**** BEGIN LOGGING AT Mon Jan 08 03:00:03 2018 Jan 08 11:52:46 hello. Jan 08 12:08:25 welcome, to the world of tomorrow! Jan 08 12:11:17 is that substantially different from the world of yesterday? Jan 08 12:12:22 Yes the world is different. I did not need to work yesterday Jan 08 12:13:26 fair point Jan 08 12:18:01 also, Futurama reference: https://youtu.be/aiwA0JrGfjA Jan 08 12:19:18 pah, I'll make my own world. With black jack and...well you know Jan 08 12:19:36 :) Jan 08 12:24:32 tbr: hm, I thought that was a quote from "futureworld" Jan 08 12:25:18 thinkfat: could be they referenced something themselves Jan 08 12:25:38 tbr: not unlikely with futurama Jan 08 12:25:45 tbr: also, 'happy new year' Jan 08 12:46:39 thinkfat: thanks, and a happy new(?) fear to you too! Jan 08 12:49:58 Anyone got experience with Xenomai on the BBB? I installed following the tutorial [here](http://www.cvra.ch/blog/2015/xenomai-installation-on-a-beaglebone-black), and everything seems okay except when I try to test it, it returns `Xenomai: /dev/rtheap is missing` Jan 08 12:50:23 this should mean that I'm not booted into the xenomai-patched kernel... but I'm not finding anywhere how to actaully choose what I boot into...? Jan 08 12:51:11 Also, the directions are somewhat dated. I did make sure to find and use more recent kernels etc., but I found out yesterday that "a few years" may well be completely obsolete :P Jan 08 12:55:39 sheim: xenomai... I haven't seen that being mentioned in any form, shape, or fashion for a long long time Jan 08 12:56:14 generally people with real-time needs either just use an -rt kernel (if that suffices) or stick the real-time parts into pruss Jan 08 12:57:37 and yeah, 2015 may as well be the neolithic age Jan 08 12:57:42 -rt kernel would be the preempt-rt right? My understanding is that they are working to having this integrated into the mainline linux kernel, but at the moment it's still a patch. Or did they already get there? Jan 08 12:58:53 zmatt: you missed an opportunity for a star wars reference, there Jan 08 12:59:06 it's still a patch afaik, but rcn builds -rt versions of both his -bone and his -ti kernels, so all you need to do is apt-get install one Jan 08 13:00:06 That's good to know. What's the difference between an -bone and -ti kernel? Jan 08 13:01:17 -bone is closer to mainline and more beaglebone-specific, while -ti covers all beaglebone/beagleboard variants (including the x15) and has more stuff that hasn't reached mainline yet Jan 08 13:01:45 e.g. remoteproc-pru is only available in -ti (while uio-pruss is available in both) Jan 08 13:01:55 Also, we've got some legacy code that's working with xenomai (on a desktop, not on the beaglebone), and I wanted to show that the beaglebone is a good alternative, that can still run our old code... hence still trying to get it running. (Although yeah, one of the reasons we're pushing BBB is to maybe move away from having an RTOS and just rely on PRUs for the RT part we need) Jan 08 13:03:03 well, xenomai is presumably still possible, but noone has put in the effort to maintain a turnkey solution for that Jan 08 13:04:13 also, as for your earlier question, the kernel version is selected in /boot/uEnv.txt Jan 08 13:04:29 Thanks, I'll try from there first then. Jan 08 13:05:12 normally when installing a kernel image package, /boot/uEnv.txt should get automatically updated though Jan 08 13:07:17 yeah, first uncommented line gives me `uname_r=4.9.49-ti-xenomai-r58`, which is what I was expecting... Jan 08 13:08:35 Incidentally... I got the Derek Molloy book and the Yoder/Kridner book, which are both from ~2015... is there an up-to-date reference you can recommend, or I guess the real option is to just try stuff and hang around here and on the google-groups huh? Jan 08 13:17:49 oh, there are still xenomai builds of recent kernels? or did you custom-build it? Jan 08 13:22:02 there are recent builds Jan 08 13:22:29 what about meltdown/spectre on BBBs? according to ARM, some action should be taken: https://developer.arm.com/support/security-update Jan 08 13:23:21 The tutorial linked before (cvra.ch), they use an old kernel. I just used `sudo apt-cache search xenomai | linux-image` and