**** BEGIN LOGGING AT Fri Sep 18 02:59:58 2015 Sep 18 12:27:01 hey Sep 18 12:27:42 can someone help me with a stream UDP from a file with beagkebone Sep 18 12:27:48 ? Sep 18 12:37:05 <_av500_> as in? Sep 18 12:37:41 <_av500_> https://www.google.com/search?q=linux+send+data+over+udp Sep 18 12:41:49 I have this error Sep 18 12:41:51 ./capture: error while loading shared libraries: libv4l2.so.0: cannot open shared object file: No such file or directory Sep 18 14:13:00 help: how do i add users in beagle bone black Sep 18 14:14:58 Cyrus_: like on every other linux system. google for adduser Sep 18 17:05:53 hey Sep 18 17:06:16 is anybody there who can help me out with this beaglebone "white" Sep 18 17:10:52 BG__, ask your question and find out.. Sep 18 17:11:42 This beaglebone white doesnt boot my friend gave it to me for free and i connect it to pc , pc doesnt recognize it Sep 18 17:11:57 only power led flashes rapidly Sep 18 17:12:39 the "white" has a built-in usb-serial adapter... fire up a serial terminal and see what it's doing.. Sep 18 17:12:51 (ftdi) Sep 18 17:15:02 guys i am a newbie Sep 18 17:15:24 BG__, windows, mac or linux? Sep 18 17:15:42 windows Sep 18 17:15:59 BG__, install terterm: https://ttssh2.osdn.jp/index.html.en Sep 18 17:24:24 yes Sep 18 17:27:39 BG__, Plug in "otg" cable, with teraterm you should see a serial connection show up in the drop down list.. Sep 18 17:30:39 when i connect it to pc it flashes once some times two and even the light goes out Sep 18 17:37:47 Really Cool! Sep 18 17:38:41 I love beaglebone, so much. Right now I'm reading "Exploring Beaglebone" from Derek Molly! Sep 18 17:44:10 It looks like Debian's UART drivers have hooks for enabling an RS-485 transmitter. Is anyone using that successfully? Is there a default pin that's already set up? Sep 18 18:10:25 rcn-ee, jkridner: would the beagleboard-x15 make a decent build machine, running constantly? Sep 18 18:11:15 looking to add more armhf nodes for the reproducible builds project: https://wiki.debian.org/ReproducibleBuilds/About Sep 18 18:11:25 https://reproducible.debian.net/unstable/index_suite_armhf_stats.html Sep 18 18:11:29 he heheheh ehhehe he Sep 18 18:12:46 really want to add more CPU types; most of the nodes are IMX6 based ... occasionally programs optimized based on CPU type and it would be nice to detect those. Sep 18 18:13:24 ds2: is that a no to my initial question? :) Sep 18 18:16:17 vagrantc: isn't a 64bit arm better? Sep 18 18:16:24 (for your compilation purposes) Sep 18 18:17:00 ds2: would be great... if you had a lead on a donation :) Sep 18 18:17:14 they are currently absurdly spendy Sep 18 18:17:19 hmmmm? Sep 18 18:17:23 what is absurdly spendy? Sep 18 18:17:31 give me a number Sep 18 18:17:48 i think the cheapest I saw was ~USD900 ? Sep 18 18:18:59 there are a few arm64 boards in the ~USD100 range, but they don't have ethernet or sata/usb3 Sep 18 18:19:59 350ish Sep 18 18:20:07 but I suppose it depends on your definition of board :D Sep 18 18:20:22 What this planet needs is a good $5 supercomputer. Sep 18 18:20:36 define 'good' :D Sep 18 18:20:44 they have them, they're just 20 years out of date Sep 18 18:20:51 Runs Linux Sep 18 18:20:54 most people carry them in their pockets Sep 18 18:21:11 if you can waive the runs linux on the nodes requirement... Sep 18 18:21:55 I'd think Linux would be a poor choice of OS on the nodes Sep 18 18:22:53 vagrantc: do you really need a bare board for compilation? Sep 18 18:23:01 ds2: what's the 350ish you're talking about? Sep 18 18:23:27 ds2: versus what? Sep 18 18:23:53 vagrantc: there are 64bit ARM devices for around that price Sep 18 18:24:11 Make the nodes completely dumb and just boot whatever application