#linuxcnc-devel | Logs for 2013-08-16

[00:18:43] <psha[work]> mhaberler: have a minute?
[00:18:59] <mhaberler> sure!
[00:19:11] <psha[work]> were there any problems in ring code?
[00:19:15] <mhaberler> zero
[00:19:16] <psha[work]> i'm going to 'legalize' it
[00:19:26] <psha[work]> push to github, add copyrights ets
[00:19:32] <psha[work]> and use in our new project :)
[00:19:39] <mhaberler> whohoo! very good. At work you mean?
[00:19:43] <psha[work]> yea
[00:19:52] <psha[work]> in place of my old propriatary code
[00:20:03] <mhaberler> excellent, extra exposure helps - it is very versatile
[00:20:33] <mhaberler> yesterday I added the mbarriers
[00:21:34] <mhaberler> https://github.com/mhaberler/linuxcnc/commit/0d2f8bb86bea6ceef2e675ab10f49eaf0f6d21fe
[00:22:19] <mhaberler> your project gets away with SPSC?
[00:46:33] <psha[work]> spsc?
[00:46:46] <psha[work]> mit license is ok for you?
[00:52:22] <mhaberler> single producer single consumer
[00:52:26] <mhaberler> yes, very ok
[00:53:51] <psha[work]> no, i hope it will be main ipc link
[00:54:11] <psha[work]> with same proberties - single-signle
[00:54:51] <mhaberler> I found no need for it either, with the exception of the error message q, and that is so low-frequency it's handled with a mutex
[00:56:09] <psha[work]> you mean for mp-sc?
[00:56:15] <mhaberler> right
[00:56:46] <mhaberler> I setup a contention canary flag, and it never triggered
[01:01:57] <memleak> has anyone got xenomai 3.8 on x86 working for linuxcnc?
[01:02:33] <memleak> i'm working on it for RTAI but all the kernel headers have been moved around.
[01:18:53] <memleak> RTAI bumped from 3.4.55 to kernel 3.4.58 (effortlessly)
[02:32:39] <KGB-linuxcnc> 03git 05unified-build-candidate-2 cc2cc1b 06linuxcnc 10debian/configure 10debian/control.in 10docs/man/man9/motion.9 10docs/src/Submakefile * Merge remote-tracking branch 'origin/master' into ubc1-rollforward
[02:32:39] <KGB-linuxcnc> 03git 05unified-build-candidate-2 8a84c6e 06linuxcnc 10tcl/mini.tcl * ui/mini: remove pause/resume sequence from redraw
[02:32:44] <KGB-linuxcnc> 03git 05unified-build-candidate-2 0d2f8bb 06linuxcnc 10src/rtapi/ring.h * rtapi/ring.h: add memory barriers
[02:32:52] <KGB-linuxcnc> 03charles 05unified-build-candidate-2 6507142 06linuxcnc 10configs/ 10(56 files in 8 dirs)
[02:32:53] <KGB-linuxcnc> configs/ARM : Move existing BeagleBone configuration files to new locations per MachineKit branch
[02:33:01] <KGB-linuxcnc> 03charles 05unified-build-candidate-2 c43bb80 06linuxcnc 10configs/ 10(37 files in 8 dirs)
[02:33:05] <KGB-linuxcnc> configs/ARM : Manually merge MachineKit configurations for ARM and BeagleBone
[02:33:08] <KGB-linuxcnc> Refer to the MachineKit branch (at cd06ec5db3147755f242c803a862a511ae78100d
[02:33:10] <KGB-linuxcnc> at the time of this commit) for details on the history of these files, due to
[02:33:13] <KGB-linuxcnc> the massive changes between the branches due to the unified RTOS build, a
[02:33:17] <KGB-linuxcnc> git merge with full history is not practical.
