**** BEGIN LOGGING AT Sun Jan 17 02:59:58 2016 Jan 17 03:01:28 what is the difference between the debian 8 (medium) root file system and the small flash? obviously7 one is smaller in size, but what is left out of the smaller one? Jan 17 03:01:43 is there a list somewhere? Jan 17 03:02:08 you mean lxqt 2g vs 4g ? Jan 17 03:02:28 i mean here: https://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-Debian8 Jan 17 03:02:51 versus here: https://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-Debian8%28smallflash%29 Jan 17 03:03:12 i guess i could just take a look at the files... Jan 17 03:04:26 debian-8.2-minimal is already quite bare Jan 17 03:04:52 but yeah I'd say just diff them Jan 17 03:07:14 right Jan 17 04:27:45 Does anyone have experience with the Waveshare 7" LCD B? I'm hoping to find drivers for it other than the full image, since I am already running another custom image Jan 17 04:57:21 maxbots|mtw: I'm guessing revision older than 2.1 ? Jan 17 05:18:57 since rev 2.1 should apparently be usable with standard drivers, for older revisions they only have custom binary drivers... they did publish the protocol, which is quite simple although it leaves some unanswered questions (like what happens if you make a second touch and then release the first one) Jan 17 05:19:08 let me look Jan 17 05:19:36 Yeah, rev 1.1 Jan 17 05:19:45 http://www.waveshare.com/w/upload/7/7c/7inch-Capacitive-touch-screen-HID-protocol.pdf Jan 17 05:19:59 someone made this -> https://github.com/derekhe/waveshare-7inch-touchscreen-driver Jan 17 05:21:24 That driver specifies Raspberry Pi, I assume the install directions work as is with the Beaglebone? Jan 17 05:21:42 unlikely, since a binary driver is tied to the specific kernel build Jan 17 05:21:48 Damn Jan 17 05:21:53 oh you mean the github thing? Jan 17 05:21:55 that might work Jan 17 05:21:58 Yeah Jan 17 05:22:35 hmm Jan 17 05:22:57 that install script ain't gonna work Jan 17 05:23:15 it assumes raspbian Jan 17 05:23:39 kk Jan 17 05:24:25 the python script can probably be made to work... it uses hidraw and uinput to convert the raw data from the touchscreen into a simulated mouse Jan 17 05:24:47 ew, this is gross Jan 17 05:24:52 os.system("modprobe uinput") Jan 17 05:24:53 os.system("chmod 666 /dev/hidraw*") Jan 17 05:24:53 os.system("chmod 666 /dev/uinput*") Jan 17 05:25:04 If I am willing to forego the touchscreen, I should be able to make this work without drivers, right? Jan 17 05:25:12 yeah then it's just a hdmi display Jan 17 05:25:38 Right now it is almost working https://www.dropbox.com/s/2pv1rfeu0ofu5cb/2016-01-16%2021.21.52.jpg?dl=0&preview=2016-01-16+21.21.52.jpg Jan 17 05:25:54 Just need to figure out why it is rotated 90 degrees and oddly sized Jan 17 05:27:20 well, my guess would be that the display is actually 480x800 :P Jan 17 05:27:36 maybe you can rotate it using xrandr (if you're using X11) Jan 17 05:28:16 or check the edid data Jan 17 05:28:26 I have no clue how this thing outputs. I think it is actually a native app. Jan 17 05:28:56 Pretty sure this is what we are looking at Jan 17 05:28:57 http://wiki.thing-printer.com/index.php?title=Toggle Jan 17 05:29:12 Unfortuantely the docs for this image are not exactly great Jan 17 05:29:25 but it is needed to use the hardware I am trying to use Jan 17 05:29:44 And no one is awake in their IRC channel :-) Jan 17 05:29:58 * zmatt shrugs Jan 17 05:30:32 I actually know nothing about your image nor your lcd screen, I just googled things, something you can do too :P Jan 17 05:30:39 I haven't used Linux in 10 years and was far from a guru even then, so it definitely has me a bit confused Jan 17 05:30:43 hehe Jan 17 05:30:52 what kernel version btw? ( uname -r ) Jan 17 05:31:08 I tried, and had seen that protocol list, just was hoping someone had more experience with it Jan 17 05:31:12 10 years, a lot has happened since then Jan 17 05:31:21 let me ssh back in Jan 17 05:32:50 4.1.6-bone15 Jan 17 05:33:53 The same company that sells this shield also sells their own LCD that I did not buy, so possible it is some weirdness due to being preconfigured for it Jan 17 05:34:52 you can try running its edid data (in /sys/class/drm/card0-HDMI-A-0/edid probably or something) through edid-decode (package edid-decode) or parse-edid (package read-edid) Jan 17 05:35:10 to see what resolution(s) the display reports