**** BEGIN LOGGING AT Wed Dec 02 03:00:59 2015 Dec 02 04:42:35 is there some simple, non-linux-based code around for the BBB that will perform minimal initialization and flash an on-board LED? Dec 02 04:42:49 i saw this: http://beagleboard.org/Support/BoneScript/demo_blinkled Dec 02 04:43:15 but i think this may be simply running a program under linux. not sure though Dec 02 04:43:31 I posted such code to E2E, both a pure assembly version and a C++ version Dec 02 04:44:06 it even toggled the leds in response to pressing the S2 button (irq-driven) Dec 02 04:45:00 zmatt: sounds like exactly what i need. can you point please me more specifically? Dec 02 04:45:37 how do you get the code loaded? using the sdcard? Dec 02 04:45:59 sheesh. can you please point me ... Dec 02 04:46:05 yes looking already Dec 02 04:46:47 I think I'll just upload it to my own server, just want to check the files to see if I do anything there that I'd consider yucky nowadays ;) Dec 02 04:47:15 cool, thanks. Dec 02 04:49:56 only thing is, I can package up the assembly version no problem... for the C++ version I can only point you to the old one I posted on E2E quite a while ago and I'm not sure whether I do things there I'd consider to be a bad idea nowadays Dec 02 04:50:06 assembly is fine Dec 02 04:50:54 preferable, even. Dec 02 04:51:05 what cross-assembler do you use? Dec 02 04:51:22 are you building under windows or linux? Dec 02 04:51:25 (linux here) Dec 02 04:53:08 linux Dec 02 04:53:18 you can use pretty much any arm gcc toolchain Dec 02 04:53:27 arm-none-eabi, arm-linux-gnueabihf, doesn't matter Dec 02 04:53:47 since I don't use any libs, headers, even startfiles Dec 02 04:55:25 i see. that would be great. Dec 02 04:56:37 i saw this, but it's not complete: https://e2e.ti.com/support/arm/sitara_arm/f/791/p/312081/1085682 Dec 02 05:01:49 hum, this is odd, my code no longer links with the current debian arm-linux-gnueabihf toolchain Dec 02 05:02:28 what link error? Dec 02 05:03:45 fixed it, was missing -static ... it probably didn't work before either with this simplified Makefile (that replaced the behemoth of a Makefile derived from my baremetal C++ codebase) Dec 02 05:04:11 static libraries? Dec 02 05:04:12 the simplified one worked with arm-none-eabi since it has no dynamic linking anyway Dec 02 05:04:18 /usr/lib/gcc-cross/arm-linux-gnueabihf/5/../../../../arm-linux-gnueabihf/bin/ld: error: no memory region specified for loadable section `.interp' Dec 02 05:04:45 like .interp ? what? oh crap it's trying to use the dynamic linker? hell no Dec 02 05:05:10 I should probably *test* whether the code works now, but I'm lazy so I'll just push it to github ;) Dec 02 05:05:37 what is the output, an elf file? Dec 02 05:06:08 how do you get that loaded at boot time? Dec 02 05:06:43 https://github.com/mvduin/bbb-asm-demo Dec 02 05:06:50 it produces 3 files Dec 02 05:07:15 demo.elf (for JTAG upload or debugging), demo.bin (peripheral boot), demo.MLO (memory boot) Dec 02 05:08:10 in memory boot, how do you get it into memory? Dec 02 05:08:26 thanks, btw!Q Dec 02 05:08:28 put the demo.MLO on a fat-formatted sd card, rename it to MLO, boot the BBB from it Dec 02 05:08:49 oh, very good! Dec 02 05:08:53 I usually use peripheral boot, specifically netboot Dec 02 05:09:19 is that a boot mode? (netboot) Dec 02 05:09:46 nm. Dec 02 05:09:52 i'll do the mlo approach. Dec 02 05:09:59 yes, ROM supports booting via a ton of ways Dec 02 05:10:13 let me do some playing around. Dec 02 05:10:36 fyi, we just got our custom board a week or so and now am trying to bring up the processor. Dec 02 05:10:56 do check then whether the GPIOs being toggled are appropriate in your case Dec 02 05:11:06 what GPIOS? Dec 02 05:11:22 the leds of the BBB Dec 02 05:11:23 oh Dec 02 05:11:35 oh, yes, copied those from the BBB Dec 02 05:11:55 just check main.S to see what the program actually does Dec 02 05:12:22 is there a way to detect if the processor is running the rom boot code without loading anything? Dec 02 05:12:57 what boot mode do you have configured? (sysboot0-15) Dec 02 05:13:05 403c Dec 02 05:13:08 no Dec 02 05:13:10 403c Dec 02 05:13:21 bah! eyes are getting cloudy... Dec 02 05:13:43 with the normal boot mode switch to change it. Dec 02 05:13:59 we copied a lot of stuff from the bbb, just not everything Dec 02 05:14:09 then if ROM doesn't find anything bootable on eMMC or SD, it will try uart boot Dec 02 05:14:16 you see CCCCCC on the serial port Dec 02 05:14:29 next in the boot order is usb Dec 02 05:14:48 which serial port? Dec 02 05:14:52 uart0 Dec 02 05:15:04 same as the serial console on the BBB Dec 02 05:15:56 next in the boot order is USB... if connected as device to a host it'll appear as a network interface (RNDIS) Dec 02 05:17:49 note that iirc once enumerated as usb device the boot process will hang in attempting to boot via usb Dec 02 05:17:57 zmatt.. you talkin to urself again .. :p Dec 02 05:18:07 what is it looking for on the sdcard? i.e., how does it know to skip the sdcard or not? obviously there is the card_detected output, but does it look for a master boot record or somesuch (or a magic number)? Dec 02 05:18:23 or am I suffering over-aggressive ignore filters once again :/ Dec 02 05:18:23 yates: yes, described in detail in the TRM Dec 02 05:18:36 veremit: that Dec 02 05:18:38 ok, i'll look there. Dec 02 05:18:42 hrm great Dec 02 05:18:43 thanks matt! Dec 02 05:19:07 yates: but summary is: FAT-formatted partition, must be first partition and marked bootable Dec 02 05:19:15 looks for file named "MLO" in the root of the filesystem Dec 02 05:19:43 right, but we're not even sure our sdcard interface is good. Dec 02 05:19:52 you don't have JTAG ? Dec 02 05:20:18 we actually had a mechanical problem with the holder, and had to lift it with some funky solder work Dec 02 05:20:19 nope Dec 02 05:20:39 ... bringing up a custom board with no jtag Dec 02 05:20:41 we thought we'd wait and see if we absolutely needed it before spending the money. Dec 02 05:20:48 that's a bit like bringing up a custom kernel without a serial console Dec 02 05:21:00 yeah, i hear you. i asked for it at least 2 times, but the PM is cheap! Dec 02 05:21:00 if it works, it works Dec 02 05:21:12 program manager Dec 02 05:21:18 if it doesn't work, you're blindly groping around in the dark Dec 02 05:21:28 do you have ethernet on your board? Dec 02 05:21:32 ohhh, aahhhh, ... Dec 02 05:21:36 yes! Dec 02 05:21:46 i insisted on that - they were going to remove ti! Dec 02 05:21:50 it Dec 02 05:22:33 can you pull sysboot to something that includes ethernet? on the BBB I pull down sysboot2 and pull up sysboot0 Dec 02 05:22:55 that yields boot order { spi, mmc0, eth, uart } Dec 02 05:23:00 if we can't get the assembly code to work, then i think i'm going to insist we get a jtag Dec 02 05:23:13 zmatt: how do you boot off the eth? Dec 02 05:23:32 well first see if it attempts the boot at all Dec 02 05:23:39 how? Dec 02 05:23:44 what are we looking for? Dec 02 05:23:45 if I understand correctly you're still looking for *some* sign of life right? Dec 02 05:23:46 packets Dec 02 05:23:49 yes. Dec 02 05:23:51 bootp requests to be specific Dec 02 05:24:28 we are seeing the following: 1) the 24 MHz clkout1 is running, 2) some clocks to the SDCARD, 3) some activity on uart0 Dec 02 05:24:37 so it looks like it is alive, somewhat.. Dec 02 05:24:48 ok so ROM is running and trying to boot Dec 02 05:25:04 does the boot code require the DRAM to be present and working? (even just a few bytes of it?) Dec 02 05:25:09 no Dec 02 05:25:23 all register based? Dec 02 05:25:25 neither does my application Dec 02 05:25:30 there's 128 KB of on-chip SRAM Dec 02 05:25:34 oh! Dec 02 05:25:39 gotta read the damn trm. Dec 02 05:25:41 ... Dec 02 05:25:45 RTFM! Dec 02 05:26:01 yes, it's only like 5000 pages or something, sheesh, get reading already Dec 02 