**** BEGIN LOGGING AT Mon Nov 19 03:00:00 2018 Nov 19 03:06:32 pweh are you a bot Nov 19 03:06:49 jeez suprothunderbolt i need to get uxmux out Nov 19 03:06:52 but first i gotta do THIS Nov 19 03:08:51 mmmm... the SGX build script appears not to work for me... time to investigate... Nov 19 03:12:13 what kinda gui is it going to be? Nov 19 03:12:40 qt quick, touch screen based Nov 19 03:13:19 just a few simple interactive plots is the plan Nov 19 03:16:28 interesting Nov 19 03:16:31 plots as in like, okay Nov 19 03:16:32 cool Nov 19 03:29:56 im a bot Nov 19 03:40:45 thats what i thought Nov 19 03:51:36 im a bot Nov 19 06:27:48 KotH: allpcb are interesting... They spotted a problem with my design after panelization, there was a cut-out at the PCB border that they could not route out with their CNC because it was too thin Nov 19 06:33:00 KotH: so, I updated the gerbers and only after that I looked at how they did the panelization, realized another error and emailed them an update Nov 19 06:34:01 KotH: they reviewed the update tonight. redid the panelization and this morning I had an email asking for confirmation Nov 19 06:34:48 all that effort for 20 boards on 5 panels, for 21 USD, which already include DHL shipping Nov 19 06:35:08 surely they haven't made any money on that order Nov 19 07:30:06 thinkfat: add another 40€ for "not-customs" fee :-/ Nov 19 07:30:12 thinkfat: but yeah.. good companies do that Nov 19 07:30:38 thinkfat: i have a short list of assembly companies in .ch that carefully check designs for errors Nov 19 07:31:02 thinkfat: prototypes always went to them, even if they were twice as expensive as others Nov 19 09:55:09 uh oh, "sgx build script" ... I suspect he's talking about the really ancient one Nov 19 09:56:04 but suprothunderbolt never stays around for me to reply to what he says :-/ Nov 19 10:01:42 hi. Has anyone used buildroot with beaglebone_defconfig? Nov 19 13:24:46 umbaman: tip: don't ask "has anyone" questions since the answer to that question (either "yes, someone has" or "no, noone has") is not actually useful to you. Nov 19 13:25:10 you probably have an actual question about it. ask that question instead Nov 19 14:59:10 Hi, we are using the beagle X15 in our product and its great. I need to get in touch with someone from manufacturing or supply logistics. If we were to bulk purchase 5000 units 1: what would the price and lead time be, and 2 if it was instead 10 000 units what would the price or lead time be? Nov 19 15:00:08 Decs: ask your favorite distributor :) Nov 19 15:02:00 Thats being done, can you guys meet the demand and what is your lead time on orders that large? are we talking weeks, months or years? Nov 19 15:03:14 Decs: we are not the distributors. we are a group od random weirdos who try to answer technical questions for fun in our spare time. for business topics, really, ask the electronics distributor of your choice. those who sell beagle-branded items should be listed on the website. Nov 19 15:04:04 (being Arrow, Mouser, and DigiKey for the time being) Nov 19 15:05:05 Decs: like LetoThe2nd said, this is a community chatroom, not a sales/support division Nov 19 15:05:24 Ahh, sorry I was mistaken, I thought you guys worked at Beagleboard, do you know if there anyway to call someone from bealge sales7support? Nov 19 15:05:56 Decs: sorry to repeat. but the beagleboard thing is not exactly what you seem to think. its not a business thing that sells hardware. Nov 19 15:06:20 the design is from beagleboard.org, but they do not manufacture boards Nov 19 15:06:26 hence can't answer the questions you asked Nov 19 15:06:36 Decs: really, i listed the three "official" distributors for the x15. they do business. they do have sales departments. go ask there. Nov 19 15:06:54 Thanks :) Nov 19 15:11:19 *sigh* I really hate companies that firewall all non-whitelisted outgoing connections, and then complain when products that require internet connectivity don't work correctly Nov 19 15:11:52 zmatt: want me to chime up some LTE 5G sales tune? Nov 19 15:12:31 we're basically getting pressured to tunnel all our traffic over https... hurray, that would obviously make the internet more secure and efficient -.- Nov 19 15:12:55 absolutely! Nov 19 15:34:47 zmatt: and there i thought we were over that berlin firewall thing Nov 19 17:22:48 hey guys, looking to work with bonescript and was wondering if there is a guild on how to setup a node app with bonescript to run as service (systemd) somewhat securely (or at least best practice) Nov 19 17:26:45 well the only contribution the systemd service can make towards security is restricting privileges and thereby hopefully limiting the damage possible if your application does end up compromised Nov 19 17:28:03 yeah so, i guess my question is more so, that because the bonescript is accessing GPIO it needs to be elevated, and the real issues is i dont understand what that means well enough to do it. Some of the things i've read say use a generic