#emc-devel | Logs for 2007-04-27

[00:26:39] <cradek> yay, got my new processors with proper heat sinks - now let's see if the machine stops turning itself off
[00:27:00] <jmk-core2> mine isn't turning itself off, just slowing down
[00:27:32] <cradek> you have a heat problem on your new system?
[00:28:27] <jmk-core2> I don't think its heat
[00:28:44] <jmk-core2> but not sure yet... trying to figure out how to read the CPU temp and other sensors
[00:29:03] <jmk-core2> I think it switched down from 2.4GHz to 1.6GHz
[00:29:42] <jepler> with the standard ubuntu kernel you automatically get frequency changing when the CPU is not very utilized
[00:30:31] <jmk-core2> running "make j" last night took 35ish seconds, today its taking 47
[00:30:43] <jmk-core2> I would hope that running a make would result in the kernel going to full speed
[00:30:57] <cradek> wow that's fast
[00:31:45] <jmk-core2> ok, it looks like it did speed up during the compile...
[00:31:49] <jepler> it should quickly adapt and return to max speed
[00:31:55] <jmk-core2> (I'm using lshw, which is rather limited)
[00:32:44] <jepler> you can do something like: watch -n.1 grep MHz /proc/cpuinfo
[00:33:21] <jmk-core2> cpuinfo doesn't show the actual speed I don't think
[00:33:42] <jmk-core2> jmkasunich@solo:~$ lshw -class cpu
[00:33:43] <jmk-core2> WARNING: you should run this program as super-user.
[00:33:43] <jmk-core2> *-cpu:0
[00:33:43] <jmk-core2> product: Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
[00:33:43] <jmk-core2> vendor: Intel Corp.
[00:33:44] <jmk-core2> physical id: 1
[00:33:46] <jmk-core2> bus info: cpu@0
[00:33:47] <jmk-core2> version: 6.15.6
[00:33:49] <jmk-core2> serial: 0000-06F6-0000-0000-0000-0000
[00:33:52] <jmk-core2> size: 1596MHz
[00:33:53] <jmk-core2> capacity: 1596MHz
[00:33:56] <jmk-core2> width: 64 bits
[00:33:58] <jmk-core2> thats lshw
[00:34:00] <jmk-core2> cpuinfo still says 2.4G
[00:34:22] <cradek> I advise against make -j on a 384MB machine...
[00:34:26] <jepler> cradek: hah
[00:34:40] <cradek> it doesn't hurt as much as you think, though
[00:35:02] <cradek> (scsi fwiw)
[00:35:07] <jmk-core2> I just read something that recommend make -j X, where X - number of cpus + 1
[00:35:32] <jepler> it won't take long to compare -j, -j2, and -j3
[00:35:41] <jepler> I bet -j3 is nearly the same as -j2, and -j is slower
[00:36:01] <jmk-core2> what is bugging me is the slowdown since last night
[00:36:19] <jmk-core2> is it possible your makefile fixes would have increased make time?
[00:36:35] <jmk-core2> (I ran makes on my regular system, and there is no significant change from yesterday)
[00:36:53] <jepler> I don't think it should have changed it by much
[00:37:03] <jmk-core2> didn't think so
[00:37:38] <jmk-core2> I need to get lmsensors running and see how hot its getting
[00:38:04] <jmk-core2> and I need a trustworthy way of measuring actual clock frequency
[00:38:12] <jmk-core2> rdtsc every minute or something
[00:38:27] <jmk-core2> every second I mean
[00:39:04] <jepler> does /proc/cpuinfo say the same as lshw did?
[00:39:10] <jmk-core2> no
[00:39:53] <jepler> hm OK -- same on my system
[00:40:08] <jepler> there's this file: /sys/devices/system/cpu/*/cpufreq/scaling_cur_freq
[00:40:39] <jmk-core2> http://pastebin.ca/460335
[00:40:44] <jmk-core2> thats what I get when idle
[00:41:15] <cradek> make -j3 and make -j took within 3 secs of each other here, even though -j swapped somewhat
[00:41:43] <jmk-core2> lines 11 and 34 go to 2394MHz during a build, so lswh is apparently showing the actual freq
[00:42:11] <jtr__> jtr__ is now known as jtr
[00:43:14] <jmk-core2> the /sys files show the actual frequency
[00:43:58] <cradek> jmk-core2: did you notice there's a gnome cpu frequency scaling monitor
[00:44:28] <jmk-core2> yeah, I was going thru synaptic looking for stuff
[00:44:35] <cradek> you already have it
[00:44:43] <jepler> I have 'powernowd' installed but that may not be the default -- I don't use the gnome stuff, though
[00:44:57] <jmk-core2> cradek: how do I invoke it?
