#emc | Logs for 2006-02-07

[00:34:29] <CIA-8> 03cradek * 10emc2/debian/emc2.files: I didn't mean to check these in
[01:40:30] <lerman_> lerman_ is now known as lerman
[01:56:04] <robin_z> meep?
[01:56:11] <cradek> hi robin
[01:56:22] <robin_z> hi ;)
[01:56:31] <cradek> what happened to your s?
[01:56:52] <cradek> I guess you lost it somewhere
[01:57:01] <robin_z> my other PC is using it :)
[01:57:17] <cradek> I suppose there are only so many s in the world
[02:02:19] <robin_z> yeah, I was lucky to get one
[02:15:00] <robin_z> robin_z is now known as robin_sz
[02:35:34] <LawrenceG> anybody running the new debs (emc2, axis) under ubuntu?
[02:38:58] <jepler> LawrenceG: I have run them a little bit
[02:39:47] <robin_sz> ubuntu? eek.
[02:40:01] <robin_sz> I trod in some ubuntu once.
[02:40:12] <LawrenceG> I am having trouble getting the rt modules to install
[02:40:38] <jepler> are there errors in 'dmesg'? (type 'dmesg' at a terminal window and look at the last lines)
[02:40:51] <LawrenceG> on adeos.ko, get a files exists error
[02:41:08] <LawrenceG> and a bunch of operation not permitted on rtai_*
[02:41:26] <LawrenceG> this is in a terminal window when I run emc from the terminal
[02:42:10] <jepler> look at 'dmesg', sometimes it has additional information
[02:42:38] <jepler> from what you've told me so far, it sounds like the symptom that when emc fails to start once, it can leave some modules around, and you get this message the next time .. instead of whatever the underlying problem was.
[02:42:52] <LawrenceG> dmesg shows a whole pile of unknown symbol errors on all the rtai modules
[02:42:54] <jepler> umm, I don't know if what I just said made any sense
[02:43:22] <jepler> scroll back to the earliest errors (or use dmesg | less)
[02:44:08] <jepler> one thing I ran into on my laptop was that the very first module would complain about my 'APIC', which was disabled. Do you have a message like that?
[02:44:41] <LawrenceG> * LawrenceG looking
[02:46:36] <jepler> this is the message I got:
[02:46:37] <jepler> ./base/arch/i386/hal/hal.piped:printk("RTAI[hal]: ERROR, LOCAL APIC CONFIGURED BUT NOT AVAILABLE/ENABLED\n");
[02:47:02] <LawrenceG> hmmm.. first error in dmesg seems to be "RTAI[hal] .... yep thats the one I see
[02:48:19] <jepler> OK. The next thing to try is to boot the kernel with the "lapic" flag. Are you familiar with setting kernel flags in grub?
[02:51:38] <LawrenceG> not recently, but I can probably figure it out..... is there a special line you can paste into menu.lst
[02:51:43] <jepler> http://pastebin.com/542682
[02:52:09] <jepler> basically, copy the group of lines for the -magma kernel to above the AUTOMAGIC line, and then add 'lapic' after 'ro'
[02:53:04] <jepler> what hardware are you using (CPU)?
[02:54:06] <jepler> http://www.captain.at/programming/rtai/apic.php
[02:54:59] <LawrenceG> its an old 200mhz P?? 196mb
[02:55:12] <cradek> looks like this error would go away if I compiled the kernel without CONFIG_X86_LOCAL_APIC but I don't know what problems that might cause
[02:55:14] <LawrenceG> suspect there will be an issue with period as well
[02:55:42] <LawrenceG> rebooting now
[02:55:49] <jepler> original pentiums might not have a local APIC
[02:56:44] <jepler> bah, ignore me, I don't know why I said that
[02:58:06] <LawrenceG> I get spoiled running 2+ghz devel machine... its takes patience on the sub 500mhz boxes, but they still work!
[02:59:18] <cradek> LawrenceG: did you just install this ubuntu?
[03:00:09] <LawrenceG> I was using it as a test box for working with the surplus 3ph ac servo driver
[03:00:22] <LawrenceG> so its been around for a few months
[03:00:45] <cradek> LawrenceG: are you using the packages from http://solaris.cs.utt.ro/emc2/ ?
