**** BEGIN LOGGING AT Wed Aug 02 03:00:01 2017 Aug 02 03:31:06 woo hoo, 2 serial ports on BB blue now working nice with gps module (1 and 5) Aug 02 03:32:20 i guess that leaves i2c left, with a level shifter Aug 02 03:49:24 tomorrow… wire scanse.io lidar to 5v gps port and see if that flies. then next HC-SR04 sonars via gpio Aug 02 07:28:59 zmatt :is it possible to work with pwm for a single pulse Aug 02 07:29:10 short pulse (1ms) Aug 02 08:12:05 I'm pretty sure all the various pwm peripherals (eHRPWM, eCAP, timers 4-7) can generate a single pulse yes Aug 02 08:12:23 1ms isn't short btw Aug 02 08:14:47 1ms is a bloody fortnight .. Aug 02 08:18:40 eHRPWM wouldn't be able to make such a long pulse if it didn't have a prescaler :) Aug 02 08:19:03 (16-bit counter) Aug 02 08:46:28 anyone faced with the problem of changeing the ip of the board and forgot about what it is? and found a good way to restart it? Aug 02 08:47:50 pathrocle: ways: 1) use a serial cable to log in 2) use a network scanner 3) if it is dhcp, look it up in your dhcp server Aug 02 08:49:16 use something like bbblfs to mount it via usb on another computer and inspect the config file Aug 02 08:49:27 (just to add more options) Aug 02 08:49:45 LetoThe2nd, aham, i atm now i connect a keyboard, and i have a python sript that waits for "restart" input and rewrites the /etc/network/interfaces with a default one Aug 02 08:49:47 more is more. Aug 02 08:50:04 that sounds awful Aug 02 08:50:30 zmatt: lets number it way #4782 or so. Aug 02 08:50:50 (the something something python script that rewrites /etc/network/interfaces ) Aug 02 08:51:00 zmatt: i know. :-) Aug 02 08:52:36 well the end user that will use the board, cant do 2, 3 and i'm using 1 but i'm not sure if they will have that type of cable Aug 02 08:53:35 can't you just connect to the bbb by hostname? Aug 02 08:54:10 more ideas: install a python script that pulses out the ip over the leds. one led per octet Aug 02 08:54:32 I haven't had any need to care about the ip of a beaglebone in a long time Aug 02 08:56:09 doesn't mean we can't be creative! Aug 02 08:56:40 zmatt, how you use the hostname? you connect the board to lan and ssh into it like "ssh beaglebone" / "ping beaglebobne" ? Aug 02 08:58:57 or whatever hostname has been configured. this works if you have a router on your lan which also provides local dns service (typical for example for things like cable modems and adsl routers), or if the computer you're using can do name lookup via mDNS (default in most linux distros I suspect, standard on Mac, not standard on Windows -.- ) Aug 02 09:04:35 for mDNS you might need to explicitly use "beaglebone.local" instead of just "beaglebone" (although I'd expect just "beaglebone" to work too) Aug 02 10:48:33 I'm trying to upgrade my BBB to the latest Debian image which seems to be be 2017-03-19. When I've got the file written onto the microSD card and flashed onto the BBB Isee BeagleBoard.org Debian Image 2015-11-12 at bootup. Is this right? Aug 02 10:50:06 are you sure it flashed it to the board? Aug 02 10:50:24 you must edit configuration files to make a regular image into a flashing image Aug 02 10:52:20 No, I'm not sure: I held down the Boot button while I connected the USB. The outer LEB flashed regularly and I left it for 45 minutes, removed the card and rebooted. Aug 02 10:52:45 right, then you just booted debian off the SD card Aug 02 10:53:05 after removing it, you booted back to the original unchanged environment Aug 02 10:54:53 I opened the gz file with 7zFM and wrote the img file to the card with Win32 Disk Imager Aug 02 10:55:28 What should I have done? Aug 02 10:55:53 http://elinux.org/Beagleboard:BeagleBoneBlack_Debian#Flashing_eMMC Aug 02 10:57:04 also note that it won't work on a RevB or RevA board Aug 02 11:00:16 Peter_l: when it's flashing the leds do a very distinctive cylon / knight rider pattern Aug 02 11:00:57 the heartbeat led is just default config what you see on a running system (along with various indicators on the other leds) Aug 02 11:01:26 note that on that wiki page you can