**** BEGIN LOGGING AT Wed Mar 16 02:59:58 2016 Mar 16 04:41:33 Good morning ! Mar 16 04:42:41 * ZeekHuge has been really busy with his college stuff - minor project, assignments, labs etc ... really exhausting ... Mar 16 04:55:13 Hey m_w ! there ? Mar 16 05:04:53 I am sort of here Mar 16 05:05:11 you get any good feedback yet? Mar 16 05:05:44 ds2's question... Mar 16 05:17:49 sorry my wifi got disconnected Mar 16 05:22:43 Hello Everyone Mar 16 05:22:43 ZeekHuge: you still there? Mar 16 05:23:05 yes :) Mar 16 05:23:14 hi Mar 16 05:23:24 Can anyone here guide me regarding the "Linux kernel support for embedded devices and interfaces" projec Mar 16 05:23:30 what is dcc? Mar 16 05:23:59 I am interested in working on it but I am unable to find any more details about the project Mar 16 05:24:18 Manmohan: you have any kernel experience? Mar 16 05:24:42 Yes I have worked with arm linux kernel before Mar 16 05:25:10 to what extent? Mar 16 05:25:19 I have done some boot time opitmization, Initramfs alterations and driver module development Mar 16 05:25:50 any mainline patches? Mar 16 05:26:07 I did these as my internship at TI Mar 16 05:26:13 so the patches at not mainline Mar 16 05:26:19 they were TI specific Mar 16 05:26:41 sound like a fair amount of experience Mar 16 05:27:09 Currently I am doing a course on Linux kernel, so I also have fair idea about the memory management in the kernel Mar 16 05:27:17 m_w, did you see the post ? on the ml ? Mar 16 05:27:45 how are you wanting to contribute? what subsystems are you interested in? Mar 16 05:28:05 ZeekHuge: yes Mar 16 05:28:30 Actually that was what I wanted to ask. Mar 16 05:28:46 ZeekHuge: you should reply to Hunyue Mar 16 05:29:03 Are there any specific requirement Mar 16 05:29:38 I am also an embedded developer, I have worked mostly with Cortex-M series and arduino. Mar 16 05:29:41 ahh ... I didnt think about DMA ... Mar 16 05:30:29 Manmohan: it is just a general topic so it is up to the student to find a good place to contribute based on their experience or interests Mar 16 05:31:01 yes, please reply via email Mar 16 05:31:20 the DMA is not going to help much for the short burst transactions that go to converters Mar 16 05:31:51 Ok so any project involving development of support for a embedded device will be fine? Mar 16 05:32:02 ohk, so DMA is like a small mcu ? that executes sequentially .... is it ? Mar 16 05:32:03 Which has previously not been done Mar 16 05:32:34 Manmohan: which embedded device are you thinking? Mar 16 05:32:48 m_w: yes but neither will bouncing off the PRU Mar 16 05:33:09 ZeekHuge: yes... a small MCU that does nothing more then memcpy in response to an event Mar 16 05:34:03 ohk, so to copy data, it get interrupted, in form of DMA requests. and then copies the data Mar 16 05:34:12 yes Mar 16 05:34:21 I dont have a specific device in mind currently. Mar 16 05:34:31 so it would be a huge waste to burn the PRUSS for that purpose Mar 16 05:34:49 I am actually still confused about the requirement Mar 16 05:34:54 at higher rates, wont it get saturated ? i mean there will be a large number of interrupts per sample. Mar 16 05:34:56 can you just give a small example Mar 16 05:35:13 ZeekHuge: clarify please... not sure what you are asking Mar 16 05:36:09 since the DMA is an mcu in itself, wont it get saturated, resulting in loss of data or samples, if the data transfer rate is high ? Mar 16 05:36:27 Manmohan: improving the functionality of existing drivers perhaps or helping mainline out of tree beagle board support Mar 16 05:36:52 saturated = lot of DMA requests. Mar 16 05:37:26 By mainline drivers you mean the general SPI, UART, I2C etc. drivers right? Mar 16 05:38:08 *existing drivers Mar 16 05:38:11 mainline kernel = linus's tree Mar 16 05:38:38 ZeekHuge: not really Mar 16 05:38:39 there are features in the beagle board linux tree that are not in the mainline kernel Mar 16 05:38:51 DMA is designed for this kind of thing Mar 16 05:39:10 remember it is used for stuff like ethernet too Mar 16 05:39:31 okay Mar 16 05:39:59 nothing too significant is not in mainline Mar 16 05:40:03 I would probably find out who posted that idea and have them ellaborate Mar 16 05:40:16 I'd describe it as tweaks to suit our fickled tastes ;) Mar 16 05:40:26 yeah Mar 16 05:40:28 That would be great :) Mar 16 05:41:10 But from what I got, I think it means that some tweaks that are just for beaglebone we need to make them generic to work on other arm-linux platform as well Mar 16 05:41:18 am I right? Mar 16 05:41:55 ds2, what would you suggest then ? Mar 16 05:42:44 ZeekHuge: suggests in what sense? alt. ideas? what to eat for lunch/dinner? next year's vacation? Mar 16 05:42:51 :D Mar 16 05:43:40 Hi :) I'm new to GSoC, Chinese Students~ Mar 16 05:44:10 ds2, changes to make this idea useful. Mar 16 05:44:44 ZeekHuge: one thing that may be useful (donno what other people think) - using the PRUSS to improve sensor resolution Mar 16 05:44:54 ZeekHuge: are you familiar with oversampling? Mar 16 05:45:29 kind of , I read about it while doing research for the above idea . Mar 16 05:45:59 and also, in my course. Mar 16 05:46:15 Manmohan: send a request for more information on the mailing list Mar 16 05:46:50 one thing that might be useful is