[03:01:06] <LawrenceG> yes... kernel, modules, emc2, axis
[03:01:10] <cradek> gotcha
[03:01:30] <jepler> cradek: hm, in my linux-2.6.git tree (indeterminate version) there's a flag CONFIG_X86_GOOD_APIC which seems to force use of the local APIC .. with a scary comment
[03:01:33] <cradek> I didn't know you were doing that; I'm still a bit in the development phase with these
[03:01:46] <cradek> jepler: interesting
[03:03:25] <cradek> LawrenceG: I encourage you to use them, but beware they're still changing
[03:03:52] <LawrenceG> same problem.... "no local apic present or hradware disabled"
[03:04:11] <cradek> have a look at 'cat /proc/cmdline'
[03:04:15] <LawrenceG> showed up higher up in the desg stuff.... can play with bios settings
[03:04:31] <cradek> make sure it says lapic there
[03:05:02] <LawrenceG> says root=/dev/hda3 ro lapic
[03:05:23] <cradek> hmm, yeah maybe you need to look at the system bios
[03:05:52] <LawrenceG> I'm pretty sure all the power save stuff is shut off....
[03:06:53] <LawrenceG> no problem thats its under devel... I am very happy to see that.... it will make installs very easy.
[03:07:22] <LawrenceG> I wanted to see if I could help test it
[03:07:36] <cradek> that's great
[03:07:41] <cradek> I appreciate it
[03:07:50] <cradek> maybe there's something about APIC in your bios
[03:07:59] <cradek> note ACPI (power management) does not equal APIC
[03:10:09] <jepler> The other approach would be to rebuid the kernel and rtai without X86_FEATURE_APIC, but that would take a fortnight on your machine
[03:10:28] <LawrenceG> rats... monitor is not syncing to bios display...
[03:11:31] <LawrenceG> got it.... had to power up with monitor disconnected....
[03:14:09] <jepler> """What is APIC? APIC is a distributed set of devices that make up an interrupt controller. In current implementations, each part of a system is connected by an APIC bus. One part of the system, the "local APIC," delivers interrupts to a specific processor; for example, a machine with three processors must have three local APICs. Intel has been building local APICs into their processors since the Pentium P54c in 1994. Computers built with genuine
[03:14:18] <jepler> -- http://www.microsoft.com/whdc/system/sysperf/IO-APIC.mspx
[03:14:45] <cradek> nice find
[03:15:04] <cradek> I think I remember some of those very old boards had bios settings about interrupt routing
[03:15:30] <jepler> looks like P54c covers 75MHz and up Pentiums
[03:15:33] <jepler> http://www.x86.org/intel.doc/586manuals.htm
[03:15:45] <cradek> I have a couple P90 around I think
[03:15:51] <jepler> but some of the comments in the linux source code seem to imply that it was only usable since Pentium MMX
[03:16:10] <cradek> MMX were the P166?
[03:16:14] <jepler> * If we were told we had a good local APIC, check for buggy Pentia,
[03:16:15] <jepler> * i.e. all B steppings and the C2 stepping of P54C when using their
[03:16:15] <jepler> * integrated APIC (see 11AP erratum in "Pentium Processor
[03:16:15] <jepler> * Specification Update").
[03:16:21] <jepler> -- asm-i386/bugs.h
[03:16:28] <LawrenceG> is linux considered to be a pnp aware o/s?
[03:16:30] <cradek> yikes
[03:16:49] <jepler> cradek: I don't know about MMX and speeds. This is the list I was reading: http://www.x86.org/intel.doc/586manuals.htm
[03:16:58] <jepler> bbl
[03:17:06] <cradek> LawrenceG: I think so
[03:17:34] <cradek> LawrenceG: anything about interrupts or interrupt routing in there?
[03:18:00] <LawrenceG> will try turning it on.... that should leave the pci resource allocations to theos
[03:18:03] <cradek> LawrenceG: if your pentium is one of those blacklisted, you might be somewhat out of luck
[03:19:38] <LawrenceG> bios is from 1995... it does have the table of irq/dma allocations
[03:21:35] <LawrenceG> main proc is a p54c (on bios boot info splash)
[03:22:06] <cradek> when it boots, we'll check the stepping and see if it's blacklisted
[03:22:40] <cradek> we could also try booting with "noapic"
[03:22:56] <LawrenceG> booting now
[03:23:37] <cradek> I've got a Pentium PRO 200 board/processor here you could have if you were nearby
[03:26:21] <LawrenceG> np.... I have a dual pp200 in the junk somewhere... not sure if that would help or hurt emc
[03:26:57] <cradek> you'd also have to rebuild to get smp, and I don't know what rtai thinks of that
[03:27:11] <cradek> but you could easily run it on one processor
[03:29:46] <LawrenceG> trying boot with noapic.....
