**** BEGIN LOGGING AT Tue Oct 11 02:59:58 2016 Oct 11 06:40:54 Having tons of fun with my Beaglebone Black desktop Oct 11 06:41:19 Experimentally Oct 11 06:44:57 for science! Oct 11 06:54:28 science and freedom Oct 11 07:30:28 zmatt: im currently stuck in the kernel config menu! Oct 11 07:40:21 i try to use the debian 8.6 console-verison image with my bbgw. Oct 11 07:40:55 my problem is, i cant get the wifi to work Oct 11 07:42:17 it seems that there is no connman on this image Oct 11 07:42:56 when i "ifconfig" the bbgw doesnt find any wlan0 or something like that Oct 11 07:44:08 tommer: earlier you mentioned you successfully built a kernel using build_deb ? Oct 11 07:44:45 BBGw_Rookie: then install whatever packages are needed for wifi on the bbgw? Oct 11 07:46:04 wpa_supplicant is probably a good start Oct 11 07:46:22 although if there's no interface at all maybe firmware of some sort is needed Oct 11 07:47:25 zmatt you mentioned systemd-networkd Oct 11 07:47:38 is it the sam elike wpa_supplicant? Oct 11 07:47:56 no, it would replace ifupdown which is probably installed Oct 11 07:48:03 tlab: Seems like they are just moving some of the pinmuxes to seperate files. Oct 11 07:49:13 you always need wpa_supplicant (which is responsible for scanning for and associating with wireless networks) Oct 11 07:49:41 though first you'll need to get wlan to show up as link Oct 11 07:49:47 i read about that... using /etc/network/interfaces + wpa_supplicant Oct 11 07:50:05 yeah /etc/network/interfaces is the config file for ifupdown Oct 11 07:50:29 how do i check if wifi driver is installed? Oct 11 07:50:32 my experience with ifupdown + wpa_supplicant hasn't been particularly good, but maybe things improved since then Oct 11 07:50:59 I have no idea what's required for the wl18xx Oct 11 07:51:13 ok Oct 11 07:51:30 thx Oct 11 08:17:07 BBGw_Rookie: do you have all needed files under /lib/firmware/ti-connectivity/ Oct 11 08:19:36 zmatt: yes, and then I tried to install all the packages in the deploy folder using dpkg -i, but that didnt work. Which of the files in deploy do I need? Oct 11 08:20:00 linux-image-$version.deb Oct 11 08:20:29 what about headers and firmware? Oct 11 08:20:37 optional Oct 11 08:21:33 also, I need high resolution on the nanosleep() and usleep(), can I achieve this by changing something in the config menu? Oct 11 08:22:22 I think that mostly depends on the ability to preempt Oct 11 08:23:41 hrtimer needs to be enabled too probably but that's the default Oct 11 08:24:34 so, if I change to the "low latency desktop" it should be better? Oct 11 08:24:41 or should I change to RT ? Oct 11 08:25:13 RT can preempt more stuff than LL, assuming you set appropriate process priorities Oct 11 08:25:20 the price is a decrease in overall efficiency Oct 11 08:26:06 hmm, I got my source from the non-RT branch.. It looks like you can configure the non-rt branch to RT Oct 11 09:27:21 Morning Oct 11 09:28:16 I reverted to a previous kernel (4.1.30) on the X15 in order to get OpenCL working, However now i cant load the CMEM module, even after rebuilding it with dkms it tells me "Exec format error" Oct 11 09:29:34 Other kernel modules (debugss, temperature) both build and insert fine Oct 11 10:02:09 zmatt: I get the same errors as before: http://pastebin.com/RnccE4iQ Oct 11 10:06:54 tommer: huh, why do you have that dkms crap, those are all for the bbx15 Oct 11 10:07:22 it doesn't look like a fatal error though Oct 11 10:07:48 which filesystem image did you start with? Oct 11 10:13:52 I had to start with headers since it was needed.. and then I did dpkg -i linux-image... Oct 11 10:15:57 The only thing I did was change the $BUILD to add a custom label as you said, then ./build_deb.. when the config menu came up I only changed from voluntary preemption to "low latency desktop" Oct 11 11:20:42 zmatt: it works even with the errors.. Do you know if setting the Timer frequency higher has any