#emc-devel | Logs for 2008-10-03

[00:00:35] <skunkworks> http://www.cnczone.com/forums/showthread.php?p=508787#post508787
[00:34:30] <cradek> I wonder if he typed rick's, and the ' causes a problem
[00:35:13] <cradek> he doesn't say what version...
[04:08:29] <CIA-40> EMC: 03cmorley 07TRUNK * 10emc2/src/hal/classicladder/drawing.c: thicken layout line in edit mode
[12:54:59] <skunkworks64> http://pastebin.ca/1217931
[12:55:19] <skunkworks64> this is fresh 64 install updated - then running the hardy script.
[12:55:31] <skunkworks64> rebooting with the realtime kernel
[12:56:31] <skunkworks64> http://pastebin.ca/1217934
[12:56:32] <cradek> insmod: error inserting '/usr/realtime-2.6.24-16-rtai/modules/emc2/rtapi.ko': -1 Invalid module format
[12:56:45] <cradek> are you sure you installed the x64 emc package?
[12:56:52] <cradek> (you'd think it wouldn't let you install the wrong one)
[12:56:56] <cradek> check sources.list
[12:57:42] <skunkworks64> I was told the hardy script was the same for the i386 and 64
[12:57:51] <skunkworks64> is that the problem?
[12:58:08] <skunkworks64> how do I check the sources.list?
[13:00:23] <SWPadnos> cat /etc/apt/sources.list
[13:01:30] <skunkworks64> catrhttp://pastebin.ca/1217938
[13:01:31] <skunkworks64> http://pastebin.ca/1217938
[13:03:40] <cradek> ok, here's where it's apparent I don't know how that works
[13:03:57] <skunkworks_> join the club. ;)
[13:04:10] <cradek> what's uname -a say?
[13:04:36] <cradek> brb
[13:04:57] <skunkworks64> Linux samco-desktop 2.6.24-16-rtai #1 SMP Sat Apr 12 14:54:33 CDT 2008 x86_64 GNU/Linux
[13:05:28] <SWPadnos> sources.list looks right to me, but I wouldn't know if it had a problem either
[13:07:58] <SWPadnos> well, we know it's a symbol versioning problem. it's just a question of why :)
[13:16:18] <skunkworks_> heh
[13:24:40] <SWPadnos> skunkworks64, what's the output of dpkg -l emc2 ?
[13:25:45] <skunkworks64> http://pastebin.ca/1217956
[13:27:35] <SWPadnos> well that's no help ;)
[13:28:34] <skunkworks_> :)
[13:48:26] <cradek> skunkworks_: modinfo /usr/realtime-2.6.24-16-rtai/modules/emc2/rtapi.ko
[13:52:54] <skunkworks64> cradek: http://pastebin.ca/1217987
[13:55:28] <cradek> hm, no smoking guns. I guess it's something subtle.
[13:55:51] <skunkworks_> user error?
[13:55:53] <skunkworks_> :)
[13:56:01] <cradek> doubtful
[14:03:51] <jepler> #
[14:03:52] <jepler> [ 117.193273] rtapi: disagrees about version of symbol struct_module
[14:04:43] <cradek> yeah
[14:04:51] <cradek> kernel and emc package don't match I guess?
[14:05:27] <jepler> skunkworks_: what's the exact version of emc2 and linux-image-2.6.24-rtai installed?
[14:05:59] <SWPadnos> is there a different incantation of dpkg to tell you that?
[14:06:37] <jepler> $ dpkg -l emc2 linux-image-2.6.24-16-rtai | cat
[14:07:21] <jepler> at one point I made a broken kernel package which was on the package servers for a short time but was supposed to be removed again.