[00:45:25] <jmk-core2> you mean this one "GNOME CPUFreq Applet: is a CPU Frequency Scaling Monitor for GNOME Panel."
[00:45:29] <jepler> jmk-core2: btw what do you think of merging counter into encoder by providing an additional LUT (which ignores the B input) and which is selected by a new parameter?
[00:45:30] <cradek> first, locate your mouse and try not to drool
[00:45:42] <cradek> right click on a panel, "add to panel", ...
[00:45:51] <jmk-core2> ah, ok
[00:45:52] <jepler> that would seem to be the easiest way to get the nice new frequency estimation for 'counter'
[00:46:00] <jmk-core2> jepler: agreed
[00:46:02] <jepler> as opposed to re-copying it and re-modifying it
[00:46:14] <jmk-core2> we have x1 and x4 mode now, we just need "single-phase" mode
[00:47:09] <jepler> http://emergent.unpy.net/-files/sandbox/encoder+counter.patch
[00:47:14] <jepler> er, whoops
[00:47:16] <jepler> http://emergent.unpy.net/files/sandbox/encoder+counter.patch
[00:47:26] <jepler> hm it's empty?
[00:47:28] <jepler> i must have done something wrong
[00:48:15] <jepler> http://emergent.unpy.net/files/sandbox/encoder+counter.patch
[00:48:17] <jepler> there
[00:48:36] <jepler> lightly tested -- modifying the encoder-counter.0 test to use the new encoder counter mode passes
[00:49:24] <jmk-core2> you are still exporting both pins...
[00:49:41] <jepler> yes because it's selected by HAL parameter, not by module parameter
[00:49:44] <jmk-core2> I suppose the only way to avoid that is an insmod param, which kind of sucks
[00:50:47] <jmk-core2> why the extra masking? state = lut_ctr[state & (SM_LOOKUP_MASK & ~SM_PHASE_B_MASK)];
[00:51:07] <jepler> I was thinking: "ignore anything on PHASE_B"
[00:51:19] <jepler> I could also change the LUT so that phase B doesn't matter..
[00:51:26] <jmk-core2> thats what I was thinking
[00:51:42] <jepler> that would sure make that other code look more regular
[00:51:54] <jmk-core2> I overlooked the ~ and thought you were using phase B, which would be confusing
[00:52:03] <jepler> oh -- no, that's not at all what I intendeed
[00:52:24] <jmk-core2> ;-)
[00:57:02] <petev> jmkasunich, u there?
[00:57:09] <jmk-core2> no
[00:57:30] <jmk-core2> ;-)
[00:57:34] <jmk-core2> wassup?
[00:57:34] <petev> I have had emc unexpectedly shutdown by itself on me several times now
[00:57:58] <jmk-core2> what messages does it print?
[00:58:00] <petev> once when logging back in after the screen saver went on
[00:58:05] <jepler> petev: I saw your earlier check-in -- usually the .nml file is copied from configs/common, but it requires new lines in the top-level Makefile. Would you like me to add those lines for you?
[00:58:12] <petev> once when I plugged in a usb stick
[00:58:22] <petev> jepler, sure
[00:58:31] <petev> and once while doing nothing at all
[00:58:34] <jepler> petev: OK
[00:58:40] <jmk-core2> what messages does it print?
[00:58:50] <petev> I saved the output from the error dialog
[00:58:59] <petev> but I see nothing that looks like an error
[00:59:05] <petev> you want me to email it to you?
[00:59:08] <jmk-core2> what about the kernel log
[00:59:09] <jmk-core2> no
[00:59:17] <petev> it seemed to shut down cleanly
[00:59:28] <petev> but I can't figure out why
[00:59:33] <jmk-core2> pastebin it - I'm not the only (or even the best) person to look at it - pastebin lets everybody see
[00:59:41] <petev> ok, hang on
[01:01:21] <petev> hmm, pastebin gave me an error
[01:01:35] <jmk-core2> pastebin.ca?
[01:01:35] <petev> Query failure: Can't open file: 'pastebin.MYI'. (errno: 145)
[01:01:45] <petev> no, was using pastebin.com
[01:01:53] <jmk-core2> use .ca, it works better
[01:03:12] <petev> http://pastebin.ca/460413
[01:03:48] <petev> this was built from the latest head from a couple days ago
[01:04:26] <cradek> were you running gcode at the time?
[01:04:26] <jmk-core2> ran about a minute and then just up and died?
[01:04:27] <jepler> X connection to :0.0 broken (explicit kill or server shutdown).