also download ready-made flashers which you don't have to manually edit Aug 02 11:03:18 yeah, noticed that they started doing that again Aug 02 11:12:35 they stopped doing it? Aug 02 11:13:04 what I think is confusing is that the flasher apparently powers off when it's done nowadays Aug 02 11:13:42 so if you return after a while and find the bbb off, either flashing was successful or your power supply was too crappy Aug 02 11:16:04 yeah, for a while there were only sd-card images, something about people accidentally blowing away their emmc or some rubbish like that as justification Aug 02 11:17:51 well I can sort of understand the concern, although I think that better solutions exist than expecting people to change the /boot/uEnv.txt on the card Aug 02 11:18:50 maybe the flasher could "prompt" using the leds and require the S2 button to be pressed again as confirmation to start flashing Aug 02 11:19:41 http://beagleboard.org/latest-images Aug 02 11:19:42 that would also prevent the reflash-reboot loop that at least one user managed to experience Aug 02 11:19:49 there it still says no flasher Aug 02 11:19:58 well it links to the wiki page for flashers Aug 02 11:20:15 yeah, so maybe I misremembered about the wiki not having flashers Aug 02 11:20:29 no iirc it also lacked flashers for a while Aug 02 11:20:37 dunno when that was Aug 02 12:22:43 I'm in the process of updating my Beaglebone black Debian image. When I try to edit the uEnv.txt file it asks for a password. I have searched for the password but havn't found it yt. Aug 02 12:23:15 Can anyone help please? Aug 02 12:23:57 how are you trying to edit the file? Aug 02 12:24:33 Give me just a minute please Aug 02 12:24:50 the default user ("debian") has sudo rights, so hopefully you're being asked its password ("temppwd" by default) Aug 02 12:25:56 sudo gedit uEnv.txt Aug 02 12:26:32 ok I'll try that thanks!! Aug 02 12:36:27 Thanks Matt, that password worked great!! Aug 02 12:51:08 hai,i am using beaglebone black,i connected BBB to my PC via USB-OTG.After my board get completely up,i can able to interact with BBB via /dev/ttyACM0.if i also want to see the bootlog from MLO onwards through the same interface(USB-OTG not via USB-SERIAL)what should i have to do??please help me.anything i have to enable in u-boot??)) Aug 02 12:56:01 jagadeesh: you need a TTL serial port adapter. FTDI style are the most common Aug 02 12:56:34 http://elinux.org/Beagleboard:BeagleBone_Black_Accessories#Standard_FTDI_Cable Aug 02 12:57:06 something like a bus pirate will also work Aug 02 13:05:00 NishanthMenon: morning! I've seen your name in the context of AVS so I was wondering if you can clarify something for me... I'm involved with a project that uses the omap5. The datasheet is very clear that AVS should always be used during normal operation, and is required for OPP_OD. However, this isn't being done and in fact the efuse registers that are supposed to contain the AVS voltages read as zero. Aug 02 13:05:06 I'm not really sure how to feel ... Aug 02 13:05:09 ... about this Aug 02 13:05:44 (yuck, how did that get split up twice...) Aug 02 13:06:14 the omap5 uEVM does seem to run happily at 1.5 GHz Aug 02 13:06:28 zmatt, that wont be good -> unless you are using extreamely old device, you should'nt get such samples. the other possibility is that the efuse offsets were changed (and i never got informed) Aug 02 13:07:00 I'm pretty sure these were supplied as "production" parts Aug 02 13:07:15 as much as omap5 can ever be considered to have been in production anyway Aug 02 13:07:25 zmatt, i was'nt informed - wont be new since I am not involved in older SoCs anymore, so probably some support team guys have missed that detail to check. i'd contact your TI contact on it Aug 02 13:07:27 (is it? its status is really unclear) Aug 02 13:08:03 zmatt, o5 was a pretty good chip when i worked on it at least. w.r.t status and industry it is targetted, i'd suggest contacting your ti guy about it. Aug 02 13:08:34 zmatt, dra7 / am57 is a derivative from omap5 just fyi.. Aug 02 13:08:40 it's a shame it