**** BEGIN LOGGING AT Thu Jul 31 03:00:01 2014 Jul 31 12:26:25 Any one here can help? Jul 31 12:26:34 depends Jul 31 12:27:04 OK. My BeagleBone Black won't boot fron microSD card. Jul 31 12:27:24 There are no USER leds. Jul 31 12:27:31 only solid power led Jul 31 12:27:46 connect a serial console to see whats going on Jul 31 12:28:01 to the PC? Jul 31 12:28:14 jonathan_tmp: and what is on the SD card? Jul 31 12:28:53 Debian (BeagleBone, BeagleBone Black - 2GB SD) 2014-05-14 Jul 31 12:29:08 and you wrote it to the card how? Jul 31 12:29:15 flashed using dd Jul 31 12:29:35 have the exact command line you used at hand? Jul 31 12:29:49 wait Jul 31 12:30:51 Hi Jul 31 12:31:15 Is it possible to have PWM output from PRU while reading I2C? modes seems to be conflicting and disabling my I2C Jul 31 12:31:33 "sudo dd bs=1m if=~/BBB/bone-debian-7.5-2014-05-14-2gb.img of=/dev/sdc" Jul 31 12:32:46 tbr: so? Jul 31 12:34:17 jonathan_tmp: that would look correct. which brings us back to attaching to the console Jul 31 12:34:47 jonathan_tmp, why are you suggesting outdated Debian? ;) Jul 31 12:34:57 ? Jul 31 12:35:05 7.5 is stable.. its old like hell Jul 31 12:35:16 well nvm, I dont even know why it is called stable Jul 31 12:35:23 mkad: but it won't boot! Jul 31 12:35:25 should be called "old" Jul 31 12:35:41 oh I see Jul 31 12:35:44 mkad: that's besides the point here... Jul 31 12:36:02 USB serial will work? Jul 31 12:36:03 tbr, ok just my thoughts Jul 31 12:36:21 jonathan_tmp: you need a "ftdi" cable Jul 31 12:36:23 I'd take stability over polish any day. Give me the old stuff. I still have plenty of RH9 boxes ;) Jul 31 12:36:29 ouch... Jul 31 12:36:38 those are cheap Jul 31 12:36:42 Anyone experienced with setting modes for gpio on BBB ? Jul 31 12:37:38 How do I figure out which mode does not disable i2c, are modes for each gpio or just for set of gpio ? Jul 31 12:39:12 Decessus always depends on the job/tasks it has todo Jul 31 12:40:58 woglinde, also I miss dts files on my Debian ... Jul 31 12:42:26 mkad and I have them or what? Jul 31 12:44:10 woglinde, if 7.6 could do task foo, and this wasn't available in 7.5, but 7.5 is twice as stable as 7.6. I will install 7.5 and write the task myself. Yes it;s more work now, but less upkeep in the long term. With the exception of hardware support, there is no reason one would *have* to use a newer, less stable version for any task. Jul 31 12:44:10 woglinde, ok forget it Jul 31 12:44:53 Decessus, newer is not less stable, older is, because its obsolete Jul 31 12:45:13 So code breaks itself? Jul 31 12:45:28 Decessus, and cost maintaing obsolete code is high Jul 31 12:45:35 Decessus hm I wonder how you will teach 2.6 kernel newer features Jul 31 12:47:33 Decessus, debian stable has backports because some features are missing, again these backports are incomplete and are often patches taken/modified not always posible to merge with older version of library software, so cost of maintaining backports is higher than just upgrading to newer distro and following development, if you aim at reusable code for a longer period. Jul 31 12:48:17 Decessus, sorry for convoluted message but you get my point I hope:) Jul 31 12:51:56 Yes I do, but the only point I'm making is true 'stability'. Not with the additional variables of backports, etc. If your system doesn't crash with 7.5, and 7.6 has 3 system lock ups a day...how is it more reasonable to still use 7.6? I'm not talking about software distribution..just one controlled environment. Jul 31 13:18:43 I am claculating how much water I need to water the plants Jul 31 13:19:24 I then wait for the water sensor to tell me the amount of water was released, and I can close the tap Jul 31 13:20:07 I thought of using epoll and update a memory counter to tell how much water was supplied, Jul 31 13:20:30 this way the cpu wont poll the GPIO and will be kept idle Jul 