#linuxcnc-devel | Logs for 2016-10-17

[00:03:11] <KGB-linuxcnc> 03Dewey Garrett 05ncam_for_2.7 13e5e85 06linuxcnc 10configs/sim/axis/ncam_demo/mill-mm.ini 10configs/sim/gmoccapy/ncam_demo/mill-mm.ini 10configs/sim/gmoccapy/ncam_demo/mill.ini ncam sim configs [AXIS_n] section headings * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=13e5e85
[00:03:12] <KGB-linuxcnc> 03Dewey Garrett 05ncam_for_2.7 de53ca5 06linuxcnc 10(12 files in 6 dirs) ncam update to NativeCAM commit 632cf33 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=de53ca5
[00:03:24] <KGB-linuxcnc> 03Dewey Garrett 05master ae89934 06linuxcnc 10tcl/linuxcnc.tcl.in linuxcnc.tcl.in strip leading whitespace on [SECT] * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ae89934
[02:52:41] <linuxcnc-build> build #1173 of 1510.rip-jessie-rtpreempt-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1510.rip-jessie-rtpreempt-i386/builds/1173 blamelist: Dewey Garrett <dgarrett@panix.com>
[03:26:16] <linuxcnc-build> build #4561 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4561 blamelist: Dewey Garrett <dgarrett@panix.com>
[07:14:36] <jepler> [component "owner" changed from 4 to 3 in the result file, probably not something we intended to test]
[07:49:08] <skunkworks> zlog
[08:06:48] <tinkerer> @forum: the avatar image seems little blurred now
[08:32:28] <jepler> tinkerer: noted, frying bigger fish at the moment though
[08:32:31] <archivist> somewhat odd format http://www.collection.archivist.info/archive/grabs/Screenshot-43.png
[08:32:39] <jepler> archivist: yes it's afu sorry
[08:33:27] <jepler> I hadn't noticed that the navigation links were gone, and following internet advice on how to restore them added that banner and I have no clue why
[08:34:00] <archivist> too much added js by the look of it
[08:34:18] <jepler> do you want the goddamn admin password to this fucking website?
[08:35:15] <archivist> like I said I cannot state how long I am on the net for, £7 cash, meter on emergency, absolutely fscked here as well
[08:35:18] <tinkerer> yes, this is a common pain with cmss
[08:35:26] <jepler> then please sit back and enjoy the ride.
[08:35:32] <jepler> afk (from this window anyway)
[08:38:24] <tinkerer> good luck at shark killing ;)
[09:35:22] <jepler> tinkerer: looks like the resolution problem is fixed only when you upload a new picture in your profile
[09:57:52] <tinkerer> slightly
[09:59:27] <tinkerer> you can check it by switching between a post and the user profile
[11:29:07] <mozmck> jepler: norbert says he can't reply to your post here: https://forum.linuxcnc.org/29-forum-announcements/31701-forum-software-upgraded-new-default-theme#81708
[11:31:26] <cradek_> do you know what he means?
[11:31:43] <cradek_> I get a place to type a reply but I didn't try using it
[11:35:17] <jepler> mozmck: I think only administrators can post or reply in the announcement section
[11:35:31] <mozmck> I see.
[11:37:31] <mozmck> In case you did not get his message, he says, "I do not like the new design, as the headers aré very large and that results in a lot more scrolling."
[11:37:59] <mozmck> I would tend to agree, but that seems to be the "modern" way and may not be easily changed.
[11:38:02] <cradek> fwiw I replied to a random thread
[11:38:34] <jepler> I do not develop/maintain the theme for the forum, so more or less get a default theme
[11:38:58] <mozmck> The emails from the forum are also now html and the font is a lot larger.
[11:39:05] <jepler> *shrug*
[11:39:27] <mozmck> Yeah, and I bet making a theme is not real trivial anyhow.
[11:41:25] <cradek> I found some profiley things but don't see a way to select non-html emails or change anything about appearance
[11:41:37] <cradek> *shrug*
[11:41:53] <jepler> similarly I looked through kunena forum configuration in the administrator view and there's nothing to select plain text vs html on a global basis
[11:42:14] <jepler> there are two alternatives: live with it; take over the administration and maintainence of this albatross of a website from me and fix it yourself.
[11:42:32] <cradek> ooh ooh I know which one I want
[11:42:51] <mozmck> No big deal for me! The forum does seem to look a little nicer to me overall.
