**** BEGIN LOGGING AT Mon Oct 07 02:59:57 2019 Oct 07 07:13:10 Is available BeagleBone AI step model? Oct 07 09:11:48 Hi Oct 07 09:25:20 Mertkan: tip: don't expect any response to 'hi". if you have a question, just ask it, and be sure to have patience. plenty of knowledgeable people here, but most only glance at chat occasionally Oct 07 09:30:25 zmatt: your wise advices should be in the topic, IMHO. seems they are not reported in the mentioned links (at least they are not easily reachable, if they are somewhere there) Oct 07 09:31:05 I've said many times that something like that should be on the chat page Oct 07 09:31:11 the links are mostly useless Oct 07 10:36:31 hi I have the beagle bone AI, I reimagined it and now I cant find the IP address(the one on the doc wont work) and it won't show up as a mass storage device has anyone had this problem before Oct 07 10:37:07 austingoodall: which image did you use? Oct 07 10:37:40 the latest one in the start up guide = Debian 9.9 2019-08-03 4GB SD LXQT Oct 07 10:38:31 what sort of led activity are you seeing? does it seem to boot normally? Oct 07 10:39:00 so, you turned the sd card into a flasher card to flash onto eMMC I presume? Oct 07 10:39:49 yea 4 out of the leds are constantly on and yea it seemed to have worked according to the start up guide Oct 07 10:40:41 4 leds on sounds bad Oct 07 10:41:13 do you mean the flasher ended with 4 leds on, or do you mean right now it ends with 4 leds on? Oct 07 10:42:28 I was gonna say normally when flashing is done the beaglebone turns itself off... but I just realized that might not be true for the AI Oct 07 10:42:38 this is my first time using beaglebone, it started off with 4 leds on after power up after removing the sd card and it's been like that since this has been doing it from friday Oct 07 10:43:09 how did the flashing procedure end? Oct 07 10:46:24 the board shut down, I removed the sd card and powered up Oct 07 10:46:44 hm ok Oct 07 10:47:01 and now it just shows the four leds going on in sequence and that's it? Oct 07 10:47:37 no the 4 led are static they don't change Oct 07 10:47:55 immediately after applying power? Oct 07 10:48:43 basically, I can't connect to it but it seems to stay that was from boot up Oct 07 10:49:20 way* Oct 07 10:50:07 I'm asking what happens exactly when you power up: normally the first thing you see (after the power led turning on) is the four user leds turning on in sequence (at least it's that way on beaglebones and the bbx15 so I'm assuming also on the bbai) Oct 07 10:51:37 also, just a quick check: did you use Etcher to flash the sd card, or if not, what did you use? Oct 07 10:52:30 yea they flash in sequence then static when fully booted I'm assuming, and yes I used etcher from beaglebone website Oct 07 10:52:57 the four leds turning on is not "fully booted", it's the very start of booting Oct 07 10:53:10 it's the bootloader locating a system to boot Oct 07 10:53:58 normally they then turn off, and then leds will start blinking in various activity patterns while it's booting Oct 07 10:54:30 should I power up again with the sd card in and see if that'll fix the problem Oct 07 10:55:38 I have trouble imagining how that could help if flashing completed the first time, but it doesn't really hurt to try Oct 07 10:57:07 im retrying it now, 2 user leds are flashing in sequence Oct 07 10:57:35 does that seem like its reflashing the eMMC Oct 07 10:58:02 I'm not sure what it's supposed to describe, but no it doesn't Oct 07 10:58:46 flashing is a back-and-forth ("knight rider" / "cylon") pattern across all four leds: 0 1 2 3 2 1 0 1 2 3 2 1 0 etc Oct 07 10:59:28 then I don't know what its doing, do you have any idea on how to factory reset the board Oct 07 10:59:35 by reflashing Oct 07 11:00:27 is one of the leds (led 0, the outermost I