Pull request generated:
https://github.com/LinuxCNC/linuxcnc/pull/718
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Hi,
I have 4.19 working with RTAI, LinuxCNC needs this patch however to solve build
errors and warnings:
From 2765d6190515df6332d2a40326bf65177fb16794 Mon Sep 17 00:00:00 2001
From: Alec Ari
Date: Fri, 10 Apr 2020 00:29:43 -0500
Subject: Updates for kernel 4.19 support
Signed-off-by: Alec Ari
The PREEMPT_RT developers know what they're doing, the reason you need to
specify efi=runtime manually is because of the horrible latency that comes with
it (depending on CPU/motherboard/BIOS.)
Alec
___
Emc-developers mailing list
I'm all for Andy being the release manager, does this mean seb will share the
keys with him?
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
That's what happens when you lose your slaves! :P
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Alright, my last email had some issues and I'd like to clear that up.
The problem with the kernel being 8 byte aligned instead of 16 byte aligned is
not a "problem" but rather it'd be wasted stack space to use the full 16 bytes.
The kernel also uses no red zone, which is technically against the
Beautiful, that fixed the misalignment issue ergo no GPF hence no crash,
WINNING!!! I'll be merging the new-libm branch into master soon (tomorrow after
sleep.) Get ready to make new debs, I'm going to bump the kernel patches to the
latest of the 4.14 kernel series (after I fix some conflicts.)
Oh, I almost forgot! If it still crashes, try the `crash-test` branch of my
RTAI tree and see if that works. Beware, that branch will not compile on any
version of GCC older than 7.
Alec
___
Emc-developers mailing list
I'm curious to see if this change I made fixes it for you:
https://github.com/NTULINUX/RTAI/commit/dbbd0db2096724bb0ca7f3319157f6f6bd314be1
You need to be on the new-libm branch (do a git pull to make sure you have the
latest changes)
RTAI and LinuxCNC need fresh compile, make sure LinuxCNC
>It seems that we might need to make some decisions about whether we
>abandon kernel-mode realtime or put some effort in to future-proofing
>the build system.
Andy, this is a very easy problem to fix. The only gotcha is that it doesn't
crash on my end, so I'll need you to test the changes. Now
Would using `long long` instead of `long` be a potential fix then? I'm not a
very good C coder but I've noticed higher precision using `long long` over all
else, never seen `long long` fail for me.
The commit I referred to is this one, but andypugh probably has a better idea
on this than I do,
Hi,
I encounter problems like this all the time; you use a version of software,
update the software, and a bug gets introduced between versions when the code
very seemingly related to the problem, HASN'T CHANGED. If the driver code
hasn't changed, is it the same Linux kernel, distribution,
Hi,
So are Intel GPUs being added back into Kconfig? Can we leave nouveau (nvidia
drivers) out, or do you guys want these too?
I still haven't looked into the /dev/fb issue but I just want to be able to
plan the IPIPE Kconfig adjustments.
Alec
___
sue?
# CONFIG_FB_INTEL is not set
> On 21 Dec 2019, at 12:06 pm, bari wrote:
>
> Here is his paste:
>
> https://pastebin.com/ZC2LhMe5
>
>
>
> On 12/20/19 7:00 PM, Alec Ari via Emc-developers wrote:
>> Phill I'm not seeing your files.. Post to dpaste or something? An
My mistake,
So the config option for FBCON is CONFIG_FRAMEBUFFER_CONSOLE which is already
enabled.
Does /dev/fb0 exist?
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
Phill I'm not seeing your files.. Post to dpaste or something? And sure, I'm
not going to be busy for awhile so might as well make my time mean something.
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
Phill,
Can you post the output of:
grep "FBCON" /boot/config-*rtai*
Thank you!
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Oops..
$ find /lib/modules/*rtai* -name "fbcon.ko"
Fixed.
