#linuxcnc-devel | Logs for 2016-08-30

[00:57:51] <linuxcnc-build> build #2 of 4025.deb-jessie-armhf is complete: Failure [4failed shell_3 shell_4] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4025.deb-jessie-armhf/builds/2
[07:26:14] <jepler> it doesn't always croak at exactly the same time in the boot process
[07:26:14] <jepler> [ 3.188389] ------------[ cut here ]------------
[07:26:14] <jepler> [ 3.191535] WARNING: CPU: 5 PID: 108 at kernel/irq/handle.c:151 handle_irq_event_percpu+0x168/0x174
[07:26:22] <jepler> sometimes it starts to print a traceback with a warning
[07:36:55] <jepler> if (WARN_ONCE(!irqs_disabled(),"irq %u handler %pF enabled interrupts\n",
[07:36:58] <jepler> irq, action->handler))
[07:37:01] <jepler> local_irq_disable();
[07:41:48] <skunkworks> ummm
[07:48:23] <skunkworks> easy peesy lemon squeezy https://www.youtube.com/watch?v=yRF16VmeVPE&feature=youtu.be
[07:48:42] <skunkworks> expecially now I will never understand the 'mach is so much easier to setup'
[07:58:41] <archivist> hehe lots of glossing over not explaining much at all
[08:05:29] <archivist> and I do see the it is harder reason, ours is a separate app for setting up.....
[08:15:13] <jepler> yeah, your favorite text editor!
[08:17:51] <skunkworks> :)
[08:22:26] <skunkworks> hmm - openbox doesn't seem to launch on this laptop
[08:23:20] <jepler> what, do you expect to see something on the screen?
[08:23:26] <jepler> try right clicking for the menu
[08:23:39] <skunkworks> oh boy
[08:24:03] <skunkworks> heh - ok
[08:24:06] <jepler> yeah only a nerd could love etc
[08:26:33] <skunkworks> hmm - surprisingly easy to use..
[09:15:40] <seb_kuzminsky> linuxcnc-build: force build --branch=master 0000.checkin
[09:15:40] <linuxcnc-build> build forced [ETA 1h34m48s]
[09:15:41] <linuxcnc-build> I'll give a shout when the build finishes
[09:22:58] <seb_kuzminsky> the jessie-armhf buildslave successfully built 2.7 debs last night
[09:23:20] <seb_kuzminsky> i think getting proper air flow helped - fan control is different in the u3 jessie kernel than it was in the wheezy kernel
[09:23:29] <seb_kuzminsky> pics here if you want to see how it turned out: http://highlab.com/~seb/linuxcnc/u3-case/pics/
[09:24:59] <seb_kuzminsky> i thought about connecting each u3's serial console to a usb port on the other, but meh
[09:33:32] <skunkworks> there doesn't seem to be any x session errors either when axis starts
[09:34:18] <skunkworks> seb_kuzminsky, cool
[09:35:53] <cradek> wow, even has the power passthrough thing
[09:36:02] <cradek> haven't seen one of those in a while
[09:36:36] <skunkworks> the manager does think it is active. (why it didn't seem to want to come up at the shop) After launching axis - you cannot click on the panel bar icon - it doesn't do anything. You have to click on the desktop - then back to the panel bar icon and it will show up.
[09:36:41] <cradek> seb_kuzminsky: for some reason I was picturing some enclosure thing you had cut out of a solid aluminum block
[09:37:31] <cradek> very nice ethernet ports on the back
[09:39:05] <seb_kuzminsky> yeah i kept all the ~1992-vintage AC input stuff
[09:39:39] <seb_kuzminsky> the ethernet ports on the back required me to enlarge the existing DB-25(?) holes that were there, then make that little plate
[09:39:46] <seb_kuzminsky> it came out right the first time, i was amazed
[09:40:45] <cradek> yay!