05:26:02 ;) Dec 02 05:26:25 ok, this has been most helpful! if you're even in NC, USA, I owe you two guiness stout! Dec 02 05:26:34 ever Dec 02 05:26:42 hehe, I don't drink beer, but thanks Dec 02 05:27:18 nite! bedtime here. Dec 02 05:27:36 I'm about to push an example dnsmasq config Dec 02 05:27:38 for netbooting Dec 02 05:28:58 Hello... Dec 02 05:31:57 Okay, does anyone use Pocket Sphinx or another type of voice recognition on their BBB? Dec 02 05:32:49 I am asking b/c my alsamixer, espeak, and Pocket Sphinx are not working out well. Dec 02 05:33:11 Send support for voice recog. stuff on the BBB if you dare. Dec 02 05:43:41 pocket sphinx works fine Dec 02 05:43:54 a bit slow but functional Dec 02 05:50:12 Oh. Dec 02 05:50:14 Okay. Dec 02 05:50:29 I tried espeak and alsamixer. They say this works fine too. Dec 02 05:50:58 I am having an issue with my BBB and the alsamixer software. I can only play 8 bit sound for some reason. Dec 02 05:51:47 ds2: If you have some extra time, please let me know how I can change the alsamixer setting for bits of sound. Dec 02 05:54:12 what are you using for audio input? Dec 02 05:54:27 I don't use alsa, PS can take input from a raw audio stream Dec 02 05:54:54 I am using a USB Sound Adapter. Dec 02 05:55:16 I get sound on the BBB but it is in 8 bit. Dec 02 05:55:46 I have dealt with others who say it works and I have to configure it but the "--help" options only supply so much info. Dec 02 05:56:56 how do you know it is 8bit? Dec 02 05:57:03 I would have to hate to buy new hardware to make this sound work. Dang man. Oh... Dec 02 05:57:15 It says when it plays. It states it. Dec 02 05:57:23 how are you playing it then? Dec 02 05:57:32 most USB audio drivers are not 8 bit Dec 02 05:57:52 It plays but in 8 bit and it sounds like garble, slush, garble. Dec 02 05:58:17 how are you playing it? Dec 02 05:58:28 It is not making music from my .wav file. It is playing a constant mucus of sounds. Dec 02 05:58:42 Oh headphones. Dec 02 05:59:14 HOW Dec 02 05:59:17 I tried speakers and I tried two separate speakers. I get the same audio out of it. Dec 02 05:59:26 are you telling me you plug in headphones and magically audio comes out? Dec 02 05:59:40 The USB adapter has a plug for speakers and a mic. Dec 02 05:59:42 Set .. your beagle does not have an audio jack .. WHERE is the sound coming from? Dec 02 05:59:54 The usb adapter. Dec 02 06:00:00 progress :) Dec 02 06:00:17 I typed that up there earlier. Dec 02 06:00:24 Meany. Dec 02 06:00:46 I donno what kind of usb adapter you have. most do not play sounds without doing something Dec 02 06:00:58 prehaps you have a MP3 player on a dongle? Dec 02 06:01:20 I will tell you exactly what my prompt says when I play .wav files. Dec 02 06:01:42 No...it is a .wav file that I downloaded onto my BBB by way of a specific program. Dec 02 06:01:57 I used WinSCP. Dec 02 06:02:24 what command do you run to attempt to play the .wav file? Dec 02 06:02:39 aplay blah.wav Dec 02 06:03:10 and you've sorted an /etc/asound.conf ? Dec 02 06:03:31 Sorted it how? Dec 02 06:03:40 I can but I have not done such a thing. Dec 02 06:04:46 I have been in my...oh. You mean "cat /proc/asound/cards?" Dec 02 06:04:57 no Dec 02 06:05:01 Oh? Dec 02 06:05:03 I mean configuring ALSA Dec 02 06:05:09 so that it doesnt do stupid shit Dec 02 06:05:10 :) Dec 02 06:05:42 Ha. I need support on this I guess. I tried to use their "--help" function and the prompt only gives me so many options. Dec 02 06:05:52 I will go to /etc/asound real quick. Dec 02 06:05:54 I wouldn't expect udev/etc to set up alsa for you when you plug it in Dec 02 06:06:08 thats more like a pulseaudio trick Dec 02 06:06:59 /etc/asound.conf may not exist .. you might have to create it Dec 02 06:07:20 Yea...it does not. What do I do when I create it? Dec 02 06:08:40 I have a file I need to type into "nano" or "vim" to tell the BBB to go use USB instead of other things for sound. Dec 02 06:10:38 Unsigned 8 bit, Rate 8000 Hz, Mono Dec 02 06:10:38 ...This is what it says. Dec 02 06:11:16 niiiiice Dec 02 06:11:45 Yea right. I got garbage coming out of my headphones. Dec 02 06:12:10 I need it to say 16 bit, Rate 24000 Hz, Stereo Dec 02 06:13:17 I was very happy with life and then I started to listen to this book. Ouch, I have taken almost life's last breath doing good by the book. Dec 02 06:13:20 Ouch is right. Dec 02 06:14:07 Sound adapters, alsamixer, espeak, and the BBB have taken my nice pronunciation and turned it into %&&$, %&#@@, and %*#*%. Dec 02 06:17:01 ...I do not know if it is the adapter, the software, or a combination of it. Dec 02 06:21:19 Hell...it could be the book. I will give up for now. Thank you for your support. Dec 02 06:21:33 http://andicelabs.com/2014/03/usb-audio-beaglebone/ Dec 02 06:22:00 99% sure it'll be trying to use hdmi as your default device first off ... Dec 02 06:23:36 It does...but I can change that with typing in a file. The file reads: pcm.!defualt sysdefault: Dec 02 06:25:14 if you are playing a wave file, check the file itself Dec 02 06:25:25 it may have bad headers or missing headers (i.e. not atrue wave file) Dec 02 06:25:39 try using sox to set the same rate and regenerate the wav file Dec 02 06:25:48 in any case, I don't see the relevance of this to PS Dec 02 06:25:50 Okay. I will check the .wav file. Dec 02 06:26:53 you can also try to give parameters to aplay Dec 02 06:27:28 i.e. aplay -r 44000 dumb.wav Dec 02 06:27:51 but wav is suppose to have a header specifying the sample rate and sample type Dec 02 06:27:59 Okay...thank you. I was unaware of that. Dec 02 06:28:18 there are also raw files. sox can convert between them Dec 02 06:28:45 PS should configure ALSA for the format it wants (assuming you are using ALSA) Dec 02 06:29:14 could try something other than aplay as well ... Dec 02 06:29:21 Oh? Dec 02 06:29:24 Like... Dec 02 06:29:29 sox Dec 02 06:29:30 ;) Dec 02 06:29:39 sox plays music too? Dec 02 06:29:57 alsaplayer? vlc? Dec 02 06:30:04 sox is amazing Dec 02 06:30:10 I use alsamixer. Dec 02 06:30:16 I will try sox. Dec 02 06:30:42 alsamixer is not relevant to the is problem Dec 02 06:30:52 Oh. Dec 02 06:31:05 Okay...So, it has to be the .wav file or my hardware set up? Dec 02 06:31:28 wav file setup Dec 02 06:31:46 you can always ditch the hw Dec 02 06:32:12 I can see now where I may have saved it as a .wav file in the wrong area. Dec 02 06:32:16 audio can be done with a few dollars of interfacing on the BB* Dec 02 06:32:22 Oh? Dec 02 06:32:31 wav file support in software can vary quite a bit Dec 02 06:32:38 hence the usefulness of sox Dec 02 06:32:44 I need a mic and a speaker. Dec 02 06:33:10 My mic is for the machine understanding commands and the speaker is for it to reply. Dec 02 06:33:17 you can do audio out with a transistor (<$0.25) and a few resistors Dec 02 06:33:24 audio in can be done via the ADC Dec 02 06:33:34 The bot is built and the motors and motor controllers work. Dec 02 06:33:35 that's my setup. got better things to do with the USB port :D Dec 02 06:33:53 where Geographically are you? Dec 02 06:34:08 America! Dec 02 06:34:12 what state? Dec 02 06:34:18 Louisiana! Dec 02 06:34:38 Where there are no nice shops for electronics. Dec 02 06:34:42 working on something not too far from that Dec 02 06:34:46 Oh? Dec 02 06:34:50 who needs local shops.Mouser and Digikey :D Dec 02 06:34:55 Yea...I know. Dec 02 06:35:13 got the BBB stuff working; need to build the mechanicals. Dec 02 06:35:25 what are you using for motor and motor controller Dec 02 06:35:26 ? Dec 02 06:35:27 I use both. I am actually looking for element14 to come through with a 12-pin, two-row 0.05" set of headers. Dec 02 06:35:39 I have a bot with two wheels. Dec 02 06:36:06 Newark is good for those odder headers Dec 02 06:36:09 self balancing? Dec 02 06:36:23 No...it has a roller for the third wheel. Dec 02 06:36:41 that's