[03:30:02] <cradek> did you check your processor stepping (cat /proc/cpuinfo)
[03:30:16] <cradek> would like to know if it's one listed in bugs.h
[03:30:26] <LawrenceG> will in a moment
[03:31:00] <LawrenceG> itel pentium stepping 0c flashed by on booting
[03:36:42] <LawrenceG> ok.... intel, family 5, model 2, pentium 75-200, stepping 12, 199.453mhz
[03:37:30] <cradek> hmm, I wonder if stepping 12 is B or C2
[03:41:30] <cradek> LawrenceG: does noapic have any effect?
[03:41:51] <jepler> http://pastebin.com/542718
[03:41:58] <LawrenceG> still failed..... looking through dmesg now
[03:42:18] <jepler> it looks like 5 / 2 / 12 doesn't trigger the test
[03:43:09] <LawrenceG> no panic anyway!
[03:43:12] <jepler> cradek: are you sure that the person who used 'noapic' was actually able to run emc, or just to boot with the realtime kernel? The way I read the source to rtai3.3test4 it won't load the realtime modules without an APIC
[03:43:28] <cradek> jepler: I'm 100% sure
[03:43:46] <jepler> LawrenceG: well, that code is actually #if'd out, because CONFIG_X86_GOOD_APIC isn't defined in cradek's kernel
[03:43:47] <cradek> jepler: it surprised me too
[03:44:23] <cradek> maybe I should rebuild with 586MMX and GOOD_APIC
[03:45:24] <cradek> althought that still wouldn't have helped jmk's machine which needed 'noapic'
[03:45:31] <jepler> I'm pretty sure LawrenceG's machine isn't MMX
[03:45:35] <jepler> so it won't help him either
[03:45:52] <cradek> hmm.
[03:46:14] <cradek> this is a mess and I don't remember it ever being a problem with 2.4 kernels.
[03:46:45] <cradek> It would be nice if there was a combination of settings that would give me that old behavior, whatever it was
[03:46:58] <cradek> of course maybe I've just been lucky for a long time
[03:48:06] <jepler> rebuild with CONFIG_X86_LOCAL_APIC off?
[03:48:07] <LawrenceG> dmesg still gives RTAI[hal]: ERROR local apic configured but not available/enabled
[03:53:31] <jepler> goodnight folks
[03:53:42] <Jymmm> G'Night jepler
[03:54:09] <LawrenceG> thanks for helping
[03:54:33] <cradek> LawrenceG: I'm going to build a new kernel package that should fix this.
[03:54:41] <cradek> LawrenceG: it will take a little while.
[03:54:59] <LawrenceG> I can test it whenever you say
[03:55:20] <cradek> ok I will let you know
[03:55:25] <cradek> it will probably be done in an hour or less
[03:55:30] <LawrenceG> not nearly as long as building on this box!
[03:55:46] <cradek> yeah, days faster probably
[03:56:01] <jepler> I was joking when I said "fortnight", but probably not by more than a factor of 10...
[03:56:24] <jepler> ugh, another 190 megs of updates on my fedora machine
[03:56:30] <LawrenceG> I started palying with emc in about 1997 and it took 2 weeks to get my first running system so you are right
[03:56:32] <cradek> it would probably take no more than a day or two.
[03:56:49] <cradek> jepler: holy crap
[03:57:41] <jepler> looks like they upgraded kde from 3.5.0 to 3.5.1
[03:57:46] <jepler> == 150 megs
[03:57:59] <cradek> LawrenceG: does that machine have working network so you can use apt-get upgrade?
[03:58:09] <cradek> jepler: do you use kde?
[03:58:13] <jepler> cradek: nope!
[03:58:20] <cradek> jepler: that's pretty funny then
[03:58:22] <A-L-P-H-A> when stuff updates, does the old stuff get purged?
[03:58:46] <LawrenceG> cradek: yes... all setup with the .ro repository
[03:58:51] <cradek> LawrenceG: cool
[03:58:55] <cradek> A-L-P-H-A: "depends"
[03:58:58] <jepler> If I'd known then what I know now, I might have installed ubuntu on this machine
[03:59:21] <cradek> jepler: of course it's never too late
[03:59:26] <A-L-P-H-A> cradek, oh
[03:59:39] <cradek> A-L-P-H-A: what specifically do you mean? I can give a better answer.
[04:00:01] <A-L-P-H-A> like, if I do an apt-get update kde, will it clean everything up nicely?
[04:00:37] <cradek> apt-get upgrade replaces old packages with the new. It does it carefully and nicely.