31 13:21:05 didnt we discuss that before? Jul 31 13:21:28 av500 dejavu? Jul 31 13:21:36 I can then craft the function to count to a given number indicating the amount of water I seek Jul 31 13:22:06 this function will then send a release to the main program socket Jul 31 13:22:16 lmao Jul 31 13:22:23 av500: yes Jul 31 13:22:44 yet I prefer to use better coding, and the BBB Jul 31 13:23:52 so? Jul 31 13:24:16 in general having an epoll loop will keep the cpu coller Jul 31 13:24:20 cooler Jul 31 13:25:35 even in the case of a push button, the epoll approach makes more sense Jul 31 13:25:38 Does anyone know how to figure out from this https://github.com/mashu/bbb-prupwm which channel corresponds to which PRU output gpio ? Jul 31 13:25:46 I am lost Jul 31 13:25:55 if linux offer a debounce , I prefer to use that Jul 31 13:27:59 ok nvm I think I will brueforce it with the scope Jul 31 13:28:29 stupid way but I cant find any information on that Jul 31 14:54:51 Anyone here have experience using the rs485 with bbb? I'm unable to toggle rts high automagically, any advice is much appreciated :-) Jul 31 16:07:40 can someone tell me why is my BBB in just flashed debian wheezy 7.5 shutting down by itself after some minutes ? Jul 31 16:07:48 has anyone experienced any similar behavior ? Jul 31 16:08:01 I haven't touched it since the ugrade Jul 31 16:08:03 every time i plug in my BBBrC, a new usbmodem device appears in my network list. I now have like 30 of them. Any way to tell OSX to recognize the usb identifiers and always assign it to the same network entry? Jul 31 16:08:05 upgrade* Jul 31 16:42:14 freezevee: I've heard that before, but only once. Jul 31 16:42:40 tbr: I removed the SD and it seems it stays stable Jul 31 16:42:42 wtf ? Jul 31 16:42:56 jedahan: interesting, I wonder why that happens. it's probably the CDC_ACM device Jul 31 18:20:17 av500, how do I know which channel of pru drives which gpio ? Jul 31 18:22:03 ok nvm found it Jul 31 18:23:05 <_av500_> always glad to help Jul 31 18:24:06 mkad: do you know how you communicate between pru and kernel/userland? Jul 31 18:31:44 cuba, yes Jul 31 18:38:56 cuba, but cant get thing working anyway.. Jul 31 18:47:52 how would you do it mkad ? Jul 31 18:48:56 cuba, https://github.com/mashu/bbb-prupwm Jul 31 18:49:04 cuba, using this code Jul 31 18:51:20 so shared memory Jul 31 18:51:24 cuba, I can get only on P9_25 desired pwm signal Jul 31 18:51:26 but how would you know if using the pru Jul 31 18:51:29 if the data is ready Jul 31 18:51:49 *when Jul 31 18:52:04 I dont know what you talk about pru is reading regs only Jul 31 18:52:10 so no need to know that Jul 31 18:52:20 yes i am not talking about the pwm thing Jul 31 18:52:31 but if you do something else with the pru Jul 31 18:53:19 I want to reuse ready code I dont want to learn how to code for pru Jul 31 18:55:35 cuba, I see this code works for one pin but not for others Jul 31 18:56:00 I give up for today Jul 31 18:57:32 cuba: pronounced "cooba" Jul 31 18:57:37 a la cuba libre Jul 31 18:58:12 or cuba, a la cuba the counry Jul 31 18:58:15 hrm. [ 2845.581891] omap_hsmmc 481d8000.mmc: unable to obtain RX DMA engine channel 3730343760 Jul 31 20:04:59 how can I free up disk space from debian wheezy ? Jul 31 20:05:30 apt-get --purge autoremove $packages_you_do_not_want Jul 31 20:05:54 any ideas ? Jul 31 20:06:29 I am stucj in E: Write error - write (28: No space left on device) Jul 31 20:06:33 stuck* Jul 31 20:06:40 and I don't know what to rm-rf Jul 31 20:06:51 clean and autoclean before removing packages maybe ? Jul 31 20:07:08 I rm'ed all that cloud9 and nodejs crap as well Jul 31 20:07:28 das_: clean did a bit Jul 31 20:07:35 100mb Jul 31 20:08:35 clean clears your cache; autoclean removes deprecated stuff; autoremove deletes unnecessary dependencies Jul 31 20:09:04 IIRC there's a bunch of useless stuff in /opt but don't