#linuxcnc-devel | Logs for 2013-12-31

[02:15:42] <memleak> well im back to some RTAI development for awhile
[02:16:24] <memleak> been a few months since i last touched anything so i'm kind of out of date, but feel in a great mood to go back to it for awhile
[02:17:35] <memleak> hey seb_kuzminsky if i restabilize all the latest RTAI code and make it all rock solid like a did before, you think you could repackage some debs?
[05:08:20] <skunkworks> Tom_itx: had that for 9 years or so... (so far)
[05:09:52] <archivist> they are 99% front loader over here, some get bearing problems but I bet all get some design related failure
[05:27:35] <skunkworks> can't say we have had any real issues with them yet. Knock on wood..
[06:36:29] <skunkworks> pcw_home, New updated install of 12.04.3 - pulled the rt-preemt as you had showed me. Built linuxcnc. set the nic ip4 to manual,,, (ip address, subnet, gateway) and set ip6 to ignore.
[06:36:49] <skunkworks> latency seems to be around 70us or so.
[06:37:00] <skunkworks> this is my current hal file
[06:37:01] <skunkworks> http://pastebin.ca/2520687
[06:37:15] <skunkworks> I run it with halrun -I -f save.hal
[06:37:38] <skunkworks> the watchdog bites almost instantly. If I set the timeout to 1000000000 - it flashes once.
[06:37:45] <skunkworks> what am I missing?
[06:41:12] <skunkworks> could the firmware in the card?
[06:41:24] <skunkworks> http://pastebin.ca/2460941?srch=7i80
[06:43:22] <skunkworks> this is on the atom board
[07:27:38] <skunkworks> logger[mah
[07:27:42] <skunkworks> logger[mah],
[07:27:42] <logger[mah]> skunkworks: Log stored at http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-12-31.html
[08:31:37] <KGB-linuxcnc> 03Norbert Schechner 05master dfd4d41 06linuxcnc 10share/gscreen/skins/gmoccapy/gmoccapy_handler.py 10share/gscreen/skins/gmoccapy/release_notes.txt gmoccapy_0_9_9_7_4 - toolchange dialog with description * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=dfd4d41
[09:04:06] <eric_unterhausen> memleak, I tried the 32 bit rtai 3.4 kernel and xenomai kernel on ubuntu precise, but I got latency spikes due to the radeon driver
[09:05:05] <eric_unterhausen> vesa didn't have those spikes, but I hosed the system trying to switch to software opengl
[09:05:39] <CaptHindsight> eric_unterhausen: I thought you were using the RTAI kernel, sorry
[09:05:54] <eric_unterhausen> I used both
[09:05:58] <CaptHindsight> oh, you tried both
[09:05:59] <cradek> I found if I didn't check the "use proprietary software to make my MP3s play right" checkbutton when installing, I didn't get the terrible nvidia driver
[09:06:19] <eric_unterhausen> it was the open-source driver
[09:06:28] <eric_unterhausen> closed source is fglrx
[09:06:32] <cradek> oh ok. in lucid you used to be able to switch to software rendering by simply installing a package...?
[09:06:49] <cradek> do you know what you did to hose it?
[09:06:53] <CaptHindsight> eric_unterhausen: he'll be around in a few hours, what hardware is this again?
[09:07:26] <eric_unterhausen> it's a asrock 790gx dual core athlon
[09:07:32] <cradek> looks like it is/was called libgl1-mesa-swx11
[09:07:49] <eric_unterhausen> cradek, I tried installing that, there was some conflict
[09:08:00] <eric_unterhausen> trying to resolve the conflict removed linux :)
[09:08:02] <cradek> yes, that's correct, as it should uninstall other stuff
[09:08:23] <cradek> ok I'm back to not knowing what you did...
[09:08:52] <eric_unterhausen> I tried uninstalling the conflict, it uninstalled everything that depended on it
[09:09:20] <eric_unterhausen> which included linux-desktop and just about everything required for a graphical linux system
[09:10:10] <cradek> like what command did you type to try uninstalling the conflict?
[09:10:11] <eric_unterhausen> my conflict resolution skills not that great
[09:10:40] <eric_unterhausen> apt-get uninstall mesa-something-glx
[09:11:07] <cradek> that's not what you want to do. if you install -swx11, it will replace the non-swx11 packages that conflict
[09:11:29] <eric_unterhausen> it refused to do that, should I have forced?
[09:12:45] <jepler> I expect apt to say something like this: http://pastebin.com/QVugXTSC
[09:12:54] <cradek> probably not. perhaps they have broken it in later versions.