creating a library that can create 16bit data from a 14bit or less inertial sensor (say, accel, gyro, etc) Mar 16 05:47:02 I already sent a mail regarding that on 10th March Mar 16 05:47:13 no responses? Mar 16 05:47:15 Hanxiao: hi Mar 16 05:47:15 most of those sensors can run much faster then needed Mar 16 05:47:17 nope Mar 16 05:47:21 and there is a MAC on the PRUSS so... Mar 16 05:47:29 yes... Mar 16 05:47:31 Hanxiao: I am Chinese too Mar 16 05:47:33 Should I send another mail? Mar 16 05:47:41 yes Mar 16 05:47:54 ok will do that. Mar 16 05:48:03 Thanks alot :) Mar 16 05:48:27 Yes, so good to see the mentor from my tageting project here~ Mar 16 05:48:41 ZeekHuge: what might be interesting is to configure say an accel for +/-8G and have it provide the same resolution as the +/-2G (you oversample to create the extra bits) Mar 16 05:48:49 so that gives you extra dynamic range Mar 16 05:50:04 that would still involve communicating through the McSPI right ? Mar 16 05:50:16 nope Mar 16 05:50:23 no reason to waste the McSPI Mar 16 05:50:37 as there is limitation to the bitbanging speed by the PRUs ? Mar 16 05:50:46 the PRUSS is built for bitbanging Mar 16 05:51:13 there is a hardware shift function, IIRC Mar 16 05:51:23 but I don't see that as a huge barrier regardless Mar 16 05:51:45 McSPI usage is a waste of resources and is expensive (out of PRUSS access hits a shared bus Mar 16 05:54:06 ohk, and ds2, how about a mentor for this ? m_w , what do you think ? Mar 16 05:55:23 well it is interesting Mar 16 05:57:00 I was kind of the messenager on that offloading idea Mar 16 05:57:16 so don't shoot me for wasting your time :) Mar 16 05:58:07 No problem, that made me read a lot about IIO, drivers, PRUs and SPI. :) Mar 16 05:58:13 I think that the idea of offloading sensor communication to a real-time core has its merit Mar 16 05:58:22 i could help I suppose Mar 16 05:58:42 there needs to be more then just comms Mar 16 05:59:33 how much data can the PRU buffer? Mar 16 05:59:55 they have 30 * 32 bit registers Mar 16 06:00:13 and 3*30*32bit register banks called scratchpad Mar 16 06:00:54 perhaps we could do filtering Mar 16 06:01:15 filtering? Mar 16 06:01:16 a single scratchpad, ie 30*32bit register can be swapped in parallel and in single cycle. Mar 16 06:01:32 FIR, IIR Mar 16 06:01:42 how many taps FIR? Mar 16 06:01:47 and at what datarate? Mar 16 06:02:02 good questions Mar 16 06:02:19 I don't see that much value in running filters on there as quite a few of the sensors have filtering ability Mar 16 06:02:32 this is true Mar 16 06:02:38 sensor fusion? Mar 16 06:03:33 HE HE HE HE HE Mar 16 06:03:42 good luck fitting a decent fusion in the PRUSS Mar 16 06:04:03 I suppose it depends, what do you want out of the fusion? Mar 16 06:04:10 a heading? a quat? Mar 16 06:04:10 :-/ Mar 16 06:04:23 there are single chip sensors that will output heads and quats Mar 16 06:04:32 that is also true Mar 16 06:06:34 ohk, so should i start my research on the oversampling and resolution idea ? Mar 16 06:07:24 * ZeekHuge will be leaving for college in 15 minutes, need to talk to the HOD about the college project. Mar 16 06:08:00 ds2, since I am starting late on this, any useful links/resources ? Mar 16 06:10:04 nothing off hand, sorry Mar 16 06:10:44 ahh... ohk, m_w do you have anything ? Mar 16 06:12:02 oversampling is standard signal processing technique Mar 16 06:12:31 http://www.atmel.com/images/doc8003.pdf Mar 16 06:12:44 http://www.ti.com/lit/an/slaa323/slaa323.pdf Mar 16 06:13:51 bust out any data acquisition book Mar 16 06:14:10 Ohk, will see in the college library :) Mar 16 06:14:12 thank you Mar 16 06:14:22 will be leaving in 10 minutes :) Mar 16 06:14:32 https://www.silabs.com/Support%20Documents/TechnicalDocs/an118.pdf Mar 16 06:15:26 I will contact the IIO maintainers about the offloading thing Mar 16 06:16:43 maybe you could do something with a FPGA cape if the PRU is a waste of time Mar 16 06:18:53 I am sorry m_w, I couldnt contact the IIO maintainers. I am really busy with some stupid college stuff ! Mar 16 06:33:11 alexhiam, it would be great if you would say something on the idea of "oversampling and increasing resolution" suggested by ds2 . :) Mar 16 06:33:27 Thank you m_w , thank you ds2 :) Mar 16 06:33:37 * ZeekHuge is leaving. Mar 16 08:54:35 ds2, You there? Mar 16 09:13:14 * nerdboy still has a copy of schwarz Mar 16 09:14:11 Hey nerdboy , would you like to say something on the idea suggested by ds2 ? any thing you would like to share on that ? Mar 16 09:14:34 not before sleep... Mar 16 09:20:38 ask me again tomorrow/later Mar 16 09:20:49 sure :) Mar 16 13:50:54 jkridner: got the irc-gitter bridge up and running Mar 16 13:51:17 #beagle-gsoc bridging to here right now: https://gitter.im/graycatlabs/beagle-gsoc Mar 16 13:53:38 and now I have this annoying irc bot Mar 16 13:53:50 gcl-bot: .weather Mar 16 13:53:50 Yahoo! Weather - Brattleboro, VT: Cloudy, 8°C (46°F), Humidity: 71%, Light breeze 3.1m/s (↓) Mar 16 13:54:52 gcl-bot: .w beagleboard Mar 16 13:54:53 [WIKIPEDIA] BeagleBoard | "The BeagleBoard is a low-power open-source hardware single-board computer produced by Texas Instruments in association with Digi-Key and Newark element14. The BeagleBoard was also designed with open