**** BEGIN LOGGING AT Wed Jun 08 02:59:59 2016 Jun 08 08:02:26 Hi, I need hardware advice to adapt pwm signal please. Jun 08 08:03:08 I don't know if this circuit http://i.imgur.com/al3Kqyh.jpg will destroy bbb pins Jun 08 08:08:02 dirkk: what are you planning to do with Vout? Jun 08 08:08:20 dirkk: the circuit looks a bit strange to me, but I see no potential for harming the bbb Jun 08 08:09:39 Ionakka: Vout will be filtered Jun 08 08:10:01 so you were afraid that when beagle supplies the pwm it could break? Jun 08 08:10:19 dirkk: keep in mind that this circuit has very asymmetrical output impedance for low vs high Jun 08 08:10:30 since it's basically open collector with weak pull-up Jun 08 08:11:15 so if you'd follow that up with an rc filter or such, you're going to get a lower value than you might expect Jun 08 08:12:53 (an active filter with high-impedance input would not have such a problem) Jun 08 08:13:12 zmatt: I'm not familiar with the notion of balanced circuit. Jun 08 08:14:05 zmatt: you mean add an opamp with Vout as input ? Jun 08 08:14:51 dirkk: imagine just putting a capacitor over Vout... when the pwm signal is low, Q2 will drive Vout low and the capacitor would discharge quickly, but when the pwm signal is high, the capacitor only gets slowly charged via resistor R4 Jun 08 08:22:13 an active low-pass filter using an opamp ( https://en.wikipedia.org/wiki/Low-pass_filter#Active_electronic_realization ) could replace the transistors and perform filtering at the same time, although this requires a bit more care to keep the bbb safe Jun 08 08:22:13 [WIKIPEDIA] Low-pass filter#Active electronic realization | "A low-pass filter is a filter that passes signals with a frequency lower than a certain cutoff frequency and attenuates signals with frequencies higher than the cutoff frequency. The amount of attenuation for each frequency depends on the filter design. The filter is sometimes called a high-cut filter..." Jun 08 08:22:30 can someone kick that fucking bot? :P Jun 08 08:22:44 zmatt: haha ;) Jun 08 08:24:16 zmatt: ok so R1 in wiki page is R3 in my circuit ? am I right ? Jun 08 08:24:39 in particular, a resistor from the (-) input of the opamp to ground seems like a good idea to me: combined with R2 this would make a voltage divider to avoid the (12V) Vout being presented on the BBB pin, which would be a very bad idea even with a big series resistor (R1) Jun 08 08:25:36 the two circuits aren't really comparable, but R1 actually serves the same purpose in both cases: limiting the output current Jun 08 08:28:46 note that the design parameters (cutoff frequency and passband gain) still don't fix component selection: if you multiply both resistors by some factor and divide the capacitance by the same factor then the net result is the "same" filter, but with different input impedance Jun 08 08:29:29 higher impedance means less current drawn from vin, but also greater noise sensitivity Jun 08 08:33:37 adding the resistor to ground, let's call it R3, doesn't hugely impact the analysis I think... from the filter's point of view it is like vin is multiplied by a factor of R3/(R1+R3) and R1 is replaced by 1/(1/R1+1/R3) Jun 08 08:39:01 alternatively, you can probably find a level-shifting push-pull driver and then you can use whatever kind of filter you want without worries Jun 08 08:41:48 zmatt: if I use just the opamp whenI put the 12V ? level-shifting push-pull why not but don't know ref which suits to the bbb Jun 08 08:41:58 *where I put the 12V Jun 08 08:42:13 the supply of the opamp Jun 08 08:42:42 and you'd need to select the right resistor values to get the desired output gain Jun 08 08:44:13 I think there are also free programs that can simulate such circuits to verify them Jun 08 08:44:18 zmatt: ok a rail to rail