**** BEGIN LOGGING AT Sat Dec 22 03:00:03 2018 Dec 22 03:00:37 is omaplfb meant to be loaded? Dec 22 03:01:04 because I see a few examples with that using the SGX module but i'm not sure if that is an older version Dec 22 03:01:08 I don't have that module Dec 22 03:01:25 no that's something related to the old drivers, it no longer exists Dec 22 03:01:36 ahh okay Dec 22 03:02:32 so, I just found a thread on e2e with the same error, and it seems TI did patch weston for their sdk Dec 22 03:02:43 https://e2e.ti.com/support/processors/f/791/p/675976/2493812#2493812 Dec 22 03:03:57 ahh. So just need to patch weston. Though maybe I'll build KMScube first Dec 22 03:04:23 now that I know I'm not missing some other module Dec 22 03:04:25 yes even that thread recommends first testing kmscube to verify the drivers are working Dec 22 03:11:34 kmscube works when I change the pixel format Dec 22 03:11:55 woot :) Dec 22 03:19:23 this looks a bit crazy. It's a repo of patches... I swear I haven't seen this repo of patches approach until I looked at the BBB related stuff ;) Dec 22 03:20:34 not sure how to know which version of the weston/wayland repo this applies to... I guess i just try the current one Dec 22 03:24:09 zmatt looks to be designed for ethercat and high speed sync. As for media industrial machine vision is very big in robotics applications at the moment. Dec 22 03:24:36 suprothunderbolt: their sdk uses openembedded stuff I think Dec 22 03:25:43 oh so I build the current openembedded mirrror? Dec 22 03:26:11 Is there a reason to use the BBB standard debian repo vs the TI sdk? Dec 22 03:26:14 I don't know anything about how this stuff works Dec 22 03:27:33 yes, one is debian, the other is yocto/arago/openembedded/whatever Dec 22 03:28:13 the former is supported here, the latter on TI's support forum Dec 22 03:54:43 zmatt you are correct I must have mixed up with a different processor I had read on earlier. Sorry about that. Dec 22 04:05:15 GenTooMan: vayu (am57xx/dra7xx/tda2) is currently the most "media/graphics"-heavy SoC TI has at the moment, targeting automotive infotainment and automotive vision/ADAS applications Dec 22 04:07:59 the am5728/am5729 on the beagleboard-x15 includes most of that functionality, except for the four EVEs (although on my x15 one of the EVEs actually seems to be operational; but I don't know if that's true for all x15 boards in general) Dec 22 04:50:54 Is there any particular guarantee of read/write atomicity when loading/storing to DDR from the PRU? Dec 22 04:51:57 individual loads and stores will be atomic if they're small enough and properly aligned. other than that, nope Dec 22 04:51:59 I'm guessing there has to some sort of documentation describing if/when the CPU and PRU will have consistent views of memory, but I haven't yet found it anywhere in TI's docs Dec 22 04:52:59 if I can get atomic loads and stores of 32 bit properly aligned values, that'd be just fine Dec 22 04:53:16 what I'm also wondering is if write caching in the CPU can cause the PRU to have an inconsistent view of memory Dec 22 04:55:30 your question is a bit too vague to give an answer. it is relatively easy to construct a mental model of pru/cortex-a8 interaction via shared memory, but it still requires some care Dec 22 04:56:03 the biggest dangers are when interacting via one than one mechanism, e.g. ddr memory and pruss memory, or memory and interrupts Dec 22 04:57:23 so in the bigger picture, I'm looking into overhauling my PRU/CPU interactions into something message-based, but I'm wanting to stick with UIO if possible Dec 22 04:57:46 which means building my own (fairly simple) message-passing system Dec 22 04:58:00 I have a message-passing example in py-uio Dec 22 04:58:15 using a ringbuffer Dec 22 04:58:37 and I'm wondering if there's any formal description anywhere of the consistency guarantees the AM335x provides Dec 22 04:58:48 interesting, I'll have to take a look Dec 22 04:59:46 but yea, things like... if I write a bunch of memory (a "message") on the CPU, write a pointer to that memory somewhere in the PRU's memory, then