think) flashing in a heartbeat pattern? Oct 07 11:01:01 (might be the innermost actually, not sure) Oct 07 11:03:38 just to be clear, after booting from SD card the first time you modified /boot/uEnv.txt to turn the SD card into a flasher card? Oct 07 11:07:06 oh also I just realized you may want to ensure the BBAI is adequately cooled while reflashing (maybe aim a fan at it if you have one) ... it would be unfortunate if it turned itself off because it overheated rather than because it was done flashing :P Oct 07 11:07:54 oh he left, nice Oct 07 12:02:42 Hi everyone. I tried to run xen hypervisor on beaglebone black. I got a error while booting xen saying "CPU does not support virtualization". Have anyone tried to run xen on beaglebone black? Does beaglebone black support virtualization? Oct 07 12:03:03 dharani: the BBB doesn't have HW virtualization, BB X15 has Oct 07 12:03:08 (IIRC) Oct 07 12:06:46 LetoThe2nd Thanks for your reply. What does it lack for HW virtualization? BBB has armv7 based processor. I know armv7 supports hw vitualization. Oct 07 12:06:57 so does the BB-AI Oct 07 12:07:39 dharani: armv7 CAN support HW virtualization, but its not mandatory. Oct 07 12:08:05 the first core to support hardware virtualization out of the ARM Cortex-A series with ARMv7 architecture was A15 IIRC. Neither A8 nor A9 do support hardware accelerated virtualization to my knowledge. Oct 07 12:08:32 tbr: thats what i think too Oct 07 12:09:00 dharani: and i've never heard that the BBB would have LPAE, so its out anyways: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0438i/CHDCHAED.html Oct 07 12:11:25 plus, not mentioned here: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0438i/CHDCHAED.html Oct 07 12:14:38 How can i check this on the board? or for the processor i want? To check does it support virtualization extention or not? Oct 07 12:15:21 dharani: read the documentation. Oct 07 12:15:32 plus, reading /proc/cpuinfo (probably) Oct 07 12:19:30 Thank you both. Your answers are really helpful. Oct 07 12:29:57 i'm trying to adapt an old dt overlay working well on 3.8 kernels. i'm using BB-UART4-RS485-00A0.dts as an example (i need to enable an rs485 using a gpio for the re/de signal), but dtc seems to dislike the #includes: https://paste.debian.net/hidden/b1047a8e/ Oct 07 12:30:32 i have the headers for the current running kernel installed. what am i missing? Oct 07 12:30:37 correct, #includes are handled by the C preprocessor, not by dtc itself Oct 07 12:30:41 so you need a suitable makefile Oct 07 12:31:37 oh, i see. any pointer to shamelessly copy from? Oct 07 12:32:12 you could just clone the bb.org-overlays repo and place your file in src/arm/ and then do "make src/arm/YOUR-OVERLAY.dtbo" Oct 07 12:33:10 i was just reading the makefile from there. the way you suggested is probably the quickest to avoid adaptations Oct 07 13:09:54 LetoThe2nd: yeah the BBB definitely doesn't have LPAE/VE (I think every cortex-A that has LPAE also had VE?) Oct 07 13:09:57 *has Oct 07 13:10:36 zmatt: no idea, its neither my area of expertise nor of interest. Oct 07 13:11:50 what tbr said sounds right... A8 and A9 lack LPAE/VE and remaining cortex-A cpus (A15 and later) do have it Oct 07 13:17:10 ah and the A5 also lacks LPAE/VE Oct 07 13:54:01 also IIRC there was something easy to stumble upon with VE depending on the SoC being in "the right mode". I remember hrw trying to get it to work on his ARM Chromebook and it wasn't so easy. Oct 07 13:54:15 something about boot loader and secure domain foo Oct 07 14:03:01 on the am572x u-boot needs to do a secure monitor call to switch to hypervisor mode Oct 07 15:51:39 I am a newbie here. I do not know how to flash a beaglebone AI's DSPs, and if I can override the provided OpenCV engines with my own. Where can I find instructions to