what i am shooting for too Dec 02 06:36:57 A self-balancing or specific headers? Dec 02 06:37:20 2 driven wheel + a free roller Dec 02 06:37:32 Oh...they have a ton of those on the market. Dec 02 06:37:54 i think we are desync/ed. Dec 02 06:38:02 I went to robotshop.com and they had enough to pick from... Dec 02 06:38:11 Oh... Dec 02 06:38:27 What kind of motor are you going for? Dec 02 06:39:05 I have two Tamiya motors Dec 02 06:39:48 small robot? Dec 02 06:39:53 Yea. Dec 02 06:40:19 It is about 7 x 5. Dec 02 06:40:46 how are you powering it? that doesn't leave much room for batteries Dec 02 06:41:19 I know...I have to use batteries for the motor controllers and I will have a power bank for the BBB. Dec 02 06:42:07 what was your reasoning for picking those motors? Dec 02 06:43:04 I am going with a continuous rotation servo to keep it simple Dec 02 06:43:20 None...it came with my set up. Dec 02 06:43:52 Ohhhh you got kit :D Dec 02 06:43:57 They are DAGU and not Tamiya. SOrry. Dec 02 06:44:22 Sort of. It was a Chassie and motors with some rims. Dec 02 06:44:36 It is a Magician Chassis. Dec 02 06:44:51 I think they sell a lot of them. Dec 02 06:45:10 I felt adventurous and decided to build the chasis but that's taking forever :/ Dec 02 06:45:21 Augh...what did you use? Dec 02 06:45:32 Wood or plastic? Dec 02 06:45:42 Metal? Dec 02 06:45:44 metal Dec 02 06:45:47 Yea boy! Dec 02 06:45:55 sheetmetal to make a tubular frame Dec 02 06:46:01 Oh. Dec 02 06:46:13 figure it was the easiest way to mount the servos Dec 02 06:46:27 I do not have any laser cutter or high pressured water cutter. Dec 02 06:46:40 shears from the hw store Dec 02 06:46:45 I cannot use the heavy metal yet. Dec 02 06:46:49 and a metal brake from harbor freight Dec 02 06:46:55 no, this is thin sheet metal Dec 02 06:47:00 Cool... Dec 02 06:47:13 I always wanted one of those water cutters. Dec 02 06:47:19 Those things look impressive. Dec 02 06:47:28 $$$$$ and takes a lot of space Dec 02 06:47:38 I know. I looked 'em up. Too much. Dec 02 06:47:39 easier to mill them on a bridgeport Dec 02 06:48:00 or CNC them on a HAAS. No need to be fancy with a waterjet Dec 02 06:48:05 People are into routers. I like laser and water. Dec 02 06:48:33 I see people make these homemade waterjets and xyz routers. Dec 02 06:48:49 it is cool. I think the FPGA and BBB can attempt this. Dec 02 06:48:54 routers don't work well on heavier metal Dec 02 06:49:08 Yep. Dec 02 06:49:13 would much rather have a bridgeport then a flimsy router Dec 02 06:49:46 I have a trimmer. I do not have a router. It sucks. So, I have to think small. Dec 02 06:50:10 as in those rotozip thing? Dec 02 06:50:16 Trim here, cut there, route this...but I have to remember to keep it small. Thinner than sheet metal too. Dec 02 06:50:40 I got that but I was talking about the multi-tool I have. Dec 02 06:51:09 ... Dec 02 06:51:43 it isn't the tools itself. it is more important to know how to apply it Dec 02 06:51:53 yea. You are very right. Dec 02 06:52:09 Power on bigger items is not always a winner. Dec 02 06:52:38 So anyway... Dec 02 06:52:54 Good luck sorting out the audio... should be a fun journey Dec 02 06:53:02 I got this bot, the assembly is done, the hardware is here and I am working with it, but... Dec 02 06:53:08 Oh...thank you. Dec 02 06:53:41 I found where my audio "alsa" is located in my directories/tree. Dec 02 06:54:24 I saw the software behind it and I viewed it with nano. Dec 02 06:54:56 I was going to change some. I figured it would show the 8 vs. 16 bit portion of code in there. Dec 02 06:55:15 ...I was wrong. But thank you, I will keep plugging away. Dec 02 07:58:09 http://alsa.opensrc.org/Aplay shows a "help" page with some nice info. on alsamixer and the Utils from alsa. Enjoy... Dec 02 07:58:48 Oh and if you are using alsamixer, the 16 bit version is in there. Boo-Ya, my skiggity! Dec 02 08:09:25 I jumped the gun again. Dang! Dec 02 08:20:56 Well, I found their wiki. This is good news. Yet...I am stuck again. Onto plugging! boy! Dec 02 08:29:13 Okay, okay, okay. I got it now. Dec 02 08:29:49 I have a .wma file and not a .wav file. I think this is my concern. Do you know where I can exchange .wma into .wav files? Dec 02 08:33:28 Set_ .. ffmpeg is your friend Dec 02 08:33:45 but it shouldn't make a difference .. most players can play anything these days Dec 02 08:35:04 That is what I thought but it distinctly told me to use a .wav file. Dec 02 08:35:26 books/tutorials such wonderful things........ Dec 02 08:35:29 I am about to use media.io. Dec 02 08:35:48 Yea...I know. I just wanted some input on this one. Dec 02 08:36:10 Thank you again. Dec 02 09:00:40 Why me? Dec 02 09:03:57 because the universe hates you, deeply. Dec 02 09:06:12 I know. Dec 02 09:07:48 Okay...get this. I got alsamixer. It works. We all know. I got me a file converter. The .wav file has been made. When I "aplay myfile.wav" is when I get playback:2715:read error. Dec 02 09:07:52 Help a brother! Dec 02 09:08:32 I looked up that specific error. Notta. Dec 02 09:11:09 why dont you use something like mplayer,mpv,... and play the file directly instead of using aplay which is only ment as a demonstrator how to use alsa? Dec 02 09:11:50 Oh...I need to use this file to record sound, write sound to verbal instruction, and get feedback from my verbal instruction. Dec 02 09:12:14 alsamixer does this partially with espeak and PocketSphinx. Dec 02 09:15:32 I have some good news though. I got the mic to work now. Dec 02 09:15:34 Boy! Dec 02 10:14:43 is this irc channel the right place to talk about the BB-X15 kernel devicetree files ? Dec 02 10:15:03 possibly, yeah Dec 02 10:16:12 i had a look into them, and the cpu thermal zone is defined in omap4-cpu-thermal.dtsi Dec 02 10:16:19 * tbr should dust off his X15 again and play with it Dec 02 10:16:48 i don't understand why alert temp is defined to 100°C and crit temp to 125°C Dec 02 10:17:55 it seems to me crit temp should be teemp junction, and afaik it's 105°C (online datasheet) Dec 02 10:21:38 there is currently a discussion about temperature measurements and shutdown on the X15 group/mailing list. have you checked that? Dec 02 10:22:07 yes, that's why i talk about it Dec 02 10:24:41 in fact, i saw a problem which seems similar with the dragonboard 410c, and they seem to have fixed it after fixing those values https://www.96boards.org/forums/topic/recommended-heatsinkfan-for-410c/#post-9838 Dec 02 10:25:48 might want to bring it up on the list. I'm not sure if NishanthMenon reads hilights here (also texas is still asleep) Dec 02 10:37:34 i just tried to reply to joe...@gmail.com message in the list but i don't see my message there Dec 02 10:37:51 i guess there is some filter in place or so Dec 02 10:39:10 not sure. Dec 02 10:39:15 google groups is weird Dec 02 10:39:26 I only use its mailinglist mode Dec 02 10:40:05 btw: I do think you have a point. 