set a CPU->PRU interrupt, what steps do I need to take to make sure the PRU can read the message as I intend? Dec 22 04:59:47 on the SoC level, read/write requests and replies are basically just packets that travel across a network with no reordering Dec 22 05:00:13 and/or is it even possible for part of the message to be present in main memory and visible to the PRU while other parts are still in the CPU's L2 cache or somesuch Dec 22 05:00:43 uio memory regions are never cached Dec 22 05:01:09 that much makes sense Dec 22 05:01:31 but what happens if the message is large enough that I need to keep in in DDR? Dec 22 05:02:13 (or am I just looking for memory consistency problems that aren't there because cache consistency in multi-socket AMD64 chips has made me paranoid?) Dec 22 05:02:53 in fact, they're mapped as "strongly ordered" by default, which means every load/store byte/halfword/word/doubleword instruction is performed synchronously to memory, and the instruction doesn't complete until the reply is back Dec 22 05:04:43 for performance reasons I personally patch my kernel to make uio (and /dev/mem) use "device" type instead, which is also what the kernel uses and similar to how pru operates: loads/stores will still be performed as-is and in order, but it will not wait on stores Dec 22 05:07:01 so, when communicating via a ringbuffer of messages.... if you first write a message to the ringbuffer and then update the head-pointer, then anyone who sees the updated pointer will also read the correct data Dec 22 05:07:28 this is simply because memory accesses will not get reordered Dec 22 05:07:44 I follow that much Dec 22 05:08:18 hrmmmm Dec 22 05:09:14 one way to look at my question is to suppose the ringbuffer lives in shared PRU memory but the message I'm sending contains pointers to more data in the DDR Dec 22 05:09:15 however if, after updating the pointer, you notify the other party via some out-of-band mechanism like an interrupt, it is conceivably possible that when it reads the pointer, it will see the old value and think there's no new message Dec 22 05:11:04 though when notifying the cortex-a8 via interrupt this will probably never happen in practice, simply because its interrupt latency is really high Dec 22 05:13:36 as long as you use a strongly-ordered mapping on the cortex-a8 (which, as I said, is the default for uio) this problem also won't happen for cortex-a8 -> pru messages since the cortex-a8 will use non-posted writes (i.e. writes with completion reply) and wait for their completion Dec 22 05:14:28 for pru, if you want to be 100% sure your writes have arrives at the target, just perform a read (to anywhere in the same target) Dec 22 05:15:01 interesting - this is starting to make more sense Dec 22 05:15:48 so you're saying the contiguous physical memory that the UIO driver reserves at boot as main-DDR scratch space is also mapped by default to basically bypass the cache? Dec 22 05:16:14 sorry, the first time I read that I thought you were just referring to the CPU's accesses to the PRU registers and PRU RAM Dec 22 05:16:41 so, if you have pru write data to target A and want to store a pointer to that data in target B, it should first read from target A to ensure the data has landed before writing the pointer to target B. this will ensure that anyone who chases the pointer will see the data Dec 22 05:17:56 I'm pretty sure all memory regions mapped by uio are mapped as strongly-ordered, regardless of whether they're memory-mapped registers or actual memory Dec 22 05:18:00 lemme check Dec 22 05:20:40 yep Dec 22 06:09:06 got weston working :) Dec 22 06:09:18 with SGX Dec 22 06:09:59 thanks fro the help zmatt :) Dec 22 06:11:49 woot! maybe put the requires steps somewhere, or at least put the patched weston on github for reference? Dec 22 06:12:11 will do Dec 22 06:13:00 is there an easy way to generate debs from a autotools package? Dec 22 06:13:24 then I could upload the debs because the main annoying thing is getting all the dependencies to build wayland Dec 22 06:18:53 no idea Dec 22 06:19:24 I'll see if I