account with no access and run the app via systemd with that, elevated, however, with my level of knowledge, that isnt really enough for me to go on Nov 19 17:28:26 gpios don't necessarily require elevated privileges Nov 19 17:28:30 depends on the udev rules Nov 19 17:28:40 hmm well, i can say that if i run my app without sudo it crashes Nov 19 17:29:57 I don't know what the default privileges are. I thought that current images give gpio access to group 'gpio' by default but I haven't checked or anything Nov 19 17:30:23 hmm in poking around i dont think thats the case, but not totally sure, i did grab a fresh image a couple of weeks ago Nov 19 17:31:38 on our systems I do assign gpio access to group 'gpio', and I use a custom devicetree where individual gpios are named and configured. this also means userspace doesn't have the right to change their direction, which I view as rather important since changing an input to output could easily damage the AM335x and/or external hardware Nov 19 17:33:53 lol i sorta want to show you what im doing, but im mildly ashamed and scared Nov 19 17:34:47 are you doing it as a professional or a hobbyest rofl Nov 19 17:34:51 hobbiest8 Nov 19 17:34:51 my DT and udev rules combined result in: https://pastebin.com/raw/BHuEEuMj Nov 19 17:34:53 YUK Nov 19 17:34:53 hobby Nov 19 17:35:03 then don't worry about it Nov 19 17:35:46 well that depends on the situation Nov 19 17:35:56 i work more in devops... so really i just need to know how to script (python/ruby whatever) and i was more a windows guy, (until i was dumped in devops) so my linux knowledge is ok but not nearly as strong as windows... and i've been trying to get on top of node for a while.. and its kinda painful Nov 19 17:36:12 im building an aqua/hydro-ponics manager Nov 19 17:36:15 because you're not a programmer Nov 19 17:36:20 ? Nov 19 17:36:27 def not a programmer Nov 19 17:36:31 did you move from IT to dev ops Nov 19 17:36:39 huh crazy Nov 19 17:36:44 yeah, but i "can" program (C# for a while) Nov 19 17:37:13 basically im at a 'jr dev' sort of level, understand and converse on most topics, cant really implement all of it Nov 19 17:37:20 https://github.com/jrich523/aquaman Nov 19 17:37:53 like i grasp the idea of dependency injection, but no way i could write it :) Nov 19 17:37:57 jrich523: in general, if your application doesn't work without root privileges, step 1 would be to find out what it's trying to access that it can't when run as normal user Nov 19 17:38:07 the 'debian' user by default has pretty far-reaching privileges already Nov 19 17:38:22 (use the "id" command to see the massive list of groups it's in) Nov 19 17:38:40 yeah and right now im using that account (debian) Nov 19 17:39:13 oh, there is a GPIO group :) Nov 19 17:40:38 yep, there is Nov 19 17:41:35 so, in talking about the udev stuff, im not sure i could go to that level since the hardware is configured (defined) by a config file, with the goal to make it easy to reconfigure it easily without dicking with too much code Nov 19 17:41:59 yep that will significantly limit your ability to restrict the app's privileges Nov 19 17:42:32 although having the ability to mess with pins still doesn't have to imply root privileges or the ability to mess with the system in general Nov 19 17:43:17 yeah i guess maybe my goal was more so to understand whats going on (linux wise) since in reality security of this device isnt all that important :) Nov 19 17:43:39 sounds like a useful thing to figure out Nov 19 17:43:44 i mean worst case someone could ruin my crop... which, i'll likely do on my own anyways :) Nov 19 17:43:49 lol Nov 19 17:45:04 as long as you include a series resistor (4.7k or so) on all input pins (i.e. that are driven by external hardware), that would indeed be the worst they can do Nov 19 17:46:45 most of it is really straight forward, the only write type stuff is a relay outlet box (fans/pumps/lights) and then sensors Nov 19 17:46:57 https://photos.app.goo.gl/D1FwEzumoJQt55Bb6 if you're totally bored :) Nov 19 17:47:14 outputs (e.g. relays) are not the concern, inputs are Nov 19 17:47:32 btw with relays, do be careful to avoid drawing too much current from beaglebone's gpios Nov 19 17:47:54 (they are specified up to 4 or 6 mA depending on pin) Nov 19 17:47:55 total or per pin? Nov 19 17:47:56 omh so Nov 19 17:48:02 there's these new mems relays Nov 19 17:48:03 HF Nov 19 17:48:04 per pin Nov 19 17:48:05 hi current Nov 19 17:48:10 hi voltage Nov 19 17:48:11 like 200V Nov 19 17:48:25 like 5 of them in a 3mm by 3mm package Nov 19 17:48:39 * ayjay_t drools Nov 19 17:48:45 probably gonna be like 50 bucks each tho Nov 19 17:48:57 the lights and stuff are LED so power wise, US so its all 120, and most of its like 1-5amp per device, and hardware wise (outlets/wire etc) should be good for 10amp per Nov 19 17:49:06 at that point, i could just by a prebuilt relay box :) Nov 19 17:49:11 and apparently, according to ben krasnow, 2% helium air concentration is enough