125 °C sounds a tad high for a soc... Dec 02 10:41:51 from what i understand, 125°C is the highest temperature the on-chip sensor can read Dec 02 10:47:24 to send a message there, i just need to send a mail to beagleboard-x15@googlegroups.com, right ? Dec 02 10:48:32 after subscribing, yes. else it will end up in a moderation queue AFAIU Dec 02 10:51:24 ok thx Dec 02 10:51:56 my subscription request is still not accepted, that's why my message doesn't appear in the list Dec 02 12:24:53 tbr, I am awake now. Dec 02 12:26:50 nobe, yeah - the initial 125C is based on autograde SoC variant. I think you are correct that the industrial variant of SoC is 105C and the dts should have an override Dec 02 12:27:34 nobe, i think the trouble started by having to reuse much of the configurations across SoC variations.. Dec 02 12:30:29 going offline for an hr or so.. will be back. nobe : ping me if you need me once I am back.. Dec 02 16:32:11 hi all Dec 02 16:32:29 I have one question about beaglboard green Dec 02 16:33:05 please, anybody may help me? Dec 02 16:55:36 Hello.. it may sound stupid, but... capemanager sucesfully loaded a dtbo that I can't find on disk, in /lib/firmware/. Have dtbo been placed somewhere else recently? Dec 02 17:16:44 Just probing around, how much demand do you guys think there is for a kernel module for hd44780 compatible LCDs? I am working on one for work, and we are thinking about publishing it open source. It loads all information from the device tree about which gpio to use and the screen dimensions. It currently only operates in 4-bit mode(read, write, and hopefully soon some busy flag checking) as that is what our hardware is set up for here at wo Dec 02 17:18:00 abferm: sounds cool. I think I have two of those and was too lazy to do anything with them so far :D Dec 02 17:25:52 abferm: never seen on ein the kernel, I think it could be appreciated Dec 02 17:40:23 anyone have experiece with BeagleBoard-X15 ? Dec 02 17:40:38 vmcsnekke: I do, but that's not fair. :-) Dec 02 17:40:54 I've sent out a few dozen boards so far to early developers. Dec 02 17:41:11 * jkridner looks at _av500_ to see if he ever powered it up. Dec 02 17:41:23 abferm: using lcdc in "lidd" (remote framebuffer) mode you mean? Dec 02 17:41:36 ok, TRM says the first stage bootloader (u-boot I guess) cannot be larger than 504k, so do we need 2 bootloaders ? Dec 02 17:41:56 vmcsnekke: we use SPL as the first stage bootloader. Dec 02 17:42:02 what is SPL ? Dec 02 17:42:07 support is already in mainline u-boot. Dec 02 17:42:20 it is a special configuration of u-boot to make it small. Dec 02 17:42:23 wait what, u-boot does not fit in half a fucking megabyte?!? seriously? Dec 02 17:42:33 we essentially use u-boot (SPL) to load u-boot (normal). Dec 02 17:43:03 zmatt, i wanted to ask same question Dec 02 17:43:14 zmatt: it is rather scalable. u-boot is like a freaking operating system without a scheduler. loaded with drivers, just no traditional OS stuff. Dec 02 17:43:16 the plain u-boot does not fir in 504k ??? Dec 02 17:43:38 jkridner: you could fit a multitasking OS in that much space just fine Dec 02 17:43:42 I don't know how big it is right now. I just know it can grow pretty big. Dec 02 17:43:48 zmatt: true. Dec 02 17:45:15 how to select between SPL and non-SPL ? Dec 02 17:45:15 different build target ? Dec 02 17:45:15 again, it has a ton of drivers and junk. very scalable though, thus SPL is the same code as u-boot, just a build config away. Dec 02 17:45:15 vmcsnekke: yup. Dec 02 17:45:15 abferm: but just fyi, lcdc has explicit support for the Hitachi protocol (and a few others) Dec 02 17:45:15 ok, thx a lot guys! Dec 02 17:46:12 I believe https://eewiki.net/display/linuxonarm/BeagleBoard-X15#BeagleBoard-X15-Bootloader:U-Boot instructions builds both, SPL and u-boot.img Dec 02 17:46:43 * jkridner wonders why it pulls from a TI fork. :( Dec 02 17:47:10 mainline u-boot does have x15 support. Dec 02 17:48:37 be sure to check out patches like http://git.denx.de/cgi-bin/gitweb.cgi?p=u-boot/u-boot-arm.git;a=commit;h=74cc8b097d9aff8b8671396f66ece4cf2de0a36d in the mainline Dec 02 17:49:07 most of the u-boot.img in debian are ~450-500k Dec 02 17:49:25 yeah, i've seen it, it sure looks like ti has done some serious open-source/community/call-it-whatever support businuss, which is good imo Dec 02 17:49:49 mainline X15 patches: http://git.denx.de/cgi-bin/gitweb.cgi?p=u-boot.git&a=search&h=HEAD&st=commit&s=x15 Dec 02 17:50:01 oh wait, more like ~225-350 Dec 02 17:50:31 did the FCC issues for the beagle-x15 get worked out? Dec 02 17:50:33 vmcsnekke: some thanks trini Dec 02 17:50:48 vagrantc: still working it. HDMI is the culprit. Dec 02 17:51:39 * vagrantc wonders why people can't just live with serial console Dec 02 17:53:32 lol Dec 02 17:53:34 Because touch screen serial consoles are so utterly annoying ;) Dec 02 17:54:07 if you need hdmi .. use a Rpi *cough, splutter* lol XD Dec 02 17:54:37 well to be fair the x15 is a *bit* more video-oriented than the bbb, or even the rpi :P Dec 02 17:54:58 zmatt .. you mean it actualyl has some ;P Dec 02 17:55:19 Eh, the x15 seems to be a decent ARM workstation . Dec 02 17:55:59 Spidler_ .. probably true . but shame about the price-point :) Dec 02 17:56:04 veremit: it supports 4 independent digital video outputs (hdmi + three 24-bit wide parallel video ports available via expansion headers) Dec 02 17:56:07 :P Dec 02 17:56:23 zmatt .. not bad :] Dec 02 17:57:29 although to use all of those I think you'd also sacrifice all of its 10 video input ports Dec 02 17:57:51 ah it's only dual-core :/ but the dsp cores are NIIICE Dec 02 17:57:53 but you'd need to check the pinmux tables to see what exactly can be used at the same time Dec 02 17:58:05 zmatt .. you haven't completed those yet?! ;] Dec 02 17:58:17 zmatt: I'm not using lcdc at all. Just changing pinmuxes and bit-banging gpio from kernel space. Dec 02 17:58:30 abferm: *shudder* Dec 02 17:58:39 veremit: I made a decent start.... https://goo.gl/S4t2lw Dec 02 17:58:49 zmatt .. good, good :) lol Dec 02 17:58:56 tea . brb Dec 02 18:00:02 It is a character LCD...I haven't seen any implementations anywhere. Dec 02 18:00:10 the signals tab still needs more work, but already shows the available iosets for the vin/vout ports (scroll to near the bottom) Dec 02 18:00:19 At least not read/write capable. Dec 02 18:02:00 abferm: see the LCDC chapter of the TRM, e.g. section 13.3.4 Dec 02 18:02:51 zmatt: but is it implemented in the ti lcdc driver? Dec 02 18:02:53 I've done a tiny bit of playing with it Dec 02 18:03:09 veremit: no, it's a totally completely different mode of operation of the lcdc Dec 02 18:03:26 zmatt: in which case he's starting from scratch anyway Dec 02 18:03:33 basically it's two different peripherals in one, with a global pinmux-bit between them at the toplevel Dec 02 18:03:55 [he] with no gender bias.... Dec 02 18:04:48 rcn been back since t'g ? Dec 02 18:04:52 veremit: yeah but constructing a bus protocol with gpio is obviously a lot more bother than having hardware do it for you Dec 02 18:05:04 plus you get support for dma-writes Dec 02 18:05:11 zmatt: woo. Dec 02 18:05:44 zmatt .. the only thing is, would you have to disable the existing lcdc to re-use it as a lcd driver? Dec 02 18:06:03 you'd change its "compatible" in DT Dec 02 18:06:07 and, more importantly, Does it support 8 bit?! lol Dec 02 18:06:21 working with nibbles is a pain Dec 02 18:06:48 4-bit and 8-bit yes Dec 02 18:07:26 see also the LCDC tab of https://goo.gl/Jkcg0w :P Dec 02 18:07:26 * veremit adds DT to the 'must-learn' list under python .. Dec 02 18:12:13 DT itself is really dumb and simple; what matters is how the kernel/drivers interpret the DT data Dec 02 18:12:55 zmatt.. presumably you got some limited control over that too Dec 02 18:13:21 for which you can check the binding docs in Documentation/devicetree/bindings/, then pee over it, and check the sources to see how they interpret it in reality Dec 02 18:14:13 sure you have control over that, it's called "patching the kernel" ;) Dec 02 18:27:32 zmatt .. its not like we adopt mainline kernel anyway .. anywhere .. anyhow lol Dec 02 18:35:09 pretty close, ask Robert... Dec 02 18:37:31 zmatt: Your kidding me... I wish I had seen that before I designed my hardware... Dec 02 18:41:17 lesson learned: next time, be sure to read the TRM cover to cover ;-) Dec 02 18:41:28 except for overlays maybe, most everything else was "almost" there... Dec 02 18:42:19 zmatt: the other way to look at this is my code should port to any hardware, not just am335x platforms. Dec 02 18:45:42 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Dec 02 18:45:50 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ... Dec 02 18:45:50 whoooops Dec 02 18:45:51 ... \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Dec 02 18:45:57 \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ... Dec 02 18:45:59 ... \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ Dec 02 18:46:05 wirry Dec 02 18:46:07 something went very wrong there :$ Dec 02 18:46:42 yeah Dec 02 18:46:44 ;) Dec 02 18:50:15 abferm: yeah, lcdc would limit your audience, although iirc omap's dss has something similar (RFB) Dec 02 18:50:41 but of course not compatible Dec 02 18:57:43 maybe future expansion could support lcdc if present. Dec 02 19:00:24 at least I'd suggest keeping a clean boundary along those lines Dec 02 19:01:20 since there may be more ways to create the actual bus signalling required Dec 02 19:07:55 <[Butch]> Folks, I have a super weird question for you. Dec 02 19:08:08 <[Butch]> I’m unable to run any bonescript from withing C9. Dec 02 19:08:38 <[Butch]> If I try the simplest program, I get the evil “object has no method ‘pinMode’” Dec 02 19:08:43 <[Butch]> Even something as simple as this: Dec 02 19:08:44 <[Butch]> var b = require('bonescript'); Dec 02 19:08:44 <[Butch]> var inputPin = 'P8_19'; Dec 02 19:08:45 <[Butch]> b.pinMode(inputPin, b.INPUT); Dec 02 19:09:01 <[Butch]> However, if I run the examples in the web page on the bone, things work. Dec 02 19:09:03 <[Butch]> Thoughts? Dec 02 19:09:17 <[Butch]> Well, thoughts other than, “He’s an idiot.” Dec 02 19:46:52 [Butch] .. probably an environment or permissions issue? Dec 02 19:48:33 <[Butch]> But c9 is supposed to be self-contained, right? I shouldn’t be able to casually screw something up? Dec 02 19:55:54 <[Butch]> I’m trying to update it. We’ll see if that fixes anything. Dec 02 20:03:48 [Butch].. lots of silly things to "go wrong" under linux Dec 02 20:04:05 rcn-ee - just the man I want to see .. :) how goes? Dec 02 20:06:26 good how you veremit Dec 02 20:09:02 not so bad, keepin busy, as ever :) Dec 02 20:09:45 wondered whether you'd met the aw h3 on anything yet? Dec 02 20:09:55 yeah things got busy over thanksgiving. ;) Dec 02 20:10:26 I figured .. hope you got some time off :) Dec 02 20:10:28 haven't picked up a h3 yet.. was looking at one.. Dec 02 20:11:05 I bought one .. ona whim .. orange pi plus .. quite nice board Dec 02 20:11:27 yeah that was the one i was looking at.. Dec 02 20:11:48 recommend it .. just .. the sw support isn't .. great, shall we say Dec 02 20:12:12 although things are definitely moving .. kernel support and uboot are both coming up Dec 02 20:12:35 yeah. ;) this is my current project, was helping mirkix on his cape, then next thing i knew i started ordering copter parts: https://github.com/mirkix/BBBMINI/blob/master/doc/pic/bbbminipcbquad.png Dec 02 20:12:53 lol @ copters ... Dec 02 20:13:05 beagle powered! ;) Dec 02 20:13:56 looks pretty awesome .. is it flying? Dec 02 20:17:37 veremit, yeap... flying on fedex's transport. ;) Dec 02 20:17:38 a swimming quad-copter Dec 02 20:17:48 it'll be a bucket of parts for next few weeks. Dec 02 20:19:29 ah hehe Dec 02 20:19:40 work-in-progress then :) Dec 02 20:23:47 <[Butch]> Well, the update of c9 didn’t help. :-( Dec 02 20:25:37 [Butch]: nope its more fundamental than that ... if the web page version works .. its the way you're doing it Dec 02 20:26:25 rcn-ee: so can I interest you in making a build package for the opi .. ;) Dec 02 20:27:11 I've got an image working on mine .. some guy in the forum has done a lot of work tidying up the distros .. think he's got a buildroot or similar going Dec 02 20:27:26 but its stuck at 3.4 kernel and some ancient uboot iirc. Dec 02 20:27:44 hold up .. lets fire the damn thing up Dec 02 20:28:07 it doesn't like my dvi video switch though .. at all .. Dec 02 20:32:05 rcn-ee: flashed the eMMC ok, networking, wifi all work .. running jessie with xfce currently here. Dec 02 20:32:18 rcn-ee: but its got this horrible boot process Dec 02 20:32:41 zmatt: you awake? Dec 02 20:33:03 .. he was .. Dec 02 20:33:03 <[Butch]> veremit: I don’t know what I could be doing wrong. I connect to c9 and just run one of the existing programs that has worked in the past. It blows up because apparently the bonescript library isn’t loading. At least that’s my guess, since calling b.pinMode fails with the dreaded “method not found.” Dec 02 20:35:55 [Butch]: I'm unfamiliar with c9 .. but you run everything through the webIDE, yes? Dec 02 20:36:40 or you -can- run through the webIDE .. and possibly else-how too .. Dec 02 20:37:09 <[Butch]> veremit: Yes, I can run it through the BBB web interface. Dec 02 20:37:52 [Butch]: this applies to your test script too? or just the examples? Dec 02 20:38:10 <[Butch]> Yes, the test script, too. Dec 02 20:39:30 so .. you can't run -outside- the webIDE? Dec 02 20:41:01 * veremit thinks .. surely cloud9 IS the webIDE .. Dec 02 20:41:16 or I'm confusing with node.js .. which is likely. Dec 02 20:49:17 <[Butch]> veremit: 1) go to web page of BBB (i.e. 192.168.7.2) and run one of the bonescript examples that are there. That works, including the super simple code I showed above. 2) go to 192.168.7.2:3000 (C9) and try to run one of the existing code examples or create a new file with the super simple code above. This fails with the above error. Dec 02 20:50:55 spartacus: yeah Dec 02 20:53:31 [Butch]: ah .. that would suggest that c9 is broken somehow .. it's as though its not 'locating' the bonescript library or somesuch Dec 02 20:53:48 <[Butch]> Agreed Dec 02 20:54:06 <[Butch]> But that’s just not supposed to happen. :-) It’s supposed to be completely self contained. Dec 02 20:54:18 <[Butch]> Anyway, it doesn’t matter. Sounds like we don’t have any ideas how to fix it. Dec 02 20:54:59 [Butch]: just gotta locate the source of the problem ... Dec 02 20:55:29 <[Butch]> Yeah, I know. I just don’t have a lot of time to fix it righ tnow. Dec 02 20:55:34 <[Butch]> right now Dec 02 20:56:15 hmm and my beagle doesn't have 'stock' on it currently .. now where IS that board .. Dec 02 20:57:39 [Butch]: thing is, if it doesn't locate/load the library I would expect the "require" to fail already Dec 02 21:00:43 zmatt .. that's what's quite weird .. Dec 02 21:01:00 found me bbb .. now all I need is a stock uSD image .. Dec 02 21:01:15 on the other hand Dec 02 21:01:18 javascript sucks. Dec 02 21:01:59 zmatt .. nearly as much as perl... *cough* Dec 02 21:02:09 although both have their uses Dec 02 21:03:33 they're horrid in very different ways though Dec 02 21:08:53 broadly agreed Dec 02 21:15:37 bbl, need to catch train Dec 02 21:25:54 zmatt: pm me please. Dec 02 21:58:24 rcn-ee: hi there Dec 02 21:59:03 I am following your stable-kernel commits, but maybe you can tell me what is new compare to 4.0.2? Dec 02 21:59:40 citylight2, well, all the 4.1.x+ have capemgr... Dec 02 22:00:08 so one can easly change the gpio from the command line - right? Dec 02 22:00:13 anything else? Dec 02 22:01:43 like the cool ground breaking stuff? Mostly borring stuff.. Dec 02 22:02:00 go on, borring stuff please Dec 02 22:02:23 udev? Dec 02 22:02:28 connman? Dec 02 22:02:33 I hear "git log" provides boring stuff Dec 02 22:03:04 well udev/connman are userspace.. Dec 02 22:03:17 opencl is cool, but that's on the x15.. Dec 02 22:03:29 hmm let me see Dec 02 22:05:43 is there a realistic expected release date for the x15? Dec 02 22:05:52 zq: lolol Dec 02 22:06:09 veremit: ? Dec 02 22:06:19 lol = laugh-out-loud :] Dec 02 22:06:49 maybe if we poke jkridner he may be able to shed a little light on that .. but last I heard there were EMI issues Dec 02 22:07:25 I fear the addition of "random" capacitors on the video signals [zmatt] Dec 02 22:08:27 zq, it failed fcc, going thru another board spin... Dec 02 22:10:17 rcn-ee .. that kinda thing bites :/ did u see my packaging comment?! :p Dec 02 22:11:36 what obstacles are there to drawing up a simple breakout board the x15's processor? Dec 02 22:11:59 the am572* Dec 02 22:12:17 arm processor + breakout != "simple" Dec 