can have got at it but for the moment I'll put up the repos. Dec 22 06:53:37 wiley: https://pastebin.com/raw/u8c0XmBZ Dec 22 07:00:57 wiley: there are still a few cases inside write_barrier() I need to check or think about... modern CPUs are annoying Dec 22 07:01:52 on PRU all of the asm() can be ignored, it only requires the dummy read if buf and head aren't in the same target Dec 22 07:08:18 bbl Dec 22 07:21:27 zmatt, interesting, thanks Dec 22 07:23:28 zmatt, I'll probably do some glaring at that tomorrow Dec 22 08:00:24 wiley: note that, apart from the aforementioned pru case, you can safely ignore all the barrier-stuff since you're using strongly-ordered mappings on the cortex-a8 anyway. I included it mostly because I want to know it myself Dec 22 08:00:39 makes sense, yea Dec 22 08:01:06 it might become relevant if you want to increase performance of shared memory access Dec 22 08:01:16 (writing to strongly-ordered memory is sloooooow) Dec 22 08:01:27 interesting Dec 22 08:01:33 that's something I'd have to profile Dec 22 08:02:18 but I'd expect my memory bandwidth needs in the message-passing system to be similar to what I have now with the much hackier and less flexible interface, and what I have now isn't enough of a bottleneck that it's ever mattered Dec 22 08:03:55 store instructions to non-strongly-ordered memory typically take only 1 cycle (until you manage to fill up the store buffer), whereas a strongly-ordered write is as slow as a read (iirc on the order of 150ns but don't hold me to it) Dec 22 08:04:43 yeah it might not matter of course, that depends on the application Dec 22 08:05:01 does it take a full kernel patch to map another way or is it possible to tweak things through something like arguments to mmap? Dec 22 08:05:27 it takes a kernel patch, but it's a pretty tiny one Dec 22 08:05:48 hrmmm Dec 22 08:06:44 makes me wonder if it could be a cmdline parameter for uio_pruss that could be checked in upstream Dec 22 08:06:47 if you're currently using a kernel without any custom patches then I can imagine that might make it unappealing. I already have a ton of patches anyway, so for me it really doesn't matter Dec 22 08:07:00 yea, I'm on RCN's standard kernels Dec 22 08:08:13 anyway, I'm off for the night Dec 22 08:08:19 good night! Dec 22 08:08:24 thanks for all the help - it's been very interesting Dec 22 08:20:14 How can I contribute to this organisation? Dec 22 13:45:38 hi is there a page for getting started with the debian 9.5 IOT image, i.e. (ssh)ing user name default password, how to sudo ...etc? Dec 22 13:51:28 http://beagleboard.org/getting-started -- this generally applies **** ENDING LOGGING AT Sat Dec 22 18:13:37 2018 **** BEGIN LOGGING AT Sat Dec 22 18:16:09 2018 Dec 22 22:11:28 hey Guys I need a hand with beaglebone black, i just hookied it up via HDMI to my TV, and i got nothing, no video Dec 22 22:13:10 hello everyoen Dec 22 22:13:34 does the system otherwise seem to boot normally? can you ssh into it? Dec 22 22:14:14 zMatt how are you we spoke before about the debian password.... i have all solid Blue lights Dec 22 22:14:54 ehh what? all solid blue lights? that sounds bad Dec 22 22:15:26 you mean that when you power it on, the blue lights turn on in sequence and then don't turn off again? Dec 22 22:15:29 beside The ethernet connecor there are 3 solid blue lights Dec 22 22:15:57 which of the four isn't on? Dec 22 22:17:05 no, oohhh ok now we are geting some where i pressed the boot button then powered up i got the linux penguin Dec 22 22:17:26 oh, you're booting from sd card? Dec 22 22:17:38 maybe there's an ancient bootloader on eMMC that's causing boot problems Dec 22 22:17:46 in the upper orner then it dissapeared now i have a cursor on the screen Dec 22 22:17:53 yes booting from the sd card Dec 22 22:18:00 in that case, the fix is to either reflash or at least erase eMMC Dec 22 22:18:33 then it should boot normally without having to use the S2 button Dec 22 22:18:48 (and if you reflash eMMC you will also no longer need to boot from sd card) Dec 22 22:20:10 erasing eMMC (assuming you don't