**** BEGIN LOGGING AT Sat Jun 11 02:59:58 2016 Jun 11 03:21:19 wow the tutorials online really really aren't keeping up with all the advancements made for embedded linux Jun 11 03:21:56 i'm googling 2012 Jun 11 03:26:40 it's even worse because when i had my 2gb bbb all of those tutorials worked... Jun 11 06:11:25 ahh! av500 i figured out who you are. Jun 11 06:11:55 av500: hello Jun 11 06:14:54 <_av500_> mph: hi :) Jun 11 06:15:03 <_av500_> your nick so short Jun 11 06:15:11 you want one of my cheap arse pnp machiens. Jun 11 06:15:12 lol Jun 11 06:15:26 <_av500_> mph: I need some PCBs with WS2812B fabbed Jun 11 06:15:35 oh. Jun 11 06:15:50 btw, i've been banned from this channel for some reason a long time ago Jun 11 06:16:01 i think my client was massively misbehaving Jun 11 06:16:06 <_av500_> ah Jun 11 06:16:23 i used the web interface to join today Jun 11 06:16:51 Error(474): #beagle Cannot join channel (+b) - you are banned Jun 11 06:17:08 username is mrpackethead Jun 11 06:17:11 can you fix that for me Jun 11 06:17:12 ? Jun 11 06:17:18 <_av500_> sec Jun 11 06:19:33 hey... Jun 11 06:19:36 welcome back Jun 11 06:19:56 <_av500_> hi Jun 11 06:20:03 what did i miss in the last forever Jun 11 06:20:09 lol. Jun 11 08:10:08 mrpackethead: the x15 has been released, but based on a keystone processor instead of the am572x and running Windows EC instead of linux ;P Jun 11 08:11:20 no not much really Jun 11 08:16:46 yes that's what happens when you own a macbook and it triggers a wifi wake up every 5min Jun 11 08:18:07 or worse. Jun 11 08:18:23 whhhat is going on you guys are talking but i only see veremit and zmatt sending messages Jun 11 08:19:28 that's all there is .. you have to scroll a long way to get the signal from the noise :D Jun 11 08:19:53 I've got my join/part/quit messages shaded so feint .. I can actually see the chat :D Jun 11 08:26:31 so i disabled my universalcape and hdmi, stuck a probe into my lcd clock and vsync Jun 11 08:26:38 and loaded this through config: http://pastebin.com/G79HAMn3 Jun 11 08:27:06 i was expecting (hoping?) to see some action but saw nada. modprobe tilcdc was fine Jun 11 08:27:41 the fragment is overriding whats defiined for lcdc in am33xx.dtsi Jun 11 08:29:51 meh for what its worth i just started reading linux drivers 3 Jun 11 08:33:49 i'm reading through tda998x_drv.c to see how they do things... Jun 11 08:37:48 i guess ill try to imitate tda998x drm specific code, id like to see a /dev/fb0 or /dev/dri but nothing so far.. i'm too new to this Jun 11 08:56:53 ayjay: what are you trying to do? the rgb signals come right out the header .. we used a serializer chip to generate LVDS from them years ago from a BBB .. worked nicely with a 10" chalkboard tft touch display :D Jun 11 08:57:25 hello Jun 11 08:57:46 would it be bad if i plug in a sdcard while the bbb is running? Jun 11 08:57:59 i don't want to boot from it, i just want to store data Jun 11 08:58:27 belgin: no it won't mount it unless you ask it .. the interface should be hot-plug safe Jun 11 08:58:59 i know. Will it detect the device and I'll only have to run `mount'? Jun 11 08:59:26 i guess i'll just see for myself Jun 11 08:59:28 should do . never had any problems with uSDs for storage on any system .. Jun 11 08:59:28 thanks Jun 11 09:00:24 after all, its only like plugging into your pc ... Jun 11 09:00:27 one more thing. i recently bought an intel compute stick and it only lasted for a month or so before breaking, meanwhile the bbb is still going strong after about half a year of non stop use Jun 11 09:00:46 hooray for bbb Jun 11 09:00:48 bbb is based on an industrial cpu .. doubt your stick is .. ;P Jun 11 09:01:07 cool Jun 11 