confused, as the build process (make install) complains
about not having the kernel headers. But as far as I can see none of the
kernel building guides shows how to generate the headers.
Am I missing something? Is this why I failed before?

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Thu, 23 May 2019 at 05:35, Randy Steiner via Emc-developers <
emc-developers@lists.sourceforge.net> wrote:

Also I ran across another dual-kernel version for RT, https://xenomai.org/.
> has this been looked at before?


Yes, the LinuxCNC RTAPI also supports Xenomai.

See, for example:
https://github.com/LinuxCNC/linuxcnc/blob/master/src/configure.ac#L299

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Thu, 23 May 2019 at 04:46, Chris Radek  wrote:

>
> > I have tried to respin that ISO using the archive repositiories. So far
> > that isn't going all that well.
>
> I will work on this part.
>

Already done, I have the .iso and .iso.zsynch files ready and tested.

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[Gene Heskett]

On Thursday 23 May 2019 12:34:16 am Randy Steiner via Emc-developers 
wrote:

>  Robert, that is a good link and starts to confirm something else I
> had found on tuning preempt_rt. As the new guy I will likely post
> stuff that others already know and have reviewed. From what I can see
> it looks like RTAI just had a release at version 5.2 with support for
> kernel patches up to 4.14.111 For preempt_rt it looks since each IRQ
> handler shows as a process they can be prioritized separately.  In
> peoples testing, has specific tuning been done to see  if that impacts
> the paraport performance since that appears to be a concern point?  If
> not, I can try that thread and see if I can vary the results. Also I
> ran across another dual-kernel version for RT, https://xenomai.org/. 
> has this been looked at before? Thanks!
> Randy

Yes Randy, but its been several years now. I built a xenomai kernel for 
TLM, probably close to a decade back, and it worked well at the time. 
TLM is what my always drive broken 7x12 has been called, The Little 
Monster, but I've rebuilt it better and some hal trickery there to 
control the accel of the 1 hp spindle motor, plus getting rid of the 
willow tree the compound was, and giving it tapered gibs. It is pretty 
solid now, and a usefull tool today.  I made a lot of stuff while 
converting the old Sheldon on it.

Anyway the xenomai kernel drove it thru the parport just fine at the 
time. I would certainly encourage experimentation to be done by someone 
with a software stepper setup. But with the advent of the D525MW 
motherboards, and the Mesa 5i25's, I no longer have a software stepping 
setup to test it on.

> On Wednesday, May 22, 2019, 10:19:38 PM MDT, Robert Murphy
>  wrote:
>
>  Sorry for this second post..regarding latency, would this be an
> option
>
> http://www.machinekit.io/docs/hal/threads-and-latency/
>
> On 23/5/19 2:16 pm, Robert Murphy wrote:
> > Could not the repos be setup from a preseed ? (I'm don't know a
> > great deal about the LiveCD but I did make a custom stretch install
> > disk for private use using a preseed file to config desktop,
> > extra packages and repos)
> >
> > On 23/5/19 1:45 pm, Chris Radek wrote:
> >> On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:
> >>> We have a LiveCD based on Wheezy and RTAI but that is currently
> >>> somewhat
> >>> broken (the apt sources list points at files that are no longer
> >>> there now
> >>> that Wheezy is a long way past EOL)
> >>> I have tried to respin that ISO using the archive repositiories.
> >>> So far that isn't going all that well.
> >>
> >> I will work on this part.
[...]

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)
Genes Web page 



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

Re: [Emc-developers] Release Platforms

[Gene Heskett]

On Thursday 23 May 2019 12:18:50 am Robert Murphy wrote:

> Sorry for this second post..regarding latency, would this be an
> option
>
> http://www.machinekit.io/docs/hal/threads-and-latency/
>
This is an interesting little treatise. Let me suggest some further 
development to allow it to determine the best configuration and 
implement that variation, depending on the cache construction used in 
the cpu it finds itself running on.