source software development in mind, and as a way of demonstrating the Texas Instrument's OMAP3530 system..." | https://en.wikipedia.org/wiki/BeagleBoard Mar 16 14:00:52 ZeekHuge: what about oversampling? Mar 16 14:16:15 gcl-bot: .w beaglebone Mar 16 14:16:15 [WIKIPEDIA] BeagleBoard | "The BeagleBoard is a low-power open-source hardware single-board computer produced by Texas Instruments in association with Digi-Key and Newark element14. The BeagleBoard was also designed with open source software development in mind, and as a way of demonstrating the Texas Instrument's OMAP3530 system..." | https://en.wikipedia.org/wiki/BeagleBoard Mar 16 14:16:48 alexhiam: shouldn't the bridge go to beagleboard/beagle-gsoc? Mar 16 14:17:13 yeah, I was just testing with my own gitter Mar 16 14:17:29 k. give me some install instructions and I will replicate. Mar 16 14:20:01 jkridner: if you create beagleboard/beagle-gsoc I can just keep running it on my server Mar 16 14:20:43 doh... beagle-gsoc is a private repo. Mar 16 14:21:07 oh right... can there be a public room with the same name? Mar 16 14:21:32 I'm also a member of that repo though, shouldn't I be able to see the room? Mar 16 14:23:23 maybe https://github.com/jadonk/gsoc-application should have been moved over to beagleboard? Mar 16 14:23:58 I guess we could create a new repo. Mar 16 14:24:13 we don't need a repo for it though, you can just create a room Mar 16 14:24:54 oh? Mar 16 14:25:19 I went ahead and created a repo. Mar 16 14:25:21 yeah, when you click 'create a room' in gitter you can select repository or channel Mar 16 14:29:11 restarting the bot... Mar 16 14:29:58 is it working? Mar 16 14:30:11 [alexanderhiam] :thumbsup: Mar 16 14:31:09 jkrinder: you want #beagle as well? Mar 16 14:35:55 [alexanderhiam] oh... it's not gonna look too pretty in irc when someone posts code in gitter or slack... Mar 16 14:36:51 [alexanderhiam] ```def foo(): return "this won't look good in irc"``` Mar 16 14:37:49 and it doesn't get highlighted on slack :( Mar 16 14:39:00 perhaps the bot should be a bit smarter and detect code, and maybe just put a link to the gitter/slack room or something Mar 16 14:41:03 [alexanderhiam] [README](https://files.gitter.im/beagleboard/beagle-gsoc/cIuG/README) Mar 16 14:41:23 oh cool, gitter file uploads bridge nicely Mar 16 14:43:31 alexhiam, We would be using Kbuild for compiling the driver right? Mar 16 14:44:53 yeah Mar 16 14:46:36 Okay,Thanks.I would be putting the hello world driver up on github in a couple of hours and then share the link Mar 16 14:47:39 great Mar 16 14:55:35 alexhiam: not sure what repo would make sense there. Mar 16 14:55:48 just a general channel? Mar 16 14:56:13 k. https://gitter.im/beagleboard exists now. Mar 16 14:57:41 so you can create room -> channel -> set owner to beagleboard, name it and set permission to public Mar 16 14:57:59 guess I just need to figure out how to set permission to public. Mar 16 14:58:39 once you set the owner you'll have the option Mar 16 15:00:59 https://gitter.im/beagleboard?utm_source=share-link&utm_medium=link&utm_campaign=share-link Mar 16 15:01:43 still only seeing beagle-gsoc Mar 16 15:03:03 guess it can't be made public... Mar 16 15:03:06 made beagleboard/general Mar 16 15:04:50 restarting the bot... Mar 16 15:05:29 ok, should be working Mar 16 15:36:06 jkridner: we are having gsoc meeting in about 25 minutes today, right? Mar 16 15:38:00 bradfa: hi, thank you for your email. Mar 16 15:43:23 pmezydlo: you're welcome Mar 16 15:45:00 In spi.c file has no function on the slave mode. Better to add new features to spi.c or write new lib for spi slave? Mar 16 15:45:54 *new functions? Mar 16 15:49:26 bradfa: no Mar 16 15:49:32 in 1h10 Mar 16 15:50:10 av500: ok, stupid daylight savings time... Mar 16 15:50:43 * av500 puts all his saved time in a bank Mar 16 15:51:23 pmezydlo: I'm not sure what the best way to implement spi slave would be, from a code point of view, but a good action to take would be to review other previously failed spi slave proposals and understand their shortcomings first Mar 16 15:54:15 oh, crap, its wednesday Mar 16 15:54:44 * jkridner feels overwhelmed. Mar 16 15:54:50 folks are in town for meetings today. Mar 16 15:55:20 av500: is the official meeting time 1700 UTC or based on "9am west coast" time per the email? Mar 16 15:55:43 jkridner is the time keeper Mar 16 15:55:47 k Mar 16 15:55:50 he knows how to send google invites Mar 16 15:56:03 but he rarely actually DOES Mar 16 15:56:06 * bradfa doesn't use The Google calendar... Mar 16 15:56:08 I think in the past we've used a google event, which follows the timezone of whoever created it Mar 16 15:56:31 * bradfa is behind the times :) Mar 16 15:56:33 bradfa: I have no idea how we handled the DST jump in the past Mar 16 15:56:36 so we've moved it with daylight savings I think... Mar 16 15:56:40 yes Mar 16 15:56:43 but EU DST :) Mar 16 15:56:52 its 6pm MY time Mar 16 15:56:53 ohhhh Mar 16 15:56:55 that much I knopw Mar 16 15:56:56 that much I know Mar 16 15:57:06 av500: if we just wait another couple weeks so that we're behind schedule, then all the DST will line up and fix everything ;) Mar 16 15:57:24 bradfa: the old "move GSOC to another date" theme Mar 16 15:57:27 wont happen :) Mar 