**** BEGIN LOGGING AT Thu Mar 16 03:00:03 2017 Mar 16 10:33:06 how is RNDIS interaface treated by Osx?? Mar 16 10:33:27 I mean as a CDC device or a HID device? Mar 16 10:57:54 can anyone please help??? Mar 16 14:24:27 anyone else notice that the sticker on the beaglebone blue says "beaglebonblue Mar 16 14:24:30 " Mar 16 14:24:54 hope nobody tries to eat it Mar 16 14:39:54 m_w: Well, I didn't notice until now.. The sticker in the mail in the gsoc google group looks OK though. Mar 16 14:41:53 But good to see you here, I got some questions relating to the serial terminal server project. Mar 16 14:44:09 I thought about different solutions for I/O buffering, the first one uses the PRU to read and buffer UART inputs; But as the software should be "generic for Linux based terminal servers" I guess we should not rely on the PRU. Mar 16 14:47:52 So the second approach would modify the UART driver (or use a daemon) to do the buffering. zeekhuge pointed out that it might even be possible to use DMA to copy the UART's FIFO to memory, but I was not able to find information on this one yet. Mar 16 14:48:40 m_w: propbalby noone read it as it is upside down :) kind of funny though Mar 16 14:50:08 Basically, the question is what do you think makes sense to retain the serial port connection and log all the I/O? Am I moving in the right direction? Mar 16 14:56:11 not sure why additional buffering would be needed Mar 16 14:59:35 there is a 16 byte fifo for each UART in the hardware Mar 16 15:01:29 are we worried that too much simultaneous serial activity will cause buffer overrun? Mar 16 15:02:01 Right, but the project description says something about buffered past I/O. Don't we need to maintain a buffer in memory in order to log all I/O on the UART pins? That way we can read this log when we connect to the bb with our pc Mar 16 15:02:40 let me look at the project description Mar 16 15:03:00 I copied it over from last years project list Mar 16 15:04:03 so we would want to buffer the data into a file in userspace Mar 16 15:04:19 That's what I thought Mar 16 15:04:35 no need for the PRU Mar 16 15:07:14 Would a user process be sufficient to read the I/O? I'm a bit worried about missing messages when a different process is dispatched.. Mar 16 15:07:22 https://github.com/beagleboard/linux/blob/4.4/drivers/tty/serial/omap-serial.c#L946 Mar 16 15:07:34 looks like DMA is already supported by the drivers Mar 16 15:08:03 so .. probably you will just need to configure the UART to write data onto a file. Mar 16 15:08:44 Oh that sounds nice. I'm currently learning about drivers in general, until now I have not looked into the UART driver Mar 16 15:22:34 OK I looked through it, but I will probably need some time to figure out how it really works. Mar 16 15:23:37 Anyway, the DMA approach sounds like a plan to me. Mar 16 15:26:24 I am pretty sure it is taken care of Mar 16 15:26:29 is is enabled by default so you don't have to worry Mar 16 15:26:32 I think you will just need to worry about how to configure the UART from the userspace. The drivers will do the DMA thing themselves. Mar 16 15:26:51 yeah .. what m_w said Mar 16 15:27:03 ordsen: you have a beaglebone black to experiment with? Mar 16 15:27:43 I just ordered one as I have only worked with Arduino before, so yes Mar 16 15:35:36 you probably don't have a serial port cape I am guessing Mar 16 15:36:33 no Mar 16 15:38:34 did you get the USB to TTL cable for the Linux console? Mar 16 15:40:33 no, not yet. Mar 16 15:42:25 But for experimenting before the project starts, I could try to connect my Arduino via UART Mar 16 15:43:04 do you have a 3.3V arduino? Mar 16 15:43:32 no but I have a logic level converter Mar 16 15:44:18 that would be a good experiment Mar 16 15:47:08 OK, I will report what