From: Sam Moore Date: Wed, 25 Sep 2013 17:00:10 +0000 (+0800) Subject: Automatic commit of irc logs X-Git-Url: https://git.ucc.asn.au/?a=commitdiff_plain;h=b7b12523983c5d6fef8d0e6fb6cae9f1b4e7d6f1;p=matches%2FMCTX3420.git Automatic commit of irc logs --- diff --git a/irc/log b/irc/log index 03ae27b..c9821f8 100644 --- a/irc/log +++ b/irc/log @@ -3183,3 +3183,261 @@ 19:17 < jtanx> i'll stick with debian (but do the same thing) 21:07 -!- jtanx [~asfa@220-253-203-242.dyn.iinet.net.au] has quit ["ChatZilla 0.9.90.1 [Firefox 24.0/20130910160258]"] 21:34 -!- MctxBot [~twang@220-253-203-242.dyn.iinet.net.au] has quit [Ping timeout] +--- Day changed Wed Sep 25 2013 +08:41 -!- MctxBot [~twang@220-253-203-242.dyn.iinet.net.au] has joined #mctxuwa_softdev +11:31 -!- jtanx [~asfa@130.95.54.13] has joined #mctxuwa_softdev +12:15 < jtanx> I think I know why we were having issues with pwm yesterday +12:16 < jtanx> if you do this command +12:16 < jtanx> echo bone_pwm_P8_13 > /sys/devices/bone_capemgr.8/slots +12:16 < jtanx> you make it unavailable from /sys/class/pwm +12:16 < jtanx> so in the run.sh script it was exporting all the pwm devices via the first method +12:16 < jtanx> and then it becomes unavailable via sysfs +12:17 < jtanx> anyway... I tried booting from the rcn image +12:17 < jtanx> it comes with pwm enabled already +12:17 < jtanx> and via that capemgr I got pwm to work +12:17 < jtanx> I don't know what /sys/class/pwm/pwm0 corresponds to (which pin) +12:19 < jtanx> the electronics teams' bbb wasn't done properly when we tried to upgrade the kernel +12:19 < jtanx> probably something to do with the device tree stuff +12:19 < jtanx> so I flashed it with the rcn image (which runs 3.8.13-bone26 +12:20 < jtanx> (demo image from here) +12:20 < jtanx> elinux.org/BeagleBoardDebian +12:47 -!- jtanx [~asfa@130.95.54.13] has quit [Ping timeout] +13:09 -!- jtanx_ [~asfa@130.95.54.13] has joined #mctxuwa_softdev +13:09 -!- jtanx_ is now known as jtanx +13:16 < jtanx> oh +13:16 < jtanx> so it now works +13:16 < jtanx> echo bone_pwm_P9_22 > slots +13:16 < jtanx> if I do that line for pwm0 +13:26 < jtanx> oh right +13:26 < jtanx> echo bone_pwm_P9_21 >slots +13:26 < jtanx> for pwm1 +13:30 < jtanx> ahhhhhh +13:30 < jtanx> if you comment out the line +13:30 < jtanx> modprobe pwm_test +13:30 < jtanx> from run.sh +13:30 < jtanx> it works +13:43 < jtanx> geeze kernel 3.8 has issues with usb hotplugging +13:43 < jtanx> https://groups.google.com/forum/?fromgroups#!searchin/beagleboard/usb/beagleboard/8aalvyWwaig/MUXAPuMTSOYJ +13:43 < jtanx> which explains why we're having issues with cameras +13:43 < jtanx> (partly at least) +13:47 < jtanx> and now pwms not working again +13:48 < jtanx> via sysfs +13:50 < jtanx> oh +13:50 < jtanx> I know why +13:51 < jtanx> you have to export it /sys/class/pwm +13:51 < jtanx> first +13:51 < jtanx> *before* you do stuff like echo bone_pwm_P9_21 >slots +13:52 < jtanx> yep +13:53 < jtanx> so the order is: echo 0/1 > /sys/class/pwm/export +13:53 < jtanx> then do that other stuff +13:57 < jtanx> egh +13:57 < jtanx> finnicky +13:57 < jtanx> ok I have to stop now +14:14 -!- jtanx [~asfa@130.95.54.13] has quit [Ping timeout] +14:15 -!- jtanx [~asfa@130.95.218.72] has joined #mctxuwa_softdev +15:46 -!- jtanx [~asfa@130.95.218.72] has quit [Ping timeout] +16:03 -!