[01:04:52] <petev> yeah, I was in another window (show machine) and it just shut down
[01:05:14] <jmk-core2> anything that kills the GUI will kill all of EMC
[01:05:24] <petev> x was fine
[01:05:31] <petev> the machine didn't shut down, just emc
[01:05:39] <petev> it was very strange
[01:05:40] <cradek> anything on stdout?
[01:05:48] <jmk-core2> the EMC GUI I mean
[01:05:52] <petev> no, I had run from an icon
[01:05:57] <jepler> if this message can be believed, something cut the connection between the GUI and the X server
[01:05:58] <jmk-core2> if you kill axis (or tkemc, etc) it will kill everything
[01:06:12] <petev> yeah, I understand that
[01:06:22] <cradek> petev: check .xsession-errors
[01:06:24] <petev> but it happened 3 times now and I can't see a reason for it
[01:06:38] <cradek> and if that doesn't help, watch stdout and reproduce it
[01:06:38] <petev> once it happened right when I inserted an USB stick
[01:06:49] <petev> cradek, ok, will do
[01:07:19] <jmk-core2> what distro and what emc? dapper & 2.1.4? or something else?
[01:07:42] <petev> I installed from the live ubuntu emc cd, then let it install all the updates
[01:08:00] <petev> emc is the head rev from a couple of days ago
[01:08:01] <jmk-core2> dapper and 2.1.4 then, I believe
[01:08:11] <jmk-core2> recent download of the livecd?
[01:08:17] <cradek> EMC2 - pre-2.2 CVS HEAD
[01:08:20] <petev> no, a month or so old
[01:08:32] <jmk-core2> oh, you aren't running the installed version
[01:08:38] <petev> no
[01:08:42] <cradek> he's said that several times now :-)
[01:08:46] <jmk-core2> sorry
[01:09:30] <cradek> petev: there was a recent crash in AXIS fixed - but it was only triggered by running long gcode programs containing lots of switches between arcs and lines
[01:09:50] <jepler> (did I back-port that fix)
[01:09:56] <petev> I need to put the mill on the net so I can check logs while I sit here and watch the game ;-)
[01:10:18] <petev> cradek, hmm, I wasn't running anything
[01:10:23] <petev> mostly checking hal config
[01:10:30] <petev> and testing IO, etc.
[01:10:56] <petev> probably had the show machine window open most of the time
[01:13:06] <jepler> try a different user interface? try without running 'boss', in case it's tied to the failure in some non-obvious way?
[01:13:21] <petev> ok, I'll try tkemc later
[01:14:08] <jepler> good luck
[01:14:58] <cradek> that is definitely strange - hope stdout shows something
[01:15:06] <petev> cradek, you pobably know the answer to this
[01:15:23] <petev> how do I get the USB touchscreen to be the same device all the time?
[01:15:28] <petev> I made a udev rule
[01:15:35] <petev> but it still doesn't work right
[01:15:45] <petev> it's like a crap shoot
[01:15:52] <petev> it's drivning me crazy
[01:16:04] <jepler> it appears as a /dev/ttyUSBx ?
[01:16:21] <petev> it's an event input device
[01:16:26] <jepler> oh
[01:16:43] <petev> I think its /dev/input/eventX
[01:16:48] <petev> or something like that
[01:17:01] <petev> I mapped it to /dev/input/touchscreen0
[01:17:19] <petev> but sometimes when I plug a mouse or usb stick in things get messed up
[01:17:23] <petev> can't figure out why
[01:18:16] <jepler> you can write a rule similar to this one, to create a symlink for a particular device:
[01:18:19] <jepler> # Create /dev/pilot symlink for Palm Pilots
[01:18:20] <jepler> KERNEL=="ttyUSB*", SYSFS{product}=="Palm Handheld*", \
[01:18:23] <jepler> SYMLINK+="pilot"
[01:19:50] <jepler> petev: are you using it as an X pointer device, or some other way? X seemed to automatically recognize pointer devices (but I only tried mice)
[01:20:25] <petev> yes, as X pointer
[01:20:27] <petev> always core
[01:20:29] <petev> SUBSYSTEMS=="usb", DRIVERS=="usbtouchscreen", KERNEL=="event*", SYMLINK+="input/touchscreen"
[01:20:37] <petev> that's what I put in for udev
[01:21:01] <jepler> should be DRIVER not DRIVERS
[01:21:11] <petev> hmm
[01:21:21] <jepler> same with SUBSYSTEMS
[01:21:26] <petev> it does make a link, but maybe thats the problem
[01:21:39] <petev> I'll try that
[01:21:59] <petev> thanks
[01:32:24] <cradek> ovl max: 11528
[01:33:59] <jmk-core2> cradek: nice
[01:37:30] <skunkworks> cradek: no reboot?
