**** BEGIN LOGGING AT Mon Jan 12 02:59:59 2015 Jan 12 04:19:46 I'm having trouble loading the ADC drivers. Here is the command I'm using and the dmesg output: http://pastebin.com/svCMEjqp Jan 12 04:21:22 I'm trying to get the ADC running on two channels, then read the values from the PRU Jan 12 06:22:08 Running debian on my BBB, I'm having a strange problem with the system clock constantly getting set such that the day is doubled and the last two digits of the year are doubled. this happens roughtly every four days. What's going on? Jan 12 06:22:44 this happened before and after I installed NTP Jan 12 09:18:41 Need help to automount android device on beaglebone. I am using debian wheezy. Tried mtpfs but did't work properly Jan 12 10:01:31 Hello ! I am new to beagleBone and i am working on a project using that. Jan 12 10:01:38 I need help. Jan 12 10:02:08 state your case of emergency and wait for rescue Jan 12 10:03:33 actually, i read that PRUs are processors with low latency I/O Jan 12 10:03:50 but if we have to do a strict tiem job Jan 12 10:04:04 they're processors without interrupts, and a lot of other limitations. Jan 12 10:04:12 then how would we use C to program it ? Jan 12 10:04:25 that means you can basically count cycles and know for sure what your maximum latency is. Jan 12 10:05:06 you program them using an appropriate compiler and loader mechanism. all that is already documented. Jan 12 10:05:38 but counting cycles in C program ? how to do that ? cycles can only be counted using assembly ! right ? Jan 12 10:06:21 what keeps you from countiong assembly cycles after you compiled the c code? Jan 12 10:07:43 cycles can't be counted at all on any kind of platform with any kind of branch prediction and cache Jan 12 10:08:11 unless part of the program you are counting is completly linear and has no memory accesses Jan 12 10:08:17 jacekowski: right. now aplly your last post to the PRUs Jan 12 10:08:23 PRUs? Jan 12 10:08:45 jacekowski: the discussion was on the topic "PRU" Jan 12 10:08:56 well , i have almost no experience in assembly . Ok so suppose, i have the assembly code after compiling my C program but what if i want to increase or decrease the number of cycles between two different steps ? Jan 12 10:09:11 which could only be done using assembly ? Jan 12 10:09:13 ZeekHuge: you can add a nop in C Jan 12 10:09:20 how to impelment that in Jan 12 10:09:20 C Jan 12 10:09:37 ZeekHuge: asm("nop"); Jan 12 10:09:53 and if i want to decrease number of cycles ? Jan 12 10:10:14 you can't Jan 12 10:10:20 unless you optimise your program Jan 12 10:11:44 can manuall optimization reach the precision that assembly would attain ? Jan 12 10:12:07 manuall optimization of C ! Jan 12 10:12:09 if you do not change your toolchain Jan 12 10:12:22 oteherwise a new version of compiler might change things Jan 12 10:13:08 so what compiler should i use ? the most efficient one ? Jan 12 10:13:24 that's a hard question Jan 12 10:15:45 so the only solution to my problem is "learn assembly ", i guess . (right ?0 Jan 12 10:17:08 what are you trying to do? Jan 12 10:17:22 generally cycle accurate things are not needed Jan 12 10:17:34 and typically all you need for things to be not slower than Jan 12 10:17:44 actually, i want to do fast image processing ! Jan 12 10:18:28 on the BBB? Jan 12 10:18:34 where would the image come from? Jan 12 10:19:16 ZeekHuge: there is dedicated hw for image processing iirc Jan 12 10:19:38 on the BBB? no Jan 12 10:19:38 IVA it's called iirc Jan 12 10:19:43 not on the BBB Jan 12 10:19:48 yes on BBB, the BBB would itself take it using a may be 5 MP camera Jan 12 10:19:53 no Jan 12 10:20:02 so it's just BB that had IVA? Jan 12 10:20:05 