**** BEGIN LOGGING AT Sun Jun 10 03:00:05 2018 Jun 10 04:40:46 e Jun 10 04:44:12 m Jun 10 13:04:38 hello, I was install opensprinkler on my BBB, and I know it has it's own webserver, but I'm still getting the BeagleBone 101 webpage, any know how to turn it off? Jun 10 13:04:56 s/install/installing Jun 10 13:07:54 the service name is probably apache or something like that. try finding the exact service name with: systemctl | grep -i apache Jun 10 13:08:15 then stop and disable it with: sudo systemctl disable --now followed by the service name Jun 10 13:08:54 (if your systemd is old to support "disable --now", use separate "stop" and "disable" commands) Jun 10 13:11:14 I can see apache2 failed to load Jun 10 13:12:00 and stopping it doesn't change anything Jun 10 13:12:42 huh, maybe it's a different webserver? it's been aaaages since I looked at the stuff installed on the default/iot image Jun 10 13:18:01 you can get a list of all listening tcp ports along with the process that owns it using: sudo ss --tcp --listening -p -n Jun 10 13:21:55 if you're unsure to which service this process belongs, systemd-cgls shows a tree of all system units and the processes that belong to it Jun 10 13:22:34 once you know the service that is responsible, you can stop and disable it Jun 10 13:24:39 i installed vnc server in Beagleboneblack but i got error like xrdb: Can't open display '' xsetroot: unable to open display '' Jun 10 13:24:47 how can i solve Jun 10 13:40:48 OK, I can disable some stuff following this: Jun 10 13:40:50 https://www.element14.com/community/community/designcenter/single-board-computers/next-gen_beaglebone/blog/2013/11/20/beaglebone-web-server--setup Jun 10 13:41:33 but now I don't see the opensprinkler page either, looking into it... Jun 10 13:41:55 uhh, that's information from 2013 Jun 10 13:42:02 if any part of it worked for you, that was pure luck Jun 10 13:42:27 it's giving instructions for angstrom linux rather than debian Jun 10 13:42:47 also, there's some really bad advice there Jun 10 13:42:51 avahi-daemon should *not* be disabled Jun 10 13:42:58 if you did, please enable it again Jun 10 13:46:10 no, I think it's working now, I just had to do http://IP:8080 Jun 10 13:46:43 specifying the correct port on which the server is actually listening helps yes :P Jun 10 14:00:27 yeah, I figured out the avahi part and turned that back on Jun 10 14:09:07 how can i connect remote access from beaglebone to other system Jun 10 14:31:10 greetings, just purchased a BB wireless, connected the device to my mac and I am unable to enable the wireless network. Jun 10 14:31:35 can you log in via usb networking? Jun 10 14:31:40 yes Jun 10 14:31:43 I m now Jun 10 14:31:55 does 'ip link' list a wlan0 interface or an eth0 interface? Jun 10 14:32:18 (among the various interfaces listed) Jun 10 14:32:30 only the eth0 interface Jun 10 14:33:04 yeah, your beaglebone thinks it's a regular BBB rather than a BBB-wireless due to factory-misprogrammed board identification EEPROM Jun 10 14:33:09 BTW, I have a head on the BB Jun 10 14:33:23 a what? Jun 10 14:33:45 Head, a monitor, keyboard and mouse. Jun 10 14:33:53 Old term, Im old Jun 10 14:34:08 oh yeah lol, I know that term just didn't expect it in this context Jun 10 14:34:15 LOL Jun 10 14:34:23 not sure why you'd want one, but good for you ;) Jun 10 14:34:45 only for fixing the network ssue. Jun 10 14:35:32 I have the BB usb into my mac, however, the usb0 network, while set properly, is also not working. Jun 10 14:37:05 if I can't reach a beaglebone via network at all I'd use the serial console to fix it... I guess keyboard