impact? Oct 11 11:35:17 not on anything that uses hrtimers Oct 11 11:35:59 well, it has an impact, it interrupts your code more often hence will increase timing jitter probably Oct 11 11:36:58 ok, i'll leave it then Oct 11 11:37:09 "I had to start with headers since it was needed" ... by? Oct 11 11:37:25 I think only dkms needs it... the solution is to remove dkms Oct 11 11:38:50 dpkg -i linux-image.. complained that the headers were missing Oct 11 11:39:26 like I said, probably that error came from dkms (just like the other errors you showed) Oct 11 11:39:40 I've never needed to install the headers package Oct 11 11:39:52 12:07 < zmatt> which filesystem image did you start with? Oct 11 11:41:23 started with the newest 4.4 if thats whats you're asking? Oct 11 11:41:31 no Oct 11 11:41:36 filesystem, not kernel Oct 11 11:43:15 ext4 Oct 11 11:43:36 I mean, what did you originally download and flash onto your bbb Oct 11 11:44:23 Havent flashed the emmc, running from SD card.. That was the newest debian image from the beaglebone website Oct 11 11:44:37 lxqt ? Oct 11 11:45:09 what was the filename (.img.xz) Oct 11 11:47:21 bone-debian-8.4-lxqt-4gb-armhf-2016-05-13-4gb.img Oct 11 11:47:43 hm ok odd that it contains those dkms modules Oct 11 11:51:16 a bit weird yes.. Still getting quite a bit of jitter with the low latency setup.. maybe I'll try with the RT.. Not really sure why there is a separate RT branch though.. Oct 11 11:52:08 well the branch also includes the defconfig Oct 11 11:52:21 define "quite a bit of jitter" Oct 11 11:55:31 never mind, its just me spawning a new thread for every readout.. :D Oct 11 11:55:56 yeah that won't help Oct 11 11:56:45 I still need two threads for doing the readout, so the question is how can I make sure the readouts gets synchronized without having to spawn a thread for each readout? Oct 11 12:01:17 there's various ways to do so Oct 11 12:02:24 man pthread_cond_broadcast Oct 11 12:02:27 for example Oct 11 12:02:39 or eventfd Oct 11 12:02:54 but probably pthread_cond is faster Oct 11 12:55:01 thanks :) I Oct 11 12:55:04 i'll check it out! Oct 11 13:41:03 Not sure why I need a mutex for wakening multiple threads using pthread_cond_broadcast.. There is no common resource to protect Oct 11 13:51:36 Hi, I disabled hdmi and emmc in uenv.txt, now my BBB doesn't connect on eth1, Serial connection brings me to (initramfs) prompt (not regular promt) and there is no /boot dir. Any idea where I might find the uEnv.txt? Oct 11 13:52:45 PRIVMSG zmatt Thanks for your answer and sorry for the late reply Oct 11 14:21:57 kedidie_: you should find it on your uSD card Oct 11 14:22:05 in the boot partition Oct 11 14:22:24 may be FAT may not .. Oct 11 14:25:59 there is no sd card Oct 11 14:26:21 and there is no /boot Oct 11 14:27:27 how does it boot?! Oct 11 14:27:37 with disabled emmc :P Oct 11 14:28:14 I'm assuming its still ussing emmc to boot ... Oct 11 14:28:30 which is, of course, going to fall flat on its face at some point inthe boot process :D Oct 11 14:29:05 you may be unlucky in that the 'boot.scr' is hard-written to the eMMC directly if it's an older board Oct 11 14:29:20 prior to uEnv.txt being used instead Oct 11 14:30:09 tommer: yes there is. the conditional variable you're supposed to use with a pthread condition. check the manpage Oct 11 14:30:40 kedidie_: you may be better off posting your uboot output to pastebin/etc Oct 11 14:30:47 and linking here Oct 11 14:30:50 http://pastebin.com/g5Yt3X8U Oct 11 14:30:56 jinx Oct 11 14:30:58 :D Oct 11 14:31:04 srry shitty internet connection Oct 11 14:32:18 aye as I suspected .. uboot has found 'a' /boot .. "mmc1(part 0) is current device" Oct 11 14:32:46 but its nowhere to be seen once booted Oct 11 14:32:47 found a kernel Oct 11 14:33:24 jjjj Oct 11 14:33:24 key is in your kernel cmdline .. as printed .. "console=ttyO0,115200n8 capemgr.disable_partno=BB-BONELT-HDMI,BB-BONELT-HDMIN,BB-BONE-EMMC-2G