the BBB has not camera interface Jan 12 10:20:08 jacekowski: yes Jan 12 10:20:15 omap3 had it Jan 12 10:20:19 not the AM335x Jan 12 10:20:24 that's a bit of a downgrade Jan 12 10:20:45 in some ways yes Jan 12 10:20:59 but then, not many people used the IVA on BB Jan 12 10:21:41 well, i can get that image through either usb or SPI ! Jan 12 10:21:50 from the camera Jan 12 10:21:57 HI anyone experience with beaglebone mtp? Jan 12 10:22:07 mtp? Jan 12 10:22:42 media transfer protocol Jan 12 10:22:46 that's just usb Jan 12 10:22:54 ZeekHuge: 5MP over SPI? Jan 12 10:22:54 used in android device Jan 12 10:22:56 and protocol itself is crap Jan 12 10:23:08 MTP in BBB is the same as MTP in any linux Jan 12 10:23:11 mostly it sucks Jan 12 10:23:24 jacekowski: actually, its not crap Jan 12 10:23:35 need help to automount android phone to beaglebone. I am using debian wheezy Jan 12 10:23:39 av500: why not? Jan 12 10:23:49 av500: it's slow to list files and generally sucks Jan 12 10:23:51 kk_: ask in #debian Jan 12 10:23:55 jacekowski: it is not slow Jan 12 10:24:00 av500: btw. are you planning to attend fosdem this year? Jan 12 10:24:03 if implemented properly Jan 12 10:24:11 libmtp is slow Jan 12 10:24:21 ya it stuks Jan 12 10:24:23 and the google android side of MTP is too Jan 12 10:24:32 very slow Jan 12 10:24:33 and windows implementation of mtp sucks as well Jan 12 10:24:46 jacekowski: we did MTP devices with ARM9s that hosted 10000s of files Jan 12 10:24:52 why we didn't just steal NFS and use that Jan 12 10:24:53 and it worked nicely Jan 12 10:25:06 jacekowski: NFS works so nice out of the box on windows Jan 12 10:25:21 Please any suggession? Jan 12 10:25:26 anyway, I use only WIFI now :) Jan 12 10:25:31 kk_: ask in #debian Jan 12 10:25:34 its a debian issue Jan 12 10:25:39 av500: actually, NFS seems to work better than MTP Jan 12 10:25:47 debian on BBB is debian Jan 12 10:26:08 av500 hm but debian supports not the kernel Jan 12 10:26:08 jacekowski:for certain use cases Jan 12 10:26:11 for bbb Jan 12 10:26:11 av500: anyways, are you going to fosdem this year? Jan 12 10:26:15 yes Jan 12 10:26:29 using NFS means you files are not indexed by Android Jan 12 10:26:38 so they do not show up e.g. in Photos or music Jan 12 10:26:49 unless you force a full reindex by hand Jan 12 10:26:56 you could use inotify for that Jan 12 10:27:04 sure Jan 12 10:27:13 still there is no NFS out of the box on windows Jan 12 10:27:20 besides, files accesible over MTP are not always updated on android Jan 12 10:27:20 and most ppl still run windows Jan 12 10:27:33 as in, mtp process on android very often uses some old file list Jan 12 10:27:46 it should not Jan 12 10:27:51 it does Jan 12 10:27:55 i've tested it many times Jan 12 10:27:56 it queries the same database that e.g. photos uses Jan 12 10:27:58 and it's a known problem Jan 12 10:28:12 as said, goog did a shit job on it Jan 12 10:28:18 we fixed many MTP issues in our builds Jan 12 10:28:20 https://code.google.com/p/android/issues/detail?id=38282 Jan 12 10:28:27 like files >4GB Jan 12 10:28:41 goog did a halfhearted implementation of it Jan 12 10:28:45 reported 3 years ago and still not fixed Jan 12 10:29:16 yes Jan 12 10:29:19 goog is like that Jan 12 10:29:22 they do not care Jan 12 10:29:27 I talked to the eng that did MTP Jan 12 10:29:36 he said, he is now on another task Jan 12 10:29:42 so its unmaintained Jan 12 10:29:50 but you would expect that people that work on android also use it Jan 12 10:29:54 that was 2ys ago Jan 12 10:29:55 and had the same problem Jan 12 10:30:07 you would think that one of them would get annoyed enough to fix it Jan 12 10:30:08 jacekowski: google people have all thier data in