#linuxcnc-devel Logs

Jan 10 2018

12:07 AM seb_kuzminsky: ugh, i upgraded the big gantry machine at the hackspace from 2.8.0~pre1.3266.ga3b7e to .3496.ga67d00c and it broke everything
12:09 AM seb_kuzminsky: now as soon as i power the machine on (F2 in Axis) one of the gantry joint motors lurches and the machine ferrors and powers off the amps
12:29 AM seb_kuzminsky: there are some changes to motion regarding joint jogging, synchronized homing, probing, wheel jogging
12:29 AM seb_kuzminsky: that seems the most likely
12:29 AM seb_kuzminsky: bisecting the history would be more enjoyable if this machine built faster...
12:51 AM seb_kuzminsky: went back to 2.8.0~pre1.3266.ga3b7e and it's happy again
01:53 AM seb_kuzminsky: sim/axis/gantry works fine in master
04:32 AM linuxcnc-build: build #3468 of 1405.rip-wheezy-armhf is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1405.rip-wheezy-armhf/builds/3468 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>
05:45 AM linuxcnc-build: build #5316 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/5316 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>
07:51 AM JT-Shop: pcw_home: I'm flashing a 7i92M and it's stuck on Verifying EEPROM about 3/4 of the way, any ideas?
07:56 AM skunkworks_: so - the debian stretch uspace kernel doesn't have the meltdown mitigation?
07:56 AM skunkworks: Kernel/User page tables isolation: enabled in dmesg
07:56 AM skunkworks: * I don't see
07:57 AM skunkworks_: (I really just wanted to see if it effected latency)
08:12 AM skunkworks: ah - dist upgrade
08:12 AM skunkworks: [ 0.000000] Kernel/User page tables isolation: enabled
08:16 AM skunkworks_: so far so good. 10663
08:24 AM skunkworks_: 36us
08:24 AM jepler: skunkworks_: thanks for testing. it probably merits a new live image.
08:24 AM skunkworks_: so far so good.
08:24 AM jepler: in theory it adds work to each context switch so I'd not be surprised if it adds a few us.
08:26 AM skunkworks_: windows 7 virtual machine running too
08:27 AM skunkworks_: still peaking at 36
08:27 AM skunkworks_: poor laptop - fan is almost full speed
08:34 AM pcw_home: JT-Shop: hit control C and retry
08:35 AM JT-Shop: ok
08:39 AM JT-Shop: pcw_home: thanks that worked
08:41 AM pcw_home: Ive seen this on Windows (which drops packets when it feels like it) but not Linux
08:41 AM skunkworks_: cradek, room in the shop? https://lincoln.craigslist.org/tls/d/warner-swasey-4-machine-lathe/6450728591.html
08:41 AM pcw_home: Mesaflash assumes a perfect connection and will hang forever waiting for a response if a packet is ever dropped
08:42 AM JT-Shop: I might need a new ethernet cable...
08:45 AM pcw_home: might cycle the power and ping the card for a bit to make sure everything is OK
08:48 AM JT-Shop: ok
01:23 PM jepler: "r10" images have the kernel with "meltdown" mitigation http://www.linuxcnc.org/testing-stretch-rtpreempt/
01:26 PM skunkworks_: Thanks jeff
02:00 PM seb_kuzminsky: yay!
02:00 PM seb_kuzminsky: do you think it's time to advertise those images more widely?
02:01 PM seb_kuzminsky: the concern with those images is that we think they'll have generally worse latency than the wheezy/rtai images we're currently promoting, right?
02:21 PM hazzy: On my Atom D525MW the latency is worse with with the stretch than with the wheezy ISO, but on my i5 laptop the stretch r9 ISO seems better. I think that maybe preempt might be better than RTAI on certain more modern machines ...
02:25 PM seb_kuzminsky: it sure varies from machine to machine. it'd be interesting if newer machines did better with rt-preempt generally
04:25 PM MarkusBec: did somebody testet what impect the last microcode update have with preempt or rtai kernel
04:26 PM MarkusBec: it should double the time of a context switch
04:35 PM seb_kuzminsky: yikes
04:43 PM skunkworks: I did some testing this morning.. It still worked for this laptop
04:43 PM skunkworks: (rt_preempt)
07:35 PM skunkworks: 41us
08:08 PM pcw_home: all day?
08:58 PM jepler: seb_kuzminsky: yes I'd like to publicize it more but I'm also a coward
11:23 PM seb_kuzminsky: sure, the scary thing about users is they find bugs