Jan 17 05:35:25 thx Jan 17 05:35:39 edid-decode seems to work better afaict Jan 17 05:37:19 kk, will try to play with it for a bit Jan 17 05:38:11 card0-HDMI-A-1 sorry, 1-based numbering there it seems Jan 17 05:41:06 if the application operates directly on the framebuffer, check /sys/class/graphics/fb0 Jan 17 05:41:49 you may be able to echo one of the lines listed in the 'modes' file to the 'mode' file, or try fiddling with 'rotate' Jan 17 05:42:55 there's no rotation support in hardware though, so if it's a framebuffer application then rotating things may need to be done in the application Jan 17 05:43:24 kk. Think I will wait until the people in their IRC channel wake up. Jan 17 05:43:53 Unfortunately I got in early on the kickstarter, so I guess I am only 1 of 18 people who's boards ahve shipped, so not much user base yet Jan 17 05:44:26 Anyway, I really appreciate your help! Jan 17 05:44:41 still, if it's a kickstarted I'd expect some level of support directly from the developers Jan 17 05:44:52 *kickstarter Jan 17 05:44:56 Yeah, but they are asleep right now I believe Jan 17 05:45:20 yes, annoyingly we need to do that now and then Jan 17 05:45:36 about 6:45AM in Norway, so not unreasonable that they are not online Jan 17 05:46:10 well, depends a bit on their sleeping habits, if any Jan 17 05:46:51 Heh, true. But I can see no good excuse for anyone to be awake then... 4, 5am I might still be up, but by 6:45 they should be asleep :-) Jan 17 05:47:08 it's 6:45 here Jan 17 05:47:10 :P Jan 17 05:47:16 PM or AM? Jan 17 05:47:27 no. Jan 17 05:47:29 :P Jan 17 05:47:40 Probably AM and you don't use 12 hour time Jan 17 05:47:48 exactly Jan 17 05:47:51 Gotcha Jan 17 05:48:00 Where are you? Jan 17 05:48:05 .nl Jan 17 05:48:13 So not far from them Jan 17 05:49:11 depending on one's definition of "not far" Jan 17 05:49:19 By US standards :-) Jan 17 05:50:22 13 hours by car from Oslo to Amsterdam, that is not bad Jan 17 05:51:15 not exactly a short drive Jan 17 05:51:42 No, but having driven across most of the US, it is not bad Jan 17 05:51:59 fair enough Jan 17 05:52:24 I live in the midwest of the US, try doing 13 hours through nothing but flat Jan 17 05:53:12 no thanks :P Jan 17 05:53:19 Good idea :-) Jan 17 05:53:31 I get to do that a couple times a year Jan 17 05:53:50 I don't have a driver's license anyway, I stick to public transit Jan 17 05:54:00 Each way, so 26 hours in under a week Jan 17 05:54:25 Yeah, that is alot easier there Jan 17 05:54:36 Lots of cities in the US are transit friendly, not mine Jan 17 05:56:28 according to google maps I could also get to oslo by public transit, though that includes an 8h20m bus drive, which is something I wouldn't exactly look forward to Jan 17 05:57:13 Yeah, says about 20 hours total by transit Jan 17 05:57:23 or two by plane Jan 17 05:58:15 Or 197 hours to walk Jan 17 05:58:49 swimming across the north sea is not given as an option I guess ;) Jan 17 05:59:06 Lol, I assume you would take the ferry :-) Jan 17 05:59:36 You said no drivers license, I assume also no private yacht, right? Jan 17 05:59:50 hehe, no Jan 17 06:00:05 Damn, was a good idea while it lasted :-) Jan 17 06:00:26 fortunately, I no urgent need to be in oslo anyway Jan 17 06:00:36 Well, there is that Jan 17 06:00:44 *I have Jan 17 06:00:56 Yeah Jan 17 06:02:47 cd Anyway, I better run. Thanks for all your help. I'll lurk in case anyone else has any ideas, otherwise I'll wait for them to wake up Jan 17 06:02:57 you're welcome Jan 17 07:38:07 Good Day, Jan 17 07:41:09 I am having trouble connecting to beagle board on startup, windows 10, beagle board xm rev c. got it as a gift. windows only gives error that usb device malfunctioned. Jan 17 07:41:22 Can any one please assist, Jan 17 07:41:43 Drivers installed from beagleboard.org Jan 17 07:42:58 vintage hardware Jan 17 07:43:18 do you have something to connect to the RS-232C serial port? Jan 17 07:44:24 Unfortunately no Rs-232c connectivity, Jan 17 07:44:43 Is there any interface options via LAN? Jan 17 07:45:00 get a usb-to-serial cable then. It makes everything better. Jan 17 07:45:21 Which Distribution are you running on the BBxm? Jan 17 07:45:48 I will do, is all the software stored on the micro sd? as the board does not have one installed. Jan 17 07:46:33 What image should be installed on the sd card? as the previous owner lost