the cloud Jan 12 10:30:11 they dont care for MTP Jan 12 10:30:25 I use some wifi transfer tool Jan 12 10:30:32 works like a charm Jan 12 10:30:43 https://code.google.com/p/android/issues/detail?id=24713 Jan 12 10:30:57 as said Jan 12 10:30:59 wifi is even slower than mtp Jan 12 10:31:02 mtp in android is shit Jan 12 10:31:07 :) Jan 12 10:31:14 not disputing that Jan 12 10:31:23 but MTP can work nicely Jan 12 10:31:28 been there, done that Jan 12 10:31:38 got a certification Jan 12 10:31:44 that was like 5ys ago? Jan 12 10:31:48 more maybe Jan 12 10:31:53 mtp to my sony camera is slow as well Jan 12 10:32:19 but luckily there, i can switch that to use normal usb storage mode Jan 12 10:32:30 the actualy MTP data transfer is one giant bulk transaction Jan 12 10:32:45 so trasnfer of large files should max out USB speed Jan 12 10:33:02 yeah, but it's all about all the other operations Jan 12 10:33:08 yes Jan 12 10:33:08 like listing folder with 1000 files in it Jan 12 10:33:15 that sucks, big time Jan 12 10:33:15 for metadata one can do it stupidly Jan 12 10:33:18 or cleverly Jan 12 10:33:26 on both sides, server and client Jan 12 10:33:40 libmtp used to query ALL files at connect Jan 12 10:33:50 which was pretty stupidfor a 100GB device Jan 12 10:33:56 considering the fact that microsoft came up with this thing Jan 12 10:34:04 as said, the spec is quite OK Jan 12 10:34:05 i would expect their implementation to be pretty good Jan 12 10:34:25 what they did for windows media player was also OK Jan 12 10:34:34 the explorer shell extension was not so good Jan 12 10:34:40 then ppl stopped using MTP Jan 12 10:34:43 then time passed Jan 12 10:34:48 then google picked it up Jan 12 10:34:50 and the fact that you can't use it like normal filesystem Jan 12 10:34:56 on windows Jan 12 10:34:56 that is the point Jan 12 10:35:08 there was no normal filesystem on some devices Jan 12 10:35:23 but i mean things like compatibility Jan 12 10:35:32 some media players that did MTP stored all files as random numbers in a flat folder Jan 12 10:35:42 and metadata was all in the DB Jan 12 10:35:43 so for a lot of software i have to copy stuff from MTP devices to my hdd and then open them Jan 12 10:35:50 yes Jan 12 10:36:07 as said, MTP does not assume there is an fs as backend storage Jan 12 10:36:11 it might be a database Jan 12 10:36:23 its how it works Jan 12 10:36:29 it knows objects, not files Jan 12 10:36:44 but i should be able to do in simple C on windows open("\\mtp_device\file.jpg") Jan 12 10:36:51 but i can't Jan 12 10:36:54 indeed Jan 12 10:36:59 not part of the spec Jan 12 10:37:07 i can do it with cifs, i can do it with NFS, i can do it with usb storage Jan 12 10:37:20 I dont dispute that Jan 12 10:37:25 on linux luckily it can be mounted with fuse Jan 12 10:37:30 so it is more usable Jan 12 10:37:31 as said, it was not part of the design specs Jan 12 10:37:43 it knows only objects and metadata Jan 12 10:37:58 and it evolved from PTP which came from digicams Jan 12 10:39:00 does BB have a stand this year? Jan 12 10:39:39 not sure Jan 12 10:39:47 probably not Jan 12 11:03:03 does anyone here know how to load a device tree overlay in 3.19 kernel? Jan 12 11:04:14 can i ask for help now ? Jan 12 11:05:49 ZeekHuge: you can always ask Jan 12 12:10:24 I saw some vedio tutorial using open cv to do image processing. Jan 12 12:10:59 sure Jan 12 12:11:04 Can this same be done if we instruct the BBB to take an image through a camera and then process it ? Jan 12 12:11:32 yes Jan 12 12:11:46 it all depends Jan 12 12:11:57 its just takes a bit longer then on your i7 Jan 12 12:12:25 And that image would be of about 5mb ( from a 5mp camera ) Jan 12 12:12:34 It depends on ??? Jan 12 12:12:55 what you do to the image Jan 12 12:13:48 the BBB is not a good platform for image processing Jan 12 12:15:03 Ok ! Actually i want BBB to match the face in the image to the face in its database (in the micro sd) Jan 12 12:15:33 uuh thats cpu intensive Jan 12 12:15:50 What if i employ two BBB to process a single image, may be by dividing the image in two parts ? Jan 12 12:17:13 what framerate do you need anyway? Jan 12 12:18:00 depending on your opencv algo a singe BBB might take as long as 10s to process a singe image Jan 12 12:19:27 ZeekHuge: you better start developing the software on your PC and then check the processing/memory ressources you need Jan 12 12:19:29 May be 24fps. How would that matter ? Jan 12 12:20:59 ... Jan 12 12:21:40 Actually, i have bought BBB to do this task, and therefore, so wouldn't be able to do this on any another platform Jan 12 12:22:09 To start , i can us a pc. But the final product should be based on BBB Jan 12 12:22:37 well, find out the computational complexity Jan 12 12:22:39 of your task Jan 12 12:22:49 as said, it all depends Jan 12 12:23:10 ZeekHuge: see the BBB as a PC with limited power, you can do anything with it just a lot slower, and 24 times per second is probably too often Jan 12 12:24:36 Ok i'll do that, so if i use BBB for image processing, what part of the BBB will actually do that processing ? Jan 12 12:24:50 the ARM cpu Jan 12 12:25:11 The processor running the Os or the PRUs ? Jan 12 12:28:14 the main processor Jan 12 12:28:27 * stt_michael chuckes to himself Jan 12 12:29:15 I am completely newbie, so my question may sound insane , but i have only this way to learn ! Jan 12 12:31:44 BBB processor is internally divided in 3 parts (right ?) The one that would do the os work and other activities on it (at 1ghz). and the other 2 are PRUs running at 200MHz independent of the os thing (right ?) Jan 12 12:32:03 yes Jan 12 12:32:10 but ignore the PRUs for image processing Jan 12 12:32:22 they are not meant for that Jan 12 12:32:36 and being a newbie the last thing you want it to juggle three cpus Jan 12 12:36:05 Actually i am a newbie in such a complex architecture . i have experience with simpler MCUs ( 8051, AVR atmega ICs , PICmcu ) Jan 12 12:37:00 Ok what if a decrease my frame rate to 1fps ? Can it be attained ? Jan 12 12:37:16 On BBB Jan 12 12:37:56 And i really wish if i could employ those two 2 PRUs too for this !! Jan 12 12:38:49 ZeekHuge: I have no idea how complex your algo is Jan 12 12:38:55 neither have you Jan 12 12:39:04 so frame rate guessing is just guessing Jan 12 12:39:12 find out how fast it is on your PC Jan 12 12:39:18 thats easily done Jan 12 12:40:03 Ok ! I' ll do that ! Thank you !! Jan 12 12:40:12 de rien Jan 12 12:41:04 I'l soon be back after testing it on my pc and yes, Really, big thank you !! Jan 12 12:54:28 if you want an image processing cpu .. the DM365/8 works :) Jan 12 12:54:56 but then its got the core hardware and encoding engines to do it .. unlike a low power arm core Jan 12 12:57:33 DM36x has not much image processing Jan 12 12:57:41 it has fixed in hardware encoders/decoders Jan 12 12:57:46 no face matching hardware Jan 12 12:57:51 and its a measly arm9 Jan 12 12:58:57 its got face matching :p Jan 12 12:59:03 oh :) Jan 12 12:59:11 well .. I'd like to see you do image processing on the beagle ;) Jan 12 12:59:12 match my surpprised face then Jan 12 12:59:49 the encoders are done in firmware as far as I can see Jan 12 13:00:07 so you can use the pre-built or make your own if you're hardcore Jan 12 13:01:53 nice Jan 12 13:05:27 is there any other python libraries for gpio programming in python other than Adafruit's? Jan 12 13:07:58 no