**** BEGIN LOGGING AT Tue Jul 10 03:00:02 2018 Jul 10 03:50:38 Hello Jul 10 03:52:43 I have plugged in a beagleboard-blue into the USB port of my compter but I do not see the external device on my computer. what can be wrong? Jul 10 03:57:29 w Jul 10 05:32:27 I was upgrading the packages on my beagleboard blue and it stopped halfway through. now the computer doesn't even recognize the board when I plug it in. Has anybody seen this issue before? Jul 10 07:58:31 I am getting acquainted with BBB Jul 10 07:58:57 I wish to communicate between 2 BBB over Ethernet by Message Passing Jul 10 07:59:46 How to configure the node descriptors to send and receive messages between the boards Jul 10 08:00:37 [13:25] I am getting acquainted with BBB [13:25] I wish to communicate between 2 BBB over Ethernet by Message Passing [13:26] How to configure the node descriptors to send and receive messages between the boards Jul 10 13:44:55 has anyone managed to build and install SGX drivers as mentioned here? https://elinux.org/BeagleBoardDebian#SGX_Drivers Jul 10 13:45:33 weston-wayland and qt on beaglebone black? Jul 10 13:47:33 it seems likely that there have been people who managed that Jul 10 13:47:42 the instructions there are for the old drivers though Jul 10 13:48:00 (which won't work on current -ti kernels) Jul 10 13:53:42 cant Qt do direct GLES ? i wonder if it's worth having the extra wayland layer there Jul 10 13:53:58 it definitely can Jul 10 13:54:36 wayland/weston sounds wasteful on so little ram Jul 10 13:55:01 yeah we don't use wayland, just qt5 with its eglfs backend Jul 10 15:03:25 my computer does not recognize my beaglebone blue when I plug it in. Does anybody know how to solve this? Jul 10 15:14:19 kim_: do any LEDs light up? Jul 10 15:15:08 yes Jul 10 15:15:39 kim_: which operating system are you running on your host machine? Jul 10 15:15:51 windows 10 Jul 10 15:16:18 i was running apt-get upgrade when it stopped halfway through and lost connection Jul 10 15:16:37 ah, well… Jul 10 15:16:46 do you have a usb-to-serial adapter? Jul 10 15:17:47 i can try to get one. what are you thinking? Jul 10 15:18:41 if you are working on embedded devices, then having a USB to 3.3/5V UART adapter is vital Jul 10 15:19:11 you could check though if the virtual serial port of the device shows up Jul 10 15:22:26 i see Jul 10 15:23:02 i tried a different microUSB cable and that did not change anything Jul 10 15:23:10 hi Jul 10 15:23:26 has anyone ever tried to run any sorts of optimization program on BBBL Jul 10 15:24:14 prashin: I'm sure someone has. hint: you should be far more specific on *what* kind of optimization Jul 10 15:24:50 kim_: changing the USB cable will not change anything if the operating system is now in an inconsistent state. Jul 10 15:24:55 so I designed an MPC controller for my hexacopter Jul 10 15:25:05 in Matlab Jul 10 15:25:36 that requires quadratic optimization, and now I intend to port to BBBL Jul 10 15:26:18 hence wanted to know if any library is available online , or anyone has tried to perform optimization and what was the performance Jul 10 15:26:26 hope that helps Jul 10 15:33:50 tbr: could you explain a bit more what I would be seeing ideally through the UART? Jul 10 15:34:35 kim_: you should get a login prompt (you might need to press [enter] to make it appear) Jul 10 15:34:42 is there any way to do a hard reset on the entire system without being able to see it show up on my computer? Jul 10 16:52:07 what do you mean? Jul 10 17:04:00 kim_: did any errors occur during the apt-get upgrade? Jul 10 17:04:10 like, "disk is full" ? :P Jul 10 17:04:29 that's a scenario wherein apt is known to be able to a giant mess Jul 10 17:06:27 assuming that's the reason it "stopped halfway