**** BEGIN LOGGING AT Mon Mar 05 02:59:58 2012 Mar 05 04:03:39 4! Mar 05 04:04:22 3!! Mar 05 04:04:43 2!?? Mar 05 04:34:42 1!? Mar 05 04:35:22 lift off? Mar 05 05:14:05 can anybody help me getting started with the beagle board? Mar 05 05:15:52 i need to use it in a robot controlled via wifi Mar 05 05:25:18 can anybody help me getting started with the beagle board? i need to use it in a robot controlled via wifi Mar 05 05:26:21 follow beagle board homepage Mar 05 05:31:29 i wanted to know if development can be done completely on the serial link to a host computer...without attaching ny monitors or other io devices to the beaglebord Mar 05 05:34:16 jungleboy: yes of course Mar 05 05:37:09 also...i want to stream videos from my robot via the wifi kink to a laptop......what do i need to do for this? Mar 05 05:37:23 wifi *link Mar 05 05:38:54 well a camera and a wifi link Mar 05 05:43:09 how would i program this? Mar 05 05:43:19 is opencv a good choice? Mar 05 05:47:26 webcam streaming is a solved problem under linux in general Mar 05 05:48:10 you may need to adjust stream parameters to not overload the cpu Mar 05 08:07:41 Hey all, noob back. I got an LED blinking via the command line (yay!) and my cross compiler at least partially set up (yay!) but when I copy the compiled program to my BB xM, it shows up in an ls, etc, but when I try to run it with ./gpio_test, it just returns "file not found" :/ Mar 05 08:07:50 Which seems a very strange error to me Mar 05 08:08:43 Running file on it gives me this: ELF 32-bit LSB executable, ARM, version 1 (SYSV), dynamically linked (uses shared libs), for GNU/Linux 2.6.16, not stripped Mar 05 08:09:18 Hmm, which my BB gives a uname of 2.6.32, I don't know if that will break things or not Mar 05 08:13:56 is your libraries the same on your xM as your x-compile system? Mar 05 08:13:58 if not, fix that Mar 05 08:14:12 I'm not sure if I've got my cc environment set up correctly, but I installed CodeSourcery and am using -march=armv7-a -mcpu=cortex-a8 with the $(SYSROOT) vars as the CodeSourcery docs told me tod Mar 05 08:14:52 I'm not sure how to tell. I'm using the libc from CodeSourcery as my sysroot Mar 05 08:15:34 and your board has? Mar 05 08:15:56 This is my makefile http://pastebin.com/sQEkX7VQ Mar 05 08:16:05 It's running Angstrom Mar 05 08:16:14 recent angstrom? Mar 05 08:16:15 So it's using the Angstrom libc, I suppose Mar 05 08:16:25 Whatever came with rev c4 Mar 05 08:16:28 * fiveofoh checks Mar 05 08:16:31 or ancient angstrom setup to use CS? Mar 05 08:16:45 RevC4 normally doesn't ship with anything unless you bought it from SC Mar 05 08:17:31 err sorry Mar 05 08:17:41 It's an xM, rev C Mar 05 08:17:45 Not a BB rev C4 Mar 05 08:17:45 Oh Mar 05 08:17:53 those are not setup for CS Mar 05 08:18:11 What *are* they set up for? Mar 05 08:18:13 so either change your compiler to use the right libaries, use the ANgstrom SDK stuff, or change your libs on the target Mar 05 08:18:26 someversion of plain GCC of non CS lineage Mar 05 08:18:32 That sounds good to me Mar 05 08:18:47 Should I just hunt for cross-compiling on Angstrom then? Mar 05 08:19:12 yes... Angstrom should be able to generate the a SDK package with the x-compilers stuff Mar 05 08:19:25 not sure if the online builder does that...someone else will have to comment Mar 05 08:19:26 Angstrom armv7 toolchain Mar 05 08:19:34 Looks like that's what I want Mar 05 08:19:42 looks reasonable Mar 05 08:20:29 'k... it is 0020, will probally drop off Mar 05 08:20:40 Hokay, off to set up a new toolchain Mar 05 08:20:48 Thanks though! This sounds like it should be helpful :) Mar 05 08:20:59 Good night Mar 05 08:21:58 <_tasslehoff_> I've seen a couple of "NETDEV WATCHDOG: eth0 (smsc95xx): transmit queue 0 timed out" on my 2.6.39 kernel over the last few days. It was during automatic testing, but probably happened during semi-high network load with high cpu-load. Anyone seen a patch for such an issue? Mar 05 08:44:43 Woohoo! Mar 05 08:44:44 Blinking light Mar 05 08:44:45 ! Mar 05 08:44:49 Thanks, ds2 :) Mar 05 08:44:59 The angstrom toolchain did the trick Mar 05 08:45:11 Now that I can compile things, it's time to get things actually working :P Mar 05 08:47:30 koen: so in the end psp + patches is more secure than a kernel on GH :) Mar 05 08:56:08 Hu everyone! Say please how can I debug my BB Rev B4. I have eclipse, JLink, usb and serial connection with board Mar 05 08:56:22 printf? Mar 05 08:58:30 I hope that there are another way, more suitable for bare metal programming Mar 05 08:58:35 or no? Mar 05 08:59:00 bare metal? Mar 05 09:01:19 I mean work witrh board on low level, like developing bootloaders Mar 05 09:02:40 jtag could help Mar 05 09:06:22 but can I use my jlink for it? i met diff decisions (dsteram and ds-5, TinCanTools etc), but nothing aboyt using jlink adapters Mar 05 09:07:09 if the A8 is supported... Mar 05 09:11:31 MrTea: I would ask the provider of your jlink if cortexA8 and/or omap3 is supported Mar 05 09:12:52 cortex a8 is supported Mar 05 09:13:49 so? Mar 05 09:15:15 so as now I need to connect it right with eclipse Mar 05 09:15:28 #eclipse then Mar 05 09:16:09 but really i dont know how Mar 05 09:16:19 neither do I :) Mar 05 09:17:15 ok, thanks) Mar 05 09:24:47 Hello. I have a rev B4 beagleboard with angstrom and I'm trying to establish a usbnetwork witm my PC(ubuntu). I powered BB via power jack, connected via serial port and usb. When I open interface configuration on my BB, it shows local loopback and usb0. When i do the same on my PC, it only shows eth0(current working ethernet interface), local loopback and wlan) Mar 05 09:25:20 What should I do to make usb0 an active inteface on my PC? Mar 05 10:33:07 mru: it's funny how conversations with ARM developers quickly turn to "Hey, do you know Mans?" Mar 05 10:36:21 hehe Mar 05 10:51:31 oO( how comes prpplague is awake now?) Mar 05 10:53:24 insomnia Mar 05 10:53:53 and a better chance to order an rpi Mar 05 10:54:14 *sing* i can't get no sleep Mar 05 11:02:44 Hello again. I need to install gdb on my rev B4 beagleboard. How to do it, if my BB is not connected to network? Are there any step-by-step instructions? Mar 05 11:04:32 1) connect it to a network Mar 05 11:04:34 or Mar 05 11:04:42 2) include gdb with narcissus Mar 05 11:04:42 or Mar 05 11:04:49 3) download image that already has gdb Mar 05 11:07:40 Thanks. It seems that I'll have to set up network anyway. Mar 05 11:07:58 Could very well do it now Mar 05 11:08:41 <_tasslehoff_> koen: there are some scripts/setup that's only ran the first time angstrom starts, right? Mar 05 11:09:54 * _tasslehoff_ tries to find out why his hwclock gets messed up after installing a new image Mar 05 11:26:01 * _tasslehoff_ blames /etc/timestamp Mar 05 11:35:47 _tasslehoff_: /etc/timestamp gets read on each boot, it should only reset the clock is timestamp is newer than the current date Mar 05 11:36:21 _tasslehoff_: there were some recent changes to the format in oe-core, I need to check if that didn't break things Mar 05 11:46:55 <_tasslehoff_> koen: I'm on oe-classic. running on rootfs1 I extract a new upgrade to a rootfs2 partition. that one has a timestamp from 1 march (when I created to rootfs), and it seems to be used. Mar 05 11:47:24 <_tasslehoff_> bah. that ended up being confusing :) Mar 05 12:18:02 <_tasslehoff_> and now I