[09:40:46] <seb_kuzminsky> the new power supply is ~3x the current and 1/2 the volume of the old one (but the new one has no 12)
[09:41:12] <cradek> my sherline interface is in a similar box with similar modifications
[09:41:15] <cradek> they're really convenient
[09:41:22] <seb_kuzminsky> they're so good :-)
[09:41:41] <seb_kuzminsky> i made one for the robot arm too - it's got a 7i43 and a 7i54 in it
[09:41:52] <cradek> neat
[09:42:50] <jepler> seb_kuzminsky: I expected a bigger fan somewhere
[09:42:51] <seb_kuzm1nsky> huh, irssi crashed here
[09:43:27] <seb_kuzm1nsky> that little one between the AC and the ethernet is pretty high flow
[09:43:47] <jepler> I hope it makes those builders more stable
[09:43:52] <seb_kuzm1nsky> yeah
[09:45:03] <jepler> I'm bummed my rt kernel for the xu4 didn't work the first time.
[09:45:34] <jepler> if I can figure out *where* the problem is happening I bet I can make a stab at fixing it though
[09:45:45] <jepler> presumably it's in the xu4-specific drivers, and that patchset is actually pretty small
[09:45:49] <jepler> in number of commits anyway
[09:46:45] <jepler> 15 files changed, 5436 insertions(+), 69 deletions(-)
[09:46:58] <seb_kuzm1nsky> yikes
[09:47:08] <seb_kuzm1nsky> does it run without those patches?
[09:47:21] <jepler> don't panic, 4853 of those lines are odroidxu4_defconfig
[09:47:22] <seb_kuzm1nsky> that's a pretty inconvenient bisect process :-/
[09:47:33] <seb_kuzm1nsky> oh, that's good
[09:47:37] <seb_kuzm1nsky> bbl
[09:48:16] <jepler> the first kernel I built was 4.6.3+xu4 patches, and that worked. I went from there to 4.6.7-rt11. So one obvious next step would be to build 4.6.7 and see if that works
[10:04:15] <skunkworks> jepler, what device is this?
[10:06:06] <jepler> skunkworks: odroid xu4
[10:06:50] <jepler> 8 cores, 2GB, pi-like 40-pin I/O header via daughterboard
[10:07:17] <jepler> so I would imagine it driving a 7i90 via spi, just like I did with the u3
[10:13:44] <skunkworks> that is awesome - does the video seem ok?
[10:18:17] <skunkworks> my goodness. I thought maybe I could use devilspie to maximize axis. In the directions on devilspie it says to see what applications are running by doing a xlsclients -l
[10:18:26] <skunkworks> axis doesn't show up as an application
[10:22:05] <skunkworks> never mind. I think I found it
[10:24:34] <JT-Shop> https://forum.linuxcnc.org/forum/21-axis/29735-axis-modifications
[10:27:00] <linuxcnc-build> Hey! build 0000.checkin #4502 is complete: Success [3build successful]
[10:27:00] <linuxcnc-build> Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4502
[10:32:33] <skunkworks> JT-Shop, did you have any issues with axis starting minimized?
[10:34:19] <JT-Shop> I used devilspie on one machine
[10:35:09] <JT-Shop> the link has 3 different ways to start maximized
[10:35:29] <JT-Shop> IIRC the third one works the best
[10:36:47] <JT-Shop> and it doesn't work with linuxmint 17.3
[10:38:11] <skunkworks> The tk one?
[10:42:31] <JT-Shop> yea the maxgeo one
[10:44:22] <JT-Shop> this one works root_window.tk.call("wm","geometry",".","1200x600")
[10:44:33] <JT-Shop> adjusting the size for your monitor
[10:45:13] <skunkworks> none of it works (it resizes - you just still have to click on the taskbar icon for axis to come up.)
[10:45:15] <skunkworks> Just odd
[10:57:01] <skunkworks> fullsize=maxgeo.split(' ')[0] + 'x' + maxgeo.split(' ')[1]
[10:57:23] <skunkworks> errors with tuplet doesn't contain split
[10:58:07] <skunkworks> from the maximize code
[10:58:10] <skunkworks> maxgeo=root_window.tk.call("wm","maxsize",".")
