**** BEGIN LOGGING AT Tue Jul 28 02:59:58 2015 Jul 28 03:26:02 hah, never underestimate the ability of a misguided Windows admin to make things difficult on their users. Jul 28 03:26:40 uhuh Jul 28 03:27:56 I wouldn't feel any urgent desire to use a Windows system as host for BBB development in the first place anyway Jul 28 03:28:55 concur. Jul 28 03:29:02 but sometimes that's what's around Jul 28 03:48:46 mostly windows PC around at work too, for linuxy things they either ssh to the designated "development" beaglebone, or to bob Jul 28 03:49:00 (bob, the builder, where a cross-compile environment is available) Jul 28 03:51:21 did you know you can actually enable focus-follows-mouse in Windows via regedit? Jul 28 03:52:11 (though some applications really don't interact well with it, including current versions of Visual Studio) Jul 28 04:14:43 What are some good capes to use for bi directional motor control Jul 28 06:48:45 hjgh Jul 28 06:52:09 I just bought a BBB for learning Embedded linux and linux device drivers. I want to write a driver for UART in BBB, but as in x86 architecture the hardware used is PC16550d and its corresponding registers can be found in /proc/ioports, can anyone help me get started with information on hardware present for UART on BBB and how to check its corresponding address. Jul 28 06:52:43 sachin: sure, go read the am3XXX TRM. or the already existing uart drivers. Jul 28 06:53:27 sachin: of all possible choices of peripheral to practice on... I would not recommend the UART Jul 28 06:53:33 it is terrible Jul 28 06:53:54 the whole Ethernet switch subsystem is probably easier to configure correctly than an UART Jul 28 06:54:44 zmatt: hey, don't confuse reality with learning :) Jul 28 06:55:49 note that you can also practice using uio: that lets you map a peripheral into userspace and receive IRQs by reading from an fd Jul 28 06:56:22 then at least you know how the peripheral behaves before trying again in kernel space Jul 28 06:57:25 LetoThe2nd: the existing uart driver contains bugs Jul 28 06:57:31 the initialization procedure in the TRM also Jul 28 06:58:10 I don't blame them, the uart really is just horrible Jul 28 06:59:33 especially since its register map morphs, with some registers whether other registers are visible or readonly Jul 28 06:59:35 I have a sensor for testing which interfaces via UART i.e BNO05, So I just want to interface by writing a out of kernel tree driver module Jul 28 07:00:16 sachin: to get an impression, see this post (and probably the whole thread): http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/379360/1342615#1342615 Jul 28 07:00:41 A simple RX and TX communication just to get me started on BBB Jul 28 07:00:42 sachin: better write something userspace then. if its some serial protocol on top of an uart, putting it into the kernel is mostly pointless. Jul 28 07:01:14 although I later realized things can be simplified, especially if you don't give a shit about manual control of modem control signals (which is almost always true) Jul 28 07:01:28 LetoThe2nd: well, the TTY layer does have a _lot_ of overhead iirc Jul 28 07:02:08 zmatt: which certainly will kick your ass with an orientation sensro... NOT Jul 28 07:02:29 LetoThe2nd: well, maybe he wants to poll it really fast? O:) Jul 28 07:03:21 or just wants to practice directly working with a peripheral Jul 28 07:03:34 long story short: reading uart peripheral docs can be interesting to learn, but is totally the wrong way here (IMHO) Jul 28 07:03:34 but the uart... aiai Jul 28 07:04:05 the TRM chapter of the UART is confusing and contains some very bad advice Jul 28 07:04:21 zmatt: sounds like real life, huh? Jul 28 07:04:27 I'm looking for my latest version of the simplified UART init procedure Jul 28 07:04:31 after init the UART is okay Jul 28 07:05:11 at least, as long as you don't do more than 8 consecutive posted writes, since that corrupts the FIFO pointer logic Jul 28 07:05:46 (inserting dummy writes fixes the problem, as does using DMA) Jul 28 07:06:00 luckily most UART code is slow enough to never notice this Jul 28 07:06:40 oh, and don't look at TI's StarterWare for guidance