#emc-devel | Logs for 2006-12-23

[00:00:01] <jmkasunich> that doesn't count the DC bus supply for the spindle, probalby at least a 6" cube
[00:00:10] <jmkasunich> or the DB resistors and such, another 6" cube
[00:18:54] <skunkworks> it is never big enough
[00:19:14] <jmkasunich> I'm trying to figure out if it will be cool enough
[00:19:41] <skunkworks> did you decide on a motor mount? you didn't want it to stick out too far (x)
[00:19:50] <jmkasunich> the spindle drive has a 9x12" (approx) heatsink that I was hoping to install with the fins outside and the sparky bits on the inside
[00:20:10] <jmkasunich> gonna put the motor in the standard location
[00:20:32] <jmkasunich> once I have the electrical box there, it isn't so bad - the chip guard can be installed from the top of the box
[00:21:53] <jmkasunich> upon closer inspection of the spindle drive, it would be very hard if not impossible to have the fins stick out while keeping a seal
[00:22:07] <jmkasunich> the electronics go too close to the edge of the heatsink, no place to seal it
[00:26:49] <skunkworks> your having too much fun. I am jelous
[00:27:46] <jmkasunich> I don't think its fun right now
[00:27:55] <jmkasunich> too many details
[00:28:57] <skunkworks> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?RtaiSteps
[00:29:15] <jmkasunich> _that_ is too many details
[00:29:18] <skunkworks> is that where alex would have put the file if he had gotten to it?
[00:29:19] <jmkasunich> are you doing that?
[00:29:30] <skunkworks> no
[00:29:31] <jmkasunich> which file, the SMI thing?
[00:29:33] <skunkworks> yes
[00:30:07] <jmkasunich> the other box has been doing memtest for 2-1/2 hours with no probs, lemme reboot it and get the file from there
[00:31:14] <skunkworks> thank you
[00:37:47] <jmkasunich> on the way
[00:37:56] <jmkasunich> and so am I (to meet my wife for dinner)
[00:37:59] <jmkasunich> later...
[00:39:52] <skunkworks> thanks john
[11:20:19] <alex_joni> logger_dev: bookmark
[11:20:19] <alex_joni> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2006-12-23.txt
[14:03:54] <skunkworks> So I sudo insmod rtai_smi.ko and ran the latency test all night. No overruns. ovl max was 26994.
[14:05:09] <skunkworks> I had to disable the screen saver - the second the screen would start to fade out - the overuns would increase by hundreds
[14:07:58] <skunkworks> now insmod was from my desktop. I don;t know where to put it. when I search for it I find it in /usr/realtime-2.6... but from what I read yesterday - it should be in the /lib/module
[14:09:08] <skunkworks> I did copy over the module in lib/modules - but after a reboot I was still getting latency errors. That is when I did the insmod - then no overruns. (no reboot yet)
[14:14:38] <skunkworks> seems to hover around 12000
[14:14:42] <skunkworks> most of the time
[14:17:26] <skunkworks> pretty cool - great work.
[14:19:21] <alex_joni> skunkworks: nice
[14:19:28] <alex_joni> I think I was wrong about /lib/modules
[14:19:51] <alex_joni> I would put it in the realtime script
[14:20:08] <skunkworks> ok....
[14:20:09] <alex_joni> skunkworks: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting#Some_Intel_boards_have_issues_with_the_SMI_System_Management_Interrupt
[14:20:26] <alex_joni> can you upload the binary rtai_smi.ko ?
[14:20:43] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?action=upload
[14:21:14] <skunkworks> oh - cool - nice work
[14:21:55] <skunkworks> I did copy it into the /usr/realtime-2.6.15-magma/modules/rtai_smi.ko
[14:24:38] <skunkworks> but I thought after a reboot I still had overruns. It wasn't until I did a insmod did It stop.
[14:24:51] <skunkworks> (have not rebooted yet)
[14:25:34] <skunkworks> since the insmod
[14:29:00] <skunkworks> I am actually surfing and everything with now overruns
[14:29:24] <skunkworks> no overruns
[14:33:41] <skunkworks> going to reboot and see
[14:51:40] <skunkworks> Ok the reboot I get overruns. until I do a insmod from the desktop
[14:52:09] <skunkworks> So I am guessing I have not coppied it correctily
[14:53:21] <alex_joni> it's not a matter of copying
[14:53:26] <alex_joni> you need to install it too
[14:53:47] <alex_joni> there are loads of modules available, but you only install the ones you need (think network drivers))
[14:54:09] <skunkworks> ok - makes sense
[14:54:14] <alex_joni> the fact that the rtai_smi.ko module exists doesn't mean it gets used.. you need to tell it that it should get loaded
[14:54:24] <alex_joni> there are various ways to do that
[14:54:57] <alex_joni> 1. put it in some custom script, so that it gets insmoded at load time (e.g. make a small script in /etc/init.d/ and insmod it from there)
[14:55:15] <alex_joni> 2. add it to /etc/init.d/realtime and have it loaded along with the other scripts
[15:16:41] <skunkworks> sorry - was away.
