**** BEGIN LOGGING AT Mon Sep 25 03:00:01 2017 Sep 25 03:24:37 Hello. I am running a bone black rev B and I have just updated the eMMC with a recent kernel. I have a custom uEnv.txt on an SD card that TFTP loads RTEMS test code I am working with, Sometimes, say 10 out of 500+ runs the load falis and the board boots to Linux in the eMMC. How can I avoid this and have the TFTP board retry forever or reset? Sep 25 03:25:49 hard to say without knowing what exactly is going wrong, do you have a capture of the output on the serial console? Sep 25 03:27:12 my guess would be that the ethernet phy gets borked due to improper reset (this is a known issue), therefore tftp fails, and the boot script proceeds to fall back to booting from eMMC Sep 25 03:29:56 The reason for the TFTP failure could be related to the testing tool I have. I have a custom local TFTP server that serves RTEMS tests one after the other that is part of the RTEMS test tool. I should be to sort that side out. If the TFTP retries or the board resets then my test harness will recover. I was wondering how I can configure this. Sep 25 03:30:55 And yes the PHY could hiccup, I may have to power cycle the board to recover from a test failure if the test locks up the board. Sep 25 03:31:55 yeah, recovering the phy requires asserting the reset line or power cycling the board. (asserting the reset line has the benefit of guaranteeing the phy gets properly initialized this time) Sep 25 03:33:01 if you want to reset the board instead of booting from mmc, a simple trick would be to set the u-boot env var that contains the mmc booting command sequence to "reset" Sep 25 03:33:04 Being able to reset automatically is in the plans, I however need a suitable board etc. Until a networked power switch is working. This is part of an automatic tester. Sep 25 03:33:32 Ah OK, so if the TFTP fails the uEnv.txt script just continues? Sep 25 03:33:44 what does your uEnv.txt look like right now? Sep 25 03:33:59 uenvcmd=echo Booting RTEMS BBB from net; set autoload no; dhcp; set serverip 10.10.5.2; set tftpdstp 55567; tftpboot bbb/rtems.img; bootm; Sep 25 03:34:18 ohh, just add reset; at the end Sep 25 03:34:37 if bootm succeeds, it'll never reach the reset command Sep 25 03:35:18 Nice. I will give that a go. Thanks. I will report back after a test run which takes about 1hr with 530 or so tests. Sep 25 03:40:52 zmatt: thanks again, the board is now reseting on a timeout. Sep 25 03:43:37 that was a fast hour ;) Sep 25 03:44:19 I tested the timeout while waiting for the updated tests to build :) Sep 25 03:46:48 so, if you do run into problems with the phy getting confused, note that the reset can be asserted by pulling P9.10 low... that may be simple to arrange in a test setup Sep 25 03:46:59 power cycling can be done using P9.09 Sep 25 03:49:21 (and people have managed to fix the phy issue entirely by removing the large caps on the reset line) Sep 25 03:50:45 Do you have a circuit reference? Sep 25 03:50:54 for? Sep 25 03:51:13 The caps? Sep 25 03:53:00 this is the post that reported the success -> https://groups.google.com/d/msg/beagleboard/9mctrG26Mc8/HToW6Z-rAgAJ Sep 25 03:54:50 Thanks, that look good. Sep 25 03:55:26 note that removing all capacitance from the reset line might not be without risk... i.e. it might be more prone to pick up noise resulting in a spurious reset. also, when the board softresets the processor briefly drives the reset line low, and I'm not sure the phy would like that. (on an unpatched bbb the capacitance completely suppresses this pulse) Sep 25 03:56:49 in retrospect, they should never have used the processor reset line directly for the phy Sep 25 03:56:54 OK. I will see how I go. Once set up this kit will just perform the one task, testing. Sep 25 03:58:15 yeah it depends on what you're using the bbb for. e.g. we unfortunately do see this issue in our products in which we embed beaglebones, but it's rare enough to not be a huge deal Sep 25 03:59:49 I should add I see this issue on a Microzed board running a Zynq with a Marvel PHY and u-boot. A zynq custom boot loader never sees an issue with the same circuit. Sep 25 04:00:19 oh Sep 25 04:00:40 that's not the same issue then Sep 25 04:00:56 No saying there is a connection here. Just an observatoin. I should all add I have just moved to direct DC power on these boards. Sep 25 04:01:16 the phy issue I'm referring to is rather specific to the beaglebone, and not preventable or fixable using software Sep 25 04:01:44 Understood. I am sure it has been looked at in detail. Sep 25 04:01:55 or, well, a subset of messed up states of the phy can be worked around (this is already done currently), but not all Sep 25 04:02:41 it depends on how the phy gets messed up.... improper resets aren't quite consistent :) Sep 25 04:03:17 Yes and the power can change things Sep 25 04:04:46 it shouldn't matter how exactly the board gets powered, since that just affects the 5v supply while powerup of the 3.3v supply is sequenced by the PMIC Sep 25 04:06:00 I have gotten the impression though that external circuitry connected to the bbb can in some cases influence this problem, e.g. by partially charging up the 3.3v before the pmic begins to supply it, or by altering the capacitance on the reset line Sep 25 04:07:05 I have no capes for my testing Sep 