**** BEGIN LOGGING AT Sat Jan 09 02:59:58 2016 Jan 09 06:41:02 Does anyone know how to install the kernel source and headers properly on the beaglebone black in order to build a kernel module? I installed linux-headers-$(uname -r ) (I'm using kernel 4.1.8-r18) and cloned the appropriate kernel source to my BBB, but when I try to build rt2800usb, I get the error: error: linux/delay.h: No such file or directory Jan 09 06:41:51 I know the file is in /usr/src/4.1.8 ..... /include/linux Jan 09 06:42:56 I thought installing the appropriate linux-headers package would put things in the right place for me to build a kernel module, but it seems not. I've also tried the bb-get-kernel-source.sh script, but it seems to be broken for newer kernels because of changes in RCN's repos Jan 09 14:23:53 if anyone present knows about writing drivers Jan 09 14:27:34 http://www.catb.org/esr/faqs/smart-questions.html Jan 09 14:54:17 hi, I want to install ubuntu on beaglebone black, I burned the image on the sdcard, then I put it in place in the board, hold down the boot button and conncet bbb via USB to my computer. I also am connected via minicom (ttyS0). the problem is that nothing i shown in minicom, and I do not understand what is going on. does anyone in here knows what is happening? Thanks Jan 09 15:12:58 torpico: do the LEDs light up at all? Jan 09 15:14:48 ls Jan 09 15:15:03 .oO(file not found) Jan 09 15:23:13 tbr: yes they light up Jan 09 15:23:44 and you don't see any output on the debug UART? at all? Jan 09 15:23:44 0 and 2 are lighting up now. Jan 09 15:25:18 hi guys I want to get ubuntu on my beaglebone black. I write image on micro sd and after plugging beaglebone black to power its start booting. Afterwards i had to type in login and password. Yeah i was logged in. The problem is i am not getting to desktop???? always coming ubuntu@arm: So which command i have to type in to start ubuntu? Pls help Jan 09 15:26:06 could you just help me with my issue Jan 09 15:26:07 ? Jan 09 15:26:27 which ubuntu image did you download? Jan 09 15:26:32 yes Jan 09 15:26:45 http://charette.no-ip.com:81/programming/2015-06-07_BeagleBoneBlack/ Jan 09 15:27:01 WTF Jan 09 15:27:07 why? Jan 09 15:27:15 that doesn't look like a trustworthy address Jan 09 15:27:50 yeah doesnt matter but do you know something about my problem Jan 09 15:28:17 your problem is probably because of downloading a random image that nobody in here knows anything about Jan 09 15:28:52 can you give me a webpage where i can download ubuntu from? Jan 09 15:29:56 Guest23648: hey buddy can you explain how you installed ubuntu on your bbbalck? Jan 09 15:30:50 yeah i just downloaded image and buned on my sdcard. Then insert sd card into bbb. While holding the boot button i plugged in 6V Jan 09 15:30:53 sry 5V Jan 09 15:30:55 http://elinux.org/BeagleBoardUbuntu Jan 09 15:31:05 that's a sensible page to follow Jan 09 15:31:10 it should explain everything Jan 09 15:31:17 Guest23648: There is no graphic. you want it have graphic? Jan 09 15:31:41 yeah Jan 09 15:31:49 no desktop Jan 09 15:32:10 Guest23648: till I know it should not have graphic. it is only terminal Jan 09 15:32:37 what of course there is a desktop or not? Jan 09 15:32:54 Guest23648: I did the same you did. How many minutes it took till the ubuntu flashed into your BBBlack? Jan 09 15:33:51 Guest23648: As I searched in interenet for struction, nowhere said it has graphics. only terminal Jan 09 15:34:32 Guest23648: this was my question I asked here. do you know anything to help me? :p Jan 09 15:34:34 I want to install ubuntu on beaglebone black, I burned the image on the sdcard, then I put it in place in the board, hold down the boot button and conncet bbb via USB to my computer. I also am connected via minicom (ttyS0). the problem is that nothing i shown in minicom, and I do not understand what is going on. does anyone in here knows what is happening? Jan 09 15:34:43 note, that those ubuntu images are all "console" images Jan 09 15:34:51 which means they come without a graphical UI Jan 09 15:35:05 if you want one you'll have to install and set it up yourself Jan 09 15:35:17 or you could just use the default debian image with full UI Jan 09 15:37:12 tbr: any solution for poor me? Jan 09 15:37:46 torpico: did you ever answer my question? Jan 09 15:37:49 > and you don't see any output on the debug UART? at all? Jan 09 15:39:52 hello Jan 09 15:40:02 tbr: what you asked from me? Jan 09 15:40:16 tbr: what is debug UART? Jan 09 15:40:24 i'm trying to get uboot to get a serial console working with beagle Jan 09 15:40:43 i can see the text, but not stop uboot from booting. hit any key does not seem to work. Jan 09 15:42:11 davis: make sure to turn off handshake Jan 09 15:42:42 torpico: you are claiming you're "I also am connected via minicom (ttyS0)" Jan 09 15:43:12 torpico: where does that cable go for ttyS0? Jan 09 15:43:37 hmm. i am in minicom. it says 8n1, that n is no parity i guess. let me find no flow control Jan 09 15:44:10 yeap that works many thanks Jan 09 15:44:14 np Jan 09 15:44:46 tbr: SORRY BUT i DO NOT GET WHAT YOU MEAN Jan 09 15:44:55 :-[ Jan 09 15:45:13 I know I'm fool ;p Jan 09 15:45:29 torpico: which cables are connected to the BBB? Jan 09 15:46:02 tbr: only the main USB cable which belongs to BBB Jan 09 15:47:27 torpico: ok, do you have a USB-to-serial cable with "FTDI" connector? Jan 09 15:52:25 tbr: no Jan 09 15:57:03 get one :) Jan 09 15:57:40 tbr: veremit there should be a way, how you guys do this? I mean install ubuntu in bbb Jan 09 15:58:24 download an image, boot. :) Jan 09 15:58:47 but you gotta trust the image you fetch, and md5sum it for integrity, use a good known uSD card .. etc etc. Jan 09 15:59:16 but we've all got our fingers burnt with a board that won't boot .. and serial debug is invaluable Jan 09 16:00:07 veremit: is minicom sufficient to help me see what is going on in my board? Jan 09 16:00:17 minicom is fine .. use it all the time Jan 09 16:00:33 just remember to set your serial port parameters right, and restart to make sure it reads them :D Jan 09 16:01:00 veremit: and how many minutes will it take to flash image to board + I set ttyS0 for minicom ? Jan 09 16:01:03 I have had some peculiarities .. but I think that's equally with my targets, not the software Jan 09 16:01:15 dmesg | grep tty Jan 09 16:01:15 [ 0.000000] console [tty0] enabled Jan 09 16:01:16 [ 0.644353] 00:05: ttyS0 at I/O 0x3f8 (irq = 4, base_baud = 115200) is a 16550A Jan 09 16:01:17 flashing usually takes a few minutes .. Jan 09 16:01:41 there's a led sequence to watch out for .. but it does keep changing Jan 09 16:02:18 ubuntu wil be a bit slow on the beagle .. but your choice :) Jan 09 16:03:24 I just updated mine to the lxqt jessie image :) Jan 09 16:03:40 although only just sorted a uHDMI cable, so never pwered it :( Jan 09 16:03:50 stupid connector Jan 09 16:04:04 mini would have been easier Jan 09 16:04:26 Ok, I've had a look at the mainline u-boot, and to me it appears there are some efforts towards a framebuffer for am335x, but that would only solve half the problem, since BBB also has a framer chip that needs to bi initialized. Jan 09 16:08:27 eliasbakken: making good progress there Jan 09 16:08:35 Is there an easy way to configure the TDA19988 pager using just some predetermined I2C sequence perhaps? Like for a fixed resolution etc.? Jan 09 16:09:25 veremit: I wish, lol, but splash screen in u-boot is not as trivial as I thought :) Jan 09 16:09:26 eliasbakken: I thought it mostly worked it out from video timings Jan 09 16:09:37 eliasbakken: pi has one :P~ Jan 09 16:09:57 I think I've seen it on another platform too Jan 09 16:10:10 Pi has a build in framer, no? Jan 09 16:10:27 its got build-in video yes Jan 09 16:11:31 Right, but BBB uses a framer chip, so that has to be configured as well in order to display a splash-screen during u-boot start-up. Jan 09 16:11:49 For an LCD cape, it would be easier. Jan 09 16:11:50 I don't think it needs a lot . if so. Jan 09 16:11:56 no .. it really isn't ;P Jan 09 16:12:11 usually you just ping it a video signal and sync .. it works the rest out. Jan 09 16:12:38 for anything you have to load a driver of some sort Jan 09 16:12:48 and the capes have additional overhead Jan 09 16:13:19 Ok, but how about USB power, does that come on from u-boot as of now? Jan 09 16:13:40 My display is powered from USB, so that needs to be handled. Jan 09 16:13:42 client power? Jan 09 16:13:52 ouch .. really? how does it supply sufficient?! Jan 09 16:13:55 Host power Jan 09 16:14:02 woops, yes host power Jan 09 16:14:03 It's a small screen. Jan 09 16:14:04 my bad Jan 09 16:14:38 that said .. hmm how did we do .. ah, 5v direct from supply .. we used a chalkboard 10" touch panel with a beagle once Jan 09 16:15:02 that was on hdmi to start with, until we fixed up a lvds serializer to the beagle rgb pins :) Jan 09 16:15:13 I'm using a Manga Screen: http://wiki.thing-printer.com/index.php?title=Manga_Screen Jan 09 16:17:03 wow that's a nice screen Jan 09 16:17:30 must be a phone screen :) Jan 09 16:18:10 though I've no idea how they pack HD into 5" .. anyway Jan 09 16:18:17 framer init .. Jan 09 16:18:29 zmatt you there? Jan 09 16:20:24 I have a BBW that I'm using to debug and test u-boot, so I can try to get it working with an LCD3 first, that tackle the framer later, if you think it's doable! Jan 09 16:21:14 Veremit: It's not HD, only 800x480, that is enough for 4.3" : 9 Jan 09 16:21:16 :) Jan 09 16:21:32 eliasbakken: yes I know :P was just musing Jan 09 16:21:47 eliasbakken: oh its all doable :) Jan 09 16:23:05 veremit: Very good! I'll set the rest of today and tomorrow to give it a shot! Jan 09 16:23:23 if I wanted to kill the serial console on usb and replace with a dhcp server would I start by killing the getty process or would it just restart? Jan 09 16:24:16 ie. i want to use the usb as a network device instead of a serial console Jan 09 16:24:41 eliasbakken: this may be worth a read .. http://minix-i2c.blogspot.co.uk/2013/07/edid-reading.html Jan 09 16:25:15 davis: the usb is a composite 'link' yu can have mass-storage/networking/serial all on the same cable. Jan 09 16:25:36 and it already is configured to dhcp your pc when yuo plug it in :) Jan 09 16:25:50 hmm. Jan 09 16:26:03 if your drivers aren't working .. thats an entirely different issue .. but I know that's how the beagle is configured for usb client :) Jan 09 16:26:10 i was just using it to connect via serial console Jan 