either, trust me.... http://e2e.ti.com/support/embedded/starterware/f/790/p/386753/1368322#1368322 Jul 28 07:07:53 I just wanted to write a serial driver to learn interfacing a peripheral and interrupt handling and later DMA Jul 28 07:08:01 on BBB Jul 28 07:08:25 well, enough said then. Jul 28 07:09:04 So if anyone can point me to right direction for documents and datasheets to read, ill appreciate Jul 28 07:09:23 0what is unclear about TRM? Jul 28 07:09:52 LetoThe2nd: the registers they forgot to list, for example Jul 28 07:09:59 and that a datasheet contains superfluous and sometimes misleading or wrong information is real life. better learn that too. Jul 28 07:10:28 e.g. reading a squeaky clean, stripped implementation is no learning effect. Jul 28 07:10:32 sachin: I found one version of my init procedure -> http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/351658/1237650#1237650 Jul 28 07:10:43 (my $.02) Jul 28 07:10:56 LetoThe2nd: learning through the UART is cruel though Jul 28 07:11:18 So what do you suggest me to start with? Jul 28 07:12:23 * LetoThe2nd would say adc in the iio. Jul 28 07:13:29 the adc might be a nice one... it doesn't require too much setup, and seems quite well-behaved Jul 28 07:14:08 yeah. size is bearable, and the specific implementation is not a humonguous mess. Jul 28 07:14:27 dunno about the current driver, but then again I don't really understand IIO either Jul 28 07:14:51 * zmatt disabled IIO and programs the ADC from userspace, including EDMA to a circular buffer to always have the latest values available Jul 28 07:16:05 sachin: if you're really new to the am335x altogether, I suggest just browsing around and playing with things Jul 28 07:16:14 most of which can be done from userspace Jul 28 07:17:18 I personally use GPIO, ADC w/DMA, and PWM from userspace Jul 28 07:18:04 for McSPI I'm conflicted between attempting to fix the driver or just tossing it and using it from userspace Jul 28 07:18:20 I am familiar with application level for most of the drivers in linux Jul 28 07:18:49 I mean access the the drivers using C libraries. Jul 28 07:19:01 that makes you familiar with linux Jul 28 07:19:09 accessing the peripherals makes you familiar with the hardware Jul 28 07:19:17 open(), read(), write(), ioctl() calls etc. Jul 28 07:21:04 if you want a useful project: the current omap2-mcspi driver implements its own transfer_one_message method rather than using the standard one provided by spi.c Jul 28 07:21:16 thanks to which it doesn't support using GPIOs as chip-selects Jul 28 07:22:30 it also does its own dma mapping even though afaict spi.c already offers to do that Jul 28 07:23:29 afaik fixing the driver mostly means tossing away a lot of code in omap2-mcspi.c that (poorly) duplicates the work done by spi.c Jul 28 07:23:37 but I may be wrong Jul 28 07:25:28 if you want to stick to your UART project: the last link I gave contains a functional init procedure.... after init, the UART is not *too* eccentric anymore Jul 28 07:26:07 apart from not being able to cope with too many consecutive writes to the FIFO (even if it has enough space) Jul 28 07:26:09 Yes thanks I think ill stick to the UART. Jul 28 07:26:19 try it out from userspace from Jul 28 07:26:22 *first Jul 28 07:26:40 Yes Jul 28 07:26:45 * zmatt looks for the necessary device tree ingredient Jul 28 07:26:58 I need the hardware information Jul 28 07:27:16 register map is mostly in the TRM Jul 28 07:27:34 Whats a TRM? Jul 28 07:27:36 grab an omap4/5 TRM for the full register map Jul 28 07:27:43 Technical Reference Manual Jul 28 07:27:47 Ok Jul 28 07:28:05 zmatt: see why i backed out? :-) Jul 28 07:28:19 Thanks a lot Jul 28 07:28:33 sachin: learn to download TRMs of related processors and cross-reference in case of confusion Jul 28 07:28:44 and don't assume the TRM is always right Jul 28 07:29:35 for the AM335x, I also recommend downloading the omap-L137 and/or omap-L138 TRMs, especially if you ever want to do anything with the LCDC Jul 28 07:30:17 the DM814x TRM is its closest relative, and has good info on the interconnects Jul 28 07:31:17 (the DM816x is a more distant ancestor, but one of the few that