through", your options are: 1. boot from SD card and attempt to recover the mess on eMMC (I'd file this under "advanced debian system admin only"), or just backup files of importance and reflash eMMC Jul 10 17:07:08 that would be the "hard reset" you're looking for Jul 10 17:08:55 it doesn't matter what state the system on a beaglebone is in, you can always reflash it Jul 10 17:12:53 yes, i got an error. can't remember what it was exactly though Jul 10 17:13:30 to reflash the eMMC i would still need to use an SD card right? Jul 10 17:13:35 in the future you may want to capture as much output as possible to file if you want useful diagnosis or advice ;) Jul 10 17:13:46 yeah Jul 10 17:13:56 or, well, there are ways without SD card but they're a bit tricky Jul 10 17:14:44 gotcha Jul 10 17:15:50 for flashing, do i just connect through microUSB and hold the SD button and power button at the same time? Jul 10 17:16:46 sorry, i meant SD and RST button Jul 10 17:17:15 you just boot from a flasher card, which you can make from a normal ("standalone") sd card by uncommenting one line in /boot/uEnv.txt on the sd card. since you have a windows host, the easy way is to first boot from sd, log in via ssh and edit /boot/uEnv.txt, and then reboot Jul 10 17:17:23 you normally don't need to hold any buttons Jul 10 17:18:28 ROM will load u-boot from eMMC by default, but u-boot will try to boot linux from sd card before it tries to boot from eMMC Jul 10 17:20:01 holding down the SD button during power-on (not mere reset) will make ROM load u-boot from SD card rather than eMMC, which can be useful if u-boot on eMMC got corrupted somehow, or if you're trying to boot an SD card containing a wildly different system which isn't compatible with the u-boot on eMMC Jul 10 17:21:57 the SD button is sampled by hardware when the system powers up, and can (and should) be released once the power led turns on Jul 10 17:24:28 okay that makes sense! Jul 10 17:24:35 i will try it. thank you Jul 10 17:38:40 (also, in case prashin returns, or anyone with a similar question: assuming with "quadratic optimization" he means least-squares, that's purely linear algebra, hence should be easy enough to find optimized code for it. do be aware that single-precision floats can be orders of magnitude faster than double-precision floats on the cortex-a8. even better performance (and potentially better precision) ... Jul 10 17:38:46 ...may be obtained if the computation can be done with integers, but that requires careful scaling) Jul 10 17:43:48 for single precision, the neon unit of the cortex-a8 can do 2 flops/cycle, hence 2 Gflops in principle if you manage to keep the pipeline running smoothly. use of the neon unit requires either compiling with -O3 -ffast-math or use of the neon intrinsics in Jul 10 17:46:06 non-neon floating point instructions (including all double-precision ops) are terrible: 9-10 cycles for add/subtract, 10-12 (single-prec) or 11-17 cycles (double-prec) for multiply, additional penalties for denormals Jul 10 17:50:37 (a "flush denormals to zero" mode can be enabled to prevent those penalties. the neon unit always uses this mode, which is one of the reasons why -ffast-math is needed) Jul 10 21:59:53 helo Jul 10 22:14:47 250 OK Jul 10 22:15:04 elho zmatt Jul 10 22:15:43 no, the esmtp greeting is EHLO, nog ELHO :P Jul 10 22:15:47 *not Jul 10 22:15:52 turns out the ethernet device that mysteriously appeared was the linux usb gadget driver. wired ethernet seems dead. Jul 10 22:16:11 ok that proves i am human and not a process. Jul 10 22:16:38 nah you're probably just buggy software Jul 10 22:16:40 ;) Jul 10 22:17:27 been a while since i debugged any mail servers. Jul 10 22:19:08 anyway, sorry to hear about