kernel on them you need at the moment. Sep 18 18:25:06 once you do that... any OpenGL system can..... Sep 18 18:27:53 ds2: have any links? Sep 18 18:28:15 vagrantc: search for Nexus 9 Sep 18 18:28:16 :D Sep 18 18:40:51 I'm with vagrantc; my phone is orders of magnitude more powerful than a Cray 1, and highly parallel, and runs a Linux variant. Sep 18 18:41:23 vagrantc, i think i have build numbers for the main openjdk armhf package, a dual core a15 (omap_uevm) vs the quad a9 (wandboard), both sata spinng rust, i think the a15 was 30% faster.... Sep 18 18:43:29 vagrantc, here's my post: https://groups.google.com/d/msg/beagleboard/9cI499p2neo/4dEZuqDWtcwJ Sep 18 18:45:31 rcn-ee: nice. Sep 18 18:46:55 Anyone have a MIPS performance chart for the various ARM cores? Sep 18 18:48:35 CathyInBlue, not the best, (right hand side of) https://en.wikipedia.org/wiki/List_of_ARM_microarchitectures Sep 18 18:49:01 are you going to factor the 2x c6000 into it too. ;) Sep 18 18:54:50 rcn-ee: no tegra K1-64bit to compare it to? Sep 18 18:56:35 ds2, i could try it on the apm-mustang... that's 8 64bit cores.. ;) Sep 18 18:57:07 Any SBCs using the A72? Sep 18 18:58:26 CathyInBlue, the Rockchip RK3288 has that, it's a cromebook.. Sep 18 18:59:19 Not suitable to my applications. Sep 18 19:00:46 CathyInBlue, that didn't stop people from ripping out the sbc (a15) in first generation samsung cromebooks.. Sep 18 19:08:19 http://www.dhgate.com/product/in-stock-q8-rockchip-rk3288-android-4-4-tv/217574122.html <-- Could be promissing. Sep 18 19:13:11 so is that SBC the crook board? Sep 18 19:39:46 Hi Guys, can you explain to me How can I change the boot mode to only see the Terminal mode instead of access the GUI mode? Sep 18 19:50:12 Is it possible to run an iOS app on a Beaglebone black? Sep 18 19:50:55 hi everyone, I've got a question about reliability and the SD card slot. I have a few devices that are giving me the following errors when SD cards are inserted (they also won't flash from the SD card slot, uboot gives an error that it cannot find a bootable device) Sep 18 19:50:57 [ 58.671106] mmcblk0: error -84 transferring data, sector 60367744, nr 8, cmd response 0x900, card status 0xb00 [ 58.710064] mmcblk0: error -84 transferring data, sector 60367744, nr 8, cmd response 0x900, card status 0x0 [ 58.720158] blk_update_request: I/O error, dev mmcblk0, sector 60367744 Sep 18 19:51:34 i have tested the card and it works fine on my computer as well as other beagles. Does this mean the actual slot is bad on the board? Sep 18 19:53:25 @clcass, unfortunately that's not possible. Sep 18 19:55:57 beagler, it's either a bent pin, or that microSD card is shot.. Sep 18 19:58:32 hmm, seeing as it works on other beagles my guess would be bent pin. If I test the continuity between pins and the contacts of the micirosd, would that help troubleshoot? Sep 18 20:01:25 Does anyone use a beagle board to attach to a UPS and monitor power and on/off, and initiate a restart of the UPS? Sep 18 20:06:12 sasgad: no, but I really like that idea. Sep 18 20:10:10 sounds like your ups is either (1) crap or (2) broken if it doesn't manage an outage by itself .. Sep 18 20:10:12 rcn-ee, i tried removing the card, depressing some of the pins to ensure they make contact, then re-inserting the sd card and this seems better so far Sep 18 20:11:29 my desktop is attached to an apc ups, and with apcupsd I can do all sorts with it Sep 18 20:11:29 is it possible to run an iOS app on beaglebone black? the app is written in objective c. Sep 18 20:11:52 clcass : no not unless you re-compile it in ANSI C Sep 18 20:12:15 clcass, can the gnu objective c compiler build it? Sep 18 20:13:15 or maybe mono can run the