#linuxcnc-devel Logs

Oct 27 2019

12:39 PM linuxcnc-build: build #5449 of 1301.rip-precise-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1301.rip-precise-rtai-i386/builds/5449 blamelist: andypugh <andy@bodgesoc.org>
12:42 PM cerna: andypugh: It looks like the OSX is capable of soft real-time tasks. Which could be enough for machine control with MESA-like hardware with proper sanity checks (given that one could argue that the current Ethernet solution on MESA Anything I/O is soft RT), but for LPT or other kinds, I think it is insane given that the kernel can arbitrarily change the priority. But - I think - the better solution would be remote NML on
12:42 PM cerna: OSX and Linux RTOS on some small ARM chip for machine control.
12:42 PM andypugh: Well, that’s the Machinekit approach. Except their plan is to replace NML, I think.
12:43 PM linuxcnc-build: build #4062 of 1401.rip-wheezy-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-rtai-i386/builds/4062 blamelist: andypugh <andy@bodgesoc.org>
12:45 PM rene_dev_: pcw_mesa pcw_home I need a 7i76e + 7i74 firmware, do you have that?
12:46 PM andypugh: cerna: But then you might as well use GRBL or Mach3 with Smoothstepper, which already do that perfectly well.
12:46 PM cerna: Yeah well, I am active in Machinekit development, so I may be tinged with these ideas. (I have no interest in penis measurement contests.)
12:51 PM linuxcnc-build: build #3382 of 1902.clang-wheezy-rtai-i386 is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1902.clang-wheezy-rtai-i386/builds/3382 blamelist: andypugh <andy@bodgesoc.org>
12:52 PM cerna: There is some code in LinuxCNC base that indicate port to BSD. (I have no idea if it works.) So somebody tried for portability. The basic HAL should not be that much problem. Question is the whole sauce and side dish around the main parts.
12:52 PM andypugh: Do you know if Machinekit have any documentation for hal_gpio or hal_bb_gpio?
12:53 PM andypugh: (These buildbot errors are a result of me porting the hal_gpio driver to LinuxCNC from Machinekit)
12:54 PM cerna: Sorry, there is pretty much no documentation for Machinekit. Code is the only "documentation".
12:57 PM andypugh: It feels like there should be scope for a single GPIO driver for all the SBCs. But I can’t see myself writing it.
01:02 PM linuxcnc-build: build #6247 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6247 blamelist: andypugh <andy@bodgesoc.org>
01:04 PM cerna: I think that the only thing you would achieve is another form of abstraction build above what vendors are shipping their boards with. If Linux would ship it, then fine, because you would have other people maintaining it, but doing it yourself is - I think - outside the scope what LinuxCNC can push in terms of workforce.
01:07 PM cerna: Maybe some kind of meta HAL component for other HAL components?
01:10 PM andypugh: My idea was a header file with a database of GPIO port data and the corresponding /proc/cpuinfo
01:10 PM andypugh: All the GPIO drivers look much the same.
02:14 PM rmu: andypugh: does it make sense to build GPIO driver for i386 target?
02:28 PM andypugh: No
02:29 PM andypugh: But I am not sure how to tell what the target is in the makefile.
02:29 PM andypugh: Any ideas?
02:29 PM pcw_home: rene_dev_: 7i76e_7i76x1_7i74x1D.bit
04:03 PM Lcvette: pcw_home: how can I tell if a field input is bad? Currently I cannot see pin 26 and 27 that are connected properly and receiving 24v at the screw terminal correctly but not registering in Hal show? The configs were previously working.
04:05 PM pcw_home: On what I/O card?
04:05 PM Lcvette: pcw_home: 7i77 TB7 pins 26 and 27
04:06 PM Lcvette: They are connected to the scale input from the mpg selector switch
04:06 PM pcw_home: you mean inputs 26 and 27?
04:06 PM Lcvette: Yes
04:08 PM pcw_home: bad inputs are pretty unlikely I would suspect wiring or connectors
04:09 PM pcw_home: (also make sure you use "watch" )
04:09 PM Lcvette: Wiring was previously working and the terminal screws show change with the selector switch
04:09 PM Lcvette: Watch?
04:09 PM Lcvette: Oh instead of halshow
04:10 PM Lcvette: Oh in Hal show
04:10 PM Lcvette: Yes I am
04:10 PM pcw_home: halshow has show and watch options watch is realtime show is not
04:10 PM Lcvette: Red and yellow dots that change
04:11 PM Lcvette: The mpg uses a common 24v and all outputs from that same common to the 7i77
04:11 PM Lcvette: Axis selector inputs work still
04:11 PM Lcvette: They are on 20, 21, 22, 23
04:12 PM Lcvette: The wheel shows counts properly and it is stuck in x1 which is 0.0001" per pulse
04:12 PM Lcvette: Machine moves one tenth per click on the wheel
04:13 PM Lcvette: Just the scale selector switch does not change the halshow watch pins
04:13 PM pcw_home: I would check the connectors and possibly conductive dirt (the inputs are quite rugged so I would not expect some random inputs to fail)
04:14 PM Lcvette: I did have a wiring issue while connecting a probe with incorrect instructions
04:14 PM Lcvette: And it happened right after that as done or same time
04:15 PM Lcvette: I removed the probe wiring bit the mpg scale did not start working again
04:15 PM pcw_home: the inputs will take 120VAC continuously so its not likely that a wiring mistake caused damage
04:16 PM * Lcvette uploaded an image: IMG_20191025_214437.jpg (862KB) < https://matrix.org/_matrix/media/r0/download/matrix.org/dOWkkjGBFVHTgiFvPDadgrEk >
04:16 PM Lcvette: Bit it Bridger the tb1 and tb2
05:01 PM Lcvette: pcw_home: ^^^
05:35 PM andypugh: I am guessing that mozmck is not awake?
05:36 PM JT-Shop: he should be it's 5:30pm at his house
05:41 PM andypugh: But he is grey in the list
05:42 PM jthornton: he may not be home
06:17 PM linuxcnc-build: build #6250 of 0000.checkin is complete: Failure [4failed fetch branch to local git repo] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/6250 blamelist: andypugh <andy@bodgesoc.org>
06:57 PM rene_dev_: pcw_home thanks, no Idea why I could not find it :D
06:59 PM andypugh: mozmck_lp: Hi.
06:59 PM andypugh: I was going to ask you about pushing a Pi GPIO driver to 2.8
06:59 PM andypugh: But it’s too late now
07:15 PM mozmck_lp: Hi andypugh - sounds good then :-)
07:17 PM andypugh: Do you know if the makefile knows the architecture?
07:17 PM andypugh: I found some tests for rtai / uspace but not for armhf / amd64 etc
07:17 PM mozmck_lp: That I don't know. I just bought a PI4 the other day but haven't had time to look at it yet.
07:19 PM mozmck_lp: I know seb had a buildbot slave running on some arm system - but I don't think it was a pi
07:19 PM mozmck_lp: Seems like he took it offline though because it kept haven't problems or something?
07:20 PM andypugh: If you write the SD card from the image I put up, then pull 2.8, you could test the GPIO driver. I didn’t actually test that it worked, assuming that part was well tested. And I didn’t have anyhting handy to connect to the Pi
07:20 PM andypugh: I am wondeing about setting up my Udoo, Pi ad Rock64 as buildslaves.
07:24 PM mozmck_lp: If you want them to be under the buildmaster seb has you'll have to email him. I guess it would work to set up another buildbot too.
07:55 PM andypugh: Yes. If I do it then I will add them as correspondence memebers of his buildfarm