**** BEGIN LOGGING AT Wed May 09 02:59:59 2012 May 09 03:11:19 I'm curious to know how many people are using the Cloud9 IDE to develop code on their beaglebone? I am just about to start a new project and I *think* Cloud9 will work, but was wondering if thats what most people use. May 09 03:24:29 i'm not even sure what i plan to use for my project. the connection keeps resetting when i'm trying to download Angstrom. May 09 03:24:34 so i'm kinda SOL at the moment. May 09 03:26:05 if I want to add spi_test.c to my kernel build, where is the best place to put it? May 09 03:29:24 hi everybody, i have a head screatching issue with my beagleboard, would anybody be able to help me out? May 09 03:29:46 jaykay: less screatching May 09 03:29:51 :) May 09 03:30:41 i just want to see the beautiful screen of a booting beagleboard again in my life May 09 03:31:05 jaykay: angstrom ? May 09 03:31:10 yeah May 09 03:31:18 did you check the md5? May 09 03:31:28 whats md5? May 09 03:31:48 (im pretty new ) May 09 03:32:06 strata: just for sanity check, are you downloading it from here? http://downloads.angstrom-distribution.org/demo/beaglebone/ May 09 03:32:09 their server sucks donky dicks. make sure you have the whole thing. i juts had to download it to my vps in order to get the whole thing. May 09 03:32:36 http://circuitco.com/support/index.php?title=BeagleBoard-xM#Image_Files May 09 03:32:43 this is the instructions that i followed May 09 03:33:05 in order to recreate the sd card that originally came with my beagle board May 09 03:33:16 assuming that the issue im having is related with a corrupted sd card May 09 03:34:11 jaykay: I had luck with the images here: http://downloads.angstrom-distribution.org/demo/beagleboard/ May 09 03:34:35 i actually tried that too May 09 03:34:49 Have you tried the Narcissus route? May 09 03:34:52 but i think i mess up when it comes to tarring the image May 09 03:35:06 it mentions that i must MLO file and some other files May 09 03:35:17 and they must be on the fat32 partition first May 09 03:35:49 they recommend all these fancy level converters. can't i just use op-amps and pull-up/down resistors? May 09 03:35:53 yes, mlo and uboot are separate downloads May 09 03:36:08 yeah - have you figured out the mkcard.txt script to prepare your SD card? May 09 03:36:18 they're the bootloaders May 09 03:36:52 no i havent May 09 03:37:07 can somebody please explain what i need to do after properly formatting the 2 partitions May 09 03:37:15 op-amps to do what? May 09 03:37:25 i can succesfully create an sd card with fat32 and ext3 partitions May 09 03:37:33 ds2: amp some ops I presume May 09 03:37:52 why not just dd the image and have everything created? May 09 03:38:02 and the pull ups are for after the amp'ing? May 09 03:38:09 ds2: i have some shock sensors that output up to 70v. i'm using an lm339 quad comparator to level-shift the 'shocks' to a range of 0v-2.5v May 09 03:38:19 i tried doing that, following the instructions here: http://circuitco.com/support/index.php?title=BeagleBoard-xM#Image_Files May 09 03:38:27 now i just gotta figure out how to make the bone see these :) May 09 03:38:33 but i was not successfull, even though the instructions were very straight forward May 09 03:38:36 why 0v-2.5v? May 09 03:38:36 On this page: http://downloads.angstrom-distribution.org/demo/beagleboard/ May 09 03:38:43 follow the instructions under: How to Unpack and Boot the Demo Image May 09 03:38:49 much simpelr to simply boot a linux machine from an image or CD and just dd the image May 09 03:39:00 nothing to do then May 09 03:39:00 ds2: anything higer than 3.3v fries the board? May 09 03:39:16 strata: yes but 2.5v an ideal logic level May 09 03:39:31 why not shift it to 0/3.3V and be done with? May 09 03:39:40 you are using a 339 afterall May 09 03:39:45 because i don't trust the 339 :) May 09 03:39:55 The ADC is only 1.8V... May 09 03:39:57 thanks jstearns, im going to try those instructions now May 09 03:39:59 so i'm using a resistor voltage devider on the output May 09 03:40:00 just wack a coupla diodes in then May 09 03:40:05 jstearns74: not using the ADC May 09 03:40:09 ah May 09 03:40:12 im just trying to make sure that my board is not a goner May 09 03:40:26 if you are really worry about the 339, use 4 311's May 09 03:40:29 jstearns74: it's a comparator, so the swing is ether 0v or 2.5v (or if i change it to 3.3v) May 09 03:40:42 and pull them up to the voltage rails on the bone May 09 03:41:12 unless you plan to have the shocks in rapid succession May 09 03:41:15 i'm a little confused about the whole pulling up thing though. May 09 03:41:30 no point of a pull up with the 339 May 09 03:41:45 311's make more sense if you want to go that route May 09 03:42:00 let's say i'm powering the 339 with external 5v. a simple voltage divider on the output of a 339 is ok right? May 09 03:42:20 from a static stand point sure May 09 03:42:35 but will it do the task, I donno May 09 03:42:50 so i should be powering 339 via 5v rail on the bone May 09 03:42:50 I just don't see the point of using a 339 from 5V May 09 03:42:58 use the 3.3V rail May 09 03:43:02 oh true May 09 03:43:07 i agree with ds2 May 09 03:43:09 339 will work at 3.3v May 09 03:43:19 then i don't even need the divider May 09 03:43:24 if you need the margins, i.e. due to input range, use the 311 May 09 03:43:34 for extra protection, put a diode to rail. May 09 03:44:22 if you wanna.... don't load if they aren't reqd. May 09 03:44:42 I agree, amke it up with a divider and use a diode if you are worried May 09 03:44:50 make it up I mean May 09 03:44:52 yea May 09 03:45:03 i don't know why i didn't think of that before May 09 03:45:10 dividers are a bad idea May 09 03:45:17 i kept assuming the 339 wouldn't run at 3.3v May 09 03:45:28 if you don't even want to sue resistors for the clamp just put 5 diodes in series and hang it off the input May 09 03:46:58 eh? May 09 03:48:01 just to temporarily clamp the input if it gets bigger than 3v May 09 03:48:42 or i could just pull down from 5v May 09 03:50:00 Hi guys, I have a quick question regarding unpacking and booting the demo image, it says to: Format the SD card using mkcard.txt. For example: sh mkcard.txt /dev/sdX, where X is the drive letter of the SD card. On systems like Ubuntu that would look like 'sudo sh mkcard.txt /dev/sdX'. May 09 03:50:05 but ive got 2 partitions May 09 03:50:09 fat32 and ext3 May 09 03:50:15 which one do i run this command for May 09 03:50:37 just xz -dkc Angstrom-thingy.xz > /dev/sdX May 09 03:50:55 the partition table is in the image May 09 03:51:13 they are created when you write directly to the disk May 09 03:51:17 jaykay: the mkcard.txt script assumes that you have a blank card without the partitions already created. May 09 03:51:46 ok, so i should go ahead and delete the existing partitions then? May 09 03:51:53 It will create the partitions for you. BUT sense you already have the partitions created, then just copy of the reqd files! May 09 03:52:24 ok thanks, let me get back to it! May 09 03:52:25 just copy the MLO & uImage to the FAT32 partition. May 09 03:52:32 just May 09 03:52:34 those 2? May 09 03:52:44 and then tar the demo image that i downloaded May 09 03:52:50 into the ext3 partition right? May 09 03:52:56 mlo and uboot, possibly uimage if you have an older uboot May 09 03:53:59 thurbad: uimage is not required? May 09 03:54:04 actually MLO & u-boot.img May 09 03:54:13 whatever it says on those instructions... May 09 03:54:25 ok ill copy those 3 files into the fat32 partition May 09 03:54:28 newer uboot looks in /boot for uimage May 09 03:54:32 what about the demo image that i downloaded? May 09 03:54:54 $ sudo tar -xjv -C /media/rootfs -f /path/to/Angstrom-Beagleboard-demo-image*rootfs.tar.bz2 May 09 03:55:10 where /media/rootfs is your ext3 partitition May 09 03:55:11 I'll be getting my beaglebone tomorrow. May 09 03:55:18 I'm pretty excited May 09 03:56:41 ok jstearns74, im going to give that a try right now, thanks so much! May 09 03:56:57 so I have modified a kernel and done "bitbake virtual/kernel" but it only gives me a uimage and modules-xxx.tar that just contains the /lib directory May 09 03:57:34 I am not sure how to get that on the sd card and get it going on the beagleboard May 09 03:57:50 i can build images just fine and load them May 09 03:58:00 hmmm I'm curious if anyone has tried running something like virtual box on the beagle. May 09 03:58:06 That'd be interesting. May 09 03:58:32 virtualbox would be way too slow on a beagle May 09 03:59:02 yea but still pretty cool. May 09 03:59:04 :PO May 09 03:59:08 assuming you could get it built and running May 09 03:59:34 yea May 09 03:59:38 Not saying I'll even try May 09 04:00:13 all I need to beaglebone to do is allow me to read the values of 18 GPIO pins. May 09 04:01:46 KeatonT: that's way easy. lol May 09 04:02:06 Goood :P May 09 04:02:18 i was already doing that like 10 minutes after i opened the box. now i'm in PWM/i2c/interrupts hell trying to figure out how to work with other devices like a 3-axis accelerometer. May 09 04:02:47 hey strata - what are you programming in? Python? May 09 04:02:50 Yea I'm using the board to integrate into an existing alarm system. May 09 04:02:57 jstearns74: C May 09 04:03:17 Are you developing your code natively on the BB? May 09 04:03:20 So using the bone to read the values from the magnetic door sensors/switches. May 09 04:03:24 jstearns74: pretty much yea. :) May 09 04:03:30 cool. May 09 04:03:44 KeatonT: you can do that with shell scripts May 09 04:04:11 yea actually I'm planning on making a C class to mimic the classes native to the arduino. May 09 04:04:15 my first challenge is to figure out how to count how many times an input goes high. May 09 04:04:36 for PWM? May 09 04:04:39 KeatonT: bonescript is already there. you could use that. it's VERY arduino-like. May 09 04:04:51 yea I did see that. May 09 04:04:55 KeatonT: i need a way to read the frequency of a light-to-frequency convertor. May 09 04:05:14 I also need database integration so I can keep logs. May 09 04:05:27 I'll just be happy that I'll have a way of doing remote programming. May 09 04:05:34 also i have one of those ping))) range finder thingies. it's easy to use on arduino. but i am totally clueless on how to use it with a beaglebone still. :/ May 09 04:05:34 It's always a pain to plug the arduino in. May 09 04:06:24 i'm not sure if user-space is going to be fast enough. so i need to figure out how to use the interrupts/timers/whatever from userspace. May 09 04:06:34 hacking the kernel would be fun. but not practical. May 09 04:06:58 strata I think you can only have 8 interrupts totoal May 09 04:07:08 so far i've turned up tons of stuff on google: "look what i did! oops i forgot to show people how i did it or post example code." May 09 04:08:34 lol May 09 04:08:52 strata as is the life of a software engineer. May 09 04:09:01 jaykay: you still trying to boot? May 09 04:09:28 yeah, nearly there, i had something come up, but im on it now May 09 04:09:31 KeatonT: yea. i'm giving myself 1 week to figure all this stuff out. if i can't, i'm going back to arduino for this project. lol May 09 04:09:32 thanks for followup May 09 04:09:48 I was going to tell you that if that hangs for whatever reason, try this out: http://treyweaver.blogspot.com/2010/10/installing-angstrom-on-beagleboard-xm.html May 09 04:10:07 I'm sure you'll figure it out strata. May 09 04:10:09 It was my go-to reference when I first starting making my own SD cards using Narcissus. May 09 04:10:30 Unlike everything else, it always seemed to work. May 09 04:10:31 The only real difference between the two is the arduino has been made easy via abstraction. May 09 04:11:30 yea. May 09 04:12:07 like with arduino you can do attachInterrupt and call code when an input changes state. i need to figure out how to do that on this bone. May 09 04:12:50 I'm pretty sure the bone already has kernel support for interruptsups May 09 04:12:55 i thought about using select() on an open fd to the gpio's value file, but i don't think it will be fast enough for timing May 09 04:13:17 i'm working with microseconds here. it would probably get ate up in context switch May 09 04:18:51 wonderful May 09 04:18:57 another image fail. May 09 04:19:40 guess i'll have to use linux instead of freebsd May 09 04:20:35 jstearns, why would my sd card partitions appear as /dev/sdd1p1 and /dev/sdd1p2 May 09 04:20:47 last night they were /dev/sdc1 and /dev/sdc2 like in the example May 09 04:21:19 i did a reformated the card using windows May 09 04:21:23 could that be why May 09 04:22:08 not sure... Are you using a all-in-1 card reader to access the SD card? May 09 04:22:17 maybe (?) May 09 04:23:42 if you need real time performance you probably should just go back to arduino May 09 04:24:53 jay6981 would kernel support for GPIO fix the latency issues? May 09 04:25:00 nevermind my stupidity kicked in there for a sec May 09 04:25:01 or do a driver, which is harder than just buying an arduino May 09 04:25:19 everything is good, almost there lol, im a true newbie May 09 04:25:24 strata you could always connect the arduino to the USB host. May 09 04:25:40 Then you could just send data via serial and keep the main program on the bone. May 09 04:25:44 so I just tried to build the MLO and u-boot.img using OE. I put them on the SD card int he boot partition but no lights even come on when I turn on the beagleboard May 09 04:26:00 KeatonT: LOL i thought about doing that, then feeding the data via serial into the beaglebone for processing. but that's kinda like cheating. May 09 04:26:18 ti worked before when I had the old MLO, u-boot.bin, and uImage May 09 04:26:47 strata you could always do it that way till better kernel support is added. May 09 04:27:12 the kernel certainly has GPIO support May 09 04:28:44 jay6981 that's what I thought but I kept getting mixed signals from my googling. May 09 04:28:57 serekson are you also trying to create a sd card for booting angstrum May 09 04:29:43 KeatonT: for what you're doing, everything is there. May 09 04:29:48 for what i'm doing. i'm still unsure. May 09 04:30:20 what are your timing requirements? May 09 04:30:22 the only thing I'm worried about really is triggering some 5V relays. May 09 04:30:50 Not sure if 3.3V from the bone will be enough. but I'll be quite now :P May 09 04:31:30 jaykay: yeah, I can get it to work with MLO and u-boot.bin from other sources but not the ones I build May 09 04:33:10 KeatonT: you're dealing with basic gpio on/off. some opto-isolators or darlingtons feeding the relays will be fine. May 09 04:33:12 or fets May 09 04:33:36 these relays might actually trigger with 3.3V May 09 04:33:42 just have to wait and see. May 09 04:33:53 if not I'll do the opto May 09 04:33:55 lol. May 09 04:33:57 serekson, im about to try what jstearns had told May 09 04:34:03 ill keep u posted give me 2 mins May 09 04:34:08 do not trigger relays directly from the i/o pins on the beagle :) May 09 04:36:20 awe! why not :P May 09 04:36:23 hehe May 09 04:36:40 YEEEEEEEEEEEEEEEEEEEEEEEEEEEEEES May 09 04:36:46 MY BEAGLEBOARD IS BOOTING May 09 04:36:50 I LOVE YOU GUYS May 09 04:37:03 jaykay: did you just use narcissus? May 09 04:37:10 KeatonT: what happens when you apply current to an electromagnet, and then take current away from it? :) May 09 04:37:28 I was kidding :P May 09 04:37:36 I first formatted my sd card following instructions here: http://code.google.com/p/beagleboard/wiki/LinuxBootDiskFormat May 09 04:38:11 strata: these are the relays I have anyways http://www.amazon.com/SainSmart-2-Channel-Relay-Module-Arduino/dp/B0057OC6D8/ref=sr_1_1?ie=UTF8&qid=1336538270&sr=8-1 May 09 04:38:12 then I copied the MLO and ui-boot.img files to the fat32 partition May 09 04:38:16 jaykay: you got there faster than me. i confirmed md5, wrote it, it booted once, then stopped at "uncompressing kernel" May 09 04:38:21 i'm writing again May 09 04:38:36 actually i might have spoken a little early lol May 09 04:38:42 how long does the angstrom logo screen last May 09 04:38:51 the login screen? you're booted. May 09 04:38:55 can the beaglebone provide 15-20mA of current from an IO pin? May 09 04:39:09 KeatonT: yes May 09 