to break the mems oscillators for the RTC on iphones Nov 19 17:49:14 ayjay_t: 50? bummer Nov 19 17:49:22 yeah the helium thing is funny Nov 19 17:49:23 CoffeeBreakfast: i imagine because thats what analog devices are Nov 19 17:49:35 CoffeeBreakfast: the ones i was looking at tho was an add for preproduct release Nov 19 17:49:41 from a different company Nov 19 17:50:34 helium and cellphones, that rings a bell Nov 19 17:50:44 yeah uhhh Nov 19 17:50:55 it was an mri machine with a hellium link took down hospital iphones Nov 19 17:51:07 lol Nov 19 17:51:09 helium* Nov 19 17:51:22 had an MRI machine take off a door, samething right? :D Nov 19 17:51:34 those magnets dont mess around Nov 19 17:51:43 no they do not Nov 19 17:51:54 lucky it was a test run, otherwise someone would have been dead Nov 19 17:52:06 I've read plenty of reactions from other people who had similar problems in workplace situations where helium sometimes ends up being released (e.g. during refills/transfers) Nov 19 17:52:19 i mean its a super tiny molecule right? Nov 19 17:52:21 wish i had gotten to see it as it happened lol Nov 19 17:52:25 it must diffuse really easily and really quickly Nov 19 17:52:36 yeah, the problem is helium diffuses through the airtight seals Nov 19 17:52:37 i'm not saying that there aren't practices to avoid that Nov 19 17:52:45 it's quite a topic over at ifixit Nov 19 17:52:52 but consider its inert, you probably dont see the regulation around it Nov 19 17:52:59 i mean, inert Nov 19 17:53:01 non toxic Nov 19 17:53:04 not sure what official word to use Nov 19 17:53:08 not sure how important on the grand scale of things, though Nov 19 17:53:15 haha Nov 19 17:53:29 i mean, you can weaponize it right Nov 19 17:53:44 go get a baloon with helium and just release it near the pockets of a rich person you dont like Nov 19 17:53:47 VERY NAUGHTY Nov 19 17:54:16 they make a helium beer too... thats a fun google Nov 19 17:54:34 jrich523: anyway, for general understanding: devicetree declares the existence of hardware to linux, then udev will assign privileges and create symlinks based on its rules Nov 19 17:54:53 thanks zmatt Nov 19 17:55:44 jrich523: and when using groups for privileges, you can either place users in groups, or have systemd grant additional groups to a service (even if it runs under a user which isn't member of those groups) Nov 19 17:55:58 really wondering if those china pcb shops have a 24/7 schedule Nov 19 17:56:30 is there a best practice around that? Nov 19 17:56:44 don't think so, it's very situation-dependent Nov 19 17:56:56 thanks Nov 19 17:59:08 jrich523: you can also restrict device access to specific devices in the service file (which imposes restrictions on top of the filesystem permissions), along with other restrictions: https://pastebin.com/raw/yKju8e6L Nov 19 17:59:56 note however that gpios are not actually devices, they're controlled via sysfs attributes, hence only filesystem permissions (configured by udev) are relevant Nov 19 18:00:33 NoNewPrivileges=yes is pretty powerful and highly recommended whenever possible Nov 19 18:01:59 it means that neither the service nor any process it spawns can ever acquire any privileges it didn't already have, e.g. by executing a setuid executable Nov 19 18:04:45 that seems like a damn good thing... just did a bit of reading on it Nov 19 18:07:37 also, some things traditionally limited to root, e.g. binding to a port below 1024, are actually controlled by "capabilities" (CAP_NET_BIND_SERVICE for this example) and can be granted to non-root processes by adding them to CapabilityBoundingSet and AmbientCapabilities Nov 19 18:08:05 or you can have systemd open the socket for you and pass it to your service Nov 19 18:08:36 you're giving me a good grasp of how little i know on the topic lol Nov 19 18:08:50 an example for that can be found in this node module I wrote: https://github.com/dutchanddutch/node-sd-daemon Nov 19 18:09:08 it also lets you inform systemd about the status of your service Nov 19 18:11:37 neat Nov 19 18:11:58 i had started to use PM2 as the process manager, and its pretty good... but... its painfully slow on the BBB Nov 19 18:12:50 oh, I just noticed I still need to document the support for journal logging I added to node-sd-daemon Nov 19 18:12:59 and I should probably get it onto npm Nov 19 18:13:07 get to work! Nov 19 18:13:19 yeah yeah, I'll put it on the backlog Nov 19 18:13:28 a true developer :) Nov 19 18:14:34 I'd first want to fix its dependency on a custom git branch of another module Nov 19 18:34:53 I've tried to make some categories of interest for next year's GSoC. Feedback requested: https://elinux.org/BeagleBoard/GSoC/Ideas Nov 19 18:40:15 wish i was a student :) Nov 19 18:40:28 kids these days, no clue how easy they have it! Nov 19 19:03:16 Dear zmatt, I am back to trying to implement PWM on x15 board. Thank you for your help. But due to the lack of experience, I am having the hard time following the suggested steps. My