idea Jan 12 13:10:06 aswin: pybbio Jan 12 13:13:25 vvu : Do you think which one is better ? Jan 12 13:13:38 no clue, did not use them :) Jan 12 13:13:49 look at examples and choose which one you like better Jan 12 13:15:19 vvu : Thanks :) Jan 12 13:15:24 np Jan 12 13:57:11 anybody online Jan 12 13:58:03 http://www.catb.org/esr/faqs/smart-questions.html Jan 12 14:01:40 any mentors here Jan 12 14:01:48 ? Jan 12 14:03:31 ah, it's the GSoC time of the year... Jan 12 14:03:45 * tbr doesn't know that Jan 12 14:03:50 i know of a few mentals online here... Jan 12 14:03:53 yeah Jan 12 14:05:11 Chandy: wait until Google anounces the participating orgs Jan 12 14:05:29 at this point there is nothing to do GSoC related. Jan 12 14:05:32 Ok will wait Jan 12 14:05:51 but i am working BBB for my hobby project Jan 12 14:06:02 good for you :) Jan 12 14:06:02 I have some problems Jan 12 14:06:09 ask Jan 12 14:06:32 I installed ROS hydro in 12.04 ubuntuarm Jan 12 14:07:08 I want to integrate AR drone autonomy package in ARM processor Jan 12 14:07:21 Normaly it's not possible Jan 12 14:07:25 need help Jan 12 14:09:28 vvu are you geeting me? Jan 12 14:11:24 what is an "ar drone autonomy package" and is it open source? Jan 12 14:12:42 yes it's ROS package Jan 12 14:12:49 works in ubuntu Jan 12 14:13:01 open source Jan 12 14:13:08 yes Jan 12 14:13:53 https://github.com/AutonomyLab/ardrone_autonomy Jan 12 14:14:03 Chandy: whats the problem? Jan 12 14:14:35 http://wiki.ros.org/ardrone_autonomy Jan 12 14:15:27 I tryed to install ar drone autonomy in BBB but it shows the compatible issues Jan 12 14:15:50 Chandy: what kind of issue? Jan 12 14:16:03 I want to make it compatible for BBB Jan 12 14:16:55 Chandy: its even a catkin package, so whats the problem? Jan 12 14:17:28 wait Jan 12 14:23:03 I successfully installed ardrone autonomy package using this documentation Jan 12 14:23:22 http://wiki.ros.org/tum_ardrone Jan 12 14:23:45 and i run ar drone driver Jan 12 14:23:55 but it shows that stack/package ardrone_autonomy not found Jan 12 14:25:59 Humpelstilzchen have you got it Jan 12 14:28:38 Chandy: doesn't surprise me much Jan 12 14:29:15 so tell me whats the problem Jan 12 14:29:25 what i should do Jan 12 14:29:40 Chandy: from what I understood from ros this doesn't install anything, thats actually a ros-problem, not bbb Jan 12 14:29:47 Chandy: do you have a catkin directory? Jan 12 14:30:31 yes Jan 12 14:31:02 Chandy: I'ld put the the ardrone_autonomy package in src/ and then type catkin_make Jan 12 14:31:17 ok Jan 12 14:35:03 Hi all, Jan 12 14:35:35 Does anybody manage to get the firmware loading work with compat wireless for the TI WL18xx? Jan 12 14:35:59 I have bought 2 BBB. One is marked with a red line on bottom and one is with a green mark. What does that mean? Jan 12 14:36:22 Hi, i have a issue here on some freescale board but may be similar behaviour is common elsewhere, so asking here. well, tcpdump on port 5353 doesn' Jan 12 14:36:37 Hi, i have a issue here on some freescale board but may be similar behaviour is common elsewhere, so asking here. well, tcpdump on port 5353 doesn'nt show multicast packet Jan 12 14:38:07 which is unexpected. i'm running mdns on the other device and it send packets every few secs. on my device side avahi is supposed to respond. but since nothing comes up at port, i guess somethign is wrong at network driver level Jan 12 14:39:03 sorry for writing that stuff twice. Jan 12 14:43:01 LetoThe2nd: which device tree compiler to use for kernel 3.19? Jan 12 14:44:12 cityoflights2: why not the one that comes with the kernel? and why do you ask *me*? Jan 12 14:44:34 I know you, so I ask you Jan 12 14:44:43 I also know rcn-ee , but he is