04:39:26 yeah how long till the login screen shows ? i currently have the compass logo of angstrom May 09 04:39:42 im glad because previously i didnt have any output from DVI May 09 04:39:44 good my relays should work then :P May 09 04:39:55 i now know its not my board thats the problem, its me May 09 04:39:58 wehre did you get the MLO and u-boot.img May 09 04:40:01 jaykay: then welcome to my world! re-writing the sd card for about the 5th time now trying to get a working image. May 09 04:40:02 first boto can take a while May 09 04:40:05 boot May 09 04:40:31 serekson, i got the files from : http://downloads.angstrom-distribution.org/demo/beagleboard/ May 09 04:41:29 arg wtf May 09 04:41:35 mad crazy kernel panics May 09 04:42:26 yea so. it boots. i got login screen. then after about 5 seconds it just started spitting page faults May 09 04:42:37 now the sd card is corrupted again May 09 04:42:42 hmm, i cant even get login screen =/ May 09 04:43:06 lol May 09 04:43:37 so if I were trying to copy the iimage of ubuntu lets say, then i would still have the MLO file and ui-boot.img file May 09 04:43:41 in the fat32 file system right? May 09 04:43:49 jaykay: just xz it directly to the device May 09 04:44:15 the files jaykay linked to work perfectly May 09 04:44:26 now just to figure out hwo to get my modified kernel on the board May 09 04:44:38 serekson, are u able to login to agnstrom? May 09 04:44:54 jaykay: yeah, its an image I built with OE May 09 04:45:07 where can i find that image? May 09 04:45:14 and what is OE? May 09 04:45:21 openembedded May 09 04:45:25 I built it from source May 09 04:45:29 aah May 09 04:45:37 too advanced for me i think May 09 04:45:41 for now :) May 09 04:45:47 wonder if anyone else is talking about this on good ole google May 09 04:46:07 google discussion is not very helpful May 09 04:46:18 because the wait can be too long May 09 04:47:05 it doesn't make sense that it worked perfectly before putting a new Angstrom image on the sd card. May 09 04:47:22 and now it just flops May 09 04:47:28 total fail imo :) May 09 04:47:51 strata do u think its the image file thats causing the problem May 09 04:47:56 or the boot files? May 09 04:48:02 starta, are you using a beagleboard or a beaglebone? May 09 04:48:05 beaglebone May 09 04:48:19 ah im using beagleboard xm revc May 09 04:48:25 http://downloads.angstrom-distribution.org/demo/beaglebone/Angstrom-Cloud9-IDE-GNOME-eglibc-ipk-v2012.05-beaglebone-2012.04.22.img.xz May 09 04:48:27 dunno if itll make a difference May 09 04:48:28 ^ that's what i'm using May 09 04:48:31 and did you download the demo for beagleboard or beaglebone? May 09 04:48:41 thurbad: beaglebone demo May 09 04:48:54 'k just making sure May 09 04:49:18 it just kernel panics like crazy right after login prompt, corrupts the sd card, and fails to uncompress linux on all subsequent boots thereafter. May 09 04:49:34 i've verified md5. it's good. i've tried writing the sd card from 3 different machines. May 09 04:50:00 ok can somebody explain what md5 is please May 09 04:50:09 what does it mean to verify md5 May 09 04:50:17 its a hash to check if the file is valid May 09 04:50:28 jaykay: it's a hash algorithm. you compute a hash on the file and check it against the ones they have listed. it's to check to see if it's not corrupt. May 09 04:50:37 how big is your rootfs partition? May 09 04:51:12 thurbad: *shrug* i'm just doing xz -dkc Angstrom....xz > /dev/whateverithappenstobe May 09 04:51:14 md5 is like a digital signature, it's used to verify that a download is complete and untainted May 09 04:51:33 i understand May 09 04:51:53 so i can see the file listed at http://downloads.angstrom-distribution.org/demo/beagleboard/ May 09 04:51:59 and how much free space do you have after it's unarchived? May 09 04:52:02 do i just download it and use it as a script May 09 04:52:55 thurbad: not sure. i'm going to attempt doing it the other way. May 09 04:57:58 strata any updates on the angstrom booting May 09 04:57:58 is the /lib directory somehow tied to the kernel but the rest of the rootfs is independent? May 09 04:58:05 u fixed it? May 09 04:58:33 jaykay: yea. gotta switch to my arch box to use mkcard.txt because fedora wants to steal the disk before the script has a chance to create the filesystem. ultimate gayness May 09 04:58:55 lol May 09 04:59:14 if it doesn't boot after this. i'm returning it as DOA May 09 05:00:15 strata: did it not come with an image on the sd card? May 09 05:00:34 mine better damn well come with an image. :P May 09 05:00:43 I think most of them do May 09 05:00:53 just dont try and power it over usb May 09 05:01:03 last time I checked, it comes with two sd card, both are bootable May 09 05:01:20 mine wasn't getting enough power via usb and it took me 2 days to figure out what the problem was May 09 05:01:24 meklort: one is stupid broken ti stuff, though May 09 05:01:27 ah May 09 05:01:37 I haven't booted either of them May 09 05:01:48 well, other than to make sure the board started up when I first got it May 09 05:02:09 what're you doing? May 09 05:02:26 custom stuff, no os May 09 05:02:51 cool, have you seen StarterWare? May 09 05:02:57 very handy May 09 05:03:05 na, I haven't May 09 05:03:20 I've just been using the tech reference manual May 09 05:03:32 for no-OS, it's great May 09 05:03:43 I'll take a look May 09 05:03:56 simple bootloader and a lightweight HAL May 09 05:04:20 k May 09 05:06:02 well, certainly looks useful May 09 05:06:26 yeah, not least because it avoids U-Boot's BS May 09 05:07:17 right now I'm just going directly from MLO to my own code (so a little bit of u-boot, but not much) May 09 05:08:03 I've built the starterware bootloader for UART boot and put it on SD May 09 05:08:13 handy for testing May 09 05:08:35 cool, might try that May 09 05:08:56 although you should be able to boot over uart w/o any sd card May 09 05:09:04 I know the boot rom checks the uart May 09 05:09:28 i don't like initializing SDRAM manually May 09 05:09:34 ah, good point May 09 05:09:52 which is the main reason I haven't gotten rid of mol just yet May 09 05:10:30 the SDRC init for the omap3 is super easy May 09 05:10:40 and thanks for the tips, I'll definitely try some of this out May 09 05:11:06 jay6981: another source of bizarre bugs? no thanks :P May 09 05:11:15 you should see the newest x86 memory controllers :P May 09 05:11:41 heh May 09 05:11:44 x86s have BIOSes (BIOSi?) May 09 05:12:02 which are board specific May 09 05:12:17 right May 09 05:12:19 you want to make a board, you have to make a bios May 09 05:12:47 right, but the interface to them is fairly standard May 09 05:13:03 serekson: yea. it did. which worked fine. but they recommended you upgrade to the latest demo. so i did. lol May 09 05:13:24 uefi is a big mess May 09 05:13:27 "we recommend that you upgrade in order to brick your beaglebone" May 09 05:13:41 never upgrade if things are working fine May 09 05:13:54 jay6981: i needed a newer image for some PWM stuff May 09 05:14:15 ok good May 09 05:17:02 i ought to throw together a debugging MLO that blinks the GPIO LEDs and echoes stuff sent over UART May 09 05:17:54 handy thing to have May 09 05:18:53 so, dont' run "opkg upgrade" May 09 05:21:11 damn, i cant get to the login screen of angstrom =/ May 09 05:21:17 im stuck again May 09 05:22:01 I can send you the image I built with OE if you have a good way to transfer it May 09 05:22:08 I also have a BB-XM May 09 05:22:09 does the fact that i can reach the angstrom logo mean that my files (MLO & ui-boot.img files) are working as expected May 09 05:22:22 hmm May 09 05:22:30 do u have a rapidshare account by any chance? May 09 05:22:33 how large is the file? May 09 05:22:40 jaykay: it's all voodoo i tell you! May 09 05:22:45 it really is May 09 05:22:54 it has taken me forever to get to be able to build the kernel May 09 05:22:56 lol :) May 09 05:23:26 jaykay: probably May 09 05:23:31 serekson: and i'm guessing as soon as you install it, it just magically works. but an official release doesn't work? May 09 05:23:42 lol May 09 05:24:01 can you upload your magic beagle wand instead? May 09 05:24:02 but the fact that he also has a bb xm gives me hope May 09 05:24:21 what works for him should also work for me May 09 05:24:27 one would think May 09 05:25:11 if putting a new distro on the sd card is this retarded, how come people aren't saying anything about it in the forums? May 09 05:25:15 I have a 13MB tarball that should work for you if you can think of a way for me to get it to you May 09 05:25:23 maybe just coincidence that both jaykay and myself got bad litter boards? May 09 05:25:32 gmail perhaps? May 09 05:25:37 by the way what is a tarball May 09 05:25:39 strata: don't you have a bone? May 09 05:25:43 serekson: yes May 09 05:25:55 something wrong with the way you're powering it perhaps? May 09 05:26:07 as in not usb May 09 05:26:11 jaykay: basically an uncompressed FS dump May 09 05:26:21 the sd card that came with the board was working fine until 2 days ago with the 5v power supply May 09 05:26:30 serekson: i'm powering it via usb when i boot the new image. (so i can access serial console) May 09 05:26:46 I had problems powering over usb when I got mine May 09 05:26:47 thanks XMP! May 09 05:26:59 that's whynyou usually see a .gz after it; gzip is an algorithm that compresses single files May 09 05:27:00 got all sorts of kernel panics, but that was on an XM May 09 05:27:21 ok. i'll try my 5v ghetto rigged 7805 power supply May 09 05:28:23 serekson can u try sending that tarball via gmail? May 09 05:28:23 maybe that's the trick. boot it first with 5v and let it do its thing. May 09 05:28:31 definitely try 5v May 09 05:28:32 the original image worked fine over usb May 09 05:28:33 dunno what the file limit is in gmail, i hope not 10mb May 09 05:29:45 maybe you could DCC it? May 09 05:30:16 yea. 12V/1A wall-wart to 7805 :) May 09 05:30:23 didn't have a true 5v supply May 09 05:30:39 the 7805 get's REALLY hot. but it works. May 09 05:31:52 jury-rigged power supply, what could go wrong :P May 09 05:32:30 did anyone tell you these boards are sensitive wrt power? May 09 05:32:47 according to the reference manual, if an overvoltage is detected, the board will turn off and refuse to turn on while overvoltage persists. May 09 05:33:05 * XMPPwocky had an "ohcrap" moment when he plugged a 12v wallwart into his bone May 09 05:33:32 XMPPwocky: that's exactly what they said in the RM. "in case you plug in a 9v or 12v supply into your bone" May 09 05:33:35 went better than expected May 09 05:33:45 yep, heh May 09 05:33:47 there's a protection circuit for that. May 09 05:34:01 believe me. i read it like 3 times before hooking up a 7805 :) May 09 05:34:22 if the 7805 overheats to the point of fail, the worse that could happen is 12v gets through to the bone and it just turns off. May 09 05:35:02 it's kinda nice too actually because i can use 12v or 5v for external components. May 09 05:35:09 Real Men (TM) just use a capacitor, two tact switches, and finger action May 09 05:36:07 jay6981: i suppose doing serial console + powering the bone from a netbook doesn't supply enough current May 09 05:38:18 the only way I got enough power through usb was using a cellphone charger May 09 05:38:32 none of my desktops would do it May 09 05:38:37 check dmesg on the host, i'm pretty sure linux's usb drivers warn when more current than can be supplied is requested May 09 05:39:08 XMPPwocky: no need for that. i got flooded by page faults and other cool kernel panicky messages via serial console. then... dead. May 09 05:39:20 you could try a double cable May 09 05:39:21 sd card corrupt. game over. please insert coin to play again. May 09 05:39:30 one that gives extra power by plugging into another port May 09 05:40:10 strata: not the bone, the pc you plug in to May 09 05:42:05 ok got new sd card image plugged into 5v May 09 05:42:14 time to hook up serial console May 09 05:44:59 some problem May 09 05:45:05 err same May 09 05:45:21 this time i didn't get a bunch of kernel panics. and was able to login and play around May 09 05:45:27 i reset.. now it just hangs. May 09 05:45:49 i'd call it a hardware problem May 09 05:45:58 May 09 05:46:06 Uncompressing Linux... done, booting the kernel. (followed by two omap errors) and hangs May 09 05:47:42 now booting with 5v again. May 09 05:47:55 (without usb cable attached) May 09 05:48:10 led 1 is lit. hangs. May 09 05:48:28 so it's good for 1 boot. then bricked May 09 05:49:24 yeeeeah, i'm calling hardware May 09 05:49:44 now led0 and led1 are high. just sitting there May 09 05:54:14 anybody know the header pin for UART3_RXD on the beaglebone? it's mux mode 2 for the spi0_cs1.gpio0_6, but I can't seem to find that on the header May 09 05:57:39 XMPPwocky: found the problem. May 09 05:58:12 XMPPwocky: it's software. the latest angstrom demo image is fail. they posted it without testing. i just found a list of people talking about this exact same problem. they screwed up. May 09 05:59:31 are there plans to fix it? May 09 05:59:39 *shrug* May 09 05:59:55 heh May 09 05:59:59 supposedly the 2012.04.09 works. May 09 06:00:11 2012.04.22 is the bad one. (which is the one they say is the latest) May 09 06:01:12 regardless of the image being bad, shouldn't the SRM say which header is the uart? May 09 06:01:44 sorry. when i say image i mean 'the whole thing' llo May 09 06:01:53 the .xz file is what i call 'the image' May 09 06:02:09 something is wrong with it. bad build or something. May 09 06:02:24 bad physical build? May 09 06:02:27 or software build? May 09 06:02:30 software May 09 06:02:31 http://comments.gmane.org/gmane.comp.hardware.beagleboard.user/31797 May 09 06:02:34 ^ see that May 09 06:05:54 hm May 09 06:06:04 not sure if that helps May 09 06:06:24 well. i guess i could go back a little further May 09 06:06:45 let's say that I am using 2012.04.09 image, which physical pin in the header would I plug a wire to connect to the uart rxd May 09 06:07:49 i have no idea. i just got this thing _today_ May 09 06:08:08 for example, UART3_RTS is pin 34 on header p8, UART3_CTS is pin 36 on header p8, UART3_TXD is pin 42 on header p9 May 09 06:08:13 ah May 09 06:08:23 i was planning on running through the docs and figuring out how to do pwm and i2c with it. but instead, i've spent the last 6 hours just trying to get the motherf**** to boot. May 09 06:09:14 lemme check which image I'm running... May 09 06:09:39 strata: 3 people out of 30k bones out there May 09 06:10:00 I'm running 2012.02.14 May 09 06:10:14 koen: ? May 09 06:10:33 07:58 < strata> XMPPwocky: it's software. the latest angstrom demo image is fail. they posted it without testing. i just found a list of people talking about this exact same problem. they screwed up. May 09 06:10:46 it wasn't posted without testing May 09 06:11:27 strata: heh, I just realized you weren't answering me May 09 06:11:45 ak_: yea it _had_ a february image on it before i updated. May 09 06:12:10 ak_: i suppose i could just go back to that and start surfing upwards until i find a kernel that's above 3.2.5 but less than 3.2.14 :) May 09 06:12:38 according to that discussion i found, they think it's 3.2.14 May 09 06:12:41 good luck? May 09 06:12:52 hi everyone, so im looking for any image that will work on my beagleboard, im currently attempting ubuntu May 09 06:13:07 jaykay: latest angstrom demo image is the problem i think. May 09 06:13:11 however, ubuntu images are gz files, is the procedure different for gz files May 09 06:14:01 yeah i think so too =/ May 09 06:14:34 jaykay: at least that's what it appears to be with beaglebone. May 09 06:14:41 you have a XM? May 09 06:14:42 strata do u know how to extract gz files onto the ext3 partition? May 09 06:14:49 yeah i have an xm May 09 06:15:43 tar -xjv -C /path/to/rootfs -f /path/to/image.tar.bz2 May 09 06:15:48 do you know how to extract images that come in the gz file format would be a better way to ask May 09 06:16:02 or if it's just a gz file May 09 06:16:19 you can do: zcat thefile.gz > /dev/device May 09 06:16:37 ok im about to give an ubuntu image a try May 09 06:16:41 lets see if ubuntu works May 09 06:16:44 no May 09 06:16:49 not at the moment May 09 06:16:52 depends on your