I find out Mar 16 15:47:23 brb, have to reboot Mar 16 15:57:51 m_w: There is also another thing I'm not sure of: Is the goal of the project to write software which enables us to directly access (just) the UART devices on the bb from a PC? Or will we just SSH into the bb and then use the software to talk to the UART devices? Mar 16 15:58:13 The latter one is already possible Mar 16 15:59:47 both maybe Mar 16 16:00:53 the trick is making an interface that is more user friendly Mar 16 16:01:45 just ssh'ing into the target and running minicom is not going to be enough Mar 16 16:03:06 what does user friendly mean to you? Is it convenience methods which can be included e.g. into minicom? Or a UI to ease its usage? Mar 16 16:04:14 good question Mar 16 16:04:34 I think a user interface would be good to have Mar 16 16:06:19 perhaps a TUI or web interface to configure the terminals Mar 16 16:06:55 baud rate, bits, parity, logging file, etc Mar 16 16:08:09 another nice feature to have would be a hexadecimal viewer Mar 16 16:09:13 perhaps timestamps would be useful Mar 16 16:10:12 the trick is packaging the software in a way that you would expect from a consumer product Mar 16 16:12:33 So no matter whether TUI or web interface, the software runs solely on the bb. The deliverable is an image which, once installed, will provide the user immediately with an UI to configure and access its UART devices Mar 16 16:14:11 For logging I thought of something similar to the LogCat on Android devices. So there is a time stamp, name of the process (here name of the device), a tag (here configured by the user or none) and the actual message Mar 16 16:15:21 the logging could happen off the target actually Mar 16 16:15:54 that is not a bad idea as we are limited to a microsd card worth of data Mar 16 16:16:45 off the target? so we don't log which device made the message, just the message? Mar 16 16:18:11 no Mar 16 16:18:36 each device would be logged to a separate log Mar 16 16:19:02 but have the log reside on a server or host pc Mar 16 16:19:18 instead of on local storage on the beaglebone Mar 16 16:20:20 just need to find an efficient way to do such a thing Mar 16 16:20:32 just an idea Mar 16 16:21:26 Oh ok I understand Mar 16 16:25:43 perhaps you should take a look at what other commercial terminal server product provide Mar 16 16:27:07 Yeah, I was about to ask you what consumer will expect form a consumer product. maciejjo pointed me to moxa; but I will check out what else is out there Mar 16 16:30:28 As for the UI, I like the idea with the web interface. So it will be like browsing the Beagle, but instead of showing you that presentation it will bring you to the interface Mar 16 16:32:00 then you can access it from your cell phone :D Mar 16 16:34:40 yeah :D Mar 16 16:38:27 there are plenty of web interface demos for the beaglebone black Mar 16 16:40:11 ordsen: done any web programming in the past? Mar 16 16:44:48 uhm I wrote a backend in php and a small web game with javascript and HTML. Mar 16 16:48:11 okay I would take some time to get the web technology that you want to use installed on the board and get a basic interface working Mar 16 16:48:33 there are so many ways to do web interfaces Mar 16 17:03:00 alright, I already created a github repo for my gsoc preparation, I will upload my examples there so that you will be able to see them Mar 16 17:42:44 Hi, I just looked through some of the beagleboard projects and they look really exciting! Mar 16 17:44:32 I've been working on building a compiler for a course this semester and would love to be able to learn more about compiling and also writing code that is actually useful, however, I'm a little overwhelmed and not sure where to start. Mar 16 18:08:08 m_w: hi, poketbone still alive? Mar 