**** BEGIN LOGGING AT Thu Jul 23 02:59:59 2015 Jul 23 03:19:35 With the LCD Cape(s) can you set them to sleep/turn off after X time Jul 23 05:22:20 hi, thurgood Jul 23 05:25:21 heya Jul 23 06:34:30 hey all anyone up? have a question about an LCD cape Jul 23 06:35:52 then go forth and ask it Jul 23 06:40:05 I have the LCD7 Cape and noticed that the screen never sleeps or turns off, is there a place to set this in the beagleboard Debian relase? Jul 23 06:40:23 disable vampire mode? Jul 23 06:41:31 vampires sleep, they just do it during the day like bats, I thought Jul 23 06:41:47 lol... Jul 23 06:42:31 but they take off the cape when they sleep Jul 23 06:42:51 I dont want the whole board to go to sleep (running octoprint) just the LCD cape to turn off Jul 23 06:44:15 so, the LCD7 schematic, last page, shows where the backlight supply is derived from Jul 23 06:45:34 there's an inverter there driven by the signal calllled EHRPWM1A. Which is a signal on the expansion header. Jul 23 06:46:23 Ok Jul 23 06:46:26 Googling "ehrpwm1a lcd7" leads me to this page: https://groups.google.com/forum/#!topic/rowboat/IXDuQnMioOk Jul 23 06:48:14 so it looks like you can dim the backlight under software control, and dimming it to 0 should result in no pulses and thus the inverter is never enabled, no backlight, bam. Jul 23 06:49:55 ok and if I set that, then I would have to manually set it to something besides 0 to bring it back? Jul 23 06:49:59 this is all coming from someone who's never done it and is just practicing a little google-fu, by the way. It's entirely possible that I'm mostly wrong or leading you down a silly path and there might just be an echo something > /proc/sys/dev/what/ever and it'll just go. Jul 23 06:50:04 well, yeah Jul 23 06:50:55 ok I guess I should of rephrased my question a little bit more and asked if it was possible to AUTO dim and AUTO awake on mouse movement Jul 23 06:50:57 All I've shown is that the hardware exists and there's software that knows about it. Talking to that software is up to you now. :) Jul 23 06:51:09 gotcha ^5 Jul 23 06:51:24 heh. mouse movement is a bit above the solder-splashing where I play, good luck! Jul 23 07:42:42 hy there! can anybody tell me how to access the registers of the uart tty02 directly or any alternativ to tcdrain() ? Jul 23 08:05:21 MastaX: explain Jul 23 08:14:44 morning Jul 23 08:14:53 a bit confused about beagleboneblack ;x Jul 23 08:15:04 it includes 2 standard can interface pins or something? Jul 23 08:15:16 dont want to pay another 60 something for can cape Jul 23 08:15:50 check the TRM how many can interfaces the SoC has Jul 23 08:16:04 but: those are just UARTs, you still need some electrical interface Jul 23 08:16:12 (and that's what the cape does) Jul 23 08:17:55 oh Jul 23 08:18:03 it doenst have the CAN transceiver? Jul 23 08:18:14 only the mcu can controller build in Jul 23 08:19:07 no can transceivers on the BBB itself, as the SRM will easily tell you. Jul 23 08:19:19 it even has schematics, oh my! Jul 23 08:19:46 but it has 108 pages ;L Jul 23 08:20:09 * LetoThe2nd slaps superguest with th 5000+ pages omap trm Jul 23 08:20:16 ;p Jul 23 08:20:48 i still might buy it since i need some experience with embedded Jul 23 08:20:54 * KotH would have expected a sandworm eating superguest Jul 23 08:21:10 KotH: i've just had weisswürscht, don't want to ruin the taste. Jul 23 08:21:24 how do u compare bbb with rbp Jul 23 08:21:46 strcmp("bbb", "u") != 0. Jul 23 08:21:54 strcmp("bbb", "rbp") != 0. Jul 23 08:21:58 strcmp("u", "rbp") != 0. Jul 23 08:22:06 see, they're mutually exclusive. Jul 23 08:22:27 o.o Jul 23 08:23:42 wat is rbp? Jul 23 08:23:58 av500: mbp with a typo. Jul 23 08:24:24 av500: mdp with two typos Jul 23 08:24:35 wtf with 3 typos? Jul 23 08:24:48 av500: ++ Jul 23 08:26:27 raspberrypi Jul 23 08:27:26 cant pay 60 something for a serial connector and a can transiever chip which is the size of an smd resistor Jul 23 08:27:29 =/ Jul 23 08:27:44 build one yourself Jul 23 08:27:57 perhaps i will Jul 23 08:27:59 D: Jul 23 08:28:06 costs only about three bucks for parts and a box of band aids Jul 23 08:28:14 if they have the transiecer in dip Jul 23 08:28:21 i have a breadboard ;D Jul 23 08:28:50 dont think anyhhardware store has it lying around tho Jul 23 08:28:56 and they are so scarce also Jul 23 08:29:13 maybe if i am lucky i can get one today ;x Jul 23 08:32:15 size and price don't relate. have you ever bought an engagement ring? Jul 23 08:32:29 ;d Jul 23 08:32:39 no ;x Jul 23 08:32:43 never had a gf either ;'( Jul 23 08:33:08 probably because you were busy not reading a whopping 170 pages of SRM, right? Jul 23 08:33:22 i choose not to Jul 23 08:34:09 if i manage to graduate and get job then