09 16:26:30 you shuold have three "devices" .. network, serial and storage, normally. Jan 09 16:26:33 linux or windoze Jan 09 16:26:41 or mac* Jan 09 16:26:54 let me see if I can use my laptop as a usb networking client. to be honest, i want to reverse it. have the laptop be a dhcp server though. Jan 09 16:27:34 ip link set usb0 up (on my laptop failed) Jan 09 16:27:46 ah .. that means setting the usb host up differently :) Jan 09 16:28:02 although if you have the usb0 link .. shuldn't matter which end 'hosts' afaik .. Jan 09 16:28:06 which is odd, cause if I plug in my phone and enable usb tethering that command works Jan 09 16:28:29 davis: well that's a good sign ..but the phone is also dhcp server normally Jan 09 16:28:41 its a bit backwards that the 'client' end is acting as 'server' lol Jan 09 16:29:10 conceivably you might have to configure usb networkng differently to get it to work the other way Jan 09 16:29:17 yah, when my boss asked me to get this working I thought he meant usb tethering where my laptop was the client Jan 09 16:29:26 i did it with my phone rather quickly Jan 09 16:29:39 then he told me he wants to show to do it in reverse, Jan 09 16:29:45 yes .. the phone thing has been rigously developed and tested :) Jan 09 16:29:53 rigorous* Jan 09 16:30:15 but when I connect the beagle to my laptop, my laptop dmesg says usb serial support enabled. Jan 09 16:31:05 you probably need a slightly different driver Jan 09 16:31:24 I hate usb gadget for this .. too many combinations Jan 09 16:31:35 it shuold come up with a g_ ior better still, several Jan 09 16:31:49 eg. g_ether g_multi g_serial etc Jan 09 16:32:07 wonder if I can 'fly' mine quickly .. sec. Jan 09 16:32:29 hmm ls /sys/class/net on beagle lists usb0, but on my laptop its not there. Jan 09 16:33:27 * veremit vomits slightly pluggin bbb to usb .. Jan 09 16:34:14 module ftdi_sio is a serial usb driver on beagle, right? Jan 09 16:34:53 on a normal beagle .. yeah Jan 09 16:34:56 not on the BBB Jan 09 16:36:27 ok my BBB is showing up as...cdc_acm, usb-storage, rndis_host in dmesg Jan 09 16:37:06 hmm. i need to plug my phone back in and see what module gets loaded when i configure phone as usb tethering. Jan 09 16:37:07 and .. cdc_adm, cdc_ether, rndis_host rndis_wlan(?!) Jan 09 16:37:23 that's just rndis_host possibly wlan too Jan 09 16:37:35 or if its ipone its ipeth* Jan 09 16:39:44 hmm. if I plug in both the phone and beagle to the laptop, the ls /sys/class/net shows usb0. Jan 09 16:39:53 thats right .. I got /dev/ttyACM0 , /dev/sdh and enp0s18f2u1u3 ... NIIIICe lol Jan 09 16:40:08 yeah what OS you using? Jan 09 16:40:22 I don't think yur usb_gadget drivers are working Jan 09 16:40:29 lsmod on the laptop shows a cdc_ether device Jan 09 16:41:07 dmesg should tell you what that's linked to Jan 09 16:41:07 if I remove the phone, the cdc_ether is still there but the /sys/class/net device goes away Jan 09 16:41:27 rndis_host Jan 09 16:41:28 or you can ls /sys/class/net -l Jan 09 16:42:16 hmm Jan 09 16:42:21 strange Jan 09 16:42:23 i dont know what i am doing Jan 09 16:42:33 what beagle image are you using? Jan 09 16:42:43 its a debian image from last summer Jan 09 16:43:11 is there something on the beagle I need to do? Jan 09 16:43:42 the beagle does have a /sys/class/net entry Jan 09 16:43:45 hmm looks like RN pruned some stuff from elinux lol Jan 09 16:43:55 http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#2015-11-03 is what I've ben using Jan 09 16:44:06 and I've literally plugged it in, waited and done NOTHING Jan 09 16:44:24 it takes a while to boot and bring the usb up though Jan 09 16:45:37 * veremit wonders which image I booted lol Jan 09 16:46:29 this is odd Jan 09 16:46:45 ip addr on the beagle shows usb0 Jan 09 16:46:58 yeah you've got half a link missing ... Jan 09 16:47:06 but my laptop does not show usb0 unless i connect phone Jan 09 16:47:26 its not connecting Jan 09 16:47:45 when I remove the phone from laptop it says unregister rndis_host Jan 09 16:48:03 you need to read dmesg as you plug the beagle in (and wait, and wait) Jan 09 16:48:14 the phone will come up immediate ly because it has power Jan 09 16:48:30 unless you have your beagle on DC jack .. then you can unplug the usb at will Jan 09 16:48:43 otherwise you've probably broken your beagle :D Jan 09 16:48:55 i have my beagle on a dc power jack Jan 09 16:49:07 can I tail -f dmesg? Jan 09 16:49:20 um .. tail -f /var/log/messages I think .. or /var/log/dmesg Jan 09 16:49:27 can't remember which debian has .. Jan 09 16:49:50 neither Jan 09 16:49:55 it uses journalctl Jan 09 16:50:16 dmesg -w ? Jan 09 16:50:21 ugh systemd crap Jan 09 16:50:36 ok, dmesg -w works Jan 09 16:50:53 when I plug it in, it says new high-speed USB device number 19 using ehci-pci Jan 09 16:51:04 and nothing, but I will wait Jan 09 16:51:16 took a good few seconds on mne Jan 09 16:52:31 hmm on beagle (via ethernet ssh) I did a 'ip link set usb0 down' and then up. It still says down and the host dmesg is unchanged. Jan 09 16:52:57 however the usb0 on beagle has an ip address Jan 09 16:54:58 hmm. i wonder if I have the wrong cable. I have the ftdi/usb cable. Jan 09 16:55:13 i need to use the usb/micro cable, which I dont have. Jan 09 