09:01:12 veremit: i'm just trying to get RGB off the headers w/o the hdmi on Jun 11 09:01:37 so i set up the fragment and stuck a probe in the clock/vsync to check for activity but found none Jun 11 09:01:45 ayjay: you'll still need to 'activate' the hdmi, even if you don't use the framer and the physical port .. you -need- the graphics output :p Jun 11 09:02:13 otherwise you have to configure the graphics all manuallyh Jun 11 09:02:24 so that's kinda of what i'm trying to understand Jun 11 09:02:33 is there a good book to teach me the inner workings of bbb (i'm not an electrician, i just write software for a living) Jun 11 09:03:03 the rgb signals go from the CPU->framer chip->hdmi port .. the headers have the same inputs as the framer chip .. so its easier simply to configure graphics .. ignore the framer chip, and go .. Jun 11 09:03:26 well my problem was controlling the mode Jun 11 09:04:08 cmdline=video=HDMI-0:1920x1080M-16 .. oh hell I never remember the syntax Jun 11 09:04:26 but you're restricted to distinct states right Jun 11 09:04:42 like i have a 1.8inch lcd, its like 325 X 140 or something Jun 11 09:05:11 also it shuts off automatically if it doesn't detect a connection... Jun 11 09:07:57 belgin: i would rely on amazon reviews Jun 11 09:09:02 ok Jun 11 09:10:51 FARKin HELL .. here we go . fibnally ... https://www.kernel.org/doc/Documentation/fb/modedb.txt Jun 11 09:12:01 * ayjay claps Jun 11 09:13:02 ayjay: the framer will probably have a headache, and not do anything .. but we don't care .. because we have another chip attached to the signals, which should understand it Jun 11 09:13:22 ayjay: I have a couple of SPI/I2C "tiny" displays .. but never used them yet. Jun 11 09:13:34 yeah i'm going to do hook up SPI next Jun 11 09:13:39 did you get a dma buffer going with that? Jun 11 09:13:55 I will probably shell out some day and get a 4 or 7" touch panel from 4D systems if I want something for an embedded .. or just use a 7" pi tft. Jun 11 09:14:13 ayjay: no need .. just use the framebuffer. I was using X, so no work to do there Jun 11 09:14:23 i mean for SPI? Jun 11 09:14:29 oh never tried. Jun 11 09:14:39 they're still in a box on a shelf somewhere Jun 11 09:14:42 i might finish looking into that TI driver.. i've really put a ton of work into it Jun 11 09:14:48 cool. Jun 11 09:14:48 but now that you've given me this link i feel bad haha Jun 11 09:15:02 because it would be a lot easier just to tap that signal... Jun 11 09:15:11 ahhhh whyyy Jun 11 09:15:11 someone was doing a Hitachi LCD display native driver using the TIcldc Jun 11 09:15:32 because you -can- .. problem is .. -should- you?! :) Jun 11 09:15:51 thing is though, you don't learn otherwise lol Jun 11 09:15:55 been th ere, done that before :D Jun 11 09:16:07 sometimes the exercise is more important than the outcome :) Jun 11 09:16:14 s/exercise/process Jun 11 09:19:27 no kidding... the kernel is huge those Jun 11 09:19:33 i mean drm alone is too intimidating Jun 11 09:20:01 drm is a maze Jun 11 09:22:00 it also doesn't help there tends to be multiple different interfaces, one of which is the current way to with the rest being "legacy", except which is which is often hard to tell and the current one seems typically the least documented Jun 11 09:22:09 yup .. even zmatt is intimidated byh the kernel . :p Jun 11 09:26:35 ugh i'm going to bed Jun 11 09:26:37 i think Jun 11 09:27:18 i really would rather have a ti driver that just runs that subsystem on the am335x and exposes whatever userspace interface is typical Jun 11 09:27:46 i think the hdmi driver that wraps the video-out audio-out, cec&edid&etc doesn't need to be so tied to the tilcdc Jun 11 09:28:01 