it. Jan 17 07:50:27 the BBxm doesn't have any on board storage Jan 17 07:50:43 so it will not do anything useful without a SD card inserted Jan 17 07:51:49 Thank you, what image does need to be on the card? Jan 17 07:52:57 http://elinux.org/BeagleBoardDebian#Demo_Image has some hints Jan 17 07:53:19 it might not be that easy as it seems there is only a tarball, and handling that on windows is impossible Jan 17 07:55:36 this lists an old angstrom image: http://beagleboard.org/latest-images Jan 17 07:55:56 that you should be able to write out also from a windows machine Jan 17 07:56:18 Thank you so much for the assistance, Jan 17 07:57:30 I am busy downloading that image, will try to get info on how to mount the image on the sd card Jan 17 08:00:49 there is something called a disk imager, I think Jan 17 08:01:26 http://beagleboard.org/getting-started#update Jan 17 12:31:39 Hi, newbie here. Can i ask about image processing here? Jan 17 16:48:41 Hi Jan 17 16:48:57 I recently bought beagle bone black Jan 17 16:49:13 it has kernel 3.8.13 bone 79 Jan 17 16:49:24 i was looking for the source code for the same Jan 17 16:49:42 but i couldn't get it Jan 17 16:50:00 the procedure given on various sites differ Jan 17 16:50:33 and when i am following any of them it is first of all fetching around 2 gb of data Jan 17 16:50:46 any help on that Jan 17 16:58:32 the 3.8 kernel is considered quite old, quite possibly deprecated .. why do you need that version? Jan 17 16:59:08 TI have updated considerably since then, and official images are using those instead Jan 17 16:59:23 and yes, 2GB of data sounds like a kernel git checkout Jan 17 16:59:28 so that's expected Jan 17 17:00:50 Thanks Jan 17 17:01:09 so this means that i need to download the latest image and flash it Jan 17 17:01:59 my bbb came with the this version, so i was looking not to update it now Jan 17 17:04:11 actually with most devices it makes sense to update them to latest firmware/image right out of the box Jan 17 17:04:18 as factory images tend to be outdated Jan 17 17:18:38 tbr++ Jan 17 18:37:24 I suppose that is because the manufacturer only has that version too program. Makes sense. However you would think when they changed versions (IE B to C for example) they would update that. Jan 17 19:19:13 GenTooMan: I don't think necessarily 'development' and 'manufacture' end up (sometimes) in the same building even ... Jan 17 19:42:02 veremit I'm pretty sure they aren't in the case of the beagle boards. I'm just happy someone makes boards one can use without it costing an arm and a leg. Jan 17 19:42:29 GenTooMan.. no, that's why they only churn out a few hundred a week >,< Jan 17 19:43:04 there's a small group of people who hand test and program them :) Jan 17 19:43:17 they probably even know what an OS image IS! Jan 17 19:43:29 * GenTooMan is shocked! Jan 17 19:43:39 yeah :/ sad reflection on the rest of the world, though Jan 17 19:43:55 but .. realities of mass manufacture Jan 17 19:45:44 it would be nice if there were more modules available (IE COM boards / computer on a module) like the beagle bone black, commercially. Demand is not huge for them as most people either go small system or buy something that is PLC based for industrial use. Not many things are done in between those extremes. Jan 17 19:46:27 GenTooMan .. are you thinking IO-centred? there are thousands of CoM ARM boards .. but few that have good IO on-board Jan 17 19:46:48 although .. the Pi isn't -bad- and you can easily expand off, say an I2C bus .. Jan 17 19:47:23 I've not seen anything comparable to the PRU modules though Jan 17 19:48:01 You technically should NEVER use I2C in industrial (though people do use it). And yes IO is very important in industrial. 