root=UUID=d44e0a47-9e5c-4f97-8460-f231fedf924d " Oct 11 14:33:26 jjjj Oct 11 14:33:33 hm. Oct 11 14:33:48 if you do a "lsblk -f" .. you should find some partitions Oct 11 14:33:51 (I'm sorry for the noise. my ssh session hung.) Oct 11 14:34:14 phschafft: no worse than the join/leave spam ;) Oct 11 14:34:38 it kind of was, just on a diffrent level ;) Oct 11 14:35:26 lsblk is not available in initramfs shell Oct 11 14:36:07 kedidie: no probably not .. fdisk might be Oct 11 14:36:18 and blkid if you're lucky Oct 11 14:36:20 nope Oct 11 14:36:30 ugh it really is bare then. Oct 11 14:36:40 help Oct 11 14:36:46 srry wrong prompt Oct 11 14:36:56 I suggest you write a uSD card with a good image, and use that to troubleshoot Oct 11 14:37:15 ok will do, thanks man Oct 11 14:37:25 you'll then have all the tools. Oct 11 14:37:39 unlimited power Oct 11 14:37:39 don't forget to hold the 'BOOT' button when you power up to make sur the uSD boots ! Oct 11 14:38:23 ok Oct 11 15:33:31 Xabby76: please don't pm me, just communicate in this channel Oct 11 15:35:34 thinkfat: hmm, it does look like pthread_cond does more than I thought (I never use threads really so I'm somewhat ignorant of the pthreads api :P ), I'm not sure if tommer needs that Oct 11 15:37:49 zmatt: a pthread_cond is a little bit like a kernel "completion", only you have to explicitly protect the "done" variable with a mutex Oct 11 15:38:06 zmatt: to ensure exclusive access Oct 11 15:39:00 zmatt: not sure why you don't need to do that in the kernel. Maybe you have to, but it's not forcing you to use a specific method for that Oct 11 15:40:12 a wait queue embeds a spinlock Oct 11 15:40:50 ah no Oct 11 15:41:02 I mean yes it does, but it does indeed support lock-free wakeu Oct 11 15:41:03 p Oct 11 15:45:15 no it doesn't, it does take the lock Oct 11 16:30:41 zmatt you're arguing with you Oct 11 16:30:54 ayjay_t: it happens .. :p Oct 11 16:31:19 I saw nothing out of the ordinary, personally. Oct 11 16:31:46 there is a certain reasoning technique that writing something down either confirms or contradicts it in your own mind Oct 11 16:33:07 dielectic? Oct 11 16:33:25 haha dialectic.. Oct 11 16:33:38 possibly :D Oct 11 16:48:27 ayjay_t: I was just correcting myself Oct 11 16:49:06 waking up a wait queue in linux does in fact take the wait queue's spinlock. now you know :P Oct 11 16:58:46 Does anyone know where U17 on the Beaglebone Green comes from? (SSDCI3128AF) I can't find it anywhere Oct 11 16:59:37 Seems to be a clock generator chip with EMI reduction, but afaik impossible to get hold of Oct 11 17:00:08 also not a good idea, and ditched in favor of a normal crystal again on the beaglebone green wireless Oct 11 17:00:45 it's comes from here apparently: http://www.spreaddevice.com/products.asp?id=33 Oct 11 17:03:12 the description at the top of the page is very confidence inspiring, with the typos, bad grammar, and obvious whitespace errors Oct 11 17:08:40 chinglish ftw zmatt :p Oct 11 17:10:01 has somebody used buildroot with beaglebone green wireless? Oct 11 17:28:02 if i'm using a BBB as a headless wireless access point and removed everything except the networking, is there any reason i'd need the dbus-daemon running? Oct 11 17:29:06 * running debian @ 4.4.23-bone14 Oct 11 17:33:55 a lot of stuff using dbus for communication, including systemd Oct 11 17:34:00 *uses Oct 11 17:34:05 k that is what i was wondering, thanks Oct 11 18:17:00 I have a multithreaded application and my main thread needs to handle command inputs from the command line. Does anyone know of a library or something which makes this easier? I have defined a custom set of commands Oct 11 18:38:48 tommer: what does threading have to do with that? Oct 11 18:38:56 ahh, nm Oct 11 18:40:13 well no the question stands; you just use readline or some alternative in the main thread, this doesn't really have much to do with the other threads Oct 11 18:41:01 you can also