#linuxcnc-devel Logs

Apr 17 2017

#linuxcnc-devel Calendar

08:22 AM JT-Shop: seb_kuzminsky: do I need to do anything to test that?
09:01 AM pcw_home: I tried running hazzy, its very easy to test. it has a clean look that I like, but it gets its clean look by being fixed scale
09:01 AM cradek: aww
09:01 AM cradek: that is a choice that people often make
09:02 AM pcw_home: yeah
09:03 AM kwallace2: That's what PathPilot does. The system hardware is fairly well fixed (static) so a lot of work can be avoided.
09:05 AM skunkworks: pcw_home, what size screen is it designed for?
09:07 AM pcw_home: looks like maybe 4x3 aspect ratio
09:07 AM pcw_home: and small (maybe 1024 wide?)
09:08 AM pcw_home: obviously quite preliminary
09:09 AM skunkworks: sure - I think he started with the craftman screen that phizled out on the forum
09:09 AM pcw_home: you set things by clicking on the numbers
09:09 AM pcw_home: (like homing)
09:09 AM kwallace2: Does anyone happen to know if modern appliances have something like CANbus?
09:11 AM pcw_home: I thought it was car/industrial only but I know nothing about modern appliances
09:13 AM pcw_home: skunkworks: its really easy to test, just fetch/download, make one symbolic link and change the DISPLAY setting in an INI file (or run one of the provided sims)
09:30 AM skunkworks: I have been meaning to test it
10:32 AM seb_kuzminsky: kwallace2: a guy at the local hackspace has a motorized wheel chair that's all canbus
10:33 AM cradek: did it start out that way?
10:34 AM seb_kuzminsky: jthornton: i'm running the 2.7 test suite, built for uspace, on an rtai (wheezy) vm, no problems so far
10:34 AM seb_kuzminsky: cradek: yeah, it came that way from the factory
10:35 AM seb_kuzminsky: he and another dude reverse engineered the protocol and are driving it around by remote control now, flashing the lights and things, it's pretty cool :-)
10:36 AM cradek: neat
10:38 AM andypugh: Hacking wheelchairs sounds sightly evil.
10:38 AM seb_kuzminsky: well it's his wheelchair...
10:38 AM cradek: nooo having full understanding and control of machinery you depend on is imperative, not evil
10:39 AM andypugh: Yeah, I guess as long as it is your wheelchair, rather than taking remote control of other wheelchairs.
10:40 AM seb_kuzminsky: yeah hijacking other peoples wheelchairs would be evil
10:40 AM seb_kuzminsky: huh, they made Forbes, of all things: https://www.forbes.com/sites/thomasbrewster/2016/08/06/hacking-wheelchairs-xbox-def-con-zero-security/#11d62adc16f3
10:41 AM seb_kuzminsky: http://boulderhackerspace.com/2016/12/21/stephen-in-the-news/
10:43 AM andypugh: Is it fair to complain that CAN isn’t secure? Isn’t the same true of every other control system?
10:44 AM skunkworks: pm select still out of stock - now until 4.24
10:44 AM andypugh: You could potentially take charge of a car with a CAN bus, but you need physical access to the vehicle.
10:45 AM cradek: andypugh: not when your car is actually a mobile phone on wheels
10:45 AM cradek: internet-connected edutainment system
10:46 AM seb_kuzminsky: Tesla does full OTA firmware upgrades
10:46 AM skunkworks: there was some news articles where a couple guys where able to get into someones car through the cell data connection. iirc. then got into the can bus and could apply brakes and such
10:46 AM andypugh: I would hope that the WiFi bluetooth and Telegraph systems on a car don’t have access to the ABS CAN module, but might be disappointed.
10:46 AM cradek: you would probably be very disappointed
10:47 AM seb_kuzminsky: https://www.wired.com/2016/08/jeep-hackers-return-high-speed-steering-acceleration-hacks/
10:47 AM skunkworks: https://www.wired.com/2015/07/hackers-remotely-kill-jeep-highway/
10:47 AM seb_kuzminsky: they apply the brakes over the air
10:47 AM seb_kuzminsky: heh yeah
10:47 AM skunkworks: heh
10:48 AM cradek: sooooo disappointed
10:50 AM andypugh: The article seems to say that they could (a) Before a patch kill the engine wirelessly (b)Both before and after a patch operate the brakes and steering by plugging in to the CAN bus physically. They then say “How bad would it have been if we had been able to do both at the same time”
10:50 AM seb_kuzminsky: oh, i just realized that the opening of the article skunkworks linked to is an homage to Fear and Loathing in Las Vegas
10:51 AM seb_kuzminsky: http://www.rollingstone.com/politics/news/fear-and-loathing-in-las-vegas-19711111
10:58 AM skunkworks: andypugh, my rube Goldberg.. https://pastebin.com/kAmkttHc
10:59 AM skunkworks: like I say - I think peter mentioned a simpler way to reset the counts - but I don't remember what it was