your laptop's continued ethernet problems. I still don't see any plausible way for the beaglebone to be responsible in any way Jul 10 22:21:01 i Jul 10 22:22:24 i've put that on the back burner for now. the current problem is trying to get the devuan image for the bbb to start up the usb interface automatically the way the debian image does. Jul 10 22:23:15 ah, you're one of *those* people ;) Jul 10 22:23:45 its a workaround for not being able to connect via wired ethernet. Jul 10 22:23:53 cat 5/6 that is... Jul 10 22:24:13 making the usb interfaces work is pretty trivial using systemd-networkd Jul 10 22:24:51 i will probably end up copying whatever they did on the debian images. Jul 10 22:24:58 note that unless you particularly need the usb to work multiplatform (linux/mac/windows), it's much simpler to just modprobe g_ether instead of using the complicated setup used on current debian images Jul 10 22:25:36 as i said the usb networking is to make up for the lack of a working 10/100 port Jul 10 22:25:51 all it requires is one line in your /etc/modules and whatever setup is needed in your network manager Jul 10 22:26:08 it's either that or cobble a serial cable and pull out the old vt100 Jul 10 22:26:49 I'd recommend xterm over a vt100 Jul 10 22:27:16 also, especially when using ncurses-heavy programs like vim, you may want to bump up the serial port speed to 460800 Jul 10 22:27:16 yeah, but i'm one of _those_ people Jul 10 22:28:07 it turns out that this makes the console 4x faster compared to the default 115200 (who'd have thunk it) Jul 10 22:31:28 in any case i prefer the usb networking approach Jul 10 22:32:05 like I said, just get g_ether loaded to make the link show up, and then get it configured just like you would with an ethernet interface Jul 10 22:32:29 as in edit /etc/modules? Jul 10 22:32:47 yeah you can add g_ether to it to get it automatically loaded Jul 10 22:33:20 you may want to try it all manually once before doing the setup to make it happen automatically Jul 10 22:33:36 (I do hope you have that serial cable, otherwise debugging this is going to be... interesting) Jul 10 22:33:53 ok... i can do that by mounting the sd card on the laptop, but will be booting the bbb "blind" so it will be difficult to debug Jul 10 22:34:18 guess we both had the same thought... Jul 10 22:35:36 putting g_ether in /etc/modules will be easy enough and i've already edited /etc/network/interfaces Jul 10 22:36:20 it seems like that would do the job or did i forget something? Jul 10 22:36:33 should be fine Jul 10 22:40:50 it's usually named usb0 rather than eth0 Jul 10 22:41:46 yeah it is, although the actual name can vary depending on things like udev rules Jul 10 22:42:19 yes usb0 on my bbb, but en{mac-address} on the laptop. Jul 10 22:42:53 it's a different type of interface on the laptop-side. Jul 10 22:43:06 hence the different name Jul 10 22:43:17 that too Jul 10 22:43:18 oh wait that was usb0 on the beaglebone under debian, now its booting devuan, so it could be different Jul 10 22:43:27 it could be yes Jul 10 22:44:10 if devuan has gone out of their way to rename usb interfaces ... otherwise i'm pretty sure usb0 is the default in the linux kernel Jul 10 22:44:32 no, I think the beaglebone images go out of their way to suppress the default renaming done by debian, but I'm not 100% sure Jul 10 22:45:09 like zmatt said it depends on the udev rules Jul 10 22:45:11 the question isn't what the kernel name is, the question is what name is assigned by the default udev rules Jul 10 22:45:49 given that udev is part of systemd on debian, but obviously not on devuan, I don't even have a guess about what will happen Jul 10 22:46:54 i've run many plain debian images, and it's almost always