but who am i Jun 11 09:28:05 goodnight Jun 11 09:28:12 goodnight, ayjay Jun 11 09:28:16 I wonder how a "proper" sgx/lcdc pipeline is supposed to look like, since they basically run independently apart from needing to pass rendered buffers from sgx to lcdc, so that probably makes it like a multi-GPU setup and "should" use PRIME or something Jun 11 09:28:55 ayjay: well the config of lcdc and the hdmi framer, when used, are intimately tied together Jun 11 09:29:04 zmatt: dmabuf passing, very simple =) Jun 11 09:29:36 tomba: does it use that already? I remember at some point it still patches custom hooks into lcdc to latch onto Jun 11 09:30:10 depends on which kernel/userspace you mean, but yes, it uses that Jun 11 09:30:38 tomba: btw, if you want to see glitches and a kernel log spammed full of lcdc underrun errors... you need look no futher than the latest official ti linux sdk for am335x Jun 11 09:31:31 we're not seeing those... what kind of setup? Jun 11 09:32:42 running on a beaglebone at least glitches were constantly evident in the matrix gui, and if I killed that leaving just weston and opened up a terminal there and started resizing or moving it it really glitches completely to hell, at times looking like a tv image which lost its lock onto sync signals Jun 11 09:33:15 just the unmodified demo rootfs flashed onto sd card running on a standard beaglebone, hdmi screen attached Jun 11 09:33:22 ok. yeah, it's rather memory bandwidth limited. quite difficult to figure out good limits for pixel clock. Jun 11 09:34:00 but I do find it odd that BBB with the TI release would behave so badly, as that's supposed to be tested Jun 11 09:34:19 yeah, especially since the biggest culprit was easily identified Jun 11 09:34:31 it still has a cpuidle mode which unlocks the MPU PLL Jun 11 09:34:41 disabling that resolved 90% of the glitches Jun 11 09:35:22 why would mpu pll affect lcdc... Jun 11 09:36:12 dunno possibly because the latency coming out of idle causes a big pile of work to pile up? Jun 11 09:36:28 zmatt: could you send me an email about that. it's saturday, so I'm kind of not working =). tomi.valkeinen@ti.com Jun 11 09:36:28 I'm really not sure, I've never seen anything running on a beaglebone perform so horribly as that image Jun 11 09:37:02 well there's not really much more to it than what I just said, so you can just copy/paste this conversion and mail it yourself ;) Jun 11 09:37:52 ok, sure =). so what TI release is that? I'm not very familiar with those. Jun 11 09:38:40 funny thing is that I once had a standard sgx demo (an actual *nice* one, as opposed to the ones included with the sdk) running quite smoothly at 1280x1024 (snapshot of framebuffer: http://gerbil.xs4all.nl/bbb-sgx-magiclantern.png ) Jun 11 09:39:12 ti-processor-sdk-linux-am335x-evm-02.00.02.11 Jun 11 09:39:42 beaglebone black is an officially supported target Jun 11 09:39:42 thanks Jun 11 09:40:49 I also noticed the latest graphics sdk was still compiled with gcc 4.6... TI really should recompile those with the latest gcc, in my experience gcc code generation has improved quite a bit over time Jun 11 09:42:03 we're using gcc 5 for the 2016 releases. I don't know how those map to processor sdks. probably there should be a psdk out soon with 2016 base. Jun 11 09:42:59 latest release graphics sdk afaik is 5.01.01.02, which is old enough to have as highlight in the README: "NEW - 3.14 kernel support" Jun 11 09:43:36 haven't checked if the psdk sneakily includes something newer Jun 11 09:43:55 hmm... that psdk 2.00.2.11, does it have 4.1 kernel? Jun 11 09:44:34 stick to gcc 4.8 4.9 was buggy for arm .. 