**** BEGIN LOGGING AT Mon Oct 05 02:59:58 2015 Oct 05 05:05:02 hum, is hdmi output broken in 4.2.3-bone2 or is it just me? Oct 05 05:30:43 I'm trying to get Debian images to flash but it doesn't seem to want to do so. Not sure what I'm doing wrong. I'm using Flasher image, hold button apply power, wait for led, and I think its booting the old OS onboard. Oct 05 05:38:33 one way to exclude procedural errors (e.g. holding the wrong button) is by trying it without sd card inserted -- it should fail to boot (only power led on) Oct 05 05:39:17 (powering up with the sd button pressed takes eMMC entirely out of the list of boot devices) Oct 05 05:41:00 Yes ... so why does it never finish the flash? Oct 05 05:41:15 if you're powering using an AC adapter or USB charger you can then insert the card and it should boot the flasher (if you're powering via USB from a computer that won't work since it'll be stuck trying to boot via USB) Oct 05 05:42:22 it's possible that a broken flasher might boot the old OS, I'm not sure... it's possible the bootloader on the sd card might fall back to booting eMMC if it can't boot from sd card Oct 05 05:42:34 I've never dissected the flasher tbh Oct 05 05:44:09 I've tried two different firmwares now and the old OS was hosed. Now I can't flash over it or get anything else to boot ;( Oct 05 05:46:05 eMMC can't affect the BBB's ability to boot from sd card Oct 05 05:47:17 and a properly working flasher should simply overwrite eMMC, regardless of what was on it Oct 05 05:47:25 I wouldn't think. Any ideas on what's going on? I don't have a serial port to debug anything and I can't get output on the HDMI Oct 05 05:47:46 you can also try a normal bootable sd image Oct 05 05:48:04 It just blinks its little LEDs at me for hours Oct 05 05:48:53 if the flasher is running then it shows a distinctive "knight rider" sequence, and after a few minutes all leds stay on to indicate it's done Oct 05 05:49:08 how did you write the flasher to the sd card? Oct 05 05:49:27 if the leds do other kinds of blinky blinky stuff then you're staring at a running system of sorts Oct 05 05:49:27 what was the exact name of the image you wrote to the sd card? Oct 05 05:49:50 Can't try normal image for another hour .. thats when my Internet gets cheaper. And I was waiting for Knight Rider ... not seeing him. Very annoying Oct 05 05:51:32 you're still leaving tbr's questions open Oct 05 05:52:20 Yeah ... I know what all the LEDs.do. Image is BBB-eMMC-Flasher-debian-8.2-console-armhf-2015-09-27-2gb.img Oct 05 05:52:43 And I'm on my phone right now ... tough to type that filename Oct 05 05:52:59 kept wanting to autocorrect it :-) Oct 05 05:53:01 ok, that's a console image Oct 05 05:53:08 so that means no HDMI Oct 05 05:53:34 without a serial cable that means ssh'ing via ethernet is the only way in Oct 05 05:54:12 Aha! I thought they meant fb console not serial console Oct 05 05:54:29 I just don't need X and crap. Oct 05 05:55:00 hmm, I might be wrong, not sure now Oct 05 05:55:18 tbr: does the console image still have a framebuffer console? Oct 05 05:55:25 no idea Oct 05 05:55:29 * zmatt almost never has a screen attached to a BBB Oct 05 05:55:30 Grr ... the old Ubuntu and Angstrom images did Oct 05 05:55:54 uudruid-74: it might, I'm really starting to doubt my earlier statement Oct 05 05:56:05 since thinking about it, I see no reason why it wouldn't have one Oct 05 05:56:12 I won't have a screen later, but I want it long enough to set it up Oct 05 05:56:19 they probably have fbconsole, but who knows Oct 05 05:56:32 ohhh Oct 05 05:56:55 uudruid-74: also how did you write the image to sd-card? Oct 05 05:56:55 no wait, even the console image use -ti kernels, so that's not the issue either, hmm Oct 05 06:00:35 This thing has been absolute hell. I keep downgrading my plans for it. It used to work on my Samsung monitor, but it freaks out the Sharp TV ... have to reboot the TV to fix it! Now I just want it to run a backup drive via USB ... everything was set, I brought it downstairs, and it failed. Attached monitor, no output. Decided to reflash it and start over, and it wont flash. Oct 05 06:01:25 uudruid-74: a console cable is very much a lifesaver when trying to debug problems with booting or flashing Oct 05 06:01:44 otherwise you're blind Oct 05 06:02:44 And spend MORE money on this thing? Its not even a good paperweight. Oct 05 06:03:00 it's definitely not a good paperweight... it weighs far too little Oct 05 06:03:17 might also scratch the paper Oct 05 06:03:32 No .. mine is in a case Oct 05 06:04:04 a UART cable is a universal embedded tool Oct 05 06:04:11 everyone should have at least one Oct 05 06:05:07 and they can be had for as little as a 3-5 €/$/... Oct 05 06:05:20 True. I almost bought one for the router Oct 05 06:05:34 I just looked at the console image, I conjecture it *should* have a framebuffer console Oct 05 06:05:56 it's a -ti kernel and default dtb, so HDMI should be enabled Oct 05 06:06:38 Well, my other SD card is flashing (Ubuntu). Dunno why none of the Debian images work. Oct 05 06:07:12 * zmatt shrugs Oct 05 06:08:08 They ever get a working hardware graphics accell on this thing? Oct 05 06:08:40 EGL works Oct 05 06:09:39 I had the "MagicLantern" PowerVR demo running pretty smoothly at 1280x1024 with basically zero cpu load Oct 05 06:09:57 but what can take advantage of it? I never understood why they can't run X over that Oct 05 06:12:11 the problem has something to do with missing magic glue between sgx, drm, and the lcdc driver, or something like that Oct 05 06:12:14 I'm not quite sure Oct 05 06:12:58 the am335x lcdc driver doesn't give a rock solid impression in general Oct 05 06:13:45 Hell, and with the iffy compatibility ... it really freaked out my TV Oct 05 06:14:39 Guess I'm using Ubuntu on this instead of Debian since it loaded and came up. Oct 05 06:14:39 that would probably be the tda998x driver's fault Oct 05 06:16:15 (or possibly the tda19988 itself, although that thing actually seems to be quite nice) Oct 05 06:17:09 Hmm ... I keep having visions of running Plex and plugging in a USB Wifi, but every attempt at WiFi has been met with odd problems that don't make sense. Oct 05 06:17:25 I should be happy it boots again :-) Oct 05 06:17:28 I mean, it even usually manages to transmit a stable image even though lcdc is configured to shift data out on the wrong clock edge by default Oct 05 06:21:20 uudruid-74: usb and video are probably the weakest points of the BBB Oct 05 06:22:42 Great. The two things I want the most. I shoulda got the Pie huh? Oct 05 06:24:08 the rpi is actually mainly a GPU, with an ARM core stuck on the side Oct 05 06:26:13 the BBB's strength is in its I/O Oct 05 06:26:32 I mean, the PRU can freaking bitbang at 100 MHz Oct 05 06:26:48 Trying to decide how I wanna network this without dragging stuff from downstairs. Guess I have to. And maybe I was smart enough to save that smb config file to the hard drive (doubt it). Oct 05 06:28:05 I'm gonna see why combining my dtb with the ti kernel ends in "Starting kernel ..." and then death Oct 05 06:28:17 I don't have any capes and the only IO I would want is audio and I heard that another no-go for Beagle. I was hoping maybe USB Audio would be OK. The capes were all horrible.and expensive as hell. Oct 05 06:28:41 it has audio via