but my simulation doesn't work. wrong opamp I suppose Jun 08 08:44:32 I use ltspice right now Jun 08 08:44:37 that should work Jun 08 08:46:37 what exactly did you try? Jun 08 08:49:25 wiki page circuit with value RC of my filter and 10k in Vin Jun 08 08:50:01 ohh I just noticed something sneaky: that filter has negative gain Jun 08 08:50:23 try setting the negative supply of the opamp to -12V Jun 08 08:51:00 obviously since you don't want that an alternative filter is needed, but at least that should help to verify the basic idea is working Jun 08 08:54:27 https://en.wikipedia.org/wiki/Operational_amplifier#Non-inverting_amplifier Jun 08 08:54:28 [WIKIPEDIA] Operational amplifier#Non-inverting amplifier | "An operational amplifier (often op-amp or opamp) is a DC-coupled high-gain electronic voltage amplifier with a differential input and, usually, a single-ended output. In this configuration, an op-amp produces an output potential (relative to circuit ground) that is typically hundreds of thousands of..." Jun 08 08:54:33 gcl-bot: SHUT UP Jun 08 08:55:23 dirkk: looks like that non-inverting amplifier topology could be made into a filter by putting a cap in parallel to R2, but you should sim that of course Jun 08 08:56:45 zmatt: I'm trying this. still don't figure out why the first circuit output was strange (negative peak) Jun 08 08:57:03 if that works then you've got a non-inverting output, and moreover Vin is high-impedance Jun 08 08:57:53 (I'd still put a resistor in series just to be safe in case of power sequencing screwups) Jun 08 09:00:04 http://www.electronics-tutorials.ws/filter/filter_5.html scroll to "Simplified non-inverting amplifier filter circuit" Jun 08 09:05:33 who placed gcl-bot in here, in the first place? Jun 08 09:05:36 jkridner: ^^^^ Jun 08 09:06:13 dirkk: what is your goal? pwm output from BBB and drive something that needs 12V pwm ? Jun 08 09:06:33 KotH: since he's doing filtering, I think he wants a 12v "analog" output Jun 08 09:06:50 KotH: zmatt is right Jun 08 09:07:05 pilote dc with pwm Jun 08 09:07:51 ah Jun 08 09:09:15 a non-inverting active low-pass filter with passband gain 12v/3.3v seems like a good solution to perform level shifting, filtering, and buffering all at once Jun 08 09:09:22 dirkk: ok. do the following: in your circuit, replace R4 by a PNP transistor. connect the base of the new transistor and Q2 together. get rid of R3. decrease R2 to 1k Jun 08 09:10:02 dirkk: then start using R-C filter stages at Vout Jun 08 09:10:29 that would be the alternative I mentioned... a push-pull buffer followed by whatever filter Jun 08 09:10:30 dirkk: rule of thumb: increase the R by a factor of 10 with each stage Jun 08 09:11:31 dirkk: lowest resistor should not be below 1k, highest should not be much above 100k Jun 08 09:11:54 dirkk: depending on what circuit uses the voltage, you may need to have a buffer circuit there Jun 08 09:12:34 dirkk: that can be something as simple as an emitter follower, or an opamp Jun 08 09:12:57 dirkk: if you speak german, get a copy of tietze&schenk. it explains all this in detail Jun 08 09:13:48 hello all, I am writing udev rules for detecting usb ftdi device. and rule is working on if system is up and running but on a next boot it takes long time to boot device because of this rule.Does udev takes too long for detecting device? Jun 08 09:14:25 if a first-order filter suffices, wouldn't the active filter be a more elegant solution? it has high input impedance, low output impedance, and low component count Jun 08 09:14:40 though I'll admit I'm definitely no expect on analog hw Jun 08 09:14:43 kk_: udev doesnt detect devices, it gets notified by the kernel when a device is attached Jun 08 09:14:45 *expert Jun 08 09:15:00 kk_: have a look at the debug console to see