02 22:12:42 explain, please Dec 02 22:13:58 zq, lets see high speed ddr, pci-e/sata/usb3.0 timing... Dec 02 22:14:27 http://www.ti.com/tool/TMDXEVM5728 would be your second-best bet. In stock and a snip at $600 Dec 02 22:14:56 veremit, that's actually an "x15" + lcd... since it's a development kit, fcc doesn't matter. ;) Dec 02 22:15:09 * veremit imagines LoTR quote you don't simply 'breakout' an arm processor... Dec 02 22:15:32 rcn-ee .. so the beagleboards are an actual module Dec 02 22:15:48 I thought they were all dev-kits of one form .. Dec 02 22:16:21 * veremit looks for FCC approval for the BBB Dec 02 22:16:21 most boards from beagleboard.org get installed into every random "kit" /etc...so the ".org" requires fcc. ;) Dec 02 22:16:28 it's on their.. Dec 02 22:16:32 and the beaglebone white Dec 02 22:16:34 and the xm Dec 02 22:16:34 ahaaaa. Ok, that figures. Dec 02 22:16:36 and the original.. Dec 02 22:16:57 yup I see the logo on the pcb :) Dec 02 22:17:16 lets just say, we were ready on nov 15th... till that friday before when it bompbed in fcc... Dec 02 22:17:42 rcn-ee .. its never 'ready' until its undergone testing. Alas. Been there ... Dec 02 22:18:11 I need to populate P2 on 50 BBBs. Dec 02 22:18:29 ecdhe .. better get started then .. ;) Dec 02 22:18:32 So I could order the flyswatter BBB JTAG kit, but that gets me only one P2 connector. Dec 02 22:18:50 Does anyone have a digikey pn I can order? Dec 02 22:19:16 The CircuitCo BOM lists P2 as DNI so it doesn't give a manufacturer or PN. Dec 02 22:19:41 looks like smd header .. 1.27mm pitch? Dec 02 22:19:45 The BOM lists P2 as " HDR_10X2_CJTAG_0_HEADER10X2_SMD" Dec 02 22:20:26 ecdhe .. I'd imagine it sits on some 'fingers' in the test jigs Dec 02 22:20:47 We may end up with one of those too, but not for 50 units. Dec 02 22:21:12 you need jtag on all 50? Dec 02 22:21:25 They came from the factory without bootloaders, so yes. Dec 02 22:21:30 that sounds a bit .. (for want of a better word...) .. desperate Dec 02 22:21:44 as a beaglebone? Dec 02 22:21:52 =) Dec 02 22:21:59 ah, probably a .. commercial edition? Dec 02 22:22:12 Custom form factor. Dec 02 22:22:16 what else is missing?! :P Dec 02 22:22:36 It's for R&D, so we left everything on. Dec 02 22:22:36 you could serial or net boot .. Dec 02 22:22:51 I can't net boot if there isn't a bootloader, can I? Dec 02 22:23:09 erm .. I need zmatt with the TRM boot pins .. :p Dec 02 22:23:25 don't think the ROm will .. on its own .. it'll do usb iirc Dec 02 22:23:46 ecdhe .. just trying to think around the problem Dec 02 22:24:18 The JTAG connectors would help a lot. I just need a part number. Dec 02 22:24:30 *someone* has ordered one before! Dec 02 22:24:31 yeah .. you might have some luck with bbblfs over usb otg Dec 02 22:24:47 http://www.tincantools.com/wiki/Flyswatter2_BeagleBone_Black_How_To Dec 02 22:24:52 yeah I hear ya .. its just a messy giant hairy connector Dec 02 22:24:58 and for 50 boards .. ew. Dec 02 22:25:39 uSD boot -should- work .. surely ..the spl can be put on there .. Dec 02 22:25:58 uSD with a flasher image did not work. Dec 02 22:26:14 But I wouldn't expect it to if there is no boot rom. Dec 02 22:26:18 any error ? does a standalone work? Dec 02 22:26:36 no boot ROM? that's burnt into the chip at manufacture ... >,< Dec 02 22:26:48 bin it if that's missing Dec 02 22:26:58 its not an external part Dec 02 22:27:11 or a file on disk Dec 02 22:27:43 you can, btw, access the boot mode pins on the p8/p9 headers Dec 02 22:28:47 fyi.. http://processors.wiki.ti.com/index.php/AM335x_Hardware_Design_Guide#Selecting_the_Boot_Mode Dec 02 22:29:39 by all means, you can solder a connector on one .. wouldn't advise it for 50 though, personally :) Dec 02 22:39:50 veremit, are you circuitco? Dec 02 22:40:14 nope .. I think we lost all our reps from there Dec 02 22:40:21 .. who came here. Dec 02 22:40:46 Overall, this has been a somewhat rude room to me. Dec 02 22:41:42 greetings gents Dec 02 22:43:41 veremit: yes ROM supports netboot on its own Dec 02 22:43:46 BBB is a multi-domain beast. Understanding it well enough to know precisely why your code can't blink an LED means understanding signals, PCBs, electronics, ARM architecture, registers, linux kernel, linux userland, linux security, a programming language or two, etc. Dec 02 22:44:21 zmatt .. thanks, thought so. Dec 02 22:44:27 zmatt, what's the simplest way to hello-world the boot rom? Dec 02 22:45:10 it just so happens uploaded this yesterday -> https://github.com/mvduin/bbb-asm-demo Dec 02 22:46:04 it's a really really tiny baremetal app that toggles the leds in response to pressing the S2 button Dec 02 22:47:40 the makefile produces three output files (ELF, bin, and MLO). if you want to netboot, you need to serve the demo.bin via bootp+tftp, see example dnsmasq.conf Dec 02 22:47:57 and you need to pull the sysboot lines right during power-on to make ROM attempt netboot Dec 02 22:48:21 Okay! Dec 02 22:48:30 veremit, Samtec FTR-110-03-G-D-06 Dec 02 22:49:18 ecdhe .. there ya go. I wasn't intending to put you heavily off your intended path, just to verify you'd exhausted all potential other avenues :) Dec 02 22:49:42 Thanks zmatt, will try it out Dec 02 22:50:18 The aim of the channel is to help .. but occasionally becomes a bit abrupt at times, depending on the nature of some of the queries that appear! Dec 02 22:50:24 my preferred way is using a jumper to pull sysboot0 (P8.45) up to 3.3V and pressing the S2 button to pull sysboot 2 down to ground (or use another jumper) Dec 02 22:50:42 I do use a jumper with a resistor in it just to be safe Dec 02 22:51:46 alternatively, make sure the BBB won't boot a linux system that configures the lcd pins as outputs Dec 02 22:53:32 just to avoid accidents involving an pin being configured as output while there's a jumper wire connecting it to ground or vdd ;) Dec 02 22:54:29 Always good to remember zmatt! Dec 02 22:54:50 the S2 button uses a 100 Ω resistor Dec 02 22:55:32 but a much larger value works fine too... the pull resistors for the default boot mode are 100 KΩ so they're really easy to override Dec 02 22:58:05 as tempting as it might be to allow the boot control pins to be "programable" I can see why they wouldn't do that. Might be "tricky" for people inexperienced with SMT soldering to work with. Dec 02 22:58:37 GenTooMan: what do you mean? Dec 02 22:58:55 they are trivially reprogrammable using external pull resistors Dec 02 22:59:35 or, if you're experienced with SMT soldering, you can move the on-board pull resistors Dec 02 23:00:06 just pay for a bed of nails jig Dec 02 23:00:57 ds2 prefered method :D Dec 02 23:00:58 you can also jump to a different boot mode by writing to the control module register containing the sysboot bits and jumping to start of public ROM Dec 02 23:01:10 (this altered bootmode is lost on reset though) Dec 02 23:02:45 also, reexecing public ROM has never worked reliably for me once MMU and caches have been turned on, even if I properly disable and clear them again before jumping back to the public ROM entrypoint Dec 02 23:03:04 no idea why though Dec 02 23:04:00 (though I also haven't put in any real effort to figure it out) Dec 02 23:15:15 zmatt: aren't there scratchpad registers in the CM (or ist hat just the OMAPs) Dec 02 23:16:51 not sure what you're referring to, but pubrom on netra and descendents is considerably simpler than on the omaps Dec 02 23:17:19 there are registers for storing power management state info Dec 02 23:17:23 prehaps something is looking at it Dec 02 23:17:58 no, not really, there is a test for resume from standby but that's not applicable here Dec 02 23:18:24 also, it does work sorta usually and proceeds to download and boot the new image Dec 02 23:18:28 but then often crashes Dec 02 23:19:24 so it just screams "stale cache data" except I most definitely cleared those