and monitor are a usable solution too, but it would annoy me to not be able to copy-paste stuff between the beaglebone and my laptop Jun 10 14:37:39 or, in your case, the ability to copy-paste the command to write the correct data to the eeprom from irc to the beaglebone (I'm checking for you now) Jun 10 14:37:39 Yea, it is very annoying, but all I have right now Jun 10 14:37:49 thanks Jun 10 14:38:22 I plan on installing kodi on this device and maybe a NAS with Plex Jun 10 14:38:53 lol, are you sure you want to try that on a beaglebone? Jun 10 14:39:15 testing..... Jun 10 14:39:34 its strength is its i/o capabilities, it sucks at media applications Jun 10 14:41:01 anyway, you'll also need to locate testpoint TP1 and connect it to ground (which can be found e.g. on pins 1 and 2 of both expansion headers) during the eeprom write to disable its write-protection Jun 10 14:41:07 once I've tested kodi on this device, I'll more than likely move it to a toaster and then repurpose the deive in my work shop. Jun 10 14:42:14 okay, just don't be surprised if performance is... somewhere between "not great" and "nice slide-show" Jun 10 14:42:47 slide show, very funny Jun 10 14:44:43 while grounding TP1, perform (as root): Jun 10 14:45:01 echo -ne '\xAA\x55\x33\xEEA335BNLTBWA5' >/sys/bus/nvmem/devices/0-00500/nvmem Jun 10 14:45:47 after the write, double-check using a hexdump that the first 16 bytes of that file are: Jun 10 14:45:50 00000000 aa 55 33 ee 41 33 33 35 42 4e 4c 54 42 57 41 35 |.U3.A335BNLTBWA5| Jun 10 14:48:30 (if you mess up the first 12 bytes of the eeprom, you'll have a board that will require a custom-built bootloader to boot it at all anymore, so be cautious) Jun 10 14:49:06 okay, thanks. I am looking for TP1 on the board right now Jun 10 14:50:05 if I remember correctly it's a through-hole pad labeled on the top-side silkscreen, somewhere reasonably close to the board edge Jun 10 14:53:32 found it. Jun 10 14:56:01 for future readers of the irc log having the same problem, TP1 is located here: https://liktaanjeneus.nl/bbbw-tp1.jpg Jun 10 14:57:03 Im looking for a ground point Jun 10 15:00:24 I've reuploaded the image with some ground points marked in green Jun 10 15:01:19 (there are of course plenty more, but these I know for sure without having to check the schematic) Jun 10 15:02:57 it doesn't really matter if you connect it to the wrong thing anyway, the worst that can happen is that the eeprom write will be ignored Jun 10 15:03:26 thanks Jun 10 15:03:45 well I guess connecting it to 5V would probably be bad Jun 10 15:03:51 but other than that :P Jun 10 15:04:27 here goes nothing Jun 10 15:05:30 and if you do manage to misprogram eeprom, you can still fix it if you haven't rebooted yet, hence double-checking the result before rebooting is highly recommended Jun 10 15:10:46 it rebooted all on its own! HA! Jun 10 15:10:58 uh oh, you shorted something? Jun 10 15:11:14 I must have, let see what it does now Jun 10 15:11:30 I now have the GUI desktop, which I didn't habe before Jun 10 15:11:38 o.O Jun 10 15:11:52 that's... unusual Jun 10 15:12:04 maybe startup previously failed to complete properly Jun 10 15:12:45 tbh you probably want to reflash the beaglebone with a console image before you get started (check 'df -h /' to see why) Jun 10 15:42:46 well, that didnt work, I know my wire to ground is proper, since I can replicate shoring the device. Jun 10 16:07:46 what's the current eeprom contents? (specifically bytes 12-15) Jun 10 18:03:42 is there any way to tell whether i boot from emmc or sd card from inside u-boot? Jun 10 18:04:02 is there any way