what is taking time Jun 08 09:15:28 zmatt KotH Thanks for your advice Jun 08 09:15:35 zmatt: actually, if an output buffer is needed, then active filter is the better choice, yes Jun 08 09:15:53 zmatt KotH Let's simulate Jun 08 09:15:55 zmatt: i'd go for a multi-feedback structure though Jun 08 09:16:28 SUBSYSTEMS=="usb-serial", DRIVERS=="ftdi_sio", ACTION=="add", ENV{OLCDDEVNAME}="%k", RUN+="/bin/bash manual_script.sh" SUBSYSTEMS=="usb-serial", DRIVERS=="ftdi_sio", ACTION=="remove", ENV{OLCDDEVNAME}="%k", RUN+="/bin/bash manual_script.sh" Jun 08 09:16:47 it depends on how much margin there is between the pwm frequency and the bandwidth he requires Jun 08 09:17:13 if there's a big ratio there, then pretty much any filter will do Jun 08 09:17:58 kk_: ew, RUN should normally be avoided whenever possible, and running a script even more so Jun 08 09:18:15 @KotH : Sorry but debug console is not availabe. Is there any other solution Jun 08 09:19:25 kk_: "sorry i crashed into a moutain while flying blind. is there a solution not crashing into a mountain while flying bind?" Jun 08 09:20:09 zmatt: actually it depends on how much ripple you think is acceptable :) Jun 08 09:20:18 KotH: that too yes Jun 08 09:20:40 kk_: why are you invoking a script anyway? what's it doing there? Jun 08 09:21:19 @KotH Hey Thanks it works for me .After removeing RUN. Thanks a lot Jun 08 09:21:49 like I said, RUN is usually a bad idea Jun 08 09:22:19 script is running for starting firmware application. On usb attach. firmware should up and running Jun 08 09:23:04 use TAG+="systemd" Jun 08 09:23:33 then the device will become a systemd unit which you can have a service match on Jun 08 09:26:05 @KotH : Is there any alternative of RUN while writing a rule. Jun 08 09:26:27 you can also do something like ENV{SYSTEMD_WANTS}="myservice.service" Jun 08 09:26:41 to automatically start a service when the device is detected Jun 08 09:27:06 (should be combined with TAG+="systemd" otherwise it will not work) Jun 08 09:32:21 alternatively you can have the service enabled to start by default by use a BindsTo directive to make it depend on the device unit... that won't work when hotplugging but it should work if the device is present at boot (I think systemd gives devices a minute time to appear) Jun 08 09:33:43 s/by use/but use/ Jun 08 09:40:49 actually I think ttys already get a systemd tag by the default rules Jun 08 09:41:03 you can check with e.g. sudo systemctl status dev-ttyUSB0.device Jun 08 09:48:01 @zmatt using TAG and ENV i am not able to run myusb.service Jun 08 09:48:38 use systemctl to check on the status of the device and the service, check journal for errors Jun 08 09:55:22 works for me, http://pastebin.com/P7RvyEsW Jun 08 09:55:26 in combination with the udev rule Jun 08 09:55:32 SUBSYSTEM=="tty", ENV{DEVNAME}=="/dev/ttyUSB0", TAG+="systemd", ENV{SYSTEMD_WANTS}="myusb.service" Jun 08 09:57:00 (hardcoding the device name like that is a bit iffy, but okay as long as you know there will be only one ttyUSB device. it's possible to make a nicer solution of course, although the hard part would be figuring out what a good criterion is to match the device on) Jun 08 09:57:56 for USB devices: VID/PID Jun 08 09:59:12 KotH: well, if there are multiple ttyUSB devices, then they may very well have the same VID/PID Jun 08 10:00:15 true that Jun 08 10:00:22 the full device path would be a better choice probably... that only depends on where the device is located in the usb tree (i.e. which hub port) Jun 08 10:00:29 then the only way to discern them is either by using the bus path or by serial number Jun 08 10:01:06 matching on serial number would work for a one-time project but not anything that you intend to produce in any larger quantity Jun 08 10:04:35 matching on