[04:01:28] <cradek> for example if you have a config file that you changed, but the package updates it, it will show you the differences and ask which you want to keep.
[04:01:41] <A-L-P-H-A> cool
[04:02:14] <cradek> typically though, software can be updated without asking any questions like that.
[04:02:16] <fenn> unless you're running the apt-rpm patch, in which case is gleefully ransacks your system, leaving a mess that you cant even fix with a rescue disk
[04:02:30] <A-L-P-H-A> :(
[04:02:32] <cradek> what is apt-rpm?
[04:02:42] <fenn> so dont ever try to upgrade glibc with rpm
[04:02:49] <fenn> apt-get for rpm based distro's
[04:02:55] <fenn> its an ugly hack really
[04:03:01] <cradek> who would use that? that's crazy talk
[04:03:12] <A-L-P-H-A> <shrug>
[04:03:16] <A-L-P-H-A> <--- newb remember?
[04:03:27] <A-L-P-H-A> I should really turn on my CNC machine... I haven't in months.
[04:03:37] <A-L-P-H-A> stupid machine, sitting there doing dink.
[04:03:43] <fenn> stupid humans
[04:04:17] <fenn> kinkos should have cnc mills
[04:04:23] <cradek> haha
[04:04:37] <fenn> and 3d scanners
[04:05:11] <cradek> that would certainly be branching out into different markets.
[04:05:22] <fenn> they could afford it
[04:05:24] <A-L-P-H-A> yummm... 3d scanners.
[04:05:29] <A-L-P-H-A> I would so love one...
[04:05:40] <A-L-P-H-A> take apart my cellphone, and make a cover for it.
[04:08:23] <A-L-P-H-A> how cool would alu cell phone cover be?!?! :)
[04:08:30] <A-L-P-H-A> cover = case
[04:08:52] <cradek> I think my previous phone was Ti - it was very light and wore well
[04:09:02] <fenn> i think titanium is the "how cool would it be" material these days
[04:09:06] <A-L-P-H-A> :(
[04:09:13] <A-L-P-H-A> I'm too poor to make an ti case.
[04:09:17] <Jymmm> I'd like to make a wood shell for my cellphone, but too darn thin =)
[04:09:19] <Jymmm> =(
[04:09:20] <cradek> I hear Ti is cheap now
[04:09:22] <A-L-P-H-A> and besides, it would involve forming.
[04:09:25] <A-L-P-H-A> really?
[04:09:28] <fenn> not that cheap
[04:09:37] <cradek> yeah mine was thin and stamped, not milled
[04:09:44] <fenn> carbon fiber/kevlar would be more suitable for custom work
[04:10:04] <A-L-P-H-A> yeah... stamped works for large numbers.
[04:10:06] <A-L-P-H-A> not for one offs.
[04:10:21] <A-L-P-H-A> hmm... not sude how I would like a CF weaved case.
[04:10:23] <fenn> hard-coat anodized aluminum would wear well
[04:10:46] <A-L-P-H-A> I got a hard coater around here.
[04:10:48] <fenn> you'd have to have aluminum buttons to match
[04:11:37] <cradek> I think if you're going to make art, make art - not a cover for a piece of junk throwaway consumer item
[04:12:13] <A-L-P-H-A> all cell phones are.
[04:12:18] <fenn> and if you're going to make art, don't weld a bunch of rusty crap together in my shop and leave it laying around on the floor
[04:12:21] <A-L-P-H-A> are throw away.
[04:12:28] <A-L-P-H-A> try keeping a phone longer than 3 years
[04:12:32] <cradek> yeah I know, that's actually my point
[04:12:45] <A-L-P-H-A> hahaha
[04:12:50] <A-L-P-H-A> who played in your shop fenn?
[04:13:01] <fenn> * fenn sighs
[04:13:05] <cradek> it wasn't me
[04:13:08] <A-L-P-H-A> hehe
[04:13:24] <cradek> I have welded rusty crap together, but called it a repair job, not art
[04:13:25] <A-L-P-H-A> I'm going to machine some spikes for my snowboard (stomp pad)
[04:13:38] <A-L-P-H-A> tomorrow.
[04:17:50] <A-L-P-H-A> something like these. http://cgi.ebay.com/Snowboard-Stomp-pad-Dakine-stomp-pad-Studs-New_W0QQitemZ7216768726QQcategoryZ62170QQrdZ1QQcmdZViewItem
[04:19:17] <fenn> looks like a lathe job
[04:25:50] <cradek> how do those 6 things attach to whatever?