ill get a whife and engagement ring for her ;3 Jul 23 08:34:51 i have jumper wires and a breadbord Jul 23 08:35:18 no hardware store here in holland would have a dip can transiever lieing around tho ;x Jul 23 08:35:18 please note that i explicitly did not advocate any specific gender of either you or the person who gets the engagement ring. gf and wife were purely your interpretation. Jul 23 08:35:40 plus, the weisswürscht taste is now gone. Jul 23 08:35:46 i m so sorry ;x Jul 23 08:36:25 LetoThe2nd: do you want some brezel? i just bought some from the lokal brezelbäcker Jul 23 08:36:43 KotH: yes please. Jul 23 08:37:10 * KotH goes downstairs to switch on the thumber Jul 23 08:38:13 * LetoThe2nd munches on KotH+thumper+brezels Jul 23 08:49:04 if you find someone like you, you can get away with a ring made of R and C - tell that person it resonates with your heart :D Jul 23 09:07:16 how to i install yaml packages on angstrom in beagle bone black Jul 23 09:08:03 how to i install yaml packages on angstrom in beagle bone black Jul 23 09:09:51 how to i install yaml packages on angstrom in beagle bone black Jul 23 10:33:19 hari__: http://www.elinux.org/Installing_A_Software_Package_In_Angstrom Jul 23 10:34:22 (yes, that mentions a minnowboard, but 'angstrom' is whats relevant here...) Jul 23 11:27:15 I just got a BBB from Adafruit with Debian preinstalled. Any idea where I can get the linux-headers package? Jul 23 11:30:07 3.8.13-bone47 . Hmm... it seems the kernel is not installed from a package Jul 23 11:31:04 No "build" link Jul 23 11:32:27 tzafrir: apt-get install linux-headers Jul 23 11:32:36 tzafrir: as with any debian Jul 23 11:33:33 xkonni, in Debian you would try installinc linux-headers-`uname -r` Jul 23 11:34:21 There are various packages providing linux-headers, all of them for kernel 3.2 (this is a wheezy system) Jul 23 11:34:22 thats a specific one, yes... and there's also just linux-headers pointing to the latest one iirc Jul 23 11:35:03 That would be linux-headers-$arch or something similar Jul 23 11:36:01 The funny thing is that they have bothered putting some minimal files under /usr/share/doc/linux-image-3.8.13-bone47 Jul 23 11:37:01 A dummy debian changelog file and a copyright file that claims that the kernel is owned by Linus :-) Jul 23 11:37:30 But no information on where to get their sources Jul 23 11:37:44 hehe nice... and apt-cache does not show much when searchign for linux-headers? Jul 23 11:37:52 Nope. Jul 23 11:38:17 (as its a fresh install, you did apt-get update, right?) Jul 23 11:38:26 And the server they use for their repository doesn't seem to be a standard file server, so I can't poke around for extra sources Jul 23 11:39:00 https://github.com/beagleboard/linux/tree/3.8.13-bone72 a bit newer, but close Jul 23 11:39:03 yes, I have updated. Jul 23 11:39:05 71 is also available Jul 23 11:39:09 ok... mhhm Jul 23 12:30:19 hello Jul 23 12:31:45 I have built sato image for beaglebone using yocto. I'm trying to find the image to load on the sdcard for flashing a new image. Do I use the poky/build/tmp/deply/images/core-image-sato-beaglebone.tar.bz2? This seems like just a rootfs and not everything I need. Jul 23 12:51:49 Hi everyone, I tried to copy my BeagleBone according to these steps: http://elinux.org/BeagleBone_Black_Extracting_eMMC_contents Jul 23 12:52:32 Everything works the same on the copies it seems, only no driver is loaded for it in linux/windows Jul 23 12:52:53 So I can't communicate it via USB; I have no FTDI with me; does someone know a fix? Jul 23 13:29:07 does someone know how to modify the universal capes? Jul 23 13:33:55 as in? Jul 23 13:34:32 https://github.com/cdsteinkuehler/beaglebone-universal-io ? Jul 23 13:34:41 yes Jul 23 13:35:37 I took 3 of these and made one of it - it works Jul 23 13:36:00 but eQEP1 is missing at the pins Jul 23 13:36:56 when I put it in myself, the pinmux helper doesen't work for that property Jul 23 13:38:52 I already edited the config-pin Jul 23 13:55:40 Hi does anyone know why the drivers for the BeagleBone Black made by element14 aren't loaded? Only the element14 version of the BeagleBone doesn't work for me Jul 23 13:57:42 I read that as, drivers (on a PC or Mac to make it show up as a network interface) aren't loaded (when plugging the BBB into USB on the PC)? Jul 23 13:58:14 I don't have an answer, just making sure we're clear on the question! Jul 23 14:00:37 sorry, English isn't my first language. That is correct, I am using windows 8.1. With my older boards I can just plugin and use putty to connect to it. When I plug a new board made by element14 it doesn't show up and in device manager it says that the driver has not been installed Jul 23 14:03:32 Auke, have you tried with