**** BEGIN LOGGING AT Wed Jun 20 03:00:08 2018 Jun 20 03:33:02 zmatt: hmm, good point. Jun 20 09:03:51 anybody know what these numbers mean in this device tree construct "interrupts = <1 IRQ_TYPE_LEVEL_HIGH 7>;" Jun 20 11:58:18 Hi, I can't get Dlink DWA111 wifi dongle to work: https://pastebin.com/VLi8N2Am Jun 20 11:58:34 it resets/reconnects to usb as new device Jun 20 11:59:32 already tried to install firmware-ralink or linux-firmware-nonfree , no luck, any idea ? Jun 20 11:59:50 (debian8) Jun 20 12:26:06 TonyTheLion: that interrupts declaration looks like it isn't for beaglebone Jun 20 12:32:45 typically an interrupt specification only consists of two cells, not three Jun 20 12:32:48 https://www.kernel.org/doc/Documentation/devicetree/bindings/interrupt-controller/interrupts.txt Jun 20 12:39:52 fred__tv: I'm not seeing any failed attempts to load firmware, so it looks more like your device just isn't supported? Jun 20 12:40:05 the one I showed came from the gpio-keys device tree docs for my kernel version Jun 20 12:41:11 can you set in the device tree for a gpio-key the type of trigger you want? Ie level or edge? Jun 20 12:41:14 TonyTheLion: such examples probably come from a random device, so exact details like interrupts should be ignored Jun 20 12:41:34 ye its kinda confusing Jun 20 12:42:37 interrupt-specs are however not specific to a particular driver... i.e. interrupt-specs for gpio-keys on the am335x are the same as any other interrupt-spec on the am335x Jun 20 12:43:05 zmatt: https://wiki.debian.org/WiFi/rt73 USB: 07D1:3C06 D-Link System DWA-111 802.11bg Wireless Adapter [Ralink RT2571W] Jun 20 12:43:22 Debian 3.16 :-))) Jun 20 12:44:13 you're using kernel 3.16 you mean? ("debian 3.16" doesn't make sense) Jun 20 12:45:19 I'm not sure what you mean by "interrupt-specs"? Jun 20 12:45:22 TonyTheLion: for gpio-keys with an actual key/button, specifying the type of trigger wouldn't make any sense: it always triggers on both edges, performs debouncing, and generates the appropriate input event Jun 20 12:45:55 you can however also use gpio-keys with a key event generated by an interrupt rather than a gpio, which does allow any sort of interrupt specification Jun 20 12:46:54 you have to specify that in the devicetree? Jun 20 12:47:03 where else? Jun 20 12:47:31 (note: my internet is being flaky, so if I suddenly stop responding my internet probably dropped out) Jun 20 12:47:41 ok Jun 20 12:48:28 Do you have an example of how you specifiy the latter case you mentioned? Jun 20 12:48:38 is there one in your overlay-utils repo? Jun 20 12:49:41 what is your use-case exactly? the wakeup signal is coming from an external device rather than a button? Jun 20 12:50:02 ye Jun 20 12:52:30 replace the gpios property by interrupt-parent + interrupts similar to this: https://github.com/mvduin/py-uio/blob/master/dts/gpio-irq.dtsi#L27-L28 Jun 20 13:35:42 zmatt: that was the supported device list : The following list is based on the alias fields of modinfo rt73usb in Debian 3.16 (3.16.7-ckt9-3~deb8u1) kernel images. Jun 20 13:36:34 There is always something not working while trying to do something new..... Jun 20 13:37:46 ¯\_(ツ)_/¯ Jun 20 13:50:32 hi, I've a beagle bone green and if I do: Jun 20 13:51:02 echo 35 > /sys/class/gpio/export && cd /sys/class/gpio/gpio35 && echo out > direction && cat > value Jun 20 13:51:38 The pin "gpio 35" doesn't toggle when I type 0 and 1 (The "gpio 35" pin reference is here: http://wiki.seeedstudio.com/BeagleBone_Green/) Jun 20 13:51:51 and the multimeter always measuer 3.33v Jun 20 13:53:01 grep gpio-35 /sys/kernel/debug/gpio gives high when I set value to 1 and 0 when I set it to low Jun 20 13:53:06 is there some kind of offset? Jun 20 13:53:28 maybe check the pin configuration with my show-pins script: https://github.com/mvduin/bbb-pin-utils/#show-pins Jun 20 13:53:37 I have forced a wireless device into a monitor interface: iw phy phy0 interface add mon0 type monitor , ifconfig mon0 up Jun 20 13:54:09 zmatt: thanks Jun 20 13:54:14 how can I trig an event when the first packet from a mac addres is detected ?? Jun 20 13:54:33 GNUtoo: gpio35 would be listed as gpio1.03 Jun 20 13:56:02 i.e. ass soon you detect traffic from mac 11:22:33:44:55:66 execute a script Jun 20 13:56:10 so the gpio goes from 1 to 32 and not from 0 to 31? Jun 20 13:56:18 as.. :) Jun 20 13:57:14 hmm ok: P8.06 / eMMC d3 3 fast rx up 1 mmc 1 d3 mmc@481d8000 (pinmux_emmc_pins) Jun 20 13:57:22 strange that it exports gpio35 Jun 20 13:57:31 (and 34 and 36 too) Jun 20 13:59:25 gpio67 does work as it's not already used Jun 20 14:00:13 but I never seen anything like that, if the pin is used by another driver, it should refuse to export it, right? I've Linux 4.15.10 Jun 20 14:00:28 (mainline + possibly very few distro patches) Jun 20 14:00:59 Thanks a lot for the script link Jun 20 14:03:26 GNUtoo: no, 0 to 31 Jun 20 14:03:59 the kernel doesn't know which gpio corresponds to which pin, so you can export a gpio even if the pin is in use for other purpsoes Jun 20 14:04:02 *purposes Jun 20 14:04:33 if you use a beaglebone.org image, gpios for non-reserved pins are exported automatically Jun 20 14:04:38 *beagleboard.org Jun 20 14:29:40 zmatt: oh ouch Jun 20 14:30:10 so the gpio controller just exports the pin but it's not connected to anything because of the pinmux I guess Jun 20 14:30:15 yep Jun 20 14:30:59 Maybe I was making a confusion with GPIO already used somewhere else (like for leds) Jun 20 14:31:30 yeah in that case the gpio is already in use, and not just the pin Jun 20 15:16:38 someone mentioned to me that TI produces an opensource sdk is that correct? Jun 20 15:17:11 TI produces a variety of SDKs, most of which having both open- and closed-source components Jun 20 15:21:04 I was talking to a co-worker and he said that TI had its own yocto tools that were open source Jun 20 15:22:56 yocto targets many things, including various TI SoCs Jun 20 15:24:45 like every other linux distro, it is itself open source but the resulting image may include closed-source components Jun 20 16:45:22 arago? Jun 20 16:47:33 black_13: you can boot and use many TI boards with only free software Jun 20 16:47:58 black_13: some functionality will not work like 3D acceleration, but if you don't need such functionalities it should be ok Jun 20 16:48:25 also I don't know if video/audio decoding offload can work with fully free software Jun 20 16:48:51 with youcto you can generate your own SDK (including canadian cross compilation) Jun 20 16:49:36 though yocto itself is not Free Software Distribution Guidelines ("FSF") compliant as it may have non-free software in some repositories Jun 20 16:49:40 *in some layers Jun 20 16:50:38 Here I'm using a Beagle Bone green and I've also a beagleboard XM that only run 100% free software and that is "FSF" compliant (they run Parabola) Jun 20 16:53:49 GNUtoo: I have run yocto but not for beagle just to make test x86 images to run on qemu Jun 20 16:54:07 black_13: you can make your own SDK with yocto Jun 20 16:54:13 GNUtoo: have you created a complete system for the BBB Jun 20 16:54:22 black_13: define complete Jun 20 16:54:35 an sdk is just an installable toolchain Jun 20 16:54:43 something that boots and can be flashed to the BBB Jun 20 16:54:52 so you put whatever you want in it, such as QT and such Jun 20 16:54:59 of course Jun 20 16:55:01 ah that's an image not an SDK Jun 20 16:55:15 you tell me what the SDK is Jun 20 16:55:16 I didn't use yocto for that, but I used it in the past at work to do SDK Jun 20 16:55:23 (and images that boot) Jun 20 16:55:39 it's been some years, so I don't remember the exact details Jun 20 16:55:50 but I remember you can add the libraries you want in it Jun 20 16:55:51 some assembly required Jun 20 16:56:12 well, let's take a basic toolchain, with a libc, you can't do much with it Jun 20 16:56:18 you can cross-compile a hello world Jun 20 16:56:24 yes Jun 20 16:56:29 but you can't link to QT if you don't have QT in your toolchain Jun 20 16:56:33 same for a lot of libraries Jun 20 16:56:36 correct Jun 20 16:56:47 so yocto enables you to add whatever libraries you want in your toolchain Jun 20 16:57:33 so you could pick a default SDK/toolchain but depending on your needs you probably want to customize it Jun 20 16:58:56 and you have mainly 2 kinds of toolchains Jun 20 16:59:08 understood I am the point where I just want a small linux image that at best has a console a well know user and ethernet over usb stack Jun 20 16:59:17 if you have an x86 computer, you could generate a cross toolchain that runs on x86 and targets arm Jun 20 16:59:36 but you can also generate a toolchain that runs on powerpc and targets arm (it's called canadian cross compilation) Jun 20 16:59:39 i have actually dont that by hand some years ago Jun 20 16:59:45 done Jun 20 16:59:52 sorry for my spelling/grammar issues Jun 20 16:59:54 black_13: that's not an sdk, you need to build an image Jun 20 17:00:21 the sdk is the ability to create applications