[02:33:36] <KGB-linuxcnc> 03git 05unified-build-candidate-2 ad92495 06linuxcnc * Merge remote-tracking branch 'mah/MachineKit-ubc1' into ubc1-rollforward
[02:52:51] <linuxcnc-build> build #1280 of hardy-amd64-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-amd64-sim/builds/1280 blamelist: Michael Haberler <git@mah.priv.at>
[02:53:09] <linuxcnc-build> build #1276 of hardy-i386-realtime-rip is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-realtime-rip/builds/1276 blamelist: Michael Haberler <git@mah.priv.at>
[02:54:03] <linuxcnc-build> build #1278 of hardy-i386-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-sim/builds/1278 blamelist: Michael Haberler <git@mah.priv.at>
[03:26:33] <linuxcnc-build> build #1276 of lucid-i386-sim is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/lucid-i386-sim/builds/1276 blamelist: Michael Haberler <git@mah.priv.at>
[03:32:12] <linuxcnc-build> build #1274 of checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/checkin/builds/1274 blamelist: Michael Haberler <git@mah.priv.at>
[03:41:20] <linuxcnc-build> build #1279 of hardy-i386-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-sim/builds/1279 blamelist: Michael Haberler <git@mah.priv.at>, Charles Steinkuehler <charles@steinkuehler.net>
[03:41:28] <linuxcnc-build> build #1277 of hardy-i386-realtime-rip is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-realtime-rip/builds/1277 blamelist: Michael Haberler <git@mah.priv.at>, Charles Steinkuehler <charles@steinkuehler.net>
[03:41:50] <linuxcnc-build> build #1281 of hardy-amd64-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-amd64-sim/builds/1281 blamelist: Michael Haberler <git@mah.priv.at>, Charles Steinkuehler <charles@steinkuehler.net>
[04:21:08] <linuxcnc-build> build #1275 of checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/checkin/builds/1275 blamelist: Michael Haberler <git@mah.priv.at>, Charles Steinkuehler <charles@steinkuehler.net>
[05:24:45] <RicardoM> Hi is anyone available?? I'm trying to get the client.nml and server.nml combination working I already found this bug and fixed the files, wich are not updated in Git http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commitdiff;h=c2ec26d
[05:25:08] <RicardoM> I wanted to ask if anybody could help me with getting the setup working
[05:27:24] <RicardoM> with debugging enabled I'm getting the following error "libnml/cms/tcp_srv.cc 1494: Unrecognized request type received.(131072)"
[05:28:35] <RicardoM> then axis keep's doing "waiting s.axes"
[06:56:33] <KGB-linuxcnc> 03jthornton 05v2.5_branch 7a50bec 06linuxcnc 10docs/src/config/stepconf.txt * Docs: add info about using test this axis
[06:56:33] <KGB-linuxcnc> 03jthornton 05v2.5_branch 750d65a 06linuxcnc 10docs/asciidoc-markup.txt * Docs: add info on note and warning paragraph styles
[08:03:38] <KGB-linuxcnc> 03git 05unified-build-candidate-2 ec43a02 06linuxcnc 10src/hal/utils/scope.c * hal/scope: rtapi_shmem_new() with size 0 implies attach whatever the size
[08:56:00] <linuxcnc-build> build #1283 of hardy-amd64-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-amd64-sim/builds/1283 blamelist: Michael Haberler <git@mah.priv.at>
[09:02:13] <linuxcnc-build> build #1281 of hardy-i386-sim is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-sim/builds/1281 blamelist: Michael Haberler <git@mah.priv.at>
[09:02:29] <linuxcnc-build> build #1279 of hardy-i386-realtime-rip is complete: Failure [4failed install-missing-build-dependencies] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/hardy-i386-realtime-rip/builds/1279 blamelist: Michael Haberler <git@mah.priv.at>
[09:38:45] <linuxcnc-build> build #1277 of checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/checkin/builds/1277 blamelist: Michael Haberler <git@mah.priv.at>
[14:04:07] <jepler> RicardoM: I don't think anyone knows whether nml-over-the-net even works
[14:25:35] <KimK> RicardoM: Also, if it makes any difference to you, I believe the plan is to, in due time, drop nml in favor of zmq. (This is above my pay grade though, ask here if you're interested.)
[14:50:06] <KimK> memleak: Hi, no, I have not done anything about the new E350N RTAI system booting yet. Also, I have the older E350N system (the one that was at the fest) running on mhaberler's/zultron's Xenomai system, but it's not here right now. It's running 64-bits and so will probably have to be redone to incorporate Andy's (and your?) 64-bit fixes.