isn't still supported alongside the am57xx for power-constrained applications... there's a ton of overlap between the two Aug 02 13:08:47 yeah I know Aug 02 13:09:06 zmatt, cough, changing priorities? Aug 02 13:10:00 zmatt, anyways, on your question, you should mandatorily contact your TI FAE / support person about it for sure. device without class 0, is not really safe for production - generically speaking Aug 02 13:10:38 yeah I know. although I don't quite get what TI has been doing with it. It seems that they were already reorienting to more broadmarket applications with es2 since opp50 support was dropped. a new revision of the TRM was released less than a year ago. and now the product has completely disappeared from the site :/ Aug 02 13:10:43 ok Aug 02 13:11:10 zmatt, just fyi -> the same register has abb info as well -> so you should make sure both are correctly populated for each offset. Aug 02 13:13:00 I did spot those u-boot commits a while ago. good point though, lemme see what those registers say Aug 02 13:13:46 * NishanthMenon gets back to the world of simulators Aug 02 13:14:07 almost just like the real thing ;) Aug 02 13:14:35 I'm gonna check u-boot to see if it knows more about those registers than the TRM does Aug 02 13:14:41 zmatt, much better since i can capture waveforms inside the SoC ;) no better way to understand an arch ;).. ttyl.. Aug 02 13:15:44 true Aug 02 13:15:55 thanks for the info Aug 02 13:21:41 how to see boot log of BBB over USB-OTG Aug 02 13:23:23 jagadeesh: can't Aug 02 13:23:54 get a usb-serial adapter Aug 02 13:23:58 jagadeesh: get a serial cable like georgem said. Any serial cable will do Aug 02 13:24:00 plug into debug UART Aug 02 13:24:02 profit Aug 02 13:25:31 i can able to interact after the board get completely up via /dev/ttyACM0(USB-OTG) Aug 02 13:28:21 jagadeesh: yes you can Aug 02 13:29:52 but when i reset the board that ACM0 interface is getting disconnected,beacuse kernel is making that interface to get up Aug 02 13:31:12 jagadeesh: To interact with the bootloader you need a cable that connects to the UART RX/TX lines on the am335x. Aug 02 13:31:52 yes Aug 02 13:32:51 but if we make u-boot to enable the USB-OTG console..then i think we can interact with bootloder Aug 02 13:33:19 but what should we have to enable i dont know Aug 02 13:35:43 someone help me Aug 02 13:35:44 jagadeesh: even if it can be done, you're wasting your time. get a cable for $5 Aug 02 13:36:00 jagadeesh: begging won't help Aug 02 13:36:02 not a matter of money Aug 02 13:36:56 but its my company requirement Aug 02 13:37:22 NishanthMenon: uhhhh, vcores->mpu.efuse isn't being initialized on omap5, which means that https://lists.denx.de/pipermail/u-boot/2016-April/252490.html actually disabled ABB on omap5 Aug 02 13:37:37 as far as I can tell Aug 02 13:37:48 jagadeesh: well if it's a company requirement there may be another option Aug 02 13:37:57 jagadeesh: http://elinux.org/U-boot_musb_gadget_support Aug 02 13:38:00 jagadeesh: have the company buy the $5 USB serial cable. Aug 02 13:39:29 georgem: "Next steps: Merge in mainstream?" Aug 02 13:40:18 alan_o, yeah... I dunno. looks like there is MUSB gadget support in u-boot mainline. This is the only write up I can find on it though. Aug 02 13:40:19 It's interesting, but it's still not going to get you everything, like all the stuff that happens before a usb-otg driver can be loaded. Aug 02 13:40:46 yeah, of course. seems like a waste of time but if he wants to waste his time... Aug 02 13:41:28 I was saying it for his benefit rather than yours :) Aug 02 13:41:44 :) Aug 02 13:42:45 jagadeesh: yeah, I mean it just is what it is. Welcome to embedded systems: plug in your serial cable here. It's going to be the same on every platform unless you have an FTDI chip (or equivalent) soldered down to the main board. Aug 02 13:44:05 thanks Aug 02 13:45:06 zmatt, crap Aug 02 13:45:27 also, that register is undocumented in the TRM :-/ Aug 02 13:45:41 it is definitely non-zero though Aug 02 13:45:49 zmatt, i am pretty sure i'd have tested it with a scope at that time.. Aug 02 13:46:12 or i think i did.. a lot of things have happened in a year :( Aug 02 13:47:31 oh boy, upstream has a changed a bunch Aug 02 13:47:46 I'm looking at ti u-boot though Aug 02 13:48:09 digging through history to see if I can find anything odd Aug 02 13:48:50 arch/arm/mach-omap2/clocks-common.c ->mpu.efuse.reg[opp] Aug 02 13:48:50 you seemed to have added the vcores.