care about its old contents) is a simple first step though: sudo blkdiscard /dev/mmcblk1 Dec 22 22:21:27 well i dnt have a usb keyboard i only have a usb mouse attached Dec 22 22:22:15 right now it has a picture of a dog and it sasy beagleboard.org on the screen how do i know if i am booting into debian? Dec 22 22:22:17 yeah I meant via ssh. using a beaglebone by connecting keyboard and monitor is too much hassle to bother with Dec 22 22:22:44 uhh, if you flashed debian onto the sd card, you've booted into debian obviously Dec 22 22:23:37 i flashed the card hit the button and the lights ae flashing like crazy... i asum i booted from the card. any way to confirm. Dec 22 22:23:59 findmnt / Dec 22 22:24:20 the "SOURCE" should be /dev/mmcblk0p1 if you've booted from sd card Dec 22 22:24:52 zmatt gve me a sec whil i switch to the mac i have to reboot this computer i will be back in a sec Dec 22 22:25:49 im gonna try ssh from windows Dec 22 22:37:11 zmatt i forgot the root default username and password i think the password is tmppwd Dec 22 22:37:16 i cant remember the username Dec 22 22:37:19 debian Dec 22 22:37:26 you could have easily found that by googling Dec 22 22:37:45 password temppwd Dec 22 22:38:03 ive been googling and its giving mt root and toor lol Dec 22 22:38:42 ok im in Dec 22 22:38:49 I just googled "beaglebone black password" and the first google hit is http://beagleboard.org/getting-started Dec 22 22:38:58 which contains the right info Dec 22 22:39:12 i was googling debian username and password Dec 22 22:39:54 ok im in Dec 22 22:40:04 debian has no default username/password since normally you use a graphical installer for it Dec 22 22:40:24 ah ok Dec 22 22:40:51 so from here how do i flass the emmc Dec 22 22:41:32 again you could first have tried looking in obvious places... there were instructions below the download links, and there's a link to instructions on the getting-started page I just linked to Dec 22 22:43:11 , Dec 22 22:51:35 zmatt i found this, i was able to edit the /boot/uEnv.txt ##enable BBB: eMMC Flasher: #cmdline=init=/opt/scripts/tools/eMMC/init-eMMC-flasher-v3.sh Change to: ##enable BBB: eMMC Flasher: cmdline=init=/opt/scripts/tools/eMMC/init-eMMC-flasher-v3.sh Dec 22 22:51:56 i cant find the two lines in the file that i need to edit Dec 22 23:04:29 Ant__: I'm pretty sure it's all the way at the bottom Dec 22 23:12:28 i found it i rerbooted and a process started the last statement i got here is Dec 22 23:13:09 Write to {/dev/mmcblk1] Failed... Dec 22 23:13:24 Stopping Cyclone LEDs.. Dec 22 23:13:36 Setting LEDs to Dec 22 23:13:45 and thats it can i reboot this thing? Dec 22 23:14:04 write failed? that doesn't sound good Dec 22 23:14:12 "setting leds to" .. to what? Dec 22 23:14:18 are the leds on or blinking? Dec 22 23:14:26 (or did it power off?) Dec 22 23:15:40 all 4 are blinking solid Dec 22 23:15:49 okay so flashing failed Dec 22 23:15:55 there are other messages avove that Dec 22 23:15:59 Formatting Rootfs:/devmmcblkp1 Complete Dec 22 23:16:18 next question is why. do you have a really old beaglebone black with only 2 GB of eMMC maybe? (any revision older than rev C ) Dec 22 23:16:30 Creating Temporary Rootfs Directory 9/tmp/rootfs) Dec 22 23:17:01 Mounting /dev/mmcblkp1 to tmp/rootfs Dec 22 23:17:06 please stop Dec 22 23:17:11 ok Dec 22 23:17:19 just copy/paste the output to a paste site like pastebin.com Dec 22 23:17:31 oh you can't copy/paste Dec 22 23:17:32 i got this board 2 years ago Dec 22 23:17:49 no im getting this off my tv screen Dec 22 23:18:29 i cant connect to the board right now Dec 22 23:18:44 yeah I forgot you can't ssh in when running the flasher Dec 22 23:19:00 yup Dec 22 23:19:16 is there maybe a sticker on the beaglebone with a revision on it? Dec 22 23:20:28 on the ethernet jack it says lpj0011bbni 1314 Dec 22 23:20:45 that doesn't resemble any information I recognize Dec 22 23:22:21 pcb rev4b Dec 22 23:22:31 sorry b4 Dec 22 23:24:05 I don't know how pcb revisions correspond to beaglebone revisions exactly, except that beaglebone rev B and rev C have pcb revision B6, so yours