[12:14:27] <jepler> sigh, I need to work on my pitch; nobody is interested in taking over this task fro me.
[18:07:49] <dgarr> i have a patch for a recent buildbot runtests failure (master):
[18:07:49] <dgarr> http://buildbot.linuxcnc.org/buildbot/builders/1510.rip-jessie-rtpreempt-i386/builds/1173/steps/runtests/logs/stdio
[18:07:51] <dgarr> http://www.panix.com/~dgarrett/stuff/0001-tests-twopass-use-script-option-for-halcmd-show.patch
[18:07:51] <dgarr> however, i think the problem may be related to earlier commits (euid stuff 3d4fc3b) that result
[18:07:51] <dgarr> in buildbot slaves using rtpreempt to report:
[18:08:00] <dgarr> shmctl IPC_LOCK: Cannot allocate memory / shared memory segment not locked as requested / Note: Using POSIX non-realtime
[18:10:07] <tinkerer> dgarr: I think it's this stuff: https://github.com/LinuxCNC/linuxcnc/commit/5f0865a7eaa74def776f90a8b99af9db3a9578cc
[18:10:59] <tinkerer> but maybe I'm wrong
[18:22:26] <dgarr> bisecting on jessie/uspace/64bit it breaks for me at commit 3d4fc3b
[18:22:26] <dgarr> $ halrun
[18:22:26] <dgarr> halcmd: loadrt or2
[18:22:26] <dgarr> shmctl IPC_LOCK: Cannot allocate memory
[18:22:26] <dgarr> shared memory segment not locked as requested
[18:22:27] <dgarr> Note: Using POSIX non-realtime
[18:23:00] <dgarr> Linux t520 4.1.0-0.bpo.2-rt-amd64 #1 SMP PREEMPT RT Debian 4.1.6-1~bpo8+1 (2015-09-09) x86_64 GNU/Linux
[18:31:26] <jepler> I'll look into it, thanks dgarr
[18:32:09] <jepler> I didn't suspect it at first since buildbot was successful with that commit once http://buildbot.linuxcnc.org/buildbot/builders/1510.rip-jessie-rtpreempt-i386/builds/1171
[18:32:31] <jepler> though it is true that the new message is appearing in that build: shmctl IPC_LOCK: Cannot allocate memory
[18:32:35] <jepler> during runtests
[18:34:32] <cradek> stustev: so I hear there'll be beer
[18:37:13] <jepler> linuxcnc-build: force build --branch=master 0000.checkin
[18:37:14] <linuxcnc-build> build forced [ETA 1h25m02s]
[18:37:14] <linuxcnc-build> I'll give a shout when the build finishes
[18:37:30] <jepler> linuxcnc-build: force build --branch=3d4fc3b 0000.checkin
[18:37:36] <linuxcnc-build> The build has been queued, I'll give a shout when it starts
[18:37:39] <jepler> (does that work, to specify a ref with --branch?)
[18:37:49] <stustev> Hi cradek, this is Kim, setting up HexChat (for the week, anyway) on Stuart's PC. And yes, the keg is chilled. I have also set up #linuxcnc-fest so you can see who has arrived (I'll grant ops). Jon Elson has arrived, but is still hauling stuff.
[18:38:09] <cradek> yay!
[19:05:08] <linuxcnc-build> build #65 of 1540.rip-jessie-armhf is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1540.rip-jessie-armhf/builds/65
[19:09:51] <dgarr> Command terminated by signal 11
[19:11:23] <linuxcnc-build> build #4562 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4562
[19:11:23] <linuxcnc-build> build forced [ETA 1h25m02s]
[19:11:23] <linuxcnc-build> I'll give a shout when the build finishes
[19:11:35] <linuxcnc-build> build #4563 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/4563
[19:21:35] <mozmck> I'm afraid I won't make the fest after all. My dad is having some health problems and needs help.
[19:24:21] <jepler> mozmck: Understood. I hope things turn out OK.
[19:25:51] <mozmck> thank, me too. Up until today I had a room reserved at wichita inn airport for the week, so there's probably an open room there now.