*.efuse definitions for dra7 in may 2013 (commit 18c9d55ac611) Aug 02 13:49:50 omap5 never had them it seems Aug 02 13:55:33 zmatt, arch/arm/mach-omap2/omap5/hw_data.c Aug 02 13:55:52 that's where I was looking Aug 02 13:57:20 zmatt, yep, the efuse are missing.. /me kicks self. sorry about that Aug 02 13:57:59 even the non-public TRM fails to mention the existence of that efuse register :-/ Aug 02 14:18:35 zmatt, sorry got distracted, mpu.efuse.regp[OPP_NOM] and that set needs to be populated Aug 02 14:23:24 it seems abb does get enabled somewhere along the way though Aug 02 14:23:31 zmatt, if i do manage to get my hands on a newer omap5uevm, I can try and do a patch, but i dont even know at this point who is dealing with o5 at this point - i have to search. if you dont mind posting patches fixing this stuff in upstream, it'd help. Aug 02 14:25:08 uhh, I can try I guess... I'm not hugely comfortable meddling with voltage-related code I'm unfamiliar with and which involves undocumented registers :-/ Aug 02 14:26:06 I can get you root shell access on my omap5 if it's of any use Aug 02 14:26:37 zmatt, hehe.. /me quips: you dont mess with voltage stuff without an oscilloscope handy :P Aug 02 14:27:11 and something to break out that funky high-density power measurement connected on the uEVM Aug 02 14:27:12 * NishanthMenon trying to find irc log so that I can form an internal email Aug 02 14:27:38 I can get you that Aug 02 14:27:54 zmatt, it might be easier if i find a board :) i probably have put a scope on it, i should be able to find the point to measure without looking at schematics :P.. jk Aug 02 14:28:46 zmatt, http://logs.nslu2-linux.org/livelogs/beagle.txt has it Aug 02 14:28:51 those work again? Aug 02 14:29:00 zmatt, seems like Aug 02 14:29:17 indeed Aug 02 14:30:40 uevm is still for sale by ti so presumably shouldn't be hard to obtain Aug 02 14:31:15 zmatt, hehe, + point of working @ ti, i can ask inside TI to borrow a TI board :P Aug 02 14:32:19 zmatt, i will keep you posted, i just dropped a note trying to find a board that i can borrow, but thanks a ton for reporting it and apologies on my shoddy patch Aug 02 14:32:27 meh, I had hoped to find maybe working avs stuff for omap5 in the 3.8-ti kernel (which seems to be the last kernel that still really targeted the omap5), but it doesn't appear to have it Aug 02 14:33:48 ohey Aug 02 14:34:26 the avs class 0 registers being zero seems to be an artifact of the omap5 device they put on the uevm Aug 02 14:35:03 zmatt, yeah the older samples were class 1.5, where the class0 was'nt yet in the picture Aug 02 14:35:17 those are supposedly scrapped Aug 02 14:35:30 https://docs.google.com/spreadsheets/d/1lWtGYUQ54L8h7T2UKIoA2tAxaWHFIYmQijVr42mmVe0/edit?usp=sharing Aug 02 14:35:56 zmatt, TI does'nt permit me to access google docs :( Aug 02 14:36:05 at least from inside corporate network :D Aug 02 14:36:27 interestingly, std_fuse_opp_vdd_mpu_2 reads as zero on the production omap5 Aug 02 14:40:31 NishanthMenon: https://liktaanjeneus.nl/omap5-efuse.html Aug 02 14:41:12 zmatt, rofl... thanks for finding a WA ;) Aug 02 14:41:42 :) Aug 02 14:42:34 that's what corporate policies are for right... working around them. at least I didn't have to setup an IP-over-email tunnel like someone I know who only had email access at work ;) Aug 02 14:43:17 zmatt, hehe.. ok i really got to get back to my palladium stuff ttyl Aug 02 14:43:56 yeah I have other stuff also Aug 02 15:43:11 * Spidler cries Aug 02 15:43:38 zmatt: Do you know if the mmc block devices are stable these days? Aug 02 15:43:42 Or anyone else Aug 02 15:44:03 I'm having slight issues with renumbering between mmcblk0 and mmcblk1 depending on