[01:37:50] <cradek> nope it works great now
[01:38:12] <jmk-core2> what do you have to do? new CPUs?
[01:38:40] <cradek> yes
[01:38:54] <cradek> one of the old ones had a wrong heat sink, and it may have been inadequately cooled
[01:39:05] <jmk-core2> ah
[01:39:06] <cradek> or one or more of the regulators may have been wrong
[01:39:09] <cradek> not sure
[01:39:14] <skunkworks> my partime helper came up to me and said "do you know the temp light is lit on the server?" I had just been down there doing a restore of a lost file. he said it was also beeping.
[01:39:14] <jmk-core2> you got new regs too?
[01:39:30] <cradek> yes a matched set pull from ebay
[01:39:54] <cradek> skunkworks: ouch
[01:40:23] <skunkworks> one of the processor fans wasn't spinning. looks like a weekend job.
[01:40:34] <cradek> if it lasts that long
[01:40:51] <cradek> hmm, feels like time to find some dinner
[01:40:53] <skunkworks> oh - I rigged a temp fan up ;)
[01:45:55] <petev> jepler, I need to do some homework, taking the 'S' s out kept X from starting, so my rule probably wasn't doing anything before
[01:47:14] <petev> cradek, grabbed the xsession-errors log on usb stick
[01:47:27] <petev> looks like there was some gdk error
[01:48:30] <petev> http://www.pastebin.ca/460454
[02:11:03] <jmk-core2> grrr
[02:11:36] <jmk-core2> far as I can tell, if I want to support the sensors on my mobo, I gotta patch the kernel
[02:18:49] <cradek> petev: I think that's the usual crap spewed by gnome apps, I don't see any emc error in there
[02:19:40] <cradek> jmk-core2: ick
[02:20:00] <jmk-core2> apparently all the drivers for lm-sensors are kernel modules
[02:20:21] <cradek> ah building a few new modules is easy
[02:20:44] <jmk-core2> well, the sensor detection said this:
[02:20:45] <jmk-core2> Probing for `Andigilog aSC7621'... Success!
[02:20:46] <jmk-core2> (confidence 5, driver `to-be-written')
[02:21:11] <jmk-core2> and this:
[02:21:12] <jmk-core2> Intel Core family thermal sensor... Success!
[02:21:12] <jmk-core2> (driver `coretemp')
[02:21:44] <jmk-core2> jmkasunich@solo:~/emcdev/emc2head/src$ modprobe coretemp
[02:21:44] <jmk-core2> FATAL: Module coretemp not found.
[02:21:57] <jmk-core2> and of course, "to-be-written" is a bit of a problem...
[02:22:14] <cradek> Success!
[02:22:18] <cradek> !!
[02:22:50] <jmk-core2> successfully found such a chip
[02:22:58] <jmk-core2> doesn't mean theres a driver for it
[02:23:20] <jmk-core2> (I didn't post the hundred lines of "probing for foo... No"
[02:27:11] <jmk-core2> poo: http://ubuntuforums.org/showthread.php?p=2210384
[02:28:16] <cradek> welcome to the brave new world
[02:29:07] <cradek> at least the ten tries it'll take you to get it built right will go fast
[02:31:05] <cradek> http://www.debianhelp.org/node/5156
[02:31:35] <cradek> hopefully it can make it into 2.6.22, then userland support should
[02:31:36] <cradek> follow over time.
[02:33:18] <jmk-core2> thanks for the link
[02:33:21] <jmk-core2> bookmarked
[02:33:26] <jmk-core2> more than I want to tackle right now
[02:33:43] <jmk-core2> I let the sensor thing distract me from the other issue
[02:34:04] <jmk-core2> why did last nights 1:06 and 0:36 compiles become todays 1:30 and 0:47?
[02:34:28] <cradek> why not get last night's cvs
[02:34:41] <jmk-core2> I'm doing that right now
[02:34:43] <cradek> if the depends were wrong (they were), it probably didn't build everything
[02:42:29] <jmk-core2> that was it - last nights checkout compiled in 1:01 and 0:35 (regular and -j)
[02:43:56] <jmk-core2> what had me worried was the 1.6GHz at idle thing - the ratio between the build times was about the same as the ratio between clocks, so I started thinking it had switched to the lower clock and was stuck there
[03:26:24] <cradek> yay, my machine is fixed, it has been compiling for 2 hours
[03:26:34] <cradek> (previously shut off after 5 minutes of hard work)
[03:30:25] <cradek> jmk-core2: are your 4x second builds from clean and for realtime?