picked something newer that seemed reasonable Jan 08 13:25:36 *forgot a grep in there Jan 08 13:35:32 samael: meltdown is not applicable, spectre is still under review by ARM and probably very very hard to exploit on a cortex-a8 Jan 08 13:36:17 samael: regardless, workarounds for spectre are application-dependent Jan 08 13:36:56 samael: these vulnerabilities may also not be much of a concern on embedded systems in the first place, since they require that an adversary is able to execute code on your system to begin with Jan 08 13:46:03 zmatt: good, thanks. are there any official statement from bb.org? seems i can't find one Jan 08 13:48:53 I don't see any reason to expect one Jan 08 13:52:26 I mean, we're talking about systems where many people habitually run everything as root and casually use /dev/mem ... spectre wouldn't be high on my list of worries :) Jan 08 13:55:11 if you use a bbb as desktop system to browse the web (in which case I feel sorry for you anyway), then you have maybe marginally greater reason for concern, but even then I wouldn't really be worried at this time Jan 08 13:57:48 succinctly and to the point Jan 08 13:58:45 zmatt: regarding spectre, there's some patches on linux-arm mailing list that serve to invalidate the btb on switch_mm time (but that was always the case for CA8 anyway) Jan 08 13:59:05 zmatt: probably because the btb entries are not asid-tagged on ca8? Jan 08 13:59:27 zmatt: anyway there's one additional patch that invalidates the btb also on a fatal signal Jan 08 14:03:26 iirc btb invalidate is a nop on cortex-a8 by default Jan 08 14:06:03 yeah, it can be enabled via the aux control register, but this is a hazardous thing to do since then doing a btb invalidate-by-MVA writes the contents of an uninitialized debug register into the btb, potentially resulting in cpu lockup Jan 08 14:06:28 (erratum #687067) Jan 08 14:06:51 https://www.spinics.net/lists/arm-kernel/msg627034.html Jan 08 14:07:51 I don't think invalidate-by-MVA is useful, other than for setting a breakpoint ;) Jan 08 14:08:19 I wonder how much these patches affect performance on arm Jan 08 14:08:21 well, enabling btb invalidate requires verifying that invalidate-by-MVA isn't actually used anywhere Jan 08 14:08:57 Humpelst1lzchen: the ARMv8 patches a awful :-( Jan 08 14:09:20 thinkfat: any numbers? Jan 08 14:09:41 Humpelst1lzchen: there's no way to invalidate the btb on ARMv8, so they're doing tricks with entring the ATF and then switching the MMU off and on again Jan 08 14:09:48 /o\ Jan 08 14:09:50 wtf Jan 08 14:10:30 I hope this crap is conditional on a compile-time flag Jan 08 14:10:44 since I really don't care about spectre on our systems Jan 08 14:12:41 zmatt: I think the change goes together with a PSCI version bump so it's not going to be used with an unpatched ATF Jan 08 14:13:00 zmatt: but I really hope that stuff like that is conditional so that it's no even compiled in Jan 08 14:13:20 zmatt: because it's really, really ugly Jan 08 14:15:11 of course everything related to spectre/meltdown workarounds is inherently really really ugly Jan 08 14:15:32 I still think fuckwit_ would have been a much more appropriate prefix than kpti_ Jan 08 14:21:16 hi Jan 08 14:21:23 how u doin' Jan 08 14:22:34 doin' just fine, yo Jan 08 14:22:44 :P Jan 08 14:24:24 Morning Jan 08 14:24:35 is it somewhere manual how to configure uboot with support saveenv command? Jan 08 14:27:57 saveenv? Jan 08 14:28:03 is that a debian thing or BB thing? Jan 08 14:28:08 u-boot thing Jan 08 14:28:30 inisider: I think you can just enable it in your u-boot's .config file Jan 08 14:28:39 or using 'make menuconfig' Jan 08 14:30:34 zmatt: seems that there is no menuconfig target. is it possible?? Jan 08 14:32:33 I thought it was called menuconfig but maybe I misremember Jan 08 14:32:52 nope, make menuconfig works fine here Jan 08 14:37:06 thanks, seems that in uboot v2014 there was not menuconfig, for uboot v2017 it is. Jan 08 14:41:00 lol Jan 08 14:41:16 so... don't use a prehistoric u-boot version Jan 08 14:56:22 am i correct that there is no support of bbb in LK mainline? i need to get LK from git://github.com/beagleboard/kernel.git ? Jan 08 15:03:18 because i