[14:52:54] <KimK> Or else go back to 32-bit, whatever is thought to be best.
[14:53:58] <andypugh> I would be interested in seeing how it works with the fixes, but that should just be a pull and a make.
[14:57:55] <KimK> Hi Andy, yes, I'll certainly let you know what happens, after all your hard work. As you say, it should be fine though, since you've already tested it.
[15:42:07] <CaptHindsight> KimK: is this what you are using on the older system? http://wiki.linuxcnc.org/cgi-bin/wiki.pl?NewRTInstall
[17:40:00] <memleak> andypugh, i cant even get 64-bit linuxcnc tom compile..
[17:40:16] <memleak> SSE sincos issue that has been broken for several months and nobody knows the fix for.. :/
[17:41:01] <andypugh> Strange..
[17:41:22] <memleak> line 662 of src/Makefile
[17:42:02] <memleak> ifeq ($(RTARCH):$(RTAI):$(filter $(RTFLAGS),-msse),x86_64:3:)
[17:42:23] <memleak> error: SSE register return with SSE disabled
[17:42:46] <andypugh> It works for me if I 1) Install Lubuntu 64 bit 2) install the machinekit Xenomai kernel 3) Compile the unified-build-candidate-1 branch from linuxcnc.org.
[17:43:09] <memleak> ah so it only works with xenomai...
[17:43:48] <andypugh> What is not currently really working is Lubuntu 32 bit + Machinekit Xenomai. It boots and I can ssh into the machine fine, but there is absolutely no local video.
[17:44:03] <memleak> 64-bit RTAI works with linuxcnc when it compiles :P
[17:44:05] <andypugh> But this is probably an Intel GMA thing
[17:44:16] <memleak> at least with 2.6
[17:44:26] <memleak> and very well might I add!
[17:44:44] <memleak> 3.x i havent been able to test yet because 64-bit SSE support in linuxcnc is broken
[17:44:58] <andypugh> 32-bit RTAI is working for me on master, 25 and ja3 branches, on 12.04
[17:45:09] <memleak> yes 32-bit RTAI works great
[17:45:27] <andypugh> Am I the only one feeling like a rabbit in headlights with all the places LinuxCNC is going right now?
[17:46:24] <andypugh> Do any of the branches currently already have rtnet and hm2_eth?
[17:46:53] <memleak> once 64-bit RTAI at least compiles for linuxcnc i can start working on the RTAI side of things for 64-bit.
[17:48:17] <memleak> this commit might have broken it: http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commit;h=87a9b86fbcad30e273e20ec5041675c6375c3c75
[17:49:37] <memleak> "rtai for x86_64 is supposed to put -msse in $(rtai-config --module-cflags), but in rtai-modules-2.6.24-16-rtai 3.6.1-linuxcnc.2 (ubuntu hardy amd64) this is not the case." however, it is the case now (and -msse2)
[18:12:37] <skunkworks> andypugh: there is a patch on the wiki
[18:13:11] <skunkworks> I have not had any luck with it
[18:13:14] <andypugh> skunkworks: Yeah, not sure I can make it apply and compile
[18:19:54] <andypugh> A, my probkem (unexcpectedly) is the miniemc dependencies!
[18:20:11] <memleak> rtnet in RTAI still needs some work.
[18:20:33] <memleak> i think it only compiles with some hacks, been awhile since i last tested.
[18:20:35] <andypugh> I _think_ I am on Xenomai.
[18:20:59] <memleak> uname -a should tell you what you're on, if not dmesg
[18:21:05] <andypugh> And as I understand it there has been no attempt made to make 7i80 work on RTAI?
[18:21:54] <memleak> there's a way to do it using sysfs or something but i don't remember off hand, mhaberler might but he quit his client just now.
[18:21:56] <andypugh> Yes, Xenomai. I can sort-of tell that by the blank screen behind me where the dev machine is.
[18:22:31] <KimK_3> CaptHindsight: Yes, that's the Xenomai that I'm using on the older unit (unit 1?)
[18:23:43] <KimK_3> (odd, where did those two leading spaces come from?)
[18:23:56] <KimK_3> CaptHindsight: Yes, that's the Xenomai that I'm using on the older unit (unit 1?)