is a pretty old one Dec 22 23:24:13 in that case you only have 2GB of eMMC, not 4GB Dec 22 23:24:33 so the only recent image that will fit on eMMC is the console image Dec 22 23:25:00 (iot used to fit, but unfortunately it too has become too bloated) Dec 22 23:25:21 so your options are flashing a console image, or just erasing eMMC and boot from sd card instead Dec 22 23:26:04 i can hit the button to boot into it Dec 22 23:26:27 is that a question or a statement? Dec 22 23:27:05 I'm pretty sure that when you've changed an sd card to be a flasher card you can't make it boot normally unless you undo the change to /boot/uEnv.txt Dec 22 23:27:30 if you don't have a linux system where you can mount the sd card, it may be easiest to just write a new image onto sd card Dec 22 23:29:13 i wrote debian to the sd card... um how do i prevent this thing from trying to re write itself, i hit the boot button to boot to the SD card and its trying to reflash teh emmc again Dec 22 23:29:23 I just answered that Dec 22 23:29:49 the sd card wasnt loaded with a flasher Dec 22 23:30:01 you turned it into a flasher by changing its /boot/uEnv.txt Dec 22 23:30:01 it was just the lates debian Dec 22 23:30:46 ah ok so now i have to reformat the card so that its not a flasher i got you Dec 22 23:31:29 or undo the change to /boot/uEnv.txt, but that requires a linux system that can read the card Dec 22 23:34:03 i would have to reboot the computer into ios.... i have a 64gb sd card is tehre a way that when flashing the card the image will keep the remaing size of the card? is there a way to partition the card so that i can use the rest of the card for linux apps Dec 22 23:34:28 you can expand the filesystem to span the entire card Dec 22 23:34:47 https://pastebin.com/ThmLM0pX Dec 22 23:38:14 the board doesnt boot, i tried to power down power up the bbb while i am reimaging the sd card, the bbb is hanging Dec 22 23:55:17 which is the expected result since flashing failed Dec 22 23:57:14 just trying a guid i found on the internet Dec 22 23:57:27 to do what? Dec 22 23:57:38 https://old.ghielectronics.com/community/forum/topic?id=23763 Dec 22 23:57:44 reflash Dec 22 23:58:16 you already know how to reflash, you performed the steps correctly Dec 22 23:58:36 I also explained why it doesn't work: your BBB is an older revision whose eMMC is too small for current iot images Dec 22 23:59:30 i want to reflash it with the original bbb software Dec 23 00:00:08 why would you want to reflash some random old image? Dec 23 00:02:38 I'll admit I don't know why the iot images no longer fit in 2G, they did fit earlier Dec 23 00:03:12 btw I hope you're not trying to flash the actual image linked to in that post, since that's not merely an ancient image but it's also an lxqt image, which is even bigger than the iot image Dec 23 00:03:25 no idea, i am a newbe when it comes to linux and bbb Dec 23 00:03:52 then probably the easiest is to just use the latest iot image and boot from sd card Dec 23 00:04:15 thats what i am going to do once i get the old image in the bbb. Dec 23 00:04:26 which old image? and why? Dec 23 00:04:56 flashing an old image onto eMMC will cause it to not be able to properly boot a modern system from sd card Dec 23 00:05:03 (without using the S2 button) Dec 23 00:05:15 if you're going to boot from sd card, I strongly recommend erasing eMMC Dec 23 00:15:45 console 1.78gb Dec 23 00:15:57 ? Dec 23 00:16:26 the bbb wont boot up Dec 23 00:16:49 from sd card? Dec 23 00:16:56 it hangs on something that says cant load the leds something or other Dec 23 00:17:03 this is loading with out the sd card Dec 23 00:17:46 why are you expecting to be able to? Dec 23 00:18:04 did you flash an image that actually fits? if so, which? Dec 23 00:18:16 i thought all bbb can boot with out an sd card Dec 23 00:18:33 if a system is flashed onto it yes Dec 23 00:19:06 so whe you get a bbb fresh out the box and you boot it up it wont boot ? Dec 23 00:19:13 without an sd card Dec 23 00:19:29 beaglebones always initially ship with some image flashed onto them Dec 23 00:20:04 but you've