has a substantial chapter on the Debug subsystem) Jul 28 07:32:08 I am new to device driver programming so pardon me if I ask anything stupid or my technical terminology is not upto the level Jul 28 07:32:24 but often the most detailed info can be found in OMAP (before omapgeddon) or Vayu documentation... you just have to filter out the info that's not relevant to you Jul 28 07:33:04 in case of the UART, the AM335x TRM register map is missing several registers Jul 28 07:33:32 the TRMs of its siblings do mention them Jul 28 07:33:49 (the missing registers allow you to determine how much data is currently in the FIFOs) Jul 28 07:34:16 ignore the UART initialization procedure stated in the TRM, it's garbage Jul 28 07:34:48 and, general rule when working with any hardware: _check for applicable errata_ Jul 28 07:35:24 ok, thanks for the info Jul 28 07:35:39 Ill note all of the above points Jul 28 07:36:09 for background info: the first thread I linked to (with the state-diagram) was the one where I was still figuring out the UART myself in parallel to helping someone Jul 28 07:36:42 so that affects especially earlier posts in that thread Jul 28 07:40:19 sachin: btw, if you want to use uio (which is a bit nicer than using /dev/mem, and also allows you to use irqs from userspace): Jul 28 07:41:08 make sure CONFIG_UIO_PDRV_GENIRQ is included in the kernel (it is compiled as module in any not-too-ancient kernel from rcn-ee's repo) Jul 28 07:41:35 echo 'options uio_pdrv_genirq of_id=uio' >/etc/modprobe.d/uio.conf Jul 28 07:43:03 that lets you attach uio_pdrv_genirq to a peripheral by using compatible = "uio"; in device tree Jul 28 07:43:31 make sure its register range starts and ends on a page boundary Jul 28 07:43:53 (since uio can't mmap() a portion of a page, obviously) Jul 28 07:45:12 by default uio devices are just numbered (/dev/uio0 etc) which is annoying, however recent 4.x kernels make it easy to have an udev rule create an alias based on a device-tree property Jul 28 07:46:03 for example you could put add uio-alias = "foo"; in the device tree node Jul 28 07:47:26 and e.g. add the udev rule: Jul 28 07:47:42 SUBSYSTEM=="uio", SYMLINK+="uio/%s{device/of_node/uio-alias}" Jul 28 07:47:51 now you got /dev/uio/foo Jul 28 07:47:53 ta-da Jul 28 07:47:56 :) Jul 28 07:50:00 sachin: also, I don't know how familiar you are with ARM Architecture, but it's useful to know that devices are normally mapped as Device-type memory in kernelspace, but annoyingly they are mapped "Strongly-ordered" when using /dev/mem or uio Jul 28 07:52:18 this little difference makes writes to a device take 1 cycle in kernelspace (until you perform enough writes to fill up some fifo on route to the peripheral) Jul 28 07:52:34 versus about 150 cycles in userspace (the ping time to the peripheral) Jul 28 07:53:27 I'll now stop overflowing you with info :P Jul 28 08:12:00 zmatt: Thanks for all the info I really appreciate it Jul 28 08:12:50 And I am new to ARM processor, worked with ARM controller but not with Processor. Jul 28 08:25:22 sachin: a cortex-A processor is quite a different beast from a cortex-M (or older ARM7/ARM9 processors) Jul 28 08:25:26 :) Jul 28 08:28:35 * KotH finds it kinda cool, that there are arm9 uC's, with on-die ram and flash Jul 28 08:30:23 ...now if nxp would learn something from st and give those beast some reasonable amount of ram, one could actually make use of that raw processing power Jul 28 08:31:10 there are two ARM9 cores in HDVICP2 aka IVA-HD Jul 28 08:31:38 instantiated in omap4/5, dm816x (3 instances), dm814x, am572x Jul 28 08:32:03 does not surprise me Jul 28 08:32:07 arm is everywhere Jul 28 08:32:26 and it will become STANDARDS, ORDER, DECORUM in servers Jul 28 08:32:42 I guess they're either cheaper than or perform better than cortex-m for streaming data Jul 28 08:32:52 while avoid the timing uncertainty of the cortex-A Jul 28 08:32:56 more memory bw Jul 28 08:33:16 and they are cheap, if you already have licenses for the more advanced arm cores Jul 28 08:33:29 that will play a role no doubt :P Jul 28 08:34:34 in iva-hd they're just used as "sequencers" though Jul 28 08:34:39 to manage the other components Jul 28 08:35:33 each core has 16 KB data TCM (space reserved for 32 KB) Jul 28 08:35:37 32 KB instruction TCM Jul 28 08:36:20 and a connection to the SL2 RAM Jul 28 08:39:58 i can top that! Jul 28 08:40:05 the rpi has an arm9 core as usb controller! Jul 28 08:43:18 SL2 is a beast... 