can't reproduce it, so probably user error yet again Mar 05 12:29:03 try reproducing the user Mar 05 12:29:30 see also xkcd 583 Mar 05 13:05:32 the congopad: http://www.vmktech.com/ Mar 05 13:09:21 what a useless device Mar 05 13:11:47 reminds me of congatec, who always ahd bodypainting models on trade fairs during their early years. Mar 05 13:13:13 gm all Mar 05 13:13:57 LetoThe2nd: that seems much more useful Mar 05 13:14:28 mru: yes, i totally agree. Mar 05 13:17:07 hi all Mar 05 13:18:32 <_tasslehoff_> nah, there it happened again. http://pastebin.com/CkbUNHdR. looks like it sometimes fails, and uses the old /etc/timestamp from the upgrade package instead. Mar 05 13:19:07 i'm trying to natively compile a kernel module for an Angstrom distribution Mar 05 13:19:11 _tasslehoff_: /etc/timestamp is added by OE (or narcissus), not via a package Mar 05 13:19:16 but i had no succes until now Mar 05 13:19:23 stefan88_15: right, cross compile it Mar 05 13:19:50 it is just a module, i don't wanna use oe and stuff :) Mar 05 13:20:06 fine, suit yourself Mar 05 13:20:30 <_tasslehoff_> koen: I meant *my* package, and by package I meant rootfs.tar.bz2 :) Mar 05 13:20:31 so, it can't be done natively? Mar 05 13:21:10 stefan88_15: sure it can, it's just stupid Mar 05 13:21:17 <_tasslehoff_> which has an /etc/timestamp added by OE Mar 05 13:22:08 koen: why is it stupid?:) Mar 05 13:22:49 oO( if you have to ask why compiling natively is stupid, you are not an... ;) ) Mar 05 13:23:10 i've installed kernel-headers but it seams that files like init.h, module.h etc are not pressent Mar 05 13:23:53 Leto: it is a simple <100line module for pwm, how long do u think it will take compiling it? Mar 05 13:24:28 or there is something else and i don;t get your irony? Mar 05 13:24:31 stefan88_15: a) think about tabcompletion to write my name. then highlighting will work. b) that was an insider joke. Mar 05 13:27:07 right, thx for your help Mar 05 13:27:51 you're welcome :) Mar 05 13:36:08 <_tasslehoff_> hm. I actually insert the rtc-twl.ko in the beginning of hwclock.sh, so I suspect a race condition here. Mar 05 13:36:48 <_tasslehoff_> so, user error again, just a bit more advanced than I thought Mar 05 13:40:38 _tasslehoff_: you are getting more advanced Mar 05 13:46:07 _tasslehoff_: I'd recommend building in the rtc-twl module Mar 05 13:46:22 <_tasslehoff_> koen: already in progress :) Mar 05 13:46:29 _tasslehoff_: that way the kernel takes care of a lof of rtc details Mar 05 13:48:34 koen: iirc udev had issues with rtc not as module Mar 05 13:49:04 well, just some fail-log, anything serious Mar 05 13:50:15 I'm quite sure that's only with old udevs :) Mar 05 13:50:39 kay told me to build it in statically last year Mar 05 13:50:44 and he's the udev maintainer Mar 05 13:53:42 ok, so i'm gonna use OE for cross-compiling my module Mar 05 13:53:59 first, i have to built the kernel, right? (bitbake virtual/kernel) Mar 05 13:55:53 any idea how long this takes? Mar 05 14:02:17 hey. the beaglebone system reference says the max supply voltage applyable is 3.3V. The 3.3V line from the board is actually 3.6V, is this safe to feed the GPIO ? Mar 05 14:38:38 anyone here use qemu to test and build out images? Mar 05 14:38:48 why would anyone do that? Mar 05 14:40:26 save time potentially? i have always used pre built images and then configured onboard Mar 05 14:40:42 save time???!?!!??eleventyone!! Mar 05 14:41:03 it takes forever, so instead i would like to emulate and build on faster hardware Mar 05 14:41:34 I've done that while waiting for my first beagle to arrive Mar 05 14:42:52 looks like qemu will boot the sdcard directly, sweet. Mar 05 14:43:27 koen: do you build uinput as module? Mar 05 14:43:56 (still wrt udev) Mar 05 14:44:05 real hardware is faster than qemu Mar 05 14:45:16 what about faster real hardware running qmeu? Mar 05 14:46:11 uh? Mar 05 14:46:50 at best it should run at the same speed if it's emulating correctly Mar 05 14:46:58 ^_^ Mar 05 14:47:09 * mranostay yawns Mar 05 14:47:42 you'd need a *very* fast computer to emulate a beagle faster than an actual beagle Mar 05 14:47:57 what about the SD card bottleneck, would it relieve any of that? Mar 05 14:48:12 also, use a crosss-compiler Mar 05 14:50:14 would i be better off chrooting into the sd card an building out stuff? Mar 05 14:51:18 wasnt there this nifty setup where you build in qemu, but the actual compile is done using a cross compiler? Mar 05 14:53:11 if you built in qemu, would there be a need to cross compile? Mar 05 14:54:13 yes Mar 05 14:54:19 coz compile in qemu is slow Mar 05 14:54:32 so there is a trick to invoke the host cross compiler Mar 05 14:54:38 for the actual compile Mar 05 14:54:50 so stupid automake projects still get the correct config Mar 05 14:55:08 one of these "build systems" does that, forgot which one Mar 05 14:56:10 I /finally/ figured out how to link against all the libraries I need to cross compile, and feel like an idiot for not doing it sooner Mar 05 14:56:20 wouldnt you be stick in the qemu box? how would yet get out and invoke the host cc? Mar 05 14:56:37 I guess somebody edited source code to make it work Mar 05 14:56:39 stuck* Mar 05 15:02:45 i wonder if i can setup my x86 ubuntu server to automatically CC anything run from apt on my xm. That would be neat Mar 05 15:02:55 hello Mar 05 15:03:21 i'm at my girlfriend house Mar 05 15:03:29 hi snowrichard Mar 05 15:03:37 did you ever get your internet connection fixed? Mar 05 15:03:44 av500: I wouldn't call scarebox nifty Mar 05 15:04:01 av500: it breaks down quite fast when using e.g. python in your buildscripts Mar 05 15:04:05 my novatel 2200 delivered saturday but i was not there Mar 05 15:04:05 i'll get it hooked up tommorrow Mar 05 15:04:13 and like the scarecrow, it doesn't actually work Mar 05 15:04:26 koen: well, the concept is nifty, the implemtation, I haz no idea Mar 05 15:04:41 we just got engaged Mar 05 15:04:43 why does python break it? Mar 05 15:04:48 snowrichard: congrats Mar 05 15:04:56 snowrichard: congrats! Mar 05 15:05:13 snowrichard: does she know about your dark secret? Mar 05 15:05:18 hanging out here... Mar 05 15:05:30 don't know which one you mean but yes Mar 05 15:05:36 we've been talking since 03 Mar 05 15:05:45 av500: it runs the host python, so you get leakage from your host system if you want to be clever and use python for the size of long Mar 05 15:06:01 koen: ah Mar 05 15:06:15 i downloaded the cross-lfs book Mar 05 15:06:18 i may do that at some point Mar 05 15:06:26 koen: well, that breaks the concept of doinf the config on the emu Mar 05 15:06:44 av500: 3 guesses why python modules in maemo/mer/meh had weird behavioour Mar 05 15:06:53 ubuntu working great though Mar 05 15:07:18 koen: I can imagine Mar 05 15:07:26 but I assumed only the cc would be on host Mar 05 15:07:32 koen: "because it's maemo"? Mar 05 15:07:34 debian squeeze work a little faster Mar 05 15:07:53 but maemo does not need wakelocks! Mar 05 15:07:54 mru: you're cheating Mar 05 15:08:11 i want a rasberry PI - the 25 dollar flah Mar 05 15:08:21 flash sized thingy Mar 05 15:08:24 no, you don't Mar 05 15:08:28 lol Mar 05 15:08:34 it's slower than a beagle Mar 05 15:08:36 the beagle is fine Mar 05 15:08:37 lol Mar 05 15:08:48 its not flash sized Mar 05 15:08:50 it's slow and closed Mar 05 15:08:55 i do