understanding that to implement PWM I need to 1) Install proper uboot; 2) Modify the existing device tree; 3) write a C-program to do it programmatically is this correct? Though I am still puzzled with the need for steps 1 and 2, I would like to follo Nov 19 19:04:56 your message got cut off at "I would like to follo" Nov 19 19:05:35 but in general, on ARM-based embedded systems you need Device Tree declarations to declare what your hardware setup looks like Nov 19 19:06:14 and configure the pins of the SoC Nov 19 19:07:05 on the beagleboard-x15 the pin configuration is preferably done in u-boot instead because doing so at a later time is known to be able to cause glitches during reconfiguration Nov 19 19:08:31 for your pwm pin that might not matter too much, and doing it in DT may be fine. however, u-boot still contains default pin configuration for the expansion headers, intended for the AM572x EVM LCD board, and if you want to use the expansion headers for a different purpose, it is probably a good idea to disable that at the very least Nov 19 19:10:07 once the pwm controller is enabled in DT, it will show up in sysfs and you can use it Nov 19 19:10:17 your help is appreciated. Step#1: https://github.com/RobertCNelson/dtb-rebuilder do I need to pickup 4.14.x? Nov 19 19:10:33 4.14-ti most likely Nov 19 19:10:47 assuming you're using a 4.14-ti series kernel Nov 19 19:11:45 uname -> Linux BeagleBoard-X15 4.14.71-ti-r80 Nov 19 19:11:56 that would be a 4.14-ti kernel yes Nov 19 19:14:11 Thanks! Give me a few minutes to look at this... Nov 19 19:15:14 hello everyone Nov 19 19:15:46 Do I need to compile it on x15? Nov 19 19:16:31 dreamhiker: doesn't matter where you compile it Nov 19 19:17:38 you'll need dtc of course (apt-get install device-tree-compiler) Nov 19 19:25:28 jrich523: :-) Nov 19 19:28:19 back in my day we had to break old electronics to have anything to play with lol Nov 19 19:46:12 I have built it on x15. I can see a few *beaglle-x15.dtb files. Which one is mine? Or maybe I just need to do "make Install"? Nov 19 19:49:13 depends on board revision Nov 19 19:50:25 you can see which one you're using: perl -pe 's/\0/\n/g' /proc/device-tree/model Nov 19 19:51:09 the three am57xx-beagle-x15*.dtb files all have a different 'model' property Nov 19 20:00:18 TI AM5728 BeagleBoard-X15 rev C: Shall I use am57xx-beagle-x15-revc.dtb? Nov 19 20:08:50 shall I just copy am57xx-beagle-x15-revc.dtb to /boot/dtbs/4.14.71-ti-r80 and to reboot the board? Nov 19 20:15:02 well I don't really recommend modifying/replacing any of the standard dtbs. the example I gave (https://pastebin.com/MagZyG75) was to create a dtb with a custom name, which you can then place in /boot/dtbs (no need to put it in a kernel-version-specific subdir) and use by configuring "dtb=am57xx-beagle-x15-custom.dtb" in /boot/uEnv.txt Nov 19 20:15:20 that way it also survives kernel upgrades Nov 19 20:23:39 hello Nov 19 20:23:53 is someone there? Nov 19 20:24:02 probably Nov 19 20:24:25 nope, all digital illusions Nov 19 20:24:29 but that isnt all the important, unless you're looking to just drink coffee and chat Nov 19 20:24:45 haha ok... im wondering if any of you had measure the power consumption of the beaglebone Nov 19 20:25:04 anywhere between 0 and 2A Nov 19 20:25:27 i want to use it as an OBC in a cubesat project Nov 19 20:28:34 depends on which functions are used, which kernel is used, cpu frequency (which may depend on cpu load if the ondemand governor is used), etc Nov 19 20:32:36 Do you know which kernel is the best for low power applications? Nov 19 20:33:11 probably 4.9-ti or 4.14-ti series Nov 19 20:33:57 I haven't really given much attention to power management yet, other people here might have a more informed opinion Nov 19 20:35:02 Thank you anyways Nov 19 20:42:17 Sorry, I got confused. Should I take am57xx-beagle-x15-revc.dtb (that I just compiled) or I need to take your sample, name it am57xx-beagle-x15-custom.dts and to compile it? Or maybe smth else? Nov 19 20:57:28 dreamhiker: I mean... if you just recompiled the unmodified dtb, what would that accomplish? Nov 19 20:58:38 and creating a modified dtb without changing its name is unwise because it can create confusion, and because your changes would be lost if you update the kernel Nov 19 21:00:51 dreamhiker: and yes my suggestion was to take the contents of my pastebin, save it as am57xx-beagle-x15-custom.dts (in src/arm in the dtb-rebuilder project) and compile it to create am57xx-beagle-x15-custom.dtb Nov 19 21:00:53 Are you saying that https://github.com/RobertCNelson/dtb-rebuilder is the source files for the dtb's that I currently have? Nov 19 21:00:57 yes Nov 19 21:01:11 Cool! Thanks! Nov 19 21:01:34 note that my example #includes am57xx-beagle-x15-revc.dts and then just adds some stuff on top of that Nov 19 21:02:31 (if you had a different board revision, you'd change that #include) Nov 19 21:03:52 So are you saying that I basically need to replace am57xx-beagle-x15-revc.dts with