**** BEGIN LOGGING AT Mon Oct 31 03:00:01 2016 Oct 31 04:52:55 how do we feel about apt dist-upgrade Oct 31 04:53:42 does anyone use it? Oct 31 06:07:03 eh, yes, generally always Oct 31 06:07:31 you just need to check its proposal a bit and not just blindly press enter like you can probably safely do with upgrade Oct 31 06:10:04 aptitude full-upgrade can be slightly smarter with solving complicated upgrade scenarios, but in my experience aptitude is just too horribly slow Oct 31 06:10:09 on a bbb Oct 31 06:10:40 (aptitude's equivalent for "apt-get upgrade" is aptitude safe-upgrade) Oct 31 06:11:00 safe-upgrade and full-upgrade are indeed better descriptions than upgrade and dist-upgrade Oct 31 08:57:29 Hello People Oct 31 09:10:22 mau :) Oct 31 09:12:54 nyaaa Oct 31 09:35:35 Is it possible to use the 'boot' button on a BBB as a user button (once the boot process is finished)? Oct 31 09:36:27 I want to use the BBB for a demo and it would be neat to set it running just by pressing one of the buttons on the board itself Oct 31 09:37:16 Rather than connecting kbd/monitor, or connecting an external button to the expansion port with a bit of breadboard. Oct 31 09:37:34 Thanks for any advice! Oct 31 09:41:41 look at the schematics and see where it is connected to Oct 31 09:50:04 I suspect it can be used as regular gpio Oct 31 09:50:15 but yes, schematics should answer that Oct 31 09:58:05 I looked at the schematics (it's on sheet 6) and that button forms part of the 'boot configuration matrix' (I think that's what the docs call it). But it's a bit beyond my experience - I couldn't quite understand if that matrix was fundamental to the CPU or whether it's something the bootloaded works with. Oct 31 09:58:26 If you think it's likely to be possible I will look some more and try to figure it out. Oct 31 10:15:18 OK, the schematics call it SYS_BOOT2, and they've put LCD_DATA2 next to it. The manual says LCD_DATA2 can also be GPIO2_8 in mode 7 ... so, yes it should be possible? What do you think? Oct 31 10:16:23 yup, sounds like it Oct 31 10:16:37 does it say if you can put an interrupt on it? Oct 31 10:28:11 Hmm. I think so. I'm looking at page 3370 of document SPRUF98Y. Oct 31 10:29:08 then it should be all smooth sailing :) Oct 31 10:33:27 tbr: Thanks. I need to go off and lie down in a darkened room for a while now. Oct 31 14:07:19 zmatt: the problem is i'm running out of space during the upgrade process Oct 31 14:07:45 zmatt: even though, according to apt, the whole thing only needs a couple MB more at the end of it all (not sure how accurate that is) Oct 31 15:03:29 man for some reason the last write i do to text files before reset are never written after reset unless i run commands before reseting Oct 31 15:03:44 okay not *always* Oct 31 15:06:58 you do understand how buffers work, do you? Oct 31 15:07:37 yeah i do Oct 31 15:07:59 do you think i mean write commands? Oct 31 15:09:14 yes Oct 31 15:09:32 for a glimpse of people who didnt understand buffers, please read https://lwn.net/Articles/322823/ Oct 31 15:10:39 oh okay Oct 31 15:10:40 5 seconds. Oct 31 15:10:45 for filesystem writes Oct 31 15:11:08 that's default Oct 31 15:11:11 it can take longer Oct 31 15:11:17 seconds.. minutes even Oct 31 15:11:35 you really should do an fsync if you want to have your data written down and be sure they are there Oct 31 15:11:41 cool Oct 31 15:11:55 or in other words: you didnt understand buffers (yet) ;) Oct 31 15:12:01 i was gonna say, RUDE Oct 31 15:12:10 *g* Oct 31 15:12:21 seriously though, just because i understand buffers, doesn't mean i'd