want an xm Mar 05 15:09:15 what's closed about a binary blob for the GPU to boot the arm? Mar 05 15:09:23 I heard meego had scripts for binary blobs Mar 05 15:09:32 i want to build car entertainment system with voice recognition Mar 05 15:09:41 say 'hey DJ play pink floyd Mar 05 15:09:57 nice, once you have that, talk to my CEO Mar 05 15:09:59 he wants that too Mar 05 15:10:08 cheap please Mar 05 15:10:14 * mru wants driving directions to the dark side of the moon Mar 05 15:10:14 if i build it they will come Mar 05 15:10:24 snowrichard: I think you want *working* voice recognition :) Mar 05 15:10:25 i want a dollar for each unit sold lol Mar 05 15:10:32 mru: follow the saucers Mar 05 15:10:36 in reverse direction Mar 05 15:10:52 yeah you know like star trek Mar 05 15:10:56 one can always mount a hidden Siri somewhere.... Mar 05 15:11:03 behind the dash Mar 05 15:11:25 look, the $25 rpi and that 600$ iphone..... Mar 05 15:11:33 well i could o the voice part in an android app that uses wifi to talk to the media server Mar 05 15:11:50 or just run android and use the built in voice features Mar 05 15:12:00 android is not working on my beagle Mar 05 15:12:06 usb dies Mar 05 15:12:10 smplman: you have to 1) be online and 2) they do voice to text but not semantics Mar 05 15:12:43 av500: theres a nice app out called vlingo, i agree with having to have a dta conntection Mar 05 15:12:49 i have a sick puppy lol Mar 05 15:13:22 snowrichard: android crawls on the xm Mar 05 15:13:24 smplman: recognizing a defined set of command is way easier that "play pink floyd" Mar 05 15:13:27 than Mar 05 15:13:52 yes i realize Mar 05 15:13:53 true Mar 05 15:14:03 my friends phone is ringing Mar 05 15:14:07 but i don't touch it Mar 05 15:14:15 av500: the saucers full of secrets? Mar 05 15:14:29 the ones that hid on the dark side for 70ys Mar 05 15:14:52 can't seem them, they're obscured by clouds Mar 05 15:14:59 I'm in england, remember Mar 05 15:15:10 I feel for you Mar 05 15:15:39 we can't decide yet whether i should move up her, or her down there, or go somewhere ele Mar 05 15:15:43 else even Mar 05 15:15:56 i like the small town she wants to live in though Mar 05 15:16:18 more people talk in a smaller town Mar 05 15:16:30 less people, but they talk more :) Mar 05 15:16:47 exactly Mar 05 15:16:58 towns are too small, need a city Mar 05 15:17:15 she had bad experiences in arizona in big city Mar 05 15:17:21 she would not do that again Mar 05 15:17:38 there are good cities and bad ones Mar 05 15:17:44 and i'll never live in houston again Mar 05 15:17:54 never been to houston Mar 05 15:17:59 snowrichard: bad experience in general or with the city? Mar 05 15:18:00 suck Mar 05 15:18:02 crack rock Mar 05 15:18:08 robbery mugging so on Mar 05 15:18:19 there murder rate is high Mar 05 16:12:26 interesting that the zigbee home automation profile supports measurement and sensing for temperature, pressure, luminance, flow, humidity, and occupancy ... but not current. Mar 05 16:13:11 well, missuse "flow" for current Mar 05 16:13:14 electron flow Mar 05 16:13:43 hm good point Mar 05 16:14:05 and voltage can be electron pressure Mar 05 16:14:06 is what what flow is for Mar 05 16:14:29 no. the units of theri flow are cubic meters per hour. Mar 05 16:14:47 well, electrons have a volume :) Mar 05 16:14:56 be creative Mar 05 16:16:59 a high cholesterol diet can limit electron flow in household tubes Mar 05 16:23:07 hm there's a kW and kWh analog attribute but it's not tied to the home automation profile Mar 05 16:23:10 I wonder if one could use it anywa Mar 05 16:23:12 y Mar 05 16:23:18 (and still be within spec) Mar 05 16:24:13 SilicaGel: I found zigbee interesting for about an hour or so once when I had to review our stack's linux integration…then totally lost interest when I read about the licensing restrictions Mar 05 16:24:33 which are what, pay us money or you can't use it? Mar 05 16:25:01 mdp: ZAP-HLOS or the proper linux one? Mar 05 16:25:32 SilicaGel: the specification makes zigbee FOSS incompatible Mar 05 16:25:45 koen, the ZAP crap Mar 05 16:25:56 hm so how is there a ilnux stack then, it's just restricted to non commercial use? Mar 05 16:26:42 koen, it's a steaming pile..I diplomatically explained how it was complately unacceptable to that team after being solicited for feedback..they responded with, "we don't have time to address these issues" Mar 05 16:27:06 koen, so I asked them, "why did you waste my time asking for feedback?" Mar 05 16:27:33 mdp: I worked on ZAP-HLOS a bit to make it work on beagle/angstrom Mar 05 16:27:38 * SilicaGel frantically looking to figure out what you're talking about Mar 05 16:27:53 mdp: and it worked :) Mar 05 16:28:09 koen, "it works" Mar 05 16:28:19 hi all, I have an issue with a gst-launch line : ERROR: from element /GstPipeline:pipeline0/GstTIVidenc1:tividenc10: failed to create video encoder: h264enc Mar 05 16:28:50 koen, the design is oriented around an RTOS or uC model..completely linux unfriendly Mar 05 16:29:06 all compile time stuff throughout the "HAL"..blech Mar 05 16:29:16 mdp: I know Mar 05 16:29:27 yeah, you know all this Mar 05 16:29:32 here is my line : gst-launch v4l2src always-copy=FALSE input-src=COMPONENT num-buffers=2000 ! TIVidResize contiguousInputFrame=TRUE ! 'video/x-raw-yuv,width=640,height=480,framerate=25/1' ! ffmpegcolorspace ! TIVidenc1 codecName=h264enc engineName=codecServer byteStream=FALSE ! qtmux ! filesink location=yuv/test.mp4 Mar 05 16:29:32 mdp: anything in TI that has 'hlos' in the src tree should be avoided :) Mar 05 16:30:13 mdp: usually our cousins in WBU have a linux-friendly rewrite ready when asked Mar 05 16:30:16 this is a leopardboard, but nobody on leopard channel... Anyway, when replacing TIVidenc1 by a simple filesink it works well Mar 05 16:31:27 SilicaGel: http://freaklabs.org/index.php/Blog/Zigbee/Zigbee-Linux-and-the-GPL.html Mar 05 16:32:10 koen, I'd wager that zigbee on top of 802.15.4 isn't on the WBU radar map..but I could be wrong Mar 05 16:32:38 prolly not Mar 05 16:32:42 nfs likely is Mar 05 16:33:40 SilicaGel: that will explain why all open efforts have moved to 6LoWPAN…but it doesn't solve the issue of the huge deployment base of zigbee in power applications Mar 05 16:33:57 we are unfortunately fscked there for FOSS work Mar 05 16:34:30 s/nfs/nfc/ Mar 05 16:34:31 yeah Mar 05 16:34:41 6LoWPAN doesn't define the profiles /object models though do they Mar 05 16:34:43 hello guys, I was trying to use pyaudio module and I thought that it was present in BeagleBoard... but when I installed python-audio and tried to load pyAudio... it failed... is it a different module or just has a different name? Mar 05 16:35:54 SilicaGel: correct..the real value in zigbee is the ability to interact with the power application profiles deployed in the smart meters that have standardized atm on zigbee Mar 05 16:36:16 yeah Mar 05 16:36:40 SilicaGel: even out here in Ohio, the small city I live outside of (pop 20Kish) is rolling out smart meters this year with zigbee Mar 05 16:36:47 no kidding Mar 05 16:36:51 so it's extremely pervasive just based on that Mar 05 16:37:10 this thread is 2009; there are no updates to their policy on this? Mar 05 16:37:19 nope Mar 05 16:37:26 did they even respond? Mar 05 16:37:28 to say 'tough shit" ? Mar 05 16:37:44 mdp: wadsworth is cutting