16 18:21:42 pmezydlo: I think so Mar 16 18:22:41 Could you tell me more about porting the icestorm toolchain to the BeagleBone Mar 16 18:22:45 if you have time Mar 16 18:23:16 well it is just a matter of compiling the tools using the arm compiler Mar 16 18:23:34 I am pretty sure someone has done it with the raspi Mar 16 18:24:40 now all is clear Mar 16 18:26:56 do we plan to add external oscillator on BeagleWire Mar 16 18:29:11 50 MHz should be good for me, I'm not sure if iCE40 has oscillator and pll Mar 16 18:53:29 lsnaik: it looks okay so far Mar 16 18:53:51 but why add another data server? Mar 16 18:54:12 seems like you mentioned quake catcher and then forgot about it Mar 16 19:20:35 pretty much Mar 16 19:21:33 and you don't need to priv-mesg this is a community collaboration kind-of-a-thing Mar 16 19:23:01 the bionc client normally sends/receives data but a way to archive local data should be available Mar 16 19:23:34 maybe a dav/other client interface Mar 16 19:23:48 keep-it-simple-and-robust... Mar 16 19:24:23 pmezydlo: 50Mhz is probably fine Mar 16 19:25:03 I will look into the hardware stuff a bit more now than I am settled into my new place Mar 16 19:25:04 all that stuff (dav/samba/etc) is already builtin to the os so just include sane example config/document i think Mar 16 19:25:56 *but* local web server display/config is definitely "in" Mar 16 19:26:58 m_w: yeah We have a lot of time until gsoc Mar 16 19:27:48 well the proposal submission process starts soon Mar 16 19:28:59 relocated? Mar 16 19:29:37 m_w: my proposal is almost ready Mar 16 19:30:38 m_w: where do you live now? Mar 16 19:30:58 in 2 days I will send my proposal to you. Could you check it Mar 16 19:32:06 Carbondale, IL again Mar 16 19:32:18 sure Mar 16 19:32:36 nerdboy: okay, I will work on your mentioned points and update my proposal Mar 16 19:33:04 for now I have it in google docs, I think I will make elinux page after your review Mar 16 19:38:03 m_w: We can make the board during community bonding, if the project has been accepted Mar 16 19:38:15 okay Mar 16 19:38:25 I would like to make the board either way actually Mar 16 19:39:46 lsnaik: bonus points from jason of you use the existing beagle-y js web stack Mar 16 19:39:59 bonescript too Mar 16 19:40:23 also bonus points for good use of PRUs Mar 16 19:40:48 maybe dfts in this case Mar 16 19:41:56 looks like we might need to use a BGA for a decent LUT count Mar 16 19:42:45 iCE40HX8k? Mar 16 19:44:07 I do not want my bridge to use more than 500LUTs Mar 16 19:45:02 yes the 8k Mar 16 19:45:54 I want to be close to meeting the specs of this board: https://www.element14.com/community/docs/DOC-69215/l/fpga-development-board-cape-for-the-beaglebone Mar 16 19:46:22 minus the differential signalling stuff Mar 16 19:47:17 sdram or sram? I think sram is better for saving LUTs Mar 16 19:47:44 nerdboy: okay Mar 16 19:47:57 well that is true but then how much can we get at a reasonable price? Mar 16 19:48:07 nerdboy: dfts? Mar 16 19:48:37 pmezydlo: and does that limit the applications? Mar 16 19:50:43 time domain <-> frequency domain Mar 16 19:50:55 m_w: no but We must use an additional block for refreshing, Mar 16 19:51:28 nerdboy: okay Mar 16 19:52:07 hard to tell how many LUTs something takes Mar 16 19:52:19 https://en.wikipedia.org/wiki/Fourier_transform Mar 16 19:52:20 [WIKIPEDIA] Fourier transform | "The Fourier transform decomposes a function of time (a signal) into the frequencies that make it up, in a way similar to how a musical chord can be expressed as the frequencies (or pitches) of its constituent notes. The Fourier transform of a function of time itself is a complex-valued function of frequency..." Mar 16 19:52:49 lsnaik: keep in mind it has to make sense Mar 16 19:53:32 * nerdboy