Dec 02 23:19:56 but dunno, I didn't really care since simply resetting works fine Dec 02 23:22:04 using the reset button is slow due to eth phy reset, but if you perform a software reset then the pulse is short enough the phy doesn't seem to notice Dec 02 23:22:29 lol "doesn't -notice-" Dec 02 23:22:37 and rom has already downloaded and executed by app again before my finger has let go of the 'r' key Dec 02 23:24:45 veremit: the phy needs something like a 20 ms reset pulse Dec 02 23:25:15 Oh fun "race" conditions :D Dec 02 23:25:22 the cpu asserts it for 0.25 μs or so Dec 02 23:25:55 assuming the fatso external cap doesn't gobble it up entirely Dec 02 23:26:25 which reminds me, I should put a scope on the reset pin to see what happens during warm reset Dec 02 23:28:38 stop tempting me to read the user guide zmatt. :D Dec 02 23:31:10 GenTooMan: lol Dec 02 23:31:50 zmatt.. assuming there's no proper 'reset' chip/system to fully reset all that should be... Dec 02 23:32:05 250ns is *** all thuogh Dec 02 23:32:59 veremit: tbh I don't think the cpu can drive the reset line low in that time Dec 02 23:33:09 due to the huge capacitance they put on it on the bbb Dec 02 23:33:31 I'd love to be a fly-on-the-wall of a zmatt/Gerald meeting .. lol Dec 02 23:33:45 https://goo.gl/photos/jBCLk28t2VVrxjWn8 green is power-on-reset, brown is reset Dec 02 23:34:08 ""wtf did you do DEFGHIJKLMNOP.... ??"" Dec 02 23:34:39 (this shows full power-up sequence; blue is 1.8V supplies, cyan is 3.3V supplies, yellow is current measurement) Dec 02 23:35:13 so it's 50ns per division? Dec 02 23:35:21 brown? Dec 02 23:35:34 whatever color you want to call it Dec 02 23:35:47 GenTooMan: ehh, hell no Dec 02 23:35:50 wth is the red trace? Dec 02 23:35:59 veremit: that would be the one Dec 02 23:36:06 reset Dec 02 23:36:20 nice RC time constant there lol Dec 02 23:36:41 GenTooMan: where'd you get that number? it's eh... lemme see... Dec 02 23:36:58 what does it look like when you manually assert it though .. otherwise thats a boring standard capacitor charging curve ... Dec 02 23:38:17 T = 1/(2*pi * R * C) Dec 02 23:38:30 or is that F? (brain fried) Dec 02 23:40:06 think its T Dec 02 23:40:33 5 ms per major division Dec 02 23:41:36 That is a big damn cap... I wonder what TI recommends ... Dec 02 23:41:51 I'm trying to change the mode of the USB OTG to host mode so as to have two USB hosts on my BBB. I've already made the electrical changes on my board, but I've read that I need to change a setting in the device tree to switch the mode of USB0 to host (http://pastebin.com/SeSDdC4A). I'm using RCN's 4.1.8 kernel and was wondering if it is possible to make this change with the capemgr, since it seems to allow you to dynamically compi Dec 02 23:42:25 spiderkeys: I don't think you can get away with using an overlay for that Dec 02 23:42:27 device tree, but I'm not really sure how to go about doing that in the latest version. Does anyone have a good reference for writing that fragment and using the capemgr to patch it through the kernel Dec 02 23:42:56 spiderkeys: you can't make arbitrary changes to existing nodes and expect the kernel to pick up on that Dec 02 23:43:00 Ah, ok. So I'll probably have to modify the am3xx.dtsi and reload the whole thing? Dec 02 23:43:18 heh, you make it sound like it's a big effort Dec 02 23:43:28 I've just never done it before Dec 02 23:43:32 I understand Dec 02 23:45:17 GenTooMan: I'm not even sure an external cap is normally required at all... perhaps just to avoid picking up noise Dec 02 23:46:38 ah sorry, the phy needs "only" a 100 μs reset Dec 02 23:46:55 but it does require the power supplies to be stable for 25 ms Dec 02 23:47:19 stable-ish.... ("at 80%") Dec 02 23:47:37 and that is actually a close call Dec 02 23:48:42 and with the reset line having an curve like this, the actual timing of the reset release of the phy and cpu probably varies significantly Dec 02 23:49:04 (depending on die variation and temperature) Dec 02 23:49:49 Although looking at the version of am33xx.dtsi for my kernel version, I don't even see anything about the "port0-mode" setting described in the diff https://github.com/beagleboard/linux/blob/4.1.8-ti-r18/arch/arm/boot/dts/am33xx.dtsi Dec 02 23:50:08 zmatt: do you know if that setting name changed? Dec 02 23:50:10 spiderkeys: it's probably in one of the bbb files Dec 02 23:50:21 since the usage of the ports is board-dependent Dec 02 23:51:14 am335x-bone-common.dtsi Dec 02 23:51:42 and it's not called port0-mode indeed Dec 02 23:51:48 but search for usb Dec 02 23:51:59 Ah yea, peripheral vs host Dec 02 23:53:24 GenTooMan: the phy reset should never have been tied to cpu reset to begin with Dec 02 23:56:15 zmatt, GenTooMan: indeed .. should have a reset circuit of some form. Dec 02 23:56:36 just tie it to gpio and have the cpu perform the reset Dec 02 23:56:41 when necessary Dec 02 23:56:59 (i.e. if the reset bits indicate a cold reset) Dec 02 23:58:10 if you use a separate reset circuit, you'd still need to have the cpu monitor it to avoid driving the mii interface pins before the phy is out of reset Dec 02 23:58:21 otherwise its strapping options would get messed up Dec 02 23:58:27 ah bloody hell lol Dec 02 23:59:16 considering the curve on the reset line, that's probably what's causing the mysterious ethernet mis-strapping for some people Dec 02 23:59:18 zmatt you won't end up with a unknown on that gpio as it undergoes reset? Dec 02 23:59:55 there's always some damn compromise lol Dec 02 23:59:58 the reset state of all gpios is documented (high-Z, usually with weak internal pull-up or pull-down) Dec 03 00:00:09 *all ios Dec 03 00:00:23 (there's nothing specifically "gp" about them during reset :P ) Dec 03 00:00:33 well .. high-Z isn't exactly high-or low for a reset signal, is it?! :P Dec 03 00:01:19 if it's normally high and has internal pull-up by default during reset, then nothing will happen during reset as desired Dec 03 00:01:36 ah yes internal pull-up Dec 03 00:01:48 you can also reenforce (or override) the internal pulls with external ones Dec 03 00:01:52 but we want to assert it at reset? Dec 03 00:01:56 no Dec 03 00:02:04 I don't at least Dec 03 00:02:13 we're not resetting the eth phy .. Dec 03 00:02:19 hold on. lol. Dec 03 00:02:32 software resets the phy, once, after power-on-reset Dec 03 00:02:39 there's absolutely no reason to reset the phy on every reboot Dec 03 00:02:52 no, I Agree Dec 03 00:02:57 in fact the soft resets being done right now by bootloader and kernel are a significant contribution to startup time Dec 03 00:03:16 and, as you suggest, one potential reason for partial/complete failure Dec 03 00:03:40 well that happens during power-up, not during softreset Dec 03 00:03:52 hmm true .. but could one cause .. Dec 03 00:03:59 a softreset doesn't resample strapping options Dec 03 00:04:02 .. probably .. nods Dec 03 00:04:08 correct. Dec 03 00:05:30 Hello Dec 03 00:06:27 a reboot however *should* have pulsed the reset line low.... but the phy doesn't seem to reset, so either the pulse is too short (sounds like a recipe for undefined behaviour to me) or the capacitance is just too large Dec 03 00:07:25 maybe if I find a moment later this week I'll hook up the scope to check it Dec 03 00:10:20 fighting ACPI makes one very tired :( Dec 03 00:11:00 "ACPI is a complete design disaster in every way. But we're kind of stuck with it. If any Intel people are listening to this and you had anything to do with ACPI, shoot yourself now, before you reproduce." -- Linus Torvalds Dec 03 00:11:45 zmatt .. I fear something worse would have taken its place. Dec 03 00:12:25 or shall we say 'different' .. in our comparison of 'things that are aweful' Dec 03 00:12:37 EFI ? Dec 03 00:12:41 ;) Dec 03 00:12:48 like .. efi ;) Dec 03 00:13:10 yeah I remember acpi bein fun when you put win98 on a pc .. lol Dec 03 00:13:30 