16 15:57:31 we all just need to start living by av500 time Mar 16 15:57:32 evne under new leadership Mar 16 16:02:32 meet now... meet again in an hour.... meet every hour for the next 14 weeks. :-D Mar 16 16:03:22 how many students are here who have made drafts of proposals? Mar 16 16:03:34 any mentors here not invited to #beagle-mentors? Mar 16 16:04:05 anyone wanting to become a mentor, but still a mystery to /me and av500 (and pdp7 if he'll start showing up!) Mar 16 16:04:09 moin Mar 16 16:04:15 pdp7: ping, ping, ping :-D Mar 16 16:04:50 * jkridner wonders if we have more bots than people now. :-) Mar 16 16:04:59 now we can ping pdp7 here, on slack and on gitter :P Mar 16 16:05:16 * av500 is a bot Mar 16 16:05:44 gcl-bot: .w bot Mar 16 16:05:45 [WIKIPEDIA] Bot | "Bot or BOT, shortened from "robot" may refer to:..." | https://en.wikipedia.org/wiki/Bot Mar 16 16:05:51 gcl-bot: .w robot Mar 16 16:05:52 [WIKIPEDIA] Robot | "A robot is a mechanical or virtual artificial agent, usually an electro-mechanical machine that is guided by a computer program or electronic circuitry. Robots can be autonomous or semi-autonomous and range from humanoids such as Honda's Advanced Step in Innovative Mobility (ASIMO) and TOSY's TOSY Ping..." | https://en.wikipedia.org/wiki/Robot Mar 16 16:06:10 good old gcl-bot Mar 16 16:06:47 alexhiam, Just wanted to ask you,regarding Kconfig file for the hello world driver.Is it required?I mean I cannot think of an appropriate directory where to put my code whose kconfig file I should try to modify. Mar 16 16:07:04 * jkridner just released 3 messages (including 1 repeat) to the list. Mar 16 16:07:25 * jkridner wonders if we are discussing projects today at all or if he has a few minutes to take on other tasks. Mar 16 16:07:31 chanakya_vc: oh, no, only when you go to upstream. For that you can just compile it as a standalone module Mar 16 16:07:56 jkridner: why not start with the other items first, then projects after that? Mar 16 16:08:09 what other items? Mar 16 16:08:22 jkridner: sorry, I misunderstood you Mar 16 16:08:35 jkridner: I thought you meant you had other items to discuss Mar 16 16:08:59 So I should only write the make file?.It is not required to edit config symbols?Could you guide me to some sources where I can clear this a bit? Mar 16 16:10:02 alexhiam: what should be the name of the so called "modified strawson api's for blue"? Mar 16 16:10:36 chanakya_vc: the Kconfig stuff is for compiling the kernel. You don't need to do that, you can just compile it as a module, e.g.: http://www.tldp.org/LDP/lkmpg/2.6/html/x181.html Mar 16 16:11:23 kiran4399: you mean the project title? It depends on what you are proposing to work on Mar 16 16:11:46 but that title doesn't imply that there would be kernel driver work, etc Mar 16 16:12:50 alexhiam: OK.. so shall I simply call them beaglebone blue APIs? Mar 16 16:12:50 alexhiam, Ohh.I read that you had the option to declare whether you would load a file as built in or module(tristate variable) so I thought it was required. Mar 16 16:13:03 Good morninh Mar 16 16:13:20 *morning Mar 16 16:13:23 alexhiam,Anyway I will go with the link you just gave. Mar 16 16:13:27 are we meeting today? Mar 16 16:13:41 Abhishek_: currently, yes. Mar 16 16:14:13 yeah, DST's in effect now I guess Mar 16 16:14:18 chanakya_vc: yeah, in the kernel build config you can select drivers as modules Mar 16 16:14:27 Hi all. Mar 16 16:14:28 exactly.... and different all across the world. :( Mar 16 16:15:15 alexhiam: one more thing.. when I write the interface of ardupilot for beaglebone blue... I am planning to test ardupilot by mounting beaglebone blue on my quadcopter.. Mar 16 16:15:24 alexhiam: would that be a good testing platform.. Mar 16 16:15:26 alexhiam, Would I be required to push this upstream in my project? Mar 16 16:15:37 kiran4399: are you also looking to work on ROS and ardupilot support? Maybe something more along the lines of 'BeagleBone Blue software support', or 'Improving development experience on the BeagleBone Blue'? Mar 16 16:16:12 Anyone questions on the audio projects? Mar 16 16:16:17 kiran4399: depends on how much weight it can hold, the BeagleBone is pretty hefty for that Mar 16 16:16:28 chanakya_vc: the hello, world driver? Mar 16 16:16:41 or "How to not turn black and blue from using the Blue?" :D Mar 16 16:16:54 alexhiam,Oh no! I mean the final driver :) Mar 16 16:17:04 chanakya_vc: yeah Mar 16 16:17:26 ds2: lol, perfect Mar 16 16:17:34 alexhiam: I think it can hold the blue as we are using both the IMU and the flight computer as the beaglebone blue... Mar 16 16:17:37 alexhiam,Okay. Mar 16 16:17:50 alexhiam: so basically.. I am testing only arducopter.. but I think it would be sufficient.. Mar 16 16:18:07 kiran4399: right. Yeah, some field testing on a quadcopter would be great Mar 16 16:18:24 it would also be a test of the MPU-9250 driver Mar 16 16:19:02 ds2,I just wanted to ask you when you said that we require Dt support rather than DT overlay.The DT overlay just has to set the pinmux register to let certain pins act as i/o pins for PRU'S right? Mar 16 16:19:51 DT Mar 16 16:20:02 some folks may not liketo dynamically load it Mar 16 16:20:19 alexhiam: yeah.. and after testing and development of ardupilot on