[03:35:55] <jmk-core2> from clean for sim
[03:36:04] <jmk-core2> (since I don't have a RT kernel)
[03:37:01] <cradek> ah, I bet that's a lot faster, since it avoids the evil that is kbuild
[03:37:08] <jmk-core2> heh
[03:37:29] <cradek> even using my 3 machines I can only get down to 1:30 but for a realtime build
[03:37:28] <jmk-core2> ISTR there wasn't a big difference between RT and sim builds on the other system
[03:37:29] <jepler> I think that's about what I got on my dual core amd system, also for sim
[03:38:31] <jmkasunich> cradek: how long does a sim build take with 3 machines?
[03:38:41] <cradek> I'm going to try that
[03:38:57] <jmkasunich> just found this in my notes:
[03:38:57] <jmkasunich> compile EMC2 RT
[03:38:57] <jmkasunich> 7:15/3:45/0:47
[03:38:57] <jmkasunich> compile EMC2 sim
[03:38:57] <jmkasunich> 6:03/3:08/0:44
[03:39:12] <jmkasunich> thats the old machine, with the VMs running but idle
[03:39:32] <jmkasunich> real/user/sys
[03:41:22] <cradek> distcc may not be doing much for me - that machine may be on 10Mbit
[03:56:15] <cradek> about 1:25
[03:56:37] <jmk-core2> so proportionally sim is just a little faster
[03:56:50] <cradek> yeah, not much
[03:57:14] <cradek> so in three (nice, I think) machines i have less processing power than your $400 bought?
[03:57:30] <jmk-core2> looks that way
[03:57:36] <jmk-core2> more than 400 tho
[03:58:02] <jmk-core2> I bought a 199 mobo, not a 119 one (doubt it makes much difference)
[03:58:37] <jmk-core2> 200 mobo, 235 cpu, 115 IIRC ram, and 169 for the 10KRPM disk
[03:58:55] <jmk-core2> more like $700-750
[03:59:16] <jmk-core2> still, you can get a log of bang for the buck right now
[03:59:24] <jmk-core2> lot even
[03:59:35] <cradek> yes
[04:01:36] <cradek> I'd need a new case/ps too
[04:01:46] <cradek> don't think I want to spend that to save a minute here and there
[04:02:01] <cradek> besides I can always compile at work where I have many machines like yours :-)
[04:02:29] <jmk-core2> the best deal would be a 119 mobo and a 230 CPU, assuming you have suitable ram (DDR2)
[04:02:35] <jmk-core2> use existing disks, etc
[04:02:53] <cradek> I'd need video too, but they're cheap
[04:02:58] <cradek> say how do you like yours
[04:03:04] <jmk-core2> the video?
[04:03:08] <cradek> yes
[04:03:23] <jmk-core2> haven't loaded a real driver for it, its just as slow with glxgears as the old machine
[04:03:26] <cradek> ah
[04:03:35] <jmk-core2> haven't tried the screensaver version
[04:03:41] <jmk-core2> where does that live?
[04:03:50] <cradek> using the stock ubuntu kernels, it's easy to enable the bianry driver
[04:03:56] <cradek> /usr/lib/xscreensaver/gears
[04:04:23] <jmk-core2> 31 fps
[04:04:33] <jmk-core2> about double what I got on the other machine
[04:04:48] <jmk-core2> how easy is it to do the bineray driver?
[04:05:01] <cradek> do you have linux-restricted-modules-`uname -r` installed already?
[04:06:07] <cradek> https://help.ubuntu.com/community/BinaryDriverHowto/Nvidia
[04:06:40] <jmk-core2> yeah, got the modules
[04:10:42] <jmk-core2> my card isn't on this list: http://http.download.nvidia.com/XFree86/Linux-x86/1.0-8762/README/appendix-a.html
[04:10:47] <jmk-core2> (chipset, not card)
[04:10:51] <jmk-core2> I have a 7100GS
[04:11:06] <cradek> is the ubuntu version 8762?
[04:12:12] <jmk-core2> ubuntu version?
[04:12:42] <cradek> it's 8776
[04:12:59] <jmk-core2> oh, that list is for 8762
[04:13:22] <cradek> check 8776
[04:14:21] <jmk-core2> nope
[04:14:28] <jmk-core2> skips from 6800 to 7300
[04:14:31] <jmk-core2> dammit
[04:14:39] <jmk-core2> the 7300 is only another $20
[04:15:21] <jmk-core2> well, I already installed it, can't hurt to try ;-)
[04:15:27] <cradek> yeah it'll probably work
[04:15:30] <jmk-core2> restarting
[04:18:07] <cradek> uh-oh
[04:18:12] <jmk-solo> ?
