**** BEGIN LOGGING AT Tue Nov 01 03:00:02 2016 Nov 01 04:16:14 Hiya! Really excited to get my beaglebone for christmas, but I'm just now reading about the BBB wireless... It seems like it's mainly swapping the ethernet out for a wifi/bluetooth interface, but I'm very confused about all this stuff about "Octavo Systems OSD3358 SiP" Nov 01 04:16:38 What is that stuff all about? Does that somehow change the performance of the board, or is that just related to the wifi/bluetooth stuff? Nov 01 04:17:12 The site is very nonspecific about that stuff :( Nov 01 04:18:46 its just the name for the chipset including the BBB processor and TI's wifi chip and the PMIC all on one die by the looks of things Nov 01 04:18:57 http://octavosystems.com/octavo_products/osd335x/ tells you a bit more about it Nov 01 04:19:37 hey zmatt .. wonder if they've got a better implementatin of those 3v3 rails!? ;) Nov 01 04:20:59 ah no the wifi package is separate Nov 01 04:21:15 so its just the arm7, pmic, ram Nov 01 04:22:26 Allie1122: it just means instead of 3/4 separate chips .. they use one bigger chip with the same functions on the same silicon die Nov 01 04:24:48 together with losing the LAN PHY .. they make enough extra board space for the wifi/bluetooth chipset Nov 01 04:25:02 its a cunning tradeoff Nov 01 04:27:04 does that have a benefit for performance though, veremit? Nov 01 04:27:32 Allie1122: probably not noticeably .. but there may be some minor improvements Nov 01 04:27:34 for my applications, i REALLY dont want to have wifi signals near my board, so the wifi feature is a no-go, but I'm worried now that the wifi has better performance due to this chip Nov 01 04:27:55 do you think they'll be re-releasing the BBB with that new chip? Nov 01 04:28:03 nope Nov 01 04:28:14 the am335x has a very long lifetime still associated with it Nov 01 04:28:23 this is my first prototyping board and i really dont want it to be obsolete when i get it :( Nov 01 04:28:27 its a mature design Nov 01 04:28:57 do you think they will stop supporting the old design with this new one? Nov 01 04:29:13 its a completely different board made by a different company Nov 01 04:29:18 there is no overlap whatsoever Nov 01 04:29:36 it shares some heritage, that's all Nov 01 04:30:27 i guess, this is all just kinda overwhelming :( Nov 01 04:30:38 i've never really dabbled with electronics before, it's exciting but worrying too Nov 01 04:30:45 na not really :p Nov 01 04:30:50 just take it one step at a time Nov 01 04:31:01 its a whole new world to explore :D Nov 01 04:31:18 i guess >.< i was looking to get this kit, idk if you think it might be worth it: https://www.adafruit.com/products/703 Nov 01 04:32:14 and yeah! it's really cool, and i'm excited to start tweaking and making stuff- i wanna use the BBB as a hybrid digital synth / microcontroller for external analog circuits, it seems like it was made for this almost :S Nov 01 04:32:39 i'm even considering using the PWM pins as cheap square oscillators xD Nov 01 04:32:48 you'll be good with any adafruit stuff .. they're good guys over there Nov 01 04:33:01 phew, good :) i like their site, but it's nice to hear they have a good reputation Nov 01 04:33:37 gosh, i wish it was christmas already! lol Nov 01 04:33:43 Ladyada (or Limor as she's otherwise known) is a very smart lady Nov 01 04:33:44 anyways, thanks for answering my question :) Nov 01 04:34:13 but yeah .. the original Black is selling fast, and shows no signs of going anywhere Nov 01 04:34:20 phew Nov 01 04:34:26 I have two :) Nov 01 04:34:38 i am curious about the performance differences, but honestly, the black is more than enough for my purposes Nov 01 04:34:53 but one is sittin as a demo unit .. the other is doing just network duties. But I look forward to doing something wth the PRUs one day Nov 01 04:34:55 i still laugh a little seeing the word "1ghz", since i've been living in audio-land for so long Nov 01 04:35:03 idk what i'll do with all those cycles! Nov 01 04:35:15 well .. linux occupies quite a lot :D Nov 01 04:35:23 i was worried about that ;-; Nov 01 04:35:29 how much would you say? Nov 01 04:35:31 I would advise against bare-metal programming :D Nov 01 04:35:50 just at this stage .. work up from an arduino perhaps if you're really keen on trying that Nov 01 04:36:21 linux shou;dnt be using more than 3-10% normally .. depending on your applications/install/etc. Nov 01 04:36:32 i didn't really wanna, but i was hoping i could boot supercollider (a synthesis program) on my BBB with a little CPU overhead Nov 01 04:36:35 oh, 10% should be fine Nov 01 04:36:44 i was worried it was taking up half the cycles or something >.