boards May 09 06:17:01 what do u mean? May 09 06:17:31 Robert Nelson May 08 09:27AM -0500 May 09 06:17:31 > 4.6.3 (Ubuntu/Linaro 4.6.3-1ubuntu4) ) #1 Fri Apr 13 04:55:05 UTC 2012 May 09 06:17:32 > [ 0.000000] CPU: ARMv7 Processor [413fc082] revision 2 (ARMv7), May 09 06:17:32 > cr=10c53c7d May 09 06:17:33 May 09 06:17:33 This is one of the odd issues i was seeing to, it should be fixed in May 09 06:17:34 3.2.0-psp8.. May 09 06:17:45 Beaglebone freezes after login ubuntu May 09 06:18:07 i have a beagleboard xm rev c May 09 06:18:18 will i be ok? May 09 06:18:19 ok then you are probably good to go May 09 06:18:24 all I have is a bone May 09 06:22:44 ok guys i think the tar method is not gonna work here because there is .RAW file inside the gz file May 09 06:23:07 does anybody know how i can extract an image in the file format .raw May 09 06:25:59 the .raw file is just a raw disk image. you can just dd it to the sd card May 09 06:29:04 ddif=blah.raw of=/dev/sdX bs=1M May 09 06:29:18 then fdisk /dev/sdX and p to see the results May 09 06:32:06 so the ddif command does the partitionaing all over May 09 06:32:10 partitioning* May 09 06:33:55 I think it's dd if=blah.raw May 09 06:36:23 DiskDestroyer InputFile=blah.raw OutputFile=/dev/sdX BlockSize=1Mega May 09 06:58:06 Has anyone suceeded in running a newer linux-kernel (3.4 or 3.2) with the angstrom-rootfs from openembedded on a beagleboard? May 09 07:00:15 i want to see my beagleboard get passed the angstrom logo and take me to the login screen May 09 07:00:20 this will seriously make my month May 09 07:22:00 sp00k: 3.2.16 is the default kernel for beagleboard in angstrom May 09 07:59:09 hi koen, I am using openembedded to build the kernel and the rootfs (from git://github.com/Angstrom-distribution/setup-scripts.git) May 09 08:00:11 if I do a "bitbake --show-versions | grep linux" I see "linux-omap :2.6.39-r119o" as kernel version for the used kernel... May 09 08:01:49 so how do I bitbake a newer kernel version and a angstrom-rootfs that will work with it? May 09 08:02:25 opkg update? May 09 08:03:26 I would like to build a 3.x kernel from scratch and not updating it afterwards on the board May 09 08:03:35 git clone May 09 08:04:06 git clone what? May 09 08:04:12 a kernel? May 09 08:05:10 sp00k: I very much doubt current angstrom builds .39 May 09 08:05:14 its 3.x for sure May 09 08:05:20 https://github.com/koenkooi/linux/branches May 09 08:05:29 I already managed to build the newest kernel for a beagleboard, but no with all the patches from koen - so I guess its better to use the openembedded environment May 09 08:06:02 you don't now how old the image was, google offers everything from the beginning of time May 09 08:07:44 it even has an e-mail from me from jan 1992 May 09 08:08:05 @av500: just to be sure: when I build angstrom as mentioned in http://www.angstrom-distribution.org/building-angstrom I would end up with a 3.x kernel? May 09 08:08:51 yes May 09 08:14:03 av500: do you now by chance the exact kernel build version? May 09 08:14:08 no May 09 08:14:23 do you have a hint where I can lookup this version? May 09 08:14:42 in the agnstrom/oe git May 09 08:14:52 should be part of the kernel recipe for the beagle May 09 08:16:05 3.2.16 May 09 08:16:22 thats also the last mainline which works here May 09 08:16:32 (on a b*board) May 09 08:20:22 thanks, I now found it in meta-ti/recipes-kernel/linux/linux-mainline_3.2.bb (and it seems that beagleboard is using mainline) May 09 08:40:56 test May 09 08:42:31 looks like #beagle is the only irc-channel known by most people ;) May 09 08:48:06 i have build an angstrom image and the kernel is 3.0.x May 09 08:57:24 uff...i'm trying build qt with plugin-gfx-powervr and this require the source GLES/eglplatform.h that is in OMAP35x_Graphics_SDK how can i include some reference to this folder in the recipe? May 09 08:58:31 find / -name eglplatform.h May 09 09:00:48 aholler: /tmp-angstrom_v2012_05-eglibc/sysroots/beagleboard/usr/include/EGL/eglplatform.h May 09 09:01:15 in /tmp? May 09 09:01:28 didn't get installed somewhere else? May 09 09:02:06 cp -ar /tmp-angstrom_v2012_05-eglibc/sysroots/beagleboard/usr/include/EGL /usr/include May 09 09:02:14 aholler: uh yes... May 09 09:02:32 or make a symlinl May 09 09:04:01 aholler: the correct file is this /setup-scripts/build/tmp-angstrom_v2012_05-eglibc/work/armv7a-angstrom-linux-gnueabi/libgles-omap3-4.05.00.03-r27/packages-split/libgles-omap3-dev/usr/include/EGL/eglplatform.h May 09 09:05:05 aholler: i have to create a link on my /usr/lib in host ? i think i have to write something in the qt-embedded.bb recipe no? May 09 09:05:28 no May 09 09:06:02 the qt-build-script just wants stuff in usr/include/EGL May 09 09:06:23 so check out how you get the sdk to install the files there or do it by hand May 09 09:06:54 s/EGL/GLES/ May 09 09:06:58 aholler: i have the sdk just installed... May 09 09:07:10 aholler: now i try to create a link to usr/lib and rebuil May 09 09:07:13 rebuild* May 09 09:07:25 don't confuse lib with include May 09 09:07:33 yes sorry include May 09 09:08:28 the configure for qt has some other options too, so you could add the include-path in configure for qt May 09 09:08:33 in the recipe May 09 09:10:22 aholler: this same interesting...in what file? May 09 09:12:06 qt4 includes GLES support by default in angstrom May 09 09:12:11 no need to change anything May 09 09:12:38 sorry...crash May 09 09:13:27 no, crash as often as you like May 09 09:13:30 no problem* May 09 09:14:21 av500: i mean crash the irc client May 09 09:18:05 So am I insane for considering building a cape with an AVR on it, and stackable arduino-compatible headers? May 09 09:18:14 no May 09 09:18:28 It legitimately does seem to be the best way to make a Beaglebone capable of using Arduino shields - an AVR is actually cheaper than the level shifter ICs. May 09 09:18:46 nickjohnson: but TI makes good money from level shifters May 09 09:18:50 Heh May 09 09:18:51 aholler: mmm its more complicate... if i use /configure i know how can i setup the correct path...but with recipes its make me crazy i found qt4.inc and qt-4.8.0.inc but i don't understand where put some code May 09 09:19:13 Well, I'll still need to level shift between the AVR and the host 'bone, but I think the single mosfet trick will work fine for that May 09 09:19:33 for i2c, yes May 09 09:19:45 I was meaning for UART May 09 09:19:56 probably too May 09 09:19:57 I'm not sure there's any point in interfacing the two I2C buses May 09 09:20:13 what uart level is the avr? 3.3? May 09 09:20:23 I'm curious how practical it would be to add the AVR's GPIO pins to sysfs, though. I can't find any decent docs on the gpio module, though it definitely supports some sorts of GPIO expansion modules, over I2C May 09 09:20:38 av500: You can power the AVR from 1.8 to 5v, but the IOs are only 5v tolerant if it's being powered at 5v May 09 09:20:49 just make the avr replicate one of the gpio expanders protocols :) May 09 09:20:53 So I figure it makes the most sense to make it run at 5v and level-shift the uart May 09 09:20:55 or add another one May 09 09:21:17 LetoThe2nd: Right, but I need to figure out what those are, and if there are any that work over UART serial, and how to add them to the kernel. I suspect it requires a recompile May 09 09:21:35 Anyone know where I can find the beaglebone-specific kernel source repository? May 09 09:21:39 just use i2c May 09 09:21:45 yes, recompile is needed May 09 09:21:52 also writing the kernel driver for it May 09 09:21:53 afaik you need pdata set, so yes, -> recompile May 09 09:22:03 av500: Of the whole kernel, or just the module? Presumably the latter... May 09 09:22:10 depends May 09 09:22:16 LetoThe2nd: It sounds like you know more about it than me - where do I look? May 09 09:22:42 aholler: That would cause problems if you want to use an Arduino shield that uses the I2C lines (or the analog lines that are multiplexed with them) May 09 09:23:00 bitbang i2c on the avr May 09 09:23:15 hey i have a question about gpio. is there a better way to do this: http://pastebin.com/kUHEDDwD May 09 09:23:18 and on the beagle, using sysfs :) May 09 09:24:02 aholler: Again, I can't do anything that would use pins commonly used by Arduino shields May 09 09:24:06 sysfs, the new kernel-api ;) May 09 09:24:09 nickjohnson: maybe something like http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=blob;f=drivers/gpio/gpio-max7300.c;h=a5ca0ab1b372cb419a73cc7fd8101258e311c76e;hb=HEAD May 09 09:24:18 nickjohnson: and the platforms using it, repectively. May 09 09:24:31 that code is to run one of those parallax PING))) range finders. it's working. but there's some hiccups in the latency (sometimes it's off by about 50-60 usec) May 09 09:24:40 nickjohnson: is avr gpios over sysfs really worth it? May 09 09:24:45 Pretty much all I can do is use D0 and D1 as a UART to the bb, since all the arduino clones use the hardware uart that way May 09 09:25:00 strata: you dont get high precission over sysfs May 09 09:25:05 av500: I'm just thinking about how I can make it as easy to use as possible for people who just want to use the hardware, but don't want to program an arduino May 09 09:25:16 Whether it's worth it depends on how hard it turns out to be ;) May 09 09:25:25 LetoThe2nd: thanks May 09 09:25:25 well, providing a small lib that talks to arduino over uart May 09 09:25:43 av500: how can i get higher precision with gpio? May 09 09:25:43 is as good as providing a smalle lib that uses sysfs May 09 09:25:49 av500: Right, I can do that too. But people already know how to use sysfs May 09 09:25:50 strata: kernel driver May 09 09:25:58 nickjohnson: you'd be surprised :) May 09 09:26:22 As you know from my argument yesterday, I'm not a huge fan of sysfs, but it's hard to beat if you just want to poke at individual pins on a casual basis May 09 09:26:34 It's also more or less language agnostic, unlike any library I write :) May 09 09:26:42 true May 09 09:26:51 av500: is there better way to do it from userspace? or do i have to start writing linux modules? lol May 09 09:26:56 maybe May 09 09:27:32 LetoThe2nd: Hm, doesn't tell me much. What I'm curious about is where/how you configure an individual platform's GPIOs May 09 09:27:46 i'm actually surprised how precise it is from userspace. i expected it to be alot worse. i might just roll with this. May 09 09:27:57 strata: You can use /dev/mem. After you spend an hour arguing with people here about whether that's a hack or not ;) May 09 09:28:03 the value only quirks out maybe 1/50 times May 09 09:28:14 there is nothing to argue May 09 09:28:29 aholler: its always the ecaps, you know. May 09 09:28:30 nickjohnson: using /dev/mem is bypassing like 20 additional lines of code? May 09 09:28:41 there's still context switch overhead May 09 09:28:46 even sysfs is bad. userspace shouldn't be able to kill the hw May 09 09:29:08 strata: Yes, but there isn't the entire VFS API May 09 09:29:24 I doubt that add must overhead May 09 09:29:25 ok question. is there a way to frequency count pulses on either a pwm line or gpio line through sysfs? May 09 09:29:38 nope May 09 09:29:45 I really have to benchmark it to prove you all wrong (or right). First I have to find a uSD card, though May 09 09:30:00 strata: you could do stuff like make the kernel non tickless and increase HZ May 09 09:30:09 and run your stuff as SCHED_FIFO May 09 09:30:14 that can help May 09 09:30:18 av500: ok so you're saying if i want to take servo input, i haz to write kernel drivers :) May 09 09:30:24 its still no garantee for real time May 09 09:30:38 strata: You can read interrupts using sysfs (apparently) May 09 09:30:49 nickjohnson: oh really? May 09 09:31:00 And you may be able to set interrupts on the PWM functionality (I'm not familiar enough with the beaglebone architecture to say for certain) May 09 09:31:11 nickjohnson: would an interrupt in sysfs give me a running count of how many times a pin went high/low ? May 09 09:31:43 My mistake, not sysfs: http://bec-systems.com/site/281/how-to-implement-an-interrupt-driven-gpio-input-in-linux May 09 09:31:50 strata: No, but it would tell you each time the pin changed May 09 09:32:07 strata: no May 09 09:32:10 nickjohnson: hmm... that might work for what i'm trying to do. (light-frequency converter) May 09 09:32:16 koen: I have ogre 1.8 rc1 working in oe-classic with GLES, but only if I disable FrameBuffer Objects. Do you know/recall if the version in 2011.03-maintenance has some shortcomings? May 09 09:32:17 no May 09 09:32:37 strata: if you need garanteed timing, a kernel IRQ handler is needed May 09 09:32:44 strata: If you're designing hardware, though, I think you'd be better using an MCU to do that work for you, and talk to it over a serial protocol to ask for results May 09 09:32:47 tasslehoff: apart from being out of date, no May 09 09:32:55 unless its slow enough for user space and you can tolerate the occasional glitch May 09 09:33:15 You could potentially miss interrupts entirely if you're not scheduled for long enough. Depends on how critical that is to you. May 09 09:33:22 yep May 09 09:33:35 wow May 09 09:33:54 if you don't need a kernel, don't use one May 09 09:34:21 my project is somewhat of an alarm system. i have several different kinds of sensors. i made an err assumption that i could re-implement the project on a beaglebone. (it's already running on arduino) May 09 09:34:36 if you need one, he needs time to do stuff. usb, networking and such don't run by themself May 09 09:35:11 so a better idea would be to use the arduino to grab all the sensor data and feed it into the beaglebone via serial? May 09 09:35:38 strata: How are you interfacing your sensors with the beaglebone? May 09 09:35:56 strata: thats one option May 09 09:35:58 strata: That's a plausible option, yes May 09 09:36:17 But if you're designing a cape to interface with your sensors already, it may be easier to add a low power MCU, like an ATTiny, to do the work May 09 09:36:22 aholler: i copy file in /usr/include but nothing change GLES/eglplatform.h: No such file or directory May 09 09:36:23 the other one is that all the arduino-replacers get together and write some nice kernel code for the benefit of all May 09 09:36:24 nickjohnson: so far i've managed to get some of them working. (a PIR motion sensor, an ultrasonic ping device) May 09 09:36:53 av500: they would need to know how linux works ;) May 09 09:37:06 aholler: its not a secret May 09 09:37:18 but it looks like one May 09 09:37:31 nickjohnson: i like the idea of using the ATTiny on a cape :) May 09 09:37:37 and all of them have a deadline of 2 weeks May 09 09:37:44 aholler: but i found that the file is in buildimage /usr/include (maybe i have to setup correct link to this file not the file in my host) May 09 09:37:44 aholler: best place to hide stuff is in the open May 09 09:37:48 av500: They (we?) seem to fall into two categories. Some people really seem to like the idioms of the Arduino - things like 'setup' and 'loop' and 'digitalWrite'. Others, like me, just like the ease of use of that platform. Personally I really like the idea of writing my embedded stuff in Python, using a Pythonic API, rather than something pretending to be an arduino. May 09 09:38:15 strata: I built a motor driver shield for the Arduino using an ATTiny2313, and I can vouch for that approach - it works really well. May 09 09:38:24 using python is insane ;) May 09 09:38:46 Depends on your clockspeed ;) May 09 09:38:52 ok so my plan has thus changed. i can still use the beagle for processing the data May 09 09:39:12 python is broken by tabs May 09 09:39:19 and multiprocessors May 09 09:39:23 Oh dear May 09 09:39:23 -> #language_wars May 09 09:39:34 * nickjohnson is not going to get into an indentation debate May 09 09:39:36 in fact. i will still be making use of code on the beagle and some of its gpio's May 09 09:39:36 aholler: sorry, but boring May 09 09:39:43 so not all is lost! May 09 09:42:52 that is no language war. May 09 09:43:11 aholler: it is still booring May 09 09:43:17 boooooring even May 09 09:43:23 using python to process something in "somewhat" realtime is just stupid May 09 09:43:40 depends on the definition of "somewhat" May 