your sample, that let us call am57xx-beagle-x15-custom.dts? Nov 19 21:04:03 what? no? Nov 19 21:04:10 that's not at all what I said Nov 19 21:04:19 I do understand keeping the original and the modified separately Nov 19 21:04:54 my example effectively embeds the contents of am57xx-beagle-x15-revc.dts by using the #include directive (which works the same as in C) Nov 19 21:07:56 oops, sorry. I could have been more attentive :) Nov 19 21:26:10 zmatt, Thanks! I have created am57xx-beagle-x15-custom.dtb and added the line to uEnv.txt. Do I need to reboot the board? Nov 19 21:27:01 of course Nov 19 21:34:33 Can i check that uboot is using now the custom dt? Nov 19 21:36:43 if you have a serial console cable you can check the output from u-boot. otherwise just check for its effects, e.g. a device showing up in /sys/class/pwm Nov 19 21:37:03 there's also a filesystem representation of DT at /proc/device-tree Nov 19 21:47:57 Good idea! I will install the serial cable. Nov 19 21:49:22 Is there an easy way to check it in /proc/device-tree Nov 19 21:50:00 checking /sys/class/pwm is easier in this case. there should be a single device there Nov 19 21:51:15 it should also show up as /sys/bus/platform/devices/4843e100.ecap or something like that Nov 19 22:02:17 dreamhiker: .. so? did you find the device? Nov 19 22:22:50 pwmchip0 -> ../../devices/platform/44000000.ocp/4843e000.epwmss/4843e100.ecap/pwm/pwmchip0 Nov 19 22:23:47 well, there ya go Nov 19 22:26:27 Thanks! Nov 19 22:28:42 Now, my understanding that I can setup PWM by writing manually some files in pwmchip0, as well as programmatically, using some different mechanisms. Is this correct? Nov 19 22:28:50 then, inside that directory, write 0 to the export attribute to export its first (and only) pwm channel Nov 19 22:29:02 a subdirectory will appear (pwm-0:0) Nov 19 22:30:17 Which header pins can I see PWM? I need to ask somebody to wire them. Nov 19 22:30:19 inside that, write the desired period (1/frequency) in nanoseconds to the 'period' attribute Nov 19 22:30:33 the pin number is in a comment in the dts Nov 19 22:32:30 hmm, an udev rule could make this nicer to deal with Nov 19 22:36:50 Sorry, it is my next chore to understand your custom.dts file (and hopefully you will help me). Nov 19 22:37:43 For now, would be great if you just can tell me where to find those pins :) Nov 19 22:38:53 really, it was too much effort to check which of the four comments in this 22-line file contained the pin number, I have to look that up for you? Nov 19 22:38:59 there you go, line 19: // P18.12 Nov 19 22:40:29 Thanks! Nov 19 23:33:36 dreamhiker: if you create a .rules file in /etc/udev/rules.d/ with this contents: https://pastebin.com/raw/AETTNxnp then the pwm channel will get automatically exported at boot, and a symlink to it will be created at /dev/pwm/ecap0 Nov 20 00:53:54 Haven't succedded with OpenGL yet... so can the opengl stuff also only be used without X? and also eQEP isn't working smoothly for me... maybe due to circuit design. Nov 20 01:00:04 suprothunderbolt: correct, the current sgx drivers do not support x11, only direct output (drm or gbm backends) and wayland Nov 20 01:00:21 also, opengl es1/es2, not opengl Nov 20 01:00:42 I recall seeing you say something about eqep but I don't remember quite what Nov 20 01:01:38 you had problems at high rates or something? Nov 20 01:02:00 oh wayland would be fine Nov 20 01:02:37 zmatt: the eQEP problem is that whne I turn the knobs swiftly they output large negative numbers Nov 20 01:02:53 using the standard driver setup through the device tree Nov 20 01:03:17 if i move the knobs slowly everything is fine Nov 20 01:03:38 could easily also be a driver issue, iirc the driver is a bit iffy Nov 20 01:04:32 circuit wise I've just got the pins connected up directly to encoder, no filtering caps on them or anything. Nov 20 01:05:28 most of the example circuits I've seen don't have any but the manufacturer's data sheets does in one example. Nov 20 01:06:07 hmm, if it were signal integrity issues then they'd probably already show up at low rates Nov 20 01:06:28 any idea how fast (in steps/second) you're rotating in your tests? Nov 20 01:08:36 you could try py-uio's eqep test to see what the peripheral itself is saying Nov 20 01:10:07 are you using it in absolute or relative mode? Nov 20 01:10:51 I'm using absolute mode Nov 20 01:12:05 and I'm not spinning it very fast. It's a control knob and and turning it slower than i turn the knobs on my oscilloscope. Nov 20 01:12:24 how many steps per full rotation? Nov 20 01:12:53 one thing about the driver, i'm not sure how to specify anything about the steps per rotation, if needed. It's a 24 step encoder Nov 20 01:13:12 it has no way to specify that Nov 20 01:14:27 py-uio does include an eqep example for such an application... I wrote it to test eqep with a knob we use in a device Nov 20 01:15:23 in the end we ended up using gpios and software decoding instead since the rate is low anyway and