**** BEGIN LOGGING AT Wed Jun 24 02:59:58 2015 Jun 24 03:49:27 i am new to beaglebone..since some days i am trying to connect my beaglebone rev a6 with my laptop (win 8.1) Jun 24 03:49:54 but not able toconnect it with 192.168.7.2 Jun 24 05:26:25 hi all. has anyone of you made experiences in using PTP and the BBB? Jun 24 05:55:17 precision time protocol? Jun 24 05:56:08 the photo thing Jun 24 05:56:16 usb Jun 24 05:56:23 maybe Jun 24 05:56:29 People Terrorizing PAsta Jun 24 05:56:44 * LetoThe2nd loves TLAs Jun 24 05:57:03 People Toasting Pi Jun 24 05:57:09 people trolling people Jun 24 06:25:00 hi there Jun 24 06:39:17 hello sir Jun 24 06:41:11 are you there?? i have some doubt regarding Beagleboard-xM Jun 24 06:45:24 Arvind_: ask Jun 24 06:45:27 dont ask to ask Jun 24 06:48:16 is this board support windows 7 Jun 24 06:52:15 Arvind_: windows 7 running on the bpard? Jun 24 06:52:17 board Jun 24 06:52:18 no Jun 24 06:52:59 you can connect it to a windows 7 system though. If it runs the right software Jun 24 06:58:29 sir for this which driver it requires Jun 24 07:00:42 drivers? Jun 24 07:00:46 you dont need drivers Jun 24 07:01:05 if you connect the BBB to ethernet, you can talk to it over the network Jun 24 07:04:25 gee, he could have just read the bb.org pages... Jun 24 07:18:55 tbr sorry - was afk - yes precision time Jun 24 07:19:07 I need to sync the time of my bbb with the time of my laptop Jun 24 07:19:09 :/ Jun 24 07:19:19 on the bbb is debian installed and on my laptop ubuntu Jun 24 07:21:53 sla89: You can either use NTP (install ntpd on laptop) or tlsdate. Tlsdate will set time by issuing a https request towards your laptop (requires a webserver with SSL enabled) Jun 24 07:22:19 Chrille how is the precision of tlsdate Jun 24 07:22:24 and ntp? Jun 24 07:22:42 sla89: tlsdate is 1 second, ntp is fraction of a second Jun 24 07:23:01 ntp is the normal way to do things Jun 24 07:23:22 if you want to use ptp, you'd better have good reasons Jun 24 07:38:11 Chrille okay Jun 24 07:38:24 Chrille and when I need precision < 1m? Jun 24 07:38:26 Chrille and when I need precision < 1ms? Jun 24 07:39:35 i have NTP now running with an offset of 80 ms Jun 24 07:39:35 sla89: Try reading http://www.ntp.org/ntpfaq/NTP-s-sw-clocks-quality.htm about NTP precision Jun 24 07:39:57 You can easily get better than 1 ms with NTP Jun 24 07:41:25 nettime synchro under 1ms why would one need this? Jun 24 07:44:15 maybe he wants to put it into a particle accelerator experiment ;) Jun 24 07:45:39 i need to do some one-way network measurement Jun 24 07:45:52 therefore i need to take time on my laptop and bbb Jun 24 07:45:53 ? Jun 24 07:45:56 and compare those Jun 24 07:46:07 udp? Jun 24 07:46:10 RTT does not say anything in my case Jun 24 07:46:25 udp? Jun 24 07:46:33 i dont have to measure time differences Jun 24 07:46:43 i have to measure the needed time from host a to host b (latency) Jun 24 07:47:02 for my own protocol Jun 24 07:47:33 I cannot add a timestamp to my protocol since this would falsify my measurement Jun 24 07:54:52 sla89 hm okay Jun 24 07:55:23 but network traffic under 1ms even in local lan is not likely to take place Jun 24 07:55:55 woglinde I know that Jun 24 07:57:00 okay Jun 24 07:58:16 wogline maybe I wasn't clear enough in my description. I have to do some network measurements in one direction (host A to host B). Before the packet is sent with the specific data the system time is stored in memory. On system B the system time is taken when the packet received and was processed and also stored in memory. This is going to happen for about 5 minutes. Afterwards I can compare the times and the difference between time A a Jun 24 07:58:18 per packet Jun 24 07:58:54 Everything would be easy, if I can measure the RTT, but in my scenario the RTT does not say anything about the latency. Jun 24 07:59:16 RTT is equal to ping - Packet goes from host a to b and back to a Jun 24 07:59:25 and when it was sent and received by A the time is taken Jun 24 08:02:53 woglinde: ptp acheives <1ms sync Jun 24 08:03:31 woglinde: i think, w/o special hardware it's around a few 10us to 100us Jun 24 08:03:43 woglinde: with real hardware [tm] you can get it down to 100ns Jun 24 08:04:24 sla89: are you trying to do ptp by hand? if so, stop. you will not be able to beat it Jun 24 08:04:41 nope - I am not Jun 24 08:05:15 ah.. sorry, i misread what you wrote Jun 24 08:05:18 Its a students project where the BBB is used to control a small robot via wifi Jun 24 08:05:29 eh.. Jun 24 08:05:33 and one requirement is to measure the latency for the commands sent via wifi Jun 24 08:05:44 so, use a wire Jun 24 08:05:44 no problem :) Jun 24 08:05:46 i hope you run ptp over wire? Jun 24 08:05:59 I am not running PTP at all at the moment ^^ Jun 24 08:06:22 you dont need PTP Jun 24 08:06:34 the idea was to sync the time before the measurements were done via a cross over wire Jun 24 08:06:35 run a GPIO between host and robot Jun 24 08:07:05 set gpio high when you send a packet Jun 24 08:07:15 measure the delta on the other side Jun 24 08:07:17 move on Jun 24 08:07:26 what av500 said Jun 24 08:07:30 far simpler Jun 24 08:07:46 never done this before Jun 24 08:07:47 and with a good oscilloscope you can readily measure delay down to 100ns Jun 24 08:07:56 how can I achieve it in detail? Jun 24 08:07:59 actually, you can do that even with a cheap scope Jun 24 08:08:05 how can i put gpio high with my laptop Jun 24 08:08:13 use the parallel port :) Jun 24 08:08:22 +1 Jun 24 08:08:49 or then NumLock LED Jun 24 08:08:51 the* Jun 24 08:10:07 to verify you can sniff the packets off the air using gnuradio and some sdr Jun 24 08:10:09 i dont have :( Jun 24 08:10:47 you dont have a numlock LED? Jun 24 08:12:00 i have no parallel port on my laptop Jun 24 08:12:06 but i have a numlock led Jun 24 08:12:20 KotH thanks for the hint Jun 24 08:12:45 serial port Jun 24 08:12:46 whatever Jun 24 08:12:54 he dont have that either Jun 24 08:13:05 he has a laptop, not a museum piece Jun 24 08:13:12 then use a machine/board that has :) Jun 24 08:13:42 KotH can I sniff all packets with gnuradio? does it fetch everything around? Jun 24 08:15:03 sla89: ignore the gnuradio for now Jun 24 08:15:07 you dont have that either Jun 24 08:16:03 :( Jun 24 08:16:24 is it possible via USB? Jun 24 08:16:40 sla89: why not measure RTT and half it? Jun 24 08:16:51 because its not the same Jun 24 08:17:08 when the packet is on its way back some jitter can happen or other stuff Jun 24 08:18:02 that can happen in the other direction too Jun 24 08:18:05 no? Jun 24 08:18:28 yeah Jun 24 08:18:41 but assume sending from A to B needs 10 ms Jun 24 08:18:49 back from B to A 7 ms Jun 24 08:18:52 would be 17 MS Jun 24 08:18:53 right Jun 24 08:18:58 it would Jun 24 08:19:04 you cannot half it - its not the same Jun 24 08:19:14 well, then do this: Jun 24 08:19:25 on the PC, make numlock light up when you send the packet Jun 24 08:19:35 on the BBB make some GPIO light up when you recieve it Jun 24 08:19:41 hook both to a scope Jun 24 08:20:19 then you can see the delta Jun 24 08:20:20 use two bbbs, see, you can use