[14:07:40] <jepler> either I'm wrong and it's still on the package servers, *or* I accidentally compiled emc 2.2.6 against the bad package
[14:07:42] <SWPadnos> hmmm. it only listed "emc2 2.2.6" using dpkg -l emc2
[14:10:00] <skunkworks64> http://pastebin.ca/1218005
[14:10:33] <jepler> OK, downgrade to emc 2.2.5: sudo apt-get install emc2=1:2.2.5
[14:12:13] <skunkworks64> that worked
[14:12:16] <skunkworks64> :)
[14:12:21] <jepler> ok, I built a bad 2.2.6 package
[14:14:10] <skunkworks64> hmm - latency is not as good. well - I should say that the servothread is 9915 and the base thread is 6948
[14:14:29] <SWPadnos> absolute crap. good thing those boards are cheap
[14:14:39] <SWPadnos> you should send them to 4 of your favorite EMC2 developers ;)
[14:14:44] <skunkworks_> a::)
[14:15:10] <SWPadnos> by the way, is that a PCIe video slot or AGP pro?
[14:15:12] <skunkworks_> does the smp kernel put realtime in one core?
[14:15:26] <SWPadnos> yes
[14:15:41] <skunkworks_> pci-e x16
[14:15:42] <jepler> it's a dual-core machine?
[14:15:51] <skunkworks_> it is a amd64
[14:16:00] <SWPadnos> should be single core
[14:16:03] <skunkworks_> right
[14:16:09] <SWPadnos> http://www.newegg.com/Product/Product.aspx?Item=N82E16813135091
[14:16:10] <skunkworks_> that was random - sorry
[14:16:31] <jepler> #
[14:16:32] <jepler> [ 13.182031] SMP alternatives: switching to UP code
[14:16:36] <jepler> yeah I spot that in the dmesg ^^
[14:17:18] <skunkworks_> what does that mean?
[14:17:33] <SWPadnos> it's compiled for SMP, but it only detected a single processor
[14:17:38] <skunkworks_> ah
[14:17:39] <skunkworks_> ok
[14:17:42] <SWPadnos> so it's using UniProcessor code
[14:18:00] <skunkworks_> I read as uP and it didn't make sense. ;)
[14:18:06] <SWPadnos> heh
[14:19:03] <skunkworks_> jepler: thanks.
[14:19:08] <SWPadnos> too bad that board doesn't have DVI or an LVDS connector
[14:19:16] <jepler> skunkworks_: thanks for reporting the problem!
[14:19:36] <jepler> I think I can do a 2.2.6-2 package today that would fix this
[14:19:43] <jepler> (just rebuilding against the right kernel)
[14:19:51] <jepler> would you have a chance to test that?
[14:20:26] <skunkworks_> it says - compleat tv-out/digital flat pannel solution: -built-in dual display controllers to support independent dual dispalys
[14:20:44] <skunkworks_> jepler: yes
[14:20:49] <skunkworks_> I am here all day
[14:20:52] <SWPadnos> hmmm. I don't see the connectors in the images
[14:23:32] <skunkworks_> yah - I don't know how they are doing that. I see nothing in the manual
[14:23:38] <skunkworks_> on how they would be hooked up
[14:23:53] <SWPadnos> I guess you could get a PCIe video card ;)
[14:24:26] <skunkworks_> yes - maybe that is just spewed out from the chipset info.
[14:24:37] <skunkworks_> not applicable
[14:25:06] <SWPadnos> I'd love to stick that in my panel PC, and still use the touch panel
[14:25:26] <SWPadnos> should be a little faster than the celeron 500 that's in there now ;)
[14:26:05] <skunkworks_> just a bit.
[14:26:20] <skunkworks_> seems to run pretty decent on the 512mb of ram it came with.
[14:27:13] <SWPadnos> hmmm. I think I have an extra 1 or 2 GB of DDR333 sitting around too
[14:27:31] <skunkworks64> http://imagebin.ca/view/qYqNnvQg.html
[14:28:20] <SWPadnos> hmmm. I don't recall the incantation, but can you try that with a 20 or 15 uS base period?
[14:28:38] <SWPadnos> maybe even 10, if you think the machine won't crash
[14:29:26] <skunkworks_> do you have any idea? No man page.