non-IOS specific code? Sep 18 20:13:30 veremit and rcn-ee - thx for the replies, i'll look into them Sep 18 20:16:11 clcass, port it to qt5... then you can run it on all... Sep 18 20:16:26 About the UPS thing, what I had in mind is attaching a beagleboard device that could host the APC or Tripplite software and be accessible from the network or the Internet Sep 18 20:17:44 Do those programs come in Linux binaries? Sep 18 20:18:34 I assume there are open-source versions for APC observation and/or control. They're mostly USB-based these days. Sep 18 20:18:52 will try that, thx Sep 18 20:29:20 google 'apcupsd' and see http://www.networkupstools.org/ Sep 18 20:29:44 apcupsd is a debian package Sep 18 23:18:56 how is the mainline support on the beagle-x15 for linux and u-boot ? Sep 18 23:22:33 vagrantc, here's the branch: http://git.ti.com/gitweb/?p=ti-u-boot/ti-u-boot.git;a=shortlog;h=refs/heads/ti-u-boot-2015.07 Sep 18 23:22:56 not all pinmuxing and memory fixes have hit master yet.. Sep 18 23:23:50 mainline will boot, there a few dts updates needed on mainline for the change for video output regulator Sep 18 23:27:05 git masters of libdrm/xf86-video-omap support it.. libdrm gets bumped often but xf86-video-omap needs a little nudge.. Sep 19 00:21:13 rcn-ee: btw, I noticed you switched to the -ti kernel as default rather than the -bone ... what are the pros/cons ? Sep 19 00:25:24 right now... more eye's on it.. (ti.com) and i'm shooting for a shared kernel between the x15/bb(w/b/g)... Sep 19 00:26:08 but all the overlays stuff we've done with v4.1.x-ti has been copied to v4.1.x-bone/v4.1.x-bone-rt/v4.2.x-bone/v4.3.x-bone. ;) Sep 19 00:26:38 I thought overlays made it to mainline? Sep 19 00:28:00 (though tbh I normally never use them anyway) Sep 19 00:28:21 well... we are down to 29 (smallish) patches: https://github.com/RobertCNelson/linux-dev/blob/master/patch.sh#L343-L377 Sep 19 00:28:31 and the dtc compiler changes are under review... Sep 19 00:28:37 so we are damn close. ;) Sep 19 00:29:15 merging the etnaviv looks like a bigger pain then capemanager right now.. Sep 19 00:29:57 yeah I actually got my own dtc on github that merged the upstream dtc changes with overlays and some fixes of personal annoyances Sep 19 00:31:32 the dtc is probally the most important issue, the 3.8 vs 4.1+ abi is different, so whatever gets pushed to dtc git, i'd like to make sure v4.1 works with that.. Sep 19 00:35:45 well the ones my dtc produces worked... assuming the abi different is something obvious and not some subtle corner case. I derive from pantoniou/dt-overlays5 Sep 19 00:35:54 *abi difference Sep 19 00:36:42 but it's good to know, something those still using 3.8 need to be mindful of Sep 19 00:37:18 Yeah that version is fine.. if you decompile a dtbo from the 3.8 era vs the current dtc builds, you'll notice an extra "tree" it's suppost to help the kernel overlays find the node.. Sep 19 00:39:35 btw, since you've applied the cpsw patch that fixes the interrupt handling code to not blindly assume both interfaces exist, you can now set slaves = <1>; in the ethernet node to get rid of the phy not found error at boot Sep 19 00:39:52 humm, panto's looking at useing extcon, https://github.com/pantoniou/linux-beagle-track-mainline/commit/7d456c34d3e024602da33cdfd971fdf0ba735b0c wonder if extcon supports live changes.. (that might be cape-universal) Sep 19 00:40:43 zmatt, does the 2nd mac address still populate under /sys when you do that? Sep 19 00:41:12 ehh, where? Sep 19 00:42:03 (you know each am335x actually has 3 MAC addresses allocated?) Sep 19 00:42:11 err: /proc/device-tree/ocp/ethernet@4a100000/slave@4a100300/mac-address Sep 19 00:42:42 there's a third? i've