hdmi Oct 05 06:29:30 Not IN Oct 05 06:29:59 you could use an adc channel ;) Oct 05 06:30:29 The on board ADC? Oct 05 06:30:38 I'm not being very serious Oct 05 06:30:49 Oh ...good Oct 05 06:30:51 those are 10-bit channels, not audio-quality Oct 05 06:31:12 but it can do audio input via I²S, but you need external hardware Oct 05 06:32:39 And most external hardware is incredibly expensive for a 50 cent chip. I can go USB to test things out at a fraction of the price, but you say its USB is flakey Oct 05 06:58:27 ok, the -ti kernels clearly have a better working lcdc driver than the -bone kernels Oct 05 07:01:59 difference being? Oct 05 08:04:00 Hey Oct 05 08:04:17 i don't know if this is a weird issue but Oct 05 08:04:34 i can connect to a bbb with ssh via putty Oct 05 08:04:59 but i cant open the page putting 192.168.7.2 on chrome or firefox Oct 05 08:05:41 is the web server running? Oct 05 08:06:44 * uudruid74 cant get Ethernet working. Grrrr Oct 05 08:06:53 don't know Oct 05 08:07:04 recently bought and installed drivers Oct 05 08:07:24 drivers? Oct 05 08:07:39 BBB drivers, i imagine Oct 05 08:07:49 are you on windows? Oct 05 08:08:00 yup Oct 05 08:08:03 8.1 Oct 05 08:08:08 did you restart the machine after installign drivers? Oct 05 08:08:09 I asked about the web server Oct 05 08:08:19 did you disconnect and re-connect the BBB? Oct 05 08:08:33 yeah Oct 05 08:08:46 yeah... webserver makes more sense ;-) Oct 05 08:08:54 how to check if it is running? Oct 05 08:09:13 ssh works .. ssh in Oct 05 08:09:48 if it uses apache2 Oct 05 08:09:55 there is no one installed Oct 05 08:10:28 just do a ps ... cant be much running Oct 05 08:10:50 nothin Oct 05 08:11:07 or run netstat for something on port 80 Oct 05 08:11:42 netstat -anp | leas Oct 05 08:11:46 err less Oct 05 08:12:01 I'm on my phone, sorry Oct 05 08:14:04 wb Oct 05 08:14:18 bad internet connection Oct 05 08:15:37 * uudruid74 rolls up some newspaper Oct 05 08:16:08 BAD! Bad! BAD Connection! NO! Oct 05 08:17:59 spannish connection... U.U you would be delighted Oct 05 08:19:40 Proto Recv-Q Send-Q Local Address Foreign Address State User Inode tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN 0 6307 tcp 0 0 192.168.7.2:22 192.168.7.1:2420 ESTABLISHED 0 7348 tcp6 0 0 :::22 :::* LISTEN 0 6317 udp 0 0 0.0.0.0:67 0.0.0.0: Oct 05 08:19:53 Actually I'm in the middle of nowhere. I use a repeater to get a cell signal and Internet is satellite Oct 05 08:20:54 nothin on 80 Oct 05 08:21:34 pastebin is your friend ... cause that's not readable in this font. But ... that explains why the web server doesn't respond. Its not running! Oct 05 08:22:33 You can look in /etc/init.d and see if there is a startup script for it. Likely something like /etc/init.d/httpd start Oct 05 08:25:53 looks like my flash issue was the SD Card. Now I'll see if Debian 8.2 is better than Ubuntu since my ethernet refused to work :( Oct 05 08:34:08 Yes, Debian made ethernet work Oct 05 08:36:34 Now I can do a apt-get update, upgrade, install samba and maybe plesk ... and go to bed. Config it tomorrow. Whioooo ..'' Oct 05 08:42:14 SOB! Half the repositories are IPV6 Who decided I needed IPV6 cause this BS ISP doesn't support it. Now what? Oct 05 08:42:41 ehh, all of the repos are reachable via ipv4 Oct 05 08:43:09 I know, because our brand new fiber has.... no IPv6 support Oct 05 08:43:15 (newly deployed infrastructure -.- ) Oct 05 08:44:31 btw, don't forget to upgrade to stretch if you're still on jessie :P Oct 05 08:45:17 and then ditch ifupdown and isc-dhcp-* in favor of systemd-{networkd,resolved} Oct 05 08:45:18 \o/ Oct 05 08:47:21 I think the latest images already replaced ntpd by systemd-timesyncd, but jessie's systemd is too old to make networkd a good option Oct 05 08:49:42 still? The jessie stuff was only a couple weeks old! WTF? And it was marked as preliminary Oct 05 08:50:01 jessie is debian stable Oct 05 08:50:44 Says jessie snapshot dated 9-27 ... that was 2 weeks ago Oct 05 08:50:57 yes, it's a recent image Oct 05 08:51:01 ... of debian stable Oct 05 08:51:07 which is ancient cruft by definition Oct 05 08:51:09 :P Oct 05 08:51:34 And it uses systemd? Oct 05 08:51:42 uhh, yeah? Oct 05 08:51:47 I'm gonna cry. Literally Oct 05 08:53:08 I don't want 100 damn demons on a beaglebone. I could do startup config with 10 lines in a file Oct 05 08:53:31 so far whenever I've replaced some cruft by a systemd equivalent, the result worked better and/or resulted in faster boot Oct 05 08:53:52 ifupdown + dhcp will hang for a long time during boot if you don't have ethernet plugged in Oct 05 08:55:14 I don't want any of it. Its a damn embedded system. I want 'ifconfig eth0 192.168.13.155 && route add default gw 192.168.13.1' Oct 05 08:56:59 you realize that running that in a shell is a heavier load on your system than networkd is right? (which, contrary to what the name might suggest, doesn't run as a discrete daemon) Oct 05 08:57:26 how is it a heavier load? Oct 05 08:57:42 you're doing the same thing, but in shell scripting instead of in C Oct 05 08:58:07 and using multiple process spawns to accomplish the task Oct 05 08:58:54 networkd still has to parse config files and I hate to tell you this, but its a fucking UNIX box. Unix boxes run shells. Oct 05 08:59:23 yes they do, but spawning tons of them during boot is not lightweight Oct 05 08:59:37 but, dude, config your system any way you want Oct 05 09:00:06 And when I wanna change it, its easy. Who said spawning tons of them? How many do I need for an embedded box? Oct 05 10:08:11 Night all Oct 05 13:10:10 Can anyone help me with connecting to a VNC on my BBB (im running angstrom v2012.12). I have installed x11vnc, tried to run it and connect to it using tightVNC-viewer in windows. When i connect i see the login screen saying "logging in automaticaly" and it is just stuck there. Dose anyone know what I am doing wrong? Oct 05 13:10:35 is x11vnc running on the BBB? Oct 05 13:11:05 ah I guess it is if you see something in the viewer Oct 05 13:11:20 er, why do you run 3y old angstrom? Oct 05 13:11:41 well yes it is running on the BBB. :) Oct 05 13:12:03 I do not have the SD port in my comp to update it -.-' Oct 05 13:17:18 well Oct 05 13:17:26 a card reader is like $5 Oct 05 13:19:21 Yes I know and I am going to update it when I get one. I would just like to know if I am doing anything wrong. I am using this command to start x11vnc: x11vnc -bg -o %HOME/.x11vnc.log.