away Jan 12 14:45:02 I want to over lay gpio48 and pull it up Jan 12 14:45:13 so I first need to compile the dts - right? Jan 12 14:46:28 cityoflights2: if you want to use a dts, or dtso, then you need to compile it. correct. Jan 12 14:47:37 is the format of the dtb used in kernel 3.8 the same as used in 3.19? Jan 12 14:47:52 I have bought 2 BBB. One is marked with a red line on bottom and one is with a green mark. What does that mean? Jan 12 14:48:24 cityoflights2: its basically meant to be that way, but upward compatibility has been broken several times. so don't count on it. Jan 12 14:49:02 ok, so all I need now is to figure out how to load my dtb to the kernel Jan 12 14:49:17 WL18xx I keep on getting this error message: ti-connectivity/wl18xx-fw-4.binFW is ti-connectivity/wl18xx-fw-4.bin Jan 12 14:49:17 wlcore: ERROR could not get firmware ti-connectivity/wl18xx-fw-4.bin: -2 Jan 12 14:49:17 but I do have the firmware in /lib/firmware/ti-connectivity. I'm debugging the compat_firmware_class att the moment.. Jan 12 14:50:13 I mean, the /sys/devices/bone_capemgr.9/slots doesn't exist anymore Jan 12 14:50:35 I read all the device tree papers, very long Jan 12 19:48:09 hi rcn-ee, I followed all your advice regarding the new device tree , but still need your help Jan 12 19:48:43 rcn-ee: any more hints for https://github.com/beagleboard/kernel/issues/64 ? Jan 12 19:48:54 i tried the verified answer from TI website but same issue pops again Jan 12 19:49:33 1. If I want to use 3.19 kernel , must I use a new dtc compiler? Jan 12 19:49:59 dtc 1.4.0 is fine Jan 12 19:50:05 2. can I use a file I build using dtc on a 3.8 system? Jan 12 19:50:33 vvu, it's usb, it's just plain broken in v3.8.x... copy everything from v3.14.x and it'll work fine.. Jan 12 19:51:35 so switch to new kernel ? Jan 12 19:51:50 3. as there is no more slots, how to load an overlay now? I tried the /config/devices , it doesn't work Jan 12 19:52:16 vvu, for usb.. god yes.. Jan 12 19:52:55 rcn-ee: sorry , but you are the only one here that actually understand this, I tried asking others Jan 12 19:53:02 cityLights, *.dtb is a firmware... it "should" be kernel independent, as long as anythin gyou defined in it, is "properly" upstream.. Jan 12 19:53:59 2. great I will use an overlay I build under 3.8 using the famous tutorial at http://derekmolloy.ie/gpios-on-the-beaglebone-black-using-device-tree-overlays/ Jan 12 19:54:58 now I read gpio.txt as well as a few more , but don't understand how to load the dtbo file Jan 12 19:55:52 the patch for OF: DT-Overlay configfs interface needs to be fixed.. /config/device-tree/overlays/foo so it's not enabled in my rc4 tree.. Jan 12 19:56:22 used to be echo DM-GPIO-Test > /sys/devices/bone_capemgr.9/slots Jan 12 19:56:26 so for 3.19: use the dtb-rebuilder: https://github.com/RobertCNelson/dtb-rebuilder/tree/3.19.x Jan 12 19:57:12 yes, I tried this: /config/device-tree/overlays/foo as you answered two weeks ago Jan 12 19:57:17 didn't work.... Jan 12 19:57:34 it doesn't work, as it doesn't build... Jan 12 19:58:03 there's a tight function macro i need to re-work.. Jan 12 19:58:04 ok, so git clone ; make Jan 12 19:58:39 but the r8188eu usb wifi works out of the box Jan 12 20:00:32 and the its reception is far better then previous nano wifi Jan 12 20:05:24 + ti's continuing to hack on that 3.14. tree, so they keep adding more usb workarounds... Jan 12 20:19:35 rcn-ee : I don't understand the ti line, is TI contributing to the kerenl? Jan 12 20:19:47 is it importent for BBB users? Jan 12 20:20:34 this is the base, we are using for v3.14.x: http://git.ti.com/gitweb/?p=ti-linux-kernel/ti-linux-kernel.git;a=shortlog;h=refs/heads/ti-linux-3.14.y it's