do that? Oct 07 15:51:49 Thank you in advance for the information./ Oct 07 16:29:43 ssb74: I don't really know the tools currently available for working with the DSPs, but I can tell you you don't "flash" them... code is loaded onto them at runtime Oct 07 22:57:30 Has anyone got pandas to work on the BBB or related boards? Oct 07 23:01:00 I was missing cython but I got cython and then received unlimited issues/errors when installing pandas w/ pip3. Oct 07 23:02:03 which errors ? Oct 07 23:02:26 Oh. Oct 07 23:02:55 Let me set up things again. This will take a long time. I have to use this command to recreate the errors: pip3 install pandas. Oct 07 23:03:12 pandas takes forever to download w/ pip3. Oct 07 23:03:51 you can tell pip another download source if you're not satisfied with the default one Oct 07 23:03:59 like, ask to download from github Oct 07 23:04:04 Oh. Oct 07 23:04:07 Okay. Oct 07 23:04:45 you mean. git clone https://xxx.xxx.xxx.git and then go into the directory and make or whatever? Oct 07 23:04:55 no Oct 07 23:04:59 Oh? Oct 07 23:05:00 give the git path to pip Oct 07 23:05:04 Oh. Oct 07 23:05:15 there is some doc about that Oct 07 23:05:18 Okay. I will figure out how to do that... Oct 07 23:05:26 but it won't help with the errors Oct 07 23:05:30 just with dl speed maybe Oct 07 23:05:40 Oh. Oct 07 23:05:41 Got it. Oct 07 23:06:22 Would you like to see the errors anyway or should i just try your idea? Oct 07 23:16:26 generally you want the released version, not the work-in-progress version from github Oct 07 23:16:52 Okay. Oct 07 23:17:39 and if you want feedback on errors, you're guaranteed to not get any until you share them (via e.g. pastebin) Oct 07 23:17:44 :P Oct 07 23:18:26 I just figured out that I can use: pip3 git+https://www.github.com/blah/blah. Oct 07 23:18:27 Okay. Oct 07 23:18:48 I will wait until the feedback from my pip3 install pandas comes up again. i just tried the whole scenario again. Oct 07 23:18:57 it will take some time. Oct 07 23:21:35 Do you think I should have installed pandas in a closed off env instead of system wide? Oct 07 23:22:15 I'm assuming that by "system wide" you actually mean "for the 'debian' user" Oct 07 23:22:25 (i.e. you're _not_ using sudo) Oct 07 23:22:37 Right. Oct 07 23:22:49 for debian and by not using sudo. Oct 07 23:22:54 and I see no reason to use a venv really Oct 07 23:23:03 i.e. w/ pip3. Oct 07 23:23:06 Okay. Oct 07 23:23:20 No venv for this install. Okay. Oct 07 23:23:36 I figured this may be some of my concerns if this makes sense... Oct 07 23:23:48 having said that... why not just do sudo apt install python3-pandas Oct 07 23:23:55 Aw! Oct 07 23:24:02 That might just work, too. Oct 07 23:24:07 it'll probably be an older version, but it'll be a lot quicker to install and should Just Work Oct 07 23:24:13 Okay. Oct 07 23:24:21 python3-pandas, okay. Oct 07 23:24:57 I have unmet dep. Oct 07 23:25:01 Off to test it out. Oct 07 23:25:27 that's another advantage of using apt: it will take care of deps Oct 07 23:26:32 although pip3 should do so too, for deps that are python packages anyway Oct 07 23:26:52 I also have held broken packages. Oct 07 23:27:04 uh oh, how did you manage to do that? Oct 07 23:27:11 python3-pandas-lib Oct 07 23:27:20 can you please just pastebin the full output? Oct 07 23:27:21 No installation candidate. Oct 07 23:27:23 Sure. Oct 07 23:28:47 https://pastebin.com/gYuW2f0j is what I did and what was returned. Oct 07 23:29:26 wtf Oct 07 23:29:45 maybe git clean -xdf? Oct 07 23:29:45 can you try doing "sudo apt update" and then retry sudo apt install python3-pandas Oct 07 23:29:50 Sure. Oct 07 23:30:02 none of this has anything to do with git Oct 07 23:30:08 Okay. Oct 07 23:30:44 and it doesn't make sense to manually try to add the dependencies apt already found to an explicit