%VNCDISPLAY -auth /var/run/gdm/auth-for-gdm*/database -display :0 Oct 05 13:20:51 well have you tried looking at the logfile it creates? Oct 05 13:20:59 If you think that would/should work using that command, then I will wait for my card reader to arrive.. It would just be great if i could make it work now :) Oct 05 13:21:26 do you have the board connected to a display? Oct 05 13:22:10 No i only have the USB and an ethernet cable pluged into the BBB. Oct 05 13:22:57 quite possibly the graphical interface fails for some reason then Oct 05 13:31:49 The logs dose not say that an error occurred. It states that I exited the connection which i did after several minuts. The last thing the log contains before i close the connection is "called initialize_xfixes()" Oct 05 13:34:41 Can this be a bug caused by the old/outdated angstrom version? If so ill wait untill I can update it Oct 05 13:35:46 angstrom as far as I'm aware... is at this point old and outdated Oct 05 13:35:58 debian is the default now for bbb Oct 05 13:36:54 I'll update to debian asap then. ty Oct 05 13:39:54 that doesn't mean though that one couldn't easily build an up to date angstrom image for BBB though Oct 05 13:41:31 Well, your definition of easy might be a bit different from mine :) Thank you for the help with the drives btw it worked flawlessly Oct 05 13:42:44 good to hear Oct 05 16:37:14 Hello there. When flashing eMMC from SD card on BBB, the user's diodes are supposed to flash too? Oct 05 16:38:54 yes Oct 05 16:39:17 depending on your flasher image either in a random pattern or in knight rider mode Oct 05 16:45:45 ok thanks, it appears it's completely random. I get from that, that there's no way to acquire the progress of flashing? Oct 05 16:46:22 jalooc, which image are you flashing? Oct 05 17:10:55 jalooc.. only via the serial interface Oct 05 17:11:06 ok thanks Oct 05 17:11:08 +debug Oct 05 17:21:39 I'm using the Debian 8.2 image and apt-get update cant seem to get its repositories. Refuses to connect. Anyone else seeing this? Oct 05 17:22:25 uudruid74: ethernet cable connected? Oct 05 17:22:42 Yes Oct 05 17:23:02 It works for the first couple of sites on the list Oct 05 17:23:59 pastebin the output? Oct 05 17:25:54 Just says unable to connect to ftp.us.debian.org:http: Oct 05 17:26:54 hmm syntax error in that statement... Oct 05 17:27:57 I'm having a separate issue with my router where I cant pass packets between network segments so I cant ssh in and have to use console but everything connects to Internet. Oct 05 17:28:08 veremit : What error? Oct 05 17:30:00 here is an (old) sources.list from /etc/apt/sources.list .. http://pastebin.com/Uq6NZ23q Oct 05 17:30:37 that's obviously a 7.x debian but its live on my beagle currently Oct 05 17:30:38 Hmmm .... I've got some other network issues. Damn. This works just moving the cable to my PC, but can't resolv.hosts on beagle. I have the Google DNSs in resolv.conf. I think I need a hammer Oct 05 17:30:51 sounds like it :) Oct 05 17:34:21 And my laptop can't talk to the firewall cause its Wifi is fucking up and you can only get in that way. Hmm .. I'm gonna be stupid and route the traffic through my phones Wifi using bluetooth and see if I can access the Cisco that way ... which will drop this connection Bye all! Oct 05 17:34:45 wow .. don't make it easy, will you ?! :P Oct 05 17:47:19 I am currently trying to flash my BBB for the 2. time since the LED's just start blinking in sync when it is finished instead of turning off. I am using the latest img from the web page "bone-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb.img" and I was also confused when i saw that the MD5 hash didn't mach the one listed under the img? Is the latest img corrupt? Oct 05 17:49:31 doesn't look like a flasher image from the filename .. Oct 05 17:49:50 and if your md5 doesn't match .. guessing the download didn't complete properly Oct 05 17:51:21 There is no img available for flashing only the sd option and then change the /boot/uEnv.txt to make it flash the eMMC, right? http://beagleboard.org/latest-images Oct 05 17:52:40 and the flashing of the eMMC is under getting started step 8. Oct 05 17:54:38 I would refer you to http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#2015-03-01 .. for official images Oct 05 17:55:00 as usual .. documentation is a mess :) Oct 05 17:56:43 ping rcn-ee for clarification Oct 05 17:57:03 I'd tear jkridner a new one if he was here .. Oct 05 17:57:16 Ohh, I see. I'll go ahead and DL that instead :) Is that the most up to date version? Oct 05 17:57:36 lack of consistent clear, concise instructions anywhere .. just four/five sets of completely inconsistent ones Oct 05 17:57:57 Jackix .. that would be my recommendation :) much simpler Oct 05 17:58:53 Perfect, Thank you very much for your clarifying help veremit! Oct 05 17:59:06 :) np Oct 05 17:59:52 don't forget to power off, before powering up with the 'boot' button down! Oct 05 18:00:11 will do :D Oct 05 18:00:35 its not a hard process .. just lots of gotchas! Oct 05 18:05:03 when flashing a new image on BB black, i'm getting all 4 user leds blinking.. does that mean it's complete, i was expecting them to be solid on Oct 05 18:05:26 EvilTelephone.. which image did you use? Oct 05 18:05:41 bone-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb.img Oct 05 18:05:46 ohh the irony :D Oct 05 18:05:54 hmm Oct 05 18:06:15 yeah thinkin that one might be a dud .. unless anyoe here knows different .. lol Oct 05 18:07:03 have a go with this one, EvilTelephone .. http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#2015-03-01 Oct 05 18:07:44 that's the one i used Oct 05 18:07:44 I'm guessing its the stand-alone image Oct 05 18:07:54 you want the one labelled 'flasher' Oct 05 18:08:11 oh. i see what you mean Oct 05 18:08:15 thanks, i'll give it a try Oct 05 18:08:15 yeah ... :) Oct 05 18:08:44 there seems to be a few different options to write the emmc .. personally I think the flasher images are best :) Oct 05 18:09:10 don't forget to power-off before holding down the 'boot' switch .. and you can remove the SD once done :) Oct 05 18:11:56 thanks a bunch, trying it now Oct 05 18:12:20 That 7.8 image is bad. I extracted the ext4 partition and ran fsck on it Oct 05 18:12:30 It needs to be taken down Oct 05 18:12:51 :c Oct 05 18:12:52 proabbly isn't ext4 :p Oct 05 18:13:07 and judging by your network issues, uudruid74.... lol Oct 05 18:13:08 It is Oct 05 18:13:36 Try it yourself. And md5 was verified Oct 05 18:14:14 And it wasn't downloaded from here anyway Oct 05 18:14:45 well dose the 7.8 flasher img from the link veremit provided work? Oct 05 18:15:14 alas I don't have a spare beagle to try .. but I will d/l the images to have a peek Oct 05 18:17:01 I'll try the img in X time. Im downloading it now, but my internet is having a stroke atm.. -.-' Oct 05 18:17:31 aye mine's slow too Oct 05 18:17:55 Jackix: i'm writing the image now, i'll let you know Oct 05 18:17:58 veremit You don't need a beagle to test the image. Any Linux box will do Oct 05 18:18:43 uudruid74 .. my linux box is x64 and doesn't run arm code very well ... nor does it have an mmc interface on it .. Oct 05 18:18:48 EvilTelephone: Thank you. Might save me some time :) Oct 05 18:19:17 although that doesn't stop me examining the image :P Oct 05 18:19:40 Veremit Don't boot it just run fsck on the second partition Oct 05 18:19:44 pretty sure I have kpartx on here :D Oct 05 18:20:02 You can do that without writing it to an sdcard Oct 05 18:20:12 oh rats perhaps not Oct 05 18:20:24 will have to do Olde Skool Oct 05 18:20:44 Another great tool to get at the partitions is testdisk Oct 05 18:21:23 anyhoo .. seems I gotta wait an age for download .. so I'm gonna go shower :D Oct 05 18:21:35 bbiaf Oct 05 18:21:36 please do Oct 05 18:21:47 lol Oct 05 18:25:13 Looks like network issues may have been a BT vs Wifi conflict. Oct 05 18:36:50 And maybe something needed some reboots. All working now. Slow as hell, but its satellite, so slow means its operating properly Oct 05 18:54:17 it's alive!! Oct 05 18:54:21 that image worked great Oct 05 18:55:22 Jackix: the link veremit provided worked like a charm Oct 05 18:56:30 EvilTelephone: Thank you, im still DL'ing the