[09:12:56] <jepler> which looks like it would involve a minimum of fuss to switch to software X11
[09:13:02] <jepler> this is on debian wheezy, ymmv
[09:13:19] <cradek> yes I agree that's about what should happen
[09:13:47] <cradek> if ubuntu folks screwed it up I'd like to know - it's an important consideration for our next cd
[09:13:59] <eric_unterhausen> ok, I think I can recover the system
[09:15:26] <eric_unterhausen> any time anyone asks about vesa, the ubuntu people try to help them get the accelerated drivers to work
[09:15:58] <cradek> sure, that's probably what most of their askers want
[09:18:48] <eric_unterhausen> one future consideration of ubuntu is that the 3d desktop runs horribly on some pretty serviceable computers
[09:18:56] <eric_unterhausen> they are dropping the 2d desktop
[09:19:23] <cradek> yes in their base distribution. the xubuntu distrib is still fairly normal, though
[09:22:45] <cradek> my current preference is wheezy+xfce
[09:23:01] <eric_unterhausen> I hate to admit that I'm used to unity now
[09:23:29] <cradek> I've only found minor brokennesses, and none of them were things broken in the name of beauty/modernity/looking like a mac
[09:23:53] <eric_unterhausen> I could probably get used to not-unity pretty fast though
[09:24:11] <cradek> I hear people can get used to anything
[09:24:47] <eric_unterhausen> the main thing is the mac buttons, the "menuing" system is just stupid
[09:25:47] <eric_unterhausen> replacing menus with search is a bad idea, adding search is a much better idea
[09:26:34] <eric_unterhausen> it's a fixed database which can become hosed, it happens often enough that you can find solutions
[09:27:01] <eric_unterhausen> unfortunately, you can't start a terminal if the dash database is hosed
[09:28:55] <mozmck> eric_unterhausen: you should try the Cinnamon desktop. They have search added to the menus like you mention
[09:29:11] <mozmck> Linux Mint uses bash still by default it looks like.
[09:29:11] <cradek> I hear soon they're going to use a reinvented X (not that first one, a second one) so then the decision to just avoid them will be easy
[09:29:48] <cradek> oh dash is a shell? I thought it was probably a gui thing.
[09:29:58] <cradek> it does sound like a shelly name
[09:30:32] <mozmck> well, it could also be a gui thing I guess, but ubuntu replaced bash with dash a number of versions back.
[09:30:39] <mozmck> maybe 9.x?
[09:31:07] <mozmck> I figured out how to get bash back because dash did not handle some paths correctly at the time.
[10:08:52] <eric_unterhausen> un-hosing my system has commenced
[10:11:55] <eric_unterhausen> dash is like the search box on win7, don't think it's a bash replacement
[10:53:11] <eric_unterhausen> I forgot how long ubuntu takes to install over an existing install, it's hours
[11:14:17] <CaptHindsight> if that works I'll really be surprised
[11:16:01] <eric_unterhausen> I've done it 3 times successfully
[11:16:40] <eric_unterhausen> if you are careless with apt-get, you learn such things
[11:17:17] <eric_unterhausen> here's one that will get you every time: apt-get uninstall python
[11:18:18] <eric_unterhausen> I think even the c compiler packages depend on python nowadays
[11:46:00] <eric_unterhausen> CaptHindsight: worked
[11:49:00] <eric_unterhausen> apparently under ubuntu 13, it doesn't even whine when you install over an existing filesystem
[11:52:18] <CaptHindsight> maybe they are finally fixing things
[11:52:45] <CaptHindsight> fix 10 things here and break 10 things there
[11:53:06] <CaptHindsight> why all the desktop madness?
[11:55:09] <eric_unterhausen> they want to be microsoft
[11:55:42] <eric_unterhausen> and they are doing a pretty good job of that given that microsoft has no idea what they are doing nowadays
[11:58:45] <eric_unterhausen> apt-get update says there are duplicate sources, run apt-get update to fix
[12:06:24] <Tom_itx> maybe update your source locations?
[12:06:53] <eric_unterhausen> ya, i'm trying to figure out what the graphical widget is called that does that
[12:07:08] <eric_unterhausen> would be nice if it was just called software-sources, but no such luck
[12:07:33] <Tom_itx>  /etc/apt/sources.list
[12:07:38] <Tom_itx> or
[12:07:55] <eric_unterhausen> ya, I can vi that or kill myself, similar outcomes
[12:07:56] <Tom_itx>  /etc/apt/sources.list/linuxcnc-buildbot.list
[12:08:32] <eric_unterhausen> actually, there is a /var/lib/apt/lists/ and /etc/apt/sources.list
[12:08:52] <eric_unterhausen> I'm tempted to delete the one in /var, but I have a bad history of doing the easy thing and hosing my system
[12:09:12] <Tom_itx> or rename it
[12:09:30] <eric_unterhausen> now where's the fun in that?