just throwing out general "could use X" where X is a beagle-y thing Mar 16 19:53:57 bonescript does give web access to gpios Mar 16 19:54:25 so maybe a setup/test/status thing for that Mar 16 19:55:24 nerdboy: yes, that's possible Mar 16 19:55:25 there's always some up-front analysis of problem/project requirements Mar 16 19:56:39 ie, initially in the (final) proposal, and more detailed analysis should be up-front in the project schedule Mar 16 19:57:51 okay Mar 16 19:57:53 first the analysis, then later milestones for "implement x,y,z features" since you won't know the specifics until later Mar 16 19:58:42 contingency time should probably be testing time, knowing in advance that some things will hit bumps Mar 16 19:59:24 agile-y feature/milestone adjustments will happen too... Mar 16 20:00:18 so be careful about over-scoping since time is fixed Mar 16 20:01:53 okay Mar 16 20:45:02 hey tlwoerner ready for some serious mentoring? Mar 16 20:45:22 m_w: sure, bring it! Mar 16 20:46:18 m_w: did you go to embedded world too? Mar 16 20:46:42 m_w: actually my bbblue just arrived from mouser an hour or so ago! :-D Mar 16 20:46:54 tlwoerner, nah I am saving the rest of my conference budget for other events Mar 16 20:47:11 I got a blue for free as ELC Mar 16 20:47:12 at Mar 16 20:47:29 no fair... Mar 16 20:47:33 wasn't supposed to tell anyone Mar 16 20:47:40 now I know why :D Mar 16 20:48:07 tlwoerner, you ever get around to using the robomezzi? Mar 16 20:48:31 m_w: i also got a bbgreen from seeed via mouser, and mouser made me fill out some weird export control form for the green :-S Mar 16 20:48:47 m_w: uhh.... not yet :-D Mar 16 20:49:14 tlwoerner, so you just like to collect the boards then Mar 16 20:49:33 he who dies with the most borads wins Mar 16 20:50:39 m_w: haha! i want to be cremated with all my boards!! :-S Mar 16 20:50:56 mostly to know these things are going to burn in hell with me! Mar 16 20:50:57 that would be quite toxic Mar 16 20:51:09 probably not a good idea... Mar 16 20:51:41 * nerdboy definitely knows too much Mar 16 20:51:58 excess lifetime cancer risk anyone? Mar 16 20:51:59 he doesn't have to worry about it, he'll be dead Mar 16 20:52:15 live in CA? Mar 16 20:52:40 then you probably already have cancer Mar 16 20:53:43 um, mostly it's worse elsewhere Mar 16 20:54:19 * nerdboy can't remember any top superfund sites in california Mar 16 20:54:33 or did you mean skin cancer? Mar 16 20:54:46 could be right there... Mar 16 20:55:27 I was joking about the known to the state of CA to cause cancer thing Mar 16 20:57:07 like, worse elsewhere *and* they don't have to tell you about it Mar 16 20:58:28 the info card that comes with the bbblue should metion how many amps the 12v needs i'm betting it's something like 2.5A? Mar 16 20:58:42 i don't need the 12v just to twiddle leds though? Mar 16 20:59:27 m_w: in retrospect i should have spent more time at ELC asking about the whole mentoring thing Mar 16 21:01:24 tlwoerner, the 12V is regulated down to twiddle the LEDs Mar 16 21:01:40 blue requires a 12v brick? Mar 16 21:02:06 nerdboy: that's what the sheet says Mar 16 21:04:50 gotta go, time for a fancy dinner with the missus, either McD or Swiss! getting fancy tonight! Mar 16 21:05:03 it uses a switching regulator for 5V Mar 16 21:05:07 (we got coupons for both, that's why we have a choice) Mar 16 21:05:11 udoo (arm plus sam3) needs 12v Mar 16 21:05:27 but the freescale part can run on 5v Mar 16 21:05:32 so you could probably go lower Mar 16 21:52:52 M_w free isn't quite right. Didn't you volunteer to write rest of the kernel drivers? Mar 16 21:55:04 jic23: I promised to help mainline it Mar 16 21:55:15 which is important Mar 16 21:56:00 I am not really a fan of the crazy scripts used for creating the kernel