< Nov 01 04:36:52 have you seen any tutorials on it yet? Nov 01 04:37:05 i have already read some people running SC on BBB successfully, so I know it's possible Nov 01 04:37:12 and yeah, been reading tons :D Nov 01 04:37:22 just never saw anyone say exactly how much linux uses of the cpu Nov 01 04:37:57 do the PRUs run independently of the main cpu? Nov 01 04:38:06 yes, the prus are separate cores entirely Nov 01 04:38:10 that's really neat Nov 01 04:38:16 doesn't the cpu need to talk to them though? Nov 01 04:38:20 they're special real-time processors Nov 01 04:38:24 or do they have their own interfacing? Nov 01 04:38:29 that's one bit i'm fuzzy on Nov 01 04:38:37 there's a dedicated interface .. its quite clever .. even I'm unsure Nov 01 04:38:50 seems like everyone is scared of the PRUs! :P Nov 01 04:39:02 aren't they basically just arduino-type things? Nov 01 04:39:06 the am335x actually have 4 'processors' inside .. the main arm7 a cortex-m3 or m4 'supervisor' and the two PRU cores Nov 01 04:39:11 yeah sort-of Nov 01 04:39:18 jus ... quite a lot quicker! Nov 01 04:39:23 haha, yeah Nov 01 04:40:10 i had been considering arduino for stuff before, but i found bbb and saw it was the best of both worlds, helping me experiment with analog circuitry and also have a nice portable digital synthesizer computer Nov 01 04:40:19 arduino seems very underpowered for my purposes Nov 01 04:40:37 though some people have managed some (very bleepy) musical instruments via arduino Nov 01 04:40:44 the coolest i have seen is Groovesizer Nov 01 04:40:55 https://www.youtube.com/watch?v=8N2JAbtbiMk Nov 01 04:41:12 basically just an arduino due afaik Nov 01 04:41:16 (though it has a shield) Nov 01 04:43:08 i feel like "working up" is a little dangerous for my purposes, since i dont know if i have the programming knowledge to optimize that much Nov 01 04:43:19 i think i'd end up hitting a wall with arduino quite a lot Nov 01 04:43:46 zmatt: looks like the OSD has better power control ;D Nov 01 04:54:18 nini! happy halloween! Nov 01 07:10:32 veremit: in what sense? I haven't studied it in detail but iirc my impression was that it copied more from the beaglebone than I would have liked :P Nov 01 12:43:01 hello Nov 01 12:43:07 i want to know about this site more Nov 01 12:43:18 /join Nov 01 12:46:34 seems he's a quick learner Nov 01 13:53:06 does anyone have any idea why qemu + git clone seems to be quite prone to fail when running image building scripts (RootStock-NG.sh)? Nov 01 14:27:11 hey, does anyone know how much power the BeagleBoard-X15 draws when idle? maybe just approximately. Nov 01 14:42:45 nbg, you'd have to ask someone who has one .. and that's not very many as yet Nov 01 14:56:06 Hi, Nov 01 14:56:39 I have problem with bone101 Nov 01 14:56:52 I can't see the web page on my beaglebone black Nov 01 15:00:44 http://imgur.com/a/yAUpA Nov 01 15:00:51 this is the problem Nov 01 15:17:41 zmatt: go research. The docs on the osd site are quite accessible and otherwise explicit :) Nov 01 16:43:59 Hello.! Nov 01 16:44:32 Holla! Nov 01 16:44:47 I'm building a kernel fro Beagle bone black with instructions from https://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-ARMCrossCompiler:GCC Nov 01 16:45:08 I want to add driver for codec WM8960 Nov 01 16:45:31 I can't understand why I can't select this driver in menuconfig. Nov 01 16:45:57 I'm working from Ubuntu 15.10 x64 Nov 01 16:46:16 how do you call menuconfig? Nov 01 16:46:59 when I'm running the script ./build_kernel.sh at certain point I reach menuconfig Nov 01 16:47:09 ah Nov 01 16:47:10 ok Nov 01 16:47:22 then it's just simply a depencency issue Nov 01 16:47:38 the driver probably depends on something you have not enabled Nov 01 16:47:57 ok Nov 01 16:48:32 this