[12:12:25] <eric_unterhausen> I ran out of coffee anyway, going downstairs to do it the easy way
[12:59:54] <skunkworks> pcw_home, logger[mah]
[13:00:12] <skunkworks> logger[mah],
[13:00:12] <logger[mah]> skunkworks: Log stored at http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-12-31.html
[13:01:36] <skunkworks> pcw_home, http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-12-31.html#12:20:19
[13:03:48] <skunkworks> I must be doing something stupid
[13:06:45] <pcw_home> try this: freeby.mesanet.com/hm2-pidstepper.zip
[13:06:46] <pcw_home> (you need to change the board name and mac address at a minimum)
[13:07:23] <pcw_home> (and use a stepgen config)
[13:08:56] <skunkworks> there isn't stepgens in the 7i80
[13:09:12] <skunkworks> (in my 7i80)
[13:09:36] <pcw_home> well you can update the config (via mesaflash)
[13:09:46] <skunkworks> what one is a good one to put into it?
[13:10:05] <pcw_home> do you have a DB or HD?
[13:10:11] <skunkworks> hd
[13:10:22] <skunkworks> http://pastebin.ca/2460941?srch=7i80
[13:10:44] <skunkworks> (I have the zip with all the bit files)
[13:10:55] <skunkworks> downloaded it this morning)
[13:11:15] <micges> hi guys
[13:11:38] <pcw_home> any with at least 3 stepgens should do
[13:12:29] <pcw_home> 7i80hd_16_svst4_8.bit for example
[13:48:25] <memleak> eric_unterhausen, if latency is fine with vesa why did you need to change the opengl renderer?
[13:48:59] <memleak> X automatically falls back to softpipe, llvmpipe, swrast if r300/r600/radeonsi driver is not used
[13:49:47] <memleak> you can't have hardware 3d accel without hardware 2D accel so there is no point in forcing a software-based 3d accel driver when using xf86-video-vesa
[13:50:15] <skunkworks> pcw_home, that zip seems to be corrupted..
[13:50:29] <skunkworks> (of the pid stepgen)
[13:50:43] <skunkworks> I did update the 7i80 (from windows..)
[13:51:19] <memleak> you can however have hardware 2d acceleration with software 3d accel such as xf86-video-ati paired with llvmpipe (my personal favorite)
[13:53:48] <memleak> uninstalling packages once you've hosed the 2d driver for the GPU is no longer necessary
[13:55:09] <eric_unterhausen> memleak, I no opengl would work
[13:55:27] <pcw_home> skunkworks: try refetching
[14:08:22] <skunkworks> Got it.
[14:08:29] <skunkworks> still triggers watchdog
[14:08:43] <pcw_home> weird
[14:08:49] <skunkworks> I'll say :)
[14:09:05] <skunkworks> again - I have to be doing something stupid
[14:09:35] <pcw_home> can you ping the 7I80 with 0 packet loss?
[14:09:56] <pcw_home> maybe there a power supply issue or something
[14:10:03] <pcw_home> there's
[14:10:39] <skunkworks> oh...
[14:10:57] <skunkworks> I am using a 5v charger for a cell phone..
[14:11:11] <skunkworks> but rtnet seemed to work.
[14:11:18] <skunkworks> let me measure the voltage
[14:11:57] <skunkworks> 5.08v
[14:15:50] <skunkworks> pcw_home,
[14:15:50] <skunkworks> --- ping statistics ---
[14:15:50] <skunkworks> 115149 packets transmitted, 115149 received, 0% packet loss, time 115534ms
[14:15:51] <skunkworks> rtt min/avg/max/mdev = 0.063/0.079/8.097/0.026 ms
[14:16:19] <pcw_home> all 5 loose MBs I have work for me: D525,D945GCLF,AMD E450,ASUS something and my desktop here
[14:16:32] <skunkworks> well that sucks :)
[14:18:27] <pcw_home> Andy pughs d2800 also:
[14:18:28] <pcw_home> http://www.youtube.com/watch?v=2HdikHRXnAs
[14:21:36] <pcw_home> your max ping time indicates a problem 8 ms!
[14:23:41] <skunkworks> 64 bytes from icmp_req=29097 ttl=64 time=8.09 ms
[14:24:09] <skunkworks> pcw_home, usb keyboard/mouse?
[14:24:30] <skunkworks> I have the audio, serial, printer port disabled in bios
[14:24:36] <skunkworks> also lagacy usb
[14:24:39] <pcw_home> dont think so (I have USB mice on all my test systems)
[14:25:21] <memleak> eric_unterhausen, glxinfo?
[14:25:30] <skunkworks> I could try a different power supply for grins
[14:28:53] <micges> skunkworks: bad eth cable?
