**** BEGIN LOGGING AT Mon Mar 26 03:00:03 2018 Mar 26 05:03:35 Hmmmm Mar 26 05:19:50 @wormo i have fixed the typo in my draft that lead to the wrong github acc,you shall find the repo's for previous projects if you want to now Mar 26 05:20:16 also i have changed the draft a bit with reference to whatever we discussed yesterday Mar 26 05:20:34 kindly take a look,i am prepping the final pdf Mar 26 06:01:22 spititus: muneeb17 : In any case, a complete understanding of how RPMsg and RemoteProc work is necessary beforehand. And I dont see it yet in any of those proposals. Mar 26 06:05:28 zeekhuge: I'll update my proposal now to make it more detailed. I also will add dto building/loading functionality. Also, about the DMA support, I was thinking how to use the pru_dma driver Mar 26 06:05:41 the present project uses kernel modules Mar 26 06:10:00 muneeb17: and what do you mean by that ? Mar 26 06:11:53 It uses a loadable kernel module to access the kernel level API. I was thinking about how to make a python lib to do the work Mar 26 06:14:17 see ... you really need to get this. Mar 26 06:14:49 everythin internally is done by kernel Mar 26 06:15:13 and in linux everything is a file. even the interface between a driver and userspace . Mar 26 06:15:20 they are called sysfs. Mar 26 06:15:44 so .. you will ultimately have to write to sysfs. Mar 26 06:15:54 there is no other magic that at least I know of to access Kernel space APIs in the Userspace. Mar 26 06:16:24 If it was possible, then the two separate spaces wouldnt exist. Mar 26 06:16:44 muneeb17: ^ Mar 26 06:17:15 For DMA, each example provides its own sysfs Mar 26 06:17:28 how can we make it more generalised Mar 26 06:17:54 From the previous year examples, i mean Mar 26 06:23:52 zeekhuge: ^ Mar 26 06:24:44 Naah .. Mar 26 06:25:16 that would involve writting a simple driver ... about 150 lines of code Mar 26 06:25:25 Pac23 ok, I see your sample python app Mar 26 06:25:57 muneeb17: essentially .. you will take one of the drivers from the examples, and modify it a bit to be a general purpose driver for Python Mar 26 06:26:41 You need to understan how device-attributes of a kobject in the linux kernel and sysfs are connected Mar 26 06:26:59 to do it. Mar 26 06:29:25 muneeb17: https://github.com/maciejjo/beaglebone-pru-dma/blob/master/examples/03_loopback/kmod/pru_dma_loop.c#L74 Mar 26 06:29:45 the attributes provide a sysfs interface Mar 26 06:29:50 and with the same name. Mar 26 06:31:57 zeekhuge: Thanks, I get it now. The whole process just became a lot more clearer. Mar 26 06:33:31 so, in a way, you it might be required to develop a kind of minimalistic protocol maybe (a set of known signals - I am not sure if it will really be required) and the sysfs entries will be general purpose. Mar 26 10:28:51 ds2: zeekhuge: https://elinux.org/GSoC_2018_Proposal:Beaglebone_GPU_Offload Please comment on my proposal (espcially the timeline). I have an exam right now... will edit more after 2.5 hrs... Thanks Mar 26 11:58:04 * jyothi[m] sent a long message: jyothi[m]_2018-03-26_11:58:03.txt Mar 26 11:59:20 jkridner: Mar 26 16:21:31 do I need to create the elinux wiki page with my idea before final proposal submission? Mar 26 16:24:42 raunakn9: It would be helpful to try to solicit some feedback today, though the benefit may be only minimal. Mar 26 16:25:07 jkridner: Please take a look at my proposal and comment on it. Do you think I have to improve the gaming part ? Mar 26 16:25:28 link? Mar 26 16:25:36 jkridner: Please see https://elinux.org/GSoC_2018_Proposal:Beaglebone_GPU_Offload and comment Mar 26 16:26:15 Sidharth: odd link location. Mar 26 16:26:47 Sidharth: good to see some feedback from ds2. Mar 26 16:26:53 jyothi[m]: link? Mar 26 16:27:38 jkridner: Mar 26 16:27:39 https://www.elinux.org/BeagleBoard/GSoC/ModernSpeak-Spell Mar 26 16:28:36 jkridner: Mar 26 16:28:37 https://www.elinux.org/BeagleBoard/GSoC/ModernSpeak-Spell Mar 26 16:29:11 jkridner: I will move it to BeagleBOard/GSoC/ Mar 26 16:33:36 I moved it to https://elinux.org/Beagleboard:GSoC_2018_Proposal:Beaglebone_GPU_Offload Mar 26 16:34:44 jkridner: Do you think any changes needed? I'll ask ds2 as well Mar 26 16:40:21 is it legal to use the beagle boane logo in the proposal ? Mar 26 17:02:06 Sidharth: a plain C library may be more flexible, IMO Mar 26 17:03:20 ds2: are these stereo vision project proposals largely tied to the GPU off-load proposals? Mar 26 17:03:43 Sidharth: trying for the framebuffer EGL flavor instead of Wayland may be desireable... not all BBB's have a display.... a plain FB one should suffice Mar 26 17:03:53 jkridner: not that I am aware of Mar 26 17:04:25 k, because I see Sidharth is looking at a stereo vision demo at the end of his project. Mar 26 17:04:38 Sidharth: taking it all the way to stereo vision processing my be pushing it giving the time frame and the "issues" with the SDK Mar 26 17:04:58 jkridner: yes, think that is a potential complication Mar 26 17:06:09 Sidharth: prehaps, implement a simple alg like a demo that doubles all the values or adds 1... a very simple alg so someone can easily see where the GPU stuff is and where the alg begins.... there are datatype limitations with GLES (compared to