know there was a filesystem buffer Oct 31 15:12:38 yeah.. Oct 31 15:12:53 it used to be common knowledge... because unix (and the modern windows) had that, unlike dos Oct 31 15:13:06 so you should just press the power switch Oct 31 15:13:27 meh i'd rather run the fsync Oct 31 15:13:33 just for the record: a modern OS has buffers _EVERYWHERE_ Oct 31 15:14:09 there are very few places where people actively avoid buffers (ie everyhwere where low latency is paramount), but they are really few Oct 31 15:14:29 i mean, i hate it. Oct 31 15:14:49 but i realize i'm ignorant of the issue so i wont argue with it until i understand it better Oct 31 15:14:58 well, otherwise your system comes to crawling halt Oct 31 15:15:18 the buffers are there to ensure that a slow device (and all file storages are slow) does not create a bottleneck Oct 31 15:15:24 and i've seen examples of that Oct 31 15:15:39 but i'd rather the storage medium have it's own fifo Oct 31 15:15:43 rather than a software fifo Oct 31 15:15:48 doesnt help Oct 31 15:15:53 because of powerdown? Oct 31 15:15:53 the interface is already slow Oct 31 15:16:01 that too Oct 31 15:16:13 ah you're talking about software layer comm bottlenecks i think Oct 31 15:16:27 the hardware is the bottleneck Oct 31 15:16:31 software runs fast Oct 31 15:16:34 but it runs in RAM Oct 31 15:16:39 where everything is fast Oct 31 15:16:51 once it has go out of the CPU/RAM combo, it becomes slow Oct 31 15:16:52 then you could use a hardware fifo no? Oct 31 15:17:09 although i suppose i dont know how large it would have to be, reasonably Oct 31 15:17:13 or how much more xpensive it would be Oct 31 15:17:17 exactly Oct 31 15:17:31 so better do it in software and be able to manage it according to demand Oct 31 15:17:42 it is what it is Oct 31 15:17:54 what do you think of these chips: https://www.microchip.com/design-centers/memory/serial-eeram Oct 31 15:18:07 i mean the NVSRAM Oct 31 15:18:07 beside: there are FIFOs for all slow hardware. but they are minimally sized to enable the hardware to achieve maximum throughput... which means you still have latency Oct 31 15:18:31 good point, good point Oct 31 15:18:34 better, at least for write, but the interface is still slow Oct 31 15:18:49 you dont get more then 20'ish Mbit/s to these devices Oct 31 15:19:38 oh.. sorry.. 1Mbit/s Oct 31 15:21:19 yeah the speed is really bad Oct 31 15:21:43 they are not made as mass storage devices :) Oct 31 15:21:54 no i just think its a cool feature Oct 31 15:23:09 it's just SRAM Oct 31 15:23:25 could be embedded as a cache though maybe and used to restore state on a crash Oct 31 15:23:41 actually if the speed is slow i'm not sure Oct 31 15:24:03 you probably want something like this: http://arstechnica.com/information-technology/2013/04/memory-that-never-forgets-non-volatile-dimms-hit-the-market/ :) Oct 31 15:24:42 man i have so many dreams and here i am writing javascript Oct 31 15:24:59 having dreams is good. writing javascript isnt Oct 31 15:25:26 unless your goal is to end up as a low paid devops guy Oct 31 15:26:55 i mean, unfortunately, i'm an entrepreneur, so my career will take a slightly different path than most engineers Oct 31 15:27:02 it wasn't by choice, entireluy Oct 31 15:27:09 but i've embraced it. Oct 31 15:27:28 there are these javascript guys who say that they get paid 100-130k/yr here on the east coast USA Oct 31 15:27:50 i don't know what the top of the line looks like for you folk- maybe you're embedded engineers, or semiconductor guys, i don't know Oct 31 15:35:25 global domination, obviously Oct 31 15:35:53 but