img. I don't know what is going on with my internet. Oct 05 20:00:52 Just for completeness of the earlyer problem. The md5 hash sum is also wrong for the flashing img. Not that it matters as loong as it works. (im going to flash my BBB im a sec Writing to SD now) Oct 05 20:07:14 Jackix, which image? Oct 05 20:07:46 BBB-eMMC-flasher-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb.img Oct 05 20:08:43 server: 3236bffc98b12ed67adcbaae9d0c13da2e3a23a71e89255fd0daf3ac94fc8f67 BBB-eMMC-flasher-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb.img.xz Oct 05 20:08:44 But it all started with the suggested latest img from "http://beagleboard.org/latest-images" (which dose not work) Oct 05 20:09:25 err.. i mean: 38bedfc81de00907ff2913b04bdc6fe9 BBB-eMMC-flasher-debian-7.8-lxde-4gb-armhf-2015-03-01-4gb.img.xz Oct 05 20:09:28 (md5sum) Oct 05 20:09:40 the MD5 i get from the img is "b7cab5b87dafc5a33076ea52d617f765" but the one on the webpage is "38bedfc81de00907ff2913b04bdc6fe9" Oct 05 20:10:06 Jackix, md5sum is run on "*.img.xz"... Oct 05 20:10:11 not "*.img"... Oct 05 20:10:23 OHH.. my bad.. Oct 05 20:10:55 and default one ON "http://beagleboard.org/latest-images" is a standalone (microSD) it doesn't flash unless you tell it tooo.. Oct 05 20:11:24 I tried that but it boots wrong.. get flashing LED's Oct 05 20:11:56 it Flashes* wrong if i edit the /boot/uEnv.txt Oct 05 20:11:58 Jackix, yeah... so did you login into it via, ssh/cloud9/serial/etc.. ? Oct 05 20:12:13 ssh Oct 05 20:12:35 Jackix, btw, future versions include a *.sha256sum file to make the dl verfication easier: https://rcn-ee.com/rootfs/bb.org/testing/2015-10-04/lxqt-4gb/ Oct 05 20:13:12 Jackix, once you login via ssh: do: http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#Flashing_eMMC Oct 05 20:13:14 Ohh, thats neat Oct 05 20:13:59 That was exactly what i did when tried flashing the suggested build. Oct 05 20:14:35 But it failed. EvilTelephone had the same problem right after I asked in here. Oct 05 20:15:15 So it logs to both video/serial, but it might have been fixed with "2015-07-28" http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#2015-07-28 Oct 05 20:16:28 as soon as we get "signed" winxp/7/8/8.1/10 drivers again, we will roll another "wheezy" update... Oct 05 20:16:31 I am currently flashing the BBB with the img i just downloaded (where you corrected me on the MD5 hash) and it might work Oct 05 20:17:01 strange, as soon as you make the /boot/uEnv.txt modifcation those two images are exactly the same. ;) Oct 05 20:17:10 that's the only diff.. Oct 05 20:20:44 Sorry my internet dropped. If there is no diffrence between the 2 img's then i might be in for a flashin LED show in 20 mins time. -.-' If it flashes then the sugsted img might be corrupted Oct 05 20:21:13 I'm not sure if it's at all relevant in this context, but I recall running into an issue where an image would not boot from eMMC due to missing /uEnv.txt to fix the mmcargs Oct 05 20:21:42 Jackix_, the "flasher" will blink all 4'led's together on a failure... Oct 05 20:22:28 Yes sorry switched it arround. Oct 05 20:22:37 zmatt, yeah those release images have the old bootloader /uEnv.txt trampoline set, good for Angstrom 2013.6.30+ ;) Oct 05 20:22:51 no, I'm talking about recent u-boot Oct 05 20:23:06 it thinks root=/dev/mmcblk1p1 since, well, it is mmc 1 Oct 05 20:23:11 except the kernel disagrees about that Oct 05 20:23:44 I removed all the other crap from /uEnv.txt Oct 05 20:23:45 yeah, the partuuid in v2015.10-rc4 will take care of that... (make it fun to use initrd's thou..) Oct 05 20:24:46 http://git.denx.de/?p=u-boot.git;a=blob;f=include/configs/ti_armv7_common.h;h=6dc65158994a3479b293da37d2a13ac07e43b45e;hb=HEAD#l64 Oct 05 20:26:07 rcn-ee: btw, we currently also have a bbb setup with graphics... except I get nada with a -bone kernel, only -ti kernels work Oct 05 20:26:41 -bone just produces occasional kernel messages about lcdc: timeout waiting for.. something.. don't remember :P Oct 05 20:26:43 zmatt, sgx accerleration enabled? Oct 05 20:27:05 OK. I have drivers and firmware loaded for my WiFi single using BBB Debian 8.2. This is console image. Where are the tools/steps for connecting to SSID? Oct 05 20:27:23 s/single/dongle/ Oct 05 20:27:28 zmatt, which v4.x is the bone? Oct 05 20:27:29 not afaik? your lxqt image Oct 05 20:28:32 originally 4.1.6-ti-r16, now 4.1.9-ti-r20 (actually, just moved to the -rt version) Oct 05 20:28:56 uudruid74, https://wiki.debian.org/WiFi/HowToUse (lots of ways, both connman & wpa_* is insatlled) Oct 05 20:29:21 4.2.3-bone2 failed, 4.1.10-bone too Oct 05 20:29:51 rcn-ee installed where? Tab complete doesn't find it Oct 05 20:30:06 ditto on my console image with a hdmi-enabled dtb Oct 05 20:30:09 This img dose not work either (just flashing LED's right now... ) Which img would you suggest? Oct 05 20:30:14 but... Oct 05 20:30:16 zmatt, thanks, i'll look at it tonight, was gone last week, so i just pushed v4.2.x/v4.1.x without local testing.. Oct 05 20:30:28 I will admit it's with an odd video= line Oct 05 20:30:36 Jackix_, serial log -> pastebin.com Oct 05 20:30:45 zmatt, audio? Oct 05 20:31:01 ti's got a lot of tilcd hacks... Oct 05 20:31:24 definitely no audio in my own dtb Oct 05 20:31:42 Im flashing over SD card. Where would the log be? Oct 05 20:32:06 Jackix_, live log on both the hdmi port and over the serial debug port... Oct 05 20:32:22 rcn-ee I don't even have iwconfig. Grrrrr Oct 05 20:32:32 note that the kernel difference showed even when using the same custom dtb with both... ti kernel will start yelling at me for not having the right declarations for emif sram stuff, but it worked Oct 05 20:32:35 uudruid74, "sudo "... Oct 05 20:32:46 I'm logged in as root Oct 05 20:33:21 If the commands exist, where, cause its not in sbin Oct 05 20:34:59 uudruid74, then you have the really small console.. it's a small base for everyone to build off of... not everyone uses wifi... Oct 05 20:35:54 Jackix_, if you want stable, just grab the 2015-07-28 from here: http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#2015-07-28 Oct 05 20:36:11 4 months of fixes from the 2015-03-01 release.. Oct 05 20:36:32 Network support should be top priority, otherwise I can't GET the shit installed. This is a PITA Oct 05 20:37:37 rcn-ee, Thank you very much for your time i am going to use that. Oct 05 20:38:14 uudruid74, nak... the "console" was spec'd: small as possible to just "flash" the eMMC.... Oct 05 20:40:01 rcn-ee: hmm, you have a BBBlfs-like ums thingy you can boot from SD? so people can mount their eMMC ? Oct 05 20:40:48 since then you can just qemu-user + chroot into your eMMC and use your host's networking to install packages Oct 05 20:41:46 zmatt, the console here would do that: http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#Jessie_Snapshot_console Oct 05 20:43:01 there's always the netinstall too: https://github.com/RobertCNelson/netinstall Just boots the board into debian's debian-installer.. Oct 05 20:43:15 rcn-ee: does that include wireless stuff? Oct 05 20:43:37 with the "--firmware" flag... then it adds all the wifi firmware.. Oct 05 20:43:57 the ums thing seems more convenient to me since you can also do it at a later time if for whatever reason you broke networking on your bbb Oct 05 20:44:19 too bad I consider BBBlfs