two gpios. :-) Jun 24 08:21:40 av500 itsnt it a bit inaccurate? Jun 24 08:21:47 why? Jun 24 08:22:08 LetoThe2nd hehe - that doesn't fit to the setup since the controlling part is a laptop or smartphone - not another bbb ;) Jun 24 08:22:43 on the phone, make a part of the screen go white from black Jun 24 08:22:47 put a photocell there Jun 24 08:22:53 same as on the LED Jun 24 08:23:01 since I would then measure lights coming up and down - how shall I take the time of this? Jun 24 08:23:14 you see the time on the scope Jun 24 08:23:19 the X axis is time Jun 24 08:23:23 even quite calibrated Jun 24 08:23:50 Hi - I'm trying to load ARM's C compiler with NEON stuff but it's too big for the free space on my BBB. Any suggestions? Jun 24 08:24:00 * LetoThe2nd doesn't get the point. you want to make the network, so better get two endpoints that can provide the data you need. 2xbbb + a scope or logic analyzer. logic analyzer can even be running on some laptop or phone and see: results are there. Jun 24 08:25:08 sla89: you do understand that you can show the timing relation between 2 signals on a scope? Jun 24 08:26:20 Peter___: "ARMs C compiler with NEON stuff"? in what form is that meant to be different from the gcc that comes with the stock installaion? Jun 24 08:26:57 Good question, but I found a page that compared the machine code and said ARM's was better Jun 24 08:27:34 Peter___: well the general answer for the lack of space problem is: add more storage. :-) Jun 24 08:27:43 or, in reverse: free up some space. Jun 24 08:28:04 Peter___: but if you're talking about ARMCC, be assured: its crap. Jun 24 08:28:06 I also can't get the kind of speed we need from gcc, even using -O3 and the approprrate flage at compile and link Jun 24 08:28:26 and ARM compiler has unicorns inside? Jun 24 08:28:34 and make your code magically fasdter? Jun 24 08:28:37 Why do you say ARMGCC is crap? Jun 24 08:28:45 Peter___: in about 100 of 100 cases, its not the compiler that makes your code slow, its the code that makes the code slow. Jun 24 08:28:46 its not crap Jun 24 08:28:51 but its also not magical Jun 24 08:29:15 I'm running the same sode on my old PC and on the BBB - the PC is 15 times faster Jun 24 08:29:19 code Jun 24 08:29:20 of course Jun 24 08:29:40 Peter___: read more precisely what i said. i said: "ARMCC is crap". no letter 'G' in there. and its crap because its an ancient compiler frontend on an even more ancient backend. Jun 24 08:29:42 thats not surprising Jun 24 08:29:51 I don't feel gcc is using the BBB's resorces Jun 24 08:29:59 like? Jun 24 08:30:03 what does your code do? Jun 24 08:30:08 av500 nope - didnt got it how I can achieve this. I was thinking about it but I don't get it :( Jun 24 08:30:23 sla89: you see 2 signals on a 2-channel scope Jun 24 08:30:25 sla89: ask your instructor to show you how a scope works :) Jun 24 08:30:28 one goes high at time t_1 Jun 24 08:30:34 the other at time t_2 Jun 24 08:30:38 you trigger on signal 1 Jun 24 08:30:49 then you move the cursor to be on t1 and t2 Jun 24 08:30:55 yeah - I know that Jun 24 08:30:55 it just calculates atan2(y,x) over 100 different values for y and x Jun 24 08:30:57 and the scope tells you the delta Jun 24 08:31:20 ah Jun 24 08:31:28 Peter___: the BBB has no atan in hardware Jun 24 08:31:30 and i switch the LEDs of immediatly Jun 24 08:31:37 it uses the floating point lib that comes with gcc Jun 24 08:31:49 since I will receive/send several packets Jun 24 08:31:53 sure Jun 24 08:32:08 if you do it continously you can overlay all the traces Jun 24 08:32:15 then you even see the min/max latency Jun 24 08:32:30 OK - so we have to rewrite all these functions for NEON? Jun 24 08:32:31 the armcc one might be a feq cycles faster Jun 24 08:32:47 yes Jun 24 08:33:08 http://stackoverflow.com/questions/18187492/calculate-atan2-using-simd-instructions-and-neon-arm-assemlby Jun 24 08:33:14 the compiler wont do that for you Jun 24 08:33:53 av500: hm, shouldn't -gnueabihf be able to use the floating functions automagically to build atan out of them? Jun 24 08:34:09 e.g. the problem is only "constructed atan" versus "hw atan" Jun 24 08:34:31 the only speedup is if you do SIMD Jun 24 08:34:42 otherwise, yes gcc will emit floating point code Jun 24 08:34:53 very much like armcc or on the PC Jun 24 08:35:17 thanks for confirming :) Jun 24 08:35:40 OK, thanks for explaining. We weren't sure what we would get with the Arm version of the compiler. Jun 24 08:36:23 Peter___: not 15x faster Jun 24 08:36:40 is armcc free to use now? Jun 24 08:36:45 nope Jun 24 08:36:49 well, how much faster? Jun 24 08:36:55 simple Jun 24 08:36:58 it .... depends Jun 24 08:37:00 only for <32kB baremetal Jun 24 08:37:04 on everysingle use case Jun 24 08:37:21 do not use atan at all haha Jun 24 08:37:28 but optimizing your code faster will be always better Jun 24 08:37:32 dont use atan Jun 24 08:37:33 well, how much faster would you think Jun 24 08:37:36 use a look up table Jun 24 08:37:38 woglinde: that basically solves it Jun 24 08:37:45 Peter___: I cannot say Jun 24 08:37:45 av500: dang you typed faster. Jun 24 08:37:48 nobody can Jun 24 08:38:02 its the usual tradeoff storage/speed? Jun 24 08:38:04 :-) Jun 24 08:39:02 its one solution Jun 24 08:39:06 look up table Jun 24 08:39:10 approximation Jun 24 08:39:18 changing the math to another "domain" Jun 24 08:39:27 av500 thank you for your help - I think this can do the trick :) Jun 24 08:39:33 it all depends/tm) Jun 24 08:39:35 it all depends(tm) Jun 24 08:39:44 LetoThe2nd thank you too :) Jun 24 08:41:19 sla89: np :) Jun 24 08:44:48 why is he named like a TI data sheet? Jun 24 08:49:29 who? Jun 24 08:49:45 you :) Jun 24 08:50:15 http://electronics.stackexchange.com/questions/14297/ti-manual-naming-conventions Jun 24 08:50:42 omg xD Jun 24 08:50:50 it has nothing to do with TI ;) Jun 24 08:50:59 the letters are part of my real name ^^ Jun 24 08:51:06 ;) Jun 24 08:51:11 and you are 89 years old? Jun 24 08:51:34 born 89? Jun 24 08:51:49 more likely Jun 24 08:52:05 don't discriminate against old people Jun 24 08:52:09 yeah - born 89 Jun 24 08:52:10 hehe Jun 24 08:52:28 equal chances and all that stuff Jun 24 08:53:27 does the BBB throttle the cpu speed if it is plugged to USB since it does not get enough power? Jun 24 08:53:40 especially if also a wifi dongle is used Jun 24 08:55:23 nope Jun 24 08:56:21 why do people then write to use 5V and 2A when running BBB with a wifi dongle? Jun 24 08:56:29 Yes it does as far as I know Jun 24 08:57:12 sla89: because it will just fail and crash and burn Jun 24 08:57:23 sla89: if you don't Jun 24 08:57:26 because it would exceed the power draw from the USB port Jun 24 08:58:18 that's not really managed Jun 24 08:58:38 but yes Jun 24 08:58:52 basically it would cause the voltage to drop and then things to stop working Jun 24 08:59:06 ah alright Jun 24 08:59:08 thanks :) Jun 24 09:00:54 the BBB is also limited to 500 Mhz when running on USB Jun 24 09:01:18 ? Jun 24 09:01:25 Attunga: since when Jun 24 09:01:38 that must be a fairly new feature Jun 24 09:01:39 since I read it ages ago Jun 24 09:02:01 when i was using