16:55:59 let me see if i can find one. i bet its using that port. the usb mini type connector Jan 09 16:56:07 here's my dmsg log .. http://pastebin.com/fuwVc7Ha Jan 09 16:56:17 yes, you need a USB-A to USB-miniB ... Jan 09 16:56:33 you don't get anything useful besides debug info on the serial header :p Jan 09 16:56:47 and likewise on the usb-a host port :d hehe Jan 09 16:56:53 lol, now it works Jan 09 16:57:03 thank god for the the cable bucket Jan 09 16:57:07 d'oh ! Jan 09 16:57:37 its been such a long time since i used this beagle, i forgot the usb mini port Jan 09 16:59:03 lok, nope its not up. dmesg showed a bunch of messages but they are for mass storage Jan 09 17:00:00 for some reason mine wasn't dh.... oh wait .. I need to dhcp from the client side haha Jan 09 17:00:12 the server was probablyl working :D Jan 09 17:00:38 the link was up .. just forgot I have to 'dhcpcd enp13049810924801928408124' to make it get the IP Jan 09 17:01:19 ip link set usb0 up on laptop fails, but still my host ls /sys/net/net is not showing usb0 Jan 09 17:01:35 yeah you're m issing something Jan 09 17:02:08 three devices are definitely configured for usb client Jan 09 17:03:17 i'm wondering Jan 09 17:08:49 lol, omg its weird Jan 09 17:08:52 now it works. Jan 09 17:09:10 eth1 is what its called for usb0 Jan 09 17:09:28 ie. when I plugin my phone, the usb network interface is usb0 Jan 09 17:09:46 when I plugin the beagle, the usb network interface is eth1 Jan 09 17:37:22 could be .. mine comes up at ethernet .. phone or beagle Jan 09 17:37:34 hi Jan 09 17:37:37 sorry i was away Jan 09 17:37:46 the interface naming is even more random than t was believed before Jan 09 17:37:49 the phone makes my host be usb0 Jan 09 17:37:57 the beagle makes my host be eth1 Jan 09 17:39:24 at least you know now Jan 09 17:39:34 the principles are the same whatever Jan 09 17:39:38 yes many thanks Jan 09 17:39:47 i appreciate your expertise and help Jan 09 17:40:26 i don't understand how this works, just that i it works Jan 09 17:42:21 its not that hard really Jan 09 17:43:30 https://www.kernel.org/doc/htmldocs/gadget/index.html has the background/technical Jan 09 17:43:48 ok thanks Jan 09 18:41:54 it's usb0 for me, at least once I disabled the annoying udev rule renaming it to something like en Jan 09 18:59:47 , Jan 09 19:15:43 so what's going on with beaglebone black? Jan 09 19:16:02 i bought mine when they first came out but it seems like production is stopping or something Jan 09 19:16:21 a lot of distributors have backorders but some don't, and the only recent posts about it are on this element14 website... Jan 09 19:17:00 zmatt: framer initialisation .. how much is there? can it be done via uboot? or does it need something else? Jan 09 19:17:08 ping eliasbakken: you there still? Jan 09 19:18:57 veremit: I've done it in a baremetal app, and there's also a u-boot driver I think. it just involves programming the framer (timings etc), pinmux, dma controller (mainly pointer to start/end of framebuffer), and flipping the enable-switch Jan 09 19:19:02 after that it'll run on its own Jan 09 19:19:23 zmatt: perfect, though it probably wasn't -that- hard Jan 09 19:19:28 eliasbakken: did you get that? Jan 09 19:19:35 if desired you can get an irq whenever dma completes a frame Jan 09 19:19:43 it isn't hard Jan 09 19:19:44 veremit: I'm here! I think I have the LCD controller initialized, Jan 09 19:20:37 eliasbakken: meet zmatt :) he's a beagle guru .. :p of sorts Jan 09 19:21:00 if you ensure the tilcdc driver isn't loaded and the memory used for the framebuffer isn't touched (I think you can file a reservation for it with linux) the splash screen should stay intact Jan 09 19:21:18 Nice, have you not given e advice before? Jan 09 19:21:49 Ah, OK that is good to know! I was wondering what would happen once the kernel started. Jan 09 19:23:02 if the tilcdc driver loads it'll probably take over in a rather hostile way, I don't think it's been designed to perform a clean transition Jan 09 19:24:04 That's actually OK, it will never reach that point with the current eMMC flasher script. Jan 09 19:25:16 a fancy option would be to load a PRU application to animate the framebuffer :) Jan 09 19:25:29 shh zmatt lol Jan 09 19:26:04 (a very retro way of doing animation is by using palette-mapping and changing the palette instead of or in addition to the framebuffer) Jan 09 19:26:42 that's quite cool .. very windows95 :D Jan 09 19:28:00 psh, amiga did it before it was cool ;) Jan 09 19:28:11 PRU can also be switched to run on the display PLL, allowing it to run synchronous to the display controller hence enabling it to count its progress and modify the framebuffer in a way that avoids tearing without requiring double-buffering Jan 09 19:28:20 you may want to locate someone from the demo-scene for that ;P Jan 09 19:28:27 Haha, yeah, it would be nice! But that would have to wait until the bash script takes over then, perhaps I can grab the framebuffer that gets initialixed in uboot? Jan 09 19:28:40 you can launch it from u-boot Jan 09 19:28:57 yeah I'd do it from uboot Jan 09 19:29:54 If I do manage to get that far and still have time by the end of the weekend, I'll give it a shot : ) Jan 09 19:29:59 you can alternatively