ti's 3.14.x maintance release for all am335x + devices.. Jan 12 20:21:01 previous wifi usb mini adaptor was the edimax, the dlink based on the r8188eu is a bit bigger, but far better Jan 12 20:21:34 I mean I am using in a dead spot at the house, and the edimax doens't work, but it does, so thanks Jan 12 20:26:09 wicked. the git master branch is empty at dtb-rebuilder.git Jan 12 20:26:26 lets see origin/3.19.x Jan 12 20:34:04 so does this TI kernel branch , means , new features from the same OMAP, or better operation? Jan 12 20:36:13 it's the kernel branch ti supports for their customers... In our case it's handy for usb/ethernet/pm/pru for the bbb... For the BeagleBoard-X15, well lots of things.. Jan 12 20:36:58 rcn-ee: ok I make install your dtb-rebuilder , and even rebooter, what need I do now? Jan 12 20:37:55 what do you mean? Unless you change anything, it'll be exactly the same *.dtb's that were installed... Just build newer... Jan 12 20:38:25 the dtb-rebuilder, just allows you to easily tweak the *.dtb "without" spending lots of time rebuilding the fulll kernel.. Jan 12 20:38:58 I see, but how to load my overlay? Jan 12 20:39:18 you take your overlay settings... add your pinmux changes to the *.dtb file you boot.. Jan 12 20:39:18 need I put my dtbo in /boot? Jan 12 20:39:21 and reboot? Jan 12 20:40:15 cityLights: I think rcn-ee is saying you can't use dtbo files anymore Jan 12 20:40:20 with the 3.14 kernel Jan 12 20:40:31 you have to merge everything into a single dtb file Jan 12 20:41:28 this is something that is changing, I believe the capemanager had to be rewritten and it isn't done yet Jan 12 20:41:55 cityLights, if you want to load a *.dtbo on a non 3.8 kernel: you need to fix this patch on v3.19.x: https://github.com/RobertCNelson/linux-dev/blob/master/patches/overlay/0001-OF-DT-Overlay-configfs-interface-v3.patch Jan 12 20:42:05 so I need to edit the am335x-boneblack.dts file, then compile it using dtc 1.4.0 ? Jan 12 20:42:24 ddrown: aha... Jan 12 20:43:11 rcn-ee: I just want the easiest way to set pull-up for gpio48 Jan 12 20:43:21 I have the device tree overlay file Jan 12 20:43:39 I can add these lines to a given file Jan 12 20:43:47 cityLights, use the dtb-rebuilder... just add the gpio48 pinmux... easy peasy.. Jan 12 20:43:53 in fact I sould have used the include / / Jan 12 20:44:56 sorry, but I must be missing something simple Jan 12 20:45:27 In the dtb-rebuilder directory I see binaries only in the scripts directory Jan 12 20:45:55 when you say "use the dtb-rebuilder" which binary need I use ? Jan 12 20:45:58 src/arm/ Jan 12 20:46:13 cityLights, it "builds" *.dtb's... Jan 12 20:46:29 just use 3.8.. Jan 12 20:47:17 in the dtb-rebuilder/src/arm i see just dts/dtsi/dtb files Jan 12 20:47:50 cityLights, well.. what are you expecting? Jan 12 20:48:55 I expected a file called dtb-rebuilder which is executable and allows me to load a dtb file Jan 12 20:49:05 right? Jan 12 20:50:13 on the other hand , how can I tell which dtb file my BBB used to boot, so I can add a line in it Jan 12 20:50:20 source files you need to edit are under: /src/arm/ Jan 12 20:50:26 I mean add a line in the coresponding dts Jan 12 20:50:32 commands are: make; make clean; sudo make install Jan 12 20:50:53 o, so make turns the dts->dtb ? Jan 12 20:50:54 the "bbb" loads one specific *.dtb look it up.. Jan 12 20:51:12 unless you tell it to load a different one via: "dtb=" in /boot/uEnv.txt Jan 12 20:51:29 right , I thought the bbb used a single dtb... Jan 12 20:52:11 you also incorrectly assumed dtb-rebuilder was only for the bbb.. Jan 12 20:52:46 this has been written up here: http://elinux.org/Beagleboard:Capes_3.8_to_3.14#Custom_dtb Jan 12 20:52:49