Another way to unload the servo-thread that I use on the pi running my 
Sheldon is to setup a much slower thread, and move all the manual 
control stuff to it, thereby reducing the size of the servo-thread loop 
and its execution time. Not much use of the jog dials being handled much 
more than 100 times a second, and thats a fairly complex bit of hal code 
times 2. 100/sec is faster than one can turn the MPJA dials by hand and 
still maintain control although I can run into the MAXVEL's set with the 
dials when the jog size is set high. So I have that set for a max per 
click of 20 thou, going down to .0002 thou.

I have other hal routines that report via pyvcp for such as the overshoot 
and distance traveled during that over shoot when rigid tapping, which 
lets me tap a hole to a small fraction of a turn from striking the 
bottom of the hole and breaking the tap.  All that stuff is nice, but it 
also takes the poor pi's cores a while to calculate and display.

Such "fine tuning" by the user shouldn't be discouraged, and should be 
described in the wiki, or mentioned at least in the documentation.pdf so 
the user can more easily be made aware of such "optimizations."

> On 23/5/19 2:16 pm, Robert Murphy wrote:
> > Could not the repos be setup from a preseed ? (I'm don't know a
> > great deal about the LiveCD but I did make a custom stretch install
> > disk for private use using a preseed file to config desktop,
> > extra packages and repos)
> >
> > On 23/5/19 1:45 pm, Chris Radek wrote:
> >> On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:
> >>> We have a LiveCD based on Wheezy and RTAI but that is currently
> >>> somewhat
> >>> broken (the apt sources list points at files that are no longer
> >>> there now
> >>> that Wheezy is a long way past EOL)
> >>> I have tried to respin that ISO using the archive repositiories.
> >>> So far that isn't going all that well.
> >>
> >> I will work on this part.
> >>
> >>
> >>
> >> ___
> >> Emc-developers mailing list
> >> Emc-developers@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/emc-developers
> >
> > ___
> > Emc-developers mailing list
> > Emc-developers@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-developers
>
> ___
> Emc-developers mailing list
> Emc-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-developers


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)
Genes Web page 



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

Re: [Emc-developers] Release Platforms

[bari]

That is unstable beta. It will takes weeks of cleanup to be stable.

On 5/22/19 11:34 PM, Randy Steiner via Emc-developers wrote:
> From what I can see it looks like RTAI just had a release at version 5.2 with 
> support for kernel patches up to 4.14.111



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

Re: [Emc-developers] Release Platforms

[Randy Steiner via Emc-developers]

 Robert, that is a good link and starts to confirm something else I had found 
on tuning preempt_rt.
As the new guy I will likely post stuff that others already know and have 
reviewed.
From what I can see it looks like RTAI just had a release at version 5.2 with 
support for kernel patches up to 4.14.111
For preempt_rt it looks since each IRQ handler shows as a process they can be 
prioritized separately.  In peoples testing, has specific tuning been done to 
see  if that impacts the paraport performance since that appears to be a 
concern point?  If not, I can try that thread and see if I can vary the results.
Also I ran across another dual-kernel version for RT, https://xenomai.org/.  
has this been looked at before?
Thanks!
Randy
On Wednesday, May 22, 2019, 10:19:38 PM MDT, Robert Murphy 
 wrote:  
 
 Sorry for this second post..regarding latency, would this be an option

http://www.machinekit.io/docs/hal/threads-and-latency/

On 23/5/19 2:16 pm, Robert Murphy wrote:
> Could not the repos be setup from a preseed ? (I'm don't know a great
> deal about the LiveCD but I did make a custom stretch install disk
> for private use using a preseed file to config desktop, extra
> packages and repos)
>
> On 23/5/19 1:45 pm, Chris Radek wrote:
>> On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:
>>
>>> We have a LiveCD based on Wheezy and RTAI but that is currently
>>> somewhat
>>> broken (the apt sources list points at files that are no longer
>>> there now
>>> that Wheezy is a long way past EOL)
>>> I have tried to respin that ISO using the archive repositiories. So far
>>> that isn't going all that well.
>> I will work on this part.
>>
>>
>>
>> ___
>> Emc-developers mailing list
>> Emc-developers@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/emc-developers
>
>
> ___
> Emc-developers mailing list
> Emc-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-developers