a microSD card Jul 23 14:04:11 Well I just flashed it again with a debian version that I use on the other boards Jul 23 14:04:24 Do you mean running it from a microSD card instead of eMMC? Jul 23 14:09:51 stt_michael managed to fix it by extracting the drivers and pointing to one manually Jul 23 14:11:16 Auke, yes, run from a known image .. but sounds like you cracked it :D Jul 23 14:29:28 Should I see something in dmesg on my pc when I plug in my beaglebone black via usb? Jul 23 14:29:52 Somehow the usb interface does not come up Jul 23 14:31:15 something, yes Jul 23 14:37:00 hopefully lol Jul 23 14:37:24 depends on too many things ..do you have usb gadget support in your kernel ? Jul 23 14:37:58 why nobody didn't stick one of the cheap ft230x chips on .. no NOT the expensive ft232 .. Jul 23 14:42:34 stt_michael: because that entails yet another usb socket, signal routing, etc. and raises the BoM cost. Jul 23 14:42:50 hrm I guess .. but bumping the emmc justifies it .. Jul 23 14:43:25 also usb2serial adaptors are super cheap, especially if you don't buy ftdi, but pl2303 or cpxx Jul 23 14:43:54 IIRC the eMMC was discontinued Jul 23 14:44:03 the 2 GByte one Jul 23 14:44:43 the cypress chips are the most gorgeous thing, generic USB CDC drivers, FT232-compatible footprint Jul 23 14:45:27 I hate highlights... Jul 23 14:45:57 says the person whose nick isn't a common pronoun Jul 23 14:46:34 bur neri is in generic for example Jul 23 14:47:00 Why have a highlight for a partial part of your nick? Jul 23 14:47:02 it highlights on partials?! Jul 23 14:47:19 that's got to be a setting Jul 23 14:47:20 Tab-complete works on all clients I have used; let the plebes use it, or fail to get your attention. Jul 23 14:48:14 because of some people in one channel who like to change nicks od put something at the beginning Jul 23 14:48:27 or just shorten it Jul 23 14:48:38 Still sounds like others' problem, not yours. Jul 23 14:49:02 If I plug my BBone without SD card in my computer I see it show up on lsusb, but if I plug it with the SD card it does not show up?! Jul 23 14:49:04 Why make more annoyance for yourself, for others' inability to use their clients correctly? Jul 23 14:51:16 not that big problem, but right now I am hoping that someone could tel me what I am doing wrong with the overlay... Jul 23 14:51:57 nerienna: Not giving us enough detail on what the issue is, for use to solve it. Jul 23 14:52:43 I took 3 device tree overlays from https://github.com/cdsteinkuehler/beaglebone-universal-io Jul 23 14:52:56 made one with all the pins I need Jul 23 14:53:10 still, there was no eqep1 at the pins Jul 23 14:53:42 so I wrote this property myself Jul 23 14:54:32 edited the config-pin to get in my cape Jul 23 14:54:42 jsabeaudry: sounds like a really old u-boot Jul 23 14:54:54 jsabeaudry: what's on the eMMC? Jul 23 14:56:29 but when I do config-pin -l P8_31 for example, qep dosen't appear Jul 23 14:57:30 I'm just a beginner, didn't compile a kernel till today, but everything else is working Jul 23 14:58:27 tbr, I never touched the eMMC and the SD card contains the lastest debian image. Jul 23 15:00:22 jsabeaudry: I suspect it would boot fine if you hold down the S2 button before powering up the board with the SD inserted Jul 23 15:01:09 tbr, people are informing me here that this is actually a beaglebone blue, does that change anything? Jul 23 15:01:25 blue? Jul 23 15:02:03 * tbr is not aware of such a board Jul 23 15:02:31 did you mean bluesteel, which is NOT an official beaglebone, but very very similar Jul 23 15:02:44 tbr, yes bluesteel Jul 23 15:03:32 bluesteel doesn't have an eMMC Jul 23 15:04:15 My modifies beaglebone black has no hdmi Jul 23 15:04:55 tbr, does that mean that the u-boot of the lastest image somehow turns off the usb at boot? Jul 23 15:05:58 tbr, huh, looks like booting with s2 worked, now I'm confused Jul 23 15:06:07 jsabeaudry: it means that if it boots something without SD-card inserted, that either you have a special unicorn or it's not a bluesteel Jul 23 15:06:20 is the PCB black or blue? Jul 23 15:06:26 blue Jul 23 15:07:20 so it looks like this but with the bottom right footprint populated? http://files.linuxgizmos.com/circuitco_bluesteelbasic.jpg Jul 23 15:08:50 tbr, looks exactly like the picture, the bottom right footprint is not populated Jul 23 15:08:51 or is the board plugged into some sort of cape that has an eMMC? Jul 23 15:09:02 no cape Jul 23 15:09:21 * tbr is confused Jul 23 15:09:57 lsusb shows Linux Gadget now Jul 23 15:10:04 S2 forces boot from SD-card Jul 23 15:10:27 Let me try to repower it without touching anything Jul 23 15:10:33 I'm not sure what the board is configured for that prevents it from booting without pushing S2 Jul 