quote me on that Jul 31 20:09:38 also you can dpkg --get-selections to examine all installed packages Jul 31 20:09:42 where are all these shit located ? Jul 31 20:10:04 i don't remember sorry, had to browse a bit Jul 31 20:10:04 all pages shown in 192.168.7.2 Jul 31 20:10:31 maybe web stuff is under /var/www ? or /opt/www I dunno Jul 31 20:10:36 they need to upgrade the chipset to 4gbytes ! Jul 31 20:10:50 var/www has nothing Jul 31 20:10:54 you could do some 'find' trickery to find the bigger file also Jul 31 20:11:03 files* Jul 31 20:12:18 find -size xxx Jul 31 20:12:49 das_: thanks Jul 31 20:13:00 does anyone use video streaming on their BBB ? Jul 31 20:13:07 I am looking for a way to get a nice fps Jul 31 20:13:16 freezevee: read 'man find' because you'll also need to use -exec Jul 31 20:17:30 das_: is it better to install and run from the SD card ? Jul 31 20:17:38 I got a 8Gb one Jul 31 20:18:17 I don't own a BBB; I run a first generation white one that doesn't have onboard NAND Jul 31 20:18:30 so no choice but to run from SD for ,e Jul 31 20:18:37 which I prefer anyways Jul 31 20:18:47 grr, updated deb kernel, now i dont have uio_pruss ... Jul 31 20:18:50 I don't know what to do so I am just searching Jul 31 20:19:27 I'd wager you need to take a look at installed packets and remove size offending ones Jul 31 20:30:18 freezevee: if you're worried about space, yes, SD allows you plenty of space Jul 31 20:30:21 freezevee: it is slower... Jul 31 20:30:38 vagrantc: that's what I am afraid of Jul 31 20:30:56 but use apt-get remove/autoremove to remove packages, not "rm -rf" Jul 31 20:37:43 how do i build the latest kernal from source? uname -a shows beaglebone 3.15.5-bone5, but the version from https://github.com/beagleboard/kernel.git shows 3.14 as the latest.... Jul 31 20:38:24 vagrantc: yeah that's what I am doing Jul 31 20:38:28 but is it worth it ? Jul 31 20:38:53 I mean , If you're fighting to gain 100Mb does not make sense Jul 31 20:39:19 or just go to SD boot Jul 31 20:39:30 and do whatever you want Jul 31 20:40:15 freezevee: depends on what you want. Jul 31 21:28:59 are there any guides to compiling a new kernel for the BBB? I need a kernel > 3.12 that supports uio_pruss :( Jul 31 21:45:18 is there any way to let the BBB start always by SD card ? Jul 31 21:45:24 not pressing the boot button Jul 31 21:46:07 if there's a card plugged in .. and its valid .. I think you'll find its checked Jul 31 21:46:14 the button is more an override Jul 31 21:46:37 jedahan: use robercnelson kernel? Jul 31 21:46:52 robertcnelson* Jul 31 21:47:54 jedahan: this should help: http://www.eewiki.net/display/linuxonarm/BeagleBone+Black Jul 31 21:54:15 hi Jul 31 21:54:30 can anyone here tell me about PMIC behaviour? Jul 31 21:56:04 that's a rather open-ended question Jul 31 21:56:36 well i have a big problem atm Jul 31 21:56:44 shutdown current is 40mA at 5v Jul 31 21:56:57 im trying to make something that will shutdown on battery power Jul 31 21:57:03 so i need to work out what is wrong Jul 31 21:59:02 is there a repository for the debian build? Jul 31 21:59:18 i want to check the commits for bug fixes, atm im running angstrom Jul 31 21:59:27 your best bet is gonna be making a FET switch that cuts off power even to the board Jul 31 21:59:53 thats seriously lame Jul 31 22:00:03 there is a full datasheet for the PMIC ffs Jul 31 22:00:08 40mA is very good actually Jul 31 22:00:12 wtf Jul 31 22:00:19 should be 40µA Jul 31 22:00:25 its not a PIC Jul 31 22:00:31 its an ARM processor Jul 31 22:00:32 40 mA when fully shut down? Jul 31 22:00:35 yes Jul 31 22:00:52 it is 40µA when i first connect power Jul 31 22:00:56 which is correct Jul 31 22:01:07 datasheets says OFF current is typ 6 μA Jul 31 22:01:11 after running then shutdown its stuck at 40mA Jul 31 22:01:14 exactly Jul 31 22:01:21 the state has changed Jul 31 22:01:23 hmm .. possibly some