[14:29:15] <skunkworks> I could try a different one
[14:31:56] <cradek> ping -f, and watching the terminal while jiggling things, can be useful
[14:32:27] <cradek> oh but you didn't have any drops, did you
[14:36:49] <kwallace> So far with Python and Gtk I have been massaging other people's programs or I've pasted code from programs that do what I want, but I seem to have hit an application that I can't find an example for.
[14:37:02] <kwallace> It may be fall down simple but I'm just not getting it yet. One example might be to use Glade to create a window and box, which I can do just fine, then add Mplayer into the box.
[14:37:22] <kwallace> The Mplayer command could be "mplayer tv://0". Does anyone have a link or keywords I can use to find out how to do this?
[14:37:46] <cradek> try xembed as a keyword that might help
[14:38:01] <kwallace> Thank you.
[14:38:29] <cradek> welcome
[14:39:22] <cradek> ooh, this looks useful for you: https://code.google.com/p/pormplayer/
[14:40:03] <cradek> um, I mean, as a code example
[14:40:05] <cradek> y'know
[14:40:30] <skunkworks> different cable - different power supply. still bites
[14:40:48] <skunkworks> I ran the latency test for a few hours - peaked at about 70us
[14:41:56] <cradek> normally nobody cares one bit if an ethernet packet gets delayed by 8ms... wonder what all is lurking in ethernet cards.
[14:42:37] <skunkworks> pcw and I both have the same motherboards.. his 'suposidly' works ;)
[14:42:42] <cradek> skunkworks: [if you didn't check already] does ifconfig show overrun/framing/collision errors?
[14:44:50] <skunkworks> zero errors, overruns, framing drops
[14:44:57] <cradek> collisions?
[14:45:22] <skunkworks> 0
[14:45:32] <skunkworks> that one was hiding..
[14:45:34] <cradek> huh
[14:45:55] <skunkworks> I have to be doing something wrong
[14:46:17] <skunkworks> I even disabled the network from the gui - then did a
[14:46:30] <skunkworks> sudo ifconfig eth0
[14:46:55] <cradek> personally I would not rule out network-manager screwing with it
[14:46:59] <cradek> that thing is obnoxious
[14:47:11] <skunkworks> can I uninstall that?
[14:47:25] <cradek> you have all normal wired interfaces?
[14:47:45] <skunkworks> I just have 1 - onboard
[14:47:56] <skunkworks> rtl61111 iirc
[14:48:07] <skunkworks> rtl6111
[14:48:21] <cradek> oh then sure, it's easy, apt-get remove --purge network-manager, then just specify your params in /etc/network/interfaces like normal
[14:49:08] <skunkworks> but peters works!
[14:49:25] <cradek> welllll
[14:49:31] <skunkworks> :)
[14:50:01] <cradek> you had one delay in 115149 tests?
[14:50:47] <cradek> i.e. .0008% of your packets had a problem
[14:51:13] <cradek> not seeing that problem on a different run or two is not very surprising
[14:51:44] <cradek> although in retrospect network-manager isn't going to break it so subtly I bet
[14:56:37] <skunkworks> max thread time is 830us..
[14:56:54] <skunkworks> did not help
[14:57:20] <cradek> how frequent is it? you have timestamps by your ping results
[14:58:10] <skunkworks> the second hal runs or the watchdog gets reset
[14:58:29] <skunkworks> if I increase the watchdog time out - say to 1 second - it last about 1 second
[15:02:26] <micges> skunkworks: what kernel?
[15:02:45] <micges> newest rt-preempt from mah?
[15:07:02] <skunkworks> I think so
[15:09:22] <skunkworks> freeby.mesanet.com/hm2-pidstepper.zip
[15:09:30] <skunkworks> heh
[15:09:31] <skunkworks> Linux empire-OptiPlex-380 3.12.5-rt7mah #1 SMP PREEMPT RT Thu Dec 19 18:57:34 CET 2013 i686 i686 i386 GNU/Linux
[15:09:53] <skunkworks> I even for grins enabled hyperthreading... didn't help
[15:17:26] <skunkworks> bbl
[16:21:20] <skunkworks> logger[mah]:
[16:21:21] <logger[mah]> skunkworks: Log stored at http://linuxcnc.mah.priv.at/irc/%23linuxcnc-devel/2013-12-31.html
[19:55:07] <eric_unterhausen> precise rtai kernel headers return following error on my system:
[19:55:09] <eric_unterhausen> http://pastebin.com/DAHfjMWE
[19:58:16] <eric_unterhausen> error is that linux purists need to be smacked upside the head http://pastebin.com/1zamrMwt
[19:58:33] <CaptHindsight> lol
[20:18:05] <eric_unterhausen> I guess that comes from my attempt to install the fglrx module, but it would be nice to be given a workaround
[22:51:48] <eric_unterhausen> okey, so I moved the fglrx stuff to a safe place and it installed correctly