23 15:11:11 the SRM for bluesteel should answer the configuration of the SYSBOOT pins though. If there is a SRM Jul 23 15:11:19 perhaps I was not patient enough Jul 23 15:11:47 where to buy a bluesteel? can't seem to find a distributor in europe Jul 23 15:11:53 If there is no data on the emmc, it boots from SD-card - I had to change several EMMCs Jul 23 15:12:03 note that issuing a reboot will work fine as the SoC remembers to boot from SD-card. it needs to be completely without power to revert to original SYSBOOT Jul 23 15:12:05 temperature requirements Jul 23 15:12:20 samael: probably only sold by circuitco directly Jul 23 15:13:00 nerienna: yes that's what a beaglebone black is configured for. No idea what the bluesteel is configured for in terms of boot order Jul 23 15:17:24 hello Jul 23 15:18:41 trying to remember if the AM335x has some internal memory - perhaps they used something like that Jul 23 15:19:04 it has been a month or two since I read the datasheet Jul 23 15:21:03 nerienna: it doesn't. there's only ROMBL Jul 23 15:22:44 ROM-BootLoader, some very minimal thing that bootstraps the soc enough on some 32kb internal ram to accept the intermediate loader from first mmc bus, second mmc bus, serial, usb and I thinks some funky ethernet thing Jul 23 15:26:40 hello Jul 23 15:26:51 i've tried to boot a new yocto build Jul 23 15:27:00 i have a serial port connected Jul 23 15:27:26 and when I boot via power I see the uboot prompt Jul 23 15:28:18 when I stop the boot i see it printed a message saying spl_load_image_fat_os: error reading image. does that mean my sdcard has a bad image or is that a result of stopping the boot Jul 23 15:29:01 what happens when you dont stop the boot? Jul 23 15:29:11 pastebin.com the whole log Jul 23 15:29:24 if I dont stop the boot, it seems to boot from the emc Jul 23 15:29:46 and do you see the same line? Jul 23 15:30:03 im looking one sec. it keeps scrolling Jul 23 15:30:50 davis, U-Boot SPL 2014.04 ? Jul 23 15:31:07 one sec. i'm trying to save a log with this terminal program Jul 23 15:32:03 the uboot prompt is U-Boot SPL 2014.04-00014-g47880f5 (Apr 22 2014 - 13:23:54) Jul 23 15:32:13 fixed in mainline... Jul 23 15:32:18 and the error is there: spl_load_image_fat_os: error reading image args, err - -1 Jul 23 15:32:19 about a year ago.. Jul 23 15:33:04 so I need to fix the boot loader prior to booting from sdcard? Jul 23 15:33:53 yes/no... you just need a uEnv.txt file with the variable "uenvcmd" telling u-boot what you want to do.. Jul 23 15:34:11 otherwise it's just assumes microSD is external storage and boots eMMC.. Jul 23 15:34:38 hmm. uEnv.txt file? Jul 23 15:34:56 do I use that to setenv via the prompt? Jul 23 15:35:28 no just mount the drive and "echo "uenvcmd=boot" > ./uEnv.txt" Jul 23 15:36:42 hmm. so let the beagle boot, use usb to connect to my host as a disk and then rewrite the uEnv.txt file? Jul 23 15:36:42 or just clear out the eMMC... not sure what' yocto expects for u-boot variables.. Jul 23 15:37:14 or just stick the microSD card you placed yocto on in your pc... Jul 23 15:38:54 davis, here is a generic "uEnv.txt" that will work with every generation of OS ever installed on the eMMC: https://eewiki.net/display/linuxonarm/BeagleBone+Black#BeagleBoneBlack-DealingwitholdBootloaderineMMC Jul 23 15:39:13 just adjust the variables for installed location on your yocto setup.. Jul 23 15:39:25 so i just need to put that on the sdcard which I am trying to boot? Jul 23 15:39:58 Yes... if you look at your terminal log, do you see it looking for "/uEnv.txt" ???? Jul 23 15:41:03 yes, after the fail the fail message, it says "reading uEnv.txt unable to read uEnv.txt" Jul 23 15:42:52 at the bottom of that webpage it says to copy it to /media/rootfs. Really that should be /media/boot where that is partition1 of the sdcard, right? Jul 23 15:43:58 davis, yocto is still in the stone-age... no reason for the "fat" boot partition... That wiki shows the "latest" setup... so in your case stick uEnv.txt in the fat/boot partition.. Jul 23 15:44:12 gotcha Jul 23 15:44:22 since you have the old setup, make sure you change 0:1 -> 0:2. ;) Jul 23 15:44:44 yeah to be honest, I'm doing this for work. it seems that we should be using buildroot or android. not sure why we picked yocto. Jul 23 15:45:59 ahh, i see the 0:1 in the load image lines. that makes sense. thanks Jul 23 15:50:00 hmm. well that did not work. perhaps my image is bad. gawd Jul 23 15:50:26 davis, pastebin your boot log.. Jul 23 15:54:59 Which driver is required for the ethernet-usb connection with a beaglebone? Jul 23 15:55:13 I mean which kernel config Jul 23 15:56:09 jsabeaudry, g_ether Jul 23 15:57:19 rcn-ee, CONFIG_G_ETHER? Jul 23 15:57:40 correct, that's what