dont beaglebone_defconfig in LK mainline Jan 08 15:04:07 mainline supports bbb Jan 08 15:09:09 zmatt: which one defconfig than use? Jan 08 15:09:17 no idea, I don't use mainline Jan 08 15:09:39 the beaglebone kernels have various non-mainline features that I use Jan 08 15:09:58 e.g. much nicer support for gpio Jan 08 15:10:07 which one than u use? Jan 08 15:10:26 currently a 4.9-bone series kernel with some custom patches and custom config Jan 08 15:11:57 what is repo for it? Jan 08 15:12:21 you want to specifically customize the kernel? Jan 08 15:12:28 https://github.com/beagleboard/linux Jan 08 15:12:29 this one? Jan 08 15:12:29 normally you'd just install a prebuilt kernel package Jan 08 15:13:21 if you want to build a customized kernel, it's easiest to use rcn's build repositories... either https://github.com/RobertCNelson/bb-kernel for a -bone kernel or https://github.com/RobertCNelson/ti-linux-kernel-dev for a -ti kernel. in both cases, select appropriate branch for the series yo uwant Jan 08 15:13:43 thanks. Jan 08 15:52:23 zmatt: FWIW rcn has a xenomai 3.8 kernel on bb-kernel, and several on the -ti. I'm running the build_script he includes now... btw, is there a difference between building from script, or installing via apt-get install? Jan 08 15:57:36 Hi there Jan 08 15:58:26 how Beagleboard is different from Raspberry Pi Jan 08 16:01:35 less raspberries Jan 08 16:01:40 sheim: no idea Jan 08 16:02:43 Hi everybody Jan 08 16:02:58 Boss: BeagleBone has PRU (programmable realtime units), the RPi does not. BeagleBone is better suited if you're interfacing to hardware. And I think the RPi has a bigger user-base. Jan 08 16:03:19 Anyone here cross compiled QT5 for beaglebone? Jan 08 16:03:43 sheim: Boss already left Jan 08 16:04:20 some people are incapable of waiting more than 30 seconds for an answer Jan 08 16:04:29 ....and less raspberries. Yeah, sorry, looking at different screens at the same time. Jan 08 16:06:33 lame trolling attempt, thats all. Jan 08 17:14:18 /cl Jan 08 17:25:42 What is the correspoding debian version with Kernel 4.1.18 for BBB Jan 08 18:46:11 how to use buildroot for BBB Jan 08 18:47:22 build image, put on SD card, use, ... profit! Jan 08 21:56:55 Hi, I bought Beagle Board Black BBONE-BLACK-4G - Single Board Computer, AM3358 ARM Cortex-A8 MCU,4GB eMMC On-board Flash Storage, USB Interface, it works well with Element14 manufacturer, it does not work good with Embest Manufacturer, what are the differences? Jan 08 22:02:46 Hi, I bought this part at Newark BBONE-BLACK-4G - Single Board Computer, AM3358 ARM Cortex-A8 MCU,4GB eMMC On-board Flash Storage, USB Interface. It works well with Elements14, but does not works with Embest Manufacturer. Please advise. Thanks Jan 08 22:04:57 I have no idea what you're asking Jan 08 22:05:10 Matthew___: your question is a little unclear as to what you bought and what does not work Jan 08 22:06:50 Also "does not work" is not very diagnostic. more details on what fails, how, when, with what error messages, whould be good Jan 08 22:06:53 BBB with Elements14 manufacturer works well, BBB with Embest Munufacturer does not work properly Jan 08 22:07:33 Ah so different BBB from Different manufacturers? Jan 08 22:08:28 it should not matter which manufacturer you use, it's still the same board Jan 08 22:09:02 what do you mean by "does not work properly" Jan 08 22:10:00 zmatt: any issue with knock-off clones? Jan 08 22:10:16 Do you have email so I can ask engineer to email you what we have problem are? Jan 08 22:11:44 freemor: good point... I thought I recognized the name "embest", but I don't see them listed as beagleboard logo licensee Jan 08 22:12:45 It is in the outside box of the board? Jan 08 22:13:24 Matthew___: if you want to share some text from an engineer, use a paste site like pastebin.com and paste the link here Jan 08 22:13:52 given that the board might be a dubious clone, it might be useful to take high resolution photos of the board (both sides) and link those as well Jan 08 22:14:34 ultimately, if you have problems with embest boards that are not present when using element14 boards, you may need to take it up with embest **** ENDING LOGGING AT Tue Jan 09 03:00:01 2018