the quadcopter.. when I shift to ros development.. I think I'll test the beaglebone blue ros with the quadcopter platform.. Mar 16 16:20:27 alexhiam: what do you think? Mar 16 16:20:54 alexhiam: so the quadcopter testing will be useful for both ardupilot and ros.. Mar 16 16:21:29 ds2,Okay so I if I understand you correctly,you mean my driver should have the ability to be modprobed from a DT,while giving the choice of the DT to the user? Mar 16 16:23:03 kiran4399: sounds good Mar 16 16:25:33 yes Mar 16 16:26:15 alexhiam: just tell me one thing.. how do I test ble and wifi on blue? Mar 16 16:26:35 alexhiam: I know the drivers are there.. but how do I make the hardware on blue really work with those drivers? Mar 16 16:26:44 alexhiam: that is what is confusing the most now.. Mar 16 16:27:12 alexhiam: I know how to enable the driver, write the driver and compile the kernel.. Mar 16 16:27:36 alexhiam: but I don't understand how to actually make the hardware work on blue.. Mar 16 16:27:43 kiran4399: well, WiFi's easy, just connect to a network Mar 16 16:27:50 ds2,An another question that you raised in the mail regarding the communication between the ARM and the PRUSS,as far I my understanding goes,apart from the shared mem,there is no other way right? Mar 16 16:27:53 kiran4399: how do you mean? we'll have the drivers already integrated. connman will simply turn the wifi on. hciattach will be used for BT. Mar 16 16:28:43 chanakya_vc: there are interrupts and HW mboxes....but on a fundamental level, it is mostly shmem Mar 16 16:28:57 but how the shmem is used is the real question Mar 16 16:29:06 rpmsg? other? Mar 16 16:30:58 jkridner: do we have a linux kernel branch for blue? Mar 16 16:31:27 *note to all students*: I don't see many proposals on summerofcode.withgoogle.com yet - get drafts submitted ASAP so we have time to look at them and give you feedback! Mar 16 16:31:58 kiran4399: they should be on the same branch ideally, e.g. 4.1 Mar 16 16:33:00 kiran4399: I think the only difference really should be the EEPROM contents Mar 16 16:37:03 am out in 5, in case no more Qs on audio project Mar 16 16:37:57 rma_: are you student or mentor for that? Mar 16 16:40:17 wait, does this new gsoc site encourage google docs for proposals?? :| Mar 16 16:40:28 Yeo Mar 16 16:40:30 *Yep Mar 16 16:40:43 sorry got disconnected @ds2: am mentor Mar 16 16:41:11 Abhishek_: that sucks... I quit Mar 16 16:41:31 :( Mar 16 16:41:55 hmm "final proposal in PDF format" Mar 16 16:42:28 I guess it's for the inline feedback ability? Mar 16 16:43:20 oh yeah, the comments on the site are only visible to mentors Mar 16 16:47:08 * nerdboy could use a bbb-powered cat amusement device... Mar 16 16:47:26 kiran4399: no, it is not on its own branch.... Mar 16 16:47:43 kiran4399: I will try to start adding the patches (dt). Mar 16 16:48:11 *your* work should be on a branch, then merge to "official" branch... Mar 16 16:59:49 <_av500_> aloo Mar 16 17:00:01 so, the meeting is now then...? Mar 16 17:00:07 <_av500_> NOW Mar 16 17:00:35 <_av500_> Smithers, release the hounds Mar 16 17:01:18 I thought it was an hour ago. :-) Mar 16 17:01:22 _av500_: DST came into effect recently so... Mar 16 17:01:31 shall I repeat my 3 questions? Mar 16 17:01:45 how many students are here who have made drafts of proposals? Mar 16 17:01:55 * nerdboy whacking on stuff that doesn't like gold linker Mar 16 17:01:58 any mentors here not invited to #beagle-mentors? Mar 16 17:02:13 Wormo? Mar 16 17:02:13 alexhiam: One silly question... why are the eeprom contents different for different beaglebones?? what does eeprom consist of?.. Mar 16 17:02:13 anyone wanting to become a mentor, but still a mystery to /me and av500 (and pdp7 if he'll start showing up!)? Mar 16 17:02:33 kiran4399: it allows the software to select the right .dtb for the board. Mar 16 17:02:47 it also has manufacturer/serial number info for product support/returns. Mar 16 17:03:07 kiran4399: https://github.com/beagleboard/image-builder#eeprom-database Mar 16 17:03:21 jkridner: but is there any dtb for beaglebone blue? Mar 16 17:03:31 jkridner: can you please point it to me? Mar 16 17:03:50 jkridner: I've seen the defconfig file... beagleboard_def_config.. Mar 16 17:03:51 kiran4399: not yet.... I can put it together pretty quick. Mar 16 17:04:12 so far, I've just been working with overlays, but I need to make a new master .dtb Mar 16 17:04:44 kiran4399: you can look at the other dtbs for reference: https://github.com/beagleboard/linux/tree/4.1/arch/arm/boot/dts Mar 16 17:04:45 jkridner: oh... how long it will take? Mar 16 17:05:20 kiran4399: soooo many other things going on. maybe next week? Mar 16 17:05:37 jkridner: right!! Mar 16 17:05:43 I can let you make it if you'd like and make corrections. ;-) Mar 16 17:05:59 it is the strawson robotics cape with a small number of other signals moved. Mar 16 17:06:11 well, I say small, but I really just mean not too large. Mar 16 17:06:28 MPU9250 and BMP280 on I2C... not much new there. Mar 16 17:06:36 I2C2 to be precise. Mar 16 17:06:39 alexhiam, jkridner: I tried to search for the overlay for robotics cape.. but could'nt find it.. Mar 16 17:07:19 https://github.com/StrawsonDesign/Robotics_Cape_Installer/blob/master/install_files/RoboticsCape-00A0.dts <--- needs to be rewritten for 4.1 kernel. Mar 16 17:08:01 oof, I have my name on a few overlays that probably need to be rewritten for 4.1..... Mar 16 17:08:17 jkridner: btw... there are no free gpios for blue right? Mar 16 17:08:44 not so much, no. but, SPI and UART ports should be possible to config-pin into GPIOs. Mar 16 17:09:41 really? what are all those GPIOs being used for?? Mar 16 17:10:44 alexhiam: Well.. I think there are no free gpios.......I think they will be hardwired to servo ports, motor ports, dsm2 ports, uart, can, i2c, spi etc.. Mar 16 17:10:56 alexhiam: right? Mar 16 17:10:56 yeah, I guess that is a lot of stuff... Mar 16 17:11:10 alexhiam: will there by dts files for wifi and ble also? Mar 16 17:11:46 kiran4399: yes, they will be part of the file. Mar 16 17:11:59 jkridner: are those both hic over UART? Mar 16 17:12:00 kiran4399: the current draft is am335x-bonegreen-wireless.dts Mar 16 17:12:03 hci* Mar 16 17:12:17 WiFi is over SDIO Mar 16 17:12:20 jkridner: Yeah.. I've seen beaglebone green.. pretty cool one huh?? Mar 16 17:12:23 BT/BTLE is over UART Mar 16 17:12:30 cool Mar 16 17:12:31 A2DP is over MCASP Mar 16 17:12:38 MCASP == I2S Mar 16 17:12:48 oh nice Mar 16 17:13:20 jkridner: what about wifi? Mar 16 17:13:26 https://github.com/beagleboard/linux/commit/d554a724b4fb4af6cbb42bcdc5a83acaa68be942 <-- doesn't have A2DP in the dts yet. Mar 16 17:13:26 still have to create it? Mar 16 17:13:32 what HW is doing A2DP over McASP?! Mar 16 17:13:49 the Wilink? Mar 16 17:13:49 BeagleBone Blue. Mar 16 17:13:53 WTF Mar 16 17:13:57 Sancloud? Mar 16 17:13:58 Wilink8 - WL1835MOD Mar 16 17:14:05 ds2? Mar 16 17:14:11 IIRC the WiLink8 wants A2DP over UART Mar 16 17:14:22 HFP is over McASP Mar 16 17:14:25 wut? Mar 16 17:14:29 ds2: oh? am I confused about profiles? Mar 16 17:14:43 I probably am. Mar 16 17:14:49 jkridner: A2DP (aka high quality stereo audio) is usually over HCI Mar 16 17:15:11 HFP (the SCO stuff, aka telco quality 8KHz audio) is usually TDM and hence McASP Mar 16 17:15:11 So is MCASP pretty well deprecated? Mar 16 17:15:12 no A2DP should be over a higher quality link right? Mar 16 17:15:58 what's a "higher quality link"? Mar 16 17:16:15 jkridner: blue's wifi is wl1835? Mar 16 17:16:59 yes Mar 16 17:17:10 I mean, I wouldn't imagine sending 44.1kHz, 16bit stereo audio over a UART link Mar 16 17:18:11 the blue page also says '4 quad enc in'... isn't there only 3 eQEP modules? Mar 16 17:18:15 jkridner: Oh my god!! I am having a tough time in understanding the overlays... I hope I need not work on the overlays as part of this project right?? Mar 16 17:18:38 well there's no running away from having a tough time Mar 16 17:18:57 jkridner: I am longing to see a picture of beaglebone blue.. can you please show me? :-D Mar 16 17:19:03 Abhishek_: imagine the UART with h/w flow control at 3MBaud..that should make you feel better. Mar 16 17:19:35 yeah it does lol Mar 16 17:19:55 no details on datasheet Mar 16 17:23:42 prototype not assembled: https://www.facebook.com/oshpark/posts/957930137622311 Mar 16 17:23:53 w/o wifi/bt Mar 16 17:24:57 yikes, untented vias under the bga? Mar 16 17:25:23 Mmmmmmm Mar 16 17:26:04 Abhishek_: I forget the exact baud rate for TI wilink family but it's 2M or 3M, that's pretty much what all of these parts targeted at high-end designs do with the UART interface and then as ds2 pointed out..the HFP connections that use SCO typically are on an out-of-band pcm interface. Mar 16 17:27:26 I see, interesting. Mar 16 17:27:46 the wilink can vary the speed of the UART Mar 16 17:27:57 you just loose bw as you go down Mar 16 17:28:19 gets annoying when BT gets paired up with an older chip that tops out at 115200 Mar 16 18:00:45 ds2,When you say rpmsg,you mean to say that the driver that would expose the PRU ,say as an I2C device, would basically communicate with the userland with this bus(virtual) right? Mar 16 18:02:00 no Mar 16 18:02:09 rpmsg is a messaging layer built on top of the PRUSS's shmem Mar 16 18:02:15 it is suppose to be generic so for other processors, it could be different. the API for rpmsg is the same for all processors Mar 16 18:02:32 but keep in mind rpmsg does have some overhead Mar 16 18:05:23 ds2, Okay.I am a bit confused.What I had thought so far was that my driver would write the parameters(of I2C or SPI) into certain memory registers that we would create in the PRU.Now my understanding is that rpmsg is what would facilitate this?Am I correct? Mar 16 18:06:01 chanakya_vc: rpmsg is one option for the driver<->PRU communication Mar 16 18:07:23 alexhiam, So the other day when we were talking about the adapter driver and algorithim driver,it's a different way right to do the same? Mar 16 18:09:32 chankya_vc: well... it's one way to do it. It's essentially a pub/sub messaging framework, and there's some inherent overhead involved Mar 16 18:09:37 alexhiam: between ros and ardupilot development for beaglebone blue.. which one do you think will take more time? Mar 16 18:10:33 kiran4399: no idea. You'll definitely want to do some research to figure out what will be involved in each Mar 16 18:11:19 alexhiam,So we can either go for coding an adapter driver along with an algorithim driver or code a driver that uses rpmsg