[15:16:45] <skunkworks> Thanks alex
[15:18:14] <jepler> if you're going to load it with emc_module_helper, then you'll have to add it to the module whitelist
[15:19:51] <alex_joni> yeah, so #2 is nicer, but requires recompile
[15:20:14] <skunkworks> way over my head. I see the realtime file is scritp
[15:21:12] <alex_joni> skunkworks: do you run in place?
[15:21:16] <alex_joni> or installed?
[15:21:35] <skunkworks> I just got head last night - and that is normmaly what I play with.
[15:21:48] <skunkworks> run in place
[15:21:51] <alex_joni> ok, that might be easy to change.. I'll add rtai_smi to the whitelist
[15:23:30] <alex_joni> skunkworks: for now add it manually :)
[15:23:43] <skunkworks> no problem - I can handle that :)
[15:24:17] <skunkworks> no overuns this mornings
[15:25:10] <alex_joni> good
[15:25:22] <alex_joni> you might want to check plugging in an USB stick
[15:25:32] <alex_joni> the two might be related
[15:25:32] <skunkworks> eww
[15:25:42] <alex_joni> SMI is the one that handles some USB in some cases I think
[15:26:48] <skunkworks> looks like it isn't an issue on this machine.
[15:26:50] <skunkworks> cool
[15:29:29] <skunkworks> I should be able to play with the pluto on this portable
[22:39:56] <alex_joni> cradek: mind if I add rtai_smi to the module_helper whitelist?
[22:40:06] <alex_joni> actually I think it should hold all rtai modules
[22:41:36] <cradek> sure, add that one, but I don't think it should hold anything not used with emc
[22:47:56] <alex_joni> http://rtai.dk/cgi-bin/gratiswiki.pl?Latency_Killer
[22:48:01] <alex_joni> this looks interesting enough
[22:53:53] <cradek> that looks like a nice page
[22:54:33] <alex_joni> yeah, I linked it from our Troubleshooting page
[22:54:52] <alex_joni> http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?TroubleShooting
[22:56:52] <alex_joni> cradek: the rtai_smi module fails if it doesn't find a chipset it knows
[22:57:05] <alex_joni> so we can't just blindly add it by default
[22:57:21] <alex_joni> ../bin/emc_module_helper insert /usr/realtime-2.6.15-magma/modules/rtai_smi.ko
[22:57:22] <cradek> but we could compile and distribute it.
[22:57:24] <alex_joni> insmod: error inserting '/usr/realtime-2.6.15-magma/modules/rtai_smi.ko': -1 No such device
[22:57:31] <alex_joni> it is already compiled and distributed
[22:57:38] <alex_joni> but the RTAI people did something strange
[22:57:44] <cradek> oh
[22:58:02] <alex_joni> http://cvs.gna.org/cvsweb/magma/base/arch/i386/calibration/smi-module.c?rev=1.7;content-type=text%2Fplain;cvsroot=rtai
[22:58:11] <alex_joni> that's the version that's beeing compiled and distributed
[22:58:27] <cradek> oh I see it's all disabled
[22:58:34] <alex_joni> which doesn't disable anything
[22:58:35] <alex_joni> right
[22:58:51] <alex_joni> it worked for JMK to turn on the ALL
[22:58:56] <cradek> we (you!) could easily do a new package.
[22:58:56] <alex_joni> same for skunkworks
[22:59:10] <alex_joni> _but_ I've seen at least 1-2 cases where it didn't work
[22:59:17] <alex_joni> yet it worked with turning all others on
[22:59:26] <alex_joni> some guys complaining on the rtai lists
[22:59:26] <cradek> argh.
[22:59:40] <cradek> rtai sure makes it hard to distribute anything but the source.
[22:59:55] <alex_joni> which makes me somehow weiry about distributing this
[23:00:18] <alex_joni> also.. there's a rather large FIXME comment at the end of the file
[23:00:46] <cradek> BIG FAT WARNING:
[23:00:58] <cradek> hmm
[23:01:10] <alex_joni> "this won't work in most cases, so why worry .."
[23:01:33] <alex_joni> at least that's how I read their attitude :)
[23:02:41] <alex_joni> I think the proper fix would be to have the CONFIG_* as an insmod param to the module
[23:02:50] <cradek> I agree
[23:03:26] <alex_joni> which might also break some things ..
[23:03:44] <alex_joni> although I don't expect many people to use the ubuntu packages for rtai for anything else then emc