it few months ago it would run at full speed regardless of the power source Jun 24 09:59:37 av500: yes, with GR you can sniff all packets in a channel. actually, you can sniff multiple networks that lie within the frequency range you capture Jun 24 09:59:56 av500: just makes analysis more difficult :) Jun 24 10:07:54 KotH: I know you can Jun 24 10:08:04 but one does not usually have GR Jun 24 10:09:56 av500: you can get The GR from github, like with all other code Jun 24 10:10:27 you can even clone one rtl-sdr dongle, if you need to Jun 24 12:28:15 Hello Jun 24 12:28:33 need help setting time on beaglebone black Jun 24 12:29:53 timedatectl command doesn't work Jun 24 12:32:46 bbb has no rtc Jun 24 12:32:53 use the date command Jun 24 12:32:57 man date Jun 24 12:33:16 or if you want and have a network connection install a ntpd Jun 24 12:33:28 to get the correct time over the net Jun 24 12:34:03 how can I install ntpd Jun 24 12:34:24 apt-get ? if you have installed debian or ubuntu Jun 24 12:34:41 I have debian Jun 24 12:38:25 woglinde: actually, the SoC has RTC but there is no battery Jun 24 12:39:07 jacekowski okay okay Jun 24 12:39:33 i remember someone here a while back who got the RTC working on the BBB Jun 24 12:43:19 i got ntpd working Jun 24 12:43:31 on my bbb Jun 24 12:43:36 for testing purpose Jun 24 12:44:04 normaly you just need to enter sudo apt-get install ntp Jun 24 12:44:11 in your command line Jun 24 12:44:17 thanks Jun 24 12:44:39 I've also heard that someone figured out a way to solder a battery on to keep the RTC ticking but I have yet to find any proof. Jun 24 12:44:40 and setup the configuration at /etc/ntp.conf Jun 24 12:44:43 most common rtc chips are just i2c devices, almost trivial to configure. but they won't be managed by timedatectl because it relies on /dev/rtc0 which is the internal, non-battery backed-up rtc Jun 24 12:45:38 sla89 did you get error when you did yours Jun 24 12:45:40 search for a ntp server near to your location - you can find the addresses here http://www.pool.ntp.org/ Jun 24 12:45:54 Ibrahim no - not really - what is the error you get? Jun 24 12:45:55 my is coming up with an error Jun 24 12:46:14 dpkg: error processing llvm-3.0-dev (--configure Jun 24 12:47:07 try sudo apt-get update Jun 24 12:47:09 first Jun 24 12:47:11 Errors were encountered while processing: ntp; binfmt-support; llvm-3.0-runtime llvm-runtime; llvm-3.0 llvm; llvm-3.0-dev; Jun 24 12:47:57 is it a blank BBB? fresh installed everything? Jun 24 12:48:24 have you installed llvm? Jun 24 12:49:40 i have using sudo apt-get install llvm clang Jun 24 12:49:51 before? Jun 24 12:50:09 did you get any error while installing llvm and clang? Jun 24 12:50:17 Yes, Jun 24 12:50:30 Hope I have messed up somewhere Jun 24 12:50:39 tried to man thingd Jun 24 12:51:02 do you have installed the build essentials? Jun 24 12:53:14 and why dont you use gcc? Jun 24 12:53:30 (i am only interested why - dont want you to remove it) Jun 24 12:54:24 actually Jun 24 12:55:02 same error as when I am trying to install ntp Jun 24 12:56:04 Is the bone connected to a network? Jun 24 12:56:11 Can the bone talk to the Internet? Jun 24 12:56:21 Yes Jun 24 12:56:31 Have you done 'apt-get update' successfully, without error? Jun 24 12:56:46 I have managed ping, clone files ec Jun 24 12:57:27 its strange llvm bails out at dpkg-configure Jun 24 12:57:44 woglinde: Well, we also are missing the rest of the error. Jun 24 12:57:59 Ibrahim try apt-get remove the llvm stuff for now Jun 24 12:58:45 curretly running apt-get update again Jun 24 12:58:59 someone was interested in low power shutdown on beaglebone? Jun 24 12:59:20 got the following replies Jun 24 12:59:59 W: There is no public key available for the following key IDs: 9D6D8F6BC857C906; W: There is no public key available for the following key IDs: 7638D0442B90D010; W: There is no public key available for the following key IDs: 7638D0442B90D010 Jun 24 13:00:31 Ibrahim: What mirror are you using? Jun 24 13:00:43 you have performed apt-get update? Jun 24 13:01:41 08:58 < Ibrahim> curretly running apt-get update again Jun 24 13:01:42 i did Jun 24 13:01:50 sla89: And Ibrahim has broken GPG. Jun 24 13:02:13 W: There is no public key available for the following key IDs: 9D6D8F6BC857C906; W: There is no public key available for the following key IDs: 7638D0442B90D010; W: There is no public key available for the following key IDs: 7638D0442B90D010 Jun 24 13:02:30 agmlego sorry didnt got that Jun 24 13:03:00 that is what returned after run the update Jun 24 13:03:31 Ibrahim: What mirror are you using? Jun 24 13:03:51 llvm? Jun 24 13:04:15 No, the server you download your packages from./ Jun 24 13:04:27 Configured in /etc/apt/sources.list Jun 24 13:04:42 github Jun 24 13:04:48 lol Jun 24 13:04:54 THat seems very unlikely. Jun 24 13:05:07 Ibrahim did you even understood what you were doing? Jun 24 13:05:07 I am novice Jun 24 13:05:41 ibrahim perform nano /etc/apt/sources.list Jun 24 13:05:46 and tell us what you see Jun 24 13:06:01 if nano is not available try Jun 24 13:06:08 vi /etc/apt/sources.list Jun 24 13:06:17 it will open the file in an editor Jun 24 13:06:21 :) Jun 24 13:07:05 deb http://ftp.us.debian.org/debian/ wheezy main contrib non-free; #deb-src http://ftp.us.debian.org/debian/ wheezy main contrib non-free Jun 24 13:07:57 OK, that should be fine. Jun 24 13:08:06 I found a forum thread about this issue. Jun 24 13:08:11 http://crunchbang.org/forums/viewtopic.php?id=39608 Jun 24 13:08:44 Ibrahim: 'sudo apt-get dist-upgrade' Jun 24 13:10:03 Actually, before you do that... Jun 24 13:10:23 'sudo apt-get install debian-keyring debian-archive-keyring' Jun 24 13:10:30 cool Jun 24 13:10:44 thanks a lot by the way Jun 24 13:10:45 you dont need debian-keyring Jun 24 13:11:05 This is better than other approaches because it does not install "debian-keyring", which is big and 99% of the time unnecessary. Jun 24 13:11:08 http://unix.stackexchange.com/questions/75807/no-public-key-available-on-apt-get-update Jun 24 13:11:10 see last post Jun 24 13:11:33 ... Jun 24 13:12:40 agmlego ...? Jun 24 13:12:48 you dont seem to be happy? Jun 24 13:13:09 debian-keyring is not that large. Jun 24 13:13:22 But really, debian-archive-keyring is the important one here. Jun 24 13:13:29 I think when I have written my message he has already executed the command ;) Jun 24 13:13:45 And far simpler than telling someone who had trouble looking up a config file how to manage keys with gpg... Jun 24 13:14:57 ^^ Jun 24 13:15:56 dpkg: error processing llvm-3.0-dev (--configure):.... when I run sudo apt-get install debian-archive-keyring Jun 24 13:17:25 Errors were encountered while processing: ntp binfmt-support llvm-3.0-runtime llvm-runtime llvm-3.0 llvm llvm-3.0-dev Jun 24 13:18:02 and also this before the error processing message .... any ideas please Jun 24 13:18:08 'sudo apt-get -f install' Jun 24 13:18:28 Or, you know, just reflash the thing and get back to stock. Jun 24 13:18:47 THe wonderful thing about embedded systems is the nuke-from-orbit approach is fast. Jun 24 13:21:02 I just run he command 'sudo apt-get -f