compassionate too Oct 31 15:36:02 the benevolent dictator Oct 31 15:36:47 Honestly the top of the line for me is making enough money to not have to work full time and then maybe I can get more into hobbies or entrepreneurial, or both Oct 31 15:37:10 i'm not sure those two things are in the same group Oct 31 15:37:40 I meant separately Oct 31 15:37:49 hobbies don't make most people much money Oct 31 15:38:33 lol and you don't need money to be an entrepreneur, take it from a guy living in his car at the moment Oct 31 15:38:58 You might not, but I like a certain level baseline level of comfort Oct 31 15:39:32 I would have to move somewhere warmer to do that anyway (live in my car) Oct 31 15:40:04 hahahaha Oct 31 15:40:10 i dont blame you at all Oct 31 16:43:56 hi, what is operating temprature for bbb rev c? Oct 31 17:33:32 comming from raspberry pi things with gpio where easy. now with my ne bbb i am a kind of frustrade. i read some things about device tree overlay that worked fine. now i try the sysfs to export gpios that i would like to use as input. now, the gpio_38 (P8_3) i cant access as an input. i see that it is multiplexed with mmc1_dat6. well. does this means i cant use this gpio/pin as an input ? Oct 31 18:51:37 are there any boards from beagleboard that are 64bit CPUs? Oct 31 18:51:48 (and if not, are there eventually plans to support 64 bits?) Oct 31 18:57:40 zacts, tu support 64 bits you will need a 64bit cpu. Oct 31 18:58:12 uFo-Z: I realize this, but do any of the beagleboards include a 64bit cpu? Oct 31 18:58:53 I have a Linux program that is only for 64 bit cpus, and I like the low power consumption of the beagleboards Oct 31 18:59:00 I also want to support beagle over the rpi Oct 31 18:59:11 (I like the direction of open hardware that beagle is offering) Oct 31 18:59:41 do you have the sources of this programm ? Oct 31 19:00:01 you could adapt it for 32 bit machines Oct 31 19:29:13 ayjay_t: yes, you should make sure you have plenty of free space during an apt upgrade, especially if it's a big one Oct 31 19:29:27 or do it in smaller steps Oct 31 19:38:34 yeah i'm thinking about doing it in smaller greps, just writing a script to go through dpkg -l line by line and then upgrading each one Oct 31 19:38:40 Does anyone know what privilage does a user needs to have to mmap /dev/mem on BBB 4.4.9? Oct 31 19:39:14 ls -l /dev/mem Oct 31 19:39:22 but i'm not an expert Oct 31 19:40:22 makes sense, should have guessed Oct 31 19:50:55 Are the current win64 beaglebone drivers signed? I see old posts about disabling signature checking but nothing recently Oct 31 19:51:21 my install of the latest drivers is failing fast Oct 31 20:11:56 Might anyone have an idea as to why [Clang IntelliSense] error: 'cstddef.h' file not found. Using Visual Studio 2012, latest VisualGDB Oct 31 20:12:47 You're using VS for BBB coding? (violent shiver) Oct 31 20:13:33 yes Oct 31 20:13:52 atleast atempting. Oct 31 20:59:08 where are the sources for /boot/dtbs/ located? Oct 31 20:59:57 phschafft, https://github.com/beagleboard/linux/tree/4.4/arch/arm/boot/dts Oct 31 21:00:09 ah, thank you very much rcn-ee :) Oct 31 21:09:28 first need to start network on the bb, no wonder the fetch fails. Oct 31 21:52:21 ok, that is intrivial to build, hm. Oct 31 21:52:47 phschafft, do you want an easier way? which kernel are you using? Oct 31 21:53:56 rcn-ee: intrivial as in I hoped I can just copy the files into this project repo and write a simple makefile to build it. Oct 31 21:54:09 so I now need to rethink this. Oct 31 21:54:17 phschafft, look at : https://github.com/RobertCNelson/dtb-rebuilder ;) Oct 31 21:54:56 ok. that could help :) Oct 31 21:55:07 thank you very much already. you're really helpful. :) Oct 31 21:55:15 just make sure you use a branch that matches your kernel.. Oct 31 21:58:48 yo rcn-ee. just curious, since kernel 4.8 or so (not quite sure) the eMMC started showing up consistent as mmcblk1 ... I first assumed you patched it or perhaps the linux-mmc people finally had a moment of clarity, but I see no evidence of either Oct 31 21:59:22 since I had begun to rely on it I made a proper patch btw Oct 31 22:00:04 zmatt, they had to do that, since their simultaneously mmc detection on bootup 'broke' the old numbering scheme. ;) Oct 31 22:00:11 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/mmc/core?id=520bd7a8b4152aacfbd34eb7f7a447354b631039 Oct 31 22:00:17 rcn-ee: of cause. I'm on 4.4.9-ti-r25. Oct 31 22:00:33 zmatt, and this: https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/mmc?id=9aaf3437aa72ed5370bf32c99580a3fa2c330e3d Oct 31 22:00:50 (the 2nd one i had backported to our v4.4.x-ti as i needed that to UUID issues on the x15..) Oct 31 22:01:07 phschafft, first you'll need to upgrade, there's a big "dts" delta betwen r25 and r72? Oct 31 22:01:47 phschafft, just run: cd /opt/scripts/tools/ ; git pull ; sudo ./update_kernel.sh Oct 31 22:01:49 and reboot.. Oct 31 22:03:20 rcn-ee: ehh I see no evidence of that second patch Oct 31 22:03:46 and the behaviour appeared in 4.7 sorry (I think) Oct 31 22:04:15 https://github.com/dutchanddutch/bb-kernel/blob/am33x-v4.8/patches/fixes/0003-mmc-use-of_alias-for-device-numbering.patch this is what I did Oct 31 22:04:24 it hit me in 4.5.5, (the wandboard went from: /dev/mmcblk0 -> /dev/mmcblk2) Oct 31 22:04:25 based directly on the i2c code Oct 31 22:05:01 rcn-ee: ok that's really weird since I didn't see anything to cause it Oct 31 22:05:10 * zmatt goes to look more closely Oct 31 22:06:01 Correction: the "simultaiously" probe happened in the v4.5.0-rc merge, the 2nd patch hit v4.5.3 (and v4.6.0-rc merge).. Oct 31 22:06:26 my tree is 4.8.4-bone2 Oct 31 22:06:48 in that case, microSD = /dev/mmcblk0 and eMMC = /dev/mmcblk1 with no external patches. ;) Oct 31 22:07:23 the only time you really notice.. when the microSD is not plugged, as the eMMC is /dev/mmcblk1... Oct 31 22:08:40 but *how* ? I see no evidence of it in the driver Oct 31 22:08:40 rcn-ee: it's trying to upgrade to 4.4.27-ti-r62. Oct 31 22:08:43 e.g. https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.8.4-bone2/drivers/mmc/card/block.c#L2227 Oct 31 22:08:59 phschafft, that's the one. ;) Oct 31 22:09:05 ok. Oct 31 22:09:21 let's see if that kernel update works, unlike the last one. Oct 31 22:10:36 rcn-ee: https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.8.4-bone2/drivers/mmc/core/host.c#L364 Oct 31 22:11:28 it's still just plain grabbing a number from an ida Oct 31 22:14:53 since I couldn't find any evidence of a fix having been applied, I figured it may just have been some perturbation of nondeterminism. I didn't want to gamble on that so I finally made a proper patch that uses of_alias_* to get numbering assignment from DT (and forces dynamically allocated ids to be higher than all statically assigned ones) Oct 31 22:16:33 zmatt, here's v4.8.x with just the eMMC: https://gist.github.com/RobertCNelson/f137dee9139f954bbf0a187127b2e5db Oct 31 22:18:02 but I wouldn't be surprised if the mmc-layer probably needs a more-deterministic update applied for the new defualt. Oct 31 23:12:46 upgrade seems to have worked. **** ENDING LOGGING AT Tue Nov 01 03:00:01 2016