24V /12V bus etc. Jan 17 19:51:56 what preferred io would you use then? ethercat? bit banging? Jan 17 19:53:17 and 24v is a common control voltage, but useless to any chips... Jan 17 19:57:12 the am335x supports ethercat, though the bbb doesn't Jan 17 19:57:38 zmatt: where's the restriction? the phy or other electrickery? Jan 17 19:58:22 pinout is incompatible (note: uses different pins than normal MII), and ethercat is eFUSE-disabled in the am3358 Jan 17 19:58:35 my beagle has a 3359 Jan 17 19:58:48 mine too, but pinout still gets in the way Jan 17 19:59:01 intersting, thanks for that Jan 17 19:59:04 I don't remember what exactly, but you can check my spreadsheet Jan 17 20:00:03 SPI works fine and their are SPI chips that can handle 50V and 200ma for relay outputs. However kind of specialized so one uses things like ULN2007 ussually with like a HC595 IC. Jan 17 20:00:21 ooo Classic logic :D hehe Jan 17 20:00:56 The 595 is practically a SPI IO chip the way it was designed. :D Jan 17 20:03:10 TI makes a power version of it but a bit pricey. Jan 17 20:05:23 for robust connectivity CAN is another option of course, though I'm not sure whether there's such a thing as CAN-controlled I/O (wouldn't be surprised though) Jan 17 20:07:05 You can use CAN Kingdom or other CAN based protocols. They tend to be horribly complicated. Jan 17 20:08:38 lol @ CAN Kingdom's terminology Jan 17 20:08:43 CAN has no actually physical addresses or protocol associated with it so you have to make your own. Jan 17 20:09:21 right, but e.g. for a bunch of digital outputs a single ID to which you write the data would be fine Jan 17 20:11:39 we use a lot of rs485 Jan 17 20:11:42 electrically Jan 17 20:11:46 with our own bus spec Jan 17 20:13:50 hence oddly whyI keep bangin on about TXEN lol Jan 17 20:14:15 STM32 got an easy/reliable way to do it .. we've used those a bit recently Jan 17 20:15:04 just use an uart and a gpio ? Jan 17 20:15:14 yes Jan 17 20:15:27 much like how the am335x should work Jan 17 20:15:36 I meant on the am335x Jan 17 20:15:49 well that's the theory .. I've yet to load the patch Jan 17 20:16:13 oh, if I needed an uart for any *serious* applications (not just a console) I would swiftly dump the kernel driver Jan 17 20:16:32 TX some UARTS have that as part of their RTS signal. Jan 17 20:16:38 err TXEN Jan 17 20:17:16 yeah this one doesn't so you need to wait for a TXEMPTY irq Jan 17 20:18:15 GenTooMan: sometimes .. but the timing is vey different to RTS Jan 17 20:18:47 zmatt got it :) that's mostly how the stm32 implementatin works .. but ofc you're working baremetal not through an OS Jan 17 20:19:07 if you want the transmitter to start very swiftly after being enabled, you could loop the TXEN gpout to the CTS input to stall the transmitter until TXEN is asserted Jan 17 20:19:25 you may actually be able to use a single pin for that, but you'd nede to test that Jan 17 20:20:10 zmatt: interesting thought .. you'd have to re-enable flow control Jan 17 20:20:11 does require the TXEN to be active-low Jan 17 20:20:22 enable, not re-enable Jan 17 20:20:22 yeah thats non-conventional Jan 17 20:20:36 you can't change flow control options after initialization Jan 17 20:20:39 zmatt: you get my drift .. you're not normally using flow control Jan 17 20:20:48 period. Jan 17 20:21:14 well I normally do, except for the bbb which annoyingly leaves CTS and RTS not-connected on both the serial header and the am335x Jan 17 20:21:18 I doubt the timing is -that- bad Jan 17 20:21:34 zmatt: yeah bloody cheapasses lol Jan 17 20:21:49 great fun when you want to copy-paste something into u-boot Jan 17 20:21:58 Wallop! lol Jan 17 20:22:20 yeah I don't use flow control .. its another 2 pins to connect .. 3 is handy .. 5, less so .. Jan 17 20:22:41 if you've got control of both ends of the link .. and some form of 'ack' its easy enough Jan 17 20:22:46 or you can use xon/xoff lol Jan 17 20:23:55 and w.r.t. timing depends on how important it is to Jan 17 20:24:13 minimize the txen margins around transmission Jan 17 20:24:31 I'm only working at 115200 to be compatible with PCs for debugging .. even over usb/ converters Jan 17 20:24:54 could go faster, but that's sufficient for most things .. except dmx Jan 17 20:25:00 if you're the requestor I'm guessing it's mostly important to disable tx after transmission (within the turnaround time) Jan 17 20:25:12 ^^ yes Jan 17 20:27:14 I've measured irq latency to userspace (via uio) on an rt kernel to be about 40-80 μs Jan 17 20:28:33 SCHED_FIFO: http://gerbil.xs4all.nl/irq-latency2.png Jan 17 20:28:47 SCHED_OTHER isn't much worse on average, but does have outliers: http://gerbil.xs4all.nl/irq-latency1.png Jan 17 20:32:12 (the width of the output pulse is determined by the access time from cortex-a8 to gpio controller because userspace mappings of devices are strongly-ordered) Jan 17 20:35:20 meeting the E1.20 requirements for dmx packet timing would be... challenging... to do from linux userspace on the cortex-a8 Jan 