5.3 is still maturing .. Jun 11 09:44:35 "4.1.18-gbbe8cfc" Jun 11 09:45:05 bad case of YMMV though for gcc Jun 11 09:45:07 veremit: I've always used 4.9 without problem, and actually noticed the code generation improvements around 4.9/5.x Jun 11 09:45:25 I had memory leak problems with .. either 4.8 or 4.9 .. I'd have to check which Jun 11 09:45:35 on the linaro binary toolchains, mind. Jun 11 09:45:53 I don't doubt my gentoo toolchains will probably be ok :D Jun 11 09:46:12 I use the debian standard cross-toolchains Jun 11 09:47:30 zmatt: well being the debian/systemd lover that you are .. ;P Jun 11 09:48:02 lol, I'm no debian lover, I'm just used to it Jun 11 09:48:54 APT is the worst, except that my brief encounters with its competitors generally turned out even worse Jun 11 09:49:49 zmatt: have you tried slackware? Jun 11 09:50:08 the low point being YUM in the time it didn't even track which packages were explicitly installed and which were implicitly pulled in as dependency, which made removing stuff without littering your system with garbage a real pain in the ass Jun 11 09:50:39 and most YUM developers apparently didn't see this as a ball-stabbing glaring usability flaw Jun 11 09:51:26 old colleague used to swear by slack Jun 11 09:51:37 seemed ok to me .. but I was still a linux n00b at that point Jun 11 09:51:44 bbut the tcl scripts were pretty solid we were using Jun 11 09:51:54 much better than the windows drivers and associated mess .. phew! Jun 11 09:52:25 I use buildroot =). but then, I don't exactly use it, but develop with it. Jun 11 09:52:49 belgin: I don't think I have... obviously it'll have been in the lineup last time I went browsing for a distro... that was quite a long time ago however so I can't really say why I never tried it Jun 11 09:52:50 yea I've yet to try out buildroot .. seems a cool tool Jun 11 09:53:16 tomba: is it another rolling-release like gentoo or ? Jun 11 09:53:19 zmatt: it doesn't come with a package manager Jun 11 09:53:22 I tried arch once, but got tired of "customizable" apparently meaning "no sensible defaults WHATSOEVER" Jun 11 09:53:34 woops, sorry tomba means belgin Jun 11 09:53:35 ah. Jun 11 09:53:42 package managers are useful Jun 11 09:53:44 belgin: that is not a win Jun 11 09:53:50 portage works well in gentoo Jun 11 09:54:01 zmatt: you can easily install one if you want Jun 11 09:54:24 belgin: I don't see a package manager as something to be stapled onto the side Jun 11 09:54:32 ah ok Jun 11 09:54:39 philosophically I'm kind of interested in NixOS Jun 11 09:56:10 for embedded it would also be useful to have good cross-package-manager or something... where you keep the image being built strictly needed from all the crap needed to build it Jun 11 09:56:55 zmatt: gnu guix has somewhat of a more advanced scriptable package manager Jun 11 09:59:28 so the thing is there are way too many things to try in theory, and even those that happen to turn out to not suck will take quite some time to start paying off Jun 11 09:59:52 hene it's difficult to break away from the "sub-optimal but at least I mostly know my way around the place" debian Jun 11 09:59:53 definitely "better the devil you know" ... :P Jun 11 10:00:14 problem is .. you do get backed into a corner unless you do try new things Jun 11 10:00:25 I have on occasion Jun 11 10:00:33 you never know when that breakthrough which will save you time is gonna hit Jun 11 10:00:49 I tried SmartOS on a server and really liked it, even though it's quite alien Jun 11 10:05:05 I still should give fedora another try or so on my laptop, since plenty of interesting stuff seems to originate there Jun 11 10:05:24 but I'm still hurting from the YUM debacle last time Jun 11 10:05:26 arch is