map the framebuffer in linux (using uio or /dev/mem) and modify it to indicate progress Jan 09 19:30:12 though this will lack any protection against tearing Jan 09 19:30:49 (unless palette-mapping is used and the progress update can be done in a single write) Jan 09 19:30:52 here is how to do beaglebone black usb networking with archlinux. It shows beagle as dhcp client. out of box its dhcp server. Jan 09 19:30:56 here is how to do beaglebone black usb networking with archlinux. It shows beagle as dhcp client. out of box its dhcp server. Jan 09 19:31:00 https://youtu.be/oqecSDf_y1c Jan 09 19:31:25 But before that, I need to get the controller working properly. Still missing some timing info for LCD3 Jan 09 19:33:01 Pixeldiv and "pol", anyone know what values those are? Jan 09 19:33:37 pixeldiv is probably the divider applied to the display PLL output to make the pixel clock Jan 09 19:33:41 min value is 2, linux always uses 2 Jan 09 19:34:07 Ok, pol is clock polarisation? Jan 09 19:34:53 I guess, there are actually multiple bits related to that so I'm not sure what you may be referring to Jan 09 19:35:08 i.e. the clock edge used for data/oe can be set independent of the clock edge used for hsync/vsync Jan 09 19:35:13 beagle should always be dhcp server over usb, and your PC is client Jan 09 19:35:32 davis: what you want to achieve though is the reverse for your project Jan 09 19:35:34 crap I probably need to go shopping... Jan 09 19:35:42 zmatt: sux huh lol Jan 09 19:37:30 setting up LCD-Controller for 320x240x16 (hfp=59,hbp=22,hsw=48 / vfp=23,vbp=23,vsw=2 / clk-div=2) Jan 09 19:37:35 This looks about right. Jan 09 19:38:36 I've activated a test pattern, but nothing shows up... Jan 09 19:38:57 afk for a bit, bbl Jan 09 19:40:35 Hm... I guess I need to mux stuff right as well... Jan 09 19:44:59 ok, I'm declaring failure to exit the house for shopping... only 15 mins left till they close and I have no idea what to get so I'd probably just return with a bunch of random snacks, lol Jan 09 19:45:14 zmatt: been there, done that .. wearing the t-shirt :D Jan 09 19:45:16 guess I'll make a sandwich later, or maybe order a pizza Jan 09 19:45:51 eliasbakken: preferably after configuration (to avoid driving garbage onto pins) but before enabling Jan 09 19:47:22 also, if you're using an lcd cape, check whether there's some enable-signal that needs to be asserted (I think usually after enabling lcdc) Jan 09 19:48:00 Ok, yeah. Is there somewhere I can copy the pin mux from? Jan 09 19:48:25 my spreadsheet? https://goo.gl/Jkcg0w Jan 09 19:48:43 lcdc actually has its own tab Jan 09 19:50:00 Nice! Jan 09 19:50:06 for 16-bit color they're actually consecutive entries that need to be muxed to function 0 Jan 09 19:50:51 (index 40-59 of the pinconf register array) Jan 09 19:58:06 it still amazes me how nuts the signal use for color passive-matrix is Jan 09 20:00:06 Oh, haha! Jan 09 20:00:13 I got garbage on the screen! Jan 09 20:00:41 I think some artist would love that! Jan 09 20:00:59 I've seen memory dump passed off as art before... Jan 09 20:02:03 garbage is a start :D Jan 09 20:03:05 It's a great start! Jan 09 20:03:12 garbage is close to working in fact Jan 09 20:03:41 Hi, I'm having trouble finding a program that I wrote on the c9 ide, I can find it on the ide itself but I can't find where they are stored on the board Jan 09 20:03:56 Yes! Hm.. perhaps the test pattern is not working it's magic then... Jan 09 20:04:26 eliasbakken: either that or the start/end pointers are wrong Jan 09 20:04:50 WARNING: Code silently assumes 8bit/pixel Jan 09 20:05:28 LCD3 has 16 Jan 09 20:05:31 well that's fine, if you actually configure it for 8bit/pixel :P (but that also requires setting up a palette) Jan 09 20:06:05 Oh, can I do that? I'll try Jan 09 20:06:19 well keeping it 16-bit is probably easier Jan 09 20:09:58 although using a palette isn't hard per se, and saves memory (you can use 1, 2, 4, or 8 bits/pixel)... for a fixed palette it needs to be loaded in a separate phase, for variable palette with 8-bit color and your display size the framebuffer becomes struct { u16 palette[256]; u8 pixel[240][320]; } Jan 09 20:11:35 palette uses 12-bit color with the top 4 bits being zero except for the first entry (see TRM) and to avoid getting garbage colors some bit needs to be set in the raster controller config to expand 12-bit color to 16-bit color Jan 09 20:12:34 Hi, where do I find the programs I wrote on the c9 ide on beaglebone black itself? Jan 09 20:13:05 cleansnow: we saw your question the first time, the lack of response means that nobody who read it has any idea either Jan 09 20:13:50 (replying "I don't know" would scale rather poorly with the number of people who read the question) Jan 09 20:13:51 Alright, Sorry Jan 09 20:13:54 the ide runs in a browser window .. its not -on- the beagle afaik .. unless I'm mistaken Jan 09 20:14:04 veremit: it is on the beagle Jan 09 20:14:09 poop ok Jan 09 20:14:10 my understanding was that its stored on the beagle Jan 09 20:14:15 executed in nodejs Jan 09 20:14:35 but that's pretty much all I know about c9 Jan 09 20:14:51 you can do a "find / -name *filenamehere*" Jan 09 20:15:02 in the terminal supplied by c9 and it will show you