[04:18:24] <cradek> you have a different name so I figured it didn't work
[04:18:31] <jmk-solo> oh
[04:18:42] <jmk-solo> the box is called solo (cause thats the case its in)
[04:18:51] <cradek> so it works?
[04:18:56] <jmk-solo> jymm was giving me a hard time yestaerday and I changed the name
[04:18:59] <jmk-solo> jmkasunich@solo:~$ /usr/lib/xscreensaver/gears -fps
[04:18:59] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:18:59] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:18:59] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:18:59] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:18:59] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:02] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:03] <cradek> ouch
[04:19:04] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:05] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:07] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:09] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:11] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:13] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:15] <cradek> ok we get it
[04:19:16] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:18] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:19] <jmk-solo> Xlib: extension "GLX" missing on display ":0.0".
[04:19:21] <jmk-solo> gears: display ":0.0" does not support the GLX extension.
[04:19:23] <jmk-solo> jmkasunich@solo:~$
[04:19:27] <jmk-solo> X works, but no accel, and the screensaver worked before
[04:19:38] <jmk-solo> so back to SW for me I think
[04:19:48] <cradek> check your log
[04:19:48] <jmk-solo> shoulda spent the $20
[04:19:58] <cradek> Xorg.0.log
[04:20:01] <cradek> see what it's whining about
[04:20:45] <jmk-solo> last couple lines:
[04:20:46] <jmk-solo> (II) XINPUT: Adding extended input device "Generic Keyboard" (type: KEYBOARD)
[04:20:47] <jmk-solo> (II) Configured Mouse: ps2EnableDataReporting: succeeded
[04:20:47] <jmk-solo> AUDIT: Fri Apr 27 00:17:08 2007: 30359 X: client 2 rejected from local host
[04:21:15] <jmk-solo> a bit earlier:
[04:21:15] <jmk-solo> (EE) Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
[04:21:15] <jmk-solo> error opening security policy file /etc/X11/xserver/SecurityPolicy
[04:22:13] <jmk-solo> thats the only EE entry
[04:22:26] <cradek> does xorg.conf have Driver "nvidia"?
[04:22:48] <cradek> (II) NVIDIA X Driver 1.0-8776 Mon Oct 16 21:58:46 PDT 2006
[04:22:48] <cradek> (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[04:23:11] <jmk-solo> nv in xorg.conf
[04:23:15] <cradek> doh
[04:23:16] <cradek> change it
[04:23:25] <cradek> you missed a step?
[04:23:47] <jmk-solo> I did this: sudo nvidia-glx-config enable
[04:23:55] <jmk-solo> I don't think I missed anything
[04:23:59] <cradek> that must not work
[04:24:57] <jmk-solo> like so?
[04:24:58] <jmk-solo> Section "Device"
[04:24:58] <jmk-solo> Identifier"NVIDIA Corporation NVIDIA Default Card"
[04:24:58] <jmk-solo> Driver"nvidia"
[04:24:58] <jmk-solo> BusID"PCI:1:0:0"
[04:24:58] <jmk-solo> EndSection
[04:25:05] <cradek> yes
[04:25:07] <cradek> in Module:
[04:25:10] <jmk-solo> trying again
[04:25:12] <cradek> make sure you have Load "glx"
[04:25:15] <cradek> ack
[04:28:10] <jmk-solo> couldn't find the driver
[04:28:41] <jmk-solo> (II) Setting vga for screen 0.
[04:28:41] <jmk-solo> (**) NVIDIA(0): Depth 24, (--) framebuffer bpp 32
[04:28:41] <jmk-solo> (==) NVIDIA(0): RGB weight 888
[04:28:41] <jmk-solo> (==) NVIDIA(0): Default visual is TrueColor
[04:28:41] <jmk-solo> (==) NVIDIA(0): Using gamma correction (1.0, 1.0, 1.0)
[04:28:42] <jmk-solo> (**) NVIDIA(0): Enabling RENDER acceleration
[04:28:44] <jmk-solo> (EE) NVIDIA(0): Failed to load the NVIDIA kernel module!
[04:28:48] <jmk-solo> (EE) NVIDIA(0): *** Aborting ***
[04:28:51] <jmk-solo> (II) UnloadModule: "nvidia"
[04:28:53] <jmk-solo> (II) UnloadModule: "ramdac"
[04:28:55] <jmk-solo> (II) UnloadModule: "fb"
[04:28:57] <jmk-solo> (EE) Screen(s) found, but none have a usable configuration.