moderately popular too Jun 11 10:05:33 yum is a horrible system imho Jun 11 10:05:49 redhat/centos use that too Jun 11 10:05:50 11:53 < zmatt> I tried arch once, but got tired of "customizable" apparently meaning "no sensible defaults WHATSOEVER" Jun 11 10:06:00 ah Jun 11 10:06:22 if I want Linux From Scratch I'll just do that thanks Jun 11 10:06:23 :P Jun 11 10:06:24 you should probably give gentoo a spin really .. its a a learning curve .. but it works relally well .. I think Jun 11 10:08:09 ETOOMANYDISTROS Jun 11 10:08:10 :P Jun 11 10:08:33 there is only one .. the Right one :D hahaha Jun 11 10:08:58 veremit: can't you try out NixOS for me and see if it's any good? would save me some time ;) Jun 11 10:09:32 zmatt: sure, soon as you buy me a money tree ;) Jun 11 10:09:47 atomic upgrades would certainly be a breath of fresh air compared to APT Jun 11 10:10:01 I've got proper multilib too .. Jun 11 10:10:25 Nix is quite a bit more than multilib Jun 11 10:10:56 I hear a bit about noxois and guix .. seems to be the new fashioinbabble things .. Jun 11 10:11:35 * veremit rebuilds ffmpeg, mplayer and vlc .. Jun 11 10:12:15 crazy as it may sound .. my mplayer wasn't built with pulseaudio support Jun 11 10:12:34 it sucks mplayer2 stopped development... debian now went back to mplayer1 and now I get frequent weird errors that kill the audio stream Jun 11 10:12:50 look up mpv I think it is .. friend did some research Jun 11 10:13:14 or is that the -other- fork .. Jun 11 10:13:49 no idea, I don't really follow the politics of all this stuff, I'm just annoyed when a package which used to work fine is suddenly replaced by one that doesn't work fine Jun 11 10:14:10 but it still works well enough that I can't be bothered to do more about it than complain Jun 11 10:14:11 yeah there wasa lot of fallout from the ffmpeg/libav drama Jun 11 10:15:25 next question is of course, my EFI system has a graphical setup menu... can I get synergy working there too? ;) Jun 11 10:16:26 ah so nixos is basically linux+puppet Jun 11 10:17:03 veremit: "reproducible builds" taken really seriously Jun 11 10:17:04 I don't think I'd go back to a release-based OS tbh Jun 11 10:17:34 packages don't even *see* anything other than declared dependencies Jun 11 10:18:10 what about building custom versions of packages .. or do you have to use the packaged versions? Jun 11 10:18:39 afaik prebuilt packages for common configs are just a way to speed up things to avoid building everything from scratch Jun 11 10:18:50 guix+guile .. loooovely. No wonder I see the push for gettin guile-2 sorted ... Jun 11 10:19:05 yes, there is that .. guess its just what you're used to Jun 11 10:19:25 for embedded/arm stuff .. I'd never want to 'hand-craft' like I do my desktop systems Jun 11 10:19:30 as in, you can deviate from the prebuilt path but then you'll need to do some compiling Jun 11 10:20:05 huh I have it the exact opposite... I'd prefer a lot more hand-crafting for embedded Jun 11 10:20:06 -shrug- buildig package 11 of 15 (from source..) Jun 11 10:20:32 I'm more tolerant of unnecessary crap being enabled on my desktop than on embedded Jun 11 10:20:36 zmatt: its a fair point .. you do want a bit more control .. Jun 11 10:20:55 unfortunately there's no cross-Nix yet afaik Jun 11 10:21:04 >>> Emerging (12 of 15) media-video/ffmpeg-2.6.3::gentoo Jun 11 10:21:04 * ffmpeg-2.6.3.tar.bz2 SHA256 SHA512 WHIRLPOOL size ;-) ... Jun 11 10:21:40 time to grab some food I think . :D Jun 11 10:22:24 veremit: but I think gentoo does the same right? prebuilt package available to get started more quickly, but if you deviate from the defaults then yes you'll need to compile Jun 11 10:22:34 so that aspect seems same in Nix afaict Jun 11 10:23:10 no gentoo is all source-based. There are a few default options .. but you can specify down to a compile-flag based per-package Jun 11 10:23:59 * veremit frowns at all the --disable- ..... Jun 11 10:24:48 why would you cripple a lib like ffmpeg .. that seems quite strange .. *goes to look at the package file..