edge? :) Mar 05 16:38:42 since I like to wager…I'd also wager that the founding members of the zigbee consortium have looked at this and decided it's in their best interest *NOT* to liberalize their policy Mar 05 16:39:18 from past trade association participation, generally a few companies with the board seats hold all the power on this stuff Mar 05 16:39:23 so follow the money :) Mar 05 16:39:31 yeah, same with the SAE groups I was in Mar 05 16:39:44 it was volvo, allison, catepillar Mar 05 16:40:23 mranostay, yeah, they are surprisingly cutting edge..but I think this is just more indicative of the nationwide move by utilities Mar 05 16:40:39 so is it an actual mesh then Mar 05 16:40:42 with a node at your house Mar 05 16:40:45 and another one at your neighbors? Mar 05 16:41:47 mranostay, my acquaintances in the power and comm utilities tell me they've done the same thing most of the larger utility providers have done where they have cut a deal with a distributor that also provides the zigbee enabled smart thermostats (something cheaper than the expensive Nest units) Mar 05 16:42:34 mranostay, basically, the trade rags have shown all of them the huge savings they will get and better control of their power consumption by getting these installed in their customers' home Mar 05 16:42:35 the smart energy profile of zigbee uses two levels of encryption Mar 05 16:42:37 what I've wondered is Mar 05 16:42:43 can I write software to join the network the thing is on Mar 05 16:42:47 and read the meter from my own PC Mar 05 16:42:54 my feeling is that they probably wouldn't like that. Mar 05 16:43:13 i based that on no information though, except that people are generally jerks Mar 05 16:44:51 I'm not sure what the practical nature is of putting devices on the network Mar 05 16:45:02 yeah Mar 05 16:45:05 i know for that profile though Mar 05 16:45:13 However, you *are* supposed to be able to buy a Nest thermostat and get that power data from you meter Mar 05 16:45:24 there's an encryption key for the network, and another (potentially unique) one for the device Mar 05 16:46:52 well, what kind of keys? are you talking for key exchange…or the block cipher keys for the actual session? Mar 05 16:47:27 it talks about cipher keys yeah Mar 05 16:47:32 I haven't looked at that level of the spec..but I assume there's a practical way to "pair" and get the meter to talk to a thermostat Mar 05 16:47:40 hrm interesting Mar 05 16:48:01 yeah, probably so other appendix talks about key exchange and that general process of getting devices to talk Mar 05 16:49:06 possibly your utility provides a magic value online that only you can use to complete a key exchange with the meter Mar 05 16:49:07 something like that Mar 05 16:49:51 I'm hoping zigbee dies a quick death so we can dump it in the nearest landfill Mar 05 16:50:11 hrm Mar 05 16:50:20 zigbee, xbee and (6)lowpan are all 802.15.4, right? Mar 05 16:50:27 yeah. Mar 05 16:50:46 the good news is the google's @home stuffs is using 6LoWPAN..but that's probably jus their usual eternal beta test "product" Mar 05 16:51:18 when google killed google power meter I kind of thought they wouldn't ever take this kind of stuff seiously again Mar 05 16:51:19 idunnno Mar 05 16:51:22 koen, yeppers..you can carry whatever you want on an 802.15.4 radio Mar 05 16:51:41 ah, now it starts to make sense :) Mar 05 16:51:52 SilicaGel: they have partners that do the real work and are already in that business Mar 05 16:52:25 koen, and you can tell that 802.15.4 is FOSS friendly :) that support is not a problem Mar 05 16:52:43 yeah Mar 05 16:53:02 the power here though is doing these things in an interoperable manner Mar 05 16:53:06 koen, yeah, you'll also see that people using 802.15.4 xbees don't use zigbee normally..most people only do p2p stuff Mar 05 16:53:10 I looked at 6LoWPAN and I found contiki and one other project Mar 05 16:53:15 both for uC and both look fairly terrible Mar 05 16:53:16 right Mar 05 16:53:26 and require you to use their wacky programming language + their wacky micro kernel Mar 05 16:53:29 neither of which I wanted Mar 05 16:53:35 I just wanted a friggin protocol stack Mar 05 16:53:37 there's some linux-oriented 6LoWPAN work happening Mar 05 16:53:40 iirc Mar 05 16:54:01 it helps when it's being pushed through the IETF Mar 05 16:56:33 look at the smart light bulb efforts…the dinosaurs in that business went proprietary controllers and lighting science went 6LoWPAN Mar 05 16:59:21 depressing :( Mar 05 17:03:01 SilicaGel: the nice thing is that one can solve it all by contributing to a FOSS 6LoWPAN solution Mar 05 17:04:38 SilicaGel: it's a chance to be a hero! Mar 05 17:14:14 yeah i guess Mar 05 17:15:47 the frustrating part, to me, about contributing to projects like that is the communication. Knowing where to contribute and that you aren't just wasting an enormous amount of time Mar 05 17:16:02 I mean look at the learnign curve i had to climb to create my PRU reset patch for the kernel Mar 05 17:16:14 only to have vahabi say "that's not necessary, I already fixed it another way" Mar 05 17:16:28 well that was pretty depressing :( Mar 05 17:16:45 and they still might not let you into their clique! Mar 05 17:16:51 especially after I *thought* I performed due diligence looking for whether or not I was working on aproblem that really needed solving Mar 05 17:17:08 SilicaGel: will you mentor me? Mar 05 17:17:08 So when you talk about something huge like this ... crap I don't know, I'm an outsider Mar 05 17:17:21 smplman: let me check my magic 8 ball Mar 05 17:17:24 *shake* *shake* *shake* Mar 05 17:17:29 "Outlook not so good." Mar 05 17:17:38 couldn't hurt to ask Mar 05 17:17:41 :) Mar 05 17:18:25 So you started getting into kernel development by working on a PRU reset patch? Mar 05 17:18:33 no no no Mar 05 17:18:35 haha Mar 05 17:18:44 SilicaGel: keep in mind that the am335x kernel situation is utter crap Mar 05 17:19:04 The learning curve was trying to figur eout th eproper way to do a patch within the bitbake/oe/angstrom universe, mostly Mar 05 17:19:24 ahh i see Mar 05 17:19:26 which I would NOT have been able to do were it not for jwinnebeck ... Mar 05 17:19:53 I want te get more into embedded development, but i don;t know where to start Mar 05 17:19:54 The other part of it that was 'new' for me is all this clock tree / power domain stuff. Mar 05 17:20:10 i can build kernels and root filesystems Mar 05 17:20:30 i feel like im on the edge, just dont know where to jump Mar 05 17:20:55 i got ubuntu 11.10 running on my Xoom pretty much myself since there was no docs Mar 05 17:21:11 what do you want to do? Mar 05 17:21:32 koen: yeah. I definitely didn't realize that at first. Mar 05 17:22:50 jay6981: contribute to the community Mar 05 17:23:26 guys does anyone know how to search what is inside a package? Mar 05 17:23:30 what community? Mar 05 17:23:38 I mean... I've installed python-audio for Angstrom Mar 05 17:23:52 but how do I know what is inside it? I thought it was pyAudio... Mar 05 17:25:26 opkg files PACKAGE i think Mar 05 17:25:34 smplman: it actually seems hard to contribute to these "communities". they don't really take kindly to outsiders trying to help. Mar 05 17:25:47 jay6981: the beagleboard and ubuntu communities Mar 05 17:25:59 SilicaGel: we are all outsiders to something :) Mar 05 17:26:01 