09 09:43:45 not that I advocate python May 09 09:45:25 i think i have found problem...maybe i have to apply some patch May 09 09:45:40 applying patches has solved many problems May 09 09:45:40 like 002_pvrqwswsegl.c.patch May 09 09:45:58 the problem is i don't know how can i do this :) May 09 09:49:07 for now i trying manually appling patch...but if work i have to understand how can i make bitbake doing this May 09 09:49:16 11:12 < koen> qt4 includes GLES support by default in angstrom May 09 09:49:17 11:12 < koen> no need to change anything May 09 09:49:21 like I said half an hour ago May 09 09:49:46 ooh sorry but without name i haven't see the message May 09 09:50:38 Hm, well it seems that it's possible to implement support for a GPIO expander in a kernel module, at least May 09 09:50:43 koen: the problem is i try to build plugin-gfx-powerv and this fails beacause can not fine header May 09 09:50:50 So I could provide a module people could load, rather than having to patch the existing GPIO support May 09 09:51:52 koen: but for correct build some patches are need and i don't think this patches are just applied on bitbake May 09 10:01:42 koen: ok. itching to try out the latest version to see if it works there, but the instant lockup is still in da house, I guess? May 09 10:03:35 tasslehoff: yes May 09 10:04:19 Oh, that's nasty May 09 10:04:46 3 GPIOs are reserved so the kernel can read the (3-bit) hardware revision number at runtime, and know what devices to configure May 09 10:05:02 you can free them afterwards May 09 10:05:52 av500: Aren't they permanently hooked up high or low, though? May 09 10:06:07 I would bet there are resistors May 09 10:06:40 so you can still use them as e.g. outputs if the connect hardware does not pull them in the wrong way May 09 10:06:48 True May 09 10:06:59 Still seems a bit nasty. Isn't this what the onboard EEPROM is for? May 09 10:07:03 e.g. ti used to have sysboot pins on LCD pins May 09 10:07:24 nasty is in the eye of the beholder May 09 10:08:24 Well, it won't scale beyond 8 revisions, and it seems like a particularly inefficient way to discover information about the hardware you're on May 09 10:09:14 but then its easily read from early code, like MLO May 09 10:09:35 there are always tradeoffs May 09 10:10:59 one man's nasty is another man's nice May 09 10:12:32 Going back to GPIOs, although support for individual expanders is module-based, things like GPIO offset and address are configured at compile time. So, probably no simple kernel module loading after all. May 09 10:14:00 what address would that be? May 09 10:14:18 in case of uart gpio? May 09 10:14:42 nickjohnson: if you are making a cape, you would want it to be supported by the bone kernel May 09 10:14:52 no? May 09 10:14:54 nickjohnson: beagleboard != beaglebone May 09 10:15:02 Well, no address in the case of UART. But they need a start ID for the GPIOs, at least May 09 10:15:10 use 1000 May 09 10:15:15 :) May 09 10:15:16 koen: I didn't think they were May 09 10:15:27 12:04 < nickjohnson> 3 GPIOs are reserved so the kernel can read the (3-bit) hardware revision number at runtime, and know what devices to configure May 09 10:15:39 koen: both are not arduino, thus the same :) May 09 10:15:41 av500: And kernel support's only needed if there's something specifically required. SYSFS GPIOs would be handy, but they're not essential May 09 10:15:59 koen: Oh. I assumed the 'beagle' platform config covered both. Where can I find the beaglebone's one? May 09 10:16:24 av500: ok so i got the arduino interfaced to the beaglebone nicely. if i use a usb hub, would i be able to also plug in a mass storage device to the same port (usb user port) ?? May 09 10:17:03 av500: Anyway, weren't you the one who said "av500> nickjohnson: is avr gpios over sysfs really worth it?" May 09 10:17:13 strata: yes. May 09 10:17:20 cool :) May 09 10:18:20 nickjohnson: right May 09 10:18:47 nickjohnson: if the only connection is the uart, then just a lib might do fine May 09 10:18:54 and python bindings :) May 09 10:18:59 and java one May 09 10:19:02 ones May 09 10:19:37 heh May 09 10:19:58 Another option would be to interface the I2C bus as well, with jumpers to disconnect it May 09 10:20:09 Then I can emulate a commonly available GPIO expander chip May 09 10:20:27 well, then you can just *use* commonly available GPIO expander chip May 09 10:20:28 Or you could talk to I2C enabled Arduino shields directly if you wanted May 09 10:20:36 +a May 09 10:20:46 av500: Except they don't emulate the full range of things a shield might require May 09 10:21:14 well, then the stock i2c extender driver is no use May 09 10:21:26 so what would be the point? May 09 10:22:58 av500: Two use cases. Person 1 wants to use Arduino shields, but not write Arduino code. They can use the AVR like a GPIO expander and talk to (most) shields directly. Person 2 wants the AVR as an MCU they can program to do low level tasks for them. They can replace the GPIO emulator code with their own. May 09 10:25:40 what's a gpio-emulator? you first need to define that. people do all kind with them May 09 10:25:58 i2c gpio expander May 09 10:26:03 aholler: Code for the AVR that makes it emulate a commonly available GPIO expander over I2C May 09 10:26:13 * nickjohnson ponders also hooking up ICSP, so you can program the AVR from the beaglebone directly, without bootloader May 09 10:26:59 I thought he wants to use arduiono-shields May 09 10:27:20 so he wants to emulate an arduino May 09 10:27:35 nickjohnson: you would still face the issue that by using a shield with this "emu" setup, you lose the arduino "timing" May 09 10:27:38 aholler: Yes - I'm just suggesting that with a little hardware support, the cape can ship with code that emulates a GPIO expander, so if you just want the hardware capabilities of a shield, you can talk to it easily. May 09 10:27:56 av500: Right, which is why people still retain the ability to program it themselves May 09 10:29:14 But if all you want to do is attach, say, a capacitative touch shield, you can, without having to program the AVR May 09 10:29:41 Though the more I look at commonly available shields the more I realise many of them just come down to something that communicates over UART, I2C or SPI May 09 10:30:12 GPS shield, cellular shield, and so on May 09 10:31:03 Though not universally May 09 10:31:08 reinvent usb May 09 10:32:26 heh May 09 10:32:39 The point is to take advantage of existing hardware modules, not invent new ones :) May 09 10:33:30 but you need to take advantage of existing hw May 09 10:33:37 s/hw/sw/ May 09 10:34:29 most hw doesn't work without sw. May 09 10:36:57 mdp: sorry, I edileted by accident May 09 10:36:58 right May 09 10:37:21 But for all those shields that do just use one of those serial protocols, a really dumb board that just has level converters on those pins alone would suffice May 09 10:38:17 yes May 09 10:38:31 yes, a level-shifting cape would be nice for all those folks out there May 09 10:40:05 aholler: That was my original intention, but I realised that level shifting all the digital lines requires more dollars in level shifters than an AVR costs May 09 10:40:14 And that's before you deal with the analog pins May 09 10:40:46 have a look at sparkfun 8745 May 09 10:41:01 So I think a best-of-both-worlds cape would have an AVR hooked up in an Arduino-compatible fashion, and jumperable connections for all the serial protocols so you can talk to shields (and/or the AVR) directly if you want May 09 10:41:52 aholler: Yup, that's the mosfet-type level converter I'll use (the design of) for the serial lines. But you can't really hold GPIO lines high when they're unused and still claim you're entirely compatible when you don't know what they'll be used for May 09 10:43:42 arduino-shields aren't universal-shields. May 09 10:44:12 I didn't claim they were May 09 10:44:26 But my goal is to make it easy to interface with as large a subset of them as possible May 09 10:45:00 Because there's an enormous variety of useful, readily available, already tested hardware modules out there that happen to conform to the arduino shield form factor May 09 10:45:46 I like breadboards ;) May 09 10:46:31 So do I, but my tolerance for wiring stuff up on protoboard is very low, so I tend to buy existing items, or order my own PCBs. :) May 09 10:46:44 And breadboards aren't very robust ;) May 09 10:50:35 usb2i2c: http://codeandlife.com/2012/02/22/v-usb-with-attiny45-attiny85-without-a-crystal/ May 09 10:50:38 ;) May 09 10:51:40 hi all May 09 10:52:10 I have a problem with beaglebone. I am trying to load selfcompiled u-boot threough jtag. May 09 10:53:12 Everything is ok if the SD card is in the beaglebone. But if i remove the SD card, power off the beaglebone and power on, connect to JTag load u-boot. U boot do not start May 09 10:53:47 Is it possible to load u-boot to beaglebone through JTag, if no SD card is inserted? May 09 10:54:22 sure, but you need to initialize the memory May 09 10:54:24 I think, the rom does some initialization, if it find MLO file on SD card. Is it true or am i missing something? May 09 10:55:18 the mlo does that May 09 10:55:34 sdram is usable from the beginning, but not much else May 09 10:55:41 s/sdram/sram/ May 09 10:56:05 rom loads mlo to sram, mlo initialises sdram, mlo loads u-boot to sdram May 09 10:56:14 av500, np, I didn't want to leave mine floating out there without the one I responded to :) May 09 10:59:19 MLO is an ELF file right? If i compile u-boot it creates u-boot.bin. u-boot binary and MLO file. If i try to load MLO file through JTag in ccsv5, it does not go through verification data. May 09 11:00:10 no May 09 11:00:23 omap3 cpu has no elf parser inside May 09 11:00:26 mlo is just a binary blob, if I remember correct May 09 11:00:30 check the trm May 09 11:00:50 yes u are right May 09 11:00:53 yep May 09 11:00:56 matus_: load into what? May 09 11:01:06 if you load it to sdram, you need to setup sdram May 09 11:01:10 before May 09 11:01:11 From JTAG you don't need to use signed image "MLO", you can directly load spl elf file May 09 11:01:13 aholler, however, it's not a binary blob that attacks May 09 11:01:18 it will load it to sram May 09 11:01:36 but how to setup sdram first? May 09 11:01:42 I used to do same thing during chip bringup May 09 11:02:15 I think openocd should have something to do that. But of course mlo contains the stuff too ;) May 09 11:02:33 matus_, it runs from sram, no need to set up sdram before that May 09 11:02:36 aholler: or a GEL file from CCS May 09 11:02:49 or you bdi config file May 09 11:02:50 I don't know anything about ccs May 09 11:03:10 aholler: keep it that way May 09 11:03:10 hi hvaibhav May 09 11:03:14 just that it's windows-soft which I try to avoid ;) May 09 11:03:22 its for linux too May 09 11:03:26 Ok thank you for now, keep trying May 09 11:03:28 it eclipse in a blanket May 09 11:03:31 steps would be, use spl elf file to load it to sram using ccs. CCS file will set the PC to start of spl and then you can execute it. It will initialize the SDRAM and then using CCS again you can load u-boot elf file and run it. May 09 11:03:33 it's May 09 11:03:36 hi mdp May 09 11:05:43 mdp: how are u? May 09 11:07:25 hvaibhav, I am almost awake now :) how are you today? May 09 11:08:10 fully awake ;) May 09 11:08:16 heh May 09 11:33:50 florian: ucross, discuss May 09 11:34:28 * florian looks as av500 May 09 11:34:34 at May 09 11:34:43 * av500 stares back May 09 11:35:18 http://www.mucross.com/ May 09 11:35:46 * florian feels guilty May 09 11:37:12 florian: is it oe based? May 09 11:37:19 av500: yes May 09 11:37:32 ah, ok, you are absolved May 09 11:37:54 everything else would be... well... evil at least :) May 09 12:01:04 * [IDC]Dragon summons _av500_ May 09 12:04:42 too bad May 09 12:04:46 he is asleep May 09 12:05:07 <[IDC]Dragon> ok, I'll wait until he's awake May 09 12:05:18 or you ask me :) May 09 12:07:06 <[IDC]Dragon> see PM May 09 12:12:47 ...asking for the exact steps... :) May 09 12:24:59 hm, does anyone know where does this mess come from? http://fpaste.org/yuek May 09 13:00:53 hi May 09 13:01:41 I strucked up to enable ethernet on beagle Board Rev C4 May 09 13:01:59 can any body give some pointer for this... May 09 13:02:21 How should I connect USB-ethrenet dongle to Beagle Board May 09 13:03:01 I am ruunging Ginger Bread Android over the BB May 09 13:09:52 Did you tried latest linaro android images for BB? or what image are you using? May 09 13:13:14 av500: y u not givin a real life example?!? oneeleven! May 09 13:13:48 LetoThe2nd: hes a bit of a pest May 09 13:14:05 av500: agreed. May 09 13:14:41 will DT help here? May 09 13:15:20 mdp: a well placed falling tree can help in many things May 09 13:49:49 hmmm, removing first boot makes the first boot faster May 09 13:49:52 I have Code Sourcery configured on a Windows host, everything work well, and I can debug. Now, I want to have outputs (as printf) on the main tty (/dev/tty1). Is it possible? May 09 13:50:21 (I precise that the tty is on the Beagleboard) May 09 13:50:40 it's software, anything is possible :) May 09 13:50:50 so? :D May 09 13:51:09 you can "debug" how? May 09 13:51:18 Crofton|work: going directly to second boot, clever May 09 13:51:31 Crofton|work: thats called snapshot boot May 09 13:51:37 e.g. Sony does that on some TVs May 09 13:51:45 boot, then freeze that to disk/nand May 09 13:51:53 yeah May 09 13:51:56 makes sense May 09 13:51:58 we do that on the e100 May 09 13:52:07 the master sd card already went through first boot May 09 13:52:10 av500: on my host, I have Sourcery installed. When I click on "debug", it lauches gdbserver on the Beagleboard May 09 13:52:22 Crofton|work: ah, thats to get away with all your postintall May 09 13:52:29 yeah May 09 13:52:37 it has gnome, so first boot is insane May 09 13:52:44 lpi: sorry, but Sourcery support is not our strength May 09 13:52:50 s/first boot// May 09 13:53:09 maybe ask a wizard May 09 13:53:35 or sorcerer May 09 13:53:51 codesourcery people used to have such titles on their business cards May 09 13:53:59 av500: sourcery is almost identical to Eclipse. Nothing change but the glossy interface :) May 09 13:54:06 eeeew May 09 13:54:33 av500: so if you know how to do on Eclipse, it will probably work on Sourcery :) May 09 13:54:53 luckily I don't May 09 13:55:17 * koen neither May 09 13:55:26 maybe you know how to pass parmaters to gdbserver when it is called? May 09 13:55:50 * mru suggests adding some printfs instead May 09 13:56:44 mru: if I ask, it's there is a good reason: using ncurses ^^ May 09 13:58:31 so print to a different console May 09 13:58:49 how do you do that? May 09 13:59:11 and which console are you talking about? May 09 13:59:30 simplest is to log debug to stderr and run the app with 2>/dev/tty May 09 13:59:59 or /dev/pts/ May 09 14:00:18 you can send it to an ssh session like that May 09 14:00:22 or log to a file May 09 14:00:42 gdb is launched by the host, not on the beagleboard. I can't give paramaters to gdb that is executed on the board :s May 09 14:01:20 so don't use gdb May 09 14:01:24 hint: gdb can be launched on the board too May 09 14:01:29 I tried to add tty=/dev/pts/0 (that correspond to my ssh session) to .gdbinit that is on my host, but doesn't seem to work May 09 14:01:30 usign the command "gdb" May 09 14:01:31 interactive debugging is useless anyway May 09 14:02:13 av500: so you mean that I start manually gdbserver on the board with parmateres I want, and then start manual debug on the host? May 09 14:02:29 no need for server May 09 14:02:38 gdb gdb May 09 14:03:29 is it really handy to see variables, disassembly, etc? Eclipse give a nice interface to see all this stuff in same time May 09 14:03:39 wanna to use it May 09 14:04:39 why do you want to see that? May 09 14:04:53 its like matrix characters scrolling down the screen May 09 14:04:57 looks fancy May 09 14:05:05 well, disassembly can be useful to look at May 09 14:05:11 but you don't need to run the code to do that May 09 14:05:25 av500: I think you're onto something May 09 14:07:08 av500: it's nicer to work on this kind of interface, isn't it? :D http://www.hostingpics.net/viewer.php?id=148590Sanstitre.png May 09 14:08:08 