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

Re: [Emc-developers] Release Platforms

[Robert Murphy]

Sorry for this second post..regarding latency, would this be an option

http://www.machinekit.io/docs/hal/threads-and-latency/

On 23/5/19 2:16 pm, Robert Murphy wrote:

Could not the repos be setup from a preseed ? (I'm don't know a great
deal about the LiveCD but I did make a custom stretch install disk
for private use using a preseed file to config desktop, extra
packages and repos)

On 23/5/19 1:45 pm, Chris Radek wrote:

On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:


We have a LiveCD based on Wheezy and RTAI but that is currently
somewhat
broken (the apt sources list points at files that are no longer
there now
that Wheezy is a long way past EOL)
I have tried to respin that ISO using the archive repositiories. So far
that isn't going all that well.

I will work on this part.



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



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



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

Re: [Emc-developers] Release Platforms

[Robert Murphy]

Could not the repos be setup from a preseed ? (I'm don't know a great
deal about the LiveCD but I did make a custom stretch install disk
for private use using a preseed file to config desktop, extra
packages and repos)

On 23/5/19 1:45 pm, Chris Radek wrote:

On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:


We have a LiveCD based on Wheezy and RTAI but that is currently somewhat
broken (the apt sources list points at files that are no longer there now
that Wheezy is a long way past EOL)
I have tried to respin that ISO using the archive repositiories. So far
that isn't going all that well.

I will work on this part.



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



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

Re: [Emc-developers] Release Platforms

[Chris Radek]

On Wed, May 22, 2019 at 01:31:15PM +0100, andy pugh wrote:

> We have a LiveCD based on Wheezy and RTAI but that is currently somewhat
> broken (the apt sources list points at files that are no longer there now
> that Wheezy is a long way past EOL)
> I have tried to respin that ISO using the archive repositiories. So far
> that isn't going all that well.

I will work on this part.



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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Wed, 22 May 2019 at 20:05, Peter C. Wallace  wrote:

As far as RTAI goes, Has anyone tried Bari's suggestion of NTULINUX/RTAI?
>
> https://github.com/NTULINUX/RTAI


Yes, I tried both this and the 4.9.18 patch from RTAI.org. Both run Linux
perfectly well, but crash immediately anything realtime is attempted on my
test machine.

The same machine gives 5k latency with the  stretch preempt-rt ISO and 8k
latency with the Wheezy RTAI ISO I just made.

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[Peter C. Wallace]

On Wed, 22 May 2019, Chris Morley wrote:


Date: Wed, 22 May 2019 18:20:11 +
From: Chris Morley 
Reply-To: EMC developers 
To: EMC developers 
Subject: Re: [Emc-developers] Release Platforms


From: andy pugh 
Sent: May 22, 2019 12:31 PM
To: EMC developers
Subject: [Emc-developers] Release Platforms


We have a problem...



We need a 2.7 LiveCD that works.
We will need a 2.8 LiveCD that works.


Thanks for tackling this Andy.

I would not give up the option for RTAI lightly. (I saw some talk on IRC about 
removing RTAI from the make file to simplify it)
based on the forum lots of people start with/ use the parport. (so some how 
they are getting it to work or using old systems)
AFAIK the RT kernel doesn't give good enough latency for parports.



Preempt-RT _can_ have good enough latency for parallel ports but the hardware 
with good Preempt-RT latency is different from hardware with good RTAI latency


In general my experience is that faster CPUs often have better Preempt-RT 
latency (often better than RTAI) but slower ( often passively cooled) CPUs

are much worse on Preempt-RT


As far as RTAI goes, Has anyone tried Bari's suggestion of NTULINUX/RTAI?

https://github.com/NTULINUX/RTAI





Thinking further ahead: What OS and realtime system should we plan to
release 2.8 on?


I will mention we were given permission to respin Mint:
https://forums.linuxmint.com/viewtopic.php?f=18=272185=1489079#p1489079
I think Mint is far more user friendly then Debian

I wonder if machinekit has had success with using RTAI still?