is used on the beaglebone's... Jul 23 16:00:37 rcn-ee, Can't seem to find it, could if possibly be CONFIG_USB_U_ETHER ? Jul 23 16:01:33 Here is what contains "ETHER": CONFIG_ETHERNET=y CONFIG_USB_NET_CDCETHER=y CONFIG_USB_U_ETHER=y Jul 23 16:02:45 jsabeaudry, it's under usb-gadget config, but are you trying to enable your config for host? Jul 23 16:03:20 jsabeaudry: that's for the beaglebone side, for the PC side it's CDC_Ether IIRC Jul 23 16:03:30 rcn-ee, I'm trying to get my computer to have usb0 interface Jul 23 16:04:10 while running the default bbone image Jul 23 16:04:37 Right now I'm just seeing the files and not the usb ethernet Jul 23 16:08:57 I added CDC EEM, I'll give it a spin Jul 23 16:37:50 hi, i have a beaglebone black with arch linux arm installed but when i run an egl application it fails with a mesa error message: "libEGL warning: DRI2: unable to authenticate" so i was wondering, does bbb+egl work on debian? Jul 23 16:45:45 depends on which kernel you use... Jul 23 16:45:45 krnlyng, we don't have an xorg driver.. in "theory" using wayland with the sgx drivers would "work".. Jul 23 16:47:22 using it directly on fbdev definitely works Jul 23 16:47:55 I got the MagicLantern demo running Jul 23 16:49:09 thurgood: what kernel (or debian image) do you suggest for my purpose? Jul 23 16:49:17 if you don't need windowing but just single-window-fullscreen then that suffices Jul 23 16:49:30 i see that would be fine too Jul 23 16:49:48 krnlyng, any using 3.14/4.1.x as the sgx modules are installable thru apt.. Jul 23 16:50:35 i see, thanks i will try that Jul 23 16:51:13 the userspace blobs are the only thing can't install thru apt, and there's a script to just setup *.tar with ti's blob's.. Jul 23 16:51:18 http://beagleboard.org/latest-images does this have what i need? Jul 23 16:51:25 correct Jul 23 16:51:30 rcn-ee: can you link the script? Jul 23 16:51:51 krnlyng, actually use this one: http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#Debian_Image_Testing_Snapshots Jul 23 16:51:52 or the blobs? Jul 23 16:52:25 the blob extractor is here: https://github.com/RobertCNelson/ti-linux-kernel-dev/blob/ti-linux-4.1.y/sgx_create_package.sh Jul 23 16:52:39 and make sure /etc/powervr.ini (installed as part of this) selects WindowSystem=libpvrPVR2D_FLIPWSEGL.so rather than WindowSystem=libpvrPVR2D_FRONTWSEGL.so Jul 23 16:52:45 unless you like tearing Jul 23 16:53:00 thanks! you helped me a lot Jul 23 16:57:07 (the FLIP one uses vsync-triggered double buffering) Jul 23 16:59:15 23~ Jul 23 17:49:05 Im trying to set up a dts for some GPIO's with power on defaults. I'm using debian jessie with a 3.14 kernel, an rcn build from 3/1/15. Jul 23 17:49:14 I've got the dts compiling and loading, here is the GPIO fragment, http://pastebin.com/XfPfuZ7L. Jul 23 17:49:24 The defaults don't take, and control-pin, while installed on the system, doesn't seem to work for the 3.14 kernel. Jul 23 17:55:58 I should probably have said that this is for a BBB revC Jul 23 18:16:21 gwilson: paste removed. tried running with sudo? Jul 23 18:17:10 gwilson: or just basic gpio stuff. echo ID > /sys/class/gpio/export (iirc, bbb is at work) Jul 23 18:17:42 gwilson: again, this needs to be done as root Jul 23 18:18:20 (or with a udev-rule/etc for user-access) Jul 23 18:49:05 http://krtkl.com/ thoughts? Jul 23 18:52:52 no pricing and almost no info... the image is an obvious 3D render which means no board yet either Jul 23 18:55:54 not even the fpga vendor Jul 23 18:56:52 nor the processor (if separate rather integrated) Jul 23 18:57:15 if integrated, then the combination of dual-core A9 + specified number of gates and GMACs might identify it Jul 23 18:57:56 but iirc the FPGA SoCs I've seen were *very* costly Jul 23 18:58:25 indeed, the ones with a CPU even more Jul 23 18:59:23 that's what I meant with an FPGA SoC Jul 23 19:12:29 If I plug in my ethernet usb adapter, it works. However, when I plug in my bbone, I get the filesystem mount but no sign of the ethernet-over-usb. How can I troubleshoot this past lsusb that shows the linux gadget correctly Jul 23 19:16:03 The ACM works Jul 23 19:17:34 I read the silicon errata on the AM335x yesterday Jul 23 19:17:42 there is one on usb Jul 23 19:18:56 http://www.ti.com/lit/er/sprz360f/sprz360f.pdf Jul 23 19:19:35 the usb peripheral on the am335x is ehm... well... Jul 23 19:19:37 jsabeaudry, https://gist.github.com/RobertCNelson/b2fc4c1db49b6a8a304b Jul 23 19:19:42 it has a lot of personality Jul 23 19:20:35 jsabeaudry: but I'm not quite following what you're trying to do... why ethernet-over-usb ? Jul 23 19:21:19 zmatt, i'm pretty sure he's running his own minimal kernel config on his host, so nothing works out of the box. ;) Jul 23 19:21:42 