eeeew May 09 14:08:15 what _is_ that horror? May 09 14:08:32 lpi: you debug hello.cpp much? May 09 14:09:04 lol I'm trying to setting up environment yet, final code will not be a simple hello world :) May 09 14:09:42 whar do you suggest? May 09 14:10:33 concentrate on the code May 09 14:10:35 not the debug May 09 14:10:51 worry about bridges as you burn them May 09 14:13:13 av500: really need that stuff, sorry... May 09 14:13:25 nonsense May 09 14:15:39 hi #beagle. i have realtime scheduling pain at the moment. anyone have experience with portaudio+ALSA on the beagleboard? I'm running 3.2.16 with CONFIG_PREEMPT_RT_BASE. often my portaudio callback just stops happening, seems to be related to shifts in priority, which are observable via TOP. May 09 14:15:40 I haven't 10 years of coding behind me, is it so much incomprehensible to need this kind of tools? May 09 14:15:56 we can go back coding with vi otherwise ^^ May 09 14:17:07 ed and assembler is all you need May 09 14:17:56 magnetised needle and a steady hand... May 09 14:18:26 lpi: if you have 10 years experience, it is indeed incomprehensible that you still think you need these fancy tools May 09 14:19:13 my favourite debug tool is still a scope May 09 14:19:15 then a dmm May 09 14:19:20 after than printk May 09 14:19:27 or printf May 09 14:19:36 I haven'T 10 years of coding behind me, is it so much incomprehensible to need this kind of tools? May 09 14:19:37 yeah, software can never tell you what _really_ comes out those pins May 09 14:19:38 mru: he said "I haven't" May 09 14:19:47 oh right May 09 14:19:50 :) May 09 14:20:00 well then, that's good May 09 14:20:18 changing 10 years of bad habits would be hard May 09 14:20:40 * lpi was waiting for this kind of answer May 09 14:21:59 I personally don't see anything wrong with good visual debuggers. I love visual studio's debugger when coding c#. but then again, I have never actually longed for a debugger when doing kernel stuff. May 09 14:22:23 tomba: sure May 09 14:22:39 a good gui debugger can be useful May 09 14:22:42 its the worry about such a tool before a single line of code written that I dont get May 09 14:22:44 but it is never essential May 09 14:22:56 premature optimization.... May 09 14:23:27 of course, it helps lauterbach sales.... May 09 14:23:28 or put differently, a good debugger is sometimes useful, regardless of the user interface it comes with May 09 14:23:33 av500: ah =). I didn't read the whole discussion. it's much nicer to comment when you don't know the whole issue. May 09 14:23:46 tomba: most time I can't be bothered either :) May 09 14:23:54 lpi: use SPI May 09 14:24:17 av500: joking? May 09 14:24:38 http://26-26-54.hardwarebug.org/91 May 09 14:25:36 * lpi will never understand english spoken humor... May 09 14:26:28 av500: http://www.heberger-image.fr/data/images/41092_67929_menottes_fouet.jpeg May 09 14:26:35 :D May 09 14:27:15 just type help in gdb May 09 14:27:29 lpi: pas de probleme May 09 14:28:03 av500: english please :O May 09 14:28:08 ^^ May 09 14:31:34 if I build a new kerenel with OE what do I need to change on an SD card to update to that kernel? May 09 14:31:47 zImage May 09 14:31:51 and the modules if any May 09 14:32:18 In fact, just need to know one thing: when you remote debug using Eclipse, the .gdbinit that you give in "debugger" tab is passed on the host gdb, or on the Beagleboard gdb? May 09 14:32:27 so I can just build the systemd-image and then update the uImage? May 09 14:32:42 if I have just enabled SPI May 09 14:32:47 lpi: no idea May 09 14:32:56 sounds like an #eclipse question to me May 09 14:33:05 serekson: most likely May 09 14:33:19 you might need to update the modules if the kernel changed in version May 09 14:33:40 av500: thx May 09 14:34:11 and the modules are in /lib? May 09 14:34:11 gdb has documentation May 09 14:35:10 serekson: lib/modules/foo May 09 14:35:14 ls will tell you May 09 14:36:04 aholler: not looking for gdb doc, but on what it corresponds in Eclipse. I'd already found the command I need to use, just need to know how I can pass it May 09 14:38:31 still looks like an eclispsy kind of question May 09 14:38:34 guys my understanding is that part of the purpose of the 'oebb.sh update' command is to programmatically do a git pull on all the repos in ./sources, is this correct? May 09 14:39:18 read the script and see what it does May 09 14:39:43 mru: I'm wanting confirmation, I've already looked May 09 14:40:13 I'm asking because I couldn't find the linux-mainline-3.2 recipe even after doing a fresh git pull and oebb update May 09 14:40:56 so i manually checked meta-ti and sure enough i was behind head by 5 commits. pulling in there got me the stuff I wanted May 09 14:41:44 then the script clearly didn't do that May 09 14:42:06 obviously May 09 14:42:32 but did it have a good reason that I missed May 09 14:42:45 * stuk_gen is throw his head on the wall May 09 14:43:55 that I don't know, but now there's enough information here for someone who does to comment May 09 14:44:57 spacecolonyone: are you sure? May 09 14:45:08 spacecolonyone: it has linux-mainline here just fine May 09 14:45:14 I think I just figured it out, just tried another git pull and I get one thing: bump meta-ti for beagleboard 3.2 fixes May 09 14:45:22 from you koen May 09 14:45:38 so after more operations for build qt with powervr plugin i have apply manually 2 patches to qt source and all works well and build succesfully. After i run Angstrom and try run an application with -display powervr but error says powervr: driver not found May 09 14:45:40 ty May 09 14:46:17 so copy the folder gfxdriver to /usr/lib/qtopia/plugins May 09 14:46:39 awesome, thanks I got that working. Now I want to compile the spi_test.c program to test, is there a recommended method? I have a working OE environment so I know I can use bitbake or I can compile on the beagleboard but I am wondering if one is prefered May 09 14:47:09 and this contains the libqgfxpvregl.so...but always powervr cannot find driver :( May 09 14:50:33 any suggest? May 09 14:56:51 where's boris houdleroy to reshare g+ post when you need him? May 09 14:58:05 Can anyone point me to the OMAP3530 manual which tells me how much current draw should expect on VDD_PLL1 and VDD_PLL2? May 09 14:58:18 boris is coding May 09 14:58:21 Squintz: depends May 09 14:58:33 koen, coding in that badass rockstar technology :) May 09 15:00:40 What I really need to know is how small can I make those traces? I need to make room to route something on my board and I don't think I need a huge trace for those but I am not sure. May 09 15:01:20 maybe it is stupid to ask but is there any device/component to use beaglebone debugging, like MCU debuggers? May 09 15:01:20 Ellison Wonderland May 09 15:01:34 dsq: jtag May 09 15:02:18 like this http://www.olimex.com/dev/images/AVR/AVR-JTAG-USB.jpg ? May 09 15:02:19 So do those of you using angstrom and a console on the beagleboard mostly stick with busybox? May 09 15:05:12 can anyone suggest me jtag on digikey or farnell? May 09 15:06:18 xds100v3 May 09 15:06:40 thanks koen May 09 15:11:18 so I think I need to create a systemd service for bubsybox-udhcpc. by default it points to null and I'm not having much google love. The thing I really don't get about this is why it doesn't seem like this is a more common issue: baseing console system + usb ethernet not coming up automatically May 09 15:12:17 koen: what, not the Lauterbach? May 09 15:13:26 spacecolonyone: the default angstrom images will all bring up network on boot, provided you have dhcp May 09 15:13:52 av500, digikey/farnell distribute lauterbach? May 09 15:14:23 dunno, they just call us :) May 09 15:14:54 I think dsq wanted something available at those disties May 09 15:15:20 otherwise I would push flyswatter2 for prpplague :) May 09 15:15:49 koen: that has not been my experience. I have yet to bitbake one of the basic images which does not require me to issue 'ifup eth1' after login May 09 15:15:58 mdp: maybe May 09 15:16:19 mdp: I shall trys to order "2 pounds of jtag" from my local grocer... May 09 15:16:22 -s May 09 15:16:30 spacecolonyone: systemd-image brings up my network automatically May 09 15:17:14 I am trying to boot beagleboard rev c4, nothing much comes on console May 09 15:17:45 I think it is not the case of console output going to wrong serial May 09 15:17:48 av500, thinly sliced, please! May 09 15:18:02 last message is "Uncompressing Linux... done, booting the kernel. May 09 15:18:23 * mdp is on a webex meeting where they are showing the infamous yocto video :) May 09 15:18:43 serekson: I'm going to go retrace my image recepie chain. I've got a custom image that adds in a few things by default based on console-image May 09 15:20:00 kernel is from github.com/beagleboard/kernel.git branch beagleboard-3.0 May 09 15:20:19 What could be going wrong May 09 15:20:35 spacecolonyone: the console image did not bring it up automatically when I was running that so it seems consistent May 09 15:20:47 kernel is built with omap2plus_defconfig May 09 15:22:49 but wait serekson: condole-image inherits console-base-image which is just a symlink to systemd-image May 09 15:22:55 av500, thank you for translating to proper .usian units May 09 15:24:25 mdp: hehe, its "2 Pfund" here too May 09 15:24:27 spacecolonyone: I have only had my board working for about a week so I am just speaking from my limited experience May 09 15:24:35 a "Pfund" being 0.5kg May 09 15:24:45 still in heavy use when buying e.g. bacon May 09 15:24:49 koen, can you comment? May 09 15:25:13 rather would, lol May 09 15:25:16 av500, I am now salivating, thank you very much May 09 15:25:30 17:13 < koen> spacecolonyone: the default angstrom images will all bring up network on boot, provided you have dhcp May 09 15:25:45 note the 'angstrom' qualifier May 09 15:27:02 are you implying that ./sources/meta-angstrom/recipes-images/angstrom/systemd-image.bb is not a default image May 09 15:27:13 it is May 09 15:27:19 and that one will bring up networking May 09 15:27:24 (provided you have dhcp) May 09 15:27:43 you mean a server elsewhere May 09 15:27:51 or a router May 09 15:27:57 i assume, which I do. May 09 15:28:06 ifup eth1 works just fine and dandy May 09 15:28:40 but console-image does not bring up networking w/o me manually executing ifup eth1 May 09 15:31:37 is the bitbake manual online somewhere so I can read about these files? May 09 15:31:57 I can only find one from 2004 May 09 15:33:35 hello, does any know how to add wireless-tools in systemd-gnome-image? May 09 15:38:55 goldpeer: wireless-tools are deprecated, use iw May 09 15:39:42 * mru uses wpa_supplicant May 09 15:39:46 spacecolonyone: it looks like console-base-image is the same as systemd-image and console-image is just console-base-image with task-base-extened. But I I have used both console-image and systemd-image in the last week and console-image did not bring up the network interface but systemd-image did May 09 15:40:06 but I guess that too will be eaten by systemd soon enough May 09 15:40:48 serekson, thanks. I am building systemd-image now to see how that goes May 09 15:41:21 * spacecolonyone goes off to look at what task-base-extended really does May 09 15:41:22 I am not sure why they would be different though just by reading the .bb files May 09 15:43:49 anyone advice on the best way to compile a c program for the beagleboard? I have a working OE environment May 09 15:44:45 Hello! I'm looking for some consultation for my project. May 09 15:45:45 Anyone alive? May 09 15:46:08 Oleja: ask a meaningful question, then eventually someone might respond. May 09 15:46:32 <__av500_> or they might not May 09 15:47:51 ok, serekson: so it looks like task-base-extended is provided in task-base.bb and just adds wifi, bluetooth, and 3g if support, provided the distro and machine support them, based on my interface file I'm assuming at least the bt support is added May 09 15:48:58 I have a voice-recognition project, on which I would like to use beagleBoard. I had a bad luck with technextion TAO-3530. What I'm looking for is: Voice recognition to be done with CMU Sphinx (pocketSphinx) which will recognize some words (5-20) with different pronouncation. Then, if the word is recognized, it should play a wav sound file. Also, three light channels should be connected (red, green, blue). May 09 15:49:41 After the word is recognized, beagle should switch on a mix of color channels to produce different colors like orange. May 09 15:50:14 So, I would like to know what version of beagleboard can solve this task. May 09 15:50:15 spacecononlyone: yeah, it is conceivable that interfaces woudl be down by default when including wireless drivers, because in the default state you want all transmitters off May 09 15:50:22 koen, you mean i can add iw in systemd-gnome-image? May 09 15:51:07 i saw wireless-tools had compiled and stored in build/tmp-v2012.5/work folder May 09 15:51:16 serekson: the thing that really irks me is I cant figure out what setting I need to change in the deployed image. May 09 15:51:28 however, it is not packed into Angstrom-xxx.tar.bz May 09 15:51:43 i.e. it has not packed into sysroot. May 09 15:54:52 spacecolonyone: did you try changing /etc/network/interfaces? May 09 15:55:21 yea, many a time without success May 09 15:56:19 well, I should say that the settings would take, but that auto eth1 never would make it auto May 09 15:56:21 Oleja: technextion TAO-3530 is omap3530 May 09 15:56:26 like the beagle more or less May 09 15:56:35 so what does the beagle gain you? May 09 15:56:48 it looks like that config file is actually for the ifup and ifdown commands May 09 15:56:49 it's furrier May 09 15:57:35 right, and not really used by the systemd system May 09 15:58:13 the systemd image uses connman May 09 15:58:26 #av500 I wasn't able to reprogram GPIO to control the lights channels May 09 15:58:27 which is awesome because you can only configure it using python scritps May 09 15:58:43 Oleja: then you wont be able on the beagle either May 09 15:58:48 its the same GPIOs May 09 15:59:05 oh joy, just built systemd-image and now I don't boot May 09 15:59:47 hang at the same place I was yesterday ...done, booting the kernel. May 09 15:59:54 av500: on TAO3530, when I build a kernel, with all related stuff, it just stops to boot, and tech support takes a really long time to answer May 09 16:00:32 <__av500_> so you're coming here for the superior free support? May 09 16:02:19 _av500_ transfer to L2 support! May 09 16:02:51 ok, so I'm stumped: I've followed the same recommended, procedure I've used sucessfully many times perviously, but now I don't boot. I could really use some help debugging this May 09 16:02:59 <__av500_> L2$ support May 09 16:04:04 Anyone aware of the parameter that controls the time between 2 accesses when DMAing the GPMC? (I seem to always have a 150ns pause no matter what GPMC timings I configure) May 09 16:12:46 sure enough doing systemd-image fixed the network adapter issue May 09 16:12:58 spacecolonyone, how do you know you don't boot? May 09 16:13:37 no heatbeat LED,display doesn't come up, noting over the serial console May 09 16:14:33 same console=tty... settins in uenv.txt as for successful boots when I see boot messages May 09 16:15:23 absolutely nothing over the serial console, thats impressive May 09 16:15:24 again, I was able to boot by changing the uImage symlink from uImage-3.2.16 to 3.0.28 in /boot May 09 16:17:01 when you built the image with OE it gave you the 3.2 kernel? May 09 16:17:31 I still get 3.0.28 May 09 16:18:11 yea, it did. It looks like the updates were committed in the past couple of days May 09 16:18:44 do git pull in you oe directory then oebb.sh update May 09 16:18:47 well, if it breaks for you it makes me scared to update May 09 16:19:00 and next time you build it will default to 3.2.16 May 09 16:19:08 and if you are like me, won't boot May 09 16:19:13 :) May 09 16:19:38 I will try it, let me save an image for this vm first though May 09 16:20:48 I let opkg update me to 3.2 last night and it broke everything May 09 16:21:58 serekson_: no 3.0.28 May 09 16:22:57 what about 3.0.28? May 09 16:24:30 serekson_: same thing happened to me yesterday May 