17 20:35:30 (pru to the rescue in that case) Jan 17 20:37:18 edma and a timer might also be able to do the job Jan 17 21:02:44 [0][0] = 1er cercle en haut à gauche, [2][2] = dernier cercle en bas à droite Jan 17 21:03:15 au début tous les éléments du tableau sont initialisés à 0. Au fur et à mesure que tes joueurs cliquent, tu mets à jour avec le n° du joueur Jan 17 21:03:38 wrong channel, sorry Jan 17 21:04:05 zmatt: yeah if I'm ever misfortunate enough to do dmx in an am335x .. Ill be using the pru Jan 17 21:04:20 otherwise I'm sticking to c/asm in a micro :D Jan 17 21:05:38 well, some people also make it using an FT232R (switch to 9600 baud and send a 0-byte to make the break) ... I think the cortex-a8 on the am335x can probably generate better timing than that :P Jan 17 21:11:26 zmatt: that's one "preferred" mthod Jan 17 21:16:31 somehow I doubt that method will meet the E1.20 mark-after-break requirements (min 12 us max 88 us) Jan 17 22:31:25 zmatt .. documents I got say there is no 'maximum' .. or perhaps I'm thinking of another component of the crazy timing that dmx has .. Jan 17 22:31:37 its all really quite 'flappy' Jan 17 22:33:01 depends on revision of the standard, though iirc even the original had some maximum Jan 17 22:33:47 E1.20-2006 (RDM) does however tighten the timings over the original E1.11-2004 Jan 17 22:37:25 E1.11-2004 not specify a particular maximum MAB time (only a minimum of 8 us) but does specify a maximum frame time (from falling edge of break to next falling edge of break) Jan 17 22:37:57 namely 1s Jan 17 22:38:01 ah Jan 17 22:38:09 I need to look out my spec doc Jan 17 22:38:24 actually, min MAB for transmitters is 12 us Jan 17 22:38:50 its a pretty horrible frame format Jan 17 22:38:53 8 us for receivers, with an optional feature for receivers to accept a 4 us MAB (for compatibility with older equipment) Jan 17 22:41:29 bbl, got train to catch Jan 17 22:54:49 Hi! I'm trying to boot off the sd card on my beaglebone green, but it keeps booting on the emmc Jan 17 22:55:12 The sd card uses an image intended for the beaglebone black - should it be compatible? Jan 17 22:58:27 hey, i am writing a linux kernel module to handle interrupts created by the PRU. does the default prussdrv hold those interrupts or can I also be interrupted by them? Jan 17 23:11:56 gyscos: if an older u-boot is used and the rootfs does not specify a filesystem uuid in its /boot/uEnv.txt then it can happen easily that the wrong rootfs is used Jan 17 23:12:28 gyscos: you're forcing sd-boot by holding the S2-button down while powering on? Jan 17 23:16:19 (if not, then the eMMC u-boot is always used if installed... it may still try to boot from sd card, depending on the boot script used, which varies per u-boot version and is in all cases quite convoluted) Jan 17 23:19:24 ocamlman .. you mean does the prussdrv handle the interrupts already? I should imagine so .. you'd have to blacklist the driver if you wanted an alternative handler ... Jan 17 23:20:27 veremit, well I want to use prussdrv for loading and starting the program but I want to handle the interrupt myself. I don't know if prussdrv has set the interrupt to sharable or if it even registers itself for that interrupt until the user has indicated they want that interrupt to wake them Jan 17 23:20:46 althuogh I think there are systems "out there" that you can 'register' multiple interrupt handlers .. but that kinda warps my brain Jan 17 23:21:30 I've got burnt by nested interrupt handlers before .. very clever but often quirky modules Jan 17 23:21:52 bbl Jan 17 23:23:52 veremit I hate nested interrupts I always look for ways to avoid those like the plague. Some people insist on making them because they put too much in the interrupt (I've seen some very ugly interrupt code). Too much time in an interrupt means other interrupts have a high chance of colliding. Jan 17 23:24:19 GenTooMan: exactly the problem I was running into Jan 17 23:26:04 Interrupts should be very short IE stuff data into FIFO from serial port. Feed serial port from FIFO. Icrement a single timer value (single) and set a flag. That's about the biggest you should use. Unfortunately some people put entire network stack handling in damned interrupts. Jan 17 23:28:04 hehe Jan 18 00:16:18 I wanted to mmap /dev/mem to write to memory mapped i/o for setting pinmux. On the BB-Black