[18:24:03] <KimK_3> (OK, that's better)
[18:24:28] <memleak> tab completion too soon maybe?
[18:25:18] <skunkworks> andypugh: had to be xenomai and rtos integration
[18:25:23] <memleak> all these Xenomai people!!!!! >:(
[18:25:27] <KimK_3> Maybe. Plus, it was asleep and I jiggled it awake. So maybe bumped the space bar?
[18:25:39] <skunkworks> andypugh: tthis works http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Mesa7i80_Driver_For_Linuxcnc_On_Xenomai
[18:25:43] <memleak> i feel so alone..
[18:26:15] * memleak grabs his blanket
[18:26:36] <skunkworks> sort of. on 12.04 it has missing symbols when trying to load the driver - on 10.04 it seems to work better - but doesn't seem to send the configuration correclty and I get encoder errors
[18:26:49] <andypugh> There is nothing inherently wrong with the new eBay home page, much of what it shows is great. However _my_ search for 30 INT is on business and industrial, not "posing pouches for slender gay chaps" (30 waist, gay int, you see)
[18:27:45] <memleak> heh^
[18:27:51] <andypugh> On the plus side it does mean I spend less time on eBay when at work.
[18:28:53] <KimK_3> Ha, don't worry memleak, I have one on Xenomai and one (hopefully?) on RTAI. So now I feel like one of those big banks who donate to both political parties so no matter who wins, the bankers are on the winning side.
[18:29:53] <andypugh> http://imagebin.org/267863
[18:31:09] <andypugh> If 7i80 worked on RTAI that is where I would have started, it is more familiar and I am perfectly happy with its functionality. (Now)
[18:31:25] <memleak> 7i80 doesn't work with RTAI? isn't that just a mesa card..?
[18:31:40] <memleak> what's broken? I'll try and fix it.
[18:31:46] <memleak> KimK_3, thank you!!!
[18:31:54] <andypugh> It's an _ethernet_ Mesa card. Completely new interface, dependent on rtnet.
[18:32:17] <memleak> did you compile RTAI with rtnet support?
[18:32:24] <andypugh> But, if you want to play: http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Mesa7i80_Driver_For_Linuxcnc_On_Xenomai
[18:33:15] <andypugh> If you do want to play I am moderately sure that PCW would be happy to send you a card.
[18:33:17] <memleak> oh thats why it didnt compile for me.. i didn't have rtnet installed when i tried compiling rtnet into RTAI
[18:33:35] <memleak> well what makes rtnet cards so special?
[18:33:51] <andypugh> They plug into the ethernet port.
[18:34:02] <memleak> and what exactly doesn't function properly with rtai?
[18:34:20] <andypugh> I have no idea :-(
[18:34:30] <memleak> so how do you know it's broken? ;)
[18:34:36] <andypugh> I don't
[18:34:54] <andypugh> But the hm2_eth driver (as far as I know) assumes Xenomai.
[18:35:12] <memleak> hm2_eth sounds like linuxcnc related..
[18:35:32] <andypugh> Yes.. That is the channel we are on...
[18:35:47] <memleak> if a linuxcnc dev can make hm2_eth also work with RTAI if it's the current RT-system then i'm sure i could get it working.
[18:36:15] <andypugh> Hello chicken, here is your egg?
[18:36:15] <memleak> i'm not a linuxcnc person at all, nor an RT-net person, but i know RTAI inside and out.
[18:37:21] <memleak> if hm2_eth can at least detect RTAI and utilize it's functions (assuming it's hard-coded against Xenomai) then i'll be able to work on it.
[18:37:52] <andypugh> I am not sure I have even ever seen the hm2_eth code
[18:38:04] <memleak> define how hm2_eth "assumes" Xeno.. ah.
[18:39:24] <memleak> well depending how much stuff is hard-coded against xenomai that leaves me to modify linuxcnc code, something i'm not comfortable doing..
[18:39:50] <memleak> RTAI bugs in rtnet though that cause the 7i80 to not work properly, that i can do :)
[18:39:56] <KimK_3> Re the 7i80 wiki: Ha, now *I* feel left out, andypugh and skunkworks both have 7i80 rtnet cards, guess I'd better get something going in rtnet too. I'm working through a Mesa order now, I'll talk to them after that's done.