they deploy with the debian image Mar 16 21:57:30 True enough. Mar 16 21:58:04 Fiddly but is working out which patches matter,.. Mar 16 21:58:26 For a particular usecase. Mar 16 21:58:43 So mainline away! Mar 16 21:59:54 Though sounds like we may have an intern too help out:) Mar 16 22:05:20 yeah that'll be nice Mar 16 22:10:03 Hmm be nice to tidy up that magn stuff that never got finished.. Mar 16 22:10:36 which one? Mar 16 22:11:04 Mpu9250 weirdness to handle the slave busses... Mar 16 22:11:19 ah yes Mar 16 22:11:55 Patches were nearly good to go iirc. Dropped on floor by a job change i think... Mar 16 22:12:14 it appears that the BBBlue is using I2C to the MPU9250 so it should be easier to get working Mar 16 22:12:27 No fun :) Mar 16 22:12:31 Wo Mar 16 22:12:33 but using SPI it requires some magic Mar 16 22:13:05 Was pretty well handled i think. Just last details to tidy up. Mar 16 22:13:16 I have a board that uses SPI with the MPU9250 so maybe I can push it along Mar 16 22:13:22 Ciol Mar 16 22:13:25 Cool. Mar 16 22:13:40 who did the original patch series? Mar 16 22:13:47 Err.. Mar 16 22:14:12 I don't remember seeing it on the list Mar 16 22:14:27 there are two good spi interfaces for 9259 Mar 16 22:14:39 python one and c++ one Mar 16 22:14:51 *9250 even Mar 16 22:14:52 Crestez Dan Leonard. Mar 16 22:15:13 were they using a MFD driver to handle to I2C over SPI? Mar 16 22:15:25 Yeah, but fiddly to do it generically Mar 16 22:16:00 https://git.kernel.org/pub/scm/linux/kernel/git/jic23/iio.git/log/?h=experimental-beaglefun Mar 16 22:16:26 https://github.com/Nick-Currawong/RTIMULib2 <= this one has documented support for beaglebone now Mar 16 22:16:47 Doing the mag is easy,but the slave busses are generic. Mar 16 22:17:07 Not as nasty as some of the at parts though... Mar 16 22:17:11 St Mar 16 22:17:18 nice full calibration even Mar 16 22:18:05 Will have a dig sometime.. thanks nerdboy. Mar 16 22:18:53 builds fine in beaglebone but you need cmake... Mar 16 22:19:01 s/in/on/ Mar 16 22:19:38 students take note of the pull request to update the docs Mar 16 22:20:28 Got to run. Night all. Mar 16 22:20:31 we want to see appropriate use of the tools on github Mar 17 00:10:53 Hi all, I'm another student interested in participating in GSoC 2017. I've browsed the ideas wiki page and it seems like the projects listed have already garnered quite a bit of attention on the mailing list. Mar 17 00:11:17 As a result I've spent some time thinking of my own ideas, one of which is extending and expanding Rust support for the BeagleBone. Mar 17 00:11:47 I've already used Rust to develop programs on the BeagleBone and it's been a very pleasant experience, but the currently available libraries are short in critical areas (i.e. ADC, UART, easy-to-use PWM etc). Mar 17 00:12:19 Is the extension and creation of new Rust libraries something the community would be interested in or is my attention better placed elsewhere? Mar 17 00:18:09 i'd say qualified "yes" but beagleslackbot can give a better/more definitive answer... Mar 17 00:18:26 you'd need a mentor for that too Mar 17 00:19:43 maybe you could fill up the whole time period if you did rust/lua/go Mar 17 00:20:18 There was a guy on scale that could potentially mentor Go Mar 17 00:20:23 s/on/at/ Mar 17 00:20:28 <-- bad with names Mar 17 00:23:02 Go would be fun, but afaik there's already a pretty extensive Go lib Mar 17 00:23:31 ah yes, here it is Mar 17 00:23:31 https://github.com/btittelbach/go-bbhw Mar 17 00:23:49 Seems like it could still be extended with some more capability Mar 17 00:23:58 Also, forgive me for the noob question, but who's beagleslackbot? Mar 17 00:39:20 Oops, accidentally put my laptop to sleep. Did I miss anything? **** ENDING LOGGING AT Fri Mar 17 03:00:01 2017