right? Mar 16 18:12:18 chanakya_vc: basically the driver acts to expose the PRU as an I2C, SPI, etc. device. It needs a way to pass configuration and data to and from the firmware running on the PRU (using the shared memory). rpmsg is one standard for doing that sort of thing Mar 16 18:12:57 but it's the same model regardless of how that driver<->PRU communication is implemented Mar 16 18:14:33 I think in Linux I2C terms, if using rpmsg then the rpmsg interface you build would be the "algorithm" Mar 16 18:14:54 pong pong pong Mar 16 18:15:13 but again, I'm certainly no expert on Linux I2C Mar 16 18:15:55 ahoy pdp7 Mar 16 18:18:35 heya pdp7 Mar 16 18:19:31 alexhiam,ds2, So as far as I have understood,the adapter driver basically binds itself to the shared mem.Then the algorithimic driver (or rpmsg) would basically write the required data or parameters to the required registers or memory spaces. Mar 16 18:20:31 I think for I2C in linux,both of these come in a single module.I mean I haven't found any mention of different modules. Mar 16 18:22:07 chanakya_vc: http://www.embedded-bits.co.uk/2009/i2c-in-the-2632-linux-kernel/ Mar 16 18:27:39 chanakya_vc: it sounds like you got the idea, though I'd prefer if you stopped using the term register, as it's not really correct there. With the I2C modules on the AM335x die the drivers are reading/writing registers, but with the PRU it would be a totally different thing, and trying to emulate the same model would just be adding extra overhead and complication Mar 16 18:27:55 afaik, ardupilot is either mavlink/lcm/linux or mavlink/ros Mar 16 18:28:41 i'd stay on the linux side for the core project and look at ros if there is still time... Mar 16 18:30:14 for bb black/bbmini or bb blue, you are both the autopiot and the host computer Mar 16 18:31:01 ardupilot already does black/bbmini cape and black/pixhawk-fire-cape Mar 16 18:31:35 you'd be integrating that stuff with blue hardware/sensors Mar 16 18:32:31 nerdboy: there's also the option of a standalone ros instance that's got APIs for all the blue hardware Mar 16 18:33:24 i'd make it work in linux first... Mar 16 18:33:40 unless it's a pure ros project Mar 16 18:33:58 need to see a project plan either way Mar 16 18:34:38 some hardware supports both ardupilot and librepilot Mar 16 18:34:48 eg, VR brain Mar 16 18:35:52 but that really targets the microcontroller side Mar 16 18:37:38 my thinking is that there should be as many options as possible, e.g. one user could enable the ardupilot service, another could just install use the ros interpreter (I'm assuming it's interpreted?) and run their ros robot script, and another user might keep those off and just use the C or C++ APIs to write their robot program Mar 16 18:37:51 so linux or ros, as both would be a bit much i think Mar 16 18:38:27 when I say ros I mean something like this: http://wiki.ros.org/indigo/Installation/Debian Mar 16 18:38:56 * alexhiam has absolutely no experience with ros Mar 16 18:39:36 whether or not all that can be one project, and if not what the priorities are, I don't know yet Mar 16 18:40:28 it would be ardupilot/mavros i think Mar 16 18:41:06 ros for robots has slightly different requirements than mavros Mar 16 18:42:44 dig wiki.ros.org is coming up flags: qr rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 0, ADDITIONAL: 1 Mar 16 18:44:47 http://dev.ardupilot.com/wiki/using-rosgazebo-simulator-with-sitl/ Mar 16 18:46:00 * alexhiam needs to learn more about all this robot stuff Mar 16 18:51:07 alexhiam,When I say registers,what I mean is basically write to a certain mem location.I do not really mean a writing to a register as such.You are expecting it to be done some other way? Like I am not saying that we should define registers and pointers to them in the shared mem. Mar 16 18:53:11 https://beagleros.wordpress.com/category/uncategorized/ <= apparently erle brain is bbb+pixhawk fire cape Mar 16 19:11:49 odd, dns has an answer now... Mar 16 19:54:54 chanakya_vc: right, I know what you mean, just not the right word. One way would be to define a different memory address for each individual interaction. Another way would be to use rpmsg, which I believe would be implemented on top of a circular buffer in the shared memory. You could also define your own protocol using a circular buffer and have different 'commands' to pass different things around Mar 16 19:56:19 in other words there's a lot of options, and you don't really want to be thinking about it in terms of an on-die I2C controller subsystem, because it's not very similar Mar 16 20:03:58 ds2, how will be that communicating to the ADC ? Mar 16 20:04:15 or the ADC based device for that matter .. Mar 16 20:04:36 again bitbanging SPI or I2C? Mar 16 20:18:55 alexhiam,Okay.I have just started with Digital circuits as a course,so I think that's why I am stuck at registers. :) Mar 16 20:27:21 hey m_w ! there ? Mar 16 20:35:41 alexhiam,I went through the link that you sent me.I think I have to just concentrate on writing the adapter driver and then the particular device driver(according to the link this is the one that actually implements the protocol to write on certain memory locations) Mar 16 20:37:07 I think the algorithim driver is only used when we are not