[18:40:29] <andypugh> Background. In linuxcnc hal "loadrt hostmot2" loads the high-level hostmot2 driver. Then loadrt hm2_pci handles the PCI cards and loadr hm2_7i43 handles the parallel port cards. Loadrt hm2_eth handles the cards that are on the ethernet port, via rtnet and a sprinking of magic.
[18:41:32] <memleak> sounds to me like if rtnet in RTAI work, there shouldn't be any problem with 7i80 with RTAI...
[18:41:40] <andypugh> KimK_3: You were too slow :-) The one I have was from the box of freebies that PCW brought to Wichita
[18:42:09] <KimK_3> Ha! Well, I was a
[18:42:10] <memleak> andypugh, I think the odd way I was phrasing things before threw you off. it took me awhile to get my words straight, sorry about that.
[18:42:13] <KimK_3> Oops
[18:42:52] <andypugh> memleak: The whole of the hm2_eth driver is in the patch file here. http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Mesa7i80_Driver_For_Linuxcnc_On_Xenomai
[18:43:11] <memleak> I was trying to say that if the linuxcnc side of things are fine with hm2_eth then that only leaves room for RTAI fixes.
[18:43:24] <KimK_3> Ha! Well, I was a polite fellow then. (No, after you, I insist!)
[18:43:32] <andypugh> Yes, well I am not sure that is true.
[18:44:44] <memleak> andypugh, I see nothing in the source code for 7;80 that hard-codes or "assumes" Xenomai
[18:44:58] <memleak> all linuxcnc / mesa stuff, so in theory, it should work.
[18:45:00] <andypugh> memleak: Don't assume that I know anything about anything. There is an enormous tendency in open-source projects to assume that everyone else is the expert and knows more than you. Can I just state now that I know less than you. ABout everything.
[18:45:39] <KimK_3> Ooh, nice! 7I80 -> 7i80 -> 7;80
[18:45:53] <memleak> KimK_3: hey!
[18:46:05] <KimK_3> Upper case, lower case, and even lower case.
[18:46:09] <memleak> a ; looks like an i anyway..
[18:46:29] <andypugh> Well, I probably know more than you about the inner workings of the 1916 Whyte and Poppe 9.12 litre 4-cylinder engine. But that isn't relevant right now.
[18:46:50] <memleak> andypugh, i hardly know anything about linuxcnc and xenomai.
[18:46:58] <memleak> i'm really just an RTAI developer.
[18:47:22] <andypugh> And neither do I. I know _nothing_ about Xenomai.
[18:47:35] <memleak> you use it more than i do!
[18:48:12] <andypugh> I am a src/hal/drivers/mesa_hostmot2 developer. Anything outside that particular directory is a mystery to me.
[18:49:08] <KimK_3> I just followed the Xenomai wiki page. Tried following the RTAI install, still "under construction", lol
[18:49:48] <memleak> yeah, that's why I wrote the RTAI documentation myself..
[18:50:56] <memleak> I needed to make notes for myself for when my system would crash and i'd have to start over, figured i'd publish those notes online for all to see :)
[18:51:02] <KimK_3> Now this is the link you sent earlier? I don't have it open on this browser, let me scroll back on Xchat...
[18:51:51] <andypugh> memleak: Yours were the only notes on compiling an SMP RTAI kernel that I ever got to work.
[18:52:21] <KimK_3> This one? https://github.com/ShabbyX/RTAI/blob/master/README.INSTALL#L179
[18:52:24] <memleak> SMP? i don't recall ever posted SMP-specific docs..
[18:52:29] <memleak> *posting
[18:52:34] <memleak> KimK_3, yes
[18:52:51] <memleak> after that you just compile linuxcnc as usual. ./configure make etc
[18:53:06] <andypugh> memleak: I am hopeless with names. Have I confused you with Neo?
[18:53:15] <memleak> i am neo :)
[18:53:21] <andypugh> Yeah, I thought so.
[18:54:03] <memleak> I just tend to keep my docs very universal and work with everything. never tied down to distribution or machine layout.