Alec
On Tuesday, December 17, 2019, 8:14:54 PM CST, Alec Ari via Emc-developers
wrote:
Phill,
Can you try running insmod on the fbcon driver in your RTAI kernel module
directory? Something like:
$ find /lib/mod
Phill,
Can you try running insmod on the fbcon driver in your RTAI kernel module
directory? Something like:
$ find /lib/modules/*rtai* -name "fbcon"
Then as root:
# insmod .ko
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
Does `lsmod` show fbcon?
Alec
On Monday, December 16, 2019, 9:51:15 PM CST, Phill C
wrote:
> On 17 Dec 2019, at 2:18 pm, Alec Ari via Emc-developers
> wrote:
>
> Phillip, you need the fbcon driver loaded. If this is too much of a hassle I
> can just add the DRM
Dec 2019, at 7:38 am, Alec Ari via Emc-developers
> wrote:
>
> Hi everyone,
> Are people more willing to risk possibly higher latency with RTAI to use
> Intel integrated graphics to save them the trouble of using FBDEV or VESA
> drivers? If so, I can re-enable these Kconfig option
Well, if people are only having latency spikes with parts of their system not
related to GFX, I don't see the problem with re-enabling these drivers.. If it
does become an issue, specify `nomodeset` on the kernel command line and this
should solve that. When I have time, I'll get around to it.
"But latency-test blows a 4 u-s reading out when a beep from incoming mail
occurs puttin it above 100 u-secs, and that is while running the stretch
iso"
A bit off-topic but what email client are you using? This is very odd..
Alec
___
Emc-developers
Hi everyone,
Are people more willing to risk possibly higher latency with RTAI to use Intel
integrated graphics to save them the trouble of using FBDEV or VESA drivers? If
so, I can re-enable these Kconfig options. If you haven't had latency trouble
before with Intel graphics, I can't imagine a
Ah, I guess the Debian kernel config doesn't enable FBCON so you need to tell
Debian to always load this module.
Maybe something like:
mkdir -p /etc/modules-load.d
printf "# Load fbcon driver\nfbcon\n" &> /etc/modules-load.d/fbcon.conf
If that doesn't work, post /var/log/Xorg.0.log and output
I told you the fix before, and I will tell you again. Intel and nVIDIA DRM KMS
drivers have been disabled in kernel Kconfig.
To run at 1080p, 4K, or whatever you want, use xf86-video-fbdev:
xorg.conf:
Section "Device"
Identifier "Screen" # This can be anything [A-Z] [0-9]
Driver
I should mention, I've been running Radeon cards with IPIPE/RTAI for years and
never had any trouble by the way. Too many complaints and personal experience
with Intel and nvidia drivers lead to the decision to pull them out.
Alec
___
Emc-developers
Remove the "Cheers," from the link :P
On Sunday, November 17, 2019, 3:14:27 AM CST, Phillip Carter
wrote:
https://sourceforge.net/p/emc/mailman/emc-developers/Cheers, Phill
Original message From: Marius Date:
17/11/19 7:53 pm (GMT+10:00) To:
The i915 driver has been disabled in the kernel Kconfig as this driver is known
to cause latency problems. If the distribution was properly built, you can use
vesa or fbdev and have 1080p or 4K resolution. The i915 KMS DRM driver is not
required for high resolutions. Unless you are playing
Greetings RTAI users! TESTERS WANTED!
I have just finished completely re-working the math functions in kernel space
for RTAI.
Instead of putting nearly an entire C library in kernel space, I have trimmed
the living hell out of musl and have ported it for use with RTAI.
The amount of
Thanks Andy! I will add round into musl libm kernel space.
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Hi,
I'm doing the finishing touches on a new RTAI branch that will eventually get
merged into the `master` branch.
One of the things I noticed in LinuxCNC is that in src/rtapi/rtapi_math.h there
is this line:
extern double round(double);
But this apparently isn't used anywhere. If you comment it
Hi Andy!