eqep can't give a notification on every position change (it has to be polled instead) Nov 20 01:15:26 okay, I'll check it out and see if works better. Does it just use the A and B pins or an index / strobe as well? Nov 20 01:15:35 just A/B Nov 20 01:15:37 oh okay Nov 20 01:15:50 I've never experimented with index/strobe Nov 20 01:16:01 I'm just using A / B Nov 20 01:16:29 eqep has quite a bit of functionality I've never tried Nov 20 01:18:30 https://github.com/mvduin/py-uio/blob/master/other-examples/eqep-test.py it's a really simple test that just uses position-tracking, along with a 2 Hz interval timer to periodically dump the current value, and it dumps all irqs Nov 20 01:19:18 haven't tested it recently but it should still work Nov 20 01:21:20 it needs an overlay along the lines of https://github.com/mvduin/py-uio/blob/master/dts/pwmss2.dtsi (although you can use a different pwmss instance of course, and can trim it down to just the eqep instance and its A/B pins) Nov 20 01:22:44 so, looking at the eqep driver, it isn't strange per se that negative values show up since it's a 32-bit counter which is printed using %d format Nov 20 01:23:09 it is however strange to get a large value in the first place by hand-turning a knob with so few steps per rotation Nov 20 01:24:23 so maybe you do have noise problems (not unusual for anything that's manually agitated) Nov 20 01:24:57 in which case my py-uio example would probably also spit out "quadrature decoder error" messages Nov 20 01:26:09 at least if I get that with both I'll have a clear idea it's a hardware fault. Nov 20 01:27:12 is there a good guide for installing the opengl support? I'm testing QtQuick for my UI which likes opengl. Nov 20 01:27:15 ES. Nov 20 01:27:36 I've posted an outline in the forum a while back Nov 20 01:28:12 you've installed the kernel modules? (rcn has debian packages for those iirc) Nov 20 01:28:30 or, maybe first step: which kernel are you using Nov 20 01:29:13 4.14 bone, built from the github magic patch stuff Nov 20 01:29:58 oh, ok, so you also need to manually build the modules, and probably even apply the necessary kernel patches Nov 20 01:30:09 ehh, hold on, lemme dig Nov 20 01:31:03 https://github.com/dutchanddutch/bb-kernel/commit/088d020a935c Nov 20 01:31:11 I'm having troubles getting USB Gadget to work with a combination of Windows and an X15. Nov 20 01:31:46 X15 and Linux works. BBBlack and Windows works. But, BB X15 and Windows doesn't. Nov 20 01:32:02 jimfred: o.O Nov 20 01:32:06 that sounds weirdly specific Nov 20 01:32:30 suprothunderbolt: https://groups.google.com/d/msg/beagleboard/BEipC_SFjaE/zieEuZlfDwAJ Nov 20 01:33:14 suprothunderbolt: the first link in that post should be replaced by the commit linked above, but other than that the rest of the post still applies with a bit of luck Nov 20 01:34:02 suprothunderbolt: for the userspace parts you could also try these packages I built an age ago: https://liktaanjeneus.nl/sgx/ Nov 20 01:35:09 oh nice from the perl script there I guess you were using it with QT as well ? Nov 20 01:36:11 yeah, the perl script does a binary patch to make qt5's eglfs layer use 16-bit pixel format (since they were too lazy to actually probe the drm device to check which pixel formats are supported) and is not applicable if you're using 24-bit color Nov 20 01:36:39 ah yeah I think I remember you mentioning that when I asked about colour formats Nov 20 01:41:54 slighty more opinion based question, but I'm contemplating if I should use the BBG, Pi compute module 3 or an Allwinner A64 based board. The BBG is the most expensive and slowest but has the PRUs. Nov 20 01:42:41 though I'm not actually using them for anything. I am attempting to use eQEP Nov 20 01:43:10 it also uses a well-documented industrial SoC which you can actually buy in low quantity if you ever want to move to a custom pcb instead of using a prefab module Nov 20 01:43:37 yeah, I can get the A64 is small quantities and it's easier for me to build Nov 20 01:43:45 it's already got a full board in KiCad. Nov 20 01:43:51 (whereas the rpi uses a basically undocumented SoC that doesn't even have a product page on the manufacturer's website, let alone being available) Nov 20 01:43:58 yeah I don't know anything about the allwinner one Nov 20 01:44:03 yeah, that's my main fear Nov 20 01:44:21 that if they don't have supply of the CM3 I'm really in trouble. Nov 20 01:44:30 But there's a lot of units at all the major distributors Nov 20 01:44:45 and i don't need 10000 of them or anything like that Nov 20 01:45:50 the lack of proper docs and overall behaviour of broadcom is enough for me to want to have little to do with the rpi Nov 20 01:46:01 yeah Nov 20 01:46:11 like, there isn't even an official errata doc, even though the SoC obviously has plenty of them Nov 20 01:46:23 TI seem to be a lot easier to deal with Nov 20 01:46:42 (e.g. i2c transfers getting corrupted if the i2c device uses clock stretching) Nov 20 01:46:46 though it's