that didn't work. So I used the P*_pinmux files in /sys/. Now on the BB-Green those files do not seem to be available anymore. How am I supposed to set the pinmux values? Jan 18 00:22:00 read zmatt's spreadsheet :D Jan 18 00:22:03 GenTooMan: well, if you're segregating code into two latency tiers (irq and main code) then it may also make sense to segregate it into three or more tiers, though some careful analysis should be done Jan 18 00:23:12 GenTooMan: conversely, if no code has particularly low latency requirements, having a single tier may make sense; this is what I currently have in my baremetal tests (irqs disabled, mainloop consists of WFI followed by call to irq dispatcher) Jan 18 00:23:17 veremit: I that was directed at me, do you have a link? Jan 18 00:23:38 * veremit points to zmatt ..right there :D Jan 18 00:23:51 zmatt: shevek .. needs your pinmux Jan 18 00:24:20 shevek: it sounds like you had some odd distro to have things like that in it Jan 18 00:24:21 shevek: pinmux is specified via devicetree, either the main device tree or an overlay loaded via configfs (kernel 4.x) or bone_capemgr Jan 18 00:24:46 veremit: or (makes gestures of disgust) cape-universal Jan 18 00:25:00 zmatt: of what?! Jan 18 00:25:02 zmatt: I want to be able to set most pins to input, output or pru output. Jan 18 00:25:11 easily done in DT Jan 18 00:25:28 I don't want fixed bindings; I need the user of my program to change things on the fly. Jan 18 00:25:46 Last time I tried to write a DT for that, it seemed very hard. Jan 18 00:26:07 bonescript generates and loads device tree overlays on the fly I think, but it's non-trivial indeed Jan 18 00:26:24 that does seem like a perfectly reasonable thing to do .. Jan 18 00:26:28 note that gpio is the default of non-muxed pins, so you can just export them via sysfs and set their direction Jan 18 00:26:35 and I could imagine the kernel is really clunky about it Jan 18 00:26:48 you can in fact also just change pinmux directly via /dev/mem Jan 18 00:26:49 zmatt: thats not a bad plan Jan 18 00:26:55 Yes, that works; I'm trying to get the pru outputs to work at the moment. Jan 18 00:27:11 zmatt: No, I tried. The kernel refuses to change anything when I write there. Jan 18 00:27:17 https://github.com/dutchanddutch/jbang/ Jan 18 00:27:55 If that would work, it would be perfect; that would also be a lot faster than traversing the filesystem. Jan 18 00:27:57 this is a little program I wrote for the bbb that meddles directly with pinmux Jan 18 00:28:05 (also directly with PRCM and GPIO) Jan 18 00:28:30 the thing about the control module is that it ignores unprivileged writes Jan 18 00:28:46 this I solved as follows: https://github.com/dutchanddutch/jbang/blob/master/src/hw-subarctic.cc#L17 Jan 18 00:29:16 ^_^ Jan 18 00:30:54 (actual implementation of the trick is in privileged.h in same dir) Jan 18 00:32:15 Ah, that's what I was looking for. Thanks! Jan 18 00:32:30 beware that this trick may cease to work some day Jan 18 00:32:44 technically I'd consider it to be a bug in the kernel Jan 18 00:36:07 zmatt: Yes, that makes sense. But what would be a proper way? Write a module that does it? If I map /dev/mem as root, I expect to have full control. Jan 18 00:36:41 the latter point is why I said "technically" rather than actually reporting it as a bug :P Jan 18 00:37:11 if an evildoing program runs as root and can open /dev/mem you're pretty hosed anyway Jan 18 00:38:43 Indeed. Jan 18 00:43:14 I was slightly worried when CONFIG_CPU_SW_DOMAIN_PAN was introduced (default y), but curiously it does not (currently) seem to depend on CONFIG_CPU_USE_DOMAINS Jan 18 00:49:18 as for the proper way, using DT is the only proper way and ensures the kernel knows about what's going on (and avoids conflicts) Jan 18 00:50:06 if you want that but easier, there's... I suppose... always still the possibility of using, ehm, ew, cape-universal Jan 18 00:50:48 I think I did that on the black, but it seems to have stopped working on the green. I probably did something wrong there. But I much prefer using /dev/mem anyway. Jan 18 00:51:05 well for that, see jbang :) Jan 18 00:51:29 and if you need to know the indices / mux values, my spreadsheet ( https://goo.gl/Jkcg0w ) Jan 18 00:52:16 (the three orange-marked tabs are BBB-specific, though BBG is basically the same) Jan 18 00:52:35 I already have all