I'm working on trimming musl down and extracting it's libm to it's core. This
will solve all warnings during the module build stage, and won't need to modify
LinuxCNC's code at all from it's current state. This is an on-and-off project
and it's going to take awhile to work out all the
`adduser` is one of those commands specific to Debian so this isn't friendly
across all Linux systems. Debian's `adduser` is actually a standalone perl
script. The gentoo devs for example symlinked `adduser` to `useradd` to bring
back some familiarity:
It is ill-advised to use adduser/usermod. Instead, use gpasswd to append users
to groups. This prevents changing the preexisting groups of the intended user:
https://www.redhat.com/archives/rhl-list/2004-September/msg02595.html
It's less prone to damage and is accepted across other
Great, out of nowhere I'm getting these now:
make[1]: Leaving directory '/usr/src/linux-headers-4.14.148-rtai-amd64'
WARNING: "atan2" [/home/ntu/linuxcnc/src/scarakins.ko] undefined!
WARNING: "sqrt" [/home/ntu/linuxcnc/src/scarakins.ko] undefined!
WARNING: "acos"
Hello,
I have isolated the bare minimum of math functions that need to be provided by
rtai_math.ko for LinuxCNC.
Currently, in both my master branch and RTAI mainline, rtai_math.ko depends on
a C library (such as musl) to provide the math functions. Two problems come
with this:
1.) It is
If it adds 2GB, it adds 2GB. Get a larger SD card if you need it, and your
problem has nothing to do with buildbot at all. BuildBot and installing
dependencies that dpkg catches are two completely unrelated, and different
things. This is the best I can do to help you.
Alec
Well either someone didn't talk about installing glib dev packages in the wiki
or the instructions weren't followed closely enough. Either way, you need to
install the glib dev packages. There's at least a couple hundred packages in
debian for glib, all named very similarly. This is why I hate
Hi,
sincos in LinuxCNC works as-is with my RTAI master branch here:
https://github.com/NTULINUX/RTAI
The code I was working on before was just to migrate math functions from
rtai_math.ko (musl git master) to GCC's built-in functions.
The sincos dpaste link I was working on that andypugh
Q/A issue I just noticed, linux headers and image are based off of two
different kernel releases.
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Andy,
Glad everything built well, thank you for spending the time making these!
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Hi,
For python-vte, it's labelled "debports'" not "backports" but either way, they
don't make the package available through synaptic. Maybe need to modify
sources.list or do it through the command line (like apt-get install -t
debports python-vte) I haven't looked into this much, but I haven't
andypugh, no-calibration was a test branch specifically for you to try, not
make public debs with. I'm guessing that the no-calibration branch works for
your hardware though? Can you make debs off the HEAD (tip) of the current
master branch? The commit missing from that branch is this:
Branches merged, you should not _have to_ run `make menuconfig` in RTAI for any
reason. ./configure is synced with Kconfig and there is no functional change
between Kconfig defaults and using an empty ./configure line. If ./configure
is not working for you, something is wrong.
Alec
Thank you, Robert!
Once the debs get built for RTAI itself, I'll need latency scores from a few
systems under load, most notably when opening Firefox. The Ryzen system I have,
latency will jump from 10 microseconds to 150, but on my AMD Fusion APU, I do
not have this problem. I'm not sure if
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
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.
Hey everyone,
I skimmed through this discussion a bit, I use Pluma most of the time. In the
Linux world, the never ending war has been between Vim vs. Emacs. I don't use
either, ever. If I don't have a graphical environment going, I use nano, an
editor that absolutely everyone makes fun of.
Hello,
I would just use this:
https://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-stable-rt.git/log/?h=v4.14-rt
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
If you're used to building kernels all the time, Xenomai on the Rockchip should
be a non-issue. The Xenomai mailing list is really helpful too. It'd be best to
start with PREEMPT-RT though and make sure that works first. If high latency is
the only issue during testing (no kernel panics,
There is this:
https://www.rtai.org/userfiles/downloads/RTAICONTRIB/rtai_rtnet.tar.bz2
Haven't tested it or anything.