- jtanx [~asfa@130.95.218.72] has joined #mctxuwa_softdev +16:04 < jtanx> well that was an interesting experience +16:04 < jtanx> it's most reliable when you work directly with /sys/devices/ocp2.helper/PWM9_22* +16:05 < jtanx> I think if you echo am33xx_pwm to the slots thing when it's already loaded +16:05 < jtanx> weird shit can happen +16:05 < jtanx> too +16:07 < jtanx> setting the period via sysfs (eg /sys/class/pwm) didn't work most of the time either +16:07 < jtanx> you could change duty but not period +16:07 < jtanx> although I swear I had it working at one point +16:07 < jtanx> via the other way I think it works ok +16:08 < jtanx> oh yeah, and I was doing this using the demo image from http://elinux.org/BeagleBoardDebian +16:09 < jtanx> the electrical group's one has been reflashed with that version as well +16:09 < jtanx> (for ours I worked off my sd card) +16:10 < jtanx> that image also enables the ethernet-over-usb +16:36 < jtanx> I think we have to be careful which pins we export/enable +16:37 < jtanx> https://github.com/CircuitCo/BeagleBone-Black/blob/master/BBB_SRM.pdf?raw=true +16:37 < jtanx> pages 80-82 +16:37 < jtanx> the pins have different meanings based on what mode they're in +16:41 -!- jtanx [~asfa@130.95.218.72] has quit ["ChatZilla 0.9.90.1 [Firefox 24.0/20130910160258]"] +17:59 -!- jtanx [~asfa@220-253-203-242.dyn.iinet.net.au] has joined #mctxuwa_softdev +18:05 < jtanx> ...so I brought the BBB home, even though I should be studying for 2402 :S +18:05 < jtanx> the microphone came in today too +18:19 < jtanx> ok +18:19 < jtanx> so these documents: https://github.com/derekmolloy/boneDeviceTree/tree/master/docs +18:19 < jtanx> describes what the pins are used for by default +19:00 < sam_moore> Ah... they already got the microphone +19:00 < sam_moore> Welp. Guess we're stuck with it. +19:00 < sam_moore> So... we can record <50Hz sounds reliably, maybe +19:00 < sam_moore> How useful +19:01 < sam_moore> Have I been missing out on an email stream where the sensors team actually clears things with us? +19:04 < jtanx> not that I know of +19:04 < jtanx> haha +19:04 < jtanx> in other news +19:04 < jtanx> the sensors team ordered a pressure sensor with no mount +19:04 < jtanx> adrian had a spat because it'd cost something like $200 to make the mount +19:05 < jtanx> for a $10 part +19:05 < jtanx> so he said, order it from online , I don't care if it's from overseas +19:05 < sam_moore> Oh boy +19:06 < sam_moore> If there's an issue with the camera and/or microphone they'll blame us +19:06 < jtanx> yeah +19:06 < jtanx> about that camera +19:06 < sam_moore> Oh dear... +19:06 < sam_moore> Go ahead? +19:06 < jtanx> still couldn't get it to work today +19:06 < sam_moore> God dammit +19:06 < jtanx> although I didn't spend much time on it +19:06 < jtanx> I got pwm to work +19:06 < jtanx> mostly +19:06 < sam_moore> I thought it might be something like adding the user to the "video" group +19:06 < sam_moore> That's good! +19:07 < sam_moore> What was happening? +19:07 < jtanx> yeah, the problem is it doesn't show up at all (the camera) +19:07 < sam_moore> Hmm +19:07 < jtanx> and partly because 3.8 has an issue with usb hotplugging +19:07 < sam_moore> Haha +19:07 < jtanx> about pwm +19:07 < jtanx> it seems that the sysfs method is not so reliable +19:07 < jtanx> you can get it to work +19:07 < jtanx> you have to export those first +19:07 < jtanx> so echo 0 > /sys/class/pwm/export +19:08 < jtanx> then (and only then) +19:08 < jtanx> can you do +19:08 < jtanx> that echo to the slots +19:08 < jtanx> for those pins +19:08 < jtanx> then it seems to be happy +19:08 < jtanx> if you echo am33xx_pwm to the slots when it's already enabled +19:08 < jtanx> that also screws things up +19:08 < sam_moore> Ok +19:09 < sam_moore> Thanks for working that out +19:09 < jtanx> yeah +19:09 < sam_moore> If you want to change from sysfs to the other method that's fine +19:09 < sam_moore> But sysfs was much simpler to code +19:09 < jtanx> should have spent that time studying for mech2402 though :P +19:09 < sam_moore> Because you just sprintf an integer to the path +19:09 < jtanx> yeah +19:09 < jtanx> witht he other way it's all that dynamic path crap +19:09 < sam_moore> Rather than keeping track of "bone_pwm_test_P9_22.15.arbitrary_string" crap +19:09 < sam_moore> Exactly :P +19:09 < jtanx> but +19:10 < jtanx> you can enable pwm and analogue on boot +19:10 < jtanx> if I can find the link +19:10 < sam_moore> Sure, if