rcn-ee: a filesystem mount showing up is odd too Jul 23 19:22:02 I know that at several PCs at work ethernet over USB doesn't work on each one Jul 23 19:22:12 rcn-ee, ah thanks! I'm missing huawei and phonet Jul 23 19:22:28 especially the Suse-one that I have doesn't work Jul 23 19:22:31 jsabeaudry, i doubt it's those.. Jul 23 19:24:59 jsabeaudry, here's my whole config: https://gist.github.com/RobertCNelson/b2fc4c1db49b6a8a304b#file-full-config (it's just debian.) Jul 23 19:29:06 rcn-ee: btw, ever tried the PowerVR stack without fbdev support? in theory it should be less overhead, and possibly more stable given there'd be less code attempting to glue the PowerVR crap onto existing linux code... Jul 23 19:29:53 but I'm wondering if I'm going to go down a road of despair if I'm gonna try it Jul 23 19:30:51 zmatt, i've never tried that... i've got mostly a hate relationship with pvr, so i usually don't touch it unless some emails me. ;) Jul 23 19:31:29 In dmesg I'm getting usb 1-5: bad CDC descriptors Jul 23 19:31:55 swap cables/hub... Jul 23 19:31:59 yeah it's a wonderful piece of software... compiling the included demos was also not exactly trivial Jul 23 19:32:20 jsabeaudry, the musb ip is very picky on the usb signals... Jul 23 19:32:29 still you might want to include those in your script, MagicLantern is pretty to stare at Jul 23 19:33:20 last i looked, they dropped those in 02, i didn't want to download both 01/02 just to get them.. Jul 23 19:33:39 I only have 02 Jul 23 19:33:48 there are definitely few demos Jul 23 19:33:56 only four Jul 23 19:34:00 only one worthwhile Jul 23 19:34:32 i use to always test them.. just to make sure the driver worked wonder if they just moved them around i never looked.... Jul 23 19:35:30 I had to do Jul 23 19:35:35 cd ignore/SDK_BIN/Graphics_SDK_setuplinux_5_01_01_02/GFX_Linux_SDK/OGLES2/SDKPackage/Demos/MagicLantern/OGLES2/Build/LinuxGeneric/ Jul 23 19:35:57 oh we have to build them.. that's why... Jul 23 19:36:04 rcn-ee, Do you see anything else that I might be missing in there: http://pastebin.com/MZ2juqN8 Jul 23 19:36:28 make CROSS_COMPILE=arm-linux-gnueabihf- PLATFORM=LinuxARMV7 LIBDIR="some dir containing the libs" Jul 23 19:37:05 executable appears in ../LinuxARMV7 Jul 23 19:37:08 yeah you have to build them Jul 23 19:37:10 jsabeaudry, looks fine, g_multi loaded usb-flash and usb-serial... just no networking. .;) Jul 23 19:38:01 the executable didn't work due to wrong libpath though, but relinking it on the target fixed that Jul 23 19:38:39 rcn-ee, Thanks for your help, I'll bail for now. Jul 23 19:38:48 sorry I meant executable appears in ../LinuxARMV7/ReleaseRaw/ Jul 23 19:39:09 since the weren't enough directory components yet obviously Jul 23 19:39:19 *there Jul 23 19:39:28 yeah, i'll push it to builder, then we can just have it in the repo.. Jul 23 19:39:51 except it needs the pvr libs... Jul 23 19:40:46 what are the conditions on the pvr libs? I mean, to be of any use they have to redistributable by customers Jul 23 19:41:34 I never thought about it much yet, but I just realized the fact that apparently you can't package them is a pretty bad sign Jul 23 19:43:08 wb Jul 23 19:43:35 did you still catch my question? Jul 23 19:45:59 ... at first sight using gles for a QT full-screen app seemed like a good idea, but I'm starting to get doubts Jul 23 19:46:39 zmatt, yeah due to my employment, i can't really do something ti says not to do.. (shucks! ;) ) as we are a partner... so i keep bugging them to release it for me. ;) Jul 23 19:47:05 huh, ti explicitly wants the SDK not-packaged? Jul 23 19:47:38 that's where the omap5/x15 looks so nice, it's all out: http://git.ti.com/gitweb/?p=graphics/omap5-sgx-ddk-um-linux.git;a=summary Jul 23 19:47:51 yeah some of the files are non-distrubutialbe.. sgx ligs.. Jul 23 19:48:09 the demos had a license, but now that they are source.... Jul 23 19:48:16 if they are non-distributable... what use are they? Jul 23 19:48:39 I mean, can be ship products with them? Jul 23 19:48:39 customers can 'install' them in their 'system'... the rest of us = screwed.. Jul 23 19:48:46 *we Jul 23 19:49:51 I guess I'll need to poke someone in the office to examine the relevant legalese... I'm fairly certain noone there is going to enjoy getting that request from me Jul 23 19:52:46 here's the line: "Object Code License. For the Licensed Materials provided in object code format, TI hereby grants to you a limited, non-exclusive license to reproduce and use the Licensed Materials provided to you in object code format and to distribute an unlimited number of object or executable copies of such object code Licensed Materials." Jul 23 19:54:12 I see "grant", "reproduce", "distribute" Jul 23 19:54:34 