to tell whether i boot from emmc or sd card from inside u-boot? Jun 10 18:04:21 i thought ${mmcdev} might hold the answer, but it seems to be hardcoded to "0" =( Jun 10 18:25:30 you mean from which device u-boot was loaded? Jun 10 18:29:01 yes Jun 10 18:29:14 installed Debian on a new Beaglebone Black... update fails on ti-pru-cgt-installer, tries to download a file from remote and the download fails. URL resolves, but visiting port 80 (where Debian is trying to go) fails. suggestions? Jun 10 18:29:19 that way, i can control what to boot from (including bootloader) using S2 and can use the same image for both Jun 10 18:29:39 (mine still boots from SD card using the eMMC bootloader otherwise, which i find mightily confusing) Jun 10 18:33:13 ahh, figured it out. IPv6 works, but IPv4 doesn't. Jun 10 18:34:45 Huh IPv4 is phasing out on local IP nets? Jun 10 18:40:10 zmatt: also tried with a bootscript, but it fails with 'Wrong image format for "source" command' Jun 10 18:46:16 GenTooMan: no, I enlarged my IPv4 private IP subnet at home and I forgot to reconfigure my firewall to permit routing for the entire new subnet :) Jun 10 18:46:29 GenTooMan: but all but one file on the update was accessible via IPv6 and local connectivity was fine so I couldn't tell Jun 10 18:48:00 zmatt: nvm, apparently, those need to be "compiled" Jun 10 18:49:41 bbl Jun 10 20:14:19 PhotoJim: of course you could also just have removed the package if you don't need it... but fixing your network problems is probably preferred ;) Jun 10 20:15:29 sgflt: why do you find that "mightily confusing" ? u-boot's boot script simply attempts to load linux from the mmc devices in sequential order: first mmc0 (sd card), then mmc1 (eMMC) Jun 10 20:16:40 in theory the information on how u-boot was loaded should be accessible. in particular, SPL uses it to load u-boot.img from the same device. I don't know if this information is exposed via any environment variable, but that should be easy enough to determine by dumping the environment in both cases and doing a diff Jun 10 20:17:14 I'm not sure what you're trying to do with those archaic compiled bootscripts Jun 10 20:24:22 I'm trying to get my BBB running with a kernel that supports both PRU RPROC and GLES Jun 10 20:25:30 wiley: i.e. any -ti kernel with an sgx drivers package Jun 10 20:25:36 It seems like as far as RCN's Ubuntu/Debian packages go, there are the -bone kernels that use PRU UIO and have matching ti-sgx-es8-modules packages, but the -ti kernels don't have any such thing? Jun 10 20:25:45 those are really old sgx drivers Jun 10 20:26:27 interesting... Jun 10 20:26:28 the package you're looking for it ti-sgx-ti335x-modules-$VERSION Jun 10 20:27:24 I have that, but the GLES app I'm trying to run is segfaulting at launch (even though the same app runs successfully on a -bone kernel of a similar version) Jun 10 20:27:50 are there any quick GLES diagnostics/test tools I can try to figure out whether it's my app or the platform? Jun 10 20:27:52 well the modules alone aren't sufficient obviously Jun 10 20:27:55 something like glxgears? Jun 10 20:27:59 you also need the correct userspace libs Jun 10 20:28:31 do I need to build those libs myself? Jun 10 20:28:37 they're binary-only Jun 10 20:28:54 I'm checking if rcn has a package for them Jun 10 20:31:22 doesn't seem like he does Jun 10 20:32:23 I have described (sort of) how I got them working here: https://groups.google.com/forum/#!topic/beagleboard/BEipC_SFjaE the first two links aren't important anymore since rcn builds modules-packages Jun 10 20:32:49 I also made hacky debian packages: https://liktaanjeneus.nl/data/public/sgx/ which presumably still work Jun 10 20:33:26 so the thing I think I'm not understanding is the difference between what's in ti-sgx-es8-modules and what's in ti-sgx-ti335x-modules, because I literally installed a 4.14-bone kernel with the matching ti-sgx-es8 package, observed that everything worked, installed a 4.14-ti kernel with matching ti-sgx-ti335x package, rebooted, and observed that things no longer worked Jun 10 20:33:45 are they different versions of the same drivers or completely separate stacks that need separate usermode libraries? Jun 10 20:33:51 really? the es8 stuff still works? Jun 10 20:33:55 I'm really surprised by that Jun 10 20:34:02 because that stuff is oooooooold Jun 10 20:34:18 interesting Jun 10 20:34:30 yea, I can just boot it up, crank up my app, and see goodies on the screen with no trouble Jun 10 20:34:54 I'm using the new drivers on my (custom-compiled) -bone kernel Jun 10 20:35:48 it does seem rcn doesn't have ti-sgx-ti335x-modules packages for -bone kernels... weird Jun 10 20:36:27 (I don't care enough to bother emailing him to ask him about it though) Jun 10 20:38:19 yea... Jun 10 20:38:38 everything I've seen so far seems to imply that RCN supports the -bone kernels much more heavily than -ti Jun 10 20:38:51 ... but there also seems to be a -ti kernel installed out of the box in his current images, which just confuses me Jun 10 20:39:03 it seems the opposite to me, since the -ti kernels... yeah exactly that Jun 10 20:39:22 ti kernels have been default for quite a bit now Jun 10 20:39:55 also, keep in mind the es8 modules are the obsolete drivers, while he only seems to package the new drivers for -ti kernels Jun 10 20:40:22 so I'm not sure why you have the impression that the -bone kernels are more heavily supported Jun 10 20:43:44 I guess what makes me thing -bone is preferred is that https://elinux.org/BeagleBoardDebian#SGX_Drivers seems to only be written for -bone Jun 10 20:44:31 because it's old info that has never been updated Jun 10 20:46:56 like, the examples on the wiki page for this sdk at TI still use a 2.6 kernel Jun 10 20:49:15 although possibly that's just because the wiki page is even older... the final release of this sdk is from januari 2015 Jun 10 21:05:15 should I be concerned that work on http://git.ti.com/cgit/cgit.cgi/graphics/omap5-sgx-ddk-um-linux.git/ seems to have stopped two years ago? Jun 10 21:06:04 wrong branch Jun 10 21:06:28 http://git.ti.com/cgit/cgit.cgi/graphics/omap5-sgx-ddk-um-linux.git/refs/ several branches were updated a mere 3 weeks ago Jun 10 21:07:29 and yes it's kind of a mess :P I think I used ti-img-sgx/1.14.3699939_k4.4 Jun 10 21:07:55 ergh Jun 10 21:08:16 this is making me really appreciate the es8 drivers and really wonder how TI expects anyone to build things on this board... Jun 10 21:08:28 do you think I'm better off with your repo or going upstream? Jun 10 21:08:49 if you want min-effort I suggest trying to see if my debian packages work Jun 10 21:09:10 they should, since we make active use of them Jun 10 21:09:49 so I can't browse the repo directly and I'm on Ubuntu, so I don't think apt will be happy to pull things from the repo... Jun 10 21:10:17 uhh what? it's not a repo, it's just a directory with a few debs in it Jun 10 21:10:34 you just download the .deb to the beaglebone and install it with dpkg -i Jun 10 21:10:34 ah, I was assuming it was a repo because I get a 403 at https://liktaanjeneus.nl/data/public/sgx/ Jun 10 21:10:47 I assumed you had just disabled directory listings Jun 10 21:10:53 huh why did I link that, lol Jun 10 