Chris M

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



Peter Wallace
Mesa Electronics

(\__/)
(='.'=) This is Bunny. Copy and paste bunny into your
(")_(") signature to help him gain world domination.



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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Wed, 22 May 2019 at 19:32, bari  wrote:

> https://github.com/NTULINUX/RTAI Is the last stable RTAI. linux-3.16.52
> is the latest kernel it supports.


That is next on my list, as I think that a Jessie ISO with the 3.16 RTAI
kernel should be relatively do-able.

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[bari]

We have crappy BIOS and motherboard makers to blame for this. BIOS was
always a train-wreck of spaghetti assembly. Then Intel decided to write
an entire new OS from scratch to embed into every motherboard vs just
use (LinuxBIOS) coreboot or similar.

Same cpu, chipset and memory settings should have the same results. But
BIOS/UEFI variations that are broken don't actually kill things like
power management when the GUI says choose ON or OFF. OFF might not do
anything to it.

On 5/22/19 8:16 AM, andy pugh wrote:
> On Wed, 22 May 2019 at 13:58, John Thornton  wrote:
>
>> I'd say something is better than nothing so if we can produce a LiveDVD
>> with Debian 9 and preempt-rt and LinuxCNC Uspace we should.
>
> Well, that already exists (and I have a few machines installed from it)
> http://www.linuxcnc.org/testing-stretch-rtpreempt/
>
> On one of my machines that gives the best latency numbers I have ever seen
> (< 5k) but I have also heard that it is terrible on some other machines.
>
> There is even a documentation patch lined up to make the switch:
>  https://github.com/LinuxCNC/linuxcnc/pull/500/commits
>
> I think that could happen instantly, at the cost of not offering a working
> RTAI option.
>



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

Re: [Emc-developers] Release Platforms

[Charles Steinkuehler]

On 5/22/2019 1:20 PM, Chris Morley wrote:
> I wonder if machinekit has had success with using RTAI still?

We dropped RTAI and kernel-mode HAL support ages ago.

I haven't tested recently, but Xenomai should be able to provide near
RTAI levels for latency with much easier to manage patch sets for
modern kernels if you can't get the performance you need from
PREEMPT_RT and user-space.

-- 
Charles Steinkuehler
char...@steinkuehler.net



signature.asc
Description: OpenPGP digital signature
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] Release Platforms

[bari]

https://github.com/NTULINUX/RTAI Is the last stable RTAI. linux-3.16.52
is the latest kernel it supports. We have it working with Gentoo and
Scientific Linux and it should work with any version of Debian. It will
support all the old hardware with LPT and software stepping.

Everything newer at rtai.org is unstable beta as Seb has proven. If some
big funded project comes along we might make it stable.

>From here on out preempt_rt may be used on newer hardware but leaves
software stepping with LPT questionable. It might be fine on some
hardware but much to slow on others.

-Bari


On 5/22/19 1:20 PM, Chris Morley wrote:
> From: andy pugh 
> Sent: May 22, 2019 12:31 PM
> To: EMC developers
> Subject: [Emc-developers] Release Platforms
>
>> We have a problem...
>> We need a 2.7 LiveCD that works.
>> We will need a 2.8 LiveCD that works.
> Thanks for tackling this Andy.
>
> I would not give up the option for RTAI lightly. (I saw some talk on IRC 
> about removing RTAI from the make file to simplify it)
> based on the forum lots of people start with/ use the parport. (so some how 
> they are getting it to work or using old systems)
> AFAIK the RT kernel doesn't give good enough latency for parports.
>
>> Thinking further ahead: What OS and realtime system should we plan to
>> release 2.8 on?
> I will mention we were given permission to respin Mint:
> https://forums.linuxmint.com/viewtopic.php?f=18=272185=1489079#p1489079
> I think Mint is far more user friendly then Debian
>
> I wonder if machinekit has had success with using RTAI still?
>
> Chris M
>
> ___
> Emc-developers mailing list
> Emc-developers@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-developers



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

Re: [Emc-developers] Release Platforms

[Chris Morley]

From: andy pugh 
Sent: May 22, 2019 12:31 PM
To: EMC developers
Subject: [Emc-developers] Release Platforms

>We have a problem...