annoying that the module the pocket beagle uses costs more than the pocket beagle... Nov 20 01:46:52 oh interesting. Nov 20 01:47:03 I don't need clock stretching luckly Nov 20 01:47:22 Have you used the allwinner a64 at all? Nov 20 01:47:36 I have no direct experience with any allwinner SoC Nov 20 01:48:07 how's the kernel support? ;P Nov 20 01:48:19 Really good. Nov 20 01:48:51 Olimex has a nice board that is designed with KiCad Nov 20 01:49:10 well that's a plus. I know kernel support can be a concern with some of this chinese stuff Nov 20 01:49:50 Olimex also have long term support for them which is the other challenge with fast moving chips Nov 20 01:50:57 yeah, the A64 is a "Mobile Application Processor", which sounds to me like it there's a decent risk of not having a very long lifetime Nov 20 01:51:27 yeah. But amazing cost / performance Nov 20 01:52:37 and of course the documentation available is not downloadable from allwinner but found on a random wiki with "CONFIDENTIAL" printed across every page :P Nov 20 01:52:48 hah yeah :D Nov 20 01:53:53 wiki lists GPL violations of course Nov 20 01:53:57 and you can get boards for $15... Nov 20 01:54:37 and the olimex board being designed with Kicad is a pretty big advantage for me as I don't have $7000 to spend on OrCad. Nov 20 01:55:15 some recent beaglebone variants based on the osd335x are designed in eagle I think? Nov 20 01:55:53 ... Nov 20 01:56:11 Is anyone using the dang it. Brb. Nov 20 01:56:25 yeah that module costs more than a pocket beagle... Nov 20 01:56:34 or a BBB... :( Nov 20 01:56:40 in low quantity? that sucks Nov 20 01:56:58 in quite big quanity... I had a chat with them Nov 20 01:57:12 their "deals" still are really pricey Nov 20 01:58:12 https://www2.mouser.com/search/refine.aspx?Ntk=P_MarCom&Ntt=139217159 Nov 20 01:58:27 it's cheaper to buy the BBG... Nov 20 01:58:46 and the pocket beagle, which has that module on it is a lot cheaper. Nov 20 01:59:47 direct the MOQ is 1200 units... Nov 20 02:00:53 https://eu.mouser.com/Search/Refine.aspx?Keyword=osd335&Ns=Pricing%7c0&FS=True Nov 20 02:00:56 yeah, it is kinda weird Nov 20 02:02:26 the A64 is $6 for the cpu and pmu at quantities of 1. :) Nov 20 02:03:54 and has a whole lot more performance. The CM3 is $30. Total cost for the A64 with the board / ram / flash is about $19 assembled in quantites of 100. Nov 20 02:04:40 ... Nov 20 02:04:59 Okay! Is anyone using the LoadCape yet? Nov 20 02:06:07 I think I'll attempt to get the opengl stuff working on this platform first before running to anything else! :D Nov 20 02:06:26 suprothunderbolt: if being cheap is more important than being open, and performance for media applications is more important than low-latency i/o capabilities, then go for it :) Nov 20 02:06:43 mmm... I think the A64 in some ways is more open Nov 20 02:07:02 given that I can actually change the board easily with an open tool Nov 20 02:07:30 for the BBG / BBB I need orcad. I'd much prefer KiCad. Nov 20 02:07:56 ehh, the availability of a reference design in your preferred schematic/pcb editor isn't really what I meant with openness Nov 20 02:08:30 I mean, having one in kicad would be nice. I don't know if kicad was up to the task back when the beaglebone was originally designed though Nov 20 02:08:38 true Nov 20 02:09:00 annoyingly everything seems to have close source blobs for GPU Nov 20 02:09:39 yup Nov 20 02:12:21 another thing is that I wouldn't really want to support allwinner simply because of their behaviour... (I'm reading a bit on http://linux-sunxi.org/A64 ) Nov 20 02:13:14 does allwinner contribute at all to mainline support, or is that entirely done by volunteers? Nov 20 02:14:02 seems like they have improved that stuff a bit recently but have done some sketchy stuff Nov 20 02:15:30 $ grep --count allwinnertech.com MAINTAINERS Nov 20 02:15:31 0 Nov 20 02:15:41 $ grep --count ti.com MAINTAINERS Nov 20 02:15:42 32 Nov 20 02:16:12 what's broadcom score? Nov 20 02:16:28 though TI has a lot of other products too, like it's codecs etc Nov 20 02:16:55 broadcom's score is surprisingly good Nov 20 02:16:59 yeah true Nov 20 02:17:12 interesting! I kind of put them in the evil bucket Nov 20 02:17:47 lemme see if I can make a more useful tally Nov 20 02:17:55 hah :) Nov 20 02:19:24 Wait! Nov 20 02:24:27 for? Nov 20 02:24:57 https://pastebin.com/uXKuCHS2 Nov 20 02:25:14 I think there is an odd error when updating my machine, the BBB. Nov 20 02:25:39 This is not the first time. I have received this error many times. Nov 20 02:26:42 probably not harmful. can you pastebin your /etc/initramfs-tools/initramfs.conf file? Nov 20 02:26:57 Sure. Nov 20 02:26:59 Please hold. Nov 20 02:28:43 https://pastebin.com/vkdTTDD9 Nov 20 02:28:46 brb Nov 20 02:29:05 okay, so the error is lying Nov 20 02:29:20 a bit of googling suggests it's an upstream bug, I wouldn't worry about it Nov 20 02:29:24 you're not using cryptsetup anyway Nov 20 02:30:23 Right-o. Nov 20 02:30:35 Otay...good enough for