keep your program simple and wrap it in rlwrap to get nicer commandline editing Oct 11 18:43:26 rlwrap gives you autofinish+ Oct 11 18:43:27 ? Oct 11 18:43:55 autocomplete you mean, yeah to some degree Oct 11 18:44:23 what else does it do? Oct 11 18:44:29 read its manpage? Oct 11 18:44:36 only read on git Oct 11 18:44:53 ? Oct 11 18:45:49 nvm, found the manpage now.. I'll go through it, talk tomorrow :) Oct 11 22:14:23 Bah. Have a BBB “A5B”. Isn’t booting. Get power LED, nothing else. Hooked up serial, and I see the uboot SPL banner, but NOTHING else. This is true whether there is an SD card present or not, or whether I hold the SD boot button while resetting - reset just belts out the version banner for SPL. Oct 11 22:15:12 Digging around, I see there’s a UART boot. Awesome TI wiki specifies a windows program “terraterm”. I’m not running windows. Have Linux and OSX. Oct 11 22:15:52 I can bridge the two connections with 4.7K resistors as documented elsewhere online and the device comes up and prints eight “C” characters to the serial. Oct 11 22:26:38 I can use xmodem to push the u-boot-spl.bin that I compiled from the uboot sources for the beagle, But besides it not resulting in a uboot banner, I don’t know that the build is just right for the BBB or that the xmodem send is quite right. For one, while the spl binary is about 70KB, xmodem is claiming “595 blocks” - when I’ve specified the 1KB Oct 11 22:27:03 anyone here? I’d like to get my board running again. Oct 11 22:27:26 splud: the SD boot button is sampled at power-on, not at reset Oct 11 22:28:01 if you have no card inserted then you really can't get an SPL banner if you power on with SD held Oct 11 22:28:25 if you do have a valid bootable card inserted it should simply boot in that case Oct 11 22:29:11 this is the simplest method of recovery, with USB being a good alternative. I really wouldn't try to hasslle with xmodem Oct 11 22:29:57 okay, if I disconnect power, depress the SD button and keep it held, and apply power, I get a one-line SPL header. Oct 11 22:30:13 with no card inserted? Oct 11 22:30:23 with card inserted. Oct 11 22:30:38 (and UART bootstrapping resistors removed) Oct 11 22:31:06 try first with no card inserted to confirm you get no prompt Oct 11 22:31:10 then insert card and press reset Oct 11 22:31:40 btw instead of holding the button you can alternative pull LCD_DATA2 to ground with resistor (that's also what the button does) Oct 11 22:31:45 *alternatively Oct 11 22:37:14 If I swap SD cards (different builds), I get different SPL banners. Oct 11 22:37:15 U-Boot SPL 2013.10-g78d8ebd (Mar 30 2014 - 20:46:34) Oct 11 22:37:24 U-Boot SPL 2013.04-dirty (Jul 10 2013 - 14:02:53) Oct 11 22:40:40 but nothing beyond that? Oct 11 22:40:42 but NOTHING beyond that. Both have uEnv.txt file Oct 11 22:41:27 are those cards known-good ? they're clearly very old Oct 11 22:41:28 remove card, I get the CCCC… or bridge two of the IO lines to ground with 472 resistors (4.7K) Oct 11 22:41:36 and I get _8_ C, no more. Oct 11 22:41:49 which two? Oct 11 22:42:15 how are you powering it btw? Oct 11 22:42:35 1.5A 5V wall wart. Oct 11 22:42:57 have FDTI serial adapter, have tried using both Linux workstation and OSX. Oct 11 22:43:09 ok Oct 11 22:43:33 which two did you bridge to ground to get just the 8 CCs? Oct 11 22:45:25 LCD_DATA3 and LCD_DATA2, each through their own 4.7K over to DGND Oct 11 22:47:16 uhhh, don't. that boot config is _not_ compatible with the bbb's pcb layout Oct 11 22:47:48 design I should say, schematic, whatever Oct 11 22:48:29 and the reason it stops after uart is because it'll be trying ethernet boot which occupies it a while Oct 11 22:49:12 00:41 < zmatt> are those cards known-good ? they're clearly very old Oct 11 22:50:11 was that last question directed to me? Oct 11 22:50:26 I downloaded images and wrote them last night. Oct 11 22:50:27 yes I was repeating my own question since you hadn't