256 KB (though it can be instantiated up to 512 KB in theory) interleaved over 8 memory banks, serves 18 initiators, everything 128-bit wide Jul 28 08:43:39 so that arbiter is an an 18-to-8 crossbar, 128-bit wide Jul 28 08:47:13 I think that might actually imply more data hauling capacity than the L3 Jul 28 08:47:58 sl2? l3? Jul 28 08:48:38 SL2 is the shared L2 SRAM of IVA-HD (HDVICP2) components Jul 28 08:49:17 L3 ... have you never even *browsed* any docs on any TI SoC ? :P Jul 28 08:50:11 zmatt: only that of the am335x... and only superficially... to answer questions here ;) Jul 28 08:50:45 the am335x has an L3 interconnect too, it's the main interconnect between components Jul 28 08:51:19 zmatt: do you know that i dont own any beagle*? Jul 28 08:51:31 lol Jul 28 08:52:25 then get one you silly! they're fun Jul 28 08:52:59 or get the "toy car" shown on pages 14-15 of http://focus.ti.com/asia/download/2BJ_1-2.pdf Jul 28 08:53:03 sorry, cannot afford them Jul 28 08:53:10 * KotH is a poor student Jul 28 08:53:18 (if you'd manage to get one of those, that would make you an epic god) Jul 28 08:54:19 av500: get one of the toy cars shown on pages 14/15: http://focus.ti.com/asia/download/2BJ_1-2.pdf Jul 28 08:54:26 av500: they are missing in your collection! Jul 28 08:55:27 see if they can swap out the DRA626 for a DRA628 while you're at it... dunno what features are missing in the 626 since not even a tech brief is available, but I'm sure you wouldn't want to miss out Jul 28 08:56:19 actually, I might know... the slides show little or no sign of pruss or the pwmss instances, even though they know they're on that die Jul 28 08:57:13 (the DRA62x / DM811x is sort of a hybrid between the DM814x and the AM335x) Jul 28 08:57:49 never made became available for broadmarket though, only the automotive boys get to play with them Jul 28 08:58:08 well.. just buy a couple million and you'll get them Jul 28 08:58:34 but: how many people does TI currently have working on their SoC families? Jul 28 08:58:52 they have been firing them for 3years straight Jul 28 08:59:24 avnet claims you can buy a DRA622 with minimum order quantity 1 Jul 28 08:59:36 oh my god arduino is writed in java Jul 28 08:59:45 but that's a DRA622... who knows what might be missing Jul 28 09:02:41 KotH: I actually considered buying an evm811x to play with... but they're a bit expensive as a toy just to play with Jul 28 09:08:09 it's a tragedy how TI has treated the 81xx series Jul 28 09:08:27 oh well, once the x15 comes out I'll probably forgive them instantly Jul 28 09:08:50 nyo_: ? Jul 28 09:10:03 if it comes out. what's the current estimate for when they want to announce it? Jul 28 09:20:36 "So I'm guessing that means that the Digi-Key listing with an estimated ship date of 11/2/2015 is not exactly solid yet?" Jul 28 09:20:40 Gerald> "It is as good as any. If TI supplies the parts, then that date can be met. I can't speak for TI." Jul 28 09:21:19 so they moved it again, to november Jul 28 09:25:03 ok, afk for a bit Jul 28 11:13:23 zmatt, I mean the IDE of arduino wants java :* Jul 28 12:49:42 How can i know what's wrong witn a init.d/script? I tried to make X11vnc start but it don't work, and i don't know why Jul 28 14:53:59 what's difference between bbw and bbb ? Jul 28 15:04:35 One is white, and one is black. They also use different SoCs, and they have differences in pins. Jul 28 15:05:00 owl-v-_: But I expect that typing "beaglebone white vs black" into Google will be far more instructive. Jul 28 15:10:22 I have a question, I'm trying to run a cpp program. But when I try to run it it says: "-bash: ./CVA: No such file or directory". I've copied the program from an other Beaglebone on which it did work Jul 28 15:10:30 what am I doing wrong? Jul 28 15:11:50 * agmlego gets out the crystal ball. Jul 28 15:12:49 My guess is either a) something is wrong