[10:58:10] <skunkworks> fullsize=maxgeo.split(' ')[0] + 'x' + maxgeo.split(' ')[1]
[10:58:10] <skunkworks> root_window.tk.call("wm","geometry",".",fullsize)
[11:03:16] -linuxcnc-github:#linuxcnc-devel- [13linuxcnc] 15SebKuzminsky commented on issue #130: I'm assuming this bug report is against master, since you say it works in 2.7.6.... 02https://github.com/LinuxCNC/linuxcnc/issues/130#issuecomment-243481256
[11:11:06] <skunkworks> this is infuriating. now it is getting personal
[11:11:32] <skunkworks> the maximizing works great - it just doesn't display axis.
[11:12:07] <skunkworks> once you click on the task bar axis icon - it magically appears maximized.
[11:20:52] <JT-Shop> you mean that axis starts out minimized to the task bar?
[11:21:20] <skunkworks> minimized or just not showing. Not sure.
[11:21:30] <skunkworks> but yes
[11:21:30] <JT-Shop> what os?
[11:22:11] <skunkworks> This is jessie. It does it in gnome and cinnamin
[11:22:23] <skunkworks> mate works fine
[11:23:08] <JT-Shop> yea I run xfce or something like that
[11:23:29] <JT-Shop> I tried gnome once and almost passed out
[11:24:09] <skunkworks> yah - I don't like gnome. cinnamon looks nice - more windowsy for dad
[11:24:51] <jepler> you can try any or all of these: root_window.wm_deiconify(), root_window.focus_force(), root_window.tkraise()
[11:25:35] <JT-Shop> looks like I'm running mate ls /usr/bin/*session
[11:25:50] <JT-Shop> /usr/bin/ck-launch-session /usr/bin/mate-session
[11:26:27] <jepler> also the current way to request a window be maximized in Tkinter is root_window.wm_attributes(zoomed=True) afaik
[11:26:30] <jepler> errrr
[11:27:03] <jepler> or maybe root_window.wm_attributes('-zoomed', 1)
[11:28:36] <skunkworks> would you believe none of those work?
[11:28:40] <skunkworks> go figure
[11:29:57] <skunkworks> I can alt-tab to it.
[11:29:59] <skunkworks> also
[11:34:02] <seb_kuzminsky> yay, master built on jessie-armhf
[11:34:26] <seb_kuzminsky> i wonder why it takes 1 hr on jessie and 45 minutes on wheezy (both armhf, both same kind of SD card)
[11:35:22] <seb_kuzminsky> most of the difference is time spent in user space
[11:35:28] <seb_kuzminsky> maybe the jessie gcc is just slower
[11:35:57] <seb_kuzminsky> and jessie takes > 2x the RAM, like jepler said earlier
[11:39:56] <seb_kuzminsky> i'm planning to make a 2.7.7 soon, to release cradek's fix for yet another Task regression that i introduced
[11:40:06] <seb_kuzminsky> does anyone have anything they're itching to get into 2.7?
[11:40:51] <seb_kuzminsky> 2.7.7 will also fix the tcl 'true' bug
[11:43:34] <cradek> I don't know of anything
[11:43:50] <seb_kuzminsky> ok
[11:44:57] <seb_kuzminsky> i'm also planning to release 2.6.13, and announce it as the final 2.6 release barring the discovery of any major disasters
[11:45:16] <seb_kuzminsky> 2.7.0 is 1 year old next week, seems like enough overlap
[12:02:33] <kwallace2> Hmm, G76 threading doesn't seem to check for the active plane, should it?
[12:59:33] <cradek> what would it do with the plane?
[13:23:57] <kwallace2> I'm just trying to figure out how G76 should work. There is code that seems to keep track of X, Y, and Z so I thought maybe G76 should work in other planes. Thinking more, I suppose a function only needs to check the active plane if X, Y, or Z are used but their application changes with the active plane. This doesn't seem to be the case with G76.