quirk then Jul 31 22:01:24 there is a bug Jul 31 22:01:41 s/bug/piss poor PMIC driver Jul 31 22:01:46 my guess is the PMIC is never fully shut down Jul 31 22:01:49 dump the PMIC's registers? Jul 31 22:05:14 yo mrpackethead .. you there, or is that just a computer pinging in&out Jul 31 22:06:21 zmatt: that was my plan Jul 31 22:06:26 but i dont know how Jul 31 22:06:31 there are no test pins Jul 31 22:06:35 hell, 40 mA @ 3.6 V is even a lot if the CPU were still on and merely in standby Jul 31 22:06:49 im pretty shut the cpu is down Jul 31 22:06:54 and rtc is def down Jul 31 22:07:43 all the rails seem to be down Jul 31 22:07:54 i suspect PMIC is powered up and in the wrong state Jul 31 22:07:54 yea some converter is still tickin over Jul 31 22:07:58 with rails off somehow Jul 31 22:08:26 probably some magic needed to shut the power rails off and then the pmic in the right order to avoid damage Jul 31 22:08:41 but if the pmic is any good .. it should handle some kinda timeout Jul 31 22:09:18 i guess tomorrow i could solder enamled copper wires on Jul 31 22:09:24 The /lib/firmware directory use to have the dtbos for various overlays, like UART, PWM, etc… in 3.8.13-bone59 (jessie) they aren't there. Are they somewhere else? Jul 31 22:09:24 to probe the registers on the PMIC Jul 31 22:09:32 ill try that tomorrow Jul 31 22:09:47 then i can fix the kernel and recompile Jul 31 22:09:49 you could do that via jtag if you can get power to the wakeup domain Jul 31 22:09:49 you might be lucky and find the i2c bus has the pmic on it Laurenceb__ Jul 31 22:09:57 or .. that .. :D Jul 31 22:10:00 access the i2c controller via jtag Jul 31 22:10:19 oh good plan Jul 31 22:10:35 but as you say, id need power to the wakeup domain Jul 31 22:10:50 veremit: already tried, it doesnt help me Jul 31 22:11:41 ill read through debian commits first, this issue may have been fixed Jul 31 22:11:50 hmm don't bank on it Jul 31 22:11:58 well I know on OMAPs jtag can get access even in off-mode... but I'm not sure if the same thing works on the am335x Jul 31 22:12:21 ping/pong :p Jul 31 22:12:53 lemme see if I connect under WiR whether it powers anything up besides the wakeup domain Jul 31 22:13:08 ok thanks Jul 31 22:13:15 i need to cook some food, bbl Jul 31 22:31:41 hrm. got through a full debian-installer run to eMMC using the debian-installer daily images ... but somehow i messed up the partitioning for the MLO/u-boot.img ... Jul 31 22:33:10 just get C repeating indefinitely... attempting to load from serial? Jul 31 22:34:24 yup, that's a fallback boot option Jul 31 22:34:31 i have a element14 issued BBB that has revision C, and I've connected an LCD3 cape that doesn't light up when sitting on the BBB. THere is a green power light that is lit, but the screen isn't on. Thoughts? Aug 01 00:15:43 ok wtf, is the datasheet still doesn't list the pin numbers (for indexing into pinmux array) for each ball? Aug 01 00:15:52 s/is // Aug 01 00:17:10 ah, found a header which does Aug 01 00:22:26 and lo and behold, my pmic_read() function suddenly works Aug 01 00:35:06 zmatt: interesting Aug 01 00:35:17 ive been using i2cget -f Aug 01 00:35:24 to read i2c0 Aug 01 00:36:37 this is using jtag while the target is held in reset, so I get to examine the pmic before its registers get meddled with again Aug 01 00:37:43 I have values from right after power cut before first run, I've just booted linux and shutdown again so let's see if the regs are different Aug 01 00:38:51 argh, stupid debug server from TI Aug 01 00:39:20 ok, fresh try Aug 01 00:43:52 the only difference I see is that in the first case the "USB power status change interrupt" bit is set and in the second case the "Push-button status change interrupt" Aug 01 00:44:19 but I think I have some ancient distro on it, maybe it doesn't have the pmic driver yet or something Aug 01 00:47:09 very interesting, thanks