you could do a ton of work and write a bunch of patches and they'll just ignore you or nitpick about how you didn't do it right Mar 05 17:26:06 can anyone see why I can't build python-dbus for beagleboard? "configure: error: could not find Python headers" http://pastebin.com/GEyMLzeR I was building systemd-image when it broke Mar 05 17:26:17 i agree, i seem to get flamed for asking trivial questions to geniuses Mar 05 17:26:37 i tired update and bitbake -c clean already and it's still borked Mar 05 17:26:38 it's rather sad but the FOSS community is filled with antisocial people who aren't interested in collaboration at all Mar 05 17:26:49 just trolling Mar 05 17:27:08 @w1n9zr thanks - that's a starting point at least Mar 05 17:27:23 mmmm...geniuses Mar 05 17:27:40 W1N9Zr0: check the directories under line 187 and see if the headers are there Mar 05 17:27:49 That's the most screwed up ham callsign I ever saw Mar 05 17:28:55 jay6981: the beagle community seems pretty cool though Mar 05 17:29:05 you're just saying that because we're all here! Mar 05 17:29:28 no, because I hold his family hostage Mar 05 17:29:50 the mentality is very much: if you don't know, i'm not going to tell you. reminds me of grade school. Mar 05 17:29:54 what i want to know is if could I use what I have learned from the bone to get a better job Mar 05 17:29:55 !!! Mar 05 17:30:02 probably Mar 05 17:30:03 Apparently, this zigbee stuff will NOT get me a better job Mar 05 17:30:07 beagle community eats babies Mar 05 17:30:07 it will probably get me a *worst* one Mar 05 17:30:10 -e+t Mar 05 17:30:17 do you live in the bay area? Mar 05 17:30:28 smplman - there's one file there, pyconfig.h Mar 05 17:30:43 no, but my wife would be happy as hell if I moved there. Her family lives in oakland and walnut creek. Mar 05 17:30:45 jay6981: how are closed source communities in that regards? Mar 05 17:30:52 SilicaGel: it allows you to claim to be a green engineer Mar 05 17:30:55 I never took part in one, so I would not know Mar 05 17:31:06 One problem with california is my baby silas is illegal there Mar 05 17:31:17 W1N9Zr0: honestly i fun ubuntu and im not familiar with angstrom, but it looks like you need some sort of python-dev package Mar 05 17:31:18 e Mar 05 17:31:21 1 secon i will explain Mar 05 17:31:28 av500: not sure i understand the comparison. closed source communities are motivated by somethign much more tanglible... money. Mar 05 17:31:45 https://picasaweb.google.com/108380917869049424169/Silas#5294927223224252658 Mar 05 17:31:48 that's silas :) Mar 05 17:31:53 rapid prototyping is where the money is Mar 05 17:31:58 SilicaGel: you should probably move Mar 05 17:32:01 :) Mar 05 17:32:18 bbiab Mar 05 17:32:29 jay6981: well, you said how open source communities are and I was wondering what to compare them against Mar 05 17:32:47 https://picasaweb.google.com/108380917869049424169/Akuna#5294938307899692050 that's me and my galah, his name is Akuna Mar 05 17:33:07 exotic bird smuggler? Mar 05 17:33:13 haha Mar 05 17:33:19 or you have a coal mine? Mar 05 17:33:22 I have wanted a blue and gold macaw Mar 05 17:33:23 if I moved to california I woudl be Mar 05 17:41:51 has anyone gotten mame to run on an xm? Mar 05 17:42:04 I had it running years ago Mar 05 17:42:35 SilicaGel: many hobbies are illegal in CA Mar 05 17:42:42 koen: how you like your new job? Mar 05 17:43:25 like ferrets! Mar 05 17:44:02 and UAVs Mar 05 17:44:43 darknighte: pretty much the same for a lot more $$$ so far Mar 05 17:44:54 congrats. Mar 05 17:44:57 thanks! Mar 05 17:45:05 send beer Mar 05 17:45:08 SilicaGel: it's an ok tradeoff to live in a land of milk and honey with no worries. Mar 05 17:45:29 darknighte: I might have to poke you a bit for shipping hw to the mentor office for the bspfest Mar 05 17:45:37 k. Mar 05 17:45:41 you coming? Mar 05 17:46:02 darknighte: bspfest travel approved, waiting for collab invite Mar 05 17:46:18 cool. ping jefro and he can probably get you an invite to collab. Mar 05 17:46:20 ESC is the week before in san jose Mar 05 17:46:29 I pinged jefro already :) Mar 05 17:46:45 koen, take mine..I'm sticking with vacation that week :P Mar 05 17:47:46 darknighte: btw, I took away most reasons for you to yell at me for meta-ti Mar 05 17:47:52 darknighte: besides quitting TI :) Mar 05 17:48:27 koen: I'm sure I can come up with new ones. Mar 05 17:48:57 idunno man, i had an office in sunnyvale and I did NOT like it much there Mar 05 17:55:38 5 more minutes of lunch left, just enough time for some really loud Cheap Trick Mar 05 17:57:03 I want you to want me? Mar 05 17:57:10 hehe Mar 05 17:57:16 nope. They did a cover of the beatles Day Tripper Mar 05 17:57:30 it took me so long to find out ................ but I found out. Mar 05 18:06:51 OK Mar 05 18:07:00 I'm sick of working Mar 05 18:07:06 so it's time to play PRUSS and dmtimer now Mar 05 18:07:18 12ftw Mar 05 18:07:27 shhhhhhhhhhhhhhhhhh you'll get us killed Mar 05 18:22:26 hmm omap_dm_timer_request() did indeed return succes, but it returned timer->id == 0 Mar 05 18:22:31 very peculiar Mar 05 18:22:45 now I'm going to try requesting the super secret timer that will end the world Mar 05 18:26:20 this is cool Mar 05 18:26:23 i did a dmesg -c Mar 05 18:26:39 and it COMPLETELY broke dmesg, as in, it cleared the messages alright, it cleared them and all the subsequent ones forever Mar 05 18:26:58 :/ Mar 05 18:27:24 i wonder if jwinnebeck will notice if I reboot his bone remotely >:) Mar 05 18:28:03 I did actually Mar 05 18:28:18 Windows pops up with USB errors when it boots, for some reason Mar 05 18:28:22 doh. I'm sorry. that was rude of me. Mar 05 18:28:43 that's OK I wasn't using it, I was just checking, because I'm also interested in how reliable the bone is Mar 05 18:28:47 since I've found it "dead" a few time Mar 05 18:29:00 SilicaGel: systemd-journalctl -o cat Mar 05 18:29:09 i have to figure out how you're supposed to use pr_fmt. Mar 05 18:29:13 SilicaGel: that includes everything, including dmesg Mar 05 18:29:22 ok Mar 05 18:30:36 hrm maybe KERN_INFO just isn't on Mar 05 18:31:15 SilicaGel: oh, and mkdir /var/log/journal if you want logs survive reboot Mar 05 18:31:27 cool, thanks Mar 05 18:40:14 | /home/oe-classic/oe-classic/build/tmp-angstrom_2010_x/work/usrp-e1xx-angstrom-linux-gnueabi/ti-dsplink-1_1_65_00_03-r0i/dsplink_linux_1_65_00_03/dsplink/make/start.mk:44: *** This makefile requires one of GNU make version 3.81 3.81beta1 3.90 3.92. Stop. Mar 05 18:40:20 does this make any sense? Mar 05 18:40:23 F16 Mar 05 18:41:39 urg, I have 3.82, bother Mar 05 18:41:45 Crofton: i could never get the display link stuff to compile on arm, or maby it was the touch kit stuff Mar 05 18:42:51 this is from oe classic Mar 05 18:43:01 looks like it hates make version 3.82 for some reason Mar 05 18:43:09 dsplink btq Mar 05 18:43:11 btw Mar 05 18:44:58 Crofton: cross compile potentially? Mar 05 18:56:59 <_av500_> it is cross compile Mar 05 18:57:55 <_av500_> Crofton|work: just comment that check Mar 05 18:58:26 Crofton|work: last I heard 3.82 was the latest version of gnu make Mar 05 18:59:02 and I checked 30 seconds ago Mar 05 19:08:50 hm don't get it, switched to dev_info(), dev_warn(), etc. like I was supposed to and I just get no output. switching back to printk I guess. Mar 05 19:18:41 <_av500_> SilicaGel: iirc you