21:11:09 https://liktaanjeneus.nl/sgx/ Jun 10 21:11:13 sorry, my bad Jun 10 21:11:23 wasn't paying attention Jun 10 21:11:35 lol, no worries Jun 10 21:11:37 yea, that works Jun 10 21:11:38 thanks Jun 10 21:13:00 (the perl script is in case you want to use qt5 eglfs and are, like me, too lazy to try to recompile it to fix its stubborn insistence of using 24-bit depth) Jun 10 21:13:53 You seem to have references to a sgx-init package from somewhere? Jun 10 21:13:56 (which wouldn't be a huge problem if it supported the pixel format needed for 24-bit color on the beaglebone, which it doesn't) Jun 10 21:14:07 this userspace package includes the init stuff Jun 10 21:14:26 I think at some point I wanted to split them Jun 10 21:14:26 ah, yea, my bad Jun 10 21:14:37 looks like I need to nuke a bunch of the es8 libraries to try this... Jun 10 21:14:38 or I did but never put that up... dunno, this was a year ago :P Jun 10 21:14:43 yeap Jun 10 21:14:56 having two different versions of libs around is a recipe for exciting fun Jun 10 21:15:46 ditto if you have mesa around together with these libs, which dpkg should however refuse since I added Conflicts: for those Jun 10 21:15:58 iirc Jun 10 21:18:45 maybe I should put a little bit more time into this again some time to help rcn make this stuff as simple as apt-get install Jun 10 21:18:58 it's just hard to find the time and motivation Jun 10 21:20:49 yea, I can imagine Jun 10 21:21:47 but, rest assured: it *can* work Jun 10 21:21:50 ;) Jun 10 21:23:23 LOL Jun 10 21:23:29 Failed to load shared library 'libmash-0.3.so.0' referenced by the typelib: libdrm_omap.so.1 Jun 10 21:23:34 (and the same for libclutter) Jun 10 21:23:53 (this is a python app running on top of Clutter+Mx) Jun 10 21:23:56 say what? I've never heard of "libmash" in my life Jun 10 21:24:08 I wonder if I'm going to have to rebuild the entire stack leading up to clutter... Jun 10 21:24:30 and libdrm_omap isn't even useful Jun 10 21:24:45 (on the beaglebone) Jun 10 21:25:26 we basically use it because it gives us a UI toolkit with multitouch widgets that can run on GLES Jun 10 21:25:40 (clutter, not anything OMAP-related) Jun 10 21:25:56 unfortunately, getting all that does mean it has a lot of dependencies to get there Jun 10 21:25:57 but it seems to be saying libdrm_omap is pulling in libmash? that doesn't make sense? Jun 10 21:27:01 I'm checking but I think it's actually the other way around... Jun 10 21:27:06 if you want you can probably even replace libdrm_omap.so.1 by a blank library Jun 10 21:27:43 something like gcc -shared -o blank.so -x c /dev/null or something Jun 10 21:28:18 well, the only libdrm_omap on my machine is the /usr/lib/arm-linux-gnueabihf/libdrm_omap.so.1.dummy that your package installed, so I think the error must be phrased backwards... Jun 10 21:28:30 ah I provided one Jun 10 21:28:42 did a symlink libdrm_omap.so.1 -> libdrm_omap.so.1.dummy get created? Jun 10 21:28:56 ah... no Jun 10 21:29:08 okay so maybe try running... ehh... what's it called Jun 10 21:29:16 thingy-to-update-those-symlinks Jun 10 21:29:20 ln -s ? Jun 10 21:29:20 oh Jun 10 21:29:33 sudo ldconfig ? with some options possibly? Jun 10 21:30:08 running it without arguments should be fine actually Jun 10 21:30:13 yea, without arguments worked Jun 10 21:30:43 made it further: Jun 10 21:30:58 PVR:(Error): Couldn't load WS module libpvrPVR2D_FRONTWSEGL.so [0, ] Jun 10 21:31:03 PVR:(Error): PVRSRVBridgeCall: Failed to access device. Function ID:3223086849 (strerror returns no value.). [0, ] Jun 10 21:31:14 are you sure you removed the old libraries? Jun 10 21:31:35 I