bus path is also something you get almost for free, since the device unit already gets made anyway: Jun 08 10:04:44 dev-ttyUSB0.device - FT232 Serial (UART) IC Jun 08 10:04:47 Follow: unit currently follows state of sys-devices-pci0000:00-0000:00:1d.0-usb2-2\x2d1-2\x2d1.4-2\x2d1.4:1.0-ttyUSB0-tty-ttyUSB0.device Jun 08 10:05:16 (that's on my laptop, obviously not on a bbb) Jun 08 10:06:01 anyway, gotto go, stuff to do Jun 08 10:07:45 Thanks zmatt Jun 08 10:13:52 zmatt: doing stuff is overrated ;-) Jun 08 10:13:55 zmatt: en guete! Jun 08 10:20:06 KotH: I tried this http://i.imgur.com/W6RJt3B.png . the output does not change when duty changes Jun 08 10:20:49 output is 4,XX V Jun 08 10:28:06 for how long have you simulated? Jun 08 10:29:45 you have an \tau of 20ms of your filter... so anything below 80-100ms will be too short Jun 08 10:33:20 anyone here experimented with an encrypted home on bbb with ecryptfs? will it suffer performance wise? Jun 08 10:38:35 samael: http://26-26-54.hardwarebug.org/30 Jun 08 10:40:07 Hi I want to ask is it possible to use beaglbone only for data acquisition and send data directly to a laptop connected via USB? Jun 08 10:40:36 righthearted: is there a reason why it should not be possible? Jun 08 10:41:28 I mean, without ssh connection Jun 08 10:42:39 Like access its pins "directly" Jun 08 10:43:09 righthearted: is there a reason why it should not be possible? Jun 08 10:45:06 Seems like my question is not clear. If it is possible to do so, are there any manuals on how to do so Jun 08 10:47:36 samael: if the kernel took optimal advantage of the hw aes accelerator then there should only be minor increase in latency and no decrease in throughput. in reality however the omap-aes driver doesn't support the XEX disk encryption mode that the hardware supports, and I have no idea whether it's used in any way or fashion at all for disk encryption, or even whether it would still be a net win when the Jun 08 10:47:42 driver overhead is taken into consideration Jun 08 10:49:18 KotH: and yeah, but people want me to do stuff. specifically the people that pay me Jun 08 10:50:10 righthearted: depending on what you actually want to do. Jun 08 10:53:07 zmatt: i'm experimenting right now. what would be a quick but meaningful test? Jun 08 10:53:38 i'm using bonnie++ Jun 08 10:54:28 so you have a benchmarking tool, what more do you want? :P Jun 08 10:54:46 ah, need to switch trains, afk Jun 08 10:55:31 just want to be sure it's the most appropriate tool :) Jun 08 10:55:54 bonnie* gives you good answers Jun 08 10:56:09 it tests for the performance metrics you are most likely interested in Jun 08 10:56:21 but how you interpret those results, is a different question ;) Jun 08 10:57:39 i'll proceed serially: let's face one problem at a time :) Jun 08 11:33:27 Hello everyone! question about compiling kernel modules on beagleboard xm. i have 4.4.1-armv7-x5 kernel from the debian 8.3 image on beagleboard.org and i want to compile a kernel module, but the headers for that version are not in the repo. i tried upgrading the kernel by installing a linux image for which there is a headers package in the repo, but it rendered the board unbootable. i also tried using headers from a higher version (4.4.2-armv7-x5) but it didnt w Jun 08 12:05:55 I want to connect sensors to control a model in blender and also send signals from blender to beaglebone and acttvate motors Jun 08 12:06:17 KotH: Jun 08 12:37:15 oh he left, I was gonna say "well, sounds like a plan. make it happen!" Jun 08 12:48:34 well.. some people never understand that you can make anything you can imagine. it might take some time to get it done. at times you might need to bend the laws of phyics. but in general, you can do it even if nobody else has done it before Jun 