need to do -DDEBUG Mar 05 19:18:57 even for a KERN_INFO? Hrm Mar 05 19:19:20 <_av500_> dmesg -n X Mar 05 19:19:30 7 means show everything right Mar 05 19:34:52 <_av500_> or 8 Mar 05 19:36:12 I can build you a kernel that goes to 11 Mar 05 19:36:45 I don't know, but if you try to build one that goes to 12, TI will kill you. Mar 05 19:38:48 wow Mar 05 19:38:53 apparently screwing with DRV_NAME is bad ! Mar 05 19:44:32 [ 846.997986] pruss_uio: Allocated timer 0 Mar 05 19:44:32 [ 847.002186] pruss_uio: Allocated timer 3 Mar 05 19:44:34 huh. Mar 05 19:44:40 Attempts to allocate timer 12 were a complete failure. Mar 05 19:44:51 I thought it was a one off error, but attempts to allocate timer 11 also failed Mar 05 19:44:57 Those sneaky fargin bastages Mar 05 19:48:24 [ 1075.113240] pruss_uio: Allocated timer 3, irq 69, base address fa042000 Mar 05 19:48:26 yay! Mar 05 19:48:33 let's see if that address makes any sense at all! Mar 05 19:51:10 ls -l Mar 05 19:51:58 drwx-rx-rx 2 jay6981 users 4096 Jan 1 1970 00:00 . Mar 05 19:52:25 woah Mar 05 19:52:35 sounds like he needs my real time clock cape, with integrated Xbee ! Mar 05 19:53:10 thanks mru Mar 05 19:53:21 I had a computer that would reset the clock to 2051 if left without power too long Mar 05 19:53:39 I never figured out what was special about that year Mar 05 19:53:52 is that all 1's? Mar 05 19:54:35 the day your god created the computer Mar 05 19:54:37 depends on the epoch, no? Mar 05 19:54:51 what's the binary roll over year? Mar 05 19:55:05 djlewis: time-travelling computer? Mar 05 19:55:17 ds2: 32-bit unix time rolls over in 2037 Mar 05 19:55:26 Hmmm Mar 05 19:55:30 2038 they say but is that 2^31 for reasons of time_t being originally signed? Mar 05 19:55:41 time_t is signed Mar 05 19:57:01 2^32 - 1 --> GMT: Sun, 07 Feb 2106 06:28:15 GMT Mar 05 19:57:30 if it was unsigned there would be no time before 1970 Mar 05 19:57:30 2 ^ 31 - 1 --> GMT: Tue, 19 Jan 2038 03:14:07 GMT Mar 05 19:57:32 the firmware probably stored the time in some other format Mar 05 19:57:37 so I'm not sure what the hell 2051 is Mar 05 19:57:52 was an old alpha machine Mar 05 19:58:00 hey, only the mayans would make up their own epoch! Mar 05 19:58:52 http://xkcd.com/607/ Mar 05 19:58:55 this 3D camera setup with the xM is pretty neat Mar 05 19:59:35 also, http://xkcd.com/376/ Mar 05 20:02:42 gm Mar 05 20:02:44 or so Mar 05 20:05:05 hm noooooooooo that address it gave me back as io_base for the timer makes no sense whatsoever. hrm. Mar 05 20:23:51 I think I should make a drinking game, where every time T.I. tells me "The PRUSS is not supported" everyone has to take a drink Mar 05 20:23:56 sorry, I meant TI, not T.I. Mar 05 20:24:08 Treasure Island?!? Mar 05 20:24:12 *g* Mar 05 20:24:56 I don't think you can map an interrupt > irq60 to the PRU :( which means no timer events :( Mar 05 20:29:20 why does the boot partition have to be vfat? Mar 05 20:29:32 can it be ext3? Mar 05 20:30:00 depends on how well you succeed in patching the boot ROM in the chip Mar 05 20:30:36 the default boot ROM requires vfat? Mar 05 20:31:01 no laser beams Mar 05 20:31:07 for photon emissions Mar 05 20:31:24 woglinde so get laser beams! Mar 05 20:31:40 build it into boot rom Mar 05 20:32:35 why did they choose it to default to vfat? Mar 05 20:32:45 JViz, the detailed answers are in the TRM Mar 05 20:33:11 it is a fountain of knowledge and goodliness Mar 05 20:33:21 vfat code is small and easy Mar 05 20:34:34 woglinde: ah, thank you Mar 05 20:35:07 isn't M$ suing people over using vfat? Mar 05 20:37:23 that's exfat Mar 05 20:37:28 JViz: one entity…not people..to my knowledge Mar 05 20:38:00 some patents for vfat were ended Mar 05 20:44:49 hm in the am335x how is an 'event' different from an 'interrupt' ? Mar 05 20:50:42 SilicaGel: the term 'event' is used in a few different contexts in the TRM Mar 05 20:51:07 SilicaGel: they generally talk about events in the scope of an IP block Mar 05 20:51:22 oh Mar 05 20:51:34 and then those events can be mapped to an action outside of that block…an interrupt Mar 05 20:51:40 oh ok Mar 05 20:51:41 i see Mar 05 20:51:43 ok Mar 05 20:51:52 so that's why things that the outside world consdiers an "irq" the pru considers an "event" Mar 05 20:52:00 generally the complex blocks like pruss or edma have the concept of some event mux and then funnel up to an INTC interrupt Mar 05 20:52:11 yeah ok Mar 05 20:52:14 right Mar 05 20:52:30 I'm trying to go the other way actually. I'm trying to map a timer interrupt (like timer3 on irq69) to the PRU. I'm thinking it's not possible. Mar 05 20:52:59 yeah, it's best to look at the part as a collection of loosely related IP blocks that a designer bought off-the-shelf :) Mar 05 20:53:25 this is most apparent when you go to read the USB OTG chapter on our parts Mar 05 20:54:23 but you'll find a different use of events (in the dma event context) in the edma block and a slightly different usage in the ehrpwm block Mar 05 20:55:53 yeah ok. that makes more sense. I wasn't sure if it was just the PRU, since the PRU's machine definition doesn't have any concept of interrupts or interrupt handlers. You just map an external event (irq or event) to his INTC which maps it to a CHANNEL which maps it to a bit in R30 or R31 Mar 05 20:56:10 then the PRU code just polls for that thing to have occurred, signaling, for example, that some EDMA action completed Mar 05 20:56:18 that's the way I understand it, at least Mar 05 20:56:37 you know more than me about it Mar 05 20:56:39 busy-loops ftw Mar 05 20:57:01 yeah. Well in PRUSSv2 there's an actual "wait for event" instruction which maybe puts it into a lower power state I don't know Mar 05 20:57:40 I thought I had a pretty good handle on this, too, until I found that the irq to channel mapping registers only go up to 64, but the timer IRQs start at 66 ... so I can't figure out how the hell I can route a timer irq to a pruss event Mar 05 20:58:01 which means the only way I can think of to do this is busy waiting trying to poll timers over the L3 to L4 which would SUCK Mar 05 20:58:08 I hope some v2-specific docs will come on wiki when (I think it's *when*) the pasmv2 is released Mar 05 20:58:32 mdp: yeah me too. I've been able to figure out a lot of stuff from the TRM though about what I can expect! Mar 05 21:00:17 <_av500_> there, fix the condensate pump Mar 05 21:00:19 <_av500_> fixed Mar 05 21:00:28 was it leaking? Mar 05 21:00:39 <_av500_> it was not pumping Mar 05 21:00:48 did you have a flood? Mar 05 21:00:52 <_av500_> and the joke plumbers did not connect the emergency relay Mar 05 21:00:59 <_av500_> no flood, just a trickle Mar 05 21:01:04 Is this on a high efficiency furnace? Mar 05 21:01:08 <_av500_> yep Mar 05 21:01:09 mdp: hopefully melissa can soon share her PRU wisdom legally with the hoi polloi in #beagle :) Mar 05 21:01:16 _av500_: I had a new furnace put in very soon Mar 05 21:01:19 <_av500_> and a air heat exchanger Mar 05 21:01:25 koen: do you know melissa? Mar 05 21:02:13 _av500_: I just recently had a high efficiency furnace built in. They DID hook up the emergency switch to kill the furnace and A/C. But I was worried thet hing was going to crap out on me so I bought a Little Giant spare one anyway Mar 05 21:02:36 <_av500_> spare what? Mar 05 21:02:37 I figured it's cheap insurance, in case the thing