[18:54:45] <memleak> can you post a link to the docs you're referring to?
[18:56:41] <andypugh> This one: http://code.google.com/p/neo-technical/wiki/emc2ubuntu
[18:57:29] <andypugh> It was quite some time ago
[18:57:51] <memleak> heh, that was back before RTAI wrote it's own udev rules, heh
[18:57:59] <memleak> i said heh twice?..
[18:59:22] <KimK_3> memleak: Yes, unfortunately I never saw your document before I started, I just stumbled in based on the simple advice, "Install 12.04(32), then install Seb's debs". So now I'm not quite sure where to jump to on your document, which begins with compiling, you'd think it would be obvious, but I'm not quite sure of the re-entry point.
[18:59:24] <andypugh> it works for Avril http://youtu.be/nq5gfVGJkRQ
[18:59:52] <memleak> KimK_3, the grub file :)
[19:00:02] <memleak> That's why #L179 is in the page link
[19:00:08] <memleak> line 179
[19:01:02] <memleak> andypugh, nice link!
[19:01:45] <memleak> KimK_3, please note that that grub.cfg provided will most likely not work with your hard drive partitioning scheme so you'll need to modify it.
[19:02:08] <memleak> just replace the whole file. rm -rf /boot/grub/grub.cfg && vim /boot/grub/grub.cfg
[19:02:18] <memleak> base it off the one posted, 3.4.55 should boot fine
[19:02:38] <memleak> unless seb forgot to post an initrd and the kernel expects one.. (the .img file)
[19:03:14] <memleak> i never tested his debs let alone even unpacked them. i do everything from source by hand because i can, thanks to gentoo :)
[19:08:31] <andypugh> Ah, err, I got ona youtube random walk and ended up looking at http://youtu.be/bN4t9O3sGP0
[19:25:25] <KimK_3> memleak: Sorry, I was called away for a bit. OK, I'm looking at line 179, etc. And /boot/grub2/grub.cfg will get overwritten if I do "sudo update-grub" (or whatever it is), right? (I think ubuntu links "update-grub" to the longer command you suggested earlier) So no matter how badly I screw up, I should be able to restore things with "update-grub"? Yes, I think there may be a missing file? See: http://imagebin.org/267621
[19:28:44] <memleak> you'll be able to restore it yes
[19:29:09] <memleak> depending on how seb packaged the .deb it might not have a .img file in it
[19:29:19] <memleak> so the kernel might not even expect a .img file
[19:29:26] <memleak> not sure...
[19:29:39] <andypugh> I think I will give up and get back to the tooltable work.
[19:29:41] <andypugh> make: *** No rule to make target `hal/drivers/mesa-hostmot2/hm2_eth.c', needed by `objects/rthal/drivers/mesa-hostmot2/hm2_eth.o'. Stop.
[19:29:59] <memleak> going afk for awhile
[19:30:04] <KimK_3> It looks like ...51-generic... and ...55-rtai... come in pairs, usually, except for the image, and an "api"(?)
[19:30:17] <andypugh> Goodnight chaps
[19:30:19] <KimK_3> OK, thanks for the help, goodnight
[19:31:04] <KimK_3> Sorry, make that "abi"
[19:43:18] <KimK_3> I'll be back later too.
[20:07:21] <memleak> KimK_3, if you're worried about that .img initrd / init ramfs / initial ram filesystem you can decompress the .deb with ar
[20:07:49] <memleak> ar xv *55*.deb for example
[20:08:08] <memleak> if the .img file isnt in data.tar.bz2 then seb didnt make the kernel depend on one, which is fine.
[20:08:13] <memleak> i personally never use initrds
[20:08:43] <memleak> i compile the kernel and have _it_ figure out what my machine needs.
[20:10:53] <memleak> if you want, you can use dracut or whatever ubuntu uses to generate the .img file though.