08 13:00:03 hi me again Jun 08 13:01:08 I would like to know if the pins are "safe" with this circuit http://i.imgur.com/mvsEkwN.png Jun 08 13:02:26 I supply the opamp with 5V, (for my app it's 12V but I cannot download spice model of the 12V supply opamp Jun 08 13:04:28 given the rather large size of the cap maybe use a schottky diode to vdd to discharge it in a safe way at powerdown (instead of via the internal diode of the IO, which is mainly for esd protection and not very tough) Jun 08 13:07:48 KotH: bending the laws of physics is tricky though... they might break and then physicists will get all angry at you and demand that you mend them again Jun 08 13:10:15 zmatt: where sould I put it ? before the filter ? Jun 08 13:10:22 *should Jun 08 13:52:12 that x-15 tho... Jun 08 13:52:34 anyone here using the Fedora Minimal image on a BBB? Jun 08 14:42:34 hello all. I'm running debian 8.2 with 4.1.24-bone-rt-r21 kernel. I've 2 problems: 1) the only way to enable the SPI overlay is at boot, adding to uEnv.txt "cape_enable=bone_capemgr.enable_partno=BB-SPIDEV0". I would like to enable SPI after boot, using echo "BB-SPIDEV0">$SLOTS. 2) SPI max frequency is always 16MHz, even if I modified the "spi-max- Jun 08 14:42:35 frequency" setting it to 24MHz. Can you help me to understand what's the problem. I can pastebin my dts if needed. thank you in advance Jun 08 15:06:04 hiya - anyone been using the "new" 16.04 image with wireless? Been trying it on one of our dev BBBs but cant get away from this new wlan naming :S Jun 08 15:08:28 putting net.ifnames=0 on your kernel cmdline will give you the old naming back Jun 08 15:09:41 doesnt work sadly Jun 08 15:10:17 Tried to post on the google group - but google ate my post.. will repost.. old names work on eth0 usb0 but not on wlan0 Jun 08 15:19:33 Our workarounds fixed these names for 14.04 and other kernels..just not anymore.. Kernel command line: console=ttyO0,115200n8 root=UUID=b32c4e38-221c-4cf5-b6c5-71ce64770caa ro rootfstype=ext4 rootwait coherent_pool=1M quiet cape_universal=enable net.ifnames=0 Jun 08 15:19:59 so the kernel is loaded with it..but it is still happening Jun 08 15:20:30 [ 17.406146] rtl8192cu 1-1:1.0 wlx4494fcfb3a21: renamed from wlan0 Jun 08 16:18:35 hi Jun 08 16:18:39 anybody here? Jun 08 16:18:44 shh Jun 08 16:19:41 sorry to interrupt Jun 08 16:19:44 lol Jun 08 16:19:55 I was just being silly Jun 08 16:20:16 ^^ Jun 08 16:20:30 ok, well, this is certainly looking perfectly normal... http://pastebin.com/YvFBV5cn Jun 08 16:22:23 btw I am struggling with beaglebone black and ostro project Jun 08 16:35:57 zmatt: buildscript gone wrong? Jun 08 17:18:06 KotH: no, it turns out that mkfs.ext4 nowadays makes 64bit filesystems by default, and u-boot handles those... ehm... very gracefully Jun 08 19:01:52 beaglebone green wifi, not casting ssid. any thoughts? Jun 08 20:22:24 I'm trying to set up spi0 with 6 cs-gpios, so I can speak with 6 spi devices, however I can't get it to work Jun 08 20:22:32 this is my device tree overlay: http://pastebin.com/4eUv2HvY Jun 08 20:22:42 which kernel version? Jun 08 20:22:48 wait Jun 08 20:23:22 support for cs-gpios in the mcspi driver was added relatively recently Jun 08 20:23:41 ah ok :/ Jun 08 20:25:28 "uname -r" says: 3.8.13-bone70 Jun 08 20:25:29 I think 4.4, but I'm not sure... I run 4.6 myself Jun 08 20:25:41 ah, heh no that ain't gonna work Jun 08 20:25:48 that is truly ancient Jun 08 20:26:00 ._. Jun 08 20:28:51 so your options are either upgrading or managing the chip selects manually as gpios from userspace Jun 08 20:31:32 yep, upgrading may be a good idea anyway, as the other arm board I have uses 4.4 Jun 08 20:32:18 however the device tree file on the link is ok? Jun 08 20:35:27 at first sight I don't see any obvious issue, except