#linuxcnc-devel Logs

Apr 24 2019

#linuxcnc-devel Calendar

02:23 AM rmu: Tom_L: ubuntu 14.04: nominal release date in april '14
12:52 PM memfrob: Hi andypugh, I turned off all problematic options I'm aware of that conflict with IPIPE. Posting config shortly.
12:54 PM memfrob: You will need an initrd for it btw as ATA/PATA/SATA etc are modules, just like the debian config.
12:54 PM memfrob: I took the debian stretch config and modified it for RTAI.
12:55 PM memfrob: CPUs that have more than 8 cores are not currently supported by RTAI.
12:55 PM andypugh: All sounds good.
12:58 PM memfrob: https://pastebin.com/U3tu9jHC http://dpaste.com/05JV4MT
01:00 PM memfrob: I'm leaving now, that should get you started. early printk, verbose BUG() and KALLSYMS are on, default panic timeout is at 0, so if you get a kernel panic or anything, you'll see why.
01:00 PM memfrob: Have earlyprintk=vga enabled on the kernel command line if it doesn't boot, but as this is basically the debian config, all should be fine.
01:01 PM memfrob: Hope everything checks out, take care! If you need anything, let Cpt. Hindsight know :)
01:03 PM memfrob: I forgot to turn on PANIC_ON_OOPS so set oops=panic on the kernel command line. That's about it.
01:57 PM Connor is now known as connor
02:00 PM skunkworks: cradek: you around?
02:13 PM -!- #linuxcnc-devel mode set to +v by ChanServ
02:13 PM cradek: actually yes
02:14 PM skunkworks: heh
02:14 PM skunkworks: You got the email. So m8 is 68?
02:14 PM skunkworks: Dad has no clue - figures it it from grandpa's house
02:14 PM cradek: yes
02:15 PM skunkworks: when I finally get a battery for mine - I will see if it runs.
02:15 PM skunkworks: (Although it probably should be gone through..
02:15 PM cradek: nope, don't
02:15 PM skunkworks: OK
02:15 PM skunkworks: broken stuff if it is dry?
02:16 PM cradek: if there's a little corrosion on the pivots (which is very usual if an old battery is found in it) they tend to stick in their jewels, and the "motor" is actually strong to twist them off
02:17 PM cradek: even if it doesn't overtly damage anything you don't really want to run it dry and dirty
02:17 PM skunkworks: Well - you may be getting a watch soon :)
02:18 PM cradek: I have collected some photos of that kind of damage and keep meaning to add them to my "why should I have my accutron serviced" page
02:18 PM cradek: yeah I totally know a guy :-)
02:18 PM skunkworks: heh
02:18 PM cradek: what brand battery was in it?
02:19 PM skunkworks: Dad handed me the case - which was some other brand.. When I opened it - I couldn't believe it.
02:19 PM skunkworks: I will have to look - give me a second.
02:21 PM cradek: what a cool find
02:23 PM skunkworks: reyovac rw51
02:23 PM skunkworks: rayovac
02:29 PM skunkworks: cradek: http://electronicsam.com/images/greenmachine/20190424_142421.jpg
02:30 PM cradek: interesting I've never seen that one
02:31 PM cradek: it's great that it didn't leak
02:31 PM skunkworks: Yes!
02:31 PM skunkworks: what model is that then?
02:32 PM cradek: my very old battery cross reference book says that's a suitable sub for the 214
02:32 PM cradek: the movement is 214, the actual sales model I'm not sure, the only way to find out is to dig through old ads
02:33 PM skunkworks: ah
02:55 PM skunkworks: Could it be a 240?
02:58 PM skunkworks: cradek: https://www.ebay.com/itm/1968-Bulova-Accutron-Rare-Model-240-Mens-Stainless-Steel-214-Watch-Hums-4U2Fix-/152943534647?_trksid=p2047675.m43663.l10137&nordt=true&rt=nc&orig_cvip=true
02:58 PM skunkworks: It has the dots on the face..
03:04 PM cradek: oooh https://www.mybulova.com/sites/default/files/uploaded_photos/user744/accutron%20214_0.jpg
03:04 PM cradek: looks like "model 240" is right
03:09 PM cradek: skunkworks: $110 in 1968 is $820 today
03:21 PM skunkworks: crzy.. A humming one on ebay when for 250...
03:22 PM skunkworks: oh - that was the one I linked
03:36 PM andypugh: Hmm, https://www.youtube.com/watch?v=gZw-nLf_XBg
03:37 PM andypugh: RTAI kernel, goes to blank screen and (static) cursor after “started apache htp server” https://www.youtube.com/watch?v=gZw-nLf_XBg
04:31 PM rmu: andypugh: it is probably starting X or something
04:32 PM andypugh: Yes, it boots ol in recovery mode
04:34 PM rmu: i would try booting to textmode and start X while logged in via ssh
04:51 PM CaptHindsight: just watched Andy's RTAI boot video
04:52 PM memfrob: Hi andypugh, try setting nomodeset on the kernel command line of the RTAI kernel
04:52 PM CaptHindsight: andypugh: ^^
04:54 PM andypugh: OK, give me a moment
04:54 PM rmu: did modesetting exist in kernel 3.16?