usb0 ... but i guess it wouldn't be surprising if there are custom udev rules on the beaglebone images Jul 10 22:47:06 hmm, maybe I'm confused with something else then Jul 10 22:47:18 or maybe it's like you said, only for the host side driver Jul 10 22:47:40 vagrantc note that i'm currently working with the devuan image Jul 10 22:48:08 yes, though, the debian versions were all usb0 in my experience as well. Jul 10 22:48:45 beaglebonius: i'd be surprised if devuan went out of their way to do anything in particular with the usb gadget ethernet naming conventions ... but i have little experience with devuan so maybe everything about it is suprising Jul 10 22:50:18 well on debian the en{mac-address} interface name comes from /lib/udev/rules.d/73-usb-net-by-mac.rules whose content is clearly systemd-flavored, so if this naming scheme also exists on devuan then they've replicated it with different rules Jul 10 22:51:26 or maybe not, I guess the rule could still work depending on the degree to which whatever project is maintaining a systemd-free udev is keeping compatibility with it Jul 10 22:53:06 ah, eudev says they're "tracking closely the systemd-udev developments" Jul 10 23:00:01 ok thanks zmatt. g_ether did the job. i am getting ping response over usb networking Jul 10 23:08:17 btw it is usb0 on both ends Jul 10 23:08:26 that's possible Jul 10 23:08:51 of course its possible becuase that is what is. Jul 10 23:09:44 but g_ether was the magic sauce i needed. thanks again. Jul 10 23:10:10 well, g_ether and a lucky guess about the interface name, that is.... Jul 10 23:12:05 takes a lot of years to build up that much luck Jul 10 23:14:05 i guess it could be characterized as an s/lucky/educated/ Jul 10 23:14:14 type of situation Jul 11 00:22:30 after enabling ip forwarding on box1 and doing "route add default gw box1" on box2, i cannot access internet from box2. did i miss something? Jul 11 00:23:04 output from "route" looks ok Jul 11 00:24:08 your internet modem/router has no reason to send packets meant for box2 to box1 Jul 11 00:24:18 your options are: Jul 11 00:24:58 1. bridge the links together, enable dhcp on box2 (or configure a static ip in the range of your main network) Jul 11 00:25:09 2. configure ip masquerading on box1 Jul 11 00:25:48 thanks ip masq is what i forgot Jul 11 00:26:01 3. add a static route to your internet router/modem to route packets meant for the box1-box2 network (be sure to use a different ip network) to box1 Jul 11 00:27:52 it will have to door number 2 since i don't have admin on the router Jul 11 00:56:11 I usually use bridging Jul 11 00:57:13 zmatt as i said i don't have admin on the router so door number 2 it is for me. Jul 11 00:57:45 what does that have to do with the router? Jul 11 01:05:47 Hey! Jul 11 01:06:06 Why the hell will my .service file not work? Please hold for pastebin! Jul 11 01:07:12 https://pastebin.com/Xz6knTne was found w/ journalctl -xe! Jul 11 01:07:16 ... Jul 11 01:07:20 Please send rations! Jul 11 01:07:45 brb...sorry. Jul 11 01:11:15 I think it is needing a another pastebin. Please hold. Jul 11 01:11:16 ... Jul 11 01:13:46 https://pastebin.com/tfq3G3uE is my simple .service file. I thought I needed PID instead of Type under [Service] but... Jul 11 01:16:05 I think, since it has to do w/ sockets and not multi-user.target, I may need to change the WantedBy= under [Intall]. Jul 11 01:16:10 What do you think? Jul 11 01:17:25 The files are .js and .html. Those two files combined work to make a LED on the BBBW turn on and off via application online at the address 192.168.7.2:8085. Jul 11 01:17:41 ... Jul 11 01:17:44 Rations? Jul 11 01:18:00 .... Jul 11 01:18:05 zmatt! Jul 11 01:18:20 I will give you time. I am on hold and I will keep