* Jun 11 10:25:30 ok, I thought I seemed to remember gentoo had something like Nix' binary cache also Jun 11 10:25:42 "Conceptually, Nix builds packages from source, but can transparently use binaries from a binary cache if available." Jun 11 10:29:41 you can build binary packages .. but they're more useful to deploy to another system Jun 11 10:29:53 right .. similar principle Jun 11 10:30:21 if you hav multiple machines, then having a local binary cache also seems like a way to save a lot of time Jun 11 10:31:12 yes, provided the hardware is similar (core h/w that is) Jun 11 10:31:23 indeed Jun 11 10:32:22 I'm working with a small mix of amd/intel cpus .. the gcc optimisations for one arch don't always work so well on the other lol Jun 11 10:32:49 there's a few traps to fall into, but otherwise, yeah I don't mind it. Jun 11 10:33:56 Hello... Jun 11 10:34:54 Are there complications updating with apt-get in Machinekit (the newest version)? Jun 11 10:36:54 like? Jun 11 10:38:14 I am having issues like npm and gyp building. Jun 11 10:39:21 node-gyp rebuild fails on my system. Jun 11 10:42:43 pty.js@0.3.0 install script node-gyp rebuild is where it is failing. Jun 11 10:42:46 ok, afk for a bit Jun 11 10:42:52 Okay. Jun 11 10:48:21 npm 3.8.9 and node v6.2.0 keep trying to update. bb-node-red-installer and c9-core-installer keep trying to update with error codes, too. Jun 11 10:52:09 >>> Emerging (14 of 15) media-video/vlc-2.2.1-r1::gentoo Jun 11 10:52:10 Ah super... Jun 11 11:37:55 I will be back...like a swift or something. Later. Jun 11 11:59:51 <_av500_> tomba: hi :) Jun 11 12:12:57 (warm and moist)-- Jun 11 12:13:02 at least w.r.t. the weather Jun 11 12:13:25 (in tea I'd consider them desirable properties) Jun 11 12:21:05 O,o Jun 11 12:34:37 what? you prefer chewing on dry tea leaves? :P Jun 11 12:52:12 nuu .. or eating coffee beans/grounds/granules .. ick. Jun 11 15:35:41 hi i have problem with overlays from github bb-org overlays Jun 11 15:35:57 ic chan install them and also activate them Jun 11 15:36:16 then i see them mapped in the slot Jun 11 15:36:46 but i don't get data from the mapped i/O pins Jun 11 15:39:45 hi everybody i have some trouble with overlays Jun 11 15:40:02 i installed the overlays from beaglebone github Jun 11 15:40:15 dbauernf: wait around longer for a response Jun 11 15:41:04 ok Jun 11 15:41:26 you also might want to be more specific about what's not working Jun 11 15:42:08 i can map overerlay bone_eqep2b Jun 11 15:42:23 ihave instal this from official git rep Jun 11 15:42:49 but the driver do not get the io data Jun 11 15:43:16 if i use the PyBBIO-eqep2b overlay Jun 11 15:43:29 the driver works corrrect Jun 11 15:44:47 but in this overlay the index bit is not mapped Jun 11 15:45:00 thats wy i would need the other one Jun 11 16:49:01 submit a PR .. Jun 11 18:54:28 it's alive! Jun 11 19:09:19 haha, i think clock-frequency: <800> was an illegal value :-p Jun 11 19:11:49 jeeze i broke something Jun 12 00:55:59 are you trying to overclock your BBB? Jun 12 01:12:53 well it's a valid frequency for the cpu, unless you have the low-speed-binned am335x :P Jun 12 01:14:04 or forgot to crank up the cpu voltage, but uboot does that already by default, as should the kernel when using cpufreq Jun 12 02:12:57 what who? **** ENDING LOGGING AT Sun Jun 12 02:59:58 2016