I don't know whether ti,spi-num-cs should be adjusted, and if so whether it should be 0 (the number of real chip selects used) or 1 (the number of channels used)... the only DT I have which uses cs-cpios has one "real" and one gpio chipselect so that muddies things Jun 08 20:36:00 and I'd explicitly declare pinmux for the pins used Jun 08 20:36:42 (i.e. also the ones muxed to gpio, not just those muxed to mode 0) Jun 08 20:36:46 ok, so put the directly into the spi0_pins_s0 block? Jun 08 20:36:53 *them Jun 08 20:37:12 yeah no reason to use multiple pin blocks Jun 08 20:38:44 btw if you're interested, I've made some stuff that makes writing overlays much nicer Jun 08 20:39:22 https://github.com/mvduin/py-uio/tree/master/dts ignore the examples dtsi files there since they're all for uio (since I made this stuff originally for py-uio) Jun 08 20:39:39 http://gerbil.xs4all.nl/spi1.dtsi but here's an example for enabling spi1 Jun 08 20:40:30 if you clone the git repo, go to its 'dts' subdir, download that dtsi and type make "spi1.dtbo" it builds an overlay from it Jun 08 20:41:00 bin/add-overlay spi1.dtbo then inserts the overlay using configfs (introduced in kernel 4.1) Jun 08 20:42:42 anyone running ubuntu16.04 with 4.4.9-ti-r25? or similar_ Jun 08 20:42:51 ? Jun 08 20:45:07 What ever I do I cannot get wlan0 back to being wlan0.. eth0 is eth0, usb0 usb0.. but wlan gets renamed to wl Jun 08 20:45:28 and net.ifnames=0 doesnt change that either Jun 08 20:45:29 arnvid: probably some udev rule Jun 08 20:45:33 ah, it combines the cpp preprocessor and dtc Jun 08 20:46:05 pyBlob: yes, together with a perl script that converts normal dtsi syntax to the horrid crap needed for overlays Jun 08 20:46:36 the devicetree overlay references <&spi0>, where can I find that one? Jun 08 20:46:47 zmatt: the old tricks with /etc/udev/rules.d/70-persistent-net.rules and /lib/udev/rules.d/75-persistent-net-generator.rules doesnt work.. Jun 08 20:47:54 arnvid: try hotplugging the wifi while running udevadm monitor -u -p Jun 08 20:47:58 it's the "4th" naming rule of udev that kicks in.. the "Names incorporating the interfaces’s MAC address (" Jun 08 20:48:06 lets see Jun 08 20:48:41 arnvid: if you can locate the file containing the actual rule, you can easily block or override it Jun 08 20:48:49 wow people still using 70-persisten-rules .. lol ?! Jun 08 20:48:51 you can probably also do so even otherwise actually Jun 08 20:49:17 veremit: we built our BBB systems with 12.04 originally Jun 08 20:49:26 pyBlob: overlays reference labeled nodes of the main device tree Jun 08 20:49:34 and been keeping the same "init" program since then Jun 08 20:50:09 pyBlob: usually ones in the core dtsi of the processor, am33xx.dtsi Jun 08 20:50:13 arnvid: nice Jun 08 20:51:03 pyBlob: https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.4.13-bone11/arch/arm/boot/dts/am33xx.dtsi Jun 08 20:51:47 we upgraded the base system to 14.04 earlier this year.. and made some code changes in our end until now and since 16.04 came out we wanted to look at that Jun 08 20:52:16 zmatt: /lib/systemd/network/99-default.link <-- apparently the two lines there makes it go bananas for wlan0 Jun 08 20:53:48 arnvid: those two lines look entirely sensible Jun 08 20:54:07 at least on my system Jun 08 20:54:42 [Link] NamePolicy=kernel database onboard slot path MACAddressPolicy=persistent Jun 08 20:54:47 yea .. I would think that too Jun 08 20:54:49 also, NamePolicy is not used if net.ifnames=0 Jun 08 20:55:01 but removing that file makes it come up with wlan0 vs wl1410928410 Jun 08 20:55:02 :) Jun 08 20:55:29 also: The name is not set directly, but is exported to udev as the property "ID_NET_NAME", which is, by default, used by a udev rule to set "NAME". Jun 08 20:55:59 i.e. NamePolicy merely