you had to enable it instead of plain "plug and pray" Dec 03 00:13:55 which windows has, and never will ever be, capable of. Dec 03 00:14:11 somehow they even f*ed up usb .. Dec 03 00:19:45 UEFI is a freaking pain. Dec 03 00:22:06 veremit MS messed up USB from the start. the Communication interface specified was not supported by windows at first. Then they used a proprietary one (IE one only supported by windows) for network USB devices. So most "windows" USB net devices are actually non standard and won't work with any other platform. Dec 03 00:22:10 * veremit *nod* Dec 03 00:22:44 its like bloody apple .. why re-invent the wheel and break it and make it square?! Dec 03 00:24:02 I bought a Apple NET USB device and it worked fine on my beagle board rev B board and everything else I plugged the thing into. It just worked. It even worked on a windows box no problems. But a lot of "PC" ones don't work on everything. Dec 03 00:25:13 Well sometimes people follow between the lines on the road but others wander all over. (how I look at it) usually messy too. Dec 03 00:26:06 USB is awful anyway Dec 03 00:26:20 zmatt it was MS's idea. Dec 03 00:26:25 no Dec 03 00:26:45 zmatt then who dreamed it up? Dec 03 00:26:54 http://firewireexpert.blogspot.nl/2009/05/dollar-deal-that-almost-killed-firewire.html Dec 03 00:27:07 wait that's not the right link Dec 03 00:27:16 And we should all be using betamax, not vhs .. lol Dec 03 00:27:16 wait it is Dec 03 00:27:32 phillips had a hand in usb iirc Dec 03 00:27:44 as well as intel et al Dec 03 00:29:25 usb was meant for mice and keyboard Dec 03 00:29:41 not exclusively Dec 03 00:30:32 no, but dumb devices in general... they just cough up static descriptors and leave the host to micro-manage everything Dec 03 00:32:04 USB works great for that ... Dec 03 00:32:17 anyhow, enough about hardware that sucks Dec 03 00:34:19 it is really distrubing to have to code around a closed source BIOS that does not accurately describe your hardware on your embedded device. Dec 03 00:35:03 ds2 .. what kinda bios you using for wht device? Dec 03 00:35:06 with OF, at least you can inject a potentially accurate picture Dec 03 00:35:18 ds2: have you read http://blog.invisiblethings.org/2015/10/27/x86_harmful.html ? Dec 03 00:35:21 veremit: whatever is shipped with the fishy board Dec 03 00:35:32 oo *googles* Dec 03 00:35:54 ds2: SMM is apparently also horror for people trying to do real-time stuff on x86 Dec 03 00:36:18 zmatt:SMM is a cluster-f to begin with Dec 03 00:36:34 it is like asking people to do realtime in a VM Dec 03 00:36:51 ok .. google fail .. what's the fishy board? Dec 03 00:36:53 voodoo system administration... "don't boot with an usb stick inserted, it will mess up your irq latency. inserting an usb stick after boot is no problem" stuff like that Dec 03 00:36:56 trustzones might be as nasty though Dec 03 00:37:11 veremit: a board I shall not specifically call out as a couresty to that board Dec 03 00:37:22 hahaha ok Dec 03 00:37:38 pm or /notice me then .. lol Dec 03 00:37:40 jk Dec 03 00:38:32 back to complaining about MUSB and SGX Dec 03 00:38:40 trustzone you mean... and yes it has similar problems, although it can also be very useful... it definitely sucks TI rendered it unusuable on their GP SoCs instead of leaving it open for use Dec 03 00:38:49 oh yes what a pile of shite that lot is .. *grin* Dec 03 00:39:16 zmatt .. you're just not Trying hard enough lol .. yet .. :) Dec 03 00:39:20 well I have no idea whether SGX itself is good or not, it seems pretty decent based on very limited info Dec 03 00:39:27 their code otoh /o\ Dec 03 00:41:16 veremit: I still know a potential route to gain execution control in secure world on GP dm81xx/am335x devices, but I haven't had the time or motivation to persue it Dec 03 00:42:36 even if it works at all, it probably can't be made to work without JTAG, so that makes it useless for the purpose of putting trustzone to useful work Dec 03 00:43:15 zmatt: loop back PRU pins to the JTAG port and.... Dec 03 00:43:32 ds2: no, you'd need an external mcu or so Dec 03 00:43:35 or via the M3 toggling the jtag line Dec 03 00:43:54 zmatt: why? (Boot in GP, start/configure PRU, force reset...) Dec 03 00:44:27 you need to take control of the a8 after cold reset and before public rom entry Dec 03 00:44:46 after that the SSM is locked down and it's game ovefr Dec 03 00:44:57 no way to generate a cold reset using the PRCM? Dec 03 00:45:05 yes, but it resets all cores Dec 03 00:45:14 oh DOH Dec 03 00:45:34 fine. toss in one of those 16pin sized M0's then Dec 03 00:46:08 there's still a decent chance the ever-grumpy SSM will intervene anyhow Dec 03 00:47:15 the attack vector is the fact that you have a _lot_ of control over the SoC via jtag in wait-in-reset mode, before the a8 has executed any code Dec 03 00:48:02 which means almost certainly you can cause a secrom access to fault Dec 03 00:48:40 the secrom fault handler seems to assume the secure world kernel has been initialized already and accesses data structures in a8-local SRAM Dec 03 00:48:44 including an indirect branch Dec 03 00:48:52 oooooooo Dec 03 00:49:47 now the a8-local SRAM in inaccessible at this point, but it's not explicitly reset hence will almost certainly hold its state from a previous boot if you do it right Dec 03 00:50:12 see where I'm going with this :) Dec 03 00:50:38 unless there is HW to force a clearing of SRAM Dec 03 00:50:43 there isn't Dec 03 00:50:47 that's easy enough to check Dec 03 00:51:19 A common security flaw in OS is not wiping memory when an application closes. Dec 03 00:51:26 in fact secrom overwrites only a few SRAM locations on GP devices, most of it is left intact if you reset Dec 03 00:52:18 ds2: and the nice thing is that this would happen before secom has had a chance to configure the SSM Dec 03 00:54:13 (see comments here https://github.com/dutchanddutch/jbang/blob/master/include/ti/subarctic/ctrl.h#L207 ) Dec 03 00:55:32 it isn'tnice t burn down firewalls :D Dec 03 00:55:47 once firewall_control is 0x41 it's all over Dec 03 00:58:46 this I noticed from the fact the secmon exception handlers on HS devices write some SSM bit after switching secure/public context and before exiting secmon, presumably some acknowledge that it has performed the transition and state should be locked down again Dec 03 00:58:56 the GP secmon has no such thing Dec 03 00:59:15 i.e. the SSM is most likely in a permanent lockdown state Dec 03 01:00:07 which makes perfect sense... the GP secmon contains negligible code Dec 03 01:00:53 or, well, negligible functionality, it's still quite a bit of code due to being horribly inefficient :P Dec 03 01:03:09 actually no it's also negligible code Dec 03 01:26:48 http://pastebin.com/k6CGLCAd <-- in case anyone is interested :P Dec 03 01:28:43 (pastebin's highlighting for arm asm is really crap, I just made random guesses at its rules until it gave proper highlighting, but ARM asm normally doesn't use ; as comment char) Dec 03 01:29:19 or rather, gnu asm doesn't for ARM, not sure what the official one does Dec 03 01:30:29 ah, it does, well that sort of excuses pastebin a bit I guess :P Dec 03 02:13:51 I would be depressed by reading the paper if I were younger, but those issues have long existed in systems. Complexity often leads to unsecure (insecure?) systems, More interesting however it does make me consider making computers again for fun (LOL). FPGA's four fun! LOL well one thing at a time. Dec 03 02:16:58 :) Dec 03 02:59:01 I am surprised however that hardware doesn't automatically clean memory these days on start up. "random" reset states have always been troublesome for debugging 'stupid' code (LOL). Dec 03 03:00:18 probably rather costly in terms of realestate Dec 03 03:00:48 and yet another potential attack vector as well. **** ENDING LOGGING AT Thu Dec 03 03:01:11 2015