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
Hello,
RTAI 5 is unstable at the moment so building kernels for it doesn't make much
sense. If you need RTAI, I recommend using this tree:
https://github.com/NTULINUX/RTAI (master branch or 4.9.9 release tarball)
As for the kernel not booting, make sure the drivers for your hard disk, SSD or
Hello,
Disable WithRoot with Xenomai, but leave it for systems that need it? What
system configuration would need it? I think it's fine and this whole root issue
isn't anything to worry about if it's not conflicting with anything. Xenomai
2.X is EOL so I wouldn't worry about supporting it.
What should the behavior be? What would you say is the "right way" to fix it?
It sounds like you don't approve of your solution. LinuxCNC shouldn't need root
access anyway, sudo make setuid should handle all that for you, so it looks
fine to me. If you had to manually run certain things as
What's the problem exactly? Maybe I can help?
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
Xenomai developers have been working on RTnet a lot lately, how much work would
it be to be able to use it instead of PREEMPT_RT POSIX threads over ethernet?
Just asking out of curiosity. RTAI has brought it back as well:
https://www.rtai.org/userfiles/downloads/RTAICONTRIB/rtai_rtnet.tar.bz2
Yahoo keeps screwing with the formatting of my emails after I send them. I
promise you that my line breaking skills are not that bad. Sorry about that!
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
Hello,
I had to re-write a convoluted part of configure.ac to get it to detect what
kernel I'm running. Forget the exact error, but this is very messy in the state
it's in right now:
---cut here---if test $RTS '!=' uspace; then
AC_MSG_CHECKING([for kernel version string])
if test -e
LinuxCNC being passed the wrong ./configure option shouldn't cause the result
it did. A NULL pointer deference is never an appropriate method of informing
the user that something is wrong in their configure line. Please do not accept
this as the appropriate behavior.
Alec
Hello,
I would prefer addressing this problem on the public mailing list. I do not
mind anyone contacting me directly. The version of RTAI the LinuxCNC developers
are using is a depreciated fork of RTAI 3, which is also depreciated upstream.
NULL pointer deference bugs is usually caused by an
Thanks Andy, all clear now! :)
Alec
___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers
When adding new commands, some of these things are a bit obscure to me. This
doesn't affect me personally but for others who want to add new gcodes it might
help to document the following:
1.) How to tell what the debug address should be, for example in that commit:
configs/sim/axis/lathe.ini
I think the majority of PREEMPT_RT users are Intel, I get 100 microseconds
latency on AMD hardware with my trimmed down (ideal) custom kernels. RTAI in
worst conditions reach 20 microseconds (big bloated kernel with lots of
debugging features on, Youtube, Firefox, Thunderbird, 500 instances of
I was under the impression that seb, cradek, and jepler handle PRs..
Alec
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
Wow!! Of all places, it's 15 minutes away from Porsche's factory and their
museum, I wish I could go! Sounds like a fun trip!
Alec
--
Check out the vibrant tech community on one of the world's most
engaging tech sites,
This was just added to the kernel tree about two months ago:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/gpio/gpio-raspberrypi-exp.c
Would this be of any use?
Alec
--
Check out the
I see nothing in that list that I haven't seen before.. e1000 and r8169 have
been in the tree forever.
Alec
--
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org!
The only RTnet repo I was aware about is here:
https://sourceforge.net/p/rtnet/code/ci/master/tree/
With the last change being from 4 years ago.
Alec
--
Check out the vibrant tech community on one of the world's most
RTnet only really works on 2.6 kernels. There was some work done for kernel 3.2
support but it wasn't usable. It's all EOL at this point and completely
useless, don't even think about RTnet anymore. Just use preempt-rt with POSIX
threads over the network. You will have the low-latency ethernet
I'm not sure what you mean by "sim" but all the axis simulation configs work
for me on 64-bit LinuxCNC installs. To my knowledge, there are no 64-bit
LinuxCNC and RTAI .deb packages however.
Alec Ari
--
Check out the
>To set up read access, first 'cd' to somewhere inside the clone, then issue:
>git config remote.origin.url https://github.com/LinuxCNC/linuxcnc.git
>If you are a developer with push access, you should also issue
>git config remote.origin.pushurl g...@github.com:LinuxCNC/linuxcnc.git
201 - 272 of 272 matches
Mail list logo