too unstable to recommend to people in general Oct 05 20:44:43 (though the usb_flasher isn't quite as bad as the scripts on top of it) Oct 05 20:45:37 we just need a button... "mount eMMC over usb-slave". ;) Oct 05 20:47:44 u-boot has it but it's pig slow Oct 05 20:48:46 With later images, on failure, it should "end" in that routine.... but it's slow and doesn't always work.. Oct 05 20:49:11 (still faster then the 45min Angstrom 2GB flashing routine. ;) ) Oct 05 20:50:15 I actually pondered something MLO-sized to serve eMMC over Ethernet in some form... should be relatively easy to implement efficiently and at least you can shove an MLO down an BBB's throat in a ton of ways (my fav being via ethernet) Oct 05 20:50:17 45 min for 2gb ???? Oct 05 20:50:26 it's the multi-multi-stage stuff that makes everything a headache Oct 05 20:51:28 rob_w, yeap, the old "2013-06-20" Angstrom flasher ( http://beagleboard.org/latest-images ) took 45mins.. Oct 05 20:52:03 hmm odd Oct 05 20:52:23 but taht was not copying 2gb to a sdcard? Oct 05 20:53:25 rob_w, it spends most of the time on this call: https://github.com/beagleboard/meta-beagleboard/blob/master/contrib/bone-flash-tool/emmc.sh#L57 Oct 05 20:54:08 rsync which we use by default in debian, is just faster... playing with bmap-tools, that's even faster then rsync... Oct 05 20:54:54 rcn-ee: btw, lcdc, working vs non-working: http://pastebin.com/z3mqaCMN Oct 05 20:58:13 (I just added the 'e' to the video mode since I'm at home now so have no actual cable to connect) Oct 05 20:58:15 dco clock: 846.545 MHz vs dco clock: 423.273 MHz Oct 05 20:58:23 looks obviously different to me Oct 05 20:58:24 zmatt, might be... http://git.ti.com/gitweb/?p=ti-linux-kernel/ti-linux-kernel.git;a=commit;h=ed26841aa9112816e3e6c213625bff983524ff69 Oct 05 20:59:02 rob_w: both generate the same pll output clock ultimately though Oct 05 20:59:26 all patches after 2015-06-24, isn't mainline... yuck.. http://git.ti.com/gitweb/?p=ti-linux-kernel/ti-linux-kernel.git;a=history;f=drivers/gpu/drm/tilcdc;hb=refs/heads/ti-linux-4.1.y Oct 05 21:01:00 shoudl be able to shoe horn them in mainline: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/log/drivers/gpu/drm/tilcdc ..... Oct 05 21:01:21 rcn-ee: the config done by -bone is bizarre in this case though.... wrong pixel clock for the requested mode, pixel clock way out of spec, pixel clock divider inherently invalid Oct 05 21:02:13 divider=1 is only allowed for the bus interface mode (lidd/rfb/whatever) Oct 05 21:02:40 that's where darren's patch: drm/tilcdc: rewrite pixel clock calculation seems to fit, as it's way out there from your pastebin... Oct 05 21:03:04 a complete rewrite would at least explain the noticable differences ;) Oct 05 21:03:28 data and OE change on rising edge of pixel clock Oct 05 21:03:34 (that's still the wrong clock edge btw) Oct 05 21:04:20 amazingly the hdmi framer seems able to manufacture a mostly stable image out of either, although fixing that setting does seem to result in a more stable image Oct 05 21:05:55 I wonder if that's the purpose of those (otherwise quite inexplicable) capacitors on the lcd_data lines... create just enough delay for things to work with the wrong polarity setting >< Oct 05 21:07:08 well, till rmk fixed the "tda999x" driver, (for an armanda device), a lot of things were inverted.. ;) Oct 05 21:07:47 hmm Oct 05 21:07:59 that thing *is* quite configurable Oct 05 21:08:33 and hidden behind nda's.. Oct 05 21:08:43 oh I got most of the register map Oct 05 21:08:50 (although i think most of it is now out..) Oct 05 21:11:06 On your topic, image from BBB was unusable on Sharp TV, but fine on Samsung monitor and other TVs. That was using Ubuntu 14, not sure what kernel. On another topic, is there a Plesk server for Beagle? If not, recommended replacements? Oct 05 21:11:09 I know they're using the tda to remanufacture one of the syncs, since the lcdc isn't actually capable of producing standards-compliant sync signals Oct 05 21:12:24 uudruid74: missampling would happen by the HDMI framer though Oct 05 21:13:21 I dunno, it kinda blinked until the TV gave up, but the TV actually froze and had to be rebooted Oct 05 21:13:22 so it should display equally fine or crappy on any screen. compatibility-issues are more likely due to particular timings Oct 05 21:13:35 or other trouble in the HDMI data stream Oct 05 21:13:59 timing related somewhere, I would think Oct 05 21:14:11 you can try forcing a timing mode Oct 05 21:15:10 I saw how to force resolutions, and none seemed to work right. Oct 05 21:15:26 I ended up with the weirdass 1440x1080R@50 because the monitor resizes any image to fit, and resizing 1024 lines to 1080 lines looks really really crappy Oct 05 21:16:44 it also refused low refresh rate, hence 1440x1080 (4:3) with reduced blanking Oct 05 21:17:46 Yeah, something like that. I'm trying to just find a way to stream to a Chromecast instead. Oct 05 21:18:51 the wrong output clock edge thing manifested itself to me by shifting the rightmost column to the first column of the next row Oct 05 21:19:03 and the last pixel of the image ended up at the topleft of the next frame Oct 05 21:21:36 Hmmm ... here is a Plesk for Arm7-X31, but for a QNAP Has. ' Oct 05 21:21:45 Nas Oct 05 21:21:59 rcn-ee: I saw in the register map btw that the tda has a built-in test image generator feature XD Oct 05 21:21:59 Think it could have a chance of running? Oct 05 21:24:41 rcn-ee: the old calculation looks pretty iffy indeed Oct 05 21:26:05 it asked for a rate that should result in a clkdiv of 2, but then *calculated* the clkdiv by dividing the actual pll clock output by the desired rate Oct 05 21:26:18 if the pll rate gets rounded down even slightly... voila, clkdiv=1 Oct 05 21:26:30 -.- Oct 05 21:26:50 should be a +0.5 somewhere to help that :p Oct 05 21:27:07 since there isn't a true round function Oct 05 21:27:24 thats the hack I normally use Oct 05 21:27:43 no, it should have hardcoded the 2, which is exactly what the new code does Oct 05 21:27:52 ah well there is that :p Oct 05 21:28:11 commit, vote, uprate ;D Oct 05 21:29:37 in theory it could use both dividers to generate requested rates more precisely though Oct 05 21:30:11 (always fun when you need to perform prime factorization on your intended clock ratio to figure out whether you can make it with the dividers you have) Oct 05 21:30:27 *dividers and pll Oct 05 21:31:59 (though given that the pll has a multiplier with 18 fractional bits it's really a non-issue) Oct 05 21:34:18 hmm, higher clkdiv in lcdc might also be useful in more data-hungry modes, e.g. 24-bit color, since the undivided clock is used for the module's internal functions like dma Oct 05 21:35:32 anyhow, I'm rambling Oct 05 21:35:53 no, really .. lol .. its interesting though. You grapple with many hardware issues :P Oct 05 21:36:45 time to take care of some stuff though .. bbl :) Oct 05 21:37:09 also, if anyone cares: the latency of delivering a gpio-triggered irq via uio to userspace on the latest ti-rt kernel is usually aroudn 40 microseconds Oct 05 21:37:20 though ranges up to 80 Oct 05 21:37:55 for a task with real-time scheduling that is... Oct 05 21:39:32 Ugh ... I must be crazy. I just downloaded Plesk for Qnap NAS since its Arm7/Linux based and I'm hoping I can get into their package files and