basically checked dpkg and nuked everything that had gles, egl, or mesa in the name Jun 10 21:33:00 but you installed the graphics sgx from TI, which isn't a debian package Jun 10 21:33:08 so you're going to have to manually get rid of it Jun 10 21:33:23 I think? Jun 10 21:33:38 yeah Jun 10 21:33:49 maybe check that sgx-install.sh script to see what it does Jun 10 21:34:11 *graphics sdk Jun 10 21:34:45 oh ick... Jun 10 21:35:00 yea, think I'm just going to rewrite that script to delete everything it's supposed to install Jun 10 21:35:23 probably easier to remove it manually... iirc it doesn't affect many locations Jun 10 21:35:29 but maybe I don't remember correctly Jun 10 21:35:41 ehhhh, it touches quite a bit Jun 10 21:35:46 that's unfortunate Jun 10 21:35:51 time to reflash? ;) Jun 10 21:36:07 it also uses lovely functions like install_lib that I can rewrite very quickly :) Jun 10 21:45:49 hmm, I'm beginning to consider pizza Jun 10 21:48:37 ergh Jun 10 21:48:55 after nuking everything I could find, I have three errors while launching gles1test1: Jun 10 21:49:00 PVR:(Error): Couldn't load WS module libpvrPVR2D_FRONTWSEGL.so [0, ] Jun 10 21:49:04 PVR:(Error): WSEGL_CreateWindowDrawable: Couldn't set CRTC: Invalid argument [0, ] Jun 10 21:49:10 'eglCreateWindowSurface' returned egl error 'EGL_BAD_ALLOC' (0x3003) Jun 10 21:49:39 if you run ldd on gles1test1, which libraries does it list? Jun 10 21:50:34 you didn't eradicate the old libs successfully, since libpvrPVR2D_FRONTWSEGL.so is old stuff Jun 10 21:50:53 so something is still explicitly trying to locate old stuff for some reason Jun 10 21:51:35 also, make sure the sgx-init.service is started Jun 10 21:51:42 https://pastebin.com/ARHJNsKF Jun 10 21:52:42 that looks... kinda okay? weird that it's trying to load libpvrPVR2D_FRONTWSEGL.so then. is there still some old config file around that lists that library or something? Jun 10 21:52:45 I had already initialized PVR through pvrsrvctl - looks like sgx-init is failing because it's already initialized Jun 10 21:52:52 ah ok Jun 10 21:53:17 good call - looks like it's still in powervr.ini Jun 10 21:53:20 (which I haven't touched yet) Jun 10 21:53:28 just toss it Jun 10 21:53:42 actually lemme check if I still have that file Jun 10 21:53:55 nah I don't Jun 10 21:54:00 I think? Jun 10 21:54:07 oh I do Jun 10 21:54:26 https://pastebin.com/RkzhwVbR Jun 10 21:54:53 ayyyyyyyyyyyyyy Jun 10 21:54:55 dancing triangles Jun 10 21:55:08 although... wtf? Jun 10 21:55:28 `gles1test1` terminates instantly but `gles1test --help` shows dancing triangles Jun 10 21:58:16 triangle dance? Jun 10 21:59:08 iirc not all of the tests worked equally well Jun 10 22:00:05 do you really care about gles1 anyway? (as opposed to gles2) Jun 10 22:02:15 nah, that was just the test program I caught the name of earlier Jun 10 22:02:33 my python/clutter/mash/cogl app is also running, although it's as the wrong resolution Jun 10 22:02:37 I'm guessing that's probably my own fault :) Jun 10 22:02:58 thanks so much zmatt Jun 10 22:03:06 I'm not sure I would have figured this out at all without you Jun 10 22:03:17 What are you using to set the resolution? Jun 10 22:04:09 currently digging for that - I'm afraid it might be in a config Jun 10 22:05:09 I'll admit I have no idea how it selects resolution. we use a fixed-resolution display (via lvds, not hdmi) so I've never had to deal with that Jun 10 22:05:16 ah, looks like we're running at the screen's native resolution but our magical JSON-based UI is coded for 800x480 Jun 10 22:05:40 and apparently the default behavior is to tile it for some reason :\ Jun 10 22:05:44 lol Jun 10 22:05:47 sounds good Jun 10 22:06:52 but wait, tiled? does that mean you're rendering into a pixmap and then render that to display or something? because I'm not sure how else you could manage to tile the display Jun 10 22:07:08 (but if so, that sounds really inefficient) Jun 10 22:07:34 we're a couple layers removed, but that's probably what Clutter is doing for us Jun 10 22:08:20 the UI I'm messing with is actually a 3d printer control interface, and it's mostly static content and icons with a few graphs we draw ourselves Jun 10 22:08:39 it's not the most efficient but it also doesn't need to be as long as it doesn't eat too much CPU Jun 10 22:09:37 this is the sort of project where we probably would have been happy with a Chromium shell and some HTML/JS, but that's probably even less efficient on a BBB Jun 10 22:11:56 well I'm wondering since I'm not even sure the GBMWSEGL backend supports rendering into a pixmap... so I'm wondering whether your ui uses gles at all, or whether it's actually cpu-drawn and just blitted onto the display using the gpu Jun 10 22:12:02 (which would be epic silly) Jun 10 22:12:08 (but plausible) Jun 10 22:12:47 but hopefully I'm wrong and clutter actually does clever things Jun 10 22:13:04 hrmmm Jun 10 22:13:41 well, I'm seeing 1-5% CPU usage at idle Jun 10 22:13:53 why are you seeing any cpu usage at idle? Jun 10 22:14:06 we're still rendering graphs and such Jun 10 22:14:09 ah ok Jun 10 22:14:28 well if performance is good enough then I wouldn't spend a second of thought on how it works anyway Jun 10 22:14:46 lol Jun 10 22:15:11 eh, we'll probably put a bit of thought into it before calling it good :) Jun 10 22:15:26 we're running an awful lot of code with soft-realtime constraints on a single core Jun 10 22:17:41 brb pizza Jun 10 22:18:17 Good enough sometimes turns out not too be I found so I tend to do a bit better just to have some margin when someone randomly adds a :feature: Jun 10 22:21:37 tsk tsk, premature optimization ;) Jun 10 22:33:07 I think of it as SYA (saving your arse) from stupid software people who think THERE application should have all the CPU clock. Jun 10 22:34:10 *their Jun 10 22:35:04 and there's a simple solution for that: set a cpu% quota in the systemd unit file ;) Jun 10 22:36:42 most software people like to take short cuts maybe that would tamper their zeal for "cool" instead of working. Jun 10 22:41:16 working is cool too Jun 10 23:04:04 zmatt, unfortunately, it looks like Ubuntu 18.04 changed the virtual package names just enough that your packages don't fill the dependency chain they did on 16.04 :\ Jun 10 23:04:31 ubuntu? you're using ubuntu rather than debian on the beaglebone? Jun 10 23:04:47 yea... Jun 10 23:05:15 well, I'm afraid any problems caused by using an obscure distro are your own :) Jun 10 23:05:33 I've never tested on ubuntu nor do I have any intention to Jun 10 23:05:50 lol, kk Jun 10 23:06:35 you can of course try unpacking the deb, fiddling with the control file, and building a new deb Jun 10 23:26:53 think I can get away with it by making your package provide libegl1 Jun 10 23:27:14 yeah it depends on which packages you want to install that depend on this of course Jun 10 23:27:28 basically you need to provide whatever is needed to prevent mesa from being pulled in Jun 10 23:28:17 ah yes, I needed to provide libegl1-x11 even though that's bullshit... my package description even mentions that Jun 10 23:28:32 but libegl1 itself isn't in the list since apparently nothing I needed had a dep on that **** ENDING LOGGING AT Mon Jun 11 03:00:02 2018