that's easy +19:10 < sam_moore> I figured if we put them in the /etc/init.d script that'd be fine too +19:10 < sam_moore> Actually... maybe we should put it in the /etc/init.d script +19:11 < jtanx> oh yeah +19:11 < jtanx> and the demo image from that rcn image +19:11 < sam_moore> Because if someone gets a different beaglebone then they'd have to reenable it on boot +19:11 < jtanx> is better than screwing around with recompiling kernels +19:11 < sam_moore> Can you give a link? +19:11 < jtanx> I think it's the first image that you had originally +19:11 < jtanx> http://elinux.org/BeagleBoardDebian +19:12 < jtanx> there's this script in /boot +19:12 < sam_moore> Oh +19:12 < jtanx> that allows you to copy the sd card to flash +19:12 < jtanx> it also enables the usb over ethernet +19:12 < sam_moore> Oh right, the image I downloaded before we used yours +19:12 < sam_moore> Cool +19:12 < jtanx> yeah +19:12 < jtanx> I flashed electronics' one with that +19:12 < sam_moore> Does PWM and stuff work on it? +19:13 < jtanx> probably +19:13 < jtanx> I was using the same image +19:13 < jtanx> on ours +19:13 < jtanx> you run this script and it copies exatly what's on the sd card to the internal flash +19:13 < jtanx> resizes the partition as necessary +19:13 < jtanx> http://digital-drive.com/?p=146 +19:13 < jtanx> that page shows how to enable on boot +19:13 < jtanx> it's just a change to uEnv.txt in the boot partition +19:18 < sam_moore> Good work +19:19 < sam_moore> While I remember, for multiple logins and crap... can you just try to login as a local user account? +19:19 < sam_moore> Then we could make a wrapper around adduser and deluser for the "administrator" account +19:19 < jtanx> wow +19:20 < jtanx> I don't know +19:20 < sam_moore> I was just thinking +19:20 < sam_moore> Linux has a user account system already +19:20 < jtanx> yep, but is it a good idea to be making ~300 on a BBB? +19:21 < sam_moore> Well... putting LDAP on the BBB probably won't be less intense +19:21 < sam_moore> I know it's called "Lightweight" +19:21 < sam_moore> But that's in comparison to "DAP" +19:21 < jtanx> well to be perfectly honest, adrian is asking way too much +19:21 < sam_moore> Which was designed in the 1980s by a telephone directory company and used the original OSI networking model +19:21 < jtanx> you simply can't support a 300-odd user base on something like a BBB +19:21 < sam_moore> Yeah +19:22 < sam_moore> But maybe something like 30 users would work? +19:22 < jtanx> yeah +19:22 < jtanx> let's just keep it at that limit +19:22 < sam_moore> Another thing regarding the crazy requirements... +19:22 < sam_moore> If we have multiple Beaglebones running FastCGI +19:23 < sam_moore> We can design our GUI so that it has links to the appropriate Beaglebone for each function +19:23 < sam_moore> I don't think we actually need to do anything in nginx or the Beaglebone software +19:24 < jtanx> hmm +19:24 < sam_moore> At least in terms of displaying sensor data +19:24 < sam_moore> For actuator control, we would need to introduce networking between individual beaglebones +19:24 < jtanx> it actually depends on what he means by 'extensible' and/or distributed +19:24 < jtanx> like +19:24 < jtanx> you could say this BBB is for this experiement +19:25 < jtanx> this other BBB is for this other experiment +19:25 < sam_moore> But quite frankly you'd be mad to trust a distributed system with networking delays to coordinate control over hardware +19:25 < jtanx> well yeah +19:25 < sam_moore> Well at least something like this where we care about safety +19:25 < sam_moore> But if you keep the actual control over hardware independent and on seperate devices +19:25 < jtanx> but I mean +19:26 < jtanx> wait +19:26 < jtanx> if we interpret it as meaning +19:26 < jtanx> that each BBB runs an instance of the software +19:26 < jtanx> then they would still be separate +19:26 < jtanx> as in each BBB controls one 'experiment' +19:26 < jtanx> you customise each BBB based on the experiment that needs to be done +19:26 < sam_moore> Yes, that would work +19:26 < sam_moore> Yep +19:26 < jtanx> then there's no interaction between BBBS +19:27 < jtanx> the only thing is you have some sort of control at the front +19:27 < jtanx> that determines which BBB you connect to +19:27 < sam_moore> Yes, if there's interaction between BBBs it gets problematic +19:27 < jtanx> yeah +19:27 < sam_moore> Yes, you have one BBB which gives the user the "main menu" part of the GUI +19:27 < jtanx> I reckon that's a stupid requirement to ask +19:27 < jtanx> yeah +19:27 < sam_moore> Then the others just have customised GUIs or whatever +19:28 < jtanx> once you have to get them to talk to each other, you're then having to try and invent a whole new protocol +19:28 < jtanx> for that +19:28 < sam_moore> Yeah, and it depends on exactly what the hardware is +19:29 < sam_moore> You might be able to hack it onto the web protocol (eg: BeagleBone #1 sends http://beaglebone2/api/actuators?id=X?set=Y) +19:29 < sam_moore> But... let's not think about that +19:30 < sam_moore> It's clearly beyond the scope of this project +19:31 < sam_moore> So, after all that, I reckon if we use snoopy for ADC/GPIO/PWM and spike for the dilatometer then that would be cool (probably not actually necessary though) +19:31 < jtanx> yeah +19:32 < sam_moore> The dilatomter... it's going to cause headaches if Kieren really wants to "return" an array of points +19:33 < sam_moore> If the goal is to provide the user with a demonstration of what the dilatometer is doing, then you can just edit an image +19:33 < sam_moore> If the goal is to provide more data... I don't see the point really +19:34 < sam_moore> It's going to be the same sort of distribution every time +19:34 < sam_moore> Realistically all anyone would do is average it +19:34 < sam_moore> Maybe take a standard deviation +19:36 < jtanx> I really don't know why a dilatometer's even needed +19:36 < sam_moore> Educational reasons? :P +19:37 < jtanx> haha sure +19:37 < sam_moore> Anyway, hopefully Callum will deal with the dilatometer stuff +19:37 < sam_moore> The interferometer code is a good starting point +19:39 < jtanx> Yeah +19:39 < jtanx> hopefully +19:42 < sam_moore> We should arrange some meetings next week +19:42 < sam_moore> Also I'd like to see more of the other group members committing to git and talking in this channel +19:45 < sam_moore> People are missing a lot of design decisions here :S +19:45 < jtanx> Yeah +19:51 < jtanx> Ok +19:51 < jtanx> so I made a LUT from pin number on the board to GPIO pin number +19:52 < jtanx> so if you wanted to use P8_13 +19:52 < jtanx> you can use the lut to figure out what gpio number that corresponds to +19:53 < jtanx> we should probably restrict which pins can be used +19:53 < jtanx> because quite a few are reserved +19:53 < sam_moore> Sure +19:53 < sam_moore> Remove the #defines in bbb_pin_defines.h ? +19:54 < sam_moore> Don't export those pins in pin_test.c +19:54 < sam_moore> It is only really for testing anyway +19:54 < jtanx> yeah +19:55 < sam_moore> Although... I predict if we leave it in the software, *someone* at some point will try and control hardware directly through it :P +19:55 < sam_moore> For all the educational stuff it's nice though +19:56 < sam_moore> Oh, we could have an image of the pinout diagram +19:56 < sam_moore> And when someone clicks on a part of the image they get to control that pin +19:57 < sam_moore> Anyway... I really should study for MECH2402 or I will fail it +19:57 < sam_moore> So bye +19:57 < jtanx> yeah +19:57 < jtanx> bye +20:50 -!- jtanx [~asfa@220-253-203-242.dyn.iinet.net.au] has quit ["ChatZilla 0.9.90.1 [Firefox 24.0/20130910160258]"] +21:50 -!- MctxBot [~twang@220-253-203-242.dyn.iinet.net.au] has quit [Ping timeout]