attempted to reflash it with an image that doesn't fit Dec 23 00:20:33 whatever used to be installed on your beaglebone will have been some really ancient image anyway Dec 23 00:21:34 I strongly recommend just booting from sd card from now (and erase eMMC). once you're more comfortable debian you may consider flashing a console image onto eMMC Dec 23 00:21:35 ok my other bbb just booted i going to see what was on that one Dec 23 00:22:37 how do i check to see what software is loaded on this one Dec 23 00:23:01 cat /etc/dogtag Dec 23 00:25:18 Cloud9 GNOME Image 2013.06.20 Dec 23 00:26:27 i am going to load Debian 7.5 (BeagleBone Black - 2GB eMMC) 2014-05-14 Dec 23 00:26:37 why? Dec 23 00:27:00 Aw! Dec 23 00:27:19 Oh...I see you two are busy. Sorry. Dec 23 00:27:34 why is it so important to you to boot from eMMC instead of sd card that you'd want to run ancient and unsupported software? Dec 23 00:28:02 bbl Dec 23 00:28:15 (ancient, unsupported, and unmaintained) Dec 23 00:29:48 so that if for what ever reason someone else fires up the board without hitting the button they will have atlease something Dec 23 00:30:06 i will have to write out some directions on how to start this thing Dec 23 00:31:54 if you wipe eMMC it will boot from SD card without needing to hit any button Dec 23 00:32:22 the button is only needed if there's an ancient image installed on eMMC Dec 23 00:32:30 kindof a spinoff from that comment : what's the smallest working modern image ? I think you mentioned a console build Dec 23 00:33:00 Debian 7.5 (BeagleBone Black - 2GB eMMC) 2014-05-14 same size a few years newer Dec 23 00:33:40 Ant__: please just wipe eMMC and boot from sd card Dec 23 00:33:42 ok, but nothing current ? Dec 23 00:33:52 artag: console image is small yes Dec 23 00:34:22 artag: the console images are smallest ones from rcn Dec 23 00:34:37 you could strip them down further probably, but not by much Dec 23 00:35:21 artag: presumably image builders like yocto can produce even smaller images Dec 23 00:37:18 Ant__: it is perfectly normal to boot an embedded system from sd card... there are plenty that don't even have internal storage (e.g. the pocketbeagle and the raspberry pi family) Dec 23 00:38:05 Ant__: a debian 7 system from 2014 is not something you want to inflict on yourself or others Dec 23 00:38:23 I can vouch for that fact. Dec 23 00:38:54 so that would be https://elinux.org/Beagleboard:BeagleBoneBlack_Debian#microSD.2FStandalone:_.28stretch-console.29_.28All_BeagleBone_Variants_.26_PocketBeagle.29 ? Dec 23 00:39:34 artag: yep Dec 23 00:40:24 I don't necessarily want the smallest possible, but the iot version seems to take a fair bit of the 4G. If I want to add a bunch of things (I was asking about ROS the other night) I might end up removing a load of stuff Dec 23 00:41:14 installing stuff you need is often easier than removing stuff you don't need Dec 23 00:41:30 and yeah like I said I don't know how the iot image managed to get so bloated Dec 23 00:41:32 obvious candidates given my application would be the ide, node-red and related stuff but not sure what to hi next Dec 23 00:41:39 *hit Dec 23 00:42:00 if you're going to remove all that, you're probably better off starting with the console image and installing things you do want Dec 23 00:42:16 looking at the onboard webserver I'd say a fair bit went on documentation and intro stuff Dec 23 00:43:02 which is kind of a nice start but also not wanted for a runtime system Dec 23 00:43:10 imho anyway Dec 23 00:44:04 yeah, I'm used to starting with a debian minimal image and adding what I need Dec 23 00:44:36 then you'll definitely want to start with the console image rather than the iot image Dec 23 00:45:07 i will give it a go. nothing like trying it out to see how it goes :) Dec 23 00:46:05 last time I started with a clean image I grabbed the then-latest stretch-console, removed unnecessary crap, upgraded to buster, switched to systemd-networkd as network manager, and then start installing things Dec 23 00:47:30 i haven't decided what I think of connman yet. documentation seems a bit spares but