[20:13:33] <memleak> looks like ubuntu uses mkinitramfs: http://manpages.ubuntu.com/manpages/lucid/man8/mkinitramfs.8.html
[20:14:25] <memleak> for precise / 12.04: http://manpages.ubuntu.com/manpages/precise/man8/mkinitramfs.8.html
[20:19:28] <KimK> memleak: OK, thanks. I'm hoping that I won't have to mess with it too much, and that if the missing tidbit(?) can be discovered, RTAI will just appear in the boot menu automatically. Grub2 does a good job, generally, I'm not sure why it's confused this time around. And I'm not a kernel/booting/grub2 expert, are there any general kernel tutorials you might suggest? And RTAI too, in case I have time for one more tutorial?
[20:32:37] <KimK> I'll be back (on unit 1) as KimK_2 in a few minutes. I had a laptop that was _1, but I think I'll rename the laptop now, unit 1 = _2 and unit 2 = _3 is too confusing.
[20:32:50] <memleak> I learned all about kernel compiling from this: https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/tree/README?id=refs/tags/v3.10.7
[20:33:15] <memleak> as for RTAI the README.INSTALL file should cover everything on github.
[20:34:03] <memleak> grub2 help: http://www.linuxfromscratch.org/lfs/view/development/chapter08/grub.html
[20:35:19] <memleak> my rule of advice, never follow distro-specific documentation and always use the upstream documentation by the original developers whenever possible.
[20:36:27] <mozmck> memleak: unless you are building a packaged kernel for a distribution - especially a generic one to run on as many machines as possible.
[20:37:27] <mozmck> Here's a site I got some good information from when making the kernel packages for ubuntu 10.04. http://blog.avirtualhome.com/category/ubuntu/compile-kernel-ubuntu/
[20:38:04] <memleak> https://help.ubuntu.com/community/Kernel/Compile
[20:38:59] <memleak> KimK already has a kernel though, pre-packaged, the rest is just modifying grub to actually read it. why it won't do it automatically, i have no idea.
[20:39:00] <mozmck> That page is better now than it was, but it was way behind/inaccurate. It now actually has links to the site I just sent
[20:39:25] <mozmck> ah. I don't know.
[20:40:33] <memleak> I remember back when i was an ubuntu developer working on kernels how great everything was until they kicked me out of the dev team because i exposed too many of their secrets with compiling kernels...
[20:41:36] <memleak> ever since i stopped following distro-specific documentation and you can make large generic kernels without the help of any ubuntu / debian documentation.
[20:42:49] <mozmck> kicked you out for exposing secrets?!? They should have given you an award!
[20:43:02] <memleak> if you want a .deb kernel just read man make-kpkg, no need for misleading wiki pages.
[20:43:19] <memleak> that's canonical for you..
[20:43:23] <mozmck> I used make-kpkg quite a bit before.
[20:43:30] <memleak> yeah it actually works!
[20:44:13] <mozmck> I was trying to build the 10.04 packages and had a hard time finding info. I asked on IRC channels and looked around and finally found and pieced together enough information to get them built.
[20:44:35] <mozmck> I remember having to dig through quite a bit of stuff to get it figured out.
[20:44:57] <memleak> that's all done intentionally :)
[20:45:01] <memleak> don't worry!
[20:45:04] <mozmck> bleh!
[20:45:58] <mozmck> I like debian. I first installed it from 6 floppies, and kept that computer running on debian unstable for about 12 years I think just doing incremental upgrades.
[20:46:55] <mozmck> Switched to ubuntu for my main computer though a while back because I wanted something a little more stable than debian unstable had been at times.
[20:53:54] <memleak> debian unstable has never actually been... unstable for me..
[20:54:07] <memleak> i feel like it's false advertising...
[20:54:25] <memleak> placebo effect of it being unstable perhaps?
[20:59:02] <mozmck> Well, it was a number of times for me - but only maybe a dozen or so times over 10 or 12 years.
[20:59:48] <mozmck> Various programs would quit working after an update, and several times X wouldn't start.
[21:02:48] <memleak> X not starting is a common issue for many even today, regardless of distro.
[22:30:27] <memleak> ah finally linuxcnc compiles with RTAI 64-bit: http://pastebin.com/TEZwMLYz
[22:31:28] <memleak> rtai_math loading problems now.. ugh
[22:32:03] <memleak> Unknown symbol sincos (err 0)
[22:34:02] <memleak> i should probably get RTAI working with -ffast-math too.. maybe that'll fix it.