codec work with I2c and Mcasp Nov 01 16:48:47 I have enabled them both Nov 01 16:48:55 i'm not talking about the hardware Nov 01 16:48:58 i'm talking about the code Nov 01 16:49:10 go to the sources of the driver and look what depencencies it has Nov 01 16:49:21 there is a Kconfig file somewhere that handles that Nov 01 16:49:37 yes...I have it opened if fron of me Nov 01 16:50:37 But you point something I didn't think of before. Nov 01 16:51:02 That maybe I don't have all the include stuff Nov 01 16:53:42 Xabby76, in the menuconfig program, you can type "/" to search. When you find what you are looking for in the search, it will tell you what the dependencies are and what state they are in. You should be able to quickly see what is missing. Nov 01 16:53:43 Because regarding WM8960 this is the line in Kconfig : select SND_SOC_WM8960 if I2C Nov 01 16:55:32 Thanks a lot dlech.! Nov 01 16:56:02 I will try immediately.! Nov 01 17:04:46 This is what I got from search command : http://pastebin.com/vkWEw2uA Nov 01 17:07:21 Am I missing something.? Nov 01 17:11:52 Hmm... It says you have the required dependences, however, it usually shows a path of where the setting is in the menu. So maybe this is not something you can enable directly. What happens if you enable SND_SOC_P1022_RDK? Nov 01 17:13:02 I will try right now. Thanks a lot for your precious help. Nov 01 17:16:19 I have notice now that I haven't pasted all the items from search command. Nov 01 17:17:04 I will send again the right pastebin in few minutes.Sorry for the error. Nov 01 17:26:36 This is the right pastebin : http://pastebin.com/TqMKA7WX Nov 01 17:31:53 I have enable in menuconfig SND_SOC_P1022_RDK but in codec drivers nothing change. Nov 01 17:49:26 It is strange, because if I disable I2C by hardware the available codec list doesn't change. Nov 01 18:26:54 Xabby76, I don't understand it either. I don't see why this should not be in the menu. Nov 01 18:27:31 you could try adding CONFIG_SND_SOC_WM8960=y to your .config file manually Nov 01 18:28:01 If you run menuconfig after that, it will probably try to delete it, but it may be good enough to get it to compile. Nov 01 18:41:23 +1 dlech Nov 01 18:41:29 oh Xabby76 left Nov 01 18:41:39 man these TFT drivers are like, the worst absolutely worst Nov 01 18:42:43 on the other than, the amX has my favorite SPI drivers Nov 01 18:42:47 the variable word size is really nice Nov 01 18:42:54 SPI hardware? i guess i mean Nov 01 18:43:20 who in their right mind would okay 9-bit and 10-bit words for an interface device Nov 01 19:01:40 Hi, I am working with the PRU on the latest 4.4.27 Kernel and image for BBB, I would like to load the PRU using the uio_pruss driver Nov 01 19:02:17 I am getting the famous: prussdrv_open failed Nov 01 19:02:46 lsmod shows I have uio_pruss installed and running Nov 01 19:04:03 I removed all the device trees once booted up and loaded many of the BB-BONE-PRU* variants hoping those device trees would enable access to the PRU. Nov 01 19:05:41 Has anyone succefully loaded PRU's using the latest kernel and uio_pruss drivers? Nov 01 19:12:25 veremit: yes I had seen their docs, and I just checked it again to be sure: they copied the bbb power supply scheme, annoying pmic and poorly managed 3v3b power domain included, and I think they might possibly have made it worse Nov 01 19:13:34 it also seems they're oblivious to the fact that RTC-only mode has been defeatured by TI Nov 01 19:14:57 fed007: try poking the mailing list. I know rcn has been changing stuff but I don't know what the official way is to use uio-pruss currently Nov 01 19:17:28 veremit: it'll be hilarious if/when they bother to check how much power their little toy consumes in rtc-only mode Nov 01 19:17:58 official per who? :D Nov 01 19:18:11 doesn't TI officially want remoteproc only? ;) Nov 01 19:19:20 official per rcn and the BeagleBoard.org Foundation Nov 01 19:19:49 mostly rcn, since he's causing the breakage by changing the device trees :P Nov 01 19:20:00 @zmatt - Thx - I'll try that route and of course looks like the remoteproc is becoming the increasing strategy Nov 01 19:20:47 so is he breaking the default am335x-boneblack.dtb ? Nov 01 19:21:21 fed007: I haven't looked at remoteproc much, but it seems to me it's just adding more complication to the kernel while offering less actual userspace functionality Nov 01 19:21:23 I tried sending that to dts and enabling PRU = okay then compile back to dtb Nov 01 19:21:37 ew Nov 01 19:21:52 Also that kernel seems to load something called cape_universaln Nov 01 19:22:10 that has a ton of pin exclusives... Nov 01 19:22:11 yes, joy Nov 01 19:22:27 so I am officially in PRU hell :) Nov 01 19:22:45 I got jkridner to start on http://elinux.org/Beagleboard:BeagleBone_Debian_Image_Migration and ended up writing most of it myself in the end Nov 01 19:23:13 but I don't actually really know how new users are supposed to accomplish all sorts of stuff in the current images Nov 01 19:24:09 OK here is another theoretical device tree question: If I " echo - #" all the slots until empty then load BB-BONE-PRU-01 should that not be enough to enable? Nov 01 19:24:11 I've gone my own ways and haven't really kept track of all the changes Nov 01 19:24:48 or is there some default dtb in mem that is cock blocking my PRU penetration? Nov 01 19:25:19 see the wiki page I just linked. you can disable cape universal in your /boot/uEnv.txt, although it will also get automatically disabled if you configure any other overlay to be loaded at boot Nov 01 19:26:15 checking your link... Nov 01 19:26:31 which kernel have you had the most success with? Nov 01 19:26:36 the thing is that the default DT now no longer actually includes the stuff needed for uio_pruss nor remoteproc to work, instead leaving them both in comments with "Pick one" above it (or something) Nov 01 19:27:40 And that choice should be as simple as loading an overlay . "should be" Nov 01 19:28:19 I've used most major releases since 4.1, currently on 4.8. but like I said I've been going my own way: custom kernel config and patches, and entirely custom device trees. I don't use overlays myself Nov 01 19:28:33 in theory yes, if he provided one. Nov 01 19:28:37 haha -> Help! My overlay won't load! There's just this weird cape in slot 4. Nov 01 19:31:22 I have tried , but is dtb= the spot for pure default override? Nov 01 19:31:47 dtb= specifies the main device tree yes Nov 01 19:31:47 in 4.x uEnv.xt Nov 01 19:33:00 since the dtb is loaded by u-boot that part is actually not dependent on kernel version Nov 01 19:33:31 @zmatt so essentially - going your own way is to bake in a module of sorts? Nov 01 19:33:41 ? Nov 01 19:33:50 bake what in a module? Nov 01 19:33:51 you must still need a default DT right? Nov 01 19:34:05 you need a DT Nov 01 19:34:35 normally that would be a complete DT for your hardware (or at least the parts the kernel cares about) Nov 01 19:36:14 ok does the DT need to exist to bootstrap eMMC linux loader etc..? Nov 01 19:36:42 or is it really applied after the OS is brought up? Nov 01 19:36:45 ?? Nov 01 19:37:22 So we have 4GB on board flash... Nov 01 19:37:54 the DT is needed to bootstrap I am guessing Nov 01 19:38:07 u-boot loads it along with the kernel and initramfs (if you use one), then starts the kernel, passing it pointers to the dtb (with some small modifications), the initramfs, and the kernel parameters Nov 01 19:38:43 linux needs it to know what hardware is out there, which drivers it needs to load, and those drivers need it for their configuration Nov 01 19:39:08 OK - I was wondering what the min DT could be Nov 01 19:39:28 then script in the overlay needs Nov 01 19:40:51 why would want that though? that's kinda reverse thinking... the main dtb contains as much as possible and only stuff that simply isn't known in advance (like what cape you plugged into your bbb, if any) is dealt with using overlays Nov 01 19:41:46 overlays are more limited than the main DT, have poor error checking, and tend to trigger kernel bugs since it was originally never designed to handle runtime changes to DT Nov 01 19:42:57 good point Nov 01 19:43:06 does am335x-boneblack-nhdmi-overlay.dtb mean no hdmi? Nov 01 19:43:17 there is that 'n' Nov 01 19:43:39 I am using the disable directive right now Nov 01 19:44:09 http://elinux.org/BeagleBone_DTBs Nov 01 19:44:28 nhdmi means hdmi is enabled, but not audio-over-hdmi Nov 