install' Jun 24 13:21:18 still the same error as before .... Jun 24 13:21:34 do you mean that my BBB is broken ? Jun 24 13:21:43 Ibrahim: Please pastebin the entire contents of the error. Jun 24 13:22:01 YOu keep cutting off the full text, so none of us know what error you are actually seeing. Jun 24 13:22:12 dpkg: error processing llvm-3.0-runtime (--configure): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of llvm-runtime: llvm-runtime depends on llvm-3.0-runtime (>= 3.0-4); however: Package llvm-3.0-runtime is not configured yet. dpkg: error processing llvm-runtime (--configure): dependency problems - leaving unconfigured dpkg: dependency problems prevent configuration of llvm-3.0: l Jun 24 13:22:58 Errors were encountered while processing: ntp binfmt-support llvm-3.0-runtime llvm-runtime llvm-3.0 llvm llvm-3.0-dev E: Sub-process /usr/bin/dpkg returned an error code (1) Jun 24 13:23:08 OK, that was not pastebin, but OK. Jun 24 13:23:28 Ibrahim: Just uninstall all of llvm, please. Jun 24 13:23:36 I will send ou again he error step by step Jun 24 13:23:39 preferably with purge, not remove. Jun 24 13:23:43 Ibrahim: NO. Jun 24 13:23:56 ok Jun 24 13:24:04 If you need to share large amounts of text, use http://pastebin.com Jun 24 13:24:08 And then share the link. Jun 24 13:24:19 *NOT* giant blocks of text into the client directly. Jun 24 13:24:32 it will be cut off you know ;) Jun 24 13:24:47 sla89: Yeah, or Ibrahim will get kicked for flooding. Jun 24 13:25:20 amglego now we will never know ;) ^^ Jun 24 13:25:31 sorry for the nick typo Jun 24 13:26:19 sla89: Tab-complete works in most clients. ;-P Jun 24 13:26:47 http://pastebin.com/P6E6vv6g Jun 24 13:27:06 Ibrahim: Thanks. Jun 24 13:27:10 That's the full error sorry about pastebin i didn know it before Jun 24 13:27:41 How i can unistall llvm ? Jun 24 13:27:52 ... Jun 24 13:28:01 'sudo apt-get -f purge llvm' Jun 24 13:28:28 Then follow it by 'sudo apt-get autoremove' Jun 24 13:28:38 sudo apt-get remove llvm Jun 24 13:28:40 And maybe try reading the manpages for apt-get... ;-P Jun 24 13:29:01 sla89: I want to get to a clean system. purge is the correct option. Jun 24 13:29:02 Ibrahim: trust agmlego Jun 24 13:29:22 agmlego: i have corrected me ;) Jun 24 13:29:30 Thank you. Jun 24 13:29:33 agmlego: you are right - the tab works great :D Jun 24 13:29:40 I trust him I am following his steps ! :) Thanks ! Jun 24 13:31:03 tell us if you were successful Jun 24 13:32:15 agmlego: well - I only knew the apt-get remove thing - didn't know there was a purge which also removed config files and related packages Jun 24 13:32:21 agmlego: nice to know - thanks Jun 24 13:32:34 sla89: Not a problem. ALso, read manpages. ;-P Jun 24 13:32:58 http://pastebin.com/CnBDFCj8 Jun 24 13:32:59 agmlego: on windows you have the program uninstallation function which lefts 99% of the program on your disk - so actually it is the same :P ^^ Jun 24 13:33:19 Still the same issue !! Jun 24 13:33:30 Ibrahim: why apt-get remove, when agmlego told you to do apt-get -f purge? Jun 24 13:34:46 and when you do apt-get -f purge llvm Jun 24 13:34:57 well - actually llvm isnt installed o.O Jun 24 13:35:04 Actually, there are bigger issues with that image. Jun 24 13:35:26 The led_aging.sh script needs to be changed; it is broken and causing issues. Jun 24 13:35:45 rcn suggests just reflashing the board to the latest image. Jun 24 13:35:58 http://pastebin.com/ryK9TzA6 Jun 24 13:36:17 I run also this command and has the same issues Jun 24 13:36:19 https://groups.google.com/forum/#!topic/beagleboard/LPjCn4LEY2I Jun 24 13:36:21 same issue here Jun 24 13:36:35 Ibrahim: My suggestion is as well to reflash the board to the latest debian image, found at http://beagleboard.org/latest-images Jun 24 13:36:46 sla89: Yes, that is the threaqd I am reading. Jun 24 13:36:49 you can solve it like described in the 2nd post Jun 24 13:37:18 agmlego: google is our friend, isnt it? ;)^^ Jun 24 13:37:22 Sure could, but at this point I have lost interest in helping. Jun 24 13:37:45 I am more interested in Ibrahim getting the board to a known-working state than troubleshooting every tiny issue with his broken image. Jun 24 13:37:48 ;-P Jun 24 13:38:05 And that route is to just reflash the board. Jun 24 13:38:28 yeah - would be the easiest Jun 24 13:38:35 Ibrahim: what image have you installed? Jun 24 13:38:45 can you do uname -a and show us what you have on your bbb Jun 24 13:39:09 agmlego: it was uname -a right?^^ Jun 24 13:39:09 Linux beaglebone 3.8.13-bone47 #1 SMP Fri Apr 11 01:36:09 UTC 2014 armv7l GNU/Linux Jun 24 13:39:31 That is pretty old. Jun 24 13:39:34 yep Jun 24 13:39:41 And, as demonstrated, broken. Jun 24 13:39:57 Ibrahim: have you ever flashed your bbb with the newest version Jun 24 13:40:31 or it that still the image which was installed when you got your bbb Jun 24 13:40:32 No ... i think ... i have to use an SD card to do that ? Jun 24 13:40:38 Yes. Jun 24 13:40:39 yes Jun 24 13:40:45 normally you should have one already Jun 24 13:40:51 There are excellent tutorials online on how to do this. Jun 24 13:40:59 sla89: Why? Jun 24 13:41:10 The beaglebone black does not normally come with one. Jun 24 13:41:14 It has eMMC. Jun 24 13:41:21 agmlego: when I got my there was one already included :) Jun 24 13:41:34 Lucky for you I guess. Jun 24 13:41:35 okay - maybe my girlfriend added it then Jun 24 13:41:45 mine was a present :) Jun 24 13:42:11 There isnt any other way doing it without the SD card ? Jun 24 13:42:29 Can i do it just from the connection of my USB cable ? Jun 24 13:42:32 Ibrahim: YOu can do it with an MMC writer, or with a special setup on Ethernet. Jun 24 13:42:58 Ibrahim: But I am pretty sure you will have the best time just following the usual methods with a microSD card. Jun 24 13:45:37 okk :S no other solution .... damn ... thanks for help ... the best image to install is the latest version of angstorm? Jun 24 13:45:55 NO. Jun 24 13:46:04 Ibrahim: YOu need to read more carefully. Jun 24 13:46:15 Angstrom is no longer supported in any way. Please install Debian. Jun 24 13:46:32 okkk Jun 24 13:46:56 you get it here Jun 24 13:46:57 http://beagleboard.org/latest-images Jun 24 13:47:13 Debian (BeagleBone, BeagleBone Black - 4GB SD) 2015-03-01 Jun 24 13:47:50 okk ... the debian that arleady exists in my bbb Jun 24 13:48:19 can be update with an command ... just asking if there is a solution avoiding reflashing Jun 24 13:48:40 Ibrahim: The debian that is already on your bone is broken beyond our time and interest to fix. Jun 24 13:49:11 what kernel version is in the latest debian build? Jun 24 13:49:17 YOu certainly *can* fix it. YOu just need to be more diligent about self-education and research and put more time into it than your have demonstrated you are willing to do thus far. Jun 24 13:49:58 Ibrahim: The fastest way to a working board is to flash it. Jun 24 13:50:20 Ibrahim: The best way to learning what you are doing is to spend time learning what is broken, and fixing it yourself. Jun 24 13:50:25 Ibrahim: Your choice. Jun 24 13:51:13 Ibrahim: Either way, I will no longer be actively monitoring this