systemd isn't known for that either Dec 23 00:48:22 systemd has pretty detailed manpages Dec 23 00:50:01 that's encouraging. i don't have a lot of luck with things from that stable but I'm used to sysV and bsd Dec 23 00:50:39 wasn't broke, didn't ned fixing imho but I seem to be in a minority Dec 23 00:50:41 https://pastebin.com/AeZ5tvRL something like this is a fine config if you use dhcp Dec 23 00:51:14 I like systemd and things like socket activation Dec 23 00:51:43 that seems straightforward enough. where does the wifo login go ? Dec 23 00:52:41 wifi networks (incl credentials) ? /etc/wpa_supplicant/wpa_supplicant.conf Dec 23 00:52:47 not systemd-networkd's concern Dec 23 00:53:50 once wpa supplicant connects to a wifi network, the link comes up and then systemd-networkd will do its thing (just like plugging in an ethernet cable) Dec 23 00:54:30 ok, seems reasonable Dec 23 00:57:18 i've never seen a decent explanation of wpa supplicant tbh but maybe I haven't looked hard enough. when it first came along it was pretty flaky, then the various network managers arrived and made it worse (!) and then they settled down a bit. haven't had to tweak at the wpa supplicant level since then Dec 23 00:57:55 a good while back I had tons of trouble getting connected to an eduroam network using gnome network manager Dec 23 00:58:20 then I tried just using wpa_supplicant directly, with just the bare minimum config, and it worked instantly Dec 23 00:58:31 i do occasionally do an upgrade on the laptop and it all goes pearshaped. the last one of those pushed me from ubuntu to mint but that's not perfect either Dec 23 00:59:12 unfortunately I don't qualify for an eduroam id :( Dec 23 01:01:37 https://pastebin.com/raw/tvHmXBQc Dec 23 01:01:53 yeah I don't have it anymore either, it was a long time ago Dec 23 01:03:09 ta. I seem to have totally bricked the network on this image by trying to clean out /var/lib/connman so it's time to try the console one .. Dec 23 01:03:21 it won't even talk usb widget Dec 23 01:05:37 https://pastebin.com/raw/4FDgJDS4 and this is the custom wpa_supplicant service file I use... I can't remember what I disagreed about in the default service file that made me use a custom one Dec 23 01:11:55 for usb network something like https://pastebin.com/raw/xmmH8KEG would be an example, and then put g_ether in /etc/modules Dec 23 01:19:06 all stashed away for the next run at it ... Dec 23 01:25:19 I got it! Dec 23 01:26:05 Security System w/ BBBW and Motor Control Over Wifi via Web Page! Dec 23 01:26:12 Finally. Time to take a breather. Dec 23 01:26:32 Now...what should these motors do? Dec 23 01:27:10 I need to scare the next intruder. Dec 23 01:27:29 I know. Swing a hammer? Dec 23 01:27:40 Thwack! Dec 23 01:31:01 i had a good laugh at https://www.youtube.com/watch?v=xoxhDk-hwuo today Dec 23 01:31:56 Youtube is something I stay away from. Off to watch YouTube for a change. Brb. Dec 23 01:32:25 I saw that on the national news. Dec 23 01:32:32 booted that console image. Dec 23 01:32:34 /dev/mmcblk0p1 836280 324492 452092 42% / Dec 23 01:32:36 Glitter bomb. hahaha. Dec 23 01:32:48 .. that's a bit more healthy Dec 23 01:33:36 the glitter bomb is awesome. it looks like a nuclear shockwave Dec 23 01:33:41 I was unaware of the "fart" spray. Dec 23 01:33:46 That is too much. Dec 23 01:34:09 Ha! Dec 23 01:42:04 That is funny. I watched it until the end. Dec 23 01:43:00 the thieves reaction is almost an anticlimax, after the leadup Dec 23 01:43:41 pity he didn't use a paintspray Dec 23 01:43:59 I know. I liked that the last guy stood and thought for a while. Dec 23 01:44:32 i wasn't really sure who that was. did he set it up twice ? Dec 23 01:44:40 artag: Did zmatt give you enough for setting up the BBB yet? Dec 23 01:44:47 I think so. I think he kept getting robbed. Dec 23 01:44:52 Busy street. Dec 23 01:45:41 yeah, I haven't needed the network stuff yet, the default is good enough for now but I'll probably need it later Dec 23 01:46:10 Oh. I was just wondering. I kind of interrupted him as usual. Dec 23 01:46:34 artag: Is this your