01 19:45:39 OK - I keep thinking that the only reason the pruopen fails is because somehow the hdmi is still active Nov 01 19:45:53 they don't conflict Nov 01 19:46:15 ok - only for certain I/O Nov 01 19:46:38 I thought the default open pin was in conflict Nov 01 19:49:15 pru doesn't have default pins, and doesn't necessarily need any pins whatsoever Nov 01 19:49:43 It has hookers and blackjack too Nov 01 19:49:44 also its pinmux options (if you do need pru gpios) are all over the two expansion headers Nov 01 19:52:48 i heard pru... fed007: what you working on? Nov 01 19:55:35 Well I am first just trying to blink and LED - then I am off to World Domination Nov 01 19:56:45 To recap trying to get past pruopen 'failed' on the 4.4.27 kernel Nov 01 19:57:37 @zmatt isn;t there some mailbox pin the say "look out here comes some bytes" ? Nov 01 20:00:19 fed007: ?? Nov 01 20:08:59 sry - on the phone.... Nov 01 20:22:53 Ok I am going to give things a shot by trying the default DT - not universal , see if that gets better Nov 01 21:21:18 hi there channel :-) Nov 01 21:28:53 dury: you're aware that channels are inanimate concepts and don't usually respond to "hi" ? ;-) Nov 01 21:38:47 hi there zmatt, how do you doing there? hope pretty good Nov 01 21:40:09 zmatt is doing excellent thank you for asking Nov 01 21:40:13 LOL Nov 01 21:40:14 wat Nov 01 21:40:51 zmatt, listen got my bbb plugged by usb is running jessie in it how do I access to it? Nov 01 21:41:53 ddrown, can you help in case zmatt can't, please Nov 01 21:42:22 dury, zmatt says to check this document http://beagleboard.org/static/beaglebone/latest/README.htm Nov 01 21:44:56 ddrown, did it but didn't mount the eMMC flash drive, so? Nov 01 21:45:28 dury: do you see a usb network interface? Nov 01 21:46:24 I assume you've plugged the BBB's usb into your computer's USB Nov 01 21:49:21 ddrown, the connector behind the flashing lights to computer's usb, is it clear, do u understand what I mean? Nov 01 21:51:30 ddrown, mmmm... you know... the usb cable that comes with bbb Nov 01 21:51:48 dury: like this? http://beagleboard.org/static/beaglebone/latest/Docs/images/Black-on-USB.jpg Nov 01 21:52:17 yeap that's right ddrow :-) Nov 01 21:53:10 ok, the computer you've plugged it into, does that computer have the drivers installed listed as step #2 on http://beagleboard.org/static/beaglebone/latest/README.htm Nov 01 21:55:11 ddrown, running linux Nov 01 21:55:44 dury: ok, on your linux computer, do you see the beaglebone network interface? Nov 01 21:56:42 ddrown, you mean the ethernet connector or what? Nov 01 21:57:27 dury: the text right under "step #2" is "Install the drivers for your operating system to give you network-over-USB access to your BeagleBone. " Nov 01 21:57:53 dury: the beaglebone defaults to having a virtual network over the usb port Nov 01 22:00:37 ddrown, ok, sorry... be patient please I'm doing my best.....mmmm... how can I see bbb network interface typing a command in terminal maybe Nov 01 22:01:02 dmesg - will show you your kernel messages Nov 01 22:01:09 ifconfig - will show a list of network interfaces Nov 01 22:01:34 right one second please Nov 01 22:02:08 can be also /sbin/ifconfig too? Nov 01 22:03:39 yes or /usr/sbin/ifconfig which isn't always in your default path Nov 01 22:08:14 ip addr is its preferred replacement actually Nov 01 22:10:49 it's in eth1 Nov 01 22:11:26 just plugged the network cable Nov 01 22:12:17 or not needed it? Nov 01 22:14:11 so what should I do now , please Nov 01 22:14:31 what's the next step on that document I linked? Nov 01 22:15:54 I don't know exactly if the drivers are right or not Nov 01 22:16:08 ok, do you see an ip address on eth1? Nov 01 22:17:31 not ip address on eth1, jeeesuuusss, :-( Nov 01 22:18:19 ok, so either your OS didn't use dhcp on that interface or that's not the right interface Nov 01 22:23:09 ddrow, I guess the os on eMMC have to login in 'cause the other day booted bbb with display plugged, mouse and keyboard and asked me to login Nov 01 22:23:54 I don't know if it's something to do with it Nov 01 22:24:15 see what I mean? Nov 01 22:25:25 not amstrong running on it Nov 01 22:25:54 what distro do you have on it? Nov 01 22:26:13 