channel for the day. Best of luck! Jun 24 13:51:29 agmlego: bye Jun 24 13:52:07 Thanks a lot agmlego !!!! For your help .. i will try to my best as new to that stuff Jun 24 13:52:14 Thanks again !!!!!!!!!!!!!!!! Jun 24 13:52:35 Ibrahim: dont give up - its hard at the beginning ;) we all had a lot to learn :) Jun 24 13:52:53 Ibrahim: if you have any further problems/questions feel free to ask Jun 24 13:53:07 Ibrahim: what are you going to do now? Jun 24 14:56:02 what is the meaning of step in ADC for BBB? Jun 24 14:57:32 where do you see that? Jun 24 14:57:50 new2bbb: the ADC subsystem has a 16-step sequencer Jun 24 14:58:07 av500: TRM of AM335x Jun 24 14:58:49 alexanderhiam: can you explain more on "16-step sequencer" Jun 24 15:00:17 the idea is to do something like give it a sequence of channels to convert and read them all at once (eg convert channel 2, then 3,then 6 and so on) Jun 24 15:01:14 yeah, it lets you setup a sequence of sampling operations, where each step has a selectable channel to sample, averaging settings, etc. Jun 24 15:01:42 and each step can load its data into a different result register Jun 24 15:02:21 you set that up ahead of time, then you can trigger it once by software or repeatedly with a timer Jun 24 15:02:40 if you want to fully understand the sequencer you really should read the whole ADC (touchscreen controller) section of the AM335x TRM Jun 24 15:02:58 for bbb it has 16 step registers. right? If I configure only upto 7 then what happens ? Jun 24 15:03:19 you can enable just the steps you want Jun 24 15:03:23 then you read only 7 AD channels Jun 24 15:03:41 in other words it supports that Jun 24 15:03:49 alexanderhiam: currently reading TRM :) but having doubt dats why asked Jun 24 15:04:02 sure, no problem Jun 24 15:04:41 alexanderhiam: 7 AD channels meaning 7 inputs BBB can take parallel ? Jun 24 15:05:41 well, not quite parallel, it's still sequential sampling Jun 24 15:06:20 but it lets you sample multiple channels quickly, and in real-time since it's clocked internally in the ADC subsystem Jun 24 15:06:57 they could also all be smpling the same input channel to give you a sample burst you could then take an average of or something like that Jun 24 15:07:07 say in BBB we have 7 pins for ADC ok? then if I give i/p to all of them. Then how this sequencer come into picture ? Jun 24 15:07:30 I mean want to understand with real example :) Jun 24 15:08:46 howver you want it to. You could just configure the first step to the first input, trigger a single sample and save the result, then configure the first step for the next input, etc... Jun 24 15:09:30 or you could configure the first 7 steps to each read from the different 7 inputs and store their results in the different step result registers and trigger it once Jun 24 15:09:41 it's really up to you and what makes the most sense for your application Jun 24 15:10:34 but I wonder there are only 7 pins for adc on bbb and it has 16 steps that we can configure Jun 24 15:11:31 I think I should read TRM once more. but sometimes getting confused as it is too long.. :) Jun 24 15:13:02 alexanderhiam: as of now my application is to interface adc using one posix system Jun 24 15:13:12 like I said, you could set it up to sample one channel 16 times then take an average. It's also a fairly general purpose ADC block that could be put in a different ARM that has 16 ADC input channels Jun 24 15:14:02 well, there's already an ADC kernel driver for the BeagleBone Jun 24 15:14:27 'can you point me to the link if possible Jun 24 15:16:08 alexanderhiam: well in short 1 step register is for 1 time sampling we can get up to 16. same can be done with all 7 pins. (but serially). Am I understanding right ? Jun 24 15:17:04 new2bbb: https://github.com/beagleboard/linux/blob/4.1/drivers/iio/adc/ti_am335x_adc.c Jun 24 15:17:13 yeah, that's rightr Jun 24 15:17:54 ohkk thanks alexanderhiam . got my concept clr Jun 24 15:21:52 will surely ask if getting more doubt :) Jun 24 15:27:52 alexanderhiam: i am not getting "linux/mfd/ti_am335x_tscadc.h" Jun 24 15:28:24 does anyone know how to fix the cpufreq failed to scale voltage issue. No matter what happens my cpu is running at 550 MHz. I'm using ubuntu 14.04 with a 3.8.13-bone37 kernel? Jun 24 15:31:29 new2bbb: not sure what you mean... Jun 24 15:31:55 alexanderhiam: I mean where is that file is located "linux/mfd/ti_am335x_tscadc.h" Jun 24 15:32:30 in include: https://github.com/beagleboard/linux/blob/4.1/include/linux/mfd/ti_am335x_tscadc.h Jun 24 15:32:55 ohh ya thanks alexanderhiam Jun 24 15:33:19 np Jun 24 15:33:31 I'm working on something where I'm powering a BBB using the headers, and it seems to work just fine, but for some reason serial quits working when I power it that way. Any ideas? Jun 24 15:37:35 hi Jun 24 15:38:40 hey Jun 24 15:39:07 I'm here for pessoc Jun 24 15:41:09 * tbr ponders what a pessoc is Jun 24 16:35:11 alexanderhiam: any application code for adc is there ? like simple adc input from 1 pin / Jun 24 16:36:14 rcn-ee care for some connman news? Jun 24 16:36:24 new2bbb: you mean using the driver? Or controlling the ADC module directly? Jun 24 16:36:43 yup using the driver code Jun 24 16:36:51 citylight2, i think i've caught up.. (took a vacation friday night till monday night) Jun 24 16:36:58 using "ti_am335x_tscadc.c" Jun 24 16:37:23 ok, so connman now has a new feature Jun 24 16:37:27 now just building lxqt/chromium updates for the repo, (mind numbing, need to automate it..) Jun 24 16:37:33 new2bbb: if you google "beaglebone black adc" pretty much every example will be using that library Jun 24 16:37:48 what feature did they add? ;) Jun 24 16:38:17 say you got an atheros based wifi stick Jun 24 16:38:24 alexanderhiam: that all are terminal based and it uses "/sys/bus/iio/devices/iio:device0" Jun 24 16:38:32 citylight2, so they autodetect it's ap? ;) Jun 24 16:38:40 so you can run two virtual wireless on it Jun 24 16:38:45 sweet! Jun 24 16:39:06 no , this is not connman Jun 24 16:39:06 new2bbb: the interface that driver exposes is through sysfs Jun 24 16:39:23 ok, so the kernel allows two devices Jun 24 16:39:41 so you use connman to connect to your home network Jun 24 16:39:59 you can use the other as a tether Jun 24 16:40:14 cool! Jun 24 16:40:21 this mean, that connman creates a new device called tether Jun 24 16:40:39 alexanderhiam: I mean to say any code that is using functions defined in ti_am335x_tscadc.c Jun 24 16:40:43 it checks the subnet and assigs a free subnet to tether Jun 24 16:41:06 it then uses the wireless device as a brigde Jun 24 16:41:16 new2bbb: no, it's a kernel driver, not a library Jun 24 16:41:34 assume wlan0 is connected, wlan1 will serve as an access point Jun 24 16:41:42 in one simple comman Jun 24 16:41:59 connmanctl tether wifi SSID PASSWORD Jun 24 16:42:26 now you can also do this with the usb, Jun 24 16:42:42 connmanctl tether gadget on Jun 24 16:42:45 alexanderhiam: cool ! thanks Jun 24 16:42:50 np Jun 24 16:42:56 cool, looking at: https://git.kernel.org/cgit/network/connman/connman.git/log/ still patches ? Jun 24 16:43:00 so the usb and the wifi are