>We need a 2.7 LiveCD that works.
>We will need a 2.8 LiveCD that works.

Thanks for tackling this Andy.

I would not give up the option for RTAI lightly. (I saw some talk on IRC about 
removing RTAI from the make file to simplify it)
based on the forum lots of people start with/ use the parport. (so some how 
they are getting it to work or using old systems)
AFAIK the RT kernel doesn't give good enough latency for parports.

>Thinking further ahead: What OS and realtime system should we plan to
>release 2.8 on?

I will mention we were given permission to respin Mint:
https://forums.linuxmint.com/viewtopic.php?f=18=272185=1489079#p1489079
I think Mint is far more user friendly then Debian

I wonder if machinekit has had success with using RTAI still?

Chris M

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

Re: [Emc-developers] Release Platforms

[Gene Heskett]

On Wednesday 22 May 2019 08:31:15 am andy pugh wrote:

> We have a problem...
>
> We need a 2.7 LiveCD that works.
> We will need a 2.8 LiveCD that works.
>
> Currently there is a 2.7 LiveCD running preempt-rt waiting in the
> wings. That is based on Stretch and so will be usable until June 2020.
> (ie, one more year)
>
> LinuxCNC has historically used RTAI, and this has tended to give
> better results for software stepping with the parallel port.
> We have a LiveCD based on Wheezy and RTAI but that is currently
> somewhat broken (the apt sources list points at files that are no
> longer there now that Wheezy is a long way past EOL)
> I have tried to respin that ISO using the archive repositiories. So
> far that isn't going all that well.
>
> Do we switch to the (working) preempt-rt Stretch ISO and abandon RTAI?
> Has anyone made RTAI work with a Linux version that is currently
> current? I have tried compiling for kernel 4.9 with Stretch and whilst
> it compiled and ran, and LinuxCNC compiled against it, nothing useful
> happened when I tried to run LinuxCNC.
>
> What I hoped we could do (urgently) is switch to the preempt-rt
> stretch image with an alternate download link to a (fixed ) Wheezy
> RTAI ISO with a disclaimer that we know it to be ancient.
>
> I guess we could link to it with a note that "This is so ancient that
> you will need to edit the apt sources before anything will work"
>
> Thinking further ahead: What OS and realtime system should we plan to
> release 2.8 on?

I'd vote for buster for the distro which is due to be marked stable in a 
few days, and whatever kernel in the 5.0 rt area from the linux-rt 
mailing list announcements that compiles and everything runs.  The 
current testing iso's kernel has truly amazing latency-test figures on 
this old amd phenom, might even run steppers thru the parport but its 
not been tried, but has two gotchas.

1: remove any usb-serial devices plugged in so it doesn't setup brltty 
for the braille using blind at install, its broken and will disable all 
your usb-serial adapters until you sudo apt remove brltty .

2: The hplip package, which includes hpfax, turned out to be a total 
showstopper crash, intermittent after about a days uptime, and random 
until I found an error recorded in syslog, and renamed the hpfax binary 
to hpfox. You can't remove hplip as it takes your desktop with it.  So 
just disable hpfax. It apparently goes funkity when cron calls it and it 
has nothing to do, but it kills hid-common locking you out of the 
machine, with the front panel reset button your only rescue. 

Thats it, I am using that testing install as this, my basic house machine 
right now, with an uptime of over 4 days since I found the hpfax problem 
and renamed it.

For a new install thats bound to have a wart or 6, I think its a keeper.

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)
Genes Web page 



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

Re: [Emc-developers] Release Platforms

[andy pugh]

On Wed, 22 May 2019 at 14:36, John Thornton  wrote:

> I'm all for changing the Docs to point to something that does work in
> most cases and add the link to prior version to the archives section
> along with notes about what is broken.


Do you have the access to move the various files to the correct places?

-- 
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, 1916

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

Re: [Emc-developers] Release Platforms

[John Thornton]

I'm all for changing the Docs to point to something that does work in 
most cases and add the link to prior version to the archives section 
along with notes about what is broken.