those things figured out, but thanks. Jan 18 00:52:51 okay :) Jan 18 00:53:58 This one looks very nice though, I think I'll keep it. :-) Jan 18 00:54:01 hehe Jan 18 00:54:20 it's a composite of a lot of info sources Jan 18 00:55:53 Yes, I know the sources, I've been reading them as well. Jan 18 00:56:55 (note the BBB sheet actually still some columns hidden to avoid clutter, things like power domain and boundary scan register bits) Jan 18 00:57:59 here in the office the P9/P8 tabs are used most often (typically as printouts) Jan 18 01:00:35 (apologies for occasionally using eye-stabbing colors for pins that have special constraints) Jan 18 01:01:01 Haha, no problem. Jan 18 01:44:06 Hey Yo...do we use "apititude update" now instead of "apt get update"? Jan 18 01:44:39 apt update, apt-get update, aptitude update ... they all have basically the same result (except aptitude is slower) Jan 18 01:44:58 oh...thank you Mr. Zmatt. Jan 18 01:45:27 I never used to use the "-" b/t apt and get. Jan 18 01:45:42 I always typed in "apt get update." Jan 18 01:46:09 So, I need to type in apt-get update and not apt get update. Right? Jan 18 01:46:18 it Jan 18 01:46:20 it Jan 18 01:46:21 argh] Jan 18 01:46:35 it's always been apt-get update Jan 18 01:46:45 Oh. Man, my damn memory is terrible. Jan 18 01:46:46 Sorry. Jan 18 01:47:06 I am on break and it is time to update. Thank you again zmatt. Jan 18 01:47:11 apt is a wrapper to apt-get and apt-cache meant to be slightly user-friendlier Jan 18 01:47:18 apt update will work too Jan 18 01:47:49 Cool beans. Thank you again. Jan 18 01:55:53 okay...I got problems. I cannot upgrade. I am getting asked a question. It states, "Unable to lock the administrative directory, is another process using it?" Jan 18 01:55:56 Please send guidance. Jan 18 01:58:31 * GenTooMan draws pictures of a black hole for beagle bone black data to go into. Jan 18 01:59:13 Not sure what directory it's talking about. You are using debian linux on the BBB? Jan 18 01:59:21 Yea. Jan 18 01:59:44 are you root? Jan 18 01:59:58 Yea. Jan 18 02:00:06 sounds like apt is unhappy Jan 18 02:00:18 I fixed it. Jan 18 02:00:28 then it usually means you have another apt/apt-get/aptitude in progress somewhere (left open in another terminal?) Jan 18 02:00:37 I just unplugged the BBB and plugged it in again. Bam! It works. Jan 18 02:00:49 lol NIIIIICE Jan 18 02:00:53 Yep! Jan 18 02:01:03 that's not gonna do it any harm .. lol Jan 18 02:01:17 next time consider 'reboot' instead :P Jan 18 02:01:18 That's like hitting it in the head though. Jan 18 02:01:23 Good idea! Jan 18 02:01:26 I panicked. Jan 18 02:01:30 lol Jan 18 02:01:38 GenTooMan, zmatt .. no .. banging it hard on the desk would be better Jan 18 02:01:45 I usually type shutdown -h now. Jan 18 02:01:46 good point Jan 18 02:01:56 Beagle Bone Black abuse! Jan 18 02:02:03 GenTooMan+1 Jan 18 02:02:18 That is me! Bang and Bash. Jan 18 02:02:26 No pun intended. Jan 18 02:02:30 GenTooMan: you know there's always i2c-0 and the ability to request voltage changes from the pmic... Jan 18 02:03:15 zmatt So you can use the i2c bus to turn the power off in an application for shutdown? Jan 18 02:05:03 sure thing, just write 0 to register 0x01 (PPATH) Jan 18 02:06:27 that will get you a nice clean shutdown, probably looking something like http://elinux.org/File:Bbb-c-3v3u-powercut-hdmi.png Jan 18 02:07:01 ;) Jan 18 02:09:37 Do you fellows remember me looking to find a 12-pin JTAG (two row)? Jan 18 02:10:27 12-pin ? the header on the BBB is 20-pin Jan 18 02:10:38 Standard JTAG is 20pin Jan 18 02:10:42 I know...this is for another device. Jan 18 02:10:53 It is not a standard type of "thang." Jan 18 02:10:54 GenTooMan: "standard" ... Jan 18 02:11:25 GenTooMan: ARM standard? newer ARM standard? TI standard? older TI standard? ;P Jan 18 02:11:39 I've heard of 14 but not 12. TI is 14 Jan 18 02:11:50 TI is 20 usually nowadays Jan 18 02:11:56 I know...this where the complication lies. Jan 18 02:11:57 (as on BBB) Jan 18 02:12:17 It is near impossible. I will have to make one. Jan 18 02:12:18 http://processors.wiki.ti.com/index.php/JTAG_Connectors Jan 18 02:12:51 zmatt: What 'dat my friend? Jan 18 02:13:37 Nevermind. I will go and look. Jan 18 02:14:55 Well JTAG never technically had a standard then users of the JTAG interface did "weird stuff" Jan 18 02:16:04 Some people use USB to