with $PATH on the afflicated bone or b) you failed to copy the permissions correctly or c) you fail to have the correct dynamic libraries installed. Jul 28 15:12:55 In order of likelihood. Jul 28 15:13:03 But this is a Linux thing, not a bone thing. Jul 28 16:50:04 hi how to use beagle_boneblack as web server Jul 28 16:50:24 what webserver to use Jul 28 16:50:30 it has one preinstalled in the defaulte images Jul 28 17:07:18 have you tried make your own os on bbb? Jul 28 17:10:16 ... why? Jul 28 17:12:32 for fun. Jul 28 17:12:49 anyway, how do i stop at u-boot prompt? Jul 28 17:12:58 rather than booting an os? Jul 28 17:13:16 Fail to have an OS installed. Jul 28 17:18:28 haven't done it, but I'm pretty sure you just send characters into the uart console during boot, if it sees one at the "hit any key to stop autoboot" step, it stops there and gives you a prompt Jul 28 17:28:28 just hit enter at the uboot prompt .. but you'll need to have a console connected at boot cf. serial Jul 28 17:29:55 hi Jul 28 17:31:43 myself, veremit, minicom didn't send a character. I'm using 'screen' right now. Jul 28 17:31:50 i got 42 led panel (on each panel are soldered 24 led's), each panel is 12 Volt and 2.4 watt Jul 28 17:32:17 owl-v-_: what serial adapter are you using, just curious? Jul 28 17:32:50 which is the value in Ohm of the resistor to use for each led ? Jul 28 17:33:06 myself, one from sparkfun.com ftdi with 3V Jul 28 17:33:53 natsukao_, one that does not burn led and the processor. Jul 28 17:34:04 you're seeing bootup messages, so that's good. Can't hurt to try a loopback test: remove the adapter from the bbb, wire tx to rx and send some characters in minicom/screen, see if they echo back Jul 28 17:34:50 minicom has flow control on by default iirc Jul 28 17:34:53 turn it off :D Jul 28 17:39:05 veremit: what's the command line option for turning flow-control off? Jul 28 17:39:45 I go into the options .. Ctrl-A-O Jul 28 17:39:51 serial port setup Jul 28 17:41:13 veremit: no option like 'minicom -D /dev/tty.usb -b 115200' ? Jul 28 17:41:23 btw, it's working Jul 28 17:41:54 I don't think there's an option for it .. that I know of .. Jul 28 17:43:03 once you've set it . I just save as default Jul 28 17:56:25 what are the requirement for u-boot to boot a software? (ex:kernel) Jul 28 17:57:33 you're asking questions that're answered in the u-boot docs, why? Jul 28 17:57:49 myself : most people can't be bothered to read docs Jul 28 17:58:00 myself : or search the web Jul 28 18:06:05 hi guys I have the latest debian image flashed to a microsd card and needed to see if gcc 4.9.2 is causing my segmentation fault woes. How can I go about downgrading to 4.8? Jul 28 18:08:03 have you tried apt-get ? Jul 28 18:09:10 should be able to remove the current version and install the specific one you want Jul 28 18:09:21 installing g++-4.8 that way was not sure if it would remove the existing g++/gcc at that point. Im used to using gentoo/archlinux and not too much of a fan debian based distros Jul 28 18:10:08 build gentoo for it Jul 28 18:10:10 ah, dunno a whole lot about gentoo Jul 28 18:10:18 shouldn't take more than a couple of days to compile ;) Jul 28 18:10:24 :/ Jul 28 18:10:28 can gentoo cross-compile? Jul 28 18:10:37 ofc Jul 28 18:10:46 cross-tool-ng Jul 28 18:10:53 builds its own tool-chain :D Jul 28 18:11:14 just .. takes a while Jul 28 18:11:30 *can gentoo cross-compile from a working random linux system with a cross-compiler installed Jul 28 18:11:32 it can Jul 28 18:11:33 :P Jul 28 18:11:56 zmatt : boot stick :P Jul 28 18:11:57 just never tried to get it working on the beaglebone black Jul 28 18:12:15 I've seen guides for arm boards .. not sure whether it was the wandboard/pi/beagle .. Jul 28 18:12:18 and I think gentoo nowadays even have some prebuilt binaries for the lazy fucks Jul 28 18:12:19 tried archlinux but failed with my distcc cross compile attempts and gave up Jul 28 18:12:34 mistawright: maybe yocto is something for you? Jul 28 18:12:38 ah yea you don't wanna cross-compile with distcc .. unless you're very careful how you set it up Jul 28 18:12:50 its possible .. but SO many gotchas Jul 28 18:13:11 