jessie Nov 01 22:26:23 debian jessie Nov 01 22:27:15 the beaglebone debian jessie image still has the usb virtual network configured Nov 01 22:28:39 on your desktop/laptop, have you configured the virtual usb ethernet nic? Nov 01 22:29:24 how can ddrown? Nov 01 22:29:40 tell me the command please Nov 01 22:29:41 there's usually a control panel Nov 01 22:30:00 you can do it from the command line too, but that'll be harder Nov 01 22:31:00 let's do it by control panel then Nov 01 22:31:37 where I have to go or to check Nov 01 22:31:55 I'll leave that up to you to figure out Nov 01 22:41:32 in my opinion when I switch on or plug BBB by usb to desktop/laptop it starts jessie in it, but doesn't start a session... you know what I mean? Nov 01 22:43:11 it doesn't login default session or I'm wrong? Nov 01 22:46:02 busy ddrown? Nov 01 22:46:37 gesssss I'm doing my best though... can u assist me, please Nov 01 22:49:21 yes, I have some other things to do now. I can't help at the moment Nov 01 22:50:59 ddrown, ok would you be here tomorrow to continue?. I would really appreciate it though :-) Nov 01 22:52:50 ddrown, I understand, no worries, catch you tomorrow, right? Nov 01 22:53:02 I cannot promise I will be available tomorrow Nov 01 22:54:05 anyway, thanks for your help and advice really appreciate it believe me cheeerss there good night :-) Nov 01 22:55:17 I recommend you read up on your distribution's documentation on how to setup network cards Nov 01 22:59:49 ddrown, the only way I learn is if somebody drives me rightly and then I see how it works and read about it .... sometimes for me the documentation it's hard to understand Nov 01 23:00:57 anyway thanks indeed for your patient, ddrown. good night Nov 02 00:07:11 I wonder how many people realize gpio numbers (e.g. as used by the sysfs interface) are not at all guaranteed by the kernel to be consistent on every boot (let alone exactly 32 * bank + subpin) Nov 02 00:08:21 * rajesh did not realize that... Nov 02 00:09:57 although to be fair, it is hard to imagine how the kernel could manage to probe the on-chip gpio controllers in the wrong order (or after other gpio providers, if any) Nov 02 00:18:13 zmatt: i'm using debian on a bbb, and looking at the results of `apt-cache search linux-image`... big list. Nov 02 00:18:42 obviously Nov 02 00:18:50 what is the general recommendation for choosing one image over another? or, where do we read more about supported features? Nov 02 00:19:19 apt-cache policy shows them coming from the rcn repo, but where do people go to read up about the changesets? Nov 02 00:22:02 so, you'll probably want the latest kernel from a 4.X-bone or 4.X-ti kernel series, for some value of X Nov 02 00:23:26 the number after -bone or -ti is the revision of the patchset... you can check the git log of that to know what happened I guess. for the kernel version itself I would suggest any generic information resource about what's been happening in the kernel Nov 02 00:23:27 and when i tried the latest 4.8.X-ti kernel, things didn't show up under /sys so downgraded to 4.4.X... but it sure would've bee great to see some notes... Nov 02 00:23:49 there's no 4.8-ti series yet last time I checked? Nov 02 00:24:27 zmatt: thanks but i understand what the naming convention is. i was wondering about any notes to go along with. Nov 02 00:24:47 zmatt: you're right, 4.4.9-ti-* is the latest -ti Nov 02 00:25:06 i must've tried the -rc-bone Nov 02 00:25:13 so, 4.4-ti is the current long-term-support focus Nov 02 00:25:29 although 4.4-bone should be fine too Nov 02 00:26:25 when a new major release comes out, it may take time for local patches/customizations to be forward ported to that kernel Nov 02 00:27:56 I've so far not run into anything missing in 4.8 that I care about, but that'll vary depending on use case of course Nov 02 00:28:02 4.8-bone I mean Nov 02 00:28:31 zmatt: i was trying eqep... Nov 02 00:28:43 yeah that one I noticed is missing Nov 02 00:30:19 although I don't actually care since I'd probably use uio instead the kernel driver anyway, especially since iirc it really wasn't a great driver Nov 02 00:41:59 ohey the new gpio device interface (intended to replace the deprecated sysfs