been using the 1st/2nd to generate a fake third.. (the g_ether needs two mac's) Sep 19 00:42:58 yeah, 3 consecutive numbers Sep 19 00:43:39 and in /proc/device-tree ? is u-boot populating that then? Sep 19 00:43:51 that's just the kernel.. Sep 19 00:44:47 CONFIG_PROC_DEVICETREE has to be enabled.. Sep 19 00:45:28 yeah I know, but I mean /proc/device-tree just shows the DT as the kernel got it, so if the mac-address property is there yet not in the DT source then u-boot must be adding it Sep 19 00:46:48 kernel populates that: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/net/ethernet/ti/cpsw-common.c Sep 19 00:48:28 that code is still not populating anything in DT though Sep 19 00:48:56 the kernel reading the MAC address from hw doesn't exclude u-boot from doing the same (especially since it has network support) Sep 19 00:49:37 probe of cpsw, it get's it: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/net/ethernet/ti/cpsw.c#n2056 Sep 19 00:51:19 of course, if you set it in u-boot, the kernel will just use that value... Sep 19 00:51:27 it's still just fetching data, not mutating DT Sep 19 00:51:32 but Sep 19 00:51:35 why are things caring about the mac address? Sep 19 00:51:57 in any case, it seems I only have one slave node Sep 19 00:52:05 ds2, zmatt got me interested in their being a 3rd. ;) Sep 19 00:52:32 * zmatt looks up the relevant reference Sep 19 00:53:13 and we should be able to get rid of that "phy not found error at boot" ;) Sep 19 00:53:48 blah Sep 19 00:54:44 although it's been a few weeks since i got a "phy not found" email... Sep 19 00:55:08 slaves = <1>; is the correct solution Sep 19 00:55:39 since there really is only one mii port in use (which is what they mean by that) Sep 19 00:56:36 argh, why can't I find it anymore Sep 19 00:56:56 too many 3000 pages docs. Sep 19 00:57:04 isn't that the address on that 2 wire serial bus? Sep 19 00:57:09 mdio Sep 19 00:57:12 no Sep 19 00:57:50 oh then it must refer to the hardware port used? Sep 19 00:57:53 rcn-ee: what's the delta between the two addresses on one device? Sep 19 00:58:04 ds2: MAC address is the 6-byte globally unique thing Sep 19 00:58:37 zmatt: eh? MAC addresses do not have to be globally unique (see locally admin. bit) Sep 19 00:58:51 e4 > e6... Sep 19 00:58:52 ds2: yeah ok Sep 19 00:58:59 is the third between? Sep 19 00:59:01 rcn-ee: bingo Sep 19 00:59:12 rcn-ee: it indicates the range reserved for the device Sep 19 00:59:21 ah!.. cool Sep 19 00:59:23 but I'm still trying to figure out where I read that Sep 19 00:59:40 so we always assume, x, x+1, & x+2? Sep 19 00:59:48 on am335x... Sep 19 01:01:00 yes Sep 19 01:01:11 it's probably no coincidence either that 0xe4 % 3 == 0 Sep 19 01:01:27 or it might be, who knows Sep 19 01:02:06 it is Sep 19 01:04:52 rcn-ee: you can of course also just read those two addresses from the control module... it's portable at least across the whole dm81xx/am33xx/am43xx/dra6xx family Sep 19 01:05:21 the range covered varies though... dm814x only has two addresses Sep 19 01:06:16 we could generate a locally admin address based on the board S/N Sep 19 01:06:33 you can also just generally a locally admin address by picking one at random Sep 19 01:06:54 due to the problem of DHCP, a persistant address is needed Sep 19 01:07:01 then windows/systemd users have fun issues... (if it's random between reboots) ;) Sep 19 01:07:07 ah Sep 19 01:07:14 systemd users can be fixed Sep 19 01:07:21 windows just needs to be disposed of. Sep 19 01:07:22 it's too late for them. ;) Sep 19 01:07:32 systemd users are in YOUR hands :P Sep 19 01:07:43 * ds2 pets good old SysV init Sep 19 01:07:44 :D Sep 19 01:07:47 i've been assumlated... it's too late for me