bridged to tether Jun 24 16:43:41 the feature is: when seeking for a wireless device , skip the ones in use Jun 24 16:43:56 otherwise wlan0 will serve as an access point Jun 24 16:44:54 so if you use connman you don't need dnsmasq Jun 24 16:45:04 and don't need hostapd Jun 24 16:45:16 or udhcpd ;) Jun 24 16:45:39 regarding this Jun 24 16:45:54 I just downloaded the image and dnsmasq is missing Jun 24 16:46:56 citylight2, which image? console = (currently neither, but future has udhcpd (as it's small)), lxqt = dnsmasq Jun 24 16:50:19 yes console Jun 24 16:51:02 also hostapd is missing Jun 24 16:57:04 so my problem with using connman is that tether does not use the same IP address Jun 24 16:57:15 as you use 192.168.7.2 Jun 24 16:57:42 so in order to ssh to the usb device you should use avahi Jun 24 16:57:53 but it does not work for me Jun 24 16:58:03 odd Jun 24 16:58:09 we use avahi, but it doesn't work 100%.. Jun 24 16:58:39 connman tether solution reduces a lot of the sciript Jun 24 16:59:15 yeah it does, found patric's patches, so i need to rebuild connman using git, plus his 3 patches.. Jun 24 16:59:40 use his conman-deb ... Jun 24 17:00:04 he is so sweet, this was added in 4 days Jun 24 17:00:08 amazing Jun 24 17:00:36 yet more QA is needed Jun 24 17:01:06 brb Jun 24 17:01:21 his conman-deb is nice, but i have siduction's version of connman installed, so i have to merge. ;) Jun 24 17:21:26 alexanderhiam: I wonder if I can see the code written in in_voltage0 file. in "sys/bus/iio/devices/iio\:device0/" Jun 24 18:41:43 rcn-ee: I switched to systemd-{networkd,resolved,timesyncd} a while back for ethernet... it feels weird to apt-get remove things like ifupdown, *dhcp*, ntp*, etc but it works like a charm, and noticably reduced boot time Jun 24 18:43:31 dunno if it does wireless already though Jun 24 18:48:19 ah of course it does, nearly all the work (scanning for, selecting, and connecting to a network) is done by wpa_supplicant Jun 24 18:50:35 maybe I'll give that a try on my laptop when I manage to find the courage... (given the many fights I've had with NetworkManager I'd be glad to get rid of it) Jun 24 19:30:32 zmatt, i'd like to switch to systemd-networkd, does it work with jessie, or did you have to pull stretch's systemd? Jun 24 19:30:57 i know the version in jessie, is a little 'early' for networkd... Jun 24 19:32:48 it works even in jessie, but then you lack decent tools Jun 24 19:32:52 or even basic tools Jun 24 19:33:19 or any tools whatsoever other than making the relevant config files and staring hopeful at your journal Jun 24 19:33:34 but still it works like a charm Jun 24 19:34:04 before that we had serious issues with startup taking ages for something to timeout if no ethernet was plugged in Jun 24 19:37:23 I should really upgrade those though.. I'm running stretch now on my dev bbb (and always run sid on "barebone", my own bbb) Jun 24 19:39:07 haven't been able to spend much time on systems since they dropped some "drop what you're doing, this has priority" stuff in my lap Jun 24 19:41:31 gotta do what pays the bills first. ;) Jun 24 19:42:24 as much as people dislike systemd, when configured it just works.. that dhcp delay in debian just annoys me... Jun 24 19:43:36 uhuh, and they know my head is kinda binary when it comes to topic-switching... so now I know little about power management, but have absorbed the Fast Fourier Transform into myself Jun 24 19:45:17 (poor BatBeagle has been sitting neglected on my desk, still haven't even tried booting linux on it yet) Jun 24 19:46:48 zmatt has backports newer systemd stuff? Jun 24 19:47:26 ehh, well when I first tried this I did install some newer systemd on my jessie dev-beagle Jun 24 19:47:48 now it just runs stretch Jun 24 19:48:31 oh that's also still an old systemd, hmm Jun 24 19:48:39 maybe it still has the packages I installed then Jun 24 19:49:23 but no "backporting" was involved other than just dpkg -i Jun 24 19:50:23 hm okay Jun 24 19:50:28 must look up myself Jun 24 19:50:39 i'm tempted to backport systemd by default (and stick it in the repo), do you guys think more users would like that? Jun 24 19:50:40 anyway ifupdown still works for me Jun 24 19:50:56 rcn-ee the systemd-haters likely not Jun 24 19:51:15 but if the network thingy runs much smoother as it as now Jun 24 19:51:16 well we've been systemd default since Wheezy. ;) i hope they left.. Jun 24 19:51:19 backport it Jun 24 19:52:10 most people won't know or care, they will just notice the results (if at all) Jun 24 19:53:09 and maybe a few puzzled glares from Jun 24 19:53:10 -bash: ifup: command not found Jun 24 19:54:02 well that's 2 and then me.. i see no nay's... let the record state they had a "chance" ;) Jun 24 19:55:01 I'm working on getting myself out the door to get to work... I can check then which packages I installed to get stuff like networkctl in jessie/stretch Jun 24 19:55:13 but I think it was just systemd 220 + deps Jun 24 19:55:57 i actually did that tueday, first thing i did after vacaction.. Jun 24 19:56:17 back then I pulled it from experimental Jun 24 19:57:00 sid has 220, that's the one i'll backport Jun 24 19:57:10 I'm really getting fed up with debian "unstable" being old, "testing" being ancient, and "stable" being "omg does that still exist?" Jun 24 19:57:40 zmatt, that's why we run 'testing' on our systems. ;) Jun 24 19:57:53 except during the freeze, it's "pretty new"... Jun 24 19:59:03 stable might be nice to minimize maintenance on servers... as long as you don't need anything "cutting-edge" (cough) Jun 24 19:59:06 oh that's right, 220 needs a util-linux upgrade too, i need to fix a couple scripts under /opt/scripts/ before i backport that.. (i know what the fix, just been doing other prioerites) Jun 24 19:59:25 but everything about the beaglebone is still pretty new and in flux anyway Jun 24 20:00:13 oh we have debian stretch test images out too: ;) http://elinux.org/BeagleBoardDebian#Debian_Testing_.28stretch.29 Jun 24 20:00:14 rcn-ee: FYI, I started looking a bit at the ti-opencl stuff and it seems like there are several x86-cross-build issues. Jun 24 20:00:24 issues == dependencies. Jun 24 20:00:25 (but apparently my makefiles were too cutting edge for wheezy's make to understand) Jun 24 20:00:47 rcn-ee: stretch = testing Jun 24 20:01:07 jessie is stable now Jun 24 20:01:34 jkridner, which of these is the problem child? http://git.ti.com/gitweb/?p=opencl/ti-opencl.git;a=blob;f=debian/control;hb=HEAD Jun 24 20:02:02 ti-llvm-3.3 & ti-cgt-c6x (different name) are in the repo... Jun 24 20:02:19 my thoughts, cmem/libtiulm-dev might be a pain.. Jun 24 20:02:40 not sure what pdk-keystone2 does.. Jun 24 20:04:15 "Programmers Development Kit" Jun 24 20:04:20 vague++ Jun 24 20:05:49 foudn the source; https://launchpad.net/ubuntu/+source/pdk-keystone2/03.00.04.19-0ubuntu1 Jun 24 20:06:35 is it october yet? :P I wanna play with Vayu too... I don't actually have time to play with it, but I want to anyway Jun 24 20:06:51 jkridner, zmatt's on the list right? ;) Jun 24 20:08:10 "the list".... *looks around slightly worried* Jun 24 20:09:36 so if I understand