gpio interface) has arrived... now let's see how badly it sucks Nov 02 00:42:53 where has this been mentioned? Nov 02 00:43:36 I first spotted it in the source code (and already knew it was coming), I'm checking out some slides about it right now Nov 02 00:43:41 ok Nov 02 00:46:20 lol this completely sucks Nov 02 00:47:56 is this something ti is releasing? Nov 02 00:48:31 no it's made by Linus Walleij who works for Linaro Nov 02 00:49:04 ok Nov 02 00:51:35 have they finally removed userland gpio access abomination? Nov 02 00:53:13 ds2 userland gpio access is fantastic Nov 02 00:53:20 it is downright dangerous Nov 02 00:53:26 well the bbb is an educational tool Nov 02 00:53:27 and stupid Nov 02 00:53:33 for most people anyway Nov 02 00:53:41 so it serves that purpose Nov 02 00:53:54 why not also include a bottle of parrafin/kerosene along with a match Nov 02 00:54:09 ds2: the newly introduced mechanism is pretty dangerous indeed Nov 02 00:54:32 gpio access needs to be thought out very carefully Nov 02 00:54:37 what I'm currently using otoh (enhanced sysfs + gpio-of-helper + udev) is very safe Nov 02 00:54:48 no userland gpio access is safe Nov 02 00:55:04 that's a nonsense statement Nov 02 00:55:11 uncoordinated gpio access == large potential for frying hw Nov 02 00:55:11 its more important that it be accessible to casual users than for it to be safe for your application Nov 02 00:55:22 and if they fry hardware, thats a good learning experience Nov 02 00:55:30 casual users who do not understand hardware implications is the problem Nov 02 00:55:36 yes thats why we have the BBB Nov 02 00:55:37 ds2: you didn't say uncoordinated, you said "no userland gpio access" Nov 02 00:55:44 so they can start educating themselves Nov 02 00:55:48 zmatt: yes, I said no userland gpio access Nov 02 00:55:59 using gpio-of-helper requires declarations in DT for all gpios to be exported to userspace Nov 02 00:56:00 cuz userland is swarming with folks who do not understand what it implies Nov 02 00:56:03 including pin direction Nov 02 00:56:33 it isn't just direction Nov 02 00:56:46 gpios connected to regulator enables are just as dangerous Nov 02 00:56:48 I've enhanced it and included some udev rules so I now have /dev/gpio/$label symlinking to the right pin so you also don't have to figure that out (and retain portability if you move the pin) Nov 02 00:56:59 yes, so then don't export those to userspace sillybeans Nov 02 00:57:20 you don't know that w/o knowing the HW hence the danger Nov 02 00:57:51 some parts can get rather sensitive to power sequencing Nov 02 00:57:56 and describing the HW is exactly what DT is for Nov 02 01:00:33 note that ajp_ is of course also right, you still want gpio directly exportable and controllable for experimentation Nov 02 01:00:51 yes this can be harmful, but no more than sticking wires into your hardware Nov 02 01:01:03 ah! good point Nov 02 01:01:31 i learned a lot from it myself, so thats why i defend it. i think ds2's perspective on it though is an important next thought Nov 02 01:03:31 for final applications where the hw probably no longer changes this shouldn't be used anymore except perhaps for debugging, but only root can export gpios anyway. The real thing that needs to change is teach people how to grants limited rights in a safe way and avoid running services as root just to be able to access some hardware Nov 02 01:03:43 *grant Nov 02 01:04:28 that being said, when it comes to the dirty cow exploit, i got luck since i had a natural distrust of giving casual users ssh access anway Nov 02 01:04:56 so i'm a proponent of the "total security" type attitude where you try to shut off systems as largely as possible when you dont need them Nov 02 01:05:09 and make things as restrictive as possible Nov 02 01:05:43 so keeping gpio out of userland for a real product seems like a great idea to me, tbh Nov 02 01:05:52 its just, not reasonable for the bbb Nov 02 01:06:01 i actually support bonescript, to an extent Nov 02 01:06:15 it can make perfect sense for a userspace process to need to control some particular signal Nov 02 01:06:48 yeah, it really does Nov 02 01:06:48 root is given out too indiscriminately Nov 02 01:06:54 root? Nov 