09 16:25:01 goldpeer: it's already there May 09 16:25:47 maybe 3.2 is broken? May 09 16:26:37 3.2 works for me May 09 16:26:42 3.3 was broken last time I tried May 09 16:27:15 mru, how did you get 3.2? was is a straight git pull oeb.sh update? May 09 16:27:28 or have you manually applied any patches? May 09 16:27:37 git clone from kernel.org May 09 16:28:55 let me guess, no bitbake, and standard build with x platform toolchain May 09 16:29:03 I haven't level up that far. May 09 16:31:34 isn't OE the preferred way to build the kernel? May 09 16:31:46 preferred by whom? May 09 16:31:56 by OE devs May 09 16:32:05 develoepers May 09 16:32:22 what do you guys use? May 09 16:32:37 make May 09 16:32:58 but then, I have a 20 people team to make make work :) May 09 16:33:40 <__av500_> indeed May 09 16:34:45 yeah, I am not real fired about about OE. Too many layers of abstraction, it takes me forever to dig down into the problems I encounter May 09 16:40:08 Thanks koen. May 09 16:49:13 any one has experience ad hoc network setting on Angstrom via rt3572sta? May 09 16:49:52 the problem is when I configured it into ad hoc mode, it didn't work on that. it changed to AP mode May 09 17:36:50 serekson, crosstool-ng to build the toolchain then I just crosscompile May 09 17:37:03 serekson, aka "make" May 09 17:39:10 I am considering trying the linux kernel method described here: http://elinux.org/BeagleBoard#Linux_kernel May 09 17:41:27 serekson, ah yes I guess you can use the angstrom toolchain, saves you from building it May 09 17:48:06 does anyone know how to use OE to build a C program for the beagleboard without using make? It seems like you shoudl be able to but i can't find instructions anywhere May 09 17:48:58 why not use make? May 09 17:50:31 I am not opposed to using make, it just seems like there should be an easy way with OE since I have it all setup May 09 17:50:53 like "bitbake hello.c" or something May 09 17:50:58 but it seems that is not the case May 09 17:52:05 burrito! May 09 17:56:40 serekson, my understanding is that it is in theory quite simple May 09 17:57:40 using OE it is simple? May 09 17:57:46 the main thing is to create a recipe file that does the compilation of your file for you May 09 17:57:53 in theory May 09 17:58:06 just a sec, I think I have a tab somewhere with the OE info May 09 17:58:31 by the way, I am still working on my 3.2 systemd-image May 09 17:58:35 I've yet to do it, because it is a little lower on my priority list May 09 17:58:44 yea, I've been ewatching May 09 17:59:11 I wont get to test it out utnil I get home in a 3-4 hours though May 09 17:59:15 but its almost done building May 09 17:59:43 if you do get it going I'll be sure to ask. I suspect that if I give it a few more days then whatever issue will automagically clear up May 09 18:01:24 yeah, I think it likely that there is some problem with that version and OE but I am not good enough to figure out what it is yet May 09 18:01:36 I dno't really expect it to work May 09 18:04:54 serekson: http://www.openembedded.org/wiki/How_to_create_a_bitbake_recipe_for_dummies May 09 18:05:52 it is incomplete, and there is more to it, like where to stick it so bitbake finds it (hint create your own overlay) May 09 18:06:30 the latter point is really not bad at all and has helped me keep my work compartmentalized May 09 18:08:23 my other biggest help has been find ./ -name *.bb -exec grep some_string "{}" \; -print May 09 18:13:44 thanks, I will give it a shot after this build is done May 09 18:21:12 is there any chance that the failure to boot could be caused by using ext3 for the root instead of ext4 May 09 18:21:52 I've noticed that in the u-boot environment there is: ext4load mmc ${mmcdev}:2 ${loadaddr} /boot/uImage May 09 18:22:31 This strikes me as something unrelated and I would presume backwards compatibility is maintained, just curious. May 09 18:32:56 Isn't ext4 a superset of ext3 features? May 09 18:33:13 ext4 is quite different May 09 18:33:22 ah..k May 09 18:33:31 ext3 is ext2 with a journal bolted on May 09 18:33:42 ext4 has a completely different structure May 09 18:33:43 that's the one of was thinking of then May 09 18:34:02 koen++ koen++ koen++ koen++ koen++ May 09 18:38:15 Does anyone know if a linux-omap git repository that supports 1GHz on the Beagleboard-xm? May 09 18:38:26 I tried https://github.com/beagleboard/kernel May 09 18:38:39 but the 3.0 one doesn't seem to support 1GHz May 09 18:41:09 s4wrxttcs: even TI doesn't support 1GHz on those SoCs anymore May 09 18:41:30 I thought it just required the smart reflex 1.5 stuff? May 09 18:41:39 correct May 09 18:41:42 koen: why did they drop that? May 09 18:41:49 I mean seriously, wtf? May 09 18:42:01 mru: because there is one evm in india that works without the SR1.5 stuff May 09 18:42:03 I was wondering why it never made it into the mainline May 09 18:42:09 therefore all boards work without it May 09 18:42:17 but it used to work on my xm May 09 18:42:19 now it doesn't May 09 18:42:27 mru: it took me close to a year to get them to admit sr1.5 was actually needed May 09 18:42:28 so someone removed something May 09 18:42:39 * Russ adds another thing to the list of xm issues to fix May 09 18:43:01 what I find really odd is the git repository for the sr1.5 patches is renamed to deleted May 09 18:43:34 s4wrxttcs: they wouldn't work anyway, that portion gets rewritten almost every week May 09 18:44:04 so basically I should just forget about it May 09 18:44:10 and despite all this crap, TI still has the best support of all the soc vendors May 09 18:44:11 and be happy at 800MHz May 09 18:44:18 or run an older kernel May 09 18:44:25 iirc I had 1GHz on a 3.0 kernel May 09 18:44:36 I'll my stuff is based on 3.0 May 09 18:44:46 so I just thought I could add some patches and be good to go May 09 18:46:12 so all this talk and I went to go check proc/cpuinfo May 09 18:46:24 I've got 600 bogoMIPS May 09 18:46:52 which seems low even for the 800MHz speed May 09 18:46:53 what is "one evm in inda"? May 09 18:47:14 +i May 09 18:47:17 I'm at 798 bogomips at 800 May 09 18:47:40 aholler: an evm is an expensive dev board May 09 18:47:57 great, so why am I running at 600 MHz on a clean build? May 09 18:47:57 india is a country rather far east, but not quite as far east as japan May 09 18:48:08 its at 597 bogomips right now because setting the mpurate at 1000 failed May 09 18:48:16 ahh May 09 18:48:29 make sure mpurate is set at 800 May 09 18:48:35 in uboot May 09 18:48:50 ya, I got ya. sucks May 09 18:49:33 so what is this smart reflex stuff? May 09 18:49:42 s4wrxttcs: cpufreq-set -g performance? May 09 18:50:01 spacecolonyone: missing May 09 18:50:10 spacecolonyone: smart reflex is a a way to know if your cpu needs extra power for higher frequencies or not May 09 18:50:22 clock: dpll1_ck: unable to set MPU rate to 1000: -22 May 09 18:50:27 so someone in that country was unable to add a proper smartflex-capable power-ic and therefor the support for 1GHz was dropped for the world? May 09 18:50:29 assuming his kernel has the performance governor compiled in ~.~ May 09 18:50:30 I get that message when I try an mpurate of 1ghz May 09 18:50:38 spacecolonyone: when the chip is produced some measurements are done and coded into the efuses May 09 18:50:59 using https://github.com/beagleboard/kernel May 09 18:51:05 stop creating emi... write better code ;) May 09 18:51:06 aholler: not unable, that person lucked out in the silicon lottery and didn't need the boosting for 1ghz May 09 18:51:09 the 3.0 branch with the patches applied May 09 18:51:13 aholler: therefore the problem doesn't exist May 09 18:51:29 lol May 09 18:51:36 thurbad: it's not a joke May 09 18:52:07 I know... but it's so cringe inducing I had to let it out May 09 18:52:50 cpufreq-set -g performance got me a null pointer derefrence at 0x0 May 09 18:52:54 whee May 09 18:53:12 mru: I think the core voltages in the OPP definitions got tweaked in recent kernels May 09 18:53:34 is there a way to untweak them? May 09 18:53:49 yeah, raise the voltages in the .h file :) May 09 18:54:00 or rather the magic number that corresponds to it May 09 18:56:02 mru: how well tested/used is the armv6 vfp code in libav? May 09 18:56:22 mru: DJWillis ran into a problem, but that might just be a toolchain bug May 09 18:56:38 at least some interesting stories are offered here ;) May 09 18:57:26 koen: it could be better tested for sure May 09 18:57:30 koen: what's the problem? May 09 18:57:50 koen: there is ARMv7 code in some of the ARMv6 stuff for sure ;). I looked into it and worked out a fix but also found that it has been sitting upstream as a patch so did not shout very loud about it. May 09 18:58:02 well looks like I'm in the dark. May 09 18:58:11 The beaglebone says it was delivered today. May 09 18:58:14 but it wasn't May 09 18:58:14 koen goes r-pi? ;) May 09 18:59:05 aholler: I deny everything May 09 18:59:08 mru: https://github.com/djwillis/meta-raspberrypi/commit/7dbc0ca1103d3ce445947979272038502ccf6cb0 May 09 18:59:11 koen: does it surprise you to hear that I cant boot the xMrevC with a systemd-image built today with bitbake unless I first swap out the 3.2.16 kernel for the 3.0.28? I guess what I'm really getting at is: is this so current that perhaps all the necessary patches haven't propagated through to the setup-scripts repo or some of the individual meta repos yet? May 09 18:59:11 mru: https://github.com/djwillis/meta-raspberrypi/blob/master/recipes-multimedia/libav/libav/libavcodec-ac3dsp_armv6-patch-out-armv7.patch was the one patch I did. May 09 18:59:21 Ahhh, far too slow ;-) May 09 18:59:39 spacecolonyone: I ran 3.2.16 on an xM-C and classic C5 today May 09 18:59:40 koen: deny it all ;-). Best way to be. May 09 19:00:31 yes, don't attract those users to this channel May 09 19:00:43 (too) May 09 19:01:14 DJWillis: what libav version are you using? May 09 19:01:37 koen: I followed http://www.angstrom-distribution.org/building-angstrom from scratch today, changing only virtual/kernel to systemd-image and it hangs on boot May 09 19:01:41 mru: 0.7.4 May 09 19:01:49 my oh my May 09 19:01:52 spacecolonyone: that's what I used as well May 09 19:01:58 spacecolonyone: update mlo and uboot as well? May 09 19:02:19 DJWillis: grab the latest git master May 09 19:02:23 it'll be much faster May 09 19:02:49 koen: yep, every time May 09 19:02:53 mru: it also gave me so much pain I just built with --disable-armvfp --disable-neon in the end ;-). I'll try and bump my .bbappend when I get some time later. May 09 19:03:06 well, the rpi has no neon May 09 19:03:18 to be clear I'm not asking you to solve this for me, I'm looking for advice on how to approach debugging this and figuring out wtf I've messed up. May 09 19:03:38 I have never been able to get the MLO and u-boot.img to work from OE either May 09 19:03:49 I alwasy get them from somewhere else May 09 19:03:55 spacecolonyone: but the bootlog in pastebin May 09 19:04:01 serekson ?? May 09 19:04:21 mru: I know that ;-) but without it the build seems to try and pull in neon code :o May 09 19:04:28 it always builds them for me nicely May 09 19:04:33 DJWillis: then you're holding it wrong May 09 19:04:37 quick,let's write patches to emulate neon May 09 19:04:41 or your toolchain is fucked up May 09 19:04:55 mru: I know so I let go and put it down to something being fucked up ;-) May 09 19:05:20 exactly what configure command are you using? May 09 19:05:24 spacecolonyone: it builds them for me too, but they dont work May 09 19:05:42 I havne't had a chance to troubleshoot why yet May 09 19:06:01 mru: i'll look it up in a bit, not at my build box, just noticed koen flash when I looked at my PC to check email ;-) May 09 19:06:18 if I had any actual armv6 I'd set up automated tests May 09 19:06:48 did it fail to build or fail at runtime? May 09 19:07:23 koen: nm I think the failure to boot was caused by mpurate=1000 May 09 19:08:25 I swapped the symlink back to 3.2.16 and now it boots, only other thing I changed was mpurate=800 based on the (rather disappointing) conversation a min ago May 09 19:09:03 okay, I tried the cpufreq-set -g performance May 09 19:09:06 and that worked May 09 19:09:20 so if that worked why didn't just setting the mpurate to 1000 work? May 09 19:10:27 Note - I'm using the linux-omap 3.0.29 kernel with the patches from https://github.com/beagleboard/kernel May 09 19:11:07 does anyone know how to check the frequency that the dsp is running at? May 09 19:11:14 I know how to look at the freq of the cpu May 09 19:12:04 spacecolonyone: no mpurate param needed May 09 19:12:41 s4wrxttcs: true story: mpurate uses a different path to set the frequency May 09 19:12:49 s4wrxttcs: that was fixed in recent kernels May 09 19:12:55 thanks koen May 09 19:14:25 I guess thats changed, oh well. for that matter are e.g. vram=16M defaultdisplay="dvi" and dvimode="1280x1024MR-16@60"still needed May 09 19:15:03 spacecolonyone: the default uboot env sets most of those May 09 19:15:19 spacecolonyone: just put 'dvimode=hd720' in a uEnv.txt May 09 19:16:53 well by removing mpurate=800 I'm not at 666 May 09 19:17:01 s/not/now May 09 19:17:27 what are you at without it? May 09 19:17:41 666 May 09 19:17:51 without an mpurate in uEnv May 09 19:18:00 and with it, its the same May 09 19:18:02 800 with =800 and hang with =1000 May 09 19:18:06 ah May 09 19:18:33 correction, hang on the 3.2.16 kernel, boots fine on 3.0.28 May 09 19:18:51 though I don't know what speed, as I never checked there May 09 19:19:52 cpufreq-set -g performance causes the 800 to drop down to 500 but doesn't affect the 666 rate May 09 19:20:14 is it just the mainline 3.0.28 May 09 19:20:22 or does it have the omap specific patches May 09 19:20:29 or patchset May 09 19:22:04 3.0.28 is the one in oe until yesterday/this morning. just a sec May 09 19:24:20 meta-ti/recipes-kernel/linux/linux_3.0.bb. it lists a ton of patches, so i think the answer is yes. for instance I see OMAP3-beagle-HACK-add-in-1GHz-OPP.patch May 09 19:24:20 \ May 09 19:25:00 it needs the smart reflex patches on top of that May 09 19:25:17 which if I understood correctly earlier is likely related to guy that was lucky in the Si lottery May 09 19:26:53 yea, grepping the recipe for smart or reflex gives nothing May 09 19:27:10 what I don't understand is what became of http://www.mail-archive.com/linux-omap@vger.kernel.org/msg46267.html May 09 19:27:50 it says the series is available at a git repository May 09 19:27:53 grepping more widely i see them in the linux-omap 2.6.39 and 2.6.37 recipes May 09 19:28:03 but the git repository it points to has been renamed to deleted May 09 19:28:15 like they were somehow obsolete May 09 19:28:33 but it seems like the kernel patches I was using has most of it May 09 19:28:44 at least enough to work May 09 19:29:13 my only goal is to find a kernel/patches > 3.0 kernel that has 1GHz support and PM support May 09 19:29:58 from my understanding most of the power management made it into the mainline May 09 19:30:10 but not the smart reflex stuff to support 1GHz May 09 19:30:24 but, I'm far from an expert May 09 19:30:52 yet you are from my perspective in this regard. May 09 19:31:06 aka I'm listening but have no advice May 09 19:31:19 I'm so far from an expert that the mere mention of OE puts fear into me May 09 19:32:01 you don't need to be an expert to fear oe May 09 19:35:14 lol May 09 19:35:24 so far it hasn't been that bad May 09 19:35:40 I've gotten comfortable enough with the omap platform to know of the gotchas (MUSB, some ISP limitations) May 09 19:36:02 and now for the next project I'm trying to decide between omap4460 (variscite module) versus an i.mx6 platform May 09 19:36:35 they both have the csi/mipi camera interface I need May 09 19:36:37 unfair comparison :) May 09 19:36:43 they are both cortex-a9 May 09 19:37:00 the COM modules are probably about the same May 09 19:37:07 (comparing a quad core system that has SATA with a dual core system that only has USB for disks) May 09 19:37:08 but one is quad core May 09 19:37:18 we won't use SATA though May 09 19:37:35 just boot off an SD card May 09 19:37:39 ogra_: it might be hard for you to imagine, but there are people who don't run ubuntu May 09 19:37:48 