glue it to Debian/Bone. Anyone else have a better solution for making Plesk media server run on BBB before I meet Alice in Wonderland? Oct 05 21:40:32 uudruid74: you're a funny guy... complaining about systemd being heavyweight, and then wanting to install *plesk* Oct 05 21:41:58 also, media server? Oct 05 21:42:03 Well ... I already asked about other solutions to stream to a Chromecast. Got any ideas? Oct 05 21:42:32 google corrects that to "plex media server" Oct 05 21:43:06 which I'm not familiar with, but at least makes a lot more sense than plesk Oct 05 21:43:36 yeah ...that's it, sorry. Can't spell today. Brain is toast. Migraine day Oct 05 21:44:30 If it has to trancode I'm probably toast, but I can always transcode the stuff manually on a PC using Handbrake Oct 05 21:45:15 And yeah Plex media server, not evil Plesk control panel. I don't need control panels Oct 05 21:46:46 I don't really have much advice to offer here I'm afraid Oct 05 21:47:46 other than: use Ethernet, it's actually good :P if you need wireless, consider a cheapass AP in client mode rather than using a wifi usb stick or such Oct 05 21:48:36 (really is a shame they put a 100Mbit phy on it though, cpu itself supports dual gigabit) Oct 05 21:48:43 but well, costs... Oct 05 21:57:05 <[Butch]> Sorry if I'm late to the party, but has anybody been having problems with the BBB USB gadget networking not working properly after upgrading to El Capitan? Oct 05 21:57:26 <[Butch]> I've tried re-installing the drivers, but that doesn't seem to help. Oct 05 21:57:54 I've heard more complaints Oct 05 21:58:04 <[Butch]> I have no route to 192.168.7.2 and the BBB doesn't show up in the Network Preferences Pane. Oct 05 22:00:23 <[Butch]> zmatt: more complanints than what? About what? Oct 05 22:00:29 I think linux also supports usb gadget networking as EEM Oct 05 22:00:49 I've heard other people complain about the rndis driver thing not working currently Oct 05 22:01:01 <[Butch]> zmatt: Ahhhhh, thank you. Oct 05 22:01:40 rcn-ee: is there some straightforward way to configure usbnet to use EEM instead of rndis ? Oct 05 22:01:47 * zmatt never really looked at that stuff Oct 05 22:04:02 zmatt, you need to disable: CONFIG_USB_ETH_RNDIS (we have =y) with CONFIG_USB_ETH_EEM (which is disabled)... windows loves to hate EEM, hence we disable it.. Oct 05 22:04:50 situation's the other way around on mac though Oct 05 22:09:34 zmatt. Chromecast is wireless, but I suppose ethernet eliminates USB. The problem is that most of the software wants to autodetect the chromecast, so it has to run on the same network rather than having to hop. I might be able to play some tricks on the Cisco to get a broadcast to span networks but it would be a PITA Oct 05 22:10:06 uudruid74: hmm? normally a wireless AP is a bridge, not a router Oct 05 22:13:02 Usually. The wireless AP has its own port on the Cisco and each Cisco port is another network segment. Hmmm ... wait. I have wired ports on the wireless AP I could use. Hmm ... that should work as those are bridged! Thanks for pointing out my mental bias :-) Oct 05 22:13:39 I've only once worked with a cisco device Oct 05 22:14:32 it did not take long for me to hate the thing with a passion :P Oct 05 22:15:05 (a managed switch) Oct 05 22:16:22 I don't blame you, but I got a special deal (free) on a very nice ASA security appliance. And I can have very large firewall rulesets in it ... like blocking all TOR servers :-) Oct 05 22:20:35 the beaglebone actually has a managed switch embedded too... except it only connects the single external port with the internal host, since the second external port isn't available... still, you can make vlans if you wish ;) Oct 05 22:32:41 zmatt: can the HW handle dot1q tagging directly? Oct 05 22:33:43 yup Oct 05 22:33:57 Stupid question Oct 05 22:34:37 I screwed up fstab and it wont boot. How do I get in to fix it? Oct 05 22:35:06 I'd use console cable to put it into ums mode in u-boot Oct 05 22:35:25 (u-boot's ums is slow, but fine for fixing config files) Oct 05 22:35:39 no cable Oct 05 22:35:46 option2? Oct 05 22:35:47 yeah saw that one coming ;) Oct 05 22:36:11 boot some image from sd, ssh into it, fix emmc from there? Oct 05 22:36:37 or use BBBlfs Oct 05 22:36:56 its "usb_flasher" tool also puts the BBB into ums mode Oct 05 22:37:23 Hello, Got an issue here - I flashed the newest debian form the BB webpage and can neither open http://192.168.7.2/ nor run standalone with separate display 9Only a penguin logo appears) - does anybody know what's happening? Oct 05 22:37:35 I was afraid you'd say that. And make sure the image isn't a Flasher. But boot process is the same as flashing right? Not heard of BBBlfs. It won't flash anything??? Oct 05 22:37:53 uudruid74: it includes scripts to flash images.... those scripts are awful Oct 05 22:38:09 but the "usb_flasher" program it builds, contrary to its name, won't flash anything by itself Oct 05 22:38:17 it just makes the BBB appear as mass storage device Oct 05 22:38:36 that is, if it works... it's quite brittle (but when it doesn't work, it just doesn't work, it doesn't break anything) Oct 05 22:39:08 Hmm ... other option sounds more straightforward. Will BBB recignize a 64GB SD? or does it have a 32GB limit? Oct 05 22:39:19 I actually recommend trying BBBlfs Oct 05 22:39:47 it's just a matter of: git clone, compile the usb_flasher Oct 05 22:39:57 then power up the BBB with sd-button pressed but no SD card Oct 05 22:40:22 one of the boot modes it then tries is usb Oct 05 22:41:03 OK. And then? Oct 05 22:41:27 Give it another image to boot? Oct 05 22:41:28 it appears as network interface (rndis), you then just ./usb_flasher and wait a bit, and it should disappear as network device and reappear as mass storage Oct 05 22:41:37 being your eMMC Oct 05 22:41:44 which you can then mount and inspect/fix Oct 05 22:42:05 Hmm ... that would be interesting. Will give it a shot. Thanks Oct 05 22:42:07 (I've even successfully chrooted into it with the help of qemu-user-static) Oct 05 22:43:28 usb_flasher works 100% of the time on my laptop, but 0% on some other machine... I suspect it's actually *necessary* that something like Network Manager tries to bring the ethernet interface up Oct 05 22:43:33 nice. Make make sticking Funtoo on it easier if I ever get adventurous Oct 05 22:43:43 i.e. usb_flasher implicitly relies on the initialization already being done Oct 05 22:44:18 in general, usb_flasher is very brittle... it handles the boot process like this: Oct 05 22:44:21 And I added nobootwait to the fstab line. Is that Ubuntu only? Oct 05 22:44:38 receive a packet, discard it, send a BOOTP reply Oct 05 22:44:47 receive a packet, discard it, send an ARP reply Oct 05 22:44:52 receive a packet, discard it, send a TFTP reply Oct 05 22:44:53 etc Oct 05 22:45:25 it just blindly follows the pattern without checking whether those packets are actually BOOTP/ARP/TFTP requests Oct 05 22:45:29 -.