02 01:06:58 where? Nov 02 01:07:08 you can dress it up as a led or a button or whatever, but I really don't get the point of all these dressed-up-gpio subsystems that don't actually fundamentally change the right being granted Nov 02 01:07:13 these days, root isn't even trusted. i.e. see the google android abomination where they use SELinux to neuter root Nov 02 01:07:39 i know a guy that works on hardware security for SELinux Nov 02 01:07:42 if you are doing a real product, do a proper kernel driver Nov 02 01:07:45 he strikes me as a drug user Nov 02 01:07:51 tbh Nov 02 01:08:12 i guess i mean, hardware security for android Nov 02 01:08:22 look hisname@google.com Nov 02 01:08:24 ds2: yes, let's move our whole javascript application into the kernel and do away with userspace entirely! :D Nov 02 01:08:29 LOL Nov 02 01:08:35 zmatt: they're working on it Nov 02 01:08:46 whoever "they" are, they're morons :P Nov 02 01:09:15 https://github.com/shelljs/shelljs Nov 02 01:09:23 javascript really has an insane camp Nov 02 01:09:32 ds2: and you really need to cut the nonsense about pointlessly moving code that has no need to be privileged into kernelspace Nov 02 01:09:55 i had to start learning javascript and its such a weird transition from C Nov 02 01:10:04 the scoping and types are all kinda of what the hell? Nov 02 01:10:24 but i'm honestly really excited from webassembly Nov 02 01:10:25 ajp_: at least in ES6 you can finally ditch the idiocy known as 'var' in favor of let Nov 02 01:11:12 i think they make sincere improvements, but javascript is undisciplined by itself. i love having learned C as well as I did. javascript is a hacked together language Nov 02 01:11:39 and i think that kernel is some of the most elegant code Nov 02 01:11:53 I don't know what kind of face I made when I first learned that javascript 'var' declarations leak out of the block they're in (all the way up to function scope, including the code before the declaration) but it must have been quite a sight :P Nov 02 01:12:27 scope in javascript is all messed up, especially because of first class functions, and the way function scoping works Nov 02 01:12:32 its really the worst Nov 02 01:12:51 and if you watch the v8's heads talk on improving your speeds in javascript, he basically says code javascript as if you're writing C, and the V8 optimizer will treat it as such Nov 02 01:13:11 but well, they fixed the scoping stuff at least :P Nov 02 01:13:11 and if you do, you can get to half the speed of actual C Nov 02 01:13:40 now you're still stuck with utf16 strings whose .length is actually the number of utf16 code units necessary to encode the string Nov 02 01:13:41 i'm not so confident.. Nov 02 01:14:12 i work with a lot of high level guys and they want to pass all data as ascii strings lol Nov 02 01:14:26 they'll pass a hex string in ascii, lol. 2 bytes per byte. Nov 02 01:15:38 I'd be slightly worried about putting arbitrary binary data into a js string also... probably just irrational though :P Nov 02 01:17:57 working with integer bitops is fun too Nov 02 01:18:01 node -p '4000000000 | 0' Nov 02 01:19:23 ugh, i gotta study more javascript Nov 02 01:19:25 i hate it though Nov 02 01:19:40 I avoid it when possible Nov 02 01:20:23 Code Composer Studio uses javascript (rhino) as scripting language embedded in their debug server... really a brilliant choice of language for that application domain Nov 02 01:21:20 it's really great to get an address printed to you as 1.07689e+09 or similar :P Nov 02 01:23:41 there is no need for JS to be in K space Nov 02 01:23:57 the kernel needs to expose a safe API appropriate for the device Nov 02 01:24:05 i.e. if it is a sensor, it should spit out sensor data Nov 02 01:34:41 ds2: yes, and if that sensor is sensing something boolean, a GPI is probably a perfectly appropriate API Nov 02 01:35:17 likewise a GPO for some boolean control output Nov 02 02:12:04 hey guys i was wondering if anyone could help me out with some issues I seem to be having Nov 02 02:13:15 basically my bbb wont boot from the microsd after i install some packages Nov 02 02:13:24 the screen just flickers Nov 02 02:13:29 any ideas? **** ENDING LOGGING AT Wed Nov 02 03:00:00 2016