too.. Sep 19 01:09:27 but anyhow, you have three official reserved MAC addresses per am335x Sep 19 01:09:31 so Sep 19 01:10:18 cool: http://fpaste.org/269119/ and i still get both mac's. http://fpaste.org/269120/ Sep 19 01:10:42 and no mdio missing ;) http://fpaste.org/269121/ Sep 19 01:11:26 ahh, I just noticed I actually delete the second slave node in my DT Sep 19 01:11:37 that explains its absence quite well Sep 19 01:12:18 heheh Sep 19 01:12:39 but setting slaves to <1> is enough to keep the kernel from trying to use it Sep 19 01:12:49 I just deleted it because... well, why not Sep 19 01:13:15 it's just a magic struct anyways... call it fuzz testing.. Sep 19 01:14:24 board files would avoid this crap Sep 19 01:14:25 :P Sep 19 01:14:51 we did a lot of crap in board files too. ;) Sep 19 01:15:13 yes but that is more manageable Sep 19 01:15:33 I'd prefer good metacompilation infrastructure... all this stuff being probed/scanned/constructed at *runtime* is just ridiculous Sep 19 01:15:49 thinks back to the "buddy=" and "buddy2='... manageable. ;) Sep 19 01:16:11 the whole concept of capes is broken in the first place Sep 19 01:16:25 the whole concept of a dynamic kernel is broken. ;) Sep 19 01:16:36 they are expansion boards and you need to mod the kernel to suit your HW Sep 19 01:16:43 yes. kernels should be static Sep 19 01:16:51 remember v4.1.x is the terminator... look what we did... Sep 19 01:18:34 zmatt, is that irq fix also in mainline or should i just keep the slaves = <1> for the v4.1.x-ti branch? Sep 19 01:18:56 rcn-ee: no idea what it's status is... you're the one who located it in the first place :) Sep 19 01:19:53 ah.. so it's beer time to fix my long term memory. ;) Sep 19 01:19:59 lol Sep 19 01:20:15 beer... swamp water... same thing Sep 19 01:21:25 it's selected swamp water... you don't want every animal's piss.. just one organism's Sep 19 01:23:24 also, if anyone happens to have an urge to be able to throw C++ exceptions out of synchronous signal handlers (SIGSEGV, SIGBUS, etc), I got the glibc fix working that people have been asking periodically for for the past four years Sep 19 01:24:04 (no need to patch glibc, you can just link two small files into your program) Sep 19 01:26:04 arrrgg... people still using glibc?! Sep 19 01:26:17 Paul Brook "explained" how to do it in 2011 ( https://gcc.gnu.org/ml/gcc/2011-10/msg00482.html ) but his outline is not exactly obvious to expand Sep 19 01:26:38 ds2: actually it'll work for any libc Sep 19 01:27:38 the fix is a customized arm eabi unwinding "personality routine" for the signal restorer Sep 19 01:27:57 that actually unwinds the signal frame correctly Sep 19 01:29:41 so it is arm specific Sep 19 01:29:43 basically just emulate the kernel's sys_sigreturn but acting on the unwinder's virtual regsiter state Sep 19 01:29:52 yes, because it's broken arm-specific Sep 19 01:30:00 it works fine on most other platforms already Sep 19 01:30:54 in all cases you need to compile with -fnon-call-exceptions to inform gcc that exceptions can fall out of the sky almost anywhere, and then you're good to go Sep 19 01:31:38 people use it most commonly to emulate Windows' behaviour of throwing a NullPointerException on NULL-dereference Sep 19 01:32:08 sigh.... Sep 19 01:33:50 well if you have to choose between either that or actually having to try and *fix* the crawling horror of a codebase that someone dropped into your lap? ;-) Sep 19 01:34:03 I rather fix it Sep 19 01:34:26 i am usually the receiver of such bugs Sep 19 01:35:14 what bugs? it worked "fine" on windows... Sep 19 01:35:20 ;) Sep 19 01:55:02 tell that to the angry customer ;) Sep 19 02:21:39 Hello **** ENDING LOGGING AT Sat Sep 19 02:59:59 2015