where it saved them Jan 09 20:15:03 I never use it myself, and don't have it installed on any beagle Jan 09 20:15:17 i think it's in a folder called "workspace" on some directory Jan 09 20:16:24 let me know if that works for you cleansnow Jan 09 20:16:56 Ah, found a define for BPP32. Jan 09 20:17:28 zmatt: i'd like to have a fresh install myself because of all these configuration/permissions errors i get with the factor install of c9. did you uninstall c9 or did you flash a bare image? Jan 09 20:18:02 zmatt: So having the test pattern draw in 8 bit is not a problem? Jan 09 20:18:15 eliasbakken: well, it requires different (and more complicated) setup of lcdc Jan 09 20:18:34 more importantly, it results in a framebuffer that's half the required size Jan 09 20:18:42 if you don't reconfigure lcdc Jan 09 20:18:54 Hm.. OK. This is above my head right now. Jan 09 20:19:24 eliasbakken: but it's not like it's hard to generate a test pattern yourself Jan 09 20:19:37 Haha, yeah... Jan 09 20:20:06 i am the most ignored person in this channel Jan 09 20:20:14 ayjay: Yes thank you, I was able to find it Jan 09 20:20:24 ayjay: the channel lag is quite spectacular at times :p Jan 09 20:21:05 ayjay: I saw what you said, and filed it somewhere where it might get reproduced some day as "I think someone mentioned it being in some dir named 'workspace' somewhere" :P Jan 09 20:21:33 but not being a c9 user myself it didn't pay much attention Jan 09 20:23:28 haha okay, thanks guys Jan 09 20:27:52 eliasbakken: in case there's any doubt about the framebuffer format -> http://pastebin.com/E9HB8vMX Jan 09 20:29:50 Ok, yeah, I remember that from the hardware I think. Jan 09 20:32:27 So from reading the code, the test pattern should draw something along the lines of a 2*3 square pattern with colors. Jan 09 20:33:13 I'm now seeing something, similar to some squares, but not what I expect. Jan 09 20:33:37 Only black and white squares... Jan 09 20:36:12 squares is good .. lol Jan 09 20:39:46 Yeah.. But the color definitions for the BPP16 was missing. Only white and black was defined, so I need to google for a lookup table for red, green etc... Jan 09 20:41:31 d'oh Jan 09 20:43:05 Oh, at least now I got gray too! Jan 09 20:46:30 eliasbakken: look at my pastebin Jan 09 20:46:47 it gives the format of a pixel Jan 09 20:46:54 (in separate r/g/b components) Jan 09 20:47:32 Ok, yeah, let me try that. Jan 09 20:58:17 zmatt: did you use a blank OS image or did you uninstall c9? Jan 09 20:58:44 I always start with a console image Jan 09 21:00:10 could you point me to a link to download the image? i'd like to do the same thing Jan 09 21:00:11 then remove useless crap, upgrade to stretch, enable systemd-{networkd,resolved,timesyncd}, remove more now-unnecessary crap (like ifupdown, *dhcp*, *ntp*), and start installing things Jan 09 21:00:44 i'm assuming when you say console image you mean you don't start with the image that comes with c9 Jan 09 21:00:46 http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#Jessie_Snapshot_console Jan 09 21:00:52 oh cool Jan 09 21:00:53 zmatt: I used your ixel definition for the colors, but I think there must be something else wrong! Jan 09 21:01:26 thanks zmatt Jan 09 21:02:34 Ah, think I found the culprit. using uchar as a pixel pointer instead of ushort. Jan 09 21:03:27 Haha! Jan 09 21:03:35 you know you can just literally use the definition I gave (except for putting 'struct' in front of uses of Pixel16 if using C instead of C++) Jan 09 21:03:38 :P Jan 09 21:04:01 struct Pixel I mean Jan 09 21:04:15 Ah, OK, yeah that would be way more convenient : D Jan 09 21:04:42 (may want to rename it to Color16 to disambiguate in case you ever want to try palette-mapped formats) Jan 09 21:07:47 * GenTooMan peers at zmatts frame buffer format information. "Ugh" is his only response. "Well it's 16bit but C bit packed structures I've never had particularly efficient use of, then there is no guarantee of the ordering of the fields or their size except through compiler switches" Jan 09 21:08:57 GenTooMan: how many people do you know that configure the cortex-a8 on an am335x into big-endian mode? Jan 09 21:09:21 how much do you think I care about those people? :P Jan 09 21:09:32 using bitfields is perfectly safe and well-defined in practice Jan 09 21:09:39 zmatt almost none and it's probably not that dad burned important Jan 09 21:09:39 and a lot nicer than macros Jan 09 21:10:06 though I'd still wish I could just write u5 / u6 Jan 09 21:10:48 be glad it works... my policy Jan 09 21:11:27 it's sad that one needs to be glad when a C/C++ compiler behaves in a sane way nowadays Jan 09 21:12:09 This is what I have so far: http://ink361.com/app/users/ig-270828116/eliasbakken/photos/ig-1158918486570874052_270828116 Jan 09 21:13:08 you heard about MS compiler issues a few years back? They decided to write the device drivers because they didn't realize they had a compiler bug that had to do with the optimizer not respecting the volatile keyword. Jan 09 21:13:52 GenTooMan: fortunately the semantics of "volatile" are close to undefined anyway Jan 09 21:14:37 I've seen byte-packing issues differ bewteen linux and windows .. caused a beautiful bug with an ethernet firmware update util ........ Jan 09 21:14:58 rule of thumb .. NEVER assume ANYTHINg .. lol Jan 09 21:15:25 veremit and always test to see if it works before "commiting" mayhem. Jan 09 21:15:34 GenTooMan: definitely :) Jan 09 21:15:48 and I've been burned by volatile too, not to mention that in most cases it's overly aggressive in inhibiting the optimizer (while not necessarily imposing the constraint that's actually relevant) Jan 09 21:17:31 e.g. when reading an entire struct or array of volatile u32 the compiler *ought* to use ldr instructions instead of ldm Jan 09 21:17:34 gcc uses ldm Jan 09 21:17:50 -> bus error when accessing L3 interconnect registers Jan 09 21:18:15 (ldm is allowed to use larger than 32-bit accessses) Jan 09 21:18:45 that could be messy. Jan 09 21:21:42 at the same time I don't want the optimizer-inhibition of volatile... most of the registers are sufficiently memory-like to apply the usual optimizations. All I want is a constraint on the types of access the compiler is allowed to use Jan 09 21:22:15 an even worse example is the SRAM located in the ethernet switch subsystem that's used to store DMA descriptors Jan 09 21:23:03 it's a perfectly normal SRAM, but it has no byte-enables hence attempting to write a partial word will fill the remainder of the word with garbage Jan 09 21:23:26 the DMA descriptor however has several 16-bit fields Jan 09 21:26:22 (to make the situation even more bizarre, the registers in that subsystem are perfectly fine with bytewise access, only the SRAM isn't) Jan 09 21:39:11 zmatt I'm sure that is common to all TI created stuff. Sigh. Some of the MSP430 stuff does equally strange things. Where you are supposed to have 16bit acesss it's 32bit ... etc. Jan 09 21:40:23 well actually most TI peripherals are very memory-like, with the usual exception of "write 1 to clear" registers (but there's never any reason to perform a partial write to those anyway) Jan 09 21:40:40 I almost never use volatile, just an occasional compiler barrier Jan 09 21:41:10 which makes the behaviour of that SRAM particularly jarring Jan 09 21:42:18 most TI peripherals on SoCs I've worked with I mean Jan 09 21:42:45 generally it's third-party IP that misbehaves Jan 09 21:42:55 like the FlexNOC config registers Jan 09 21:50:00 I still wonder where Unicache/AMMU comes from, since its "style" and behaviour is unlike anything I've ever seen Jan 09 22:00:29 china? :D Jan 09 22:01:13 hehe Jan 09 22:02:13 it's apparently also heavily parameterized, but I know of only two configurations (one in Tesla on omap4/5 and one in Ducati/Benelli in omap4/5, dm81xx, vayu) Jan 09 22:03:40 leaving some features unused in both instances Jan 09 22:04:02 pizza arrived! Jan 09 22:04:36 \o/ Jan 09 22:04:45 damn you :p Jan 09 22:11:06 lol what :D Jan 09 22:15:16 OK, so here is an interesting thing: I set up the framebuffer, then I get the address of that buffer, and then I use ext4load mmc 0 0x8ffca000 /kamikaze.bmp to load a BMP image into it directly. Jan 09 22:15:45 I know it's probably not the correct way to do it, but there is an image appearing on the screen. Jan 09 22:16:24 presumably with garbage in front of it (the BMP header) Jan 09 22:16:35 Yeah... Jan 09 22:17:38 That is probably what is causing the image to not be displayed correctly... Jan 09 22:19:45 so just make a raw framebuffer file by stripping off the header? I don't know the BMP format, but if there's no trailing data you can try tail -c $(( 320 * 240 * 2 )) kamikaze.bmp >kamikaze.raw Jan 09 22:20:42 some image conversion utils can probably also directly produce a raw image Jan 09 22:21:40 that's beautifully dirty :) Jan 09 22:22:15 althuogh I think that's how you can load images to uboot framebufffer anyhow iirc Jan 09 22:25:59 I hope there was nothing located *after* that buffer though, given that you're overflowing it Jan 09 22:30:15 lolol Jan 09 22:30:29 zmatt: probably nothing important ... *BOOM!* lol Jan 09 22:31:02 if BMP does have trailing data, the proper way to parse it is of course to assume significant line-to-line correlation of pixel data and use that to locate the most probable offset of the image data Jan 09 22:31:03 doesn't the mmcload have a 'size' field .. Jan 09 22:31:25 ad bmp->raw is a trivial task for gimp :D Jan 09 22:37:39 Yes, that might work, but there is a bmp loading cmmand in u-boot. Jan 09 22:45:16 hmminteresting Jan 09 22:45:58 So what I'm seeing is what looks like the test pattern appearing in the top right quadrant of the screen. Jan 09 22:46:46 But I'm seeing the buffer being set aside is 152K, which should be 320x240x16 bit right? Jan 09 22:47:17 nope... lol Jan 09 22:47:43 ah hold on .. maybe Jan 09 22:48:04 yeah give or take Jan 09 22:50:54 Here is that picture again: http://ink361.com/app/users/ig-270828116/eliasbakken/photos/ig-1158918486570874052_270828116 Jan 09 22:50:55 150 KB Jan 09 22:51:06 Yeah. Jan 09 22:51:10 (precisely) Jan 09 22:52:03 153600 Jan 09 22:52:11 B Jan 09 23:39:13 Looks like the am335x-fb driver in u-boot does not respect the bpp, instead it sets it to BPP24 no matter what... Jan 09 23:40:33 ah .. that would be unhelpful Jan 09 23:40:48 and quite idiotic Jan 09 23:40:51 interesting Jan 09 23:40:54 also, easy to fix Jan 09 23:43:05 does it get the