04:56 PM memfrob: Yes.
04:56 PM memfrob: It works on my R9 290X.
04:57 PM memfrob: andypugh, what GPU are you using?
04:57 PM rmu: hmm. debian wiki say since kernel version 2.6.29...
04:57 PM memfrob: That sounds right.
04:58 PM rmu: time flies
04:58 PM memfrob: I remember having KMS work on my old Radeon 9800 PRO AGP 8X card.
04:58 PM memfrob: 2.6.32 roughly.
04:58 PM andypugh: is that just grub, “e” then add “nomodeset” at the end of the parameters?
04:58 PM memfrob: Yep, you got it!
05:00 PM andypugh: “can’t find command “nomodeset”
05:00 PM andypugh: Hmm
05:01 PM rmu: andypugh: add it on the line with the other kernel parameters not on a separate line
05:01 PM memfrob: Yes it has to be tied to the linux /vmlinuz line
05:02 PM memfrob: linux /boot/vmlinuz-3.16.52-RTAI ro oops=panic nomodeset (for example)
05:06 PM andypugh: And we’re in!
05:06 PM andypugh: A funny corrupted graphics screen at first, but then it all seems OK
05:07 PM andypugh: So I will make that permanent and get on with RTAI and LinuxCNC
05:08 PM jthornton: Yippie!!
05:09 PM memfrob: Glad to hear!
05:24 PM andypugh: Interesting quirk, possibly in LinuxCNC, the linuxcnc config.h ends up with an unterminated “
05:24 PM andypugh: (just hand edited for the moment)
05:26 PM andypugh: 201: RTAPI KERNEL VERSION “3.16.52-RTAI
05:27 PM memfrob: LinuxCNC's configure is a bit broken, yes. I changed several lines of code into one line of `uname -r` which fixed it.
05:31 PM Tom_L: andypugh, hope you took notes...
05:31 PM Tom_L: :)
05:34 PM jthornton: yea take notes
05:37 PM CaptHindsight: if anyone sees Jepler tell him I say "Hello"
05:39 PM memfrob: Fix I needed: https://pastebin.com/HvQWvT0K http://dpaste.com/20ENVS5
05:40 PM memfrob: It works and helps prevent against people building LinuxCNC against kernels they aren't running (by mistake.)
05:42 PM Tom_L: This page is no longer available. It has either expired, been removed by its creator, or removed by one of the Pastebin staff.
05:43 PM memfrob: I had a feeling someone would have trouble with pastebin and or dpaste which is why I always post both links.
05:43 PM Tom_L: oh i had them both together
05:43 PM Tom_L: nm
05:43 PM memfrob: Heh!
05:43 PM Tom_L: been one long day
05:44 PM memfrob: I'd submit a PR but this would most likely be rejected.
05:45 PM andypugh: Hmm, I can’t help feeling that the complicated version was intended to do something slightly clever
05:45 PM memfrob: It probably is there because I'm assuming you can build LinuxCNC against a kernel you're not using.
05:45 PM andypugh: (I have no ide what, though)
05:46 PM andypugh: Yes, you can configure —with-realtime-….
05:46 PM memfrob: So while the code compiles, it will not run until you reboot into the intended kernel.
05:47 PM memfrob: 500+ undefined symbols all over the place :P
05:48 PM andypugh: Maybe somebody wanted to compile in a super-optimised non-realtime kernel then reboot into the realtime one ;-)
05:50 PM memfrob: I'm against that idea as it opens the door for error.
05:52 PM memfrob: Since the current logic is broken, then there's already error.
05:57 PM andypugh: Indeed.
05:58 PM andypugh: So, LinuxCNC compiled, but not running.
05:59 PM andypugh: https://pastebin.ubuntu.com/p/d7ChH3YCZr/
05:59 PM andypugh: Looks like the same thing doesn’t it?
06:03 PM memfrob: Yeah I don't know how you fixed it but you need to do a fix like the one I posted.
06:05 PM memfrob: Modify configure.ac -- make clean && ./autogen.sh && ./configure --your-options-here && make
06:05 PM andypugh: I just tried that, but might have the wrong quotes, I think I have ‘ rather than `
06:05 PM memfrob: I'd suggest rm -f configure to make sure a new one gets generated.
06:05 PM memfrob: Tidle, not quote.
06:06 PM memfrob: tilde, however you spell it.
06:07 PM memfrob: Wait, not a tilde. That's the wavy line.
06:07 PM andypugh: back tick, I think
06:07 PM memfrob: The one that comes up when you press tilde without pressing shift.
06:08 PM andypugh: Not helped by me haviong a Mac keyboard on the test machine, so I have to hunt for some characters. Including #
06:08 PM andypugh: (# is right-alt + £ )
06:08 PM memfrob: Oh god
06:10 PM andypugh: There is one VM on my actual Mac where I literally have to find a # somewhere else and copy-paste it.