25 04:08:01 note: I'm not sure whether this observation is true, I haven't collected any real statistics on it :) Sep 25 04:19:33 310/565 so far and no unwanted timeouts so it is working much better than before Sep 25 04:21:28 cheers :) Sep 25 07:38:05 Hello, I'm using beaglebone black with 7inch display from 4D systems (4DCAPE-70T) Sep 25 07:38:15 Final goal is to setup cross-compile environment with QtQuick support. Sep 25 07:38:33 Buildroot looked promising... But can't make display operational. Sep 25 07:39:19 Here I listed steps I made and result received on serial port: https://pastebin.com/9n7Ativt Sep 25 07:40:21 Am I missing some config in buildroot or kernel? Sep 25 07:41:10 My guess is that uEnv.txt is not configured properly, but I can't find the right one for buildroot and capemgr working together Sep 25 09:28:42 If I with to use https://github.com/beagleboard/u-boot in Build root, what is the suggested def config? Sep 25 11:12:51 a.m. time! Sep 25 11:13:39 What should the specs be for powering the BBB with an external power bank? Sep 25 11:13:50 I mean...does the power bank need specific specs? Sep 25 11:14:58 3.3v, yes. But, should there be any other things I should be aware of? Sep 25 11:17:47 Better use the barrel (5V input) connector for powering the device. Sep 25 11:18:14 Measure the current consumption with PSU, before using the battery. Sep 25 11:18:32 Then make sure that battery can support the current requirements of your setup. Sep 25 11:20:56 Take a look at page 2 of http://beagleboard.org/static/beaglebone/a3/Docs/Hardware/BONE_SCH.pdf Sep 25 11:21:21 TPS65217b have many different voltage outputs Sep 25 11:21:38 not only 3.3V... Sep 25 11:22:23 Okay. Sep 25 11:22:29 powering *directly* from a single lithium-ion cell (or several in parallel, but NOT in series) is certainly NOT for the faint of heart Sep 25 11:22:52 while powering from a managed power source like a "usb power bank" is easy and plug and play Sep 25 11:23:55 I have this power bank via usb I connect to the BBB with 2.1A and 2.4A. Sep 25 11:24:08 yes, that will work just fine Sep 25 11:24:15 Cool! Sep 25 11:35:04 If I was to use a 9v battery, what steps should I take to ensure working order? R = V/I? Sep 25 11:35:34 you don't Sep 25 11:35:38 use a 5v power source Sep 25 11:36:21 Okay but do people sell 5v power supplies? Sep 25 11:36:24 exactly Sep 25 11:36:32 yes, such as any usb power bank Sep 25 11:36:34 yes Sep 25 11:36:37 Oh. Sep 25 11:36:44 OTAY! Sounds good. Sep 25 11:37:19 I have a Micro HDMI, 5v power bank. Sep 25 11:37:28 uhh Sep 25 11:37:28 I need a new cord. Sep 25 11:37:31 Yep. Sep 25 11:37:32 micro usb you mean probably Sep 25 11:37:49 Oh, you are probably right. Sep 25 11:40:20 So, they have 5v barrel jacks to micro USB? Sep 25 11:40:57 I am going to look this one up later. Sep 25 11:44:35 running directly from li-ion ( https://nurdspace.nl/images/0/0a/Bbb-zmattified.jpg ) isn't *very* hard either, but it does require some soldering, both for the battery header itself and to work around the 3.3v regulator bug ( https://groups.google.com/d/msg/beagleboard/7sxPePT7wkM/_oNSCta5WusJ ) Sep 25 11:45:30 and there are other limitations, such as that the usb host port won't work, and the 3.3v regulator bug workaround means less current is available on the 3.3v supply for capes or other external hw Sep 25 11:46:19 ... Sep 25 11:46:44 tl;dr - not recommended if you have trouble understanding that statement Sep 25 11:46:50 like tbr said, using an usb battery pack is the easy plug-and-play solution Sep 25 11:47:02 I am going to look up 5v barrel jack to micro USB. Sep 25 11:47:03 Okay. Sep 25 11:47:17 ? just power it through the usb port Sep 25 11:47:28 USB on the BBB? Sep 25 11:47:40 yes ? the mini-usb Sep 25 11:47:46 Oh. Sep 25 11:48:10 I have a female, 5v micro USB power bank. Sep 25 11:48:52 most likely it has a normal USB-A port for devices Sep 25 11:49:06 micro-usb on power-banks is usually for charging the power bank ONLY Sep 25 11:49:18 Oh. Sep 25 11:49:23 It does. Sep 25 11:49:27 I can use that one, too. Sep 25 11:49:51 yes, a regular USB-A to mini-b cable will suffice Sep 25 11:49:57 Okay. Sep 25 11:50:26 Well, I guess I am using that USB to micro USB. Sep 25 11:51:51 http://photobucket.com/gallery/user/radioraiders/media/bWVkaWFJZDo3MzY1MTU3OA==?ref= Sep 25 12:27:51 up, up, and OTAY! Sep 25 12:29:20 a rather strange fellow Sep 25 12:37:31 yup Sep 25 13:41:18 one certainly old enough to remember "little rascals" **** BEGIN LOGGING AT Mon Sep 25 16:44:52 2017 **** BEGIN LOGGING AT Mon Sep 25 19:21:19 2017 Sep 25 19:28:32 i am getting version `GLIBCXX_3.4.21' not found errors when cross compiling for beaglebone, i can build the same prohgram on target and it works, but on ubunntu with eclipse it does nto work Sep 25 20:09:55 does gcc g++ 5.1.0 work on beaglebone Linux beaglebone 4.4.54-ti-r93? it is not available via apt-get... Sep 25 20:27:25 hello, anyone trying the grove cape? I can't seem to find a device overlay spec for it Sep 25 21:54:29 is the config for the autobuilder online anywhere? Sep 25 21:54:44 eg https://builds.beagleboard.org/ Sep 25 21:54:52 it seems like it was at one point just a buildbot Sep 25 21:55:05 would be nice to see what the stages for building an image are **** ENDING LOGGING AT Tue Sep 26 03:00:02 2017