first go-around w/ the BBB? Dec 23 01:47:59 i've had a couple knocking around for ages from when I had a look at bela. then someone asked me to build them a simple robot and bbb seemed a good candidate. Dec 23 01:48:20 i'd really like to get into pru programming but I don't really need it for this project Dec 23 01:48:25 Yep. They have a motorCape and other Capes now too. Dec 23 01:48:26 Oh. Dec 23 01:48:58 You can already use the two onboard L298 drivers for setting up motors for the bot if needed. Dec 23 01:49:18 The loadCape is cheaper but it is more GPIO than anything else. Dec 23 01:49:40 I have slacked on PRU so far. Dec 23 01:49:57 I came across some good info. a while back and then lost it. Aw! Dec 23 01:51:11 i'm actually using usb motor drivers - arduinos with h-bridges. it's not all that sensible for a bb, I could do a lot more onboard, but I'm a bit concerned I might have to put a much more powerful processor in. So I thought I'd use the bbb in a fairly generic way and if I have to do image processing it will be fairly easy to swap out Dec 23 01:51:59 Oh yea. It does. You can do speech recog. too w/ the BBB family boards. Dec 23 01:52:12 the big attraction of bb is the option to put everything on emmc instead of raspberyy pi's nasty unreliable sd card Dec 23 01:52:39 Right. eMMC! That is nice once things have completed and are ready for the final hoorah. Dec 23 01:52:58 but it's a nice excuse to learn a bit more about the bb Dec 23 01:53:14 I used a sabertooth w/ the BBB too. Have you heard of those boards? Dec 23 01:53:38 They put on some heavy duty caps and one of those Atmel processors. Dec 23 01:53:43 yeah, i possibly should have used that. I'm a bit of a sucker for home-growing everything but they look good Dec 23 01:53:52 It works w/ the BBB, too. Dec 23 01:54:41 artag: you can get industrial SLC sd cards Dec 23 01:54:57 one trap I nearly fell in was with the h-bridges. I used these lovely 30A things from ST but it turns out there's loads of chinese clones that have high rdson and get really hot Dec 23 01:55:56 30A! Dec 23 01:56:18 There are many h-bridges. Did you use a L298 or another one? Dec 23 01:56:29 wel, the raspberry pi traditionally had software problems with the sd driver that's partially handled by the gpu. I still don't trust it. And also they fall out of the slots. even on the bbb I keep picking the board up wrong and accidentally ejecting it. Dec 23 01:56:43 ah Dec 23 01:57:25 the bbb also doesn't latch the card. even without ejecting it you can simply pull it out Dec 23 01:57:49 spent a little while yesterday wondering why the bb was running okish but kept giving me I/o errors when I ran things from the shell Dec 23 01:58:00 turned out the sd card wasn't in ! Dec 23 01:58:10 lol Dec 23 01:58:59 https://www.st.com/en/automotive-analog-and-power/vnh2sp30-e.html Dec 23 02:01:24 https://www.ebay.co.uk/itm/272471330912 Dec 23 02:01:28 £2.88 ! Dec 23 02:03:03 But the sparkfun and pololu ones have genuine chips and cost more .. https://www.ebay.co.uk/itm/142502959299 Dec 23 02:03:33 zmatt do you happen to know how to remote desktop into bbb Dec 23 02:06:45 I never install a desktop environment onto a bbb Dec 23 02:07:27 a bbb isn't a desktop computer, and a desktop environment does not run comfortably on it Dec 23 02:17:23 i hae a few apps i would like to run in debian that require userinterface Dec 23 02:25:59 such as? Dec 23 02:26:50 if I really needed to run an x11 application on the beaglebone I'd probably use x11 forwarding Dec 23 02:32:18 X11 forwarding? IE set the display across your network? Dec 23 02:33:32 no just ssh -Y Dec 23 02:36:14 if you've never used it, just try it. ssh -Y to some linux system and run some simple x11 application like xeyes Dec 23 02:52:13 zmatt hmm it's been a while since I messed with X (because it's all done for me with a graphical shell) so sounds like something I should actually know instead of "huh" Dec 23 02:53:39 it does make sense after thinking about it. learn something I probably should already know all the time :D **** ENDING LOGGING AT Sun Dec 23 03:00:00 2018