[04:29:11] <cradek> did you reboot after installing the kernel stuff?
[04:29:19] <jmk-solo> no, just restarted X
[04:29:22] <jmk-solo> do I have to?
[04:29:42] <cradek> sudo /etc/init.d/linux-restricted-modules-common start
[04:29:47] <cradek> (maybe?)
[04:31:04] <jmk-solo> jmkasunich@solo:~$ sudo /etc/init.d/linux-restricted-modules-common start
[04:31:05] <jmk-solo> * Preparing restricted drivers...
[04:31:12] <jmk-solo> (it said OK)
[04:31:14] <jmk-solo> then...
[04:31:22] <jmk-solo> jmkasunich@solo:~$ modprobe -n -v nvidia
[04:31:22] <jmk-solo> FATAL: Module nvidia not found.
[04:31:22] <jmk-solo> jmkasunich@solo:~$
[04:31:30] <jmk-solo> -n is dry run
[04:31:35] <cradek> hm
[04:31:44] <cradek> just reboot?
[04:31:50] <cradek> it does some crazy runtime linking
[04:31:58] <jmk-solo> ok
[04:33:18] <jmkasunich> takes forever to reboot
[04:33:33] <jmkasunich> the bios spends about 2 minutes figuring out the CD drive
[04:34:45] <jmkasunich> http://forums.legitreviews.com/about9191.html <-- that problem
[04:35:15] <cradek> http://pastebin.ca/460608
[04:35:24] <cradek> there are some module requirements to get right
[04:36:46] <tomp> i got same prob on an ubuntu system with an ancient IBM 4 cd changer... takes a few minutes before grub appears, unplug it & its fast
[04:38:08] <jmk-solo> no joy
[04:38:09] <jmk-solo> can't find it
[04:40:35] <jmk-solo> /lib/modules/2.6.15-28-686/kernel/drivers/video/nvidia/nvidiafb.ko
[04:40:49] <jmk-solo> think it would work better if I called it nvidiafb?
[04:41:09] <jmk-solo> find found these:
[04:41:10] <jmk-solo> /lib/modules/2.6.15-28-686/kernel/drivers/char/agp/nvidia-agp.ko
[04:41:10] <jmk-solo> /lib/modules/2.6.15-28-686/kernel/drivers/char/nvram.ko
[04:41:10] <jmk-solo> /lib/modules/2.6.15-28-686/kernel/drivers/video/nvidia/nvidiafb.ko
[04:41:20] <jmk-solo> (plus others for different kernels)
[04:42:53] <jmk-solo> no, I don't think I want nvidiafb
[04:44:02] <jmk-solo> tap, tap,.... is this thing on?
[04:44:50] <cradek> none of those are it
[04:45:28] <jmk-solo> from a thread on some forum: "You can't install Nvidia's driver if your kernel supports nvidiafb."
[04:45:50] <cradek> /lib/modules/2.6.15-28-686/volatile/nvidia.ko
[04:46:29] <cradek> $ modinfo nvidia|head
[04:46:30] <cradek> filename: /lib/modules/2.6.15-28-686/volatile/nvidia.ko
[04:46:30] <cradek> license: NVIDIA
[04:46:57] <jmk-solo> I have that directory, but no nvidia stuff in it
[04:47:14] <jmk-solo> fc<severalthings>,ko
[04:47:24] <jmk-solo> dot ko
[04:47:26] <cradek> $ dpkg -l linux-restricted-modules-`uname -r`
[04:47:34] <cradek> does this show status ii?
[04:47:54] <jmk-solo> yes
[04:48:58] <jmk-solo> what is the current dapper kernel? is it 2.6.15-28-686, or 2.6.20ish?
[04:48:59] <cradek> cat /etc/defaultlinux-restricted-modules-common
[04:49:03] <cradek> 15
[04:49:12] <jmk-solo> ok
[04:49:19] <jmk-solo> dunno why I thought we were near 20
[04:49:20] <cradek> does it have DISABLED_MODULES="... nv ..."?
[04:49:58] <jmk-solo> yes
[04:50:07] <cradek> take out the nv and try again
[04:50:49] <cradek> are you cursing proprietary drivers yet? I always do
[04:50:54] <jmk-solo> reboot or just restart x?
[04:51:05] <cradek> /etc/init.d/linux-restricted-modules-common restart
[04:51:08] <cradek> then modinfo nvidia
[04:51:41] <jmk-solo> jmkasunich@solo:/lib/modules/2.6.15-28-686/volatile$ sudo /etc/init.d/linux-restricted-modules-common restart
[04:51:42] <jmk-solo> jmkasunich@solo:/lib/modules/2.6.15-28-686/volatile$ modinfo nvidia modinfo: could not find module nvidia
[04:52:01] <cradek> and it's still not in that volatile directory?