[14:29:45] <SWPadnos> look at the script. it's some option for latency-test
[14:30:43] <SWPadnos> ok, just add the time to the command line
[14:30:47] <SWPadnos> latency-test 15us
[14:31:02] <SWPadnos> you can also add another one to set the servo period
[14:46:09] <skunkworks_> cool - running at 10us
[14:49:06] <jepler> 10us period plus what latency?
[14:49:35] <skunkworks64> servo jitter is 6019 and base is 5309
[14:49:42] <skunkworks64> seems to be running ok
[14:50:00] <SWPadnos> hmmm. not much better. bummer
[14:50:22] <SWPadnos> servo is quite a bit better, but I was hoping the base jitter would have improved more
[14:50:34] <SWPadnos> you may need to get down to 7 or 8 us for that to happen though
[14:59:29] <skunkworks64> http://imagebin.ca/view/YAteyW.html
[14:59:59] <jepler> skunkworks_: ok, i think you can now: 'sudo apt-get update; sudo apt-get install emc2 '
[15:00:08] <jepler> that should install version "2.2.6-1"
[15:00:21] <SWPadnos> could you try it with 8 us? you don't need to do all the stressing if the machine runs slowly
[15:00:51] <skunkworks64> I was just runnning 7us
[15:01:06] <SWPadnos> oh. it said 10 in the screenshot
[15:01:22] <skunkworks64> yes - I ment just now ;)
[15:01:27] <SWPadnos> ah :)
[15:01:47] <skunkworks64> doesn't get better - it actually is at 7316 - so that is bad
[15:02:05] <SWPadnos> hmmm. ok - nevermind
[15:02:27] <SWPadnos> the cache may not be large enough (relative to the processor speed) to get the savings I'm looking for
[15:05:10] <skunkworks64> jepler: success
[15:13:44] <jepler> skunkworks64: great, thanks for playing
[15:22:32] <skunkworks_> No problem
[15:26:24] <skunkworks_> pulled the plug by accident
[15:29:21] <skunkworks_> who wants to field this? http://www.cnczone.com/forums/showthread.php?t=65688
[15:29:22] <skunkworks_> :)
[15:29:48] <skunkworks_> seems like a job for SWPadnos
[15:29:53] <SWPadnos> thanks
[15:30:21] <skunkworks_> no problem ;)
[15:40:37] <fenn> "RTFM"
[15:41:25] <fenn> what do people expect, really?
[15:44:20] <alex_joni> people to read to them?
[15:44:37] <alex_joni> maybe we should publish an audiobook with the integrators manual
[15:46:11] <cradek> how do I leverage EMC2 ‘s advantages while address some of the weaknesses inherit to a noncommercial GPL offerings?
[15:46:33] <cradek> smiley smiley
[15:46:35] <cradek> smiley smiley
[15:47:24] <cradek> I don't really understand the question
[15:47:35] <SWPadnos> I didn't really notice much in those strengths/weaknesses lists that had to do with EMC
[15:48:21] <cradek> guess what I'm asking, then reassure me, so I don't have to do any of the legwork myself
[15:50:13] <cradek> http://www.cnczone.com/forums/showthread.php?t=65577
[15:50:46] <cradek> this thread could use a better answer but also a better question
[15:50:56] <cradek> I don't know why he can't jog and then home
[15:51:15] <alex_joni> maybe he's not aware he can home multiple times
[15:51:23] <cradek> before homing, soft limits don't keep him from jogging
[15:51:25] <alex_joni> and has some short softlimits
[15:51:32] <alex_joni> ah.. hmm
[15:51:50] <jepler> maybe he has offsets in effect and needs to zero them
[15:52:03] <jepler> who knows
[15:52:51] <jepler> ... or do I have to use stepconf wizard to move it back to where it's supposed to be? (that's what I ended up doing to fix it the last time). ...
[15:52:56] <jepler> I wonder if stepconf always writes a fresh var file or something
[15:53:33] <jepler> (no; if I read the source right it only copies emc.var to the config if it doesn't exist already)
[15:54:06] <alex_joni> well.. you should know
[18:12:36] <skunkworks_> up to 11359
[19:10:33] <alex_joni> skunkworks_: on amd_64 ?