quits in the middle of the night on a weekend when it's -5 F out. Mar 05 21:02:59 a whole spare condesate pump Mar 05 21:03:02 <_av500_> ah Mar 05 21:03:03 condensate Mar 05 21:03:11 <_av500_> this one is ~$300 Mar 05 21:03:19 -5F... ouch Mar 05 21:03:22 SilicaGel: I worked with her a few times when I was still at TI Mar 05 21:03:25 <_av500_> but ill try to buy a spare pump for it Mar 05 21:03:26 you live in the arctic circle or something? Mar 05 21:03:32 koen, I don't believe any of that "legally" business Mar 05 21:03:34 <_av500_> dont need a 2nd plastic and tubing Mar 05 21:03:43 it's all due to being swamped Mar 05 21:03:58 _av500_: I bought one of the Little Giant ones, it looks basically like http://www.amazon.com/Little-Giant-VCMX-20ULS-C-Automatic-Condensate/dp/B0026PFJX0 Mar 05 21:04:03 but I don't remember paying that much for it, HMMMM Mar 05 21:04:24 <_av500_> well, that the thing the joke plumbers put Mar 05 21:04:34 <_av500_> i might buy another one now Mar 05 21:04:35 I must have found a deal somewhere. Part of the reason I did this, too, is I plan on screwing with it, to feed my dehumidifier into one of the spare input ports Mar 05 21:05:04 the switch that comes out is fairly low voltage ... I think it's really intended to be used for 24VAC for you to just wire in series with the hot lead of your thermostat or something like that Mar 05 21:06:03 that melissa is who answers my PRU questions on the forum almost always. And she's the one that always starts out reminding me that the PRU is not supported haha Mar 05 21:06:31 This particular problem,t hough, I really can't see an answer. Mar 05 21:10:11 one thing i notice is that CMR15 has a lot of empty address space after it before the next defined register. I almost wonder if there's secretly a CMR16 and beyond Mar 05 21:10:57 in the latest OE bits, how is the MLO built? Mar 05 21:11:12 am I understanding correctly that it is built in the u-boot tree? Mar 05 21:11:15 * cbrake keeps digging Mar 05 21:13:19 cbrake: depends on the uboot version Mar 05 21:13:31 cbrake: 2011.09 and beyond can build mlo and uboot in one go Mar 05 21:14:34 koen: must be what I have: U-Boot SPL 2011.12-dirty (Feb 24 2012 - 11:50:41) Mar 05 21:23:42 cbrake: if you are going to use 2011.12 SPL for any serious work I suggest you add this patch, otherwise SDRC timing is set up incorrectly Mar 05 21:23:52 http://www.sakoman.com/cgi-bin/gitweb.cgi?p=u-boot.git;a=commitdiff;h=797c7a2f5d90598b9db8db4bdaa300f53371328d Mar 05 21:24:24 it is fixed in current upstream, but is definitely an issue in v2011.12 Mar 05 21:25:20 sakoman_: thanks! currently just a point of interest as I'm watching the latest oe-core builds :-) Mar 05 21:26:10 cbrake: I've made the switch to yocto/oe-core for all new work Mar 05 21:26:17 sakoman_: good deal Mar 05 21:26:31 it's been interesting ;-) Mar 05 21:26:48 sakoman_: I'm still using oe-core/meta-oe/angstrom ... Mar 05 21:27:01 sakoman_: so far, its working out pretty well. I really like the systemd stuff Mar 05 21:27:16 yeah, I made that switch too Mar 05 21:27:40 sakoman_: are you using the yocto meta layer with meta-oe/angstrom? Mar 05 21:29:22 cbrake: I'm using yocto + meta-oe and meta-gnome from meta-openembedded + my own layer Mar 05 21:30:15 sakoman_: interesting, I never tried pulling in the yocto meta layer yet Mar 05 21:30:41 sakoman_: figured there would be some explosion between in and meta-oe/angstrom, but that is good if everything plays well together Mar 05 21:30:47 cbrake: yocto is jsut oe-core with extra crap you don't need Mar 05 21:31:04 don't light the wrong fuse or things will get very small Mar 05 21:35:23 I'm sure one of you guys knows off hand, which is the better or more official way to build a Beagle Bone kernel, the OE method, or from the git repo? Are they both kept in sync? Mar 05 21:36:16 wow, the RPi community is really dominating elinux.org contributions of late Mar 05 21:37:13 koen: isn't the whole point of Yocto to add in the stuff you do need? Mar 05 21:47:01 wmat: are they useful contributions? Mar 05 21:47:06 or dups of other stuff? Mar 05 21:47:12 ds2: useful Mar 05 21:47:33 ds2: although, mostly lots of information about shipping dates, local mirrors, etc. Mar 05 21:47:43 ds2: as the hardware isn't in a lot of hands yet Mar 05 21:48:18 oh...no actual technical info yet? Mar 05 21:48:33 like the low level registers on the 3D accelerator Mar 05 21:48:36 ;) Mar 05 21:48:59 * wmat looks again Mar 05 21:52:22 hmmm, that's a Braodcam part isn't it Mar 05 21:52:39 :) Mar 05 21:52:43 very big :) Mar 05 21:55:09 here's the datasheet: http://dmkenr5gtnd8f.cloudfront.net/wp-content/uploads/2012/02/BCM2835-ARM-Peripherals.pdf Mar 05 21:55:24 here's the errata: http://elinux.org/BCM2835_datasheet_errata Mar 05 21:58:21 that looks more like a propaganda flyer then a datasheet Mar 05 21:58:33 unless this is a Linux chip, not a general purpose processor Mar 05 21:59:03 There is no public datasheet Mar 05 21:59:13 requires NDA, etc, from what I've read Mar 05 21:59:39 so it is a PoS Mar 05 21:59:54 And the bootup process is weird, blobs involved for booting, loading GPU Firmware, etc. Mar 05 22:00:20 I don't know that it's a PoS. By the time the kernel (or u-boot, if you want to stick that in between) is loaded, I think it's fairly normal Mar 05 22:01:01 blah on binary blobs Mar 05 22:01:04 The RPi people say that going to a more open chip would have cost more money and blown their budget. No idea how true that is. Mar 05 22:01:58 I'm not much on blobs either, but I think most of the blob issue with the RPi is firmware loading, which from my observation, people only seem to care about if they see it Mar 05 22:02:08 it just isn't worth the trouble for an ARM11 (Correct me on this if I am wrong) Mar 05 22:02:31 waste of time to use an ancient processor that is still locked up Mar 05 22:03:02 That's where I'm starting to get to. I mean for me. Again, they're all about making a $35 computer that boots. Mar 05 22:03:17 what's the next story... going to a lower power processor will cost more so stick with lugging a car battery around? Mar 05 22:03:18 Guys like us want to hack around a bit more :) Mar 05 22:03:42 Hey, I'm just repeating their position, not indicating agreement :) Mar 05 22:03:53 *nod* Mar 05 22:04:01 Actually though, low power is one of their things. 