- Oct 05 22:45:44 but hey, it works Oct 05 22:46:07 and no idea about the nobootwait... check kernel docs? Oct 05 22:49:58 I will repeat my problem, I have no idea what's going on - I flashed the newest debian form the BB webpage and can neither open http://192.168.7.2/ nor run standalone with separate display 9Only a penguin logo appears) - does anybody know what's happening? Oct 05 22:50:07 ok Oct 05 22:50:23 * zmatt is out of steam for support for today Oct 05 22:50:41 jalooc can you ssh to that address? Oct 05 22:52:04 I will try that Oct 05 22:52:32 I have windows on my PC by the way Oct 05 22:53:36 Ewww. Format C: and install Linux :-) Otherwise, use putty to ssh. If it fails, check Windows drivers Oct 05 22:54:47 :p Oct 05 22:55:28 the windows drivers were fine two hours ago, when I tried an image from before ~3 years Oct 05 22:56:46 Ah ... which image was it? Oct 05 22:59:12 honestly I don't remember, because two hours ago I digged the BBB from the pile of my old electronics from 3 years ago :) i run it and it worked fine with the 192... address. Then I decided to update the image and now I'm stuck. Oct 05 23:00:07 When installing drivers I had a problem with signed certificates, bu I found on the forum some other version signed by Sysco or something like that and that went 100% correct Oct 05 23:01:12 After that I run BBB with the ancient image - it worked, after that I updated the image and flashing doesn't work any more. Oct 05 23:03:04 I tried both the newest image from the official BBB website, including updating the /boot/uEnv.txt file AND the way provided by Dereck Molloy with the image he linked from 03.2015 Oct 05 23:04:00 what filename? Oct 05 23:05:05 I used the 8.2 Jessie snapshot, console only. Its been a good image. The 'official' one was toast and needs to be removed Oct 05 23:05:59 ohh I see - can you link the Jessie one? Oct 05 23:06:24 yeah, hold on Oct 05 23:08:42 just checked - cannot connect with putty through ssh (port 22) Oct 05 23:08:44 Page http://elinux.org/Beagleboard:BeagleBoneBlack_Debian The exact image I used was 9/27 but they have 10/04 up now instead. Oct 05 23:09:23 I just downloaded Sun! They are insanely fast Oct 05 23:09:24 uudruid74, those get updated every monday.. :) the old ones are in the file server, the changes are rolling, not much happened last week... Oct 05 23:10:14 Well, still best to try those over the crap corrupt one on beagleboard.org That is a waste Oct 05 23:10:15 oh, and after the "knightrider" pattern during flashing, all diodes start to flash approx. 1s cycles, simultaneously, which I think means the flashing is complete, although it's contrary to the instructions, that indicate it should go completely off or stay steadily on Oct 05 23:10:40 jalooc, "simultaneously" = failure... Oct 05 23:11:07 sorry, non-english speaker, trying to write fast difficult words :) Oct 05 23:11:09 jalooc, you don't have a 2GB bbb do you ? and trying to flash a 4GB ones? Oct 05 23:11:38 that's a good question Oct 05 23:11:55 * heading home, will be back in 10 mins. ;) Oct 05 23:13:38 no, it's 4GB Oct 05 23:16:51 uudruid74 actually the image I have currently is the newest one from the page you linke.. :( : BBB Rev C (4GB eMMC) 4GB eMMC Flasher Oct 05 23:17:15 date? Oct 05 23:19:35 2015-03-01 Oct 05 23:20:07 ups sorry Oct 05 23:20:09 my mistake Oct 05 23:20:30 that's an oldie Oct 05 23:21:26 I'll try the newest one tomorrow, it's already late in my time-zone. Thanks for the help uudruid74 Oct 05 23:21:35 I think that's the broken one Oct 05 23:40:53 wb Oct 05 23:41:04 ah d'oh pingy out Oct 05 23:43:48 wb? Oct 05 23:43:58 2nd time lucky :p Oct 05 23:45:38 is using the kernel upgrade script combined with apt-get upgrade as good as flashing new images as they come out, or is it not as comprehensive Oct 05 23:45:57 EvilTelephone, it's about 95%... Oct 05 23:46:21 close enough to 100% for me! :D Oct 05 23:46:22 ty Oct 05 23:46:39 with u-boot: v2015.10-rc4 there's a nice feature for guarenteeing the eMMC get's passed to the kernel, instead of our "guess"... Oct 05 23:46:54 i don't force upgrades of u-boot, only newer releases get it.. Oct 05 23:48:07 if i generally don't have a uSD card in, would i generally be safe from a bad "guess"? Oct 05 23:49:08 That's the odd state... if the uuid value isn't present, it'll guess /dev/mmcblk"1"... when you need "/dev/mmcblk"0""... Oct 05 23:49:30 ah, gotcha Oct 05 23:49:40 how does the uuid get set reliably? Oct 05 23:49:45 is it embedded in the image now? Oct 05 23:49:48 the change in rc4, we can use the "partuuid" which can passed to the kernel intead of "/dev/mmclk"x"" Oct 05 23:49:53 ah Oct 05 23:50:32 except i noticed a fun problem... partuuid doesn't work with initrd's... (but can be decoded by the kernle) werease uuid's can't be decoded by the kernel (initrd does it)... Oct 05 23:51:32 maybe vagrantc knows... do you know where the uuid decode works in the initrd? i'd love it if "partuuid" could be used... Oct 06 00:08:49 /usr/share/initramfs-tools/scripts/functions contains resolve_device() Oct 06 00:08:54 which is used to parse root= Oct 06 00:09:22 root=UUID=yada would use /dev/disk/by-uuid/yada Oct 06 00:09:34 zmatt, yeap, cloned: https://anonscm.debian.org/cgit/kernel/initramfs-tools.git/tree/scripts/functions#n294 Oct 06 00:10:09 need to shoehorn PARTTUIID=<>, then we can ignore the uuid value and u-boot will always pick correct device! Oct 06 00:11:27 but what's the problem with UUID again? Oct 06 00:12:03 it's works great, till you don't have an initramfs... whereas partuuid doesn't need an initramfs... Oct 06 00:12:11 and partuuid crashes when using a initramfs... Oct 06 00:12:27 i'd like to use "one" ;) Oct 06 00:12:34 why would you not have an initramfs? o.O Oct 06 00:12:46 some users. .;) Oct 06 00:13:14 and u-boot now pass's the root=PARTUUID= by default... Oct 06 00:13:23 ugh Oct 06 00:13:34 i could just hack that out, but i try to merge into upstream's changes... Oct 06 00:15:09 hmm, the current initramfs script does look pretty gross... thinking about it, passing execution directly from kernel to systemd doesn't sound like such a bad idea Oct 06 00:15:37 yeah i read about that, not sure how to tell systemd use this root.. Oct 06 00:15:58 well without initramfs isn't the root already there? Oct 06 00:16:42 it's always there.. but without the uuid and depending on the insert of the microSD, eMMC = mmcblk0 (or mmcblk1) Oct 06 00:16:48 Hi everyone. My alleged beaglebone black rev C is booting showing only 1CPU in use. lscpu shows it is a single core A7 . Did I get jacked on Amazon or am I missing something? Oct 06 00:17:19 rcn-ee: except you just said u-boot now passes PARTUUID and the kernel understands that Oct 06 00:17:25 briarrose, there's another core? since when!!! Oct 06 00:17:47 zmatt, correct, so that fixes that problem.. however if you use a uuid/initramfs it's broken... Oct 06 00:18:01 I think the fs and mmc drivers are already in the kernel right? why not just ditch initramfs? Oct 06 00:18:06 what's it adding? Oct 06 00:18:13 (other than boot time) Oct 06 00:18:28 zmatt, here's my deal with partuuid/initrd/uuid: https://github.com/RobertCNelson/Bootloader-Builder/blob/master/patches/v2015.10-rc4/0001-omap5_common-uEnv.txt-bootz-n-fixes.patch#L125-L146 Oct 06 00:19:07 also, it sounds really weird the kernel doesn't know about UUIDs... devtmpfs clearly does Oct 06 00:19:11 if we can pass partuuid and not having the intrd crash, i can drop all the uuid crap.. Oct 06 00:19:49 it's silly, initrd only knows label/uuid.. kernel only knows label/partuuid... Oct 06 00:20:15 So the Cortex A8 shows as ARMv7 rev 2 (v71) rcn-ee? Oct 06 00:20:15 how do you locate a device by partuuid? Oct 06 00:20:48 