trying. Jul 11 01:18:32 or anyone w/ grace! Jul 11 01:18:36 why would you think this would do anything other than not-work? Jul 11 01:18:45 you can't execute a html file Jul 11 01:18:49 Oh! Jul 11 01:18:58 Who told you that? Jul 11 01:19:09 Or...how did you find that out? Jul 11 01:19:34 So, I would just make my .js file executable and then expect it to work? Jul 11 01:19:39 no? Jul 11 01:19:43 Oh. Jul 11 01:19:58 Hmm. Jul 11 01:20:28 Oh. Jul 11 01:20:31 ! Jul 11 01:20:50 I know what I did incorrectly. Jul 11 01:21:00 .js files can be executable using nodejs, but that's very different from how a browser runs javascript code Jul 11 01:21:12 I am using nodejs. Jul 11 01:21:14 ! Jul 11 01:21:25 I forgot to add the shebang and PATH. Jul 11 01:21:27 then why is a html file involved? Jul 11 01:21:52 I was unaware of the html file not having to be executed via chmod. Jul 11 01:22:34 as usual, I'm too confused by what you're saying to even be able to give an indication of what you're doing wrong or how to fix it Jul 11 01:23:01 I thought, since I need to make files executable for a .service file, I could do this to make a file run on boot. Jul 11 01:23:02 ... Jul 11 01:23:06 But...there are two files. Jul 11 01:23:13 one = .js Jul 11 01:23:18 another = .html Jul 11 01:23:38 I was just unaware of not making the .html file executable. Jul 11 01:23:53 this has nothing to do with a service Jul 11 01:23:56 . Jul 11 01:24:00 Oh. Jul 11 01:24:08 just get it working manually Jul 11 01:24:15 I know. It works! Jul 11 01:24:20 I can make it work. Jul 11 01:24:20 then, whatever command you use to start it manually, put that in the service file Jul 11 01:24:27 Oh! Jul 11 01:24:33 (but use the full path) Jul 11 01:24:37 Right! Jul 11 01:24:52 also, assuming you're running it as normal user (debian) right now, put User=debian in the service file Jul 11 01:25:01 Okay. Jul 11 01:25:14 and you may need to set WorkingDirectory to whatever directory you're normally in when running the command Jul 11 01:26:08 I just used the cmd chmod a+x to make it an executable (the .js file). Jul 11 01:26:24 I know what I did wrong, though. Jul 11 01:26:33 The shebang. Jul 11 01:26:44 and file path for node Jul 11 01:27:15 zmatt: Oh? Jul 11 01:27:31 you don't need a shebang nor make the .js executable though, you can just put ExecStart=/usr/bin/node file.js Jul 11 01:27:41 also, why does your service have an Alias ? that's pretty weird Jul 11 01:27:52 oneshot is also weird Jul 11 01:28:02 Okay. Jul 11 01:28:07 I was trying new things. Jul 11 01:28:39 oneshot means it's not a service, it's just a command that runs and ends (e.g. something that does one-time setup) Jul 11 01:28:52 my file is in the /home/debian/ directory. Jul 11 01:28:59 Oh? Jul 11 01:29:01 then put WorkingDirectory=/home/debian Jul 11 01:29:05 I was unaware. Jul 11 01:29:07 Okay. Jul 11 01:29:19 the default type is usually fine Jul 11 01:29:23 WorkDirectory=/home/debian Jul 11 01:29:24 Okay. Jul 11 01:29:28 Working, not Work Jul 11 01:30:46 I usually made it executable, which is not what i am going to do this time, and under the heading in the .service file, ExecStart=, I pleace my PATH. Jul 11 01:30:50 w/ file name. Jul 11 01:31:06 pleace = placed. Jul 11 01:31:52 Oh. Jul 11 01:32:11 and WorkingDirectory=/home/debian Jul 11 01:32:14 you place the command you use to run your script, i.e. "node script.js", which in the service file needs to use the full path for the executable hence "/usr/bin/node script.js" Jul 11 01:32:15 Got it. Jul 11 01:32:33 and User=debian Jul 11 01:32:45 then you got an environment fairly similar to executing the command yourself Jul 11 01:33:00 Okay. That is easy to get. Thank you. **** ENDING LOGGING AT Wed Jul 11 03:00:02 2018