[19:26:05] <mozmck> s/thank/thanks
[19:26:53] <KGB-linuxcnc> 05jepler/test-build-old-ref 3d4fc3b 06linuxcnc branch created * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3d4fc3b
[19:30:25] <jepler> I sort of suspect this is a race condition, but I'm still glad that dgarr spotted the bug I wrote wrt memory locking
[19:31:41] <jepler> for legacy reasons that I think aren't true any longer, halcmd (halcmd_systemv_nowait) temporarily shuts off its hal-component nature before starting a subprocess, then immediately recreates itself as a new component a moment later
[19:32:15] <jepler> the tests which look at component number assume that the hal_init of the newly loaded component and the hal_init of halcmd after it forks happen in a certain order
[19:32:26] <jepler> and many tests implicitly look at this because it is shown with e.g. 'halcmd show pin'
[19:36:05] <dgarr> but also latency is really bad on my jessie/uspace/64bit after commit 3d4fc3b
[19:42:39] <dgarr> isnt the report "Note: Using POSIX non-realtime" a problem when configured with --with-realtime=uspace ?
[19:42:39] <dgarr>
[19:43:37] <jepler> yes you are right
[19:51:37] <KGB-linuxcnc> 03Jeff Epler 05master ce52ea1 06linuxcnc 10src/rtapi/uspace_rtapi_app.cc uspace: fix check for setuid nature of rtapi_app * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ce52ea1
[19:51:37] <KGB-linuxcnc> 03Jeff Epler 05master f059548 06linuxcnc 10src/rtapi/uspace_rtapi_app.cc uspace: use WITH_ROOT for rtapi_open_as_root * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f059548
[19:51:37] <KGB-linuxcnc> 03Jeff Epler 05master 9799ece 06linuxcnc 10src/rtapi/uspace_common.h uspace: use root in RTAPI shmem_new so shm can be locked * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9799ece
[19:51:42] <jepler> dgarr: thank you for your patience in this matter, and please let me know if this last push still doesn't resolve the problem for you
[19:57:06] <dgarr> with those commits, no problems -- latency restored, report is "using POSIX realtime", if there still potential for a race, the commit i posted addresses different comp_ids
[19:58:58] <jepler> oh yes I think that commit is a good idea
[19:59:07] <jepler> but I see now my fingers didn't tell you that
[20:01:22] <jepler> thanks
[20:01:28] <linuxcnc-build> build #66 of 1540.rip-jessie-armhf is complete: Failure [4failed compile] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1540.rip-jessie-armhf/builds/66 blamelist: Jeff Epler <jepler@unpythonic.net>
[20:03:17] <jepler> > g++: internal compiler error: Killed (program cc1plus)
[20:03:44] <jepler> I think I'll have to try sending seb home with a different ARM builder board. I'll be taking a small boxful of the things, he can have his pick.
[20:05:09] <Roguish> jepler: quick simple question, please. sometimes linuxcnc crashes at startup and a process gets hung. how do I identify the process, and how do I kill it ???
[20:07:16] <jepler> we would prefer linuxcnc didn't crash :-P
[20:07:35] <jepler> if you are using a uspace realtime, then every process is visible in top and killable with kill or sudo kill
[20:07:59] <jepler> there are "mutexes" that sometimes need to be released manually, "halrun -U" does this
[20:08:34] <jepler> if you are aware of repeatable linuxcnc crashes that can be demonstrated without specific hardware, file issues on linuxcnc with instructions on how to reproduce.
[20:08:48] <jepler> if you are using rtai / kernel-based realtime, recovery is harder or impossible. sometimes rebooting is the only available option.
[20:11:31] <jepler> (it's best, in terms of recovering from problems, to develop non-driver components on uspace without "sudo make setuid" so that you never need sudo-steps to recover from problems; and for drivers, it's better to develop with uspace than rtai if you have the option)
[20:12:18] <jepler> you can see the linuxcnc script "Cleanup" section for another take on automatically cleaning up from a session that did not exit properly
[20:12:36] <jepler> kk# 3.2. define the cleanup function
[20:12:46] <jepler> Roguish: ^^ hope some of that brain dump helps, going afk
[20:13:03] <jepler> oh also be a 20+ year veteran of unix, that helps you figure stuff out
[20:23:32] <KGB-linuxcnc> 03Dewey Garrett 05master 964c0e4 06linuxcnc 10tests/twopass/b.hal 10tests/twopass/expected tests/twopass use script option for halcmd show * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=964c0e4
[20:24:57] <linuxcnc-build> build #4564 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4564 blamelist: Jeff Epler <jepler@unpythonic.net>
[20:25:03] <Roguish> jepler: thanks. most of that makes sense. I will look into as mush as I can. thanks again.