[13:28:23] <cradek> I haven't heard that anyone wants G76 to work in other directions. I can kinda see it being used to cut a scroll, but that's still XZ
[13:30:15] <JT-Shop> skunkworks: did you see Dewey's post on Axis?
[13:31:24] <skunkworks> ?
[13:31:43] <JT-Shop> about maximizing Axis
[13:31:47] <skunkworks> ah
[13:31:51] <skunkworks> let me try that...
[13:35:14] <skunkworks> No error - but still doesn't display axis until I alt-tab or click on the task bar
[13:35:42] <skunkworks> let me post in that thread
[16:15:51] <jepler> in 2014 someone reported a problem on mint where chrome would come up but be transparent https://forums.linuxmint.com/viewtopic.php?t=169177
[16:16:18] <jepler> it seems like in 2 years they might have solved the bug that affected chrome, but maybe try jiggling the setting they suggest near the bottom of the thread?
[16:19:35] <jepler> https://bugs.chromium.org/p/chromium/issues/detail?id=365447
[16:19:50] <jepler> though on the chromium bug they say your workaround (alt tab around to it) doesn't work
[16:19:53] <jepler> >< Note that once a window has got into the transparent state, it won't reappear even if effects are disabled - you need to disable the fade effect, close any invisible windows, and re-open them again.
[16:22:39] <cradek> is it opengl + compositing bs biting us again?
[16:27:50] <jepler> just a hypothesis I got from searching the internet
[16:32:30] <jepler> sigh with gmane gone how do I send out links to mailing list posts?
[16:32:47] <jepler> anyway just reading this one which says it's a bad idea to fork after calling mlock. this is a thing we do in hm2_eth...
[16:33:12] <jepler> https://lkml.org/lkml/2016/8/30/123
[16:33:25] <cradek> gmane is gone!?
[16:33:58] <jepler> https://developers.slashdot.org/story/16/07/28/2059249/the-end-of-gmane
[16:34:34] <jepler> (I'm sure there are better references, but that one popped up in google, third result after "GMANE - Day Is Gone lyrics | Musixmatch" since apparently it's also the name of a band)
[16:35:06] <cradek> that's a terrible band name
[16:35:30] <jepler> I guess I should switch my personal mail archiver so that it makes mail-archive.com links https://www.mail-archive.com/faq.html#msgid though I don't know that linuxcnc mailing lists are on there
[16:42:19] <jepler> looks like mail-archive does have users and developers lists, at least a couple years
[16:48:10] <skunkworks> I seem to have a problem communicating... :)
[16:48:13] <skunkworks> https://forum.linuxcnc.org/forum/21-axis/29735-axis-modifications?start=10
[16:56:56] <kwallace2> I can't resist: https://www.youtube.com/watch?v=V2f-MZ2HRHQ
[17:50:22] <jepler> next xu4 irritation: rebooting from 4.6.3+ to 4.6.3+ without power cycling, the ethernet doesn't come up
[17:50:28] <jepler> [ 4.572953] hub 5-0:1.0: 3 ports detected
[17:50:28] <jepler> [ 8.784853] xhci-hcd xhci-hcd.9.auto: Command completion event does not match command
[17:50:32] <jepler> and loops for ~30 seconds
[17:52:34] <jepler> related? pulling the power plug doesn't actually power off the thing. I read something somewhere about back-powering the device from the HDMI connector and that seems to be the case
[17:53:22] <jepler> no not related, rebooting with video unplugged doesn't matter
[21:42:05] <jepler> https://emergent.unpythonic.net/files/sandbox/xu4-rt-linuxcnc.png
[21:42:16] <jepler> .. but I had to disable HDMI to get the kernel to build(!)
[21:42:36] <jepler> I think that by following internet folklore better I'll be able to get it to work with a display enabled
[21:43:08] <jepler> note: I haven't tried stressing it yet; the device regularly gets warm enough (with fan!) to enter CPU throttling and that is probably bad for latency.
[21:43:17] <jepler> 'night!
[22:01:05] <seb_kuzminsky> neato!