correctly the issue that's stalling the bbx15 is actually also stalling the public release of the am527x ? ... for six months ... that must be a pricy issue Jun 24 20:10:46 ti just want's it perfect... cause it's a beast.. ;) Jun 24 20:11:00 *572x Jun 24 20:11:06 lol Jun 24 20:11:20 well, not as much of as beast as the dra74x ;) Jun 24 20:11:25 same die Jun 24 20:11:31 (afaik) Jun 24 20:11:46 exactly, but those need everything working, otherwise they become an am57x Jun 24 20:11:57 yeah, no EVEs for us Jun 24 20:12:32 I did notice the DM814x EVM we have has a prototype IC with all feature-bits set except customer eFUSE Jun 24 20:13:05 but stuff like VCP2, ATL, MLB etc should actually work there Jun 24 20:13:55 (i.e. DRA65x stuff) Jun 24 20:14:27 ATL still sounds interesting to me, I don't understand why it's apparently considered automotive-only Jun 24 20:15:50 someone ready "hd-audio", nope that's for car's... Jun 24 20:17:02 though I still don't know what *exactly* ATL does due to lack of info, so maybe there's a reason it would be of limited or no use for other applications Jun 24 20:17:51 sounds neat: https://lwn.net/Articles/597621/ Jun 24 20:18:22 yeah it's related to tracking a clock for asynchronous sample rate conversion purposes Jun 24 20:18:39 but that's still vague Jun 24 20:20:03 would also be useful for AVB I'd imagine Jun 24 20:20:26 which I *is* a "public feature" Jun 24 20:22:31 also, why is the driver named dra7-atl when it was already present in dra6 :P Jun 24 20:22:49 (answer: because centaurus for some reason got completely neglected by TI) Jun 24 22:37:31 does anyone here know the fix for a cpufreq cpu0: failed to scale voltage? Jun 24 22:43:35 WHOAMI Jun 24 22:47:20 I think I may know the cause. I'm running off a sd card. The kernerl in the /boot dir is initrd.img-3.13.0-armv7-x9. uname -r shows 3.8.13-bone37 .. this must be the cause. the 3.8.13-bone37 doesn't have the correct drivers for cpufreq-set ... that must be some part of the problem. Jun 24 23:18:09 wednesday: huh, what kind of a kernel is "-armv7-x9" o.O Jun 24 23:18:48 zmatt: I think its from ubuntu Jun 24 23:19:02 you generally want a beaglebone-specific kernel Jun 24 23:19:06 -bone or -ti Jun 24 23:19:52 zmatt: my system isn't running it right now though, but when I do updates it probably ended up there on the SD card in the /boot directory. Jun 24 23:20:37 shouldn't do any harm as long as it doesn't actually boot from it :P Jun 24 23:20:52 zmatt: I've been running this system for a long time, I'm sure the upgrade has moved this system through at least a version of ubuntu LTS. Jun 24 23:21:29 rcn-ee: does 3.8.13-bone37 support cpufreq? Jun 24 23:21:46 zmatt: My problem is I get dmesg errors for cpufreq cpu0: failed to scale voltage up/down. So my system is running slow and stuck at 550 MHz. Jun 24 23:21:53 wednesday: ehh Jun 24 23:22:04 it was probally broke back then Jun 24 23:22:05 wednesday: the bootloader initialized it at 1 GHz Jun 24 23:22:45 wednesday: when cpufreq is broken, the system should be stuck at fast-and-hot Jun 24 23:22:46 zmatt: You've never seen this error? Its a pain... a real pain. if you type cpufreq-info Jun 24 23:22:46 it will tell you the current speed of your cpu. Mine is stuck at 550 MHz instead of 1GHz. Jun 24 23:23:20 zmatt: its about 57.0 C right now. Jun 24 23:24:03 I've definitely seen the error often enough, dunno under which circumstances exactly though Jun 24 23:25:11 but I haven't run anything that old in quite a while Jun 24 23:25:34 zmatt: trying to find a fix has been a real issue. I solved it months ago but I forgot to document the solution, and really I can't remember if it was for my BBB or my desktop. I can't find much that is working for a solution now for the issue and so it happens that my log file fills up and then my system reboots, or gets stuck trying to reboot. Jun 24 23:25:43 wednesday, looking at the log, 37 'should' work, but that was from Jan 22, 2014 Jun 24 23:26:08 wednesday: ehh, you mean you get the error more than once? Jun 24 23:26:09 I Jun 24 23:26:14 powering via usb? Jun 24 23:26:18 I've seen the error, but then only once at boot Jun 24 23:26:20 not repeatedly Jun 24 23:26:57 zmatt: repeats over and over again ... on and on until I cpufreq-set -r -g performance Jun 24 23:26:59 log file filling up leading to reboot also doesn't make sense to me.... though I've turned off logging to file anyway, because eMMC Jun 24 23:27:24 rcn-ee: no powered hub. 5v 1AMP Jun 24 23:27:45 wednesday, you shouldn't need to use cpufreq-set: CONFIG_CPU_FREQ_DEFAULT_GOV_PERFORMANCE=y Jun 24 23:28:49 cpufreq-utils is probally switch it to userspace in the boot process, disable that.. Jun 24 23:29:36 rcn-ee: I have no idea how to do that. Jun 24 23:29:54 rcn-ee: and this seems to be a recent even after an update. Jun 24 23:33:21 https://wiki.debian.org/HowTo/CpuFrequencyScaling#Configuration Jun 24 23:33:34 rcn-ee: btw, the packages I installed back then were {systemd,systemd-sysv,libpam-systemd,libsystemd-dev,libsystemd0,libudev1,udev}_219-7 and libapparmor1_2.9.0-3+exp1 Jun 24 23:33:51 to get networkctl Jun 24 23:34:01 and you need util-linux now too ;) Jun 24 23:34:08 (came in as a dependicy) Jun 24 23:34:25 sounds sensible Jun 24 23:34:44 Just have to add the sfdisk >= 2.26.x case to the eMMC flasher. ;) Jun 24 23:35:01 as util-linux doesn't belive in keeping a stable command line.. Jun 24 23:35:05 lol Jun 24 23:39:16 unstable cmdline is important to maintain developer vigilance ;-) Jun 24 23:39:55 rcn-ee: log, 37, is that the kernel revision? Jun 24 23:40:29 550 MHz is also a weird value btw, normally it's 300 - 600 - 800 - 1000 Jun 24 23:42:44 (or 275 - 500 - 600 - 720 for rev 1.0 silicon) Jun 24 23:43:21 zmatt: I know I don't udnerstand why it says 550. Step values are what you wrote. Jun 24 23:44:04 mine are 300, 600, 800, 1000 Jun 24 23:44:29 wednesday, what version of u-boot? Jun 24 23:44:51 well you can theoretically configure other cpu speeds as well, though I don't know if the driver/tools support that Jun 24 23:45:35 (the step values are those at which it's also able to lower the cpu voltage, resulting in further power savings) Jun 24 23:46:12 in v3.8x it was just a voltage/frequency array.. you could make more steps..... Jun 24 23:46:25 it used to say soemthing else 500 MHz (agregated by hardware) ... or something like that. Jun 24 23:46:47 rcn-ee: still same in 4.x isn't it? /cpus/cpu@0/operating-points Jun 24 23:46:57 in mainline it is.. Jun 24 23:47:18 in ti's 3.14.x it's checks the core and they have actual opp points setup pin the driver based on es version... Jun 24 23:47:50 wednesday, any chance do you have a spare microSD where you could boot the latest? Jun 24 23:48:41 rcn-ee: I don't know how to check that. Jun 24 23:49:27 burn this console to a microsd card and boot ;) http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#BBW.2FBBB_.28All_Revs.29_Standalone Jun 24 23:49:47 rcn-ee: I don't know how to check the uboot version. I believe this one executed from the emmc, and not the mmcblk. Jun 24 23:49:57 serial cable? Jun 24 23:50:30 I'm not sure