#emc-devel | Logs for 2009-04-27

[00:00:16] <jepler> I'm using the thing that is created by kthread_create. I don't know which one of those names is the right one..
[00:00:43] <jepler> or whether it's the one that's appropriate for this kind of thing (basically, the same as an rtapi thread)
[00:01:57] <SWPLinux> hmmm. dunno. I haven't looked at that stuff in years
[00:02:08] <SWPLinux> not for real anyway
[00:02:59] <jepler> I've never looked before yesterday
[00:03:12] <SWPLinux> ok, then we're about even :)
[00:03:26] <jepler> I should figure out how to run this in a virtual machine instead of on my real machine
[00:03:33] <jepler> since it's going to crash hard everytime for awhile
[00:03:42] <SWPLinux> the last time I delved that far into kernel workings was when I made my own pseudo-RT module that sat under the main scheduler thread
[00:03:51] <SWPLinux> heh
[00:08:43] <jepler> hm, interesting discovery of the day: 'ifconfig eth00' will show information for eth0
[00:09:02] <SWPLinux> that sounds buggish
[00:09:13] <jepler> 'ifconfig eth0000' will too, but not 'eth01'
[00:09:24] <jepler> actually, 'eth01' shows eth1
[00:09:30] <SWPLinux> ok, that might make sense
[00:09:34] <jepler> it must be a feature
[00:09:42] <jepler> for machines with more than 10 network interfaces
[06:18:50] <KimK> Webmaster: Can http://www.cnc-workshop.com/ be updated for the 2009 CNC Workshop?
[07:29:59] <alex_joni> KimK: there is no 2009 CNC Workshop
[07:30:09] <alex_joni> and it's not our (EMC) webpage
[09:21:17] <KimK> alex_joni: Not our webpage, OK. Do you mean nothing in Galesburg this year (I know) or do you mean Wichita in May is canceled?
[09:42:00] <alex_joni> wichita is on afaik
[09:43:05] <alex_joni> details have been posted on the mailing list
[10:07:13] <KimK> Anything on the wiki for those who still haven't joined the mailing list?
[10:10:18] <alex_joni> KimK: I just looked, not much ;)
[10:12:41] <KimK> I just looked too, I only found this http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?EMC_Fest_2009 but despite its title, it's mostly an empty framework.
[10:21:12] <KimK> Thanks, have to go now, I'll be back later
[12:18:51] <jepler> I got my kernel thread thing going
[12:19:37] <jepler> as expected, it gives decent timing most of the time, but crappy timing multiple times in a half second
[12:21:10] <jepler> successive invocation times in CPU cycles, nominal = 1ms = 2201000 CPU cycles
[12:21:13] <jepler> http://pastebin.ca/1404787
[12:22:01] <jepler> now to try with the ubuntu "-rt" kernel
[12:37:14] <jepler> .. still not good on my laptop, 4ms latencies sometimes
[13:10:14] <CIA-1> EMC: 03jepler 07TRUNK * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: show the user the active units (in, mm, degree-symbol) in touch off dialog
[13:41:02] <jepler> I haven't set the scheduler priority of my thread, maybe that would make a difference
[13:46:06] <cradek> jepler: could you also add radius or diameter? (like "mm diameter")
[13:52:38] <jepler> cradek: hmm
[13:53:30] <jepler> cradek: for lathe machines, add "diameter" or "radius" to the X dimension string according to g7/g8?
[13:54:03] <cradek> yes that's it I think
[14:13:29] <CIA-1> EMC: 03jepler 07TRUNK * 10emc2/src/emc/usr_intf/axis/scripts/axis.py: in lathe mode touch off, show diameter or radius if appropriate
[14:17:39] <cradek> whee!
[14:19:46] <jepler> mind giving it a thorough testing?
[14:19:53] <jepler> I'm tempted to put it in 2.3.
[14:20:32] <cradek> ok, I'll go test it
[14:24:58] <CIA-1> EMC: 03cradek 07TRUNK * 10emc2/bin/.cvsignore: quiet
[14:25:09] <CIA-1> EMC: 03cradek 07TRUNK * 10emc2/docs/man/man9/.cvsignore: quiet
[14:42:58] <cradek> jepler: come see a thing, please
[14:52:16] <cradek> it works correctly but there's a cosmetic problem
[15:26:01] <skunkworks> is it raining hard by jepler? ;)
[15:26:24] <cradek> it's been wet the last few days.
[15:27:01] <skunkworks> it stormed really hard here sat night. The thunder was shaking the house.
[15:28:51] <skunkworks> less than ideal building weather - saturday was excellent. (until it stormed around 6:00pm)
[17:59:59] <cradek> seb_kuzminsky: http://www.usdigital.com/products/interfaces/pc/internal-card/pci-4e/
[18:01:42] <cradek> (holy crap that's expensive for what it does)
[18:22:09] <skunkworks> that and a printer port would do a ton. (but the mesa card is cheaper)
[18:32:19] <cradek> cvs should be back in a few hours. hardware failure.
[19:21:06] <seb_kuzminsky> cradek: yeah that doesnt seem like a very good deal
[19:22:01] <seb_kuzminsky> mah
[19:30:40] <cradek> hey look who's back.
[19:31:02] <jepler> I think cvs will be back in about 2 more minutes
[19:32:11] <jepler> .. cvs service should be restored
[19:32:40] <jepler> cradek: hm, but can you make sure and kick the clock at least once? the host booted with the time off by hours
[19:33:00] <jepler> and I didn't correct it until after booting the guest
[19:33:40] <cradek> it sets (steps) every 5 minutes because we never solved the drift
[19:33:53] <jepler> ok
[19:34:43] <CIA-1> EMC: 03cradek 07TRUNK * 10emc2/docs/man/man9/.cvsignore: quiet
[20:18:53] <jepler> * jepler kicks irssi
[20:22:00] <alex_joni> 22:54 -!- jepler [n=jepler@emc/developer/jepler] has quit [Read error: 111
[20:22:00] <alex_joni> (Connection refused)]
[20:22:04] <alex_joni> that's an odd error
[20:23:30] <jepler> ooh, when I sched_setscheduler, I get much better results: http://emergent.unpy.net/index.cgi-files/sandbox/ubuntu-rt-timings.png -- time between invocations, nominal=1.0, scale is big enough to show all 510 samples I took.
[20:23:50] <jepler> (this is a hal-like 'realtime' thread over the ubuntu -rt kernel)
[20:23:54] <jepler> (1ms)
[20:24:48] <cradek> hey cool.
[20:24:50] <alex_joni> cool
[20:24:59] <alex_joni> that should definately be good enough for servo-only
[20:25:03] <jepler> 510 samples isn't very many though
[20:25:06] <jepler> who knows what glitches lurk
[20:32:34] <alex_joni> hmm.. they disabled ctrl-alt-backspace by default in 9.04 :/
[20:33:59] <jepler> I'm sure it's a feature
[20:34:40] <alex_joni> yeah, probably
[21:55:15] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/img_4521-medium.jpg -- when your computer starts doing this, you'll be getting a new one soon
[21:55:19] <jepler> :-P
[21:55:22] <jepler> :-/
[21:55:23] <jepler> :-(
[21:57:54] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/IMG_4523-crop.JPG -- note good capacitor (background) and failed capacitors (foreground). It's not clear right away, but the difference in reflection is because the front caps have very domed tops
[21:58:57] <jepler> (also note the rather nice bokeh of the lens .. surprising for the cheapo bundled lens with a low-end slr)
[23:47:46] <jepler> hi skunkworks