apt command Oct 07 23:30:53 I listened to the sites of those packages. They just said to use sudo apt install matplotlib and so on. Oct 07 23:30:54 Okay. Oct 07 23:31:08 Not python3-matplotlib Oct 07 23:31:10 it already found those dependencies, it would have installed them if it didn't encounter a problem Oct 07 23:31:16 Okay. Oct 07 23:32:06 I also have failed building wheel errors using pip3 and failed cleaning errors using pip3. Oct 07 23:32:26 I have no idea what you're talking about, please don't try to summarize errors, just pastebin them in full Oct 07 23:32:36 Okay. Oct 07 23:33:08 also trying to install something via pip3 is mutually exclusive with trying to do so via apt Oct 07 23:33:31 speaking of, how is that going? did apt update fix the problem or does it persist? Oct 07 23:34:01 ohh I see the problem Oct 07 23:34:07 https://pastebin.com/RDf3V0Cg Oct 07 23:34:16 Please hold. Oct 07 23:34:21 I will see if it worked. Oct 07 23:34:41 this is weird, for some reason the armhf version of python3-pandas-lib Oct 07 23:34:47 is missing in debian stretch Oct 07 23:34:51 Right! Oct 07 23:35:21 Same error but you are right. python3-pandas-lib has no release candidate. Oct 07 23:35:26 w/ apt. Oct 07 23:35:29 very strange Oct 07 23:35:42 See. Not me. Not my doing. Oct 07 23:35:56 This is that 1% time. Oct 07 23:36:25 okay, it seems the version in apt is extremely ancient anyway, so let's get back to pip3 Oct 07 23:36:34 Okay. Oct 07 23:36:48 you still haven't pastebinned the output of that Oct 07 23:36:55 Yes. Oct 07 23:36:56 I did. Oct 07 23:37:03 oh then I'm blind Oct 07 23:37:17 https://pastebin.com/RDf3V0Cg Oct 07 23:37:35 But it is not everything. Oct 07 23:37:53 The pip3 command is still in its process of installing. Oct 07 23:38:15 set_: sudo apt install python3-pip python3-setuptools python3-wheel Oct 07 23:38:24 Aw! Oct 07 23:38:25 then try the pip3 command again Oct 07 23:38:28 Okay. Oct 07 23:39:17 I was missing python3-wheel. Oct 07 23:39:43 I should have known. Oct 07 23:43:31 just fyi, pandas takes a long time to install on pip3. Please be patient. Oct 07 23:43:53 I imagine it will yes Oct 07 23:45:09 https://pastebin.com/F2T867SE are the errors for the python3-wheel command w/ apt. Oct 07 23:46:26 wtf, can you give more context please? Oct 07 23:46:44 oh ew it's installing "recommended" (aka "useless") packages Oct 07 23:47:20 Oh. Oct 07 23:47:24 context. Oct 07 23:47:47 Okay. sudo apt install python3-wheel gave me that error from that paste. Oct 07 23:48:09 try: sudo apt -o Apt::Install-Recommends=0 install python3-wheel Oct 07 23:48:18 My pip3 and setuptools are already the newest versions. Oct 07 23:48:24 and yes but I meant give the full output, not trimmed to the part you think matters Oct 07 23:48:26 Okay. Oct 07 23:48:36 Okay. Oct 07 23:48:41 because that error looks extremely bizarre to me Oct 07 23:49:21 it's like the download vanished after it was done downloading Oct 07 23:50:38 That command worked. I have python3-wheel installed via apt now. Oct 07 23:50:53 I am using a cached version of pandas to install via pip3 now. Oct 07 23:53:16 I am going to get a bit of food. Peanut Butter and brb. I think the install should be finished by then. Brb. Oct 08 00:00:58 It is hanging right now. It may be a bit. Oct 08 00:03:01 I finally found that armf is not availble on the arch side for python3-pandas-lib. Oct 08 00:03:38 you mean what I literally said 30 minutes ago? Oct 08 00:04:17 @zmatt: Right! Oh and what does sudo apt -o Apt::Install-Recommends=0 do? Oct 08 00:04:30 I have not learned that one yet. Oct 08 00:05:02 that option suppresses the automatic installation of packages "recommended" by the package(s) you're installing, which are often useless crap Oct 08 00:05:11 Oh. Oct 08 00:05:12 Nice! Oct 08 00:05:53 yeah I have that enabled