the line seems pretty permissive to me Jul 23 19:54:39 yeah, looks like it might have changed... Jul 23 19:55:19 why hasn't anyone put up a repo with these bin's. ;) Jul 23 19:57:43 btw, since you're a bit closer to TI than I am... do you have any idea why TI has treated the 81xx series so badly? I mean, tony has recently added some mainline support for it, no help from TI afaik... official TI kernel is still from 2011. And now with the graphics SDK I've noticed ES6.x (for the 81xx series) also being omitted in some places even if they continue to list 5.x Jul 23 19:58:58 it never got a beagleboard varient.. i'm pretty sure that also ended up in the "omap" group, and they got cut after the omap4.. Jul 23 19:59:25 it's also Sitara (am38xx) and automotive (dra6xx) Jul 23 20:00:12 it's such a pity Jul 23 20:00:27 it just had to be the beagle affect, if it doesn't get a low cost board into 1000's... it just dies... Jul 23 20:00:55 the 81xx was nice, pcie/sata 1.5Ghz single core.. Jul 23 20:01:10 they were also very clean in design Jul 23 20:01:15 e.g. PRCM was actually consistent Jul 23 20:01:30 and the security was broken, so you could actually use it . ;) Jul 23 20:01:47 what security? Jul 23 20:02:00 no actually you could *use* the sec Jul 23 20:02:15 yeap, that.. Jul 23 20:02:16 though still no way to configure MReqDomain for initiators Jul 23 20:02:47 which sucks since that's the security identifier proxied by EDMA Jul 23 20:03:27 so configurable MReqDomain is absolutely required if you want memory protection between cores using DMA Jul 23 20:04:12 I really hope they loosen up on that with the am572x since with sooo many initiators, you really want memory protection Jul 23 20:04:31 otherwise stray writes from one core fucking up the state of another is going to be a nightmare to debug Jul 23 20:05:41 odly, i've been trying to get cmem setup right.. for above ^ ;) Jul 23 20:06:18 ? Jul 23 20:07:00 so we can share mem between the a15's and dsp to run opencl stuff.. Jul 23 20:07:29 I actually got the register locations in the am335x control module for MReqDomain .. but it ignores writes. or more likely, ignores non-secure writes Jul 23 20:07:38 (or perhaps all writes on GP devices) Jul 23 20:09:49 I really should find a moment to see if the potential vuln I discovered in secrom actually works... would be interesting to see what is suddenly possible if executing from secure privileged mode Jul 23 20:12:05 (to any TIer who might read that phrase and start hyperventilating: the vuln only applies to GP devices, not to HS) Jul 23 20:14:23 rcn-ee: I'm not really familiar with cmem yet... but iirc on the x15 almost everything is accessible by almost everyone? Jul 23 20:15:58 yeap, lots of masters .;) Jul 23 20:16:32 there are some limits, have dove too deep into it.. Jul 23 20:16:35 rcn-ee: and three EDMA instances... Jul 23 20:17:15 which probably still use MReqDomain as proxied identity, since there's not really any alternative Jul 23 20:17:43 (you can't proxy the true initiator ID on a FlexNOC network like you can on CBA/TeraNet since that ID is also used as destination for the reply packets) Jul 23 20:20:24 and yeah, initiator-target connectivity... that really needs an automated test script Jul 23 20:20:57 the TRM has a matrix, but it probably contains errors and omissions... they usually do Jul 23 20:25:30 which reminds me that I still want to properly map the topology the am335x... damnit, can someone please somehow manufacture more time? there clearly isn't enough of it :P Jul 23 20:29:55 zmatt: map as in decap and go at it with electron microscope? Jul 23 20:30:34 ds2: hehe, no Jul 23 20:31:18 map the l3 topology Jul 23 20:34:54 ds2: like, I once stumbled over this partial map of the centaurus L3 -> http://gerbil.xs4all.nl/centaurus-L3.png ... though it's clearly been hand-edited, so not sure how reliable it is Jul 23 20:37:40 rcn-ee: the http://github.com/beagleboard/linux does not contain unpatched mainline trees? Jul 23 20:38:35 Abhishek___, sorry it's all patch'ed tag's.. Jul 23 20:39:34 I see, np, just thought if I could save some space by not having to clone Linus' tree Jul 23 20:39:49 and keep the beagleboard.org tree Jul 23 20:40:17 Abhishek___, use git clone --reference ../linux git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git Jul 23 20:40:56 when using --reference just don't delete the reference git repo. ;) Jul 23 20:41:30 okay, so this reference would be my current beagleboard.org tree or the Linus' tree? Jul 23 20:41:40 linus Jul 23 20:42:22 Abhishek___, (this is bad example as they are both linus) use git clone --reference git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git Jul 23 20:42:57 it'll use any git objects