having a bitbanged implementation.Atleast that's what I understood of that site. Mar 16 20:38:36 chanakya_vc: no, this would be a bus driver, which consists of the algorithm and the adapter. The client and device drivers uses the corresponding bus driver for the given bus Mar 16 20:39:25 so i2c device drivers on the beaglebone are using the i2c-omap bus driver Mar 16 20:39:47 or they could use the i2c-pruss bus driver, for example Mar 16 20:40:58 the point of these different levels of abstraction is that the device driver for X i2c ADC should work regardless of the system it's running on (assuming it has an i2c bus), because the bus driver provides a common interface Mar 16 20:42:34 alexhiam, Okay let me research about this and get back to you. Mar 16 20:43:23 fyi I haven't looked at how the Linux i2c slave implementation fits into all this Mar 16 20:44:56 alexhiam, when do you think ds2 will be here ? Mar 16 20:45:22 no clue Mar 16 20:46:21 alexhiam, For the slave implementation we would have nothing but the address stored in shared mem at a particular location.And then follow the standard rules of I2C when PRU's recieve? Mar 16 20:46:32 Drivers would work both ways? Mar 16 20:48:09 https://www.kernel.org/doc/Documentation/i2c/slave-interface Mar 16 20:48:18 *Bus driver support* section Mar 16 20:49:01 alexhiam,Will look at that. Slave implementation for SPI also right? Mar 16 20:49:31 that's trickier, because no SPI slave implementation has been accepted in mainline linux Mar 16 20:50:44 pmezydlo may be looking into that as well Mar 16 20:51:28 isn't his project something related to SPI flash emulator? Mar 16 20:52:47 right, which requires slave mode spi Mar 16 20:53:41 alexhiam, Okay didn't know that.I don't think GSOC would allow us both to work on the same project? :) Mar 16 20:53:54 I mean same piece of code Mar 16 20:54:05 no, but that would be different Mar 16 20:54:23 but may be worth discussing spi slave implementations Mar 16 20:54:34 and why one hasn't been accepted in mainline yet Mar 16 20:55:38 Okay.So I guess I shall discuss with him whenever he comes online.But in the end you would want only one of us working on that right? Mar 16 20:56:49 well, that depends. Certainly two projects that involve general spi slave implementations using the PRU wouldn't work Mar 16 20:57:07 so talking amongst yourselves to figure that stuff out is important Mar 16 20:59:13 Okay. I will talk to him.And decide who is going to do that. Mar 16 21:00:37 alexhiam, It is written on the google website that we should always submit a proposal that we think can be achieved even though it might be a bit underscoped. Mar 16 21:01:19 don't worry about the google requirements at this point, just focus on getting a proposal submitted ASAP Mar 16 21:01:48 and I mean submitted to summerofcode.withgoogle.com Mar 16 21:02:43 okay. I have to figure this driver bit and I think I will be able to build on my earlier draft. Mar 16 21:03:59 even just posting what you have now on there would be good Mar 16 21:04:28 Okay I will do that right away. Mar 16 21:05:11 great Mar 16 21:06:25 ds2 has asked me a couple of questions on the ml.Would have to answer those too. Mar 16 21:11:29 alexhiam, I have uploaded my draft.It's the same document as before.I will make changes to it as soon as I figure out the driver part. Mar 16 21:12:07 awesome Mar 16 21:12:50 I hope that mentors wouldnot think that I have uploaded the same document in spite of the mistakes and observations being pointed out? Mar 16 21:13:43 don't worry. All we want at this point is for people to submit proposal drafts there!! and hardly anyone has yet :( Mar 16 21:14:06 feel free to add a note in the doc or a todo list or something Mar 16 21:18:17 pmezydlo, You there?Would you be working on the SPI slave implementation? Mar 16 21:24:04 yes, I will be working on spi slave Mar 16 21:25:04 you would be designing the driver and the bitbanged implementation? Mar 16 21:29:10 I will not use PRU Mar 16 21:29:51 Create a generic SPI slave framework in Linux with McSPI as the first hardware layer. Mar 16 21:29:52 The task is to create a driver which supports hardware controller McSPI, Mar 16 21:29:52 and to ensure optimal performance through the use of DMA and interrupt. Mar 16 21:30:25 quoted from my application Mar 16 21:30:36 Okay so yours will not be an implementation on the PRU right? Mar 16 21:31:08 Please clarify that with alexhiam. Mar 16 21:31:33 yes, exactly Mar 16 21:31:54 you are in which time zone? Mar 16 21:32:15 My time zone is UTC Mar 16 21:32:39 *UTC + 1 Mar 16 21:34:07 I am in UTC+5:30 Mar 16 21:35:01 So I was wondering if we could find any common time and discuss this with alexhiam,if you do not mind? Mar 16 21:39:53 My main task is to write the driver and API for spi slave. Mar 16 21:41:55 Okay I get it..Please tell alexhiam this.I think he is under the impression that you would be doing it for PRU also. Mar 16 21:44:37 i will not use PRU, but hardware McSPI controller. Mar 16 21:46:29 Okay cool.Thanks for clarifying :) Mar 16 21:55:07 good plan... Mar 16 23:07:01 was there a mentor meeting today? Mar 17 02:32:12 m_w: sort of Mar 17 02:32:40 little mixup with daylight stupid time change... **** ENDING LOGGING AT Thu Mar 17 02:59:58 2016