[19:10:39] <skunkworks_> yes
[19:11:07] <skunkworks_> I 'think' the i386 kernel had a bit less jitter
[19:12:15] <alex_joni> I would expect the 64 kernel to have better machine performance
[19:12:25] <alex_joni> and you haven't tested isolcpus yet
[19:12:54] <alex_joni> this is a 2+ cpu/core machine?
[19:15:16] <SWPadnos> single
[19:15:43] <SWPadnos> the 386 kernel saves fewer registers in a context change, so there may be some savings there
[19:16:07] <alex_joni> might be..
[19:16:19] <alex_joni> but both values are really good
[19:16:35] <alex_joni> and I don't think he'll run a base thread < 10000 anyways
[19:17:14] <SWPadnos> heh, probably not :)
[19:17:58] <SWPadnos> hmmm. come to think of it, it may not have been a latency change that jmkasunich and I saw. It was probably the time/tmax for the base thread
[19:40:29] <skunkworks_> this was yesterday on the i386 kernal http://www.electronicsam.com/images/KandT/testing/athlon64-3200ECSGoal3.png
[19:40:37] <skunkworks_> running for quite a while
[20:15:25] <SWPadnos> jepler, did you update the checksums / package signatures (or whatever) when you updated the 2.2.6 package today? (though you didn't touch 32-bit, did you?)
[20:17:14] <alex_joni> SWPadnos: getting updates for the new kernel on i386?
[20:17:29] <SWPadnos> no - see the discussion in #emc
[20:17:35] <SWPadnos> checksum mismatch apparently
[20:17:57] <alex_joni> hmmm.. ok
[20:18:01] <cradek> I just purged and reinstalled i686 emc2-sim_1:2.2.6 and and did not get a warning
[20:18:12] <alex_joni> sim?
[20:18:17] <cradek> yes sim
[20:18:25] <alex_joni> sim wasn't linked against the wrong kernel..
[20:18:33] <alex_joni> linked/compiled
[20:19:17] <jepler> alex and I both touched the repo today .. maybe we stepped on each others' toes
[20:19:28] <alex_joni> probably :/
[20:19:36] <SWPadnos> ouch ouch
[20:19:52] <alex_joni> it's hard to rumba when both are trying to lead
[20:20:05] <alex_joni> especially two graces like jeff & myself :P
[20:20:06] <SWPadnos> so I hear
[20:20:23] <alex_joni> jepler: so.. will you, or should I/
[20:20:24] <alex_joni> ?
[20:20:56] <jepler> alex_joni: I just started a "get", but it'll take me awhile. by contrast, you should be up to doate already
[20:21:01] <jepler> s/doate/date
[20:21:12] <jepler> so if you can (re)do it that'd be dandy and faster than having me do it
[20:21:13] <SWPadnos> marezy doates
[20:21:15] <alex_joni> hmm.. ok, so I'll run the repobuild and push it
[20:21:29] <alex_joni> (but it'll probably push the new .4 kernel on i386)
[20:23:25] <jepler> I don't get any errors on my amd64 machine
[20:23:27] <jepler> fwiw
[20:23:59] <SWPadnos> he's on i386, so maybe you got to the sums file last
[20:24:03] <jepler> could be
[20:26:25] <alex_joni> ok, pushed the new files
[20:26:59] <SWPadnos> and the checksums? :)
[20:27:10] <alex_joni> what checksums? :P
[20:27:15] <SWPadnos> heh
[20:27:26] <alex_joni> they are inside the files I just pushed :)
[20:27:30] <SWPadnos> yay
[22:29:25] <CIA-40> EMC: 03eric-johnson 07TRUNK * 10emc2/src/emc/usr_intf/emcrsh.cc: Redid/enabled command line parameters
[22:29:29] <CIA-40> EMC: 03eric-johnson 07TRUNK * 10emc2/src/hal/utils/halrmt.c: Redid/enabled command line parameters