06:10 PM andypugh: Makes editing G-code less fun.
06:15 PM memfrob: Sounds like a lot of xkbmap hacking :)
06:15 PM CaptHindsight: 🍻
06:15 PM CaptHindsight: could be worse
06:30 PM andypugh: OK, so, LinuxCNC compiled, but both LinuxCNC and latency-test crash quite badly (need to hard-reset the PC)
06:30 PM andypugh: I think I need to look at the RTAI compile.
06:31 PM memfrob: Can you try the RTAI testsuite?
06:32 PM memfrob: As root: bash /usr/realtime/testsuite/run (I believe)
06:36 PM memfrob: If that fails, it's either a kernel, BIOS or hardware problem. I didn't test that config myself but just quickly modified it based on the debian config.
06:39 PM memfrob: I'll go through the kernel config some more and see if I spot anything.
06:46 PM andypugh: literally “bash” as a command?
06:46 PM memfrob: Yes.
06:46 PM andypugh: OK.
06:46 PM andypugh: I just ran it in the terminal without the bash, and results were not great
06:46 PM memfrob: High latency?
06:47 PM andypugh: (seemed to freeze, I ctrl-C-ed it and the first line of the latency test output then appeared, then the system froze
06:47 PM memfrob: Does the system you're running had issues with RTAI in the past? Does the LinuxCNC live cd with RTAI on it work?
06:49 PM andypugh: I haven’t tried, this machine has only ever been run with preempt-rt
06:49 PM memfrob: Ok, that might be why.
06:50 PM andypugh: Worth a try, I guess
06:50 PM memfrob: For RTAI, you need a good BIOS.
06:50 PM andypugh: (this machine actually gives <6000nS latency with preempt-rt )
06:51 PM memfrob: Preempt_rt has also been tested on more x86 hardware than RTAI. Is it intel?
06:51 PM andypugh: Yes
06:53 PM andypugh: I’ll try running an RTAI LiveCD in live mode tomorrow
06:53 PM andypugh: But I think it’s time to stop for tonight.
06:56 PM andypugh: latency min -(15 digit number) latency max 8388608
06:56 PM memfrob: Ok, if it works on the live cd I'll re-work the config a bit more.
06:56 PM memfrob: But I don't think it's config related.
06:56 PM andypugh: I suspect I need to take a more careful look at RTAI itself
06:56 PM andypugh: Yes, the kernel seems otherwise OK
06:56 PM memfrob: For RTAI, the defaults using my tree are fine.
06:57 PM memfrob: Unless your CPU has more than 4 cores, then you must change that value.
06:58 PM memfrob: I have tested that tree as-is across mulitple Gentoo systems.
06:58 PM memfrob: All AMD hardware.
07:00 PM memfrob: andypugh, As a test and to debug, you can try loading the rtai modules one by one and look for any stack trace.
07:01 PM memfrob: Control+Alt+F2 -> login -> as root: insmod /usr/realtime/modules/rtai_hal.ko && insmod /usr/realtime/modules/rtai_sched.ko
07:03 PM memfrob: dmesg should say something
07:04 PM andypugh: OK, I will try that tomorrow.
07:04 PM memfrob: It might help to add && sleep 1 && dmesg to the end of that
07:04 PM memfrob: Sounds good!
07:07 PM andypugh: One bot of good news, all this fiddling has made me realise that the buzzing noise from my monitor wasn’t that the backlight inverter was dead, it was that the speaker jack plug from the MB to the (integral) monitor was pulled half out.
07:07 PM andypugh: So the machine is now a lot more bearable to work on.
07:09 PM andypugh: FWIW, the machine is one of these: https://www.alphr.com/desktop-pcs/29371/rm-one-ecoquiet-965-review
07:10 PM andypugh: Bougght as something easy to pick up and put back in the cupboard when not testing hardware. (It currently has onbaord and PCI parports, a Mesa 6i25, Mesa 7i43 and a Mesa 7i80DB connected to it.
07:11 PM andypugh: So one of each of the main Mesa interfaces (EPP, PCI and Ethernet. Only SPI missing)
07:35 PM Tom_L: does he still sell the 7i43?
07:36 PM Tom_L: the 7i90 is a cheaper counterpart iirc
07:36 PM Tom_L: (with more IO)
07:36 PM andypugh: Yes.
07:37 PM Tom_L: when i screwed mine up i replaced it with a 7i90
07:37 PM andypugh: I think that the 7i43 might still be better at something. Not sure what. I have a feelign it’s one of the fastest cards
07:37 PM Tom_L: really?
07:37 PM andypugh: (cpu clock)
10:37 PM pcw_home: The 7I90 (and all Spartan6 based cards) are faster: 100 MHz ClockLow ( CPUs Stepgens SPI etc ) 200 MHz, ClockHigh ( PWM mostly )