store images on a USB drive May 09 19:37:56 won't run ubuntu May 09 19:37:56 so what do I trust proc/cpuinfo reports bogomips of 666 cpufreq-info says I'm at 800MHz May 09 19:37:57 mru, how is that ubuntu related ? May 09 19:38:19 but needs to support a PointGrey Chameleon Camera May 09 19:38:28 ogra_: running ubuntu without a fast hard drive is painful May 09 19:38:55 the omap4460 would be the most straight forwards, but even with that there are a bunch of changes driver wise May 09 19:38:58 to the camera especially May 09 19:39:00 running anything that uses IO a lot without a fast harddrive is painful May 09 19:39:15 not true May 09 19:39:19 * ogra_ still doesnt get how what he said implied anything about ubuntu May 09 19:39:29 my 16-port GbE switch does a lot of I/O and has no hard drive May 09 19:39:40 nitpicker :P May 09 19:40:08 ogra_: there are many applications that don't need fast local storage May 09 19:40:15 sure May 09 19:40:21 but if all you know is ubuntu, you won't see them May 09 19:40:23 so no fast local storage May 09 19:40:40 just a very processor intensive camera (unless we find a way to optimize it) May 09 19:40:54 well, still comparing a quad core system to a dual core that runs at the same speed is an unfair comparison May 09 19:41:19 s4wrxttcs: you could hire someone who's good at optimising image processing code... May 09 19:41:23 * mru ducks May 09 19:41:33 its not even image processing May 09 19:41:36 (and yes, it wont matter if you only run single threaded apps) May 09 19:41:46 for whatever reason the pointgrey camera uses 100% of the cpu May 09 19:41:52 just bringing the video in May 09 19:41:55 s4rwrxttcs: sounds like you are in a similar situation, just with a different camera and you are using the CPI May 09 19:41:55 and displaying it May 09 19:41:56 <_av500_> doing what? May 09 19:42:01 probably memcpy May 09 19:42:02 I'm sure that can be optimized May 09 19:42:05 <_av500_> aye May 09 19:42:12 and it barely works on a pandaboard ES May 09 19:42:27 and will probably work even better on a variscite omap4460 module May 09 19:42:30 that runs at 1.5GHz May 09 19:42:31 as I said, there are people who will make it fast for money... May 09 19:42:34 well they advertise it May 09 19:42:41 I haven't actually seen it run at 1.5GHz May 09 19:42:46 true May 09 19:42:55 we contracted out for the ISP work on the existing product May 09 19:43:00 and I'm sure we will again May 09 19:43:27 hah in academia we have no such luxery May 09 19:44:01 but, at least in academia you don't shove the thing through abusive testing May 09 19:44:06 well not normally :-) May 09 19:44:18 spacecolonyone: Sure you do. that is what grant money is for, no? ;-P May 09 19:44:33 s4wrxttcs: well, if you need any software optimised, give me a shout May 09 19:44:54 true, but it must JUST WORK (TM) at an observatory in the middle of no where Chile for 5-8 years w/o me doing tech support May 09 19:45:07 haha May 09 19:45:16 s/shout/pile of money/ ? May 09 19:45:22 that too May 09 19:45:55 agmlego: grant money is for the $100k 12" lenses in the spectrograph and the 120k 4kx4k e2v CCDs May 09 19:46:00 and crazy users that plug in the wrong power supply May 09 19:46:01 don't like to travel to chile? May 09 19:46:25 spacecolonyone: I know. Notice the emoticon at the end. I have worked on academic projects too. May 09 19:46:37 Oh no I love traveling there, but the chef at the observatory makes these awesome empanadas! May 09 19:46:59 spacecolonyone: On the other hand, academic projects tend to benefit from related fields of study and the grad students therein in close physical proximity, or can. May 09 19:47:06 I noticed, I just feel the same way sometimes May 09 19:47:52 we've contracted out the actual camera board and analog front end May 09 19:48:07 Good choice. May 09 19:48:44 spacecolonyone: does the camera board interface to the OMAP? May 09 19:48:49 or is it connected to a PC? May 09 19:49:05 an EE has designed a board with an FPGA that takes in 8 datastreams from 8 linear cmos arrays at 10MHz and muxes it into an 8x8x512 datastream over the CPI May 09 19:49:23 from my own experience with the CPI May 09 19:49:47 its extremely sensitive to any interruption or noise on the hsync/vsync line May 09 19:50:05 it just doesn't reset like one would expect on the next vsync May 09 19:50:25 on my system I had to use a digital filter May 09 19:50:35 the sensor is about 18inches away May 09 19:50:38 so weve got (2x) set of 4 hamamatsu linear arrays -> 4channel muxing AD ADC chip -> lvds -> fpga to beagle via i2c expander and CPI May 09 19:50:42 and goes into serial LVDS May 09 19:50:56 goes into serial LVDS at the Camera May 09 19:51:09 and then 18+ inches of shielded twister pair May 09 19:51:14 before the deserializer May 09 19:51:18 and then into a CPLD May 09 19:51:26 the CPLD has the digital filter May 09 19:51:45 without the digital filter any nearby ESD (of a certain amount) would cause the video to lose sync May 09 19:51:51 we are running the LVDS over ~15' of cat5e to connect the AFE with the FPGA formatting board May 09 19:52:10 I'm just curious what will happen if there is nearby ESD May 09 19:52:53 but, I'm sure the FPGA can fix it May 09 19:52:55 if something does May 09 19:52:56 you get to replace your chips with artfully shaped sand? May 09 19:52:59 we will find out once we get the boards by from sierra circuits May 09 19:53:15 that's the awesome thing about the FPGA May 09 19:53:29 it can correct for any issues before it reaches the CPI May 09 19:53:34 ... and I get my kernel patched with ISP/CCDC support May 09 19:54:21 ...and I get uboot & MLO patched to know what our custom boards i2c commands are for bootup May 09 19:54:30 I like the ISP/CCDC interface, but it can be tricky May 09 19:54:51 are you loading the FPGA in uboot? May 09 19:55:05 oh no, May 09 19:55:35 control of the fpga is actually via a 16 bit i2c expander chip May 09 19:55:42 I was using an FPGA in a previous test version because at the time we wanted to support a CCD which I used an AFE for May 09 19:55:59 I loaded the FPGA using SPI May 09 19:56:08 and communicated with the FPGA firmware via SPI as well May 09 19:56:12 I figured that was simpler than either designing or buying IP for i2c on the cyclone May 09 19:56:23 yeah May 09 19:56:28 yea, but no spi on the camera port May 09 19:56:40 SPI was pretty easy, but either way May 09 19:57:25 the pointgrey camera was cheaper than anything I could make using the same sensor so we went away from that solution May 09 19:58:08 yea, here I need the 10MHz readout rate May 09 19:59:18 I can't wait to get an AR.Drone 2 May 09 19:59:30 to see what they use for the kernel/camera/etc May 09 19:59:36 they stream at 720P May 09 19:59:45 my goal is to stream the images to memory and then do some simple processing and discard before the next one arrives (at roughly 20kHz) May 09 19:59:52 dang May 09 19:59:53 and they use the omap3730 (as far as I know0 May 09 20:00:02 and support wireless-n May 09 20:00:15 and it has to have low-latency May 09 20:00:29 so far its the most impressive omap3730 based product I've seen May 09 20:00:31 maybe some use of the DSP? May 09 20:00:47 yeah, but in my use of the DSP May 09 20:00:49 yea, Russ I'd though about that May 09 20:00:53 I've been saying nobody uses that chip to its full potential May 09 20:00:54 I could never get any decent latency May 09 20:00:58 * Russ makes a note to order an AR.Drone 2 for "regression testing" May 09 20:01:02 but I haven't spent that much time on it May 09 20:02:07 mru: yeah, I honestly have never seen it used fully May 09 20:02:31 because everyone is busy chasing the next chip May 09 20:02:43 it definitely feels like that May 09 20:02:57 but even with the pandaboard I've never seen anything work fully May 09 20:03:08 well xbmc was pretty impressive on it May 09 20:03:29 it has hw h264 decoding, so what? May 09 20:03:48 mru: because it mostly worked May 09 20:03:51 :-) May 09 20:13:31 <_av500_> xbmc runs on the pi, so it cannot require much May 09 20:14:04 it needs a video decoder and a gpu May 09 20:14:14 both are usually driven by binary blobs May 09 20:15:37 <_av500_> thats the b in the name May 09 20:22:07 raspbinary blob May 09 20:23:08 * mdp imagines pfefferz under attack by an R-Pi May 09 20:23:42 Am I crazy for wanting to ditch busybox and go to coreutils May 09 20:24:40 depends on why May 09 20:25:47 if you need full versions, it's not crazy May 09 20:25:59 I can't quite put my finger on it yet, the only specific annoyance I've got is that ls doesn't support color May 09 20:26:38 but there are occasional things that crop up that make me worry about the future May 09 20:27:06 ok, mdp, thats the sort of answer I was looking for. May 09 20:33:04 probably you don't use a busybox build that contains *only* coreutils replacements though May 09 20:33:26 most people are using many many other replacements in busybox as well May 09 20:33:56 how do I modify the mpurate in uboot? using grep I cant find it in the source May 09 20:34:26 I see it referenced a lot, but I do not see wehre it is set May 09 20:34:29 serekso_ I just stopped autoboot and did printenv May 09 20:34:43 and i think it spat out mpurate=auto May 09 20:35:30 mdp:what do you mean? I haven't started to really look into this yet, beyond making sure there was a recipe for coreutils. I've also seen a task-busybox-replacements refrenced May 09 20:35:40 serekson_: don't use mpurate, just use cpufreq in userspace May 09 20:36:39 koen: and just write a startup script to set it? May 09 20:36:53 koen: is bogomips in proc/cpuinfo authoratative or is cpufreq-info May 09 20:37:21 serekson_: whatever floats your boat May 09 20:37:30 spacecolonyone: bogomips in /proc is based on the bogomips calculation done by the kernel on boot, I am pretty sure. May 09 20:37:35 spacecolonyone: bogomips are bogus mips May 09 20:37:43 cpufreq-info is authorative May 09 20:37:43 spacecolonyone: Not a terribly accurate number, mostly used by the system for timing. May 09 20:37:58 dependin on the kernel bogomips might or might not get updated after switching freqs May 09 20:38:11 cpufreq on te other hand should be talking directly to the processor and be more or less authoritative on the fly. May 09 20:38:12 right, that was precicely my concern was that May 09 20:38:34 on my system /proc/cpuinfo May 09 20:38:44 changes as soon as I do cpufreq-set May 09 20:39:12 s4wrxttcs: It might be that the kernel does another bogomips test when the settings are changed. May 09 20:39:14 spacecolonyone: busybox is configurable. you choose what features you want…like udhcpd..blah blah blah. whatever recipe + conf you are using has some subset of the total featureset in the busybox code enabled. May 09 20:40:08 for me bogomips doesn't change at all when i do cpufreq-set -g performance May 09 20:40:25 what about the cpu frequency itself May 09 20:41:00 yep May 09 20:41:01 I really wish USPS didn't suck. May 09 20:41:19 spacecolonyone: Yep, it changes? May 09 20:42:27 ok so everything does change when I do cpufreq-set -f [300MHz, 600MHz, 800MHz] May 09 20:42:44 spacecolonyone: Then ignore /proc May 09 20:42:54 I was thrown because I get bogomips of 250, 500 and 666 respectively May 09 20:43:22 but if I set mpurate=800 in uEnv then I get a bogomips of 798 May 09 20:44:20 i guess the real test will be down the road to see how it handles processing everything May 09 20:45:12 spacecolonyone: what ended up being your fix for the kernel not booting, was it the 1GHz mpurate? May 09 20:45:39 yes, mpurate=1000 prevents mainline 3.2.16 from booting May 09 20:46:26 and you just interrupted boot to fix it? May 09 20:47:00 no I edited it out on my dev machine May 09 20:47:32 is the boot partition mounted by default with a console image? I can't find it. May 09 20:48:56 spacecolonyone: you edited it in the source and then recompiled? May 09 20:49:15 good heavens no May 09 20:49:51 then i dont understand how you changed it May 09 20:49:53 I've got a uEnv.txt file on my boot partition and one line of that file was mpurate=1000, i deleted it May 09 20:50:22 if that file is there then uboot will pares it for settings to override the compiled defaults May 09 20:50:32 I see May 09 20:50:42 well, that is much easier May 09 20:50:42 haha May 09 20:51:26 yea, I've never found any good docs on what the valid options are for the different params and what the params are, they also seem to change a bit May 09 20:51:47 so then can you set the speed to 1GHz using cpufreq-set? May 09 20:52:10 you can get a basic feel for what the possible left-hand values are by interrupting boot and issuing printenv May 09 20:52:30 thanks, thats really good to know May 09 20:52:44 I wish i could May 09 20:53:06 I can only set it to 300MHz, 600MHz, & 800 MHz May 09 20:53:15 I just did "cpufreq-set -f 1000MHz" and it seems fine May 09 20:53:30 1GHz seems to be silently ignored and is no listed as an option May 09 20:53:59 I tried 1GHz :) May 09 20:54:01 well, I guess I am still on 3.0.28 May 09 20:54:06 yea May 09 20:54:33 for me 1000MHz just does nothing May 09 20:55:30 * spacecolonyone goes to try this with 3.0.16 May 09 20:55:38 s/16/28 May 09 20:56:20 <_av500_> ascaling_available_frequencies tells you what can be set... May 09 20:56:27 <_av500_> -a May 09 20:56:59 hmm? May 09 20:57:12 I see them listed in cpufreq-info May 09 20:57:58 serekson: going back to 3.0.28 and now the allowed range extends up to 1000MHz May 09 20:58:38 hmm, maybe I wont upgrade to 3.2 just yet May 09 20:58:49 ... but "cpufreq-set -f 1000MHz" crashes the system May 09 20:59:05 mine is still running fine May 09 20:59:12 <_av500_> btw, this is a kernel that scales up to 1.2ghz: https://www.gitorious.org/archos/archos-gpl-gen8-kernel/blobs/master/arch/arm/mach-omap2/omap3-opp.h May 09 20:59:38 i didn't think the chip could go that hig May 09 20:59:38 <_av500_> at 1.2 ghz, the dsp is scaled back full May 09 20:59:51 <_av500_> so its arm only in that state May 09 21:00:13 for power and tdp reasons? May 09 21:01:05 <_av500_> yes May 09 21:01:15 "cpufreq-set -g performance" also kills it May 09 21:01:19 <_av500_> but its OK May 09 21:01:25 <_av500_> you need 1.2ghz to render websites May 09 21:01:31 :P May 09 21:01:34 <_av500_> so you can slow the dsp May 09 21:02:15 <_av500_> i think we bought the 3630s rated at 1ghz, but 1.2 seems fine May 09 21:02:34 * mranostay pokes in May 09 21:03:01 <_av500_> of course, thats a .29 kernel, so you will all hate it May 09 21:03:26 is there anyway to quickly tell what frequency the dsp is running at? May 09 21:03:42 <_av500_> something in sysfs May 09 21:03:49 <_av500_> or debugfs May 09 21:04:04 serekson: you didn't use oe to build your kernel did, right? so did you have a patch list you compiled against? May 09 21:04:05 <_av500_> but unless you use the dsp, it should be off... May 09 21:04:21 I do use it for mjpeg compression May 09 21:04:35 hopefully h264 compression at some point May 09 21:04:43 spacecolonyone: my kernel was built with OE May 09 21:04:55 av500 there is the chance I will need the dsp to process some streaming ccdc data, so I "donts likes the sound of that" May 09 21:05:07 <_av500_> of what? May 09 21:05:29 going with a kernel that requires halting the dsp May 09 21:05:36 <_av500_> only for 1.2 May 09 21:05:41 oh May 09 21:05:43 <_av500_> 1000/800 is OK to use May 09 21:06:03 <_av500_> see the arm and dsp freq tables in the file I linked May 09 21:06:17 <_av500_> we used 1000/800 for hd decode May 09 21:08:16 see i keep thinking I need the newer kernel for catching more of the CCDC and ISP patches and drivers, but I'm really not all that sure May 09 21:08:35 <_av500_> there are patches? May 09 21:08:38 for all I know a 2.6.39 kernel would be fine May 09 21:09:02 <_av500_> what do you do on ccdc? May 09 21:10:03 we are making a custom camera sthat is going to feed in 8x512 8bit images over the CPI interface at 80MHz May 09 21:10:22 cam_pclk=80MHz May 09 21:10:51 <_av500_> cpi=parallel? May 09 21:10:56 yea May 09 21:12:25 then that gets some simple (likely just thresholding/summing at/of pixels which are known a prioi) processing and we rinse lather repeat May 09 21:14:38 av500 would you by chance know if trying the