system-wide Oct 08 00:06:04 Oh. Oct 08 00:06:06 Okay. Oct 08 00:06:21 you mean, for a particular user? Oct 08 00:06:30 (but I do pay attention to the list of packages APT mentions are recommended to check if any of them do seem to be important) Oct 08 00:06:50 there's nothing user-specific about APT, it manages system packages Oct 08 00:06:57 Oh. Oct 08 00:07:00 Okay. Oct 08 00:07:03 Understood. Oct 08 00:07:31 So, you made a bash script for apt? Oct 08 00:07:38 no Oct 08 00:08:25 Okey. I guess it is not important now. setup.py is still in the bdist_wheel portion. Oct 08 00:09:05 yeah I'm sure it'll take a while to compile stuff Oct 08 00:09:37 I noticed some keyring stuff when I installed python3-wheel. Oct 08 00:09:40 It was not alone. Oct 08 00:09:55 It had two keyring packages. Oct 08 00:10:05 while using the -o Apt::Install-Recommends=0 option? Oct 08 00:10:09 it should have prevented that Oct 08 00:10:15 Nope! It did that. Oct 08 00:11:01 I already used the comamnd clear. Oct 08 00:11:05 I am sorry. I cannot go back. Oct 08 00:11:43 comamnd = command Oct 08 00:12:06 ? Oct 08 00:12:20 python3-keyring python3-keyrings.alt python3-xdg Oct 08 00:12:21 That! Oct 08 00:12:37 That is what happened. Oct 08 00:12:44 ?? Oct 08 00:13:16 Forget it. Oct 08 00:13:25 That was just the recommended packages. Oct 08 00:13:34 I was wrong as usual, again. Oct 08 01:05:45 error: command 'arm-linux-gnueabihf-gcc' failed with exit status 4 <<--- this was the error after it failed to build pandas. Oct 08 01:06:08 There is a longer list of errors but I will need to wait until the end to see what happens. Oct 08 01:06:30 It is on numpy now. Oct 08 01:07:05 as usual that line means absolutely nothing out of context, full output is required Oct 08 01:07:10 Okay. Oct 08 01:07:58 Please stay patient. The numpy is trying to use setup.py bdist_wheel for installation. I will have to wait until the end of this process. Oct 08 01:08:19 Then, I can push out the paste. Oct 08 01:09:26 if you already have a compilation error, then the remaining output is probably irrelevant Oct 08 01:10:20 Oh. See, the output is a long list of errors. Oct 08 01:10:28 Not for the weak of heart. Oct 08 01:10:34 But. Oct 08 01:10:48 usually the first error is what matters Oct 08 01:10:51 I cannot show you this list b/c the numpy source is still installing. Oct 08 01:10:52 Oh. Oct 08 01:10:53 Okay. Oct 08 01:11:27 Well just wait for the debris to finish falling. Oct 08 01:11:37 Okay. Oct 08 01:12:17 Everytime I try to scroll up, the system brings me back down to the install section of my terminal. I think this is PuTTY's doing. Oct 08 01:12:39 probably a putty setting somewhere Oct 08 01:12:56 I got it. Oct 08 01:12:58 Please hold. Oct 08 01:13:06 but if stuff has already failed then there's probably no point in waiting and you can just cancel the rest with control-C Oct 08 01:14:10 https://pastebin.com/CEc87XVZ is the first error. Oh but things try to set up once the bdist_wheel is used. Oct 08 01:14:25 The installation process is ongoing even if bdist_wheel fails. Oct 08 01:15:50 I have had installations work and install even after the bdist_wheel has failed. It set ups and installs via setup.py after cleaning up the code w/ setup.py clean. Oct 08 01:15:51 Right? Oct 08 01:15:51 this is not an error, this is a random line of irrelevant output Oct 08 01:15:57 Hmmm. Oct 08 01:16:01 03:07 <@zmatt> as usual that line means absolutely nothing out of context, full output is required Oct 08 01:16:09 Okay. Oct 08 01:16:10 Okay. Oct 08 01:16:26 this is getting really tiresome Oct 08 01:16:38 Nevermind, forget it. Oct 08 01:16:53 do NOT try to extract something you think is relevant, since you clearly lack the ability to judge what is actually relevant Oct 08 01:17:01 full output only Oct 08 01:17:32 @zmatt: I did