veremit, this was with setting it to use a compiler to build arm binaries. cant remember if it was using the linaro toolchain or something Jul 28 18:13:13 and yea .. fyi .. I haven't tried :D Jul 28 18:13:28 I was lazy .. have always used debian/linaro/RN's images :D Jul 28 18:13:39 one day I'll do it "the gentoo way" Jul 28 18:14:00 but yea .. sucks compiling all the stuff on anything shy of a imx6-quad imho. Jul 28 18:14:22 certainly wouldn't on a single-core 800mhz/1ghz .. too f** slow Jul 28 18:14:31 * zmatt cross-compiles mostly everything Jul 28 18:14:35 hence pre-built distro's win for me. Jul 28 18:15:00 even for tiny projects cross-compiling and rsync'ing it over is faster than compiling locally Jul 28 18:15:11 zmatt : well that's a no-brainer Jul 28 18:15:17 lol Jul 28 18:15:20 (I should probably get off my lazy ass and setup NFS to remove the rsync step) Jul 28 18:15:28 zmatt : yup Jul 28 18:15:41 stick to nfs3 .. :p Jul 28 18:15:49 its piss-easy Jul 28 18:16:12 you can even nfs-boot ... Jul 28 18:16:24 ahem .. bootp Jul 28 18:16:36 best way for development Jul 28 18:17:39 yeah bootp/tftp is something I already use for baremetal dev Jul 28 18:31:11 what device do I need to mount from my beaglebone black to edit the uEnv.txt of the boot partition of the emmc? Jul 28 18:31:37 I dont have my serial cable to figure out why this dtb is causing it not to boot from the emmc Jul 28 18:34:29 get one. Jul 28 18:34:34 :P Jul 28 18:39:41 Running the beagleboard 4.1 kernel, should I have something at /sys/bus/i2c/devices/0-0024/input? In search of the mythical tps65217_pwr_but... Jul 28 18:43:14 veremit: one thing about nfs is that I worry a bit about security... I know we have at least a few managed switches, maybe I should see if I can isolate things a bit Jul 28 18:44:15 veremit: of course preferably each beagle would authenticate to kerberos, but well... Jul 28 18:44:18 :P Jul 28 18:44:35 NFS and security probably ought not be in the same sentence... ;-P Jul 28 18:44:52 agmlego: that's not quite true with nfs4 anymore afaik? Jul 28 18:45:30 agmlego: but that's why plan B would be an isolation wall from the rest of the company network Jul 28 18:46:35 just run it on a separate network & switch .. peasy. Dual lan boards are easy enough to come by Jul 28 18:47:08 zmatt .. is the tps on the i2c? Jul 28 18:47:10 veremit: yeah, but still it's suboptimal Jul 28 18:47:16 veremit: of course it is Jul 28 18:47:20 compromises :p Jul 28 18:47:22 veremit: i2c 0 Jul 28 18:47:23 zmatt: NFSv4 is better, but only if you configure it to refuse to talk anything lower. Jul 28 18:47:56 agmlego: I see no problem with that Jul 28 18:48:10 Just not often configured out of the box that way. Jul 28 18:48:14 I should crack nfs4 some day .. Jul 28 18:48:24 just probably not this year. Jul 28 18:48:37 too much other crap lol Jul 28 18:49:25 agmlego: I rarely blindly trust out-of-the-box config Jul 28 18:53:11 can't I use just smb as rootfs? ;-) Jul 28 18:53:45 I should add that the tps65217 is there at i2c/devices/0-0024, everything works including soft poweroff, just no trace of an input subsystem. Not sure if something is missing from the driver or just my config Jul 28 18:54:18 limbrik: what do you mean by soft poweroff and by input subsystem Jul 28 18:54:31 poweroff requires cooperation of the RTC Jul 28 18:55:44 (there's also seqdown but the docs make it clear it's for debugging purposes and I never dared to use it) Jul 28 18:56:11 zmatt: I'm seeing references online to there being an input/input0 node in sysfs for tps65217_pwr_but... I probably don't have the terminology exactly correct there Jul 28 18:56:57 oh, dunno, I can imagine it could produce an input device but no idea which kernels might do that, or whether this requires specific device tree config Jul 28 18:57:41 zmatt: Soft poweroff is working, which I know initiates from the rtc_omap driver, but just evidence that the tps65217 is working correctly as well I guess Jul 28 18:59:14 ah Jul 28 18:59:28 (completely insane construction btw imho) Jul 28 18:59:54 no way to shut down other than by scheduling an rtc alarm 1 second in the future to take care of it Jul 