1w is their target Mar 05 22:04:17 and I can go below 1W with a M0 ;) Mar 05 22:04:35 put in an A8 and then talk Mar 05 22:04:36 I've been meaning to look up the M0. There's so much I don't know :( Mar 05 22:04:52 or an ARM7 :) Mar 05 22:09:22 The TI forums say I'm a prodigy Mar 05 22:09:23 :-) Mar 05 22:09:25 Hey, so you know anything about my kernel quesiotn? Mar 05 22:10:04 read the recipe Mar 05 22:10:28 yeah.... I need to learn more about OE. I'm looking at the files there. Mar 05 22:10:49 I mean more about how the github repo relates to what's in OE Mar 05 22:11:09 If one or the other is better or worse, I'll use it Mar 05 22:11:31 there should be anything 'in' OE Mar 05 22:11:42 I mean what's in the BB OE repo Mar 05 22:11:54 yes, and that is what I mean Mar 05 22:12:01 look in the recipe Mar 05 22:23:32 <_av500_> rpi bom with 1ght a8 is doable these days Mar 05 22:23:35 <_av500_> 1ghz Mar 05 22:24:10 <_av500_> there are 99$ tablets now, drop lcd, ts, battery, case and retail margin and you are at $35 Mar 05 22:30:32 ds2: so a quick OE question.. In sources/meta-ti/recipes-kernel/linux/linux-ti33x-psp-3.2, are the patches there applied _before the patches in the appropriate 3.2.x directory? Mar 05 22:30:58 depends. the application is solely determined by the recipe Mar 05 22:33:53 I guess I'm not finding the recipe Mar 05 22:34:08 There's a beaglebone.conf Mar 05 22:34:12 look for a .bb file Mar 05 22:35:42 closest thing I have is beaglebone-getting-started.bb (which seems to be documentation only), then gadget-init.bb, beaglebone-tester.bb. (that's the .bb files that have the word beagle in them) Mar 05 22:36:52 I'm pretty sure this is part of it Mar 05 22:36:54 sources/meta-ti/recipes-kernel/linux/linux-ti33x-psp_3.2.bb Mar 05 22:37:09 I guess that's what I'm looking for Mar 05 22:39:56 looks about right Mar 05 22:52:13 So it looks like there's four full pages of commits between what's on github https://github.com/beagleboard/am335x-linux/ and what's on http://arago-project.org/git/projects/?p=linux-am33x.git which the OE Recipe pulls from Mar 05 23:15:57 Is anyone build angstrom OE? Mar 05 23:48:25 IS anybody active there? Mar 05 23:50:42 where? Mar 05 23:55:56 on this IRC channel Mar 05 23:57:05 there's lots of active people at various times of day Mar 05 23:58:17 but we're very stealthy, so you can't hear us Mar 06 00:00:16 DPM at work. Mar 06 00:07:11 ohh thanks for that Mar 06 00:07:44 Can anyone help regarding angstrom OE build? Mar 06 00:11:38 can you be a little more specific? Mar 06 00:13:11 sure, I want to use onboard DSP for image processing, So I am following Opencv-DSP Acceleration project to work it out Mar 06 00:13:37 I found that I need to build Angstrom OE for that Mar 06 00:14:13 no idea how one would go about building Angstrom OE Mar 06 00:14:13 I am not able to do that bcoz of limited space on my machine Mar 06 00:14:31 seem old that one would want to rebuild a build system Mar 06 00:15:58 I read that to build angstrom OE , I need to have 20 GB space and couple of hours of time Mar 06 00:16:05 IS that true? Mar 06 00:16:17 Is there is any way to do that? Mar 06 00:19:20 here is the link http://code.google.com/p/opencv-dsp-acceleration/wiki/Instruction_For_Building_Examples Mar 06 00:20:50 what do you think ? Mar 06 00:25:41 thurbad, ds2 , have any idea? Mar 06 00:27:17 never used opencv, so I dunno Mar 06 00:27:43 * SilicaGel12 stars Mar 06 00:30:30 ok Mar 06 00:30:44 thanks thurbad Mar 06 01:24:35 uh oh, my i2c-2 is gone Mar 06 01:25:28 is it somewhere else in 3.0.23 ? Mar 06 01:26:30 koen: you around? Mar 06 01:26:41 he ought to be sleeping at this hour Mar 06 01:26:50 but then again, so should I Mar 06 01:28:06 indeed Mar 06 01:28:33 I'm looking at his patches in the OE repo Mar 06 01:28:52 Most of them are not applying Mar 06 01:29:02 could easily be my fault Mar 06 01:36:39 alan_o: which patches? Mar 06 01:46:39 mranostay: so the ones in the OE repo in sources/meta-ti/recipes-kernel/linux/linux-ti33x-psp-3.2 Mar 06 01:46:57 based on the recipe (and I'm no expert at this) I did: Mar 06 01:48:31 git clone (arago-project URL) then applied the patches (git-am) in 3.2.x/ (where x is 1 through 6) then tried to git-am the patches directly in linux-ti33x-psp-3.2/ Mar 06 01:48:44 the ones directly in linux-ti33x-psp-3.2 don't apply cleanly Mar 06 01:48:55 I just did a pull Mar 06 01:49:24 and it pulled in 3.2.7 through 3.2.9 and a couple of extra patches (2) in the main directory (linux-ti33x-psp-3.2 Mar 06 01:51:30 I'm wondering if I need to grab the 09e9651bcf2ee8d86685f2a8075b that's in the linux-ti33x-psp_3.2.bb file Mar 06 01:54:50 did you check it out at the right point? Mar 06 01:55:00 That's what I'm thinking now is wrong Mar 06 01:55:19 I checked out and reset it to the 09e9651bcf2ee8d86685f2a8075b that's in the linux-ti33x-psp_3.2.bb Mar 06 01:56:19 man, now one of the 3.2.9 patches doesn't apply Mar 06 01:56:36 git-am --skip :) Mar 06 01:56:52 are they all listed in the recipe? Mar 06 01:57:06 you know... you ask bitbake nicely to do it all for you ;) Mar 06 01:57:13 That's a good qestion Mar 06 01:57:16 question Mar 06 01:57:30 if bb does it all, then I don't learn anything :) Mar 06 01:57:54 the 00* in linux-ti33x-psp-3.2 apply now Mar 06 01:58:02 just that one in 3.2.9 didn't take Mar 06 01:58:09 are you applying all of it?! Mar 06 01:58:18 that's what the recipe says Mar 06 01:58:38 It has all the ones from 3.2.1 to 3.2.9 in order Mar 06 01:58:51 'k Mar 06 01:59:27 yeah, so 0054 (which fails) is not in the recipe (but the file exists) Mar 06 01:59:27 gar Mar 06 01:59:29 :) Mar 06 02:00:18 make am335x_evm_defconfig && make -j5 Mar 06 02:00:25 let's see if it works now Mar 06 02:00:30 blame the chef, not the recipe when things don't turn out ;) Mar 06 02:00:37 That's right Mar 06 02:00:48 One day I'm gonna have to learn something about OE Mar 06 02:01:47 so this kernel from agaro-project, does it get pulled into the mainline by someone? Mar 06 02:01:55 I had a hard time finding much evidence of that on LKML Mar 06 02:02:26 Some of the patches are in there, but I'm not seeing pull request emails for it or anything Mar 06 02:04:08 ok, so all done Mar 06 02:04:12 now I have to get a beagle bone Mar 06 02:05:47 that's the easy part :D Mar 06 02:07:48 Does anybody know exactly what the MACHINE=beaglebone ./oebb.sh bitbake systemd-image command does? Also the difference between it and MACHINE=beaglebone ./oebb.sh bitbake virtual/kernel ? Mar 06 02:08:06 yes. Mar 06 02:08:14 difference is finite and bounded. Mar 06 02:08:55 I want to be able to not just generate the uImage but the root file system as well but I only see uImage in the deploy/images/beaglebone directory Mar 06 02:09:14 ah? Mar 06 02:09:24 finite and bounded? Mar 06 02:10:40 systemd-image generates the basic console rootfs along with MLO, uboot and friends Mar 06 02:13:33 where can i find the generated files? Mar 06 02:13:43 deploy/images/beaglebone Mar 06 02:14:05 I'm in there but I see uImage Mar 06 02:14:18 and modules Mar 06 02:14:46 did the bitbake systemd-image run successfully? Mar 06 02:14:48 what is the name of the rootfs file ? Mar 06 02:15:07 the full path to mine is build/tmp-angstrom_2010_x-eglibc/deploy/images/beagleboard/ Mar 06 02:15:10 Summary: There were 5 WARNING messages shown. Mar 06 02:15:16 so i am guessing yes Mar 06 02:15:37 my path is the same but it ends with beaglebone instead Mar 06 02:15:46 systemd-image-beagleboard.tar.bz2 -> Angstrom-systemd-image-eglibc-ipk-v2012.03-core-beagleboard.rootfs.tar.bz2 Mar 06 02:16:16 ok i surely do not have that file Mar 06 02:16:34 or anything near that name Mar 06 02:16:44 no rootfs name anywhere in any of my files Mar 06 02:17:39 I'm going to run MACHINE=beaglebone ./oebb.sh bitbake systemd-image Mar 06 02:17:53 again to see if it generates it this time but I doubt it Mar 06 02:19:19 http://pastebin.com/fYXrGa0u here's what mine with MACHINE=beagleboard comes out to Mar 06 02:40:20 very different than mine Mar 06 02:40:58 it is doing it again we'll see what it endup doing this time Mar 06 02:41:23 oh gosh it says Mar 06 02:41:24 NOTE: Running task 199 of 3259 Mar 06 02:41:34 it will be tomorrow when i get the results Mar 06 02:43:38 mine took about 5 hours on vmware on a core 2 quad **** ENDING LOGGING AT Tue Mar 06 02:59:58 2012