JT

On 5/22/2019 8:16 AM, andy pugh wrote:

On Wed, 22 May 2019 at 13:58, John Thornton  wrote:


I'd say something is better than nothing so if we can produce a LiveDVD
with Debian 9 and preempt-rt and LinuxCNC Uspace we should.


Well, that already exists (and I have a few machines installed from it)
http://www.linuxcnc.org/testing-stretch-rtpreempt/

On one of my machines that gives the best latency numbers I have ever seen
(< 5k) but I have also heard that it is terrible on some other machines.

There is even a documentation patch lined up to make the switch:
  https://github.com/LinuxCNC/linuxcnc/pull/500/commits

I think that could happen instantly, at the cost of not offering a working
RTAI option.




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

Re: [Emc-developers] Release Platforms

[John Thornton]

I'd say something is better than nothing so if we can produce a LiveDVD 
with Debian 9 and preempt-rt and LinuxCNC Uspace we should.


I had some partial success creating a live cd this morning with no 
desktop but with the rt kernel and LinuxCNC but it did boot so I'll 
attempt again tomorrow to add a desktop then I'd need to add the installer.


Recently I've built Lubuntu 18.04 with rt-preempt and LinuxCNC with 
surprisingly low latency. I still need to test my Debian 9 latency to 
get a 1 to 1 comparison on the same pc.


JT

On 5/22/2019 7:31 AM, andy pugh wrote:

We have a problem...

We need a 2.7 LiveCD that works.
We will need a 2.8 LiveCD that works.

Currently there is a 2.7 LiveCD running preempt-rt waiting in the wings.
That is based on Stretch and so will be usable until June 2020. (ie, one
more year)

LinuxCNC has historically used RTAI, and this has tended to give better
results for software stepping with the parallel port.
We have a LiveCD based on Wheezy and RTAI but that is currently somewhat
broken (the apt sources list points at files that are no longer there now
that Wheezy is a long way past EOL)
I have tried to respin that ISO using the archive repositiories. So far
that isn't going all that well.

Do we switch to the (working) preempt-rt Stretch ISO and abandon RTAI?
Has anyone made RTAI work with a Linux version that is currently current? I
have tried compiling for kernel 4.9 with Stretch and whilst it compiled and
ran, and LinuxCNC compiled against it, nothing useful happened when I tried
to run LinuxCNC.

What I hoped we could do (urgently) is switch to the preempt-rt stretch
image with an alternate download link to a (fixed ) Wheezy RTAI ISO with a
disclaimer that we know it to be ancient.

I guess we could link to it with a note that "This is so ancient that you
will need to edit the apt sources before anything will work"

Thinking further ahead: What OS and realtime system should we plan to
release 2.8 on?




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

[Emc-developers] Release Platforms

[andy pugh]

We have a problem...

We need a 2.7 LiveCD that works.
We will need a 2.8 LiveCD that works.

Currently there is a 2.7 LiveCD running preempt-rt waiting in the wings.
That is based on Stretch and so will be usable until June 2020. (ie, one
more year)

LinuxCNC has historically used RTAI, and this has tended to give better
results for software stepping with the parallel port.
We have a LiveCD based on Wheezy and RTAI but that is currently somewhat
broken (the apt sources list points at files that are no longer there now
that Wheezy is a long way past EOL)
I have tried to respin that ISO using the archive repositiories. So far
that isn't going all that well.

Do we switch to the (working) preempt-rt Stretch ISO and abandon RTAI?
Has anyone made RTAI work with a Linux version that is currently current? I
have tried compiling for kernel 4.9 with Stretch and whilst it compiled and
ran, and LinuxCNC compiled against it, nothing useful happened when I tried
to run LinuxCNC.

What I hoped we could do (urgently) is switch to the preempt-rt stretch
image with an alternate download link to a (fixed ) Wheezy RTAI ISO with a
disclaimer that we know it to be ancient.

I guess we could link to it with a note that "This is so ancient that you
will need to edit the apt sources before anything will work"

Thinking further ahead: What OS and realtime system should we plan to
release 2.8 on?

-- 
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, 1916

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