[04:26:33] <cradek> oh I see it says adhesive
[04:50:27] <LawrenceG> cradek: how is that kernel build going?
[04:51:08] <cradek> finished, working on updating the repository now
[04:51:44] <LawrenceG> cool.... do you have scripts set up for building the debs? or is it in the make files
[04:51:59] <cradek> it's an unholy combination of things
[04:52:26] <LawrenceG> I believe it....
[04:54:28] <cradek> ok, uploading everything now. maybe 10 mins.
[04:55:10] <LawrenceG> how close is the offical release?
[04:55:37] <cradek> I was shooting for the end of last year
[04:55:47] <cradek> but that doesn't answer your question...
[04:56:05] <cradek> mostly it's in the hands of ray and jmk, they are working on last-minute stuff.
[04:56:50] <LawrenceG> well, I have been using emc2 since about the middle of last year.... the new config stuff looks good... I'd like to play with cl a bit.
[04:57:02] <cradek> aside from the configurating tools, I think emc2 is ready
[04:57:30] <cradek> it does MANY things better than emc1 already.
[04:58:17] <LawrenceG> yes, I havent used the subroutines and conditional stuff yet.... But I have made 1/2 a dozen circuit boards
[04:58:41] <LawrenceG> some where prettier than others
[04:58:52] <cradek> that's the way it goes I'm sure
[04:59:42] <LawrenceG> getting repeatable depth control is hard on the shoptask because of the huge rack and pinion feed
[04:59:53] <LawrenceG> on the quill...
[05:00:08] <LawrenceG> would like to add a small high speed spindle and z axis
[05:10:45] <cradek> still sending.....
[05:11:21] <cradek> on my PCBs I cut .0045 deep iirc
[05:11:33] <cradek> it's pretty critical since it's a V tool
[05:12:30] <LawrenceG> yea... I did one board and it was perfect.... peeled the copper off cleanly without disturbing the fiberglass
[05:12:46] <cradek> peeled?
[05:13:04] <LawrenceG> too deep and the fiberglass messes things up
[05:13:04] <cradek> I leave the extra copper on the board - it doesn't hurt anything
[05:13:18] <cradek> sometimes I use it as the ground plane
[05:13:25] <LawrenceG> well... cut the copper isolation tracks cleanly
[05:13:38] <LawrenceG> yea I leave all the extra on as well
[05:15:07] <LawrenceG> a buddy of mine just built a small router table and cut a double sidded board... very nice... it is made from suplus printer steppers and 1/4 20 all thread with bronnze bearings running on ground rod ways
[05:15:35] <cradek> what are the leadnuts?
[05:15:50] <LawrenceG> he made anti backlash nuts from brass plumbing fittings threaded on the inside
[05:16:06] <cradek> a spring loaded deal?
[05:16:08] <LawrenceG> about a 10x10x3" work zone
[05:16:33] <cradek> sounds neat
[05:16:38] <cradek> he should post some photos
[05:16:41] <cradek> run by emc?
[05:16:50] <LawrenceG> I got him to take lots of pictures so he could do a web page.... he is running emc2/axis
[05:17:00] <cradek> cool
[05:17:36] <cradek> what did he use for stepper drivers?
[05:17:50] <LawrenceG> he even made a spindle from an old router shaft/collet driven by a treadmill motor
[05:18:11] <cradek> cool, that's closed-loop speed control then
[05:18:31] <LawrenceG> some 3955 boards ge got from pmi I think
[05:18:32] <cradek> I bet he had to gear it up a lot...
[05:19:17] <LawrenceG> treadmill motor runs at 6000rpm@90vdc... I think his speed up is between 2 and 3
[05:19:45] <cradek> 12-18k is a good range for small stuff
[05:20:02] <LawrenceG> only problem so far is the treadmill motor gets very toasty.... needs fan cooling
[05:20:21] <cradek> yeah they usually have fans on the back shaft
[05:21:27] <LawrenceG> 67mb of updates.... downloading now
[05:22:11] <cradek> no, the upload isn't done!
[05:22:11] <LawrenceG> opps
[05:22:21] <cradek> I wish there was a way I could prevent that...
[05:22:27] <LawrenceG> control c...
[05:22:34] <cradek> I bet apt wouldn't install them because the gpg sig doesn't match.
[22:16:21] <cradek> hmm, sf cvs is still down.
[22:20:42] <jepler> if you clone my git repository of emc2 (git://axis.unpy.net/emc.git) you can at least perform actions like git-diff offline
[23:04:06] <alex_joni> night all
[23:31:11] <lerman_> lerman_ is now known as lerman