makes the suggested name available to udev, there's still an udev rule doing the actual renaming Jun 08 20:56:32 zmatt: as I understand it from the udev page in this case the name is set on every boot based on the mac address Jun 08 20:57:13 actually that would require NamePolicy to be "mac" Jun 08 20:57:25 that one isn't even in the list you showed Jun 08 20:57:29 but that's whats happening Jun 08 20:57:29 wlx100d7fb73f66 Jun 08 20:59:31 i.e. udev doesn't even give a fuck what systemd is suggesting to it :P Jun 08 21:00:17 clearly on my laptop udev is straight-face ignoring systemd as well: Jun 08 21:00:19 the udev page suggest that this "functionality" ;) is not turned on by default but must be explictly turned on Jun 08 21:00:23 ID_NET_NAME=wlp2s0 Jun 08 21:00:23 ID_NET_NAME_MAC=wlx8c705a05accc Jun 08 21:00:23 ID_NET_NAME_PATH=wlp2s0 Jun 08 21:00:34 ok, then I'll try upgrading the kernel, do I have to take care of something special for that? Jun 08 21:00:35 this shows the path-based name would have been wlp2s0 Jun 08 21:00:49 zmatt: udev clearly loves us ;) Jun 08 21:00:53 and the path-based name has been chosen by systemd as suggested name, ID_NET_NAME Jun 08 21:01:02 2sec Jun 08 21:01:15 gonna re-enable that file and grab it to log Jun 08 21:01:19 hence, it got called "wlan0" Jun 08 21:01:27 ^_^ Jun 08 21:02:00 you can also do systemd test /sys/class/net/$ifname Jun 08 21:02:03 eh Jun 08 21:02:04 udevadm test Jun 08 21:03:27 pyBlob: no idea really... 3.8 is ancient history to me, and I also use a recent userspace (debian stretch) Jun 08 21:04:20 ok, so I'll just try if it works and reflash if it doesn't Jun 08 21:05:16 Network interface NamePolicy= disabled on kernel command line, ignoring. Jun 08 21:06:49 that's what I said Jun 08 21:07:37 yea but was still doing it but I think I found the actual origin now Jun 08 21:08:10 in this file /lib/udev/rules.d/73-special-net-names.rules Jun 08 21:08:18 ACTION=="add", SUBSYSTEM=="net", SUBSYSTEMS=="usb", NAME=="", ATTR{address}=="?[014589cd]:*", IMPORT{builtin}="net_id", NAME="$env{ID_NET_NAME_MAC}" Jun 08 21:09:01 udevadm test showed where it were picking it in the end Jun 08 21:09:24 tnx Jun 08 21:09:36 note that you can probably override all of it just by defining a rule that matches your device and does NAME:="wlan0" Jun 08 21:09:49 the dongles get swapped around Jun 08 21:10:05 so we need the if to be wlan0 regardless of which dongle gets put int Jun 08 21:10:14 yeah I didn't mean match a specific device Jun 08 21:10:56 zmatt: not sure how to do that:) usually dont need to fiddle so much with udev hehe Jun 08 21:11:30 ACTION=="add", SUBSYSTEM=="net", INTERFACE=="wlan0", NAME:="wlan0" Jun 08 21:11:31 possibly Jun 08 21:11:42 hm lets try Jun 08 21:13:21 or, if you consider the whole content of 73-special-net-names.rules to be a pile of garbage you can just override it by making an empty file /etc/udev/rules.d/73-special-net-names.rules Jun 08 21:13:58 prolly more likely to do that Jun 08 21:14:09 since BBB's are not likely to have idrac's any time soon Jun 08 21:14:19 or being run under vios Jun 08 21:15:43 you could see it as a challenge Jun 08 21:16:16 valid point .. Jun 08 21:16:50 plenty of challenges with BB's :) but its fun Jun 08 21:17:55 it is Jun 08 21:17:59 Wireless was always fun to overcome. Jun 08 21:18:15 worth writing up too .. there's a terrible derth of docs for BBB Jun 08 21:18:28 even if we just leave it to google to index them Jun 08 21:18:38 or there's elinux which we should make more use of Jun 08 21:19:22 there are still more idiotic udev rules left to fight Jun 08 21:19:41 When I started looking at BBB's derek molloy was my first go to place Jun 08 21:19:55 I still see some errors in journal because the rules don't understand how to parse mmcblk1boot0 