answered it yet Oct 11 22:50:57 I’d be happy to get a newer image. Oct 11 22:51:08 uhh, which did you download? I don't expect an u-boot from 2013 on a recent image Oct 11 22:52:30 Angstrom - what was originally on the board. Was trying to change the minimum Oct 11 22:52:55 Since it had been working when I last packed it away. Oct 11 22:53:49 weird Oct 11 22:53:57 angstrom has been ditched a long time ago though Oct 11 22:54:01 in favor of debian Oct 11 22:54:55 http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#microSD.2FStandalone:_.28console.29_.28BeagleBone.2FBeagleBone_Black.2FBeagleBone_Green.29 Oct 11 22:55:39 I figure you want a microsd/standalone image rather than a flasher to have the opportunity to inspect what happened to eMMC Oct 11 22:55:53 I got to some page for the downloads and it was all Angstrom. Oct 11 22:55:54 the console image is very small Oct 11 22:56:10 then you didn't get the official page but something very old Oct 11 22:56:26 which of the spl binaries corresponds to the MLO file? Oct 11 22:56:34 ? Oct 11 22:57:02 the official images are here btw: http://beagleboard.org/latest-images but I'd personally prefer the snapshots of the wiki page I linked above Oct 11 22:57:27 *on the Oct 11 22:58:12 and I have no idea what you mean with "which of the spl binaries corresponds to the MLO file?" Oct 11 22:59:12 MLO is a particular (simplistic) boot image format understood by the ROM bootloader, it will contain the u-boot spl (when using u-boot) Oct 11 23:00:20 note that jessie images don't contain an MLO file anymore as such since it uses the raw mmc boot method Oct 11 23:01:34 I know the MLO is the u-boot spl, just not sure that from the build uboot (and spl subdir) if the u-boot-spl.bin corresponds to the MLO file or not. Oct 11 23:01:51 how’s the SD booted as “raw mmc” ? Oct 11 23:02:43 I've given a brief description of the two different sd/mmc boot methods here: https://github.com/mvduin/bbb-asm-demo you can also find more details in the TRM if you care Oct 11 23:03:58 the jessie images btw place the raw mmc boot header in sector 256 (hence MLO starts in sector 257) Oct 11 23:04:25 Very werid. I replaced the MLO file with the built u-boot-spl.bin file. If I run strings on my build, I see the banner (and only ONE copy of it): U-Boot SPL 2016.07 (Oct 11 2016 - 13:27:03) Oct 11 23:04:46 u-boot-spl.bin is not an MLO file Oct 11 23:05:47 please just download a known good bootable image and flash it onto a card, you can always experiment with bootloaders later if you want once you got your bbb functional again Oct 11 23:07:16 if I remember correctly the ancient u-boot used in the angstrom era had the tendency to prefer loading u-boot from emmc instead of from microsd. since apparently the u-boot on your eMMC is somehow corrupt this explains why you have trouble booting from sd as well Oct 11 23:07:40 current versions afaik no longer have that problem Oct 11 23:07:58 Off to download yet another image… Oct 11 23:08:04 Thanks. Oct 11 23:08:12 the console image is really small Oct 11 23:25:46 I remain unclear how a new sdcard image is to change anything for me if the device isn’t even booting from it in the first place. Oct 11 23:26:06 Which is why I was trying to bootstrap it from the UART. Oct 11 23:33:44 I just explained above my hypothesis why you failed to boot from sd card Oct 11 23:35:07 "bootstrapping" via UART makes no sense, it is basically the same as using an sd card except way more involved Oct 11 23:45:58 apart from the issue described above, the only thing that could prevent booting from sd is damaged hardware Oct 11 23:48:35 Anyone willing to help with beaglesnes install on beaglboard xm with debian? Oct 12 01:16:35 Hello folks. Given that the web page status has not been updated, I am curious if the X15 boards are in production as planned? Apologies if this is not the right place to ask. **** ENDING LOGGING AT Wed Oct 12 02:59:58 2016