[04:52:11] <jmk-solo> nope
[04:52:14] <cradek> grr
[04:52:15] <jmk-solo> its like it was never installed
[04:52:27] <cradek> well it gets built (linked) every time you boot
[04:52:41] <cradek> maybe you have to try rebooting again
[04:52:50] <cradek> I don't know 100% how this all works
[04:53:15] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia
[04:53:15] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/nv-kernel.o
[04:53:15] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/nv.o
[04:53:17] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/nv-vm.o
[04:53:19] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/os-agp.o
[04:53:21] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/os-interface.o
[04:53:22] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/os-registry.o
[04:53:26] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/nv-i2c.o
[04:53:28] <jmk-solo> /lib/linux-restricted-modules/2.6.15-28-686/nvidia/nvidia.mod.o
[04:53:31] <jmk-solo> thats what the package installs, according to synaptic
[04:53:38] <jmk-solo> the mod.o becomes the kernel module maybe?
[04:53:42] <cradek> yeah that crap gets linked somehow into nvidia.ko
[04:54:05] <jmk-solo> rebooting
[04:55:32] <jmkasunich> crap, almost 1am
[04:55:49] <jmkasunich> I wanted to go to be early tonight (after being up till 5 yesterday)
[04:55:56] <jmkasunich> if this doesn't work, I'm gonna call it a night
[04:59:16] <jmk-solo> well, I'm getting good at using pico to change 'nvidia' to 'nv'
[04:59:29] <cradek> hmm
[04:59:41] <cradek> do you have the module yet?
[05:00:09] <jmk-solo> yes!
[05:00:15] <cradek> yay
[05:00:16] <jmk-solo> there is a nvidia.ko in volatile
[05:00:27] <cradek> what's the next complaint then
[05:01:05] <jmk-solo> I want xorg.0.log.old, right? the current one is from the restart with nv...
[05:01:09] <cradek> yes
[05:01:36] <jmk-solo> (EE) NVIDIA(0): Failed to load the NVIDIA kernel module!
[05:01:36] <jmk-solo> (EE) NVIDIA(0): *** Aborting ***
[05:01:36] <jmk-solo> (II) UnloadModule: "nvidia"
[05:01:36] <jmk-solo> (II) UnloadModule: "ramdac"
[05:01:36] <jmk-solo> (II) UnloadModule: "fb"
[05:01:38] <jmk-solo> (EE) Screen(s) found, but none have a usable configuration.
[05:01:52] <cradek> modinfo nvidia works, right?
[05:02:05] <jmk-solo> no
[05:02:09] <cradek> wtf
[05:02:11] <jmk-solo> modinfo: could not find module nvidia
[05:03:50] <jmk-solo> modinfo knows to recurse?
[05:04:14] <cradek> I think it uses the same scheme nodprobe does, whatever that is
[05:04:32] <jmk-solo> neither one finds it (I tried modprobe -n -v)
[05:05:15] <cradek> grep for it in /etc/modprobe.d
[05:05:19] <cradek> it could be blacklisted?
[05:06:23] <jmk-solo> jmkasunich@solo:/etc/modprobe.d$ grep nvidia *
[05:06:23] <jmk-solo> aliases:alias char-major-195-* nvidia
[05:06:23] <jmk-solo> blacklist-framebuffer:blacklist nvidiafb
[05:06:23] <jmk-solo> nvidia-kernel-nkc:alias char-major-195* nvidia
[05:06:23] <jmk-solo> jmkasunich@solo:/etc/modprobe.d$
[05:06:31] <jmk-solo> the fb is, not the retular
[05:06:53] <jmk-solo> regular
[05:07:12] <cradek> I'm stumped and tired, so let's mess with it later
[05:07:36] <cradek> (it's going to suck if it's not supported after all this)
[05:07:46] <jmk-solo> yeah
[05:07:50] <jmk-solo> goodnight
[05:07:54] <cradek> night
[15:36:08] <skunkworks> didn't we just discuss that?
[15:36:20] <jepler> yeah seems like
[15:49:49] <alex_jon1> * alex_jon1 shrugs
[15:49:53] <alex_jon1> alex_jon1 is now known as alex_joni
[15:49:58] <jepler> hi alex_joni
[15:50:10] <alex_joni> hi
[15:50:33] <alex_joni> yumm.. just had some great pizza
[19:34:04] <cradek> lerman: "O100 call [#1] (comment) [#2] [#3]" has a bug - #2 and #3 are not passed to the subroutine