of some type Jun 20 17:00:28 so again if you're satisfied with the availables default images it's fine, else you'll need to customize it a bit by creating your own bb Jun 20 17:00:29 is what you are saynig Jun 20 17:00:33 yes Jun 20 17:00:40 you've got a laptop Jun 20 17:00:45 several Jun 20 17:00:56 you compile some software for the arm device on it Jun 20 17:01:05 like you're writing a GUI for instance for some use case Jun 20 17:01:24 for such things the toolchain/sdk is very usefull Jun 20 17:01:27 I can I have really nice multi core dell machine that I have run yocto Jun 20 17:01:50 so you typically use the toolchain/sdk when you're developing an application that has to run on your ARM device Jun 20 17:02:54 for instance if you're writing an application to control a factory you use the SDK Jun 20 17:03:33 If instead you lack whatever package that you have on your laptop distro, you just write a .bb file and build it Jun 20 17:03:39 understood Jun 20 17:03:54 at this point my need pedagogy Jun 20 17:04:02 is pedagogy Jun 20 17:04:13 and if you want to include that package in an image, you create or modify an image Jun 20 17:04:21 (just to add this one more package) Jun 20 17:04:28 I have had questions come up and I didnt know the answer Jun 20 17:04:58 and it seems the BBB is a good cheap target Jun 20 17:06:33 There should also be yocto and openembedded channels, they probably know that better than me (because It's been years that I didn't use it) Jun 20 17:07:45 your right didn't think of that Jun 20 17:07:49 hi, I have a problem asking me for the root password Jun 20 17:07:59 but thanks for your time Jun 20 17:08:11 np Jun 20 19:29:32 Hello all. Maybe this is an odd one: I've been looking for a "commented dmesg", like those dvd from tv series episodes with "director's commentary"; Just for fun/learning. Never found anything, maybe because that's too boring for some people Jun 20 23:04:32 Nicolas__: heh, you mean "what the hell is all this output anyway" ? interesting though Jun 20 23:04:35 t Jun 20 23:20:25 zmatt: I find that a goldmine for learning Jun 20 23:39:29 well narrating a dmesg is a bit much to ask, and a lot of lines are not that interesting, but if there are specific lines for which you're wondering what the hell they mean you can collect them in a pastebin and I can see if I sched some light on them Jun 20 23:54:44 zmatt: Thanks Jun 21 00:08:54 zmatt: The dmesg of the day: [ 0.000000] Initializing cgroup subsys cpuset Jun 21 00:24:47 cpuset control-group subsystem lets you use control groups to restrict on which cpu cores processes may run. useful for balancing cpu load on massively multicore machines I guess, utterly useless on single-core machines like the beaglebone :P Jun 21 00:37:55 zmatt: Thanks :D Jun 21 00:42:09 Nicolas__: in case it's of any interest, here's dmesg from one of our beaglebones, which run a customized debian and custom-built kernel hence probably have a fairly different bunch of startup messages compared to the default on beaglebones. I annotated a little bit of it before I gave up on that: https://pastebin.com/RsB6N21G Jun 21 00:42:55 unlike the -ti kernel series, our kernel config is fairly specifically tuned for the beaglebone, so e.g. you don't find that silly cpuset cgroup there Jun 21 00:47:36 and if you like interesting food for thought, here's an interesting question: why does the kernel have an implementation of a mark-sweep garbage collector, and why is it located in the net/unix/ directory? ;) **** ENDING LOGGING AT Thu Jun 21 00:49:12 2018 **** BEGIN LOGGING AT Thu Jun 21 00:49:15 2018 **** ENDING LOGGING AT Thu Jun 21 00:49:16 2018 **** BEGIN LOGGING AT Thu Jun 21 00:58:16 2018 Jun 21 00:58:16 hint: you can attach file descriptors to messages sent across unix domain sockets, including file descriptors of unix domain sockets that have messages queued Jun 21 00:58:33 :) Jun 21 01:01:39 that is why is it located in net/unix, it is using possix's ipc ? Jun 21 01:02:42 unix domain sockets are just unix domain sockets Jun 21 01:03:02 posix ipc refers to different mechanisms Jun 21 01:08:13 I couldn't imagine how unix domain sockets are used in the garbage collector :( Jun 21 01:09:23 they're not "used in" the gc, the gc is *for* unix domain sockets **** ENDING LOGGING AT Thu Jun 21 01:11:42 2018 **** BEGIN LOGGING AT Thu Jun 21 02:42:17 2018 Jun 21 02:42:38 It works with the ip number, so maybe dns isn't configured properly **** ENDING LOGGING AT Thu Jun 21 03:00:01 2018