28 19:02:47 I have no input device for the power button either Jul 28 19:06:09 good point since i want to receive PMIC irqs too.. I'm gonna take a closer look Jul 28 19:16:15 zmatt: Thanks for taking a look. It appears that in drivers/mfd/tps65217.c, the 3.8 branch has an interrupt handler which is entirely missing in 4.1. So maybe the functionality just never made it in Jul 28 19:16:28 iirc the irq handler was also quite buggy Jul 28 19:16:53 what device is the boot partition of the emmc on the beaglebone black? Jul 28 19:17:34 mistawright: that's a more complicated question than you think Jul 28 19:17:45 emmc doesn't really require a boot partition Jul 28 19:18:13 or well Jul 28 19:18:23 zmatt, I was hoping to edit my uEnv.txt that is causing my board to not boot without having to flash the emmc Jul 28 19:19:44 for u-boot emmc is mmcdev 1 Jul 28 19:20:26 if booting through the microsd card how would it be represented? I do not see a /dev/mmcdev Jul 28 19:20:39 mmcblk1p1? Jul 28 19:20:46 represented where? Jul 28 19:21:25 just for clarity, boot consists of several phases Jul 28 19:21:40 i figured I could boot debian from microsd to fix the emmc boot issue Jul 28 19:21:48 MLO (aka SPL), u-boot, kernel+initrd+dt, root filesystem Jul 28 19:22:26 limbrik, that was the correct device thanks Jul 28 19:22:28 if you have a recent u-boot, you can also enter usb mass storage mode Jul 28 19:22:37 then the bbb appears like an USB stick Jul 28 19:22:47 very convenient for fixing config mistakes Jul 28 19:23:24 zmatt, how do you do that. sounds like a lifesaver Jul 28 19:23:35 you have a console cable? Jul 28 19:24:09 interrupt boot (by pressing a key when it says to) to enter the u-boot cmdline Jul 28 19:24:14 then: ums 0 mmc 1 Jul 28 19:24:40 ("ums 0 mmc 0" would turn the beaglebone into an SD card reader) Jul 28 19:24:40 thanks. I'll remember that next time i have an issue like this Jul 28 19:27:53 hmm, can you perform manual i2c transactions to a device (like the pmic) that also has a kernel driver or will the kernel object? Jul 28 20:41:44 9 Jul 28 21:05:36 you can, of course, power on with the 'boot' switch held to boot from uSD card Jul 28 21:50:59 veremit: or connect a pulldown resistor to P8 pin 43 Jul 28 21:55:59 same diff :p Jul 28 21:56:14 anyone know why lshw gives a bus error on debian wheezy? Jul 28 21:57:06 doesnt exist on my debian box Jul 28 21:57:34 exists as package Jul 28 21:57:47 ah so it does... Jul 28 21:58:22 endoped: probably because it probes the hardware in scary low-level ways and didn't bother to install a signal handler to catch the resulting bus errors Jul 28 21:58:37 but I've never used lshw (or heard of it before today) so that's just a guess Jul 28 21:59:12 spews a bunch of crap out Jul 28 21:59:18 mostly not very helpful imho Jul 28 22:00:27 no bus error here either Jul 28 22:00:49 uname -a, zmatt ? Jul 28 22:01:11 endoped: you realize wheezy is _really_ old right? Jul 28 22:01:20 3.12-0.bpo.1-amd64 #1 SMP Debian 3.12.9-1~bpo70+1 (2014-02-07) x86_64 GNU/Linux Jul 28 22:01:30 best debian vintage Jul 28 22:01:48 as opposed to regular debian vintage jessie Jul 28 22:01:51 I mean, when debian themselves call it "obsolete" ... Jul 28 22:02:05 only one step behind 'stable' ;) Jul 28 22:02:17 Linux bb938 4.1.2-bone12-dd1 #3 Tue Jul 21 17:45:02 CEST 2015 armv7l GNU/Linux Jul 28 22:02:33 huh, armv7l ... why the hell does it think that Jul 28 22:02:45 thik that's right .. its not arm5 Jul 28 22:02:50 nor 9 Jul 28 22:02:56 it's armv7 Jul 28 22:03:02 zmatt: no i didn't realize that. I just downloaded the lastest image from beagleboard.org Jul 28 22:03:34 Linux beagleblack 3.13.6-bone7.2 #1 SMP Fri Mar 21 03:59:54 GMT 2014 armv7l GNU/Linux Jul 28 22:03:36 thats mine Jul 28 22:03:39 unfortunately I don't know who's responsible for that Jul 28 22:03:41 ye olde kernel of olde Jul 28 22:04:14 didn't realize jessie was stable now. Jul 28 22:04:36 jessie isn't #official# on bealg eyet afaik Jul 28 22:04:55 veremit: weird version... usually people stick at 3.8.13 or 3.14.48 Jul 28 22:04:57 yea lshw doesn't work on my beagle Jul 28 22:05:06 zmatt : RN's old kernel build Jul 28 22:05:26 he's