bpp from the EEPROM on the board with the display? Jan 09 23:43:43 I bet its botched lol Jan 09 23:43:50 note btw that if lcdc's expectation of frame buffer size mismatches the frame buffer given (which it will in case of bpp mismatch) it will set the "lost sync" irq bit, so it's worth verifying the irq status register for confirmation Jan 09 23:44:13 (e.g. once you think the problem is fixed) Jan 09 23:45:36 Haha! Jan 09 23:45:42 That helped! Jan 09 23:46:44 Still an offset in the test pattern, but at least filling the screen with the right colors. Jan 09 23:48:10 Ok, so the boot image is now displaying, although inverted, lol! Jan 09 23:50:14 Ok, it's 1 o'clock, time for some dinner. Looking forward to trying my hand on some framer initialization maņana! Jan 09 23:51:11 hey zmatt can i ssh over usb right away with that image you pointed me to? Jan 09 23:51:13 Thanks for your help so far guys! Jan 09 23:51:30 ssh yes, over usb no Jan 09 23:51:58 ah you must need to restart the sshd after the usb0 comes up :/ use the serial console Jan 09 23:52:09 on the usb :p Jan 09 23:52:13 the console image only brings up eth0 Jan 09 23:52:40 zmatt .. I'm pretty sure I've ssh'd over the usbnet befre .. but it probably doesn't bind to it immediately Jan 09 23:53:09 How much is the MSRP for the BBB rev c? Jan 09 23:53:13 ayjay .. minicom -D /dev/ttyACM0 -b 115200 -o Jan 09 23:53:17 does 72 USD sound right? Jan 09 23:57:31 Dumle29: I'm pretty sure it's less than that Jan 09 23:57:46 hmm Jan 09 23:57:49 that's what I found it for here Jan 09 23:57:54 Denmark that is Jan 09 23:58:21 I guess it may vary Jan 10 00:01:10 ah incl VAT it may be right Jan 10 00:01:43 wait usd or eur? Jan 10 00:02:34 usd Jan 10 00:02:45 converted from dkk Jan 10 00:04:08 then it looks like we paid more, or would have if we had to pay VAT :P Jan 10 00:04:17 heh, ok :P Jan 10 00:04:30 i guess I can't complain then. Just a bit out of my pricerange Jan 10 00:04:48 I'd really love to use one for my 3D printer, as my raspberry pi is struggling. Jan 10 00:05:16 I also thought it was cheaper Jan 10 00:05:16 well, it's struggleing when I want it to serve video. the fact that the USBs and the ethernet are on the same, single USB port is bleeh Jan 10 00:05:34 how does the beaglebone handle that? Jan 10 00:05:42 is the ethernet native? Jan 10 00:06:04 ethernet is native Jan 10 00:06:23 hmm, apparently the BBB launched at $45 Jan 10 00:06:30 I wonder what happened to its price then Jan 10 00:06:57 The rev C? Jan 10 00:07:47 It's 55$ at adafruit Jan 10 00:07:50 ah yes Jan 10 00:07:57 That's the US, do vendors list it with VAT there? Jan 10 00:08:14 probably yes Jan 10 00:08:18 dunno Jan 10 00:08:33 ahh. more flash on the rev c Jan 10 00:08:37 and comes with debian Jan 10 00:08:50 "comes with" is hardly relevant Jan 10 00:09:07 ah and it's excluding VAT... which makes sense since that'll vary Jan 10 00:09:49 but yeah I know it got more expensive over time due to flash and other reasons, but I thought it wasn't a big difference Jan 10 00:09:56 but if you add tax to that, it suddenly adds up Jan 10 00:10:54 note btw usb on the bbb isn't great, though I'm not sure if it's any worse than on the rpi Jan 10 00:11:58 but you do have native ethernet... sadly only 100 Mbps since they didn't put a gigabit PHY on it Jan 10 00:16:41 though ds2 apparently managed to hack one onto it :P Jan 10 00:17:27 ds2: maybe you should make a kit for it (60 cm of wire to delay the clock included) Jan 10 00:20:11 had some odd IRC issues :/ Jan 10 01:38:49 that's a damn big inductor... http://www.navy-radio.com/commsta/cutler/cutler-102.JPG Jan 10 02:17:19 I heard about inductors like that used for low frequency high power transmitters. Jan 10 02:24:26 yep, tuning coil Jan 10 02:25:00 So imagine adjusting that thing... Jan 10 02:25:49 "Generally these failed in the winter and were hell to change." Jan 10 02:27:44 The helix has several taps that can be changed manually to provide coarse changes to the antenna tuning circuit. It is usually only necessary to change taps when changing frequency. Jan 10 02:28:00 2 MW transmitter Jan 10 02:28:30 I guess you really don't want significant energy reflected back to the transmitter Jan 10 02:35:30 definitely not... that causes fires :p Jan 10 02:35:43 and worse Jan 10 02:35:52 rf burns worse Jan 10 02:36:45 strange desk... http://www.navy-radio.com/commsta/cutler/naa-qst-3.JPG Jan 10 02:37:33 I invested in a multi-band HF transceiver before christmas .. valve output stage :) 250W Jan 10 02:37:55 And a tuning unit :P for same reason. Jan 10 02:37:59 standing inside a coax... http://www.navy-radio.com/commsta/cutler/naa-qst-4.JPG Jan 10 02:39:23 musta been a lonely job .. lol Jan 10 02:42:39 the last project my dad worked on before he died was an auto-tuner, in cooperation with PA0LL who hand-makes tuners Jan 10 02:43:13 (which allowed the tuner to reside outside near the antenna and be controlled remotely) Jan 10 02:43:23 auto-tuners are pretty cool .. the guy I bought this rig off has one Jan 10 02:43:35 internal though Jan 10 02:44:48 veremit: link is a clip I found recently Jan 10 02:45:07 project never finished unfortunately Jan 10 02:45:34 damn **** ENDING LOGGING AT Sun Jan 10 02:59:59 2016