the kernel in use. Aug 02 15:44:34 It would be TOTALLY AWESOME to use UUID or Label, but squashfs doesn't support that, so "the proper way" is "Hah-ha-ha-NO!" Aug 02 15:50:18 Spidler: latest kernels from rcn should have my patch to stabilize the block numbering Aug 02 15:52:47 zmatt: It's based on the mmc device hooked up now then? so mmc1 (internal) will always label it as mmcblk1 ? Aug 02 15:56:52 it's based on the /aliases/mmc* entries in DT Aug 02 15:57:25 those numbers are reserved for both the mmc host device and the mmcblk device Aug 02 15:58:48 any other mmc devices that might show up will get numbers allocated above those reserved. same goes if an mmc host somehow manages to spawn more than one mmcblk device Aug 02 15:59:06 Hmmm Aug 02 15:59:18 So I can in theory renumber them via the DT entries too? Aug 02 15:59:40 :( Aug 02 15:59:47 Well, that'd be cheating, but easy. And I honestly don't really care as long as what I have now is _stable_ and will stay that way Aug 02 16:00:02 mmc0 should be sd card, mmc1 is eMMC Aug 02 16:00:11 The problem is when kernel FOO needs to tell what device will be root for kernel BAR Aug 02 16:00:23 note that this also matches the hardcoded numbering in u-boot Aug 02 16:00:29 so you really want to keep that Aug 02 16:00:32 ah. Yea Aug 02 16:00:41 it also matches the TRM Aug 02 16:00:52 Also good for me Aug 02 16:01:20 I don't mind overly, just need to figure out a way to do this backwards compatible for me Aug 02 16:01:28 Downgrading broke, which was painful Aug 02 16:01:28 NishanthMenon: I finally get it now... the efuse registers present on the uEVM are for smartreflex, while the production omap5 parts only have class 0 Aug 02 16:01:39 so no smartreflex for us then I guess Aug 02 16:02:04 Spidler: well the old numbering was unstable, so any code relying on it was technically already broken ;) Aug 02 16:02:17 it's only with the patch that numbering can be safely used Aug 02 16:03:07 note that the patch has also been backported to 4.4 series Aug 02 16:03:39 if those are a concern Aug 02 16:16:29 zmatt, smartreflex is AVS technology ;) avs 1.5 as i mentioned earlier Aug 02 16:16:48 zmatt, i think i have a new board now, have to find time to do a patch and verification Aug 02 16:17:06 1.5 is basically just an optimized form of 3 right? basically caching the results of the SR cores Aug 02 16:17:33 zmatt: Yea, that's where I noticed it, something between 4.6 and onwards, and then in the backports... Aug 02 16:22:50 NishanthMenon: but yeah with smartreflex I meant the stuff with hw sensors... the terminology seems a bit hazy since i've also seen the word smartreflex used in the context of class 0, which makes no sense to me Aug 02 16:25:09 zmatt, understood. Aug 02 16:30:33 so I guess those have been definitively defeatured in production omap5, if the necessary sensor calibration values are missing from efuse Aug 02 16:32:04 odd that the uEVM has them but not class 0 values... that means u-boot actually has to get the abb values from a different register too Aug 02 16:32:13 inconvenient Aug 02 16:43:35 * Spidler sighs Aug 02 16:43:44 Seriously, the BBB's ethernet still bugs me to this day. Aug 02 16:49:05 Spidler: phy borked at power on? Aug 02 16:49:21 zmatt: Mmmmm Aug 02 16:49:25 or what? Aug 02 16:49:30 Nah, exactly that Aug 02 16:49:41 so, apparently removing two caps solves the problem Aug 02 16:50:03 And the fact that I right last month would need a similar board but with dual ethernet ports :) Aug 02 16:50:10 someone ran an automated test setup and compared before/after the procedure Aug 02 16:50:18 Hmmm. That'd be interesting Aug 02 16:50:34 Not that I really think it's feasible for me to do hardware adjustments on all my boards :( Aug 02 16:50:54 "After removing C24 and C30 we ran 1000 power cycles and had a 100% Aug 02 16:50:59 success rate - i.e. board booted and phy detected every time." Aug 02 16:51:05 "We ran the same test on an unmodified board and had a failure rate of 54/1000." Aug 02 16:51:24 Mmm. Seems in line with our failure rate too., Aug 02 16:51:47 I'm