up to 4.1 I thikn .. this was built AGES ago Jul 28 22:05:32 that's not very specific, RN is a fountain of kernels Jul 28 22:05:42 precifsely :D Jul 28 22:05:44 hehe Jul 28 22:05:58 https://github.com/RobertCNelson/linux-stable-rcn-ee/releases Jul 28 22:06:12 Fri Mar 21 03:59:54 GMT 2014 is the clue Jul 28 22:06:14 I normally run debian stretch myself Jul 28 22:06:29 unstable? or testing? Jul 28 22:06:36 stretch is testing Jul 28 22:06:44 so that's suitable for production use Jul 28 22:06:45 ah just regular old debian lol Jul 28 22:07:03 my own bbb runs sid Jul 28 22:07:05 no .. testing is not deemed 'production' .. thats why its called 'testing' lol Jul 28 22:07:11 it's debian Jul 28 22:07:25 unstable is about the only desireable flavour of debian really Jul 28 22:07:29 "unstable" is "usually works, and not too old" Jul 28 22:07:32 cos its remotely current Jul 28 22:07:59 "testing" is "not too ancient, and stable" Jul 28 22:08:14 no .. Stable is stable. The names and conventions are quite clear Jul 28 22:08:32 "stable" is "wait wtf, this version of make is so old it doesn't even understand the syntax of my makefiles?!" Jul 28 22:08:47 https://www.debian.org/releases/ Jul 28 22:09:06 I'm familiar with debian's view :P Jul 28 22:09:18 they're about one generation behind :) Jul 28 22:09:38 even sid often contains old crap Jul 28 22:09:58 hence better to run ubuntu/gentoo/anything else Jul 28 22:10:16 jessie is like running windows 7 Jul 28 22:10:26 testing ... 8 and 10 .. unstable ;) Jul 28 22:10:56 I assure you none of my debian systems resemble any version of windows ;-) Jul 28 22:11:11 and indeed I've been long intending to try some other distro Jul 28 22:11:18 get cracking Jul 28 22:11:25 gentoo isn't hard .. just have to compile everything Jul 28 22:11:47 kubuntu if you want out-the-box (I hate unity) Jul 28 22:12:07 arch/mint/etc lots of flavours Jul 28 22:12:08 I'm not going to touch ubuntu Jul 28 22:12:13 lol yeahno Jul 28 22:12:15 arch I once tried, didn't like Jul 28 22:12:45 I once tried fedora since quite a few innovative things that I liked seemed to be coming from that corner Jul 28 22:13:33 the insanity of yum was one of the things that chased me away, but afaict things have improved greatly on that point, maybe I should give it one more chance Jul 28 22:14:05 got a centos vps that confuses me a lil Jul 28 22:14:54 the initial "where the fuck is everything / how the hell am I supposed to do this here" phase is an annoying barrier Jul 28 22:15:17 I'm just very used to debian Jul 28 22:15:46 at the same time, I know that's just an initial phase to be overcome Jul 28 22:16:26 yes .. Jul 28 22:16:36 thre are better linuxes Jul 28 22:16:49 I just use it for laziness and convenience on my arm boards Jul 28 22:18:09 nixos has an interesting philosophy Jul 28 22:18:16 I don't know if that translates to a usable system Jul 28 22:23:16 lol Jul 28 22:25:40 but I like the idea Jul 28 22:46:13 hmmm I thinker their is a Gentoo Arm distro. However gentoo is too deynamic for something with a solidstate drive as it tends to use the disk a lot. Jul 28 22:49:01 there is, and yes on the disk too. Jul 28 22:53:12 You mean the LiveDVD? Jul 28 23:01:36 I mean gentoo is reliant on regular disk access for compiling/updating/etc Jul 28 23:01:46 but there definitely is a gentoo arm variant Jul 29 01:03:39 Good I was momentarily confused. I got better (heh). Although I like gentoo for it's ability to be customized it's flexibility to do so is also a weakness for a system that doesn't thrash media space. I heard someone killing his raspberry Pi SD cards by not configuring things correctly. I suppose the dynamic disk driver that they have on the LiveDVD might be helpful for adding stuff without killing the solid state drive.e Jul 29 01:04:09 just use an external usb hdd :D lol Jul 29 01:20:53 Comeon ... can't I speculate dang it! Yeah probably best if one thinks about it. Jul 29 01:23:00 anyone have any of the big little 64bit arm parts on boards yet? I noticed that ARM contributes the backend for GCC for their processors. Jul 29 01:57:15 GenTooMan: jetson tk1 sorta. **** ENDING LOGGING AT Wed Jul 29 02:59:58 2015