JTAG. Jan 18 02:16:07 JTAG was also never really meant for debugging but for boundary scan Jan 18 02:17:00 So they use it for programming debugging AND boundary scan. No problem ... really. Jan 18 02:17:31 I saw this JTAG device and it had 10, two-row configuration. Jan 18 02:17:36 Man...I was so close. Jan 18 02:18:21 I gave up to try and situate some wires myself. Jan 18 02:22:05 Well programming an programmable logic device with JTAG probably requires some "extra" things. JTAG programming of all devices requires a driver for each device. Kind of nuts that no one bothered to 'create' a standard for any of that. Jan 18 02:22:59 GenTooMan: I think they wanted to keep implementing it in the chips as simple as possible. This way, programming can be done in a way that fits the design, whatever the design is. Jan 18 02:23:59 and there are standards to describe how e.g. fpga/cpld programming should be performed Jan 18 02:24:35 using an extended bsdl file containing programming sequences Jan 18 02:24:58 (IEEE 1532) Jan 18 02:28:12 upgrade is giving me trouble with some "stuff" named connman. Jan 18 02:29:16 I guess it is stuck. Dang! Jan 18 02:30:36 It says installing new version of config file /etc/init.d/connman ... Jan 18 02:30:56 Should I wait? Jan 18 02:32:24 zmatt that will do it :D Jan 18 02:35:44 Usually my BBG is a bit quicker. I have waited for five minutes and nothing. Jan 18 02:36:09 Set_: Try pressing space and see if it still responds. If not, you lost your connection. In that case, press the power button to power it down, then again to power it back up and reconnect. Run dpkg --configure --pending to resume the installation. Jan 18 02:36:54 Okay...I did not lose power. My LEDs are blinking and my connections states on via PuTTY. Jan 18 02:37:20 Glitch. Jan 18 02:37:29 Set_: Yes, but I don't know what connman does, but it sounds like it might mess with your connection. ;-) Jan 18 02:37:46 You are right. It just sent me offline. Jan 18 02:37:54 I will redo it. Jan 18 02:38:54 heh that doesn't surprise me lol .. any connection manager unless you Ban your active connection .. will surely break it Jan 18 02:39:50 The update worked but the upgrade gave me trouble. Jan 18 02:40:18 veremit: systemd-networkd has never broken my connection just because I changed a config file (or even restart the daemon) Jan 18 02:40:30 Set_: Update is never a problem. That just downloads which packages are available. Jan 18 02:40:40 oh you and your friend poettering zmatt :p Jan 18 02:40:46 lol Jan 18 02:40:59 Huh, Poettering has friends? :-p Jan 18 02:41:03 I'm sure he wipes you $ss too Jan 18 02:41:26 shevek: apparently Jan 18 02:41:30 veremit: apparently he's also responsible for avahi, that's reason enough to not be my friend :P Jan 18 02:41:36 yikes .. avahi Jan 18 02:41:42 Damn man...I cannot get back online. Jan 18 02:41:44 and zeroconf? Jan 18 02:41:44 Boo! Jan 18 02:41:48 and all that fun? Jan 18 02:42:05 Set_: Did you reboot the device? Jan 18 02:42:42 veremit: at least the mDNS code Jan 18 02:43:00 Nope...I just powered down and plugged it back in. Jan 18 02:43:11 but but . its poettering .. it must be Good and it must Be. Jan 18 02:43:11 Zeroconf is designed by Apple AFAIK. Avahi is just an implementation for it. Jan 18 02:43:21 I am using Cloud 9 now. Jan 18 02:43:25 It works for now. Jan 18 02:43:26 Set_: That means "yes". ;-) Jan 18 02:43:31 Oh! Jan 18 02:44:19 I see there are some packages that are no longer required. Is that a custom? Jan 18 02:44:33 Some stay and some go? Jan 18 02:45:28 It tells me to use apt-get autoremove to remove those packages. Jan 18 02:45:31 Should I do that? Jan 18 02:45:40 Heh? Jan 18 02:46:07 I suspect the less you do the better. Jan 18 02:46:13 Hhahaha. Good idea! Jan 18 02:46:51 I use to say, "I can Do-EEEEE!" Now...I say, "I do not Do-EEEE!" Jan 18 02:47:10 Sorry. Jan 18 02:47:51 Could be worse I guess. Jan 18 02:48:09 if you're suddenly running openbsd, something went wrong somewhere Jan 18 02:48:35 Yep. Jan 18 02:49:44 All that just to blink an LED Set_ geesh. Jan 18 02:50:12 Yep...I got a half-bot so far. Jan 18 02:50:24 I got too much hardware so far and not enough software. Jan 18 02:50:28 I am still working. Jan 18 02:50:52 Hardware = easy-peasey, George Leweasy. Jan 18 02:51:06 software = damn it, not again **** ENDING LOGGING AT Mon Jan 18 02:59:58 2016