from the reference, thus saving lots of dl & space.. Jul 23 20:43:11 yup, that's the idea. Jul 23 20:43:20 wouldn't the reverse make more sense? Jul 23 20:43:38 i'm assuming he has the first already downloaded. ;) Jul 23 20:43:50 too late, unless he doesn't mind dl another 1gb. Jul 23 20:44:57 'k Jul 23 20:54:46 rcn-ee: he could locally clone, delete refs, and pull the other repo... would avoid the awkward dependency --reference makes Jul 23 20:55:18 (it would be nice if there were a variant of --reference that does effectively this.. making hardlinks of shared objects) Jul 23 21:01:08 ds2: not sure actually how much you'd be able to tell from a die shot, probably very little... the ones I've seen of e.g. microcontrollers just show a bunch of hard macros and an amorphous blob of logic Jul 23 21:03:19 zmatt: i had a different strategy in mind - you decap a bunch of related but different chips Jul 23 21:03:41 zmatt: the reasoning there is - different parts are documented in different ways Jul 23 21:04:13 heh Jul 23 21:05:01 a very literal meaning of mapping it out Jul 23 21:05:06 well it would settle the question which of subarctic's "ghost modules" really aren't there and which are merely disabled Jul 23 21:05:19 yep Jul 23 21:05:40 but other than that I consider the memory map to be complete Jul 23 21:06:36 ultimately it is of limited interest whether a module is absent or present if it is unusable Jul 23 21:07:27 checking current draw when toggling the module's clock enable would probably answer that more easily than decapping ;) Jul 23 21:08:25 be intersting to see if you can find a hcf bit Jul 23 21:09:11 I suppose it would say something about the design process... i.e. if they removed the PCIe module but left in its target agent and PRCM, then that means removing the latter is considerably more difficult (at that design stage) than removing the target module Jul 23 21:10:11 and I think this is the case, since I really don't believe there's a PCIe module in subarctic Jul 23 21:10:53 especially since if the PCIe PLL had been left in, it would have been saner to keep using it for Ethernet just like it was on Centaurus Jul 23 21:11:30 doesn't PCIe require different drivers? Jul 23 21:11:48 and if so, wouldn't there be hints of it in the hw datasheet for specific pins Jul 23 21:11:58 yeah I'd expect dedicated pins, and none are unaccounted for Jul 23 21:12:16 which is another clear sign PCIe isn't there Jul 23 21:12:51 also, in the ROM code related to PCIe boot, there are actual pinmux values but they are bogus Jul 23 21:13:10 one is already used by another function, the other indexes the pinmux array out-of-bounds Jul 23 21:13:12 I suppose they could just be unbonded pins Jul 23 21:13:13 iirc Jul 23 21:13:28 then there'd still be pinmux registers Jul 23 21:13:47 since they clearly made those for every I/O no matter how rediculous Jul 23 21:13:51 e.g. DDR and even PORz Jul 23 21:14:40 but all pinmux registers are accounted for Jul 23 21:14:52 sigh... even digikey is spiralling down the drain.... Jul 23 21:14:56 The page cannot be displayed because an internal server error has occurred. Jul 23 21:16:25 life is tough... Jul 23 21:44:31 a/win 6 Jul 24 00:28:20 /n Jul 24 01:14:21 hello Jul 24 01:14:43 i'm trying to get a beaglebone black working with a yocto sato build via a sdcard Jul 24 01:15:07 i have the uEnv.txt problem. Jul 24 01:15:51 the first reference I see after uboot comes up says, reading uEnv.txt and gives a byte count. Jul 24 01:16:20 the byte count matches the file size on the sdcard Jul 24 01:16:55 but later it says running uenvcmd, reading /boot/uEnv.txt, **Unable to read file /boot/uEnv.txt** Jul 24 01:17:22 however once it boots from emc, i can look /media/boot and uEnv.txt is there Jul 24 01:18:41 i'm guessing that the first pass where it read it ok, but this line within the uEnv.txt file is the problem. Jul 24 01:18:44 loaduEnvtxt=load mmc 0:1 ${loadaddr} /boot/uEnv.txt ; env import -t ${loadaddr} ${filesize}; Jul 24 01:19:13 i've tried to change that 0:1 to be 0:2 but that did not work. Jul 24 01:19:18 any idea? Jul 24 01:39:13 Is your partition containing the uEnv.txt ext2 or FAT? Jul 24 01:42:13 beware there may be plural uEnv.txt .. put something unique in one to see if it's the one being read :D Jul 24 01:47:34 * zmatt . o O ( plural uEnv.txt .... uEnv.txts? uEnvs.txt? hmm.. ) Jul 24 01:47:45 i finally got it working Jul 24 01:47:53 i'll pastebin my file Jul 24 01:47:54 one sec Jul 24 01:49:59 maybe not Jul 24 01:50:23 once i booted the yocto build, i no longer have a /media with the sd card and its two partitions mounted Jul 24 01:50:43 so I can not get the file while its in the beagle **** ENDING LOGGING AT Fri Jul 24 02:59:59 2015