11:00 AM andypugh: I don’t think that there is any need to reset the counts, all the things that use counts work on deltas, and ignore changes then count-enable is false, so you just need to net count-enable to someting like machine.is-on
11:01 AM skunkworks: maybe that was it...
12:53 PM seb_kuzminsky: jthornton: a couple of successful complete runtests runs of 2.7-uspace on an rtai machine, so i can't reproduce the crash you mentioned
01:01 PM skunkworks: pcw_home, can I flash a 5i25 bit file to a 7i92?
01:51 PM skunkworks: archivist, can you search your logs?
01:51 PM skunkworks: irc logs
01:52 PM archivist: mebe what for
01:52 PM skunkworks: references to 7i76skunktest.bit
01:55 PM archivist: ew have to do a message like ""; waiting for table scan .
01:55 PM archivist: 2 rows in set
01:56 PM archivist: http://paste.debian.net/928088/
01:57 PM skunkworks: ok - thanks. I am pretty sure that is the bitfile I want ;)
01:59 PM skunkworks: (I remember tom making me one)
02:00 PM archivist: locate file you probably still have it
02:02 PM skunkworks: I do :)
02:02 PM skunkworks: Now the question is - can I flash that to a 7i92 (it was a 5i25 bit file)
02:08 PM Tom_itx: 7i76skunktest isn't in my logs
02:08 PM Tom_itx: got another search criteria?
02:09 PM Tom_itx: http://tom-itx.no-ip.biz:81/~webpage/cnc/skunkworks/
02:09 PM * Tom_itx bet's on that one
02:10 PM Tom_itx: skunkworks, likely not without a recompile for that board
02:11 PM archivist: Tom_itx, http://tom-itx.no-ip.biz:81/~webpage/cnc/configs/sherline/bitfiles/PIN_7I76_skunktest.vhd
02:12 PM Tom_itx: that's not there now that i see
02:13 PM archivist: I must admit I cheated with the search like '%skunktest%' to catch anything
02:13 PM Tom_itx: heh
02:13 PM Tom_itx: there's stuff on the server i'm not even aware of now :D
02:14 PM Tom_itx: there is a bitfile in that directory but under another name
02:14 PM archivist: I know the problem, started migrating a customes site, reduced it from 5gig down below 1 and still some more unused stuff to remove
02:15 PM Tom_itx: i probably have more on there not attached to html files than actual web content
02:27 PM skunkworks: Tom_itx, thanks!
02:28 PM skunkworks: If you get some time - could you make one for the 7i92?
02:46 PM KGB-linuxcnc: 03Chris Morley 05master 2104916 06linuxcnc 10lib/python/hal_glib.py hal_glib -fix spelling mistake and fix forced update function * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2104916
02:48 PM andypugh: Ah, Chris_Morley: Did you see this pnnconf discussion? I am happy to push the fix if it looks like an actual fix: https://forum.linuxcnc.org/39-pncconf/32609-square-one-with-5i25-7i76?start=10#91295
03:19 PM skunkworks: put the latest rt_preempt on a system that had poor performance. So far so good. 50us - with 4.9.20
03:19 PM skunkworks: let it run overnight. this was a 2nd gen i5
03:22 PM skunkworks: It sure is easy to build a rt_preempt kernel :)
03:26 PM skunkworks: jepler, you don't have to re-compile linuxcnc if you change rt_preempt kernels? (I mean - it seemed to work)
03:26 PM skunkworks: (rip)
03:38 PM seb_kuzminsky: skunksleep: you generally don't need to rebuild linuxcnc if you switch one rt-preempt kernel for another, i think
03:41 PM jepler: skunksleep: right, rt_preempt doesn't depend on any kernel headers or any specific kernel version or module API
05:30 PM Tom_itx: is the 7i92 the same pin count as the 5i25?
05:38 PM Tom_itx: mmm, skunksleep can't seem to get that pc to boot right now
05:39 PM Tom_itx: ubuntu 10.04 says graphic mode isn't working
06:42 PM Tom_itx: log skunkworks:
07:56 PM Chris_Morley: andypugh That looks like a good fix fel free to push - thanks you.
07:57 PM Chris_Morley: man .. ya don't look at a program for a long while -forget everything about it lol
08:26 PM skunkworks: Tom_itx, I thought it did. (it has an external db25 and an internal header - just like the 5i25
09:16 PM pcw_home: "skunkworks pcw_home, can I flash a 5i25 bit file to a 7i92?"
09:16 PM pcw_home: Sure, if want brick the 7I92 and have to recover via JTAG
09:34 PM skunksleep: Probably don't want to do that....
09:35 PM skunksleep: pcw_home: thanks.. I thought I remembered some bit files work on more than 1 similar device. Maybe the 5i25 and the 6i25?
09:59 PM pcw_home: Yeah and 5I24/6I24
10:01 PM pcw_home: but the PCI interface/protocol of a 5i25 is pretty distant from the Ethernet MAC interface/protocol of a 7i92
10:01 PM pcw_home: and unlike fancier cards the 7I92 has no spare EEPROM
10:05 PM skunksleep: Heh