briarrose, "rev 2" = "revision 2" Oct 06 00:21:00 zmatt, /dev/sda1: UUID="a67deb90-f943-4bc2-adfa-dbb2d888d3f0" TYPE="ext4" PARTUUID="f46f4fca-01" Oct 06 00:21:25 rcn-ee: yes, but I mean, I have /dev/disk/by-uuid Oct 06 00:21:30 not /dev/disk/by-partuuid Oct 06 00:21:32 there isn't.. Oct 06 00:22:08 ok, thx rcn-ee ... just making sure it's a good board. I guess I expected to see ARMv8 in there. My bad. Oct 06 00:22:44 zmatt, err wait.. https://wiki.archlinux.org/index.php/Persistent_block_device_naming#by-partuuid Oct 06 00:22:48 maybe i missed a config.. Oct 06 00:23:37 briarrose, sorry, armv8 was only a dream when bbb came out.. Oct 06 00:24:02 cortex-a8 is actually the first ARMv7 cpu :P Oct 06 00:24:15 but revision 2!! so thumb2 works.... Oct 06 00:24:25 the r2 makes no sense to me at all Oct 06 00:24:45 rcn-ee No worries. I'm a newb to all this so I'll skulk back over to the corner until I derp again. :P Oct 06 00:24:48 there's no ARMv7 r2 Oct 06 00:25:03 and if there were, the Cortex-A8 wouldn't be it, since it was the first ARMv7 cpu :P Oct 06 00:25:06 think it's actually r3p2, but those other number are random in kernel.. Oct 06 00:25:08 yep Oct 06 00:25:18 the cpu rev is r3p2 indeed Oct 06 00:27:54 rcn-ee: but wait, u-boot *knows* whether an initramfs is being used right? Oct 06 00:28:01 so it could pass the right argument Oct 06 00:28:36 correct, i have a bit if/else/fi to take care of that.. ;) i'd just like to clean it up.. Oct 06 00:29:26 I'd also lean towards deliberately making them partuuid the same as the uuid, but I guess we can't force that onto users Oct 06 00:30:59 also, I should probably switch to gpt so I can actually have partuuids to begin with and participate more usefully in finding the cleanest.. ehm... least hideous way out of this Oct 06 00:31:23 yet remain backwords with angstrom's 2013.06.20 u-boot ;) Oct 06 00:31:53 why on earth do you still bother? Oct 06 00:32:36 because when i don't my email fills up quickly... i've tried to kill that twice over the last two years, but users come out of the woodwork.. Oct 06 00:32:59 but WHY would anyone keep that crap? Oct 06 00:33:07 cuz it is better Oct 06 00:33:28 ds2 wants to stay in 2013 Oct 06 00:33:36 it was a good time Oct 06 00:34:04 maybe I should port my Forth and turn it into a half-baked Open Firmware implementation ;) Oct 06 00:34:11 yet the memory address a fubared for v3.8.x+ ;) Oct 06 00:34:51 if you are going to switch, try a smaller bootloader Oct 06 00:35:07 say APEX or something that does not require 2stages Oct 06 00:35:10 I think my Forth is smaller than u-boot's SPL :P Oct 06 00:35:12 ds2's u-boot's falcon mode! Oct 06 00:35:18 spl -> kernel Oct 06 00:35:30 sure Oct 06 00:35:40 it's small and hard-coded... Oct 06 00:35:44 iirc it was 53 KB or something last time I measured.... but that may have been including the code I loaded onto the DSP Oct 06 00:35:47 ideally, it should also be APL or BSD Oct 06 00:35:49 not GPL Oct 06 00:36:46 arm port of lilo? Oct 06 00:37:01 blah Oct 06 00:37:07 it's bsd... Oct 06 00:37:24 LILO has become a PoS wit v2 Oct 06 00:38:06 other then an efi bootloader it's the only bsd on: https://en.wikipedia.org/wiki/Comparison_of_boot_loaders Oct 06 00:38:19 FASTBOOT Oct 06 00:38:19 really all we need is something that loads a fixed range of sectors from the same mmc device it was loaded from itself (info passed by bootloader)... it can skip initialization, since it was just loaded itself from it by bootROM Oct 06 00:38:31 the rest is fluff Oct 06 00:38:55 ds2, think u-boot swallowed fastboot... Oct 06 00:39:05 hmmm? Oct 06 00:39:11 there is still a standalone fastboot Oct 06 00:39:41 yeah, fastboot's being merged into u-boot... just wait till systemd takes over u-boot.. Oct 06 00:40:16 hard time seeing them allow fastboot to be GPL'ed Oct 06 00:42:04 ahh.. u-boot's just implementing the usb fastboot protocol... Oct 06 00:42:12 :) Oct 06 00:42:24 the CPU bus is rarely exposed thesedays Oct 06 00:42:40 modems used to be serial interfaced, now it is ethernet, usb, etc Oct 06 00:42:54 blah Oct 06 00:43:35 or a chip2chip interface for a stacked die ;) Oct 06 00:43:49 * zmatt still has the omap4 TRM which describes it, before it was purged from public docs Oct 06 00:44:18 :) Oct 06 00:44:39 I also suspect the "expansion port" occasionally seen in dm81xx / am335x docs is something similar for the same purpose Oct 06 00:46:12 the omap4 interface was external though, you could apparently connect two of them in "mirror mode" Oct 06 00:46:44 (intended to test the interface, but it sounds like a cool hw configuration) Oct 06 00:57:11 Well, finally got the beast fixed. It seems pretty stable, network issues fixed, etc. Now on to config samba and dlna and maybe plex (long shot) Oct 06 01:27:16 rcn-ee: btw, i2c1_pin_mux should be omitted from pinmux in u-boot for the beaglebone (all variants) Oct 06 01:27:29 rcn-ee: not only is it unnecessary, it's even the wrong pins Oct 06 01:29:45 zmatt, wonder if they were thinking of reading the cape bus... Oct 06 01:30:15 * zmatt shrugs Oct 06 01:33:18 no... not that either... Oct 06 01:33:48 face balm.. what the heck... Oct 06 01:33:56 like I said, wrong pins Oct 06 01:35:03 i actually have the same pins/i2c1 set in another overlay... but yeah, no reason for that in u-boot.. Oct 06 01:36:16 ah copy/paste, hardware on the am335x-evm.dts use it... Oct 06 01:36:42 i'll submit a patch to Tartarus to nuke that reference for the beaglebone.. Oct 06 01:38:18 zmatt, btw, the partuuid came from a fedora arm user, where dracut (fedora's initramfs) decodes partuuid.. so just ubuntu/debian is broken.. fedora works.. Oct 06 01:48:42 odd, i2c1's been enabled by default since 2012... Oct 06 01:49:45 I've known for a while, I just kept forgetting about it Oct 06 01:50:34 it's driven someone on E2E crazy because his i2c1 bus stubbornly refused to work, kernel driver reporting the bus "busy" Oct 06 01:50:49 (apparently if an input is muxed to multiple pins, they are ANDed together) Oct 06 01:52:00 the bbb has 3d acceleration it seems? i wonder if it's decent Oct 06 01:52:44 EvilTelephone, it's too lmited to be useful... Oct 06 01:53:11 rcn-ee: it seems to do a pretty decent job for GLES2 Oct 06 01:53:34 the MagicLantern demo is not trivial, and it ran smoothly for me at very decent resolution Oct 06 01:53:50 does the die get super hot? Oct 06 01:53:59 yeah, if you know it's limitations, it works... everytime some asks above ^ they want to run standard "opengl"... Oct 06 01:54:09 http://gerbil.xs4all.nl/bbb-sgx-magiclantern.png Oct 06 01:54:36 rcn-ee: i was thinking something like quake3 level of graphics.. nothing super modern Oct 06 01:55:08 EvilTelephone: it only supports GLES / GLES2 Oct 06 01:55:16 Well... if it's the "opengl es2" port of quake3... Oct 06 01:55:50 this might work https://github.com/DingoOz/Quake3-GLES-for-armv7 Oct 06 01:57:42 quake on a beagle .. awesome .. lol Oct 06 01:57:59 actually .. a port to the Pi2 would be pretty good .. Oct 06 01:57:59 lol i want to try it Oct 06 01:58:08 i mean, why the heck not Oct 06 01:58:21 quad core with the gpu .. that'd work .. Oct 06 02:06:38 ah Finally Oct 06 02:44:06 rcn-ee: ah, so with a GPT you actually do get /dev/disk/by-partuuid ... then fixing initramfs is easy indeed **** ENDING LOGGING AT Tue Oct 06 02:59:58 2015