assuming those are the caps on the reset line Aug 02 16:51:52 * zmatt checks Aug 02 16:52:14 presumably only C24 matters Aug 02 16:52:32 Hmmm. Aug 02 16:52:40 zmatt, i think i might get a socketted board as well for o5, and I am talking to product engineering folks as well to get details on efuse on production board. anyways... just fyi. eta: maybe by this weekend or next if i dont get dragged to another fire fighting Aug 02 16:52:43 I'll have to write this down for later Aug 02 16:53:08 Right now, I'm wondering if I can find the hardware supplier that makes the Netgate SG-1000 Aug 02 16:53:10 NishanthMenon: cool! Aug 02 16:55:27 zmatt, you can see me kick myself for not closing properly when I did indeed have time to do things right... :( Aug 02 16:55:34 NishanthMenon: in case you're wondering who'd use an omap5 anyway... it's a community-driven handheld, successor to the openpandora -> pyra-handheld.com Aug 02 16:55:48 zmatt, yes, i am aware of it, and I guessed this was the source Aug 02 16:55:49 I kinda ended up involved by accident :) Aug 02 16:55:53 ah cool Aug 02 16:56:37 zmatt, :P if it was'nt a community board, i'd have punted you to e2e.ti.com :P -> hey, if i spend personal time on this, it should be for others who spend personal time on passionate projects :P Aug 02 16:57:44 anyways, back to palladium :( Aug 02 16:58:10 yeah, ditto... I'm also putting in the volunteer work because the project deserves it Aug 02 17:00:37 Hello, does anyone know which Debian (or Ubuntu) version is the target OS for the beagleboard x15? Aug 02 17:03:50 there are jessie and stretch images I think Aug 02 17:05:58 monthly jessie snapshots and weekly stretch snapshots -> http://elinux.org/Beagleboard:BeagleBoneBlack_Debian Aug 02 17:11:41 any C/python/JS coder? Aug 02 17:12:00 \o Aug 02 17:12:33 I see now. Since Jessie is going into LTS, it makes sense to start working on Stretch. Aug 02 17:15:06 I've been using stretch already for a long time already Aug 02 17:15:19 ugh, sentence fail Aug 02 17:24:31 I need to chose one for the near term. Is there a reason _not_ to go with Stretch? Aug 02 17:24:57 I don't know, I've been using stretch for a long time already :) Aug 02 17:26:11 One of my main concerns is the HDMI support. I need to use the X Window system. Aug 02 17:26:31 that's really not a concern Aug 02 17:26:50 it's also not really dependent on the distro but mostly the kernel Aug 02 21:51:21 Has anyone built a recipe for scipy in Yocto or Buildroot? I know asking here is a longshot. Aug 03 00:01:53 >[ 537.103409] omap_hsmmc 48060000.mmc: prep_slave_sg() failed [ 537.103418] omap_hsmmc 48060000.mmc: MMC start dma failure Aug 03 00:02:31 Anyone know what to do about this? Running an OS from the SDCard. I/O errors and remounting root read-only follows from this error. Aug 03 00:12:45 Linux version 4.4.68+, running BeagleBone Black specifically. Aug 03 00:22:17 Guess it might be worth trying out the 4.9 kernel as per https://groups.google.com/forum/embed/?place=forum/beagleboard&showsearch=true&showpopout=true&showtabs=false&hideforumtitle=true&parenturl=http://beagleboard.org/Community/Forums#bone_forum_embed#!searchin/beagleboard/MMC$20start$20dma$20failure/beagleboard/L7piqfHiyO8/eBsLl1eDAQAJ . Will try it later, gotta run now. Aug 03 00:22:30 That's, uh, quite a link there, Google. Aug 03 00:25:37 Hmm, might also be solvable with https://groups.google.com/forum/embed/?place=forum/beagleboard&showsearch=true&showpopout=true&showtabs=false&hideforumtitle=true&parenturl=http://beagleboard.org/Community/Forums#bone_forum_embed#!searchin/beagleboard/MMC$20start$20dma$20failure/beagleboard/xpvjJAHm8mU/V6y1GY4mDgAJ Aug 03 01:11:56 Well, I was able to rsync successfully after setting vm.min_free_kbytes to 16384. I'll add it to /etc/sysctl.conf. Hopefully it doesn't happen again. ...I'll actually afk now, honest. Aug 03 02:22:35 argh that fucking dma error... I've been struck by it too, but I don't have a reproducible test case Aug 03 02:47:27 anyone here using ROS on a beaglebone? **** ENDING LOGGING AT Thu Aug 03 03:00:01 2017