kernel was as simple as changing the preferred kernel in my conf to linux-omap_2.6.39 May 09 21:14:51 <_av500_> .29 :) May 09 21:15:15 woah May 09 21:15:22 <_av500_> and its an android kernel May 09 21:15:31 <_av500_> with a lot of our stuff May 09 21:15:32 yea, I'm sure I don't want to mess with that May 09 21:15:42 <_av500_> i doubt it would work for you May 09 21:16:07 <_av500_> but it "shipped" :) May 09 21:16:27 hurray May 09 21:16:44 <_av500_> 2ys ago... May 09 21:17:05 * spacecolonyone goes back to reading about systemd and creating one's own services for ones own code May 09 21:20:29 <_av500_> gee, tizen is exciting: http://3.bp.blogspot.com/-Qjvn0H-f7UE/T6rFM0Z2cAI/AAAAAAAAApg/JatGTrz75ZM/s1600/DSC00500.JPG May 09 21:21:08 ouch May 09 21:21:34 <_av500_> these guys are so fired :) May 09 21:21:37 _av500_: to be fair one guy is looking at his phone May 09 21:21:42 LOL May 09 21:21:55 <_av500_> he fell asleep doing so May 09 21:22:18 <_av500_> see hiw the twitter timeline scrolled down... May 09 21:22:20 <_av500_> how May 09 21:22:30 <_av500_> use "enhance" May 09 21:22:34 * ogra_ is just enjoying the plenary talkof the creative head of EA games at UDS ... May 09 21:22:57 they should all just come over ... we have coffee here as well to keep them awake ;) May 09 21:23:08 I spot one female in that photo May 09 21:23:15 wait, two May 09 21:24:01 <_av500_> ah yes, Eva is there May 09 21:25:38 _av500_: which PMIC is paired with the proc? May 09 21:53:41 any immediate feedback on https://github.com/beagleboard/am335x_pru_package ? May 09 21:54:00 the source for the assembler and some of the documentation on the module are still pending approvals. May 09 21:54:28 this is what I was able to get approved for release as of now, but the assembler source and missing section of the TRM are coming. May 09 21:55:04 https://github.com/beagleboard/am335x_pru_package/blob/master/pru_sw/utils/LICENCE.txt May 09 21:55:17 you get that file *after* downloading it May 09 21:57:11 it is possible to view it ahead of time. May 09 21:57:19 I think a click-wrap will make this less usable. May 09 21:59:01 who wrote this license? May 09 21:59:22 got me. May 09 21:59:26 the "for clarification" stanzas drive me a bit batty, if you had a more clear way of stating it, just use that May 09 22:00:46 jkridner: does it support pruv2? May 09 22:01:03 yes. May 09 22:01:49 it looks like the makefile is broken May 09 22:01:50 licence? is that a typo or are we british? :) May 09 22:01:56 (besides the usual TI broken) May 09 22:03:40 do you mean https://github.com/beagleboard/am335x_pru_package/blob/master/pru_sw/example_apps/Makefile ? May 09 22:03:46 yes May 09 22:04:18 ../../path/not/in/git May 09 22:05:59 built fine for me. May 09 22:06:14 did I forget to upload something? May 09 22:06:25 I built with the Angstrom external toolchain. May 09 22:07:26 why do I need put MACHINE=beagleboard ./oebb.sh config beagleboard? i mean need that beagleboard twice? May 09 22:07:42 i saw the MACHINE is in auto.conf May 09 22:09:43 spacecolonyone: I just built the 3.2 systemd-image using OE and it seems to work except my USB is broken May 09 22:11:23 too many ../ May 09 22:12:12 er... May 09 22:12:21 koen: I had to build the library first, but other than that it just built for me. May 09 22:12:46 git clone setup-scripts; ./oebb config; mv setup-scripts somewhere else; ./oebb config; broken May 09 22:13:12 shoud each ./oebb config remove .oe/environment-angstrom first? it's not expensive anyway May 09 22:13:13 koen, its a bit twisty, but it makes sense May 09 22:13:20 er not makes sense, but works May 09 22:30:55 jkridner: thanks for the release! Are you sure the (linux) pasm is for PRU v2? It does not recognize the -PRUv2 switch. May 09 22:31:24 jkridner: but it produces the same code as the old version. So that is still working. May 09 22:31:28 the switch was removed. May 09 22:31:43 do you have any v2 instructions to try? May 09 22:32:07 I'll have to dig in my notes... May 09 22:40:22 jkridner: Switch removed, but needed in the source? I get this: algo2.p(932) Error: Instruction illegal with specified core version. May 09 22:41:05 I probably have to RT(not yet published?)FM ? May 09 22:42:09 can you file an issue on the github and include a test bit of assembly along with the command you used to invoke the assembler? May 09 22:44:04 Ok, but that will be later tomorrow. It's late now and I've got a busy day coming. May 09 23:00:53 koen: the current beagleboard demo image doesn't include 'eject'. that makes it difficult to use the bone on it. May 09 23:03:08 FWIW, 'opkg install eject' seems to be working. May 09 23:04:34 write it in perl ;) May 09 23:04:39 it is just an ioctl() May 09 23:06:06 anyone have an example of a bitbake file that compiles a .c file? May 09 23:22:36 what's wrong with the ones in the repo? May 09 23:38:34 hello fellow comrades May 09 23:39:47 does anybody have a successfully booting linux on their beagleboard xM rev C May 09 23:39:59 i am in need of help May 09 23:40:18 a lot of people do May 09 23:40:23 :) May 09 23:40:27 probably most people who have a board May 09 23:40:27 jaykay: i thought you were booting linux last night May 09 23:49:48 my damn router is acting up again May 10 00:20:30 I guess TI released a bunch of pru related stuff but it's not pasmv2 May 10 00:20:48 just reverse engineer it and be done with it May 10 00:20:56 how would i go about that May 10 00:21:09 depends on what you're missing May 10 00:21:16 haha May 10 00:21:35 the wait for event instruction might be fun May 10 00:22:05 don't misread my bad attitude for being ungrateful May 10 00:22:12 on the contrary, I'm delighted they released the stuff May 10 00:22:14 jkrinder++ May 10 00:22:50 jkridner: ^^ May 10 00:23:15 * mru suggests SilicaGel discover tab completion of nicks May 10 00:23:42 I don't need to discover it, I need to REMEMBER it! May 10 01:27:38 jaykay: did you ever get your board booting? May 10 01:28:28 i reverted mine back to stock and all is well. but now that i have a better understanding of the boot process, i'm trying the ArchLinuxARM image. May 10 01:28:54 realized i'll probably never use Cloud9 IDE or any of the other bloat Angstrom has to offer. May 10 01:32:22 mdp, hehe bin blobs attack you know May 10 01:32:50 yeeeeeees i successfully booted ubuntu 11.04 on xM rev C May 10 01:33:20 jaykay: out of curiousity what process did you use? May 10 01:38:21 well, I just used bitbake to compile a c file that runs on my beagleboard May 10 01:38:30 I am pretty excited May 10 01:38:43 maybe I can actually do somethign now May 10 01:41:12 what r u running the board on? May 10 01:41:13 angstrom? May 10 01:41:32 yeah, angstrom 3.2 kernel built with OE May 10 01:42:31 ok so this is gonna be very nooby buuttt... May 10 01:42:46 I just started a week ago, so I am right there with you May 10 01:42:47 what is the non gui version of ubuntu/angstrom called May 10 01:43:02 and what are the gui version referred to as May 10 01:43:09 well, you have the console-image and systemd-image in angstrom May 10 01:43:15 jaykay: Ubuntu calls the non-GUI version ubuntu Server. May 10 01:43:32 jaykay: But a more general term for a system without a GUI is "headless" May 10 01:43:40 not true May 10 01:43:41 so the image i initially flashed onto the sd card was non gui May 10 01:43:46 and that was "headless" May 10 01:43:51 jaykay: Yes. May 10 01:43:54 and after i installed the gui packages May 10 01:43:57 on top May 10 01:43:58 a headless system is one without a local console May 10 01:44:07 ok i understand May 10 01:44:09 mru: Common usage says otherwise. May 10 01:44:25 the linux text console certainly counts as a 'head' May 10 01:44:45 now a typical server doesn't have a monitor attached May 10 01:44:48 so it's headless May 10 01:44:57 mru: Fair enough. My mistake. May 10 01:45:04 i see, so its the monitor that determines :) May 10 01:45:26 jaykay: I misspoke, there is no real term for it. Usually, one just says that a system without a GUI...has no GUI. May 10 01:45:30 jaykay: Yeah. May 10 01:45:42 it's more or less if you can interact with it without using another computer May 10 01:45:43 i think im going to stick to the initial ubuntu 11.04 wo/ GUI May 10 01:45:56 jaykay: So, Ubunsu Server 11.04. May 10 01:45:59 *Ubuntu May 10 01:46:11 ok, yeah in that case Ubuntu Server 11.04 May 10 01:46:24 would u guys recommend that? or shall i keep on trying to boot angstrom May 10 01:46:38 jaykay: With Ubuntu, it does matter, because the Server and Desktop (and Netbook, and Education, ...) versions are different in more than just the presence or lack of a GUI. May 10 01:46:55 hmmm May 10 01:47:08 what do u use agmlego May 10 01:47:15 jaykay: Depends, what do you want to do with the system? What package manager and environment are you used to? How large a system do you want? What issues are you having with Angstrom? May 10 01:47:20 jaykay: On what? May 10 01:47:36 i cant get to the login screen May 10 01:47:43 it freezes at the angstrom logo May 10 01:47:55 Is there a login screen on the serial console? May 10 01:48:13 moreover, im having issues with screen resolution, the display stretches larger than my 22" monitor May 10 01:48:36 i can get a login console if i go with the non GUI version May 10 01:48:54 Could be an issue with the display driver in the GUI version. May 10 01:48:58 for example if i download an image from narcissus i can successfully boot, with the exception of the resolution size May 10 01:49:03 Did you try older/newer versions of the image? May 10 01:49:29 i havent tried an older image, May 10 01:49:35 let me try searching May 10 01:49:40 OK. May 10 01:50:09 Also, when you say "what do u use", which system(s) would you like information on? May 10 01:51:11 jaykay, set your screen resolution in the uEnv.txt file.. May 10 01:52:12 does uEnv.txt override any settings in u-boot? May 10 01:52:18 hmm ok, so i have currently 2 files in my fat32, mlo + u-boot.bin, May 10 01:52:25 i should add this file into the fat32 partition? May 10 01:52:52 jaykay: yes, you can specify additional boot parameters there May 10 01:53:07 ah ok let me work on that May 10 01:53:22 yeap, it does.. just create a "uEnv.txt" in the boot partition with: dvimode=1280x720MR-16@60 .... or anyother u-boot variables.. May 10 01:53:53 type 'printenv' at the u-boot prompt to get a big overview of what is available.. May 10 01:56:18 in OE how do I tell which directory in work is related to my current kernel build? May 10 01:58:05 I have problem with /dev/video0 on the beagle board xm May 10 01:58:22 the problem is that it doesn't show up May 10 01:58:52 any one home? May 10 01:59:31 Sure. May 10 02:00:11 Do you have a camera attached? Do you have a driver installed? Did udev (or something else) discover the camera and make a device node for it? Is it supposed to be at /dev/video0? May 10 02:00:27 What operating system are you running, what version, what kernel? May 10 02:01:19 Also, Guest9056, it would be easier to talk directly to you if you changed your nick to something more meaningful, like the Rusty you came in with. May 10 02:01:20 angstrom .. udev shows that it registers as a camera. May 10 02:01:44 Nickserv changed my name May 10 02:01:59 Yup, because you did not identify for the nick. May 10 02:02:48 it could be a driver issue, but I had cameras working on a different sd image with the same installed packages May 10 02:02:57 no need to install drivers May 10 02:03:05 Exactly the same packages with exactly the same versions? May 10 02:03:36 dunno.. old image got lost because of a compatriot of mine leaving. May 10 02:03:47 My guess is something changed. May 10 02:03:56 Probably either in udev or in the camera driver. May 10 02:04:07 Probably in udev, though, as that is what creates the device nodes. May 10 02:04:16 mine too... I am guessing that I missed a package on the install... but which one? May 10 02:04:23 No idea. May 10 02:04:29 The install of what? May 10 02:04:46 the SD image from the online image builder May 10 02:05:10 I guess that should rather be called the build May 10 02:05:19 Ah, yeah, narcissus. May 10 02:05:29 so, load up the old sd image, using dmesg/lsmod... figure out what's missing. ;) May 10 02:05:30 The bane of many an Angstrom user. May 10 02:05:41 rcn-ee: 22:03:36< Guest9056> dunno.. old image got lost because of a compatriot of mine leaving. May 10 02:05:55 old image is MIA May 10 02:06:19 Guest9056: Was the old image generated, or one of the stock ones you can download? May 10 02:06:31 generated May 10 02:06:34 well, with the new images, does the kernel recogonize the device an load a driver for it? May 10 02:06:47 Can you play with settings in narcissus, try a few images? May 10 02:06:54 Or that. May 10 02:07:15 your probally just missing the kernel modules, which are usually installed seperating in angstrom.. May 10 02:07:24 Guest9056: How do you know "udev shows that it registers as a camera." May 10 02:09:28 excuse me... I forget what udev said... it was dmesg May 10 02:09:40 it gave me May 10 02:09:54 registering : Camera May 10 02:10:28 so with the camera plugged in.. what is the lsusb line for it? May 10 02:10:29 Ah, but that does not mean a device node was created, or that a driver was successfully loaded. May 10 02:10:36 Is this a USB camera? May 10 02:10:42 yes May 10 02:10:47 OK. May 10 02:11:02 Useful information, given that the xM has a camera interface on it. ;-P May 10 02:11:29 does it? May 10 02:11:41 Yup. May 10 02:11:58 As well as an LCD interface. May 10 02:12:36 what kind of a camera interface. May 10 02:12:57 because I am going to be connecting RS-485 security cameras May 10 02:13:00 Standard. May 10 02:13:07 Not RS-485. May 10 02:13:10 your the one with the camera, is plugged in as usb or a 2xsomething board interface? ;) May 10 02:13:32 usb May 10 02:13:47 Guest9056: Thus, USB. May 10 02:14:06 But I make the comment about the camera interface because thta changes what tools we use to explore what is going on. May 10 02:14:15 if it's supported, there's about a half dozen different drivers... lsusb with the vid/pid would allow us to give you a hint.. May 10 02:14:18 In this case, please pastebin the output of lsusb. May 10 02:15:24 I will have to leave and get on in about 15 minutes and tell you what it is.. BB is at home May 10 02:16:07 Guest9056: OK. In the future, though, it is easier to diagnose hardware issues when you have the hardware in front of you to diagnose. ;-P May 10 02:16:48 THink of this channel as akin to going to a mechanic's: S/he will have a hard time fixing your car, if it is not there. May 10 02:19:05 anyone know how to clean the entire OE deploy directory? May 10 02:19:16 I just want to get all of the crap out of there May 10 02:20:32 serekson: As far as I can tell from about thirty seconds' Googling, you just delete the .oe directory in your home directory on the host. May 10 02:21:03 thats just the setup file for the bitbake environment May 10 02:21:21 ::shrugs:: May 10 02:21:29 I am sure there are directions out there somewhere. May 10 02:21:56 yeah, I am to but it is eluding me May 10 02:22:29 Could always just grep for directories labeled oe, and take a look at their contents. May 10 02:22:41 Modern machines can scan the entire filesystem pretty quickly. May 10 02:22:58 serekson, i usually just nuke the build directory and just rebuild it from scratch.. May 10 02:24:15 I read somewhere that you shouldn't delete thigns from OEs directories because it can get confused, but I just baked up this VM so i will give it a go May 10 02:24:21 duh, there's a readme in the deploy dir (/deploy/images/beagleboard/) May 10 02:24:28 gives you a nice hint.. May 10 02:25:25 hmm, I don't have that readme May 10 02:25:45 this is on v2012_05 of angstrom.. May 10 02:25:55 yeah May 10 02:26:39 If you just delete something from there, you'll have to run "bitbake -c clean TARGET" otherwise oe won't rebuild the files you deleted on your own.. May 10 02:26:48 TARGET=pkg name.. May 10 02:27:26 awesome, thanks I will just nuke it. I always clean before I build anyway May 10 02:27:49 specially if you have all the pkgs downloaded it doesn't take that long. ;) **** ENDING LOGGING AT Thu May 10 03:00:03 2012