actually how SPL decides which mmc device to use Jun 24 23:51:14 rcn-ee: yeah I have a plan to reload the system, but this is in production. Infact I'm using it to chat with the IRC now. Jun 24 23:51:32 SPL, isn't that smart.. it'll use u-boot.img from the same medium it found MLO Jun 24 23:51:45 rcn-ee: how does it know? Jun 24 23:51:53 oh wait the rom bootloader passes that info of course Jun 24 23:52:19 using u-boot.img from the same medium it came from would actually qualify as "smart" imho Jun 24 23:52:20 I may have a log of the uboot.. let me check. Jun 24 23:52:23 and the "interface" is still active... just a dump `pwd` ;) Jun 24 23:53:16 rcn-ee: you're not making sense now Jun 24 23:53:37 at one time i probally tried MLO (eMMC) and u-boot.img in (microSD), it didn't get beyond eMMC. ;) Jun 24 23:54:23 rcn-ee: no log of uboot. I had another issue requireming me to jump the jtag or uboot would get stuck and not boot the system. jump j1 and j4 to get past that. Jun 24 23:55:46 jtag? Jun 24 23:55:52 how did jtag enter into this story? Jun 24 23:56:30 especially since almost nobody has a BBB with a jtag connector soldered on Jun 24 23:57:19 (also, u-boot most definitely does not know or care about the state of jtag) Jun 24 23:57:38 zmatt: sorry, not jtag ... Jun 24 23:57:40 i think he means the serial, that would stop u-boot Jun 24 23:57:44 ah Jun 24 23:58:06 zmatt: uart j1 and j4. Jun 25 00:00:33 pins 1 (gnd) and 4 (rxd) of j1 you probably mean... though that should never need a jumper since pin 4 already has a pull-down resistor to ground Jun 25 00:01:31 zmatt: for some reason this is an issue. let me get a link. Jun 25 00:02:11 https://groups.google.com/forum/print/msg/beagleboard/1rbO7JGhppw/3e0LbzBUhWwJ Jun 25 00:03:18 zmatt: on my BBB if it doesn't believe there is a serial connection going on it will not boot at all. It waits in uboot and never continues. Jun 25 00:05:01 Hello Jun 25 00:05:32 the jumper doesn't make u-boot think there's a serial connection (in fact it does the opposite), rather it's apparently meant to avoid it from picking up noise during "press any key to halt" Jun 25 00:05:59 though connecting a serial cable would normally do the same Jun 25 00:07:18 (it would drive it high rather than low, but as long as it's driven either way is fine) Jun 25 00:08:04 zmatt: I found this out. One day the system wouldn't boot, I used a serial connection to see what uboot was doing and the BBB booted up fine. Then I thought it was all good. I then took serial off and it didn't boot again. repeat this several times, then searched and found someone else having the same issue. Jun 25 00:08:31 wednesday: what revision of BBB do you have? Jun 25 00:09:25 zmatt: a5a I believe. Jun 25 00:10:43 hmm that should already have the pulldown Jun 25 00:11:31 zmatt: maybe its not an a5a and its just got an a5a sticker. Oddly enough it used to not need the jump at all. One day it just needed it. Jun 25 00:11:46 same for some other people Jun 25 00:12:13 zmatt: I don't use it at all for any GPIO at all. Its just a computer to me. Jun 25 00:13:03 strange use for a BBB Jun 25 00:14:09 zmatt: at the time it was faster that my raspi, until recently its stuck at 550 MHz. I needed a low powered system, to cut cost on electricity. That's why I got the device. Jun 25 00:14:27 I mean, it's not exactly a great general-purpose computer, with its low-end-smartphone cpu, lack of gigabit ethernet, and buggy usb Jun 25 00:14:32 ok Jun 25 00:15:36 zmatt: I had no idea it was going to have such a buggy USB device, that's why I had to move to raspi devices for GPIO mixed with various USB devices and wifi. The BBB never would be stable enough, but it was fast. Jun 25 00:16:02 USB is hell Jun 25 00:16:11 if I could make the usb device even work. I would have stayed with Beaglebone as my platform. Jun 25 00:16:34 My problem is I loved this little machine so I have a hard time letting it go and not work anymore. Jun 25 00:17:33 I'm not sure how to help you, other than encouraging moving to more recent software Jun 25 00:17:52 I got 4.x kernels on all beagles here Jun 25 00:17:59 zmatt: thank you. I will do that in a few days and see if that fixes my issues. Jun 25 00:18:07 it's possible even usb might work better Jun 25 00:18:17 though I'm not sure, I actually omit usb entirely from my own kernel config Jun 25 00:19:05 the UART RX issue is weird, especially with it appearing suddenly Jun 25 00:19:05 zmatt, you 'might' save some power via; https://github.com/beagleboard/bb.org-overlays/blob/master/src/arm/dev-USB-PWR-CTL-00A1.dts Jun 25 00:19:21 zmatt: not sure why this is such an issue, but the micro usb works great. TI/Beaglebone could really use the help with a working USB and not such a power monger device. Jun 25 00:20:07 rcn-ee, zmatt: thank you for the help. ... got to run for now short drive home. Jun 25 00:20:26 rcn-ee: ah, cuts the usb 5v driver? Jun 25 00:20:55 yeap. ;) we needed a way to 'reset' a downstream usb-device.. so we found the gpio controlling the usb-pwr line. ;) Jun 25 00:21:05 rcn-ee: I also wonder in what state the usb phy is left... probably also not optimal Jun 25 00:21:20 well, as long as you unbind usb1... Jun 25 00:21:29 it's not bad... just doesn't like ti repeated to many times.. Jun 25 00:22:00 rcn-ee: there's nothing to unbind obviously Jun 25 00:22:22 # CONFIG_USB_SUPPORT is not set Jun 25 00:24:29 it concerns me a bit that apparently multiple people are having trouble with with uart0_rxd line, trouble appearing without apparent cause after a long while Jun 25 00:25:50 we've noticed with later u-boot's, it's less of an issue.. That's also why the offical "beagleboard.org" image is still May 2014, the "tester" can't stop u-boot, to flash the eeprom any more. ;) Jun 25 00:25:53 I'm wondering if it may at all be related to the strain potentially put on uart0_rxd due to power-up/down issues Jun 25 00:27:38 putting the system in rtc-only mode while having a console cable attached would be the worst-case scenario, but actually any shutdown while having a console cable attached still briefly strains the cpu pin Jun 25 00:29:42 and prior to rev A6 also every powerup (A6 fixed that but made the shutdown issue worse) Jun 25 00:30:15 with "tester" you mean automated tester? Jun 25 00:30:48 (I'd personally use ethernet booting for such an application) Jun 25 00:31:29 correct, it's the tester cco uses to test all the functionalty of every board made.. Jun 25 00:33:51 I don't see a fried console rxd as a huge obstacle there... reflash using usb or ethernet boot Jun 25 00:34:40 or netboot altogether instead of flashing... especially automated testing of images sounds like a rather big strain on eMMC Jun 25 00:35:45 but if my worries are true, that probably means it may be a good idea to warn against having a console cable attached unnecessarily Jun 25 00:56:30 during one of the many tests I did a while back I actually saw readings suggesting that 45 mA may have been flowing through the UART0_RXD pin... it survived, this time, but I doubt the pin will have appreciated it **** ENDING LOGGING AT Thu Jun 25 02:59:59 2015