not have time to fully get all the data in the output. The PuTTY emulator keeps kicking me out of cut and pasting. Oct 08 01:17:35 I will try again. Oct 08 01:18:07 again, just cancel what it's doing with control-C (if you're sure an error has occurred) Oct 08 01:19:35 https://pastebin.com/UY9FZ3Ey <<--- I think pandas_lib is missing. Oct 08 01:19:45 That is the output just for pandas. Oct 08 01:22:37 arm-linux-gnueabihf-gcc: internal compiler error: Killed (program cc1) Oct 08 01:22:43 that's the actual error Oct 08 01:23:00 Oh. Oct 08 01:23:11 I read that. I went over it and did not think twice. Oct 08 01:23:22 i.e. a compiler bug Oct 08 01:23:26 Oh. Oct 08 01:23:28 Okay. Oct 08 01:24:53 it's also possible it ran out of memory Oct 08 01:24:59 Ut oh. Oct 08 01:25:01 Let me check. Oct 08 01:25:24 yeah that's almost certainly it Oct 08 01:25:47 I'm reading that it's very memory hungry to build Oct 08 01:26:13 Oh. So I need to clear my cache? Oct 08 01:26:15 so you're not going to get this compiled on a bbb .. Oct 08 01:26:19 Dang it. Oct 08 01:29:07 sudo apt autoclean? Oct 08 01:30:08 and w/out the pandas_lib package, I am stuck. No go. Oct 08 01:30:10 Dang it. Oct 08 01:48:09 well, numpy worked. Oct 08 02:12:18 Does it make sense to cross compile pandas on your big machine instead of the BBB? Oct 08 02:22:05 Yes. Oct 08 02:22:43 But...python3-pandas-lib is not available on armf arch. for some reason. I can try I guess. I am reading about it now. Oct 08 02:27:12 that could be complicated now that I think about it. If it has to compile it doing a pip style install will not work. Nothing worthwhile is easy I guess. Oct 08 02:31:04 using distcc might work Oct 08 02:35:48 distcc. Okay. Oct 08 02:36:17 I am scouring the logs now on github for related info. Oct 08 02:37:14 ?? Oct 08 02:40:33 I am in panda online at github looking for armf ideas. I figured it was b/c of the compilation. This is why people gave up on the armf support. Oct 08 02:40:58 Maybe people are not thinking there is a big market for pandas w/ armf. Oct 08 02:41:40 I actually have no clue. I typed in "armf" in their closed/open issues. Zero inquiries came back. Oct 08 02:41:44 this isn't a pandas problem, the armhf (not "armf") build is just missing in debian stretch, not in any later version of debian Oct 08 02:41:52 Oops. yep. Oct 08 02:41:57 Oh. Oct 08 02:41:58 ! Oct 08 02:42:30 So, people just did not bother compiling it correctly. Oct 08 02:42:38 for Debian 9? Oct 08 02:42:46 armhf! Oct 08 02:42:48 Sorry. Oct 08 02:43:30 I have no idea what happened there Oct 08 02:44:05 Me neither. I am getting to the bottom of it before people officially move on. Oct 08 02:44:30 you could consider using a debian buster image instead of a debian stretch image Oct 08 02:44:34 then it should be available Oct 08 02:44:42 Oh. Okay. Oct 08 02:44:47 THat sounds less exciting. Oct 08 02:44:57 But. You are right. I should. Oct 08 02:45:19 it definitely seems like the easiest option Oct 08 02:45:58 Right. I think for peace of mind, poking around might do me some good. I might erase this issue on github. I have not submitted it yet. Oct 08 02:47:04 submit an issue? where? with pandas? you should definitely not, it's not their problem Oct 08 02:48:32 Oh. I figured it was b/c of their way of compiling the source that it was not available for Debian Stretch. Oct 08 02:49:03 a bug report on the debian package might have been appropriate if stretch were still relevant, but it isn't... you'd simply be told to upgrade to buster, where the problem no longer exists Oct 08 02:49:04 github.com/pandas-dev/pandas. Oct 08 02:49:17 Okay. Oct 08 02:49:18 Fine. Oct 08 02:56:39 Does it matter what kernel I use w/ Buster to have pandas use? Oct 08 02:57:29 Cut off "to have pandas use? **** ENDING LOGGING AT Tue Oct 08 02:59:57 2019