me. Nov 20 02:31:42 Um? Nov 20 02:32:01 Has anyone figured out how to wire up this LoadCape? Nov 20 02:32:24 I have heard many routes to take on lead connections. I have tried them all. None work. LEDs work! Nov 20 02:32:25 I don't think I've seen a reaction from jkridner[m] yet Nov 20 02:32:30 Okay. No issue. Nov 20 02:32:46 Is he still alive? Nov 20 02:34:25 Or... Nov 20 02:34:29 Is he in Japan? Nov 20 02:34:58 I have no idea where he is or what he does, but I think I've seen him active pretty recently Nov 20 02:35:23 Dang. I missed it. Otay then, I guess this little bugger is staying patient. Nov 20 02:36:16 jkridner[m]: just in case, another ping: loadcape has no schematics pdf and overall insufficient info available on how to use it Nov 20 02:36:23 also no dt overlay Nov 20 02:36:42 I was dealin' w/ the mfg. on this idea. They have ideas. Nov 20 02:37:01 I tried the ideas. It somehow erased my memory on my board. SD memory. Nov 20 02:37:21 eh? Nov 20 02:37:24 Serious! Nov 20 02:37:29 They have a website and all. Nov 20 02:37:35 ghi! Nov 20 02:37:46 link to the info you have? Nov 20 02:37:52 Okay. Please hold. Nov 20 02:38:08 zmatt: btw, as always, thanks for the help :) Nov 20 02:38:30 Oh and in no way am I responsible for what you see! https://ghielectronics.discoursehosting.net/t/re-loadcape-and-trying-seth/21797. Nov 20 02:39:01 ... Nov 20 02:40:18 Sorry for the childlike drawings. Nov 20 02:41:14 okay it does work like I assumed, but additionally it needs the supply voltage on the VIN pin for some reason Nov 20 02:41:37 Right-o. That is what they said but do all my positive leads go into VIN? Nov 20 02:42:00 Or...should all my positive leads go to the battery supply? Nov 20 02:42:16 "same difference?" Nov 20 02:42:31 that's the same thing since you're connecting those together. connecting them at the battery probably makes more sense than cramming them into that VIN terminal Nov 20 02:43:04 Okay. I can make a makeshift lead from three w/ quick pull connector. Nov 20 02:43:20 ... Nov 20 02:43:21 But... Nov 20 02:43:42 but what you said after that sounds really really really weird Nov 20 02:43:52 Wait...what part? Nov 20 02:44:08 makeshift or one out of three? Nov 20 02:44:47 if having to reflash the sd card was genuinely the result of the hardware connections rather than doing something dumb in the terminal, then I wouldn't touch that board at all until schematics are available Nov 20 02:44:59 Fine. Nov 20 02:45:08 That sounds really boring but I will listen. Nov 20 02:45:16 Dang! Nov 20 02:45:24 because that would be extremely disconcerting Nov 20 02:45:47 What...board malfunction or broken BBB? Nov 20 02:46:47 the hardware connection affecting your system in any way, let alone damage your system to the point of having to reflash Nov 20 02:47:06 I know. I lost some "nice" stuff (I think). Nov 20 02:47:17 Who knows? oh well, I can always rebuild. Nov 20 02:47:46 hey zmatt: I am going to leave you be for now. I saw you were discussing something w/ someone earlier. Nov 20 02:48:01 I was not trying to get in the way. Nov 20 02:48:27 less interruptions and more processing! Nov 20 02:51:56 Hello, I swung through a few nights ago and received a few hints that helped a lot. I have my beagle bone blue working well, and have one more question. Nov 20 02:53:11 When I installed some updates, I was asked to run an app at startup. I chose blinky to initiate hardware services. Does anyone know how I can change that and stop the service. Nov 20 02:54:05 asked to run an app at startup? o.O Nov 20 02:54:16 libroboticcape Nov 20 02:54:16 ? Nov 20 02:54:40 Yes, I think that was it libroboscape Nov 20 02:55:11 I had chosen none. Nov 20 02:55:39 I am sure if you go into your /lib/ dir and search for that libroboscape, you can find it. Nov 20 02:55:41 Let me check. Nov 20 02:55:41 I should have, well, you live, you learn. Nov 20 02:55:56 ... why on earth does libroboticscape ask for a service to run? o.O Nov 20 02:56:08 set_: config files live in /etc/ Nov 20 02:56:15 Oops! Nov 20 02:56:38 MRRSquared: maybe try grep -rs blinky /etc Nov 20 02:57:04 to try to locate the relevant config file Nov 20 02:57:08 I think it is under robotcontrol. Nov 20 02:57:13 in /etc. Nov 20 02:57:15 Well, my other issues were with getting libroboscape to access hardware, so I thought it would be nice to see it working. Now I just want it to stop. :) Nov 20 02:57:23 I will check. Thanks. Nov 20 02:57:41 It is not on mine but I did not install it. Nov 20 02:58:25 https://guitar.ucsd.edu/pyctrl/html/user_guide.html#introduction might be something worth looking into. Lots of docs. Nov 20 02:58:32 you should also be able to access the prompt again using: Nov 20 02:58:34 sudo dpkg-reconfigure librobotcontrol Nov 20 02:59:24 it sounds like it sets up a symlink at /etc/robotcontrol/link_to_startup_program ? Nov 20 02:59:44 http://www.strawsondesign.com/docs/librobotcontrol/runonboot.html Nov 20 02:59:51 Aw! **** ENDING LOGGING AT Tue Nov 20 02:59:58 2018