Aug 01 00:47:22 im running angstrom 3.8 at the moment Aug 01 00:47:40 i should have tested with the stock distro, but it was pretty broken Aug 01 00:47:41 I can give you the script to dump the regs, if you have CCS install and JTAG available Aug 01 00:47:53 *installed Aug 01 00:47:53 i couldnt boot with battery connected at all Aug 01 00:47:58 it was in a reboot loop Aug 01 00:48:06 with stock distro Aug 01 00:48:12 angstrom 3.2 or something Aug 01 00:48:24 what is CCS? Aug 01 00:48:54 can you monitor current draw of your board via the battery header? Aug 01 00:49:12 Code Composer Studio, TI's IDE... hideous thing based on eclipse, but my script just uses the debug server scripting Aug 01 00:49:15 ive got a JST connector Aug 01 00:49:16 ah Aug 01 00:49:26 ok, pastebin? Aug 01 00:51:12 im using a lipo cell with multimeter Aug 01 00:51:53 i just thought - I2c0 is connected to the id eeprom, so i can attach a logic analyser there and look at bootup and shutdown Aug 01 00:52:13 i will try that tomorrow - hardware is at work, its 2am here in uk :P Aug 01 00:54:11 http://gerbil.xs4all.nl/bbb-pmic.zip Aug 01 00:54:33 thanks Aug 01 00:54:45 you are in nl ? Aug 01 00:55:06 get some sleep lol Aug 01 00:55:09 * Laurenceb__ zzz Aug 01 00:55:19 thanks for the help Aug 01 00:55:22 the dss script is a wrapper around debug server scripting, it needs rlwrap (available as debian package, probable elsewhere too) Aug 01 00:55:40 target.ccxml may need adjustment if you have a different jtag debugger than an xds100v2 Aug 01 00:56:16 starting dss in that dir should preload your commandline with: load("debug.js"); press enter. :P Aug 01 00:57:03 hey its only 2am in England, UK Aug 01 00:57:09 if you're connected succesfully with your target then exit the debugger and power off the target then on powering it on again it will be held in reset Aug 01 00:57:59 either way, you get all pmic regs dumped with: Aug 01 00:58:03 pmic_read( 0, 32 ) Aug 01 00:58:28 though if linux is running and thinks it owns i2c0, there may be trouble, do doing it from wait-in-reset is safer :P Aug 01 00:58:35 *so Aug 01 00:59:31 (and no, this i2c "driver" is not going to win any beauty contest... but hey, it works ) Aug 01 01:03:55 sleep? at least at this hour it's a bit cooler so I can finally get anything done... during the day my brain goes into thermal shutdown :P Aug 01 01:04:33 yea been too warm lately Aug 01 01:04:55 * zmatt has been empirically trying to make sense of the behaviour of TI's I2C peripheral Aug 01 01:05:00 conclusion: it's buggy as shit Aug 01 01:05:33 my I2C sniffer/debugger is neat though, runs on a separate MCU and by stretching the I2C clock allows you to single-step through I2C transactions Aug 01 01:05:34 sounds like TI's drivers ... Aug 01 01:05:43 not the driver, the hardware Aug 01 01:05:59 ah thats debateable too wher eTI is concerned lol Aug 01 01:06:08 well Aug 01 01:07:03 it reacts to a NACKed slave address by.... locking up in a state requiring the peripheral to be reset (afaict), also without generating stop-condition of course so the bus will be considered occupied by other masters Aug 01 01:07:32 it also has a whole bunch of race conditions that are tricky to deal with Aug 01 01:07:48 in multimaster situations Aug 01 01:08:54 like, if it loses arbitration then data will still be left in its TX fifo, so if it lost arbitration due to being addressed as slave with a read transaction, it'll get whatever is left in the fifo completely for free! :D Aug 01 01:13:32 fucking up I2C periperhals seems to be a popular hobby for MCU/SoC manufacturers though... the one on STM32 is also full of bugs and race conditions, and that thing on the raspberry pi will cause random data corruption is any chip on the bus has the impudence to use clock scratching Aug 01 01:13:38 *streching Aug 01 01:13:59 **** ENDING LOGGING AT Fri Aug 01 03:00:00 2014