and friends Jun 08 21:19:57 his pages was a good introduction Jun 08 21:20:16 bbl, train to catch Jun 08 21:20:23 tnx for the help zmatt Jun 08 21:23:30 yeah molloy's stuff has been a good stepping stone. . but I don't think he's kept up with the constant updates Jun 08 21:23:55 this is a platform where the software is changing fast .. and there's no huge 'software team' backing it up .. there's like.. 2/3 guys Jun 08 21:24:07 veremit, no - but he made sense of things for us Jun 08 21:24:35 arnvid: right and that;s half the battle Jun 08 21:24:39 especially in the start with the gpio's.. Jun 08 21:24:56 after the first battle with those.. someone put out 12.04 - and from there things changed drastically Jun 08 21:25:08 then you kinda gotta come here and ask teh dumb questions, and hope for zmatt or someone else helpful lol Jun 08 21:25:49 I should put a client here and stop by more often hehe been so longe since I frequented irc Jun 08 21:26:20 won't hurt to join the idlers ;) Jun 08 21:28:06 we killed our first BB the other day.. from wear poor thing didnt want to boot anymore.. couldnt even reflash it hehe Jun 08 21:29:54 borked a few rpi's over ht years.. but the BBB's keep on going all the time. Jun 08 21:38:50 that be that industrial am335x .. can't say the same for the bcm43xx Jun 08 21:41:17 ok, it only did the update to current 3.8.13 ... so have to get a microsd for tomorrow to be able to flash it Jun 08 21:41:37 thanks and gn8 Jun 08 22:49:15 at the office I think they managed to murder half a dozen BBBs by now :/ Jun 08 22:53:33 veremit: today I took a quick look at what TI's official linux sdk demo image for the am335x looks like... lol what a mess Jun 08 22:55:06 zmatt: outrageous .. and .. typical .. respectively. Jun 08 22:55:51 glitchy graphics, choppy mouse cursor movement, boot takes ages and actually failed once or twice Jun 08 22:57:36 disabling cpuidle seemed to get rid of a large fraction of the glitches (and associated lcdc fifo underrun errors in kernel log), but not all Jun 08 22:58:02 and yeah... one of the BBBs doesn't even give a power led blip of doom anymore, just nothing Jun 08 22:59:38 and another fails to boot... you do still see CCCC on the console but the 3v3b was disturbingly low, current consumption disturbingly high, and eventually the eMMC started to emit a suspicious odour Jun 08 23:28:11 zmatt: rofl Jun 08 23:28:33 zmatt: you should feed back on their shoddy SDK and point out how slick/efficient the BBB runs .. *cough* Jun 08 23:29:04 well this was *on* a BBB, they officially support it as a target Jun 08 23:29:14 (in addition to the evm) Jun 08 23:30:54 ofc .. thats jkr's work. But that's no excuse not to build a 'working' SDK in the first place Jun 08 23:31:11 But .. tbh .. that's totally typical of a silicon vendor I fear. Jun 08 23:31:20 "It Works" but barely. Jun 08 23:32:15 you'd think someone would have tried it and noticed that cpuidle is really broken Jun 08 23:32:45 (at least, the flavor where the cortex-a8 PLL is unlocked during cpuidle) Jun 08 23:36:18 lol ..no .. Jun 08 23:36:55 btw did you see https://www.youtube.com/watch?v=hB6eY73sLV0 Jun 08 23:37:07 hold up .. just lookin up an overlay Jun 09 00:04:20 lol nice vid zmatt Jun 09 00:05:55 yes, some people do really crazy shit ^_^ Jun 09 00:07:11 "I iteratively wrote 331 bytes ..." Jun 09 00:45:56 veremit: https://www.youtube.com/watch?v=TqK-2jUQBUY this one is also interesting, showing a bit of the sort of stuff that happens when they trigger a glitch Jun 09 01:23:57 zmatt: how do you find this c*** ?! lol Jun 09 01:27:33 rofl nice hacks Jun 09 01:32:48 the same guy's explanation of the credit warp hack goes even further down the rabbit hole Jun 09 01:39:26 oh noes lol **** ENDING LOGGING AT Thu Jun 09 02:59:58 2016