CL) Mar 26 17:06:18 or even full blown OpenGL Mar 26 17:07:33 Sidharth: I'd be worried about you having enough time to implement BLAS/LAPACK... if you think that plus OpenCV stuff is really doable, I'd suggest making it a stretch goal (i.e. something you do after the GSoC period or for when you are finish with everything else Mar 26 17:08:29 Sidharth: You can leave messages here (I pop in and out) or email the list (I watch that too.) Mar 26 17:24:21 jkridner: ds2 : Mar 26 17:24:22 https://www.elinux.org/BeagleBoard/GSoC/ModernSpeak-Spell Mar 26 17:24:23 Sorry. Some issues with my network before. Mar 26 17:31:02 I made a new proposal, if someone can give it a quick look (specially ds2), it would be nice: Mar 26 17:31:03 https://elinux.org/BeagleBoard/GSoC/PRU_SPI_master_soft-peripheral_-_GSoC_2018 Mar 26 17:33:10 jyothi[m]: what HW are you using? what kind of audio interaface? Mar 26 17:35:45 hgonzalez[m]: few things should be preGSoC - most of the restudy's.... and add the SPI master framework in the kernel Mar 26 17:36:06 hgonzalez[m]: 2 weeks to do a PRU spi framework is a bit long considering what SPI is. Mar 26 17:36:11 hgonzalez: Do you have a feel for what the limitations will be regarding the PRU pins to be used? It'd be cool if the device tree overlay could specify the PRU GPIO indexes used for the SPI pins. Mar 26 17:36:54 hgoznalez[m]: how are you going to address describing the PRU pins? the gpio directive references kernel map'ed GPIOs Mar 26 17:37:32 hgonzalez: , ds2 : would this potentially also use the EDMA? Mar 26 17:37:46 * jkridner[m] loves the idea of a dynamic number of chip selects..... that'd add a lot of flexibility to end users. Mar 26 17:38:06 hgonzalez[m]: I suggest the test code be something simplier then a STM32.... a shift register + LEDs or prehaps a loopback.... Mar 26 17:38:14 of course, main goal should be to demonstrate making a kernel driver to talk to the PRU as an emulated peripheral, but usability features of the peripheral itself are nice. Mar 26 17:38:40 @jkrinder is it legal to use the beaglebone logo/icon ? Mar 26 17:38:44 jkridner[m]: isn't EDMA slighly orthogonal to this? I wouldn't care either way... want to get people point in the direction of talking to the kernel instead of the whole mess in user space Mar 26 17:38:51 yeah, like the BaconBits shift register.... that GPIO expander looks like a nice target. Mar 26 17:39:11 exactly.... HW building is outside of GSoC Mar 26 17:39:48 Pac23xyz: the "boris mascot" image is released under a creative commons attribution share alike license. The official logos are trademarked. Mar 26 17:40:18 In general, use of "boris" for commercial purposes is severly discouraged without a logo license. Mar 26 17:40:22 hgonzalez[m]: you might want to learn about a thing called a search engine (this is in regards to Contingency) Mar 26 17:41:06 for projects, it is kinda okay, but Christi doesn't really like it as it "confuses the brand". Mar 26 17:41:23 ravikp7: you around to discuss any last-minute items? Mar 26 17:41:43 i want to use it in my proposal Mar 26 17:41:48 various versions of the mascot Mar 26 17:41:49 ds2 : Bela Mini Cape for audio interfacing and OLED display Mar 26 17:42:00 that should be fine then i guess Mar 26 17:42:53 jyothi[m]: I don't think that includes a mic, does it? if not, what are you using as mic/pre-amp? Mar 26 17:44:22 ds2: yes. It does amplify the signal . Mar 26 17:44:22 It has inbuilt amplifier Mar 26 17:45:03 jyothi[m]: what kind of mic? does it have AGC? Mar 26 17:47:40 ds2: I am not sure if it AGC amplifier. The specifications say Mar 26 17:47:41 Stereo audio I/O with integrated headphone amplifier (16 bit, 44.1kHz) Mar 26 17:48:19 jkridner: forgot to include tcp/ip proxy in to-do for beagleboot, will surely add it in timeline before final submission. Couldn't work on it today due to other commitments Mar 26 17:49:40 jkridner: should be do a separate node module for the proxy server? so that it can be used without the electron app also Mar 26 17:50:45 *we do Mar 26 17:50:58 ravikp7: good question. I think this proxy would depend on using the node-usb implementation of RNDIS, so it shouldn't be separate unless that wsas broken out separate. Mar 26 17:50:59 ds2: I don't think it is AGC amplifier. Mar 26 17:50:59 Is it better to have it? Mar 26 17:52:07 ravikp7: this proxy would have the app send IP packets targeted at IP addresses other than its own virtual IP addresses over the host's IP network and leave the response ports open such that it can forward them to the board.... Mar 26 17:52:52 ravikp7: there'd never actually be a tie-in to the host's networking stack as a new networking adapter... only a program sending IP packets. Mar 26 17:53:24 Servers on the board wouldn't really be shared outside of the app. Mar 26 17:54:06 headphone amp != mic amp Mar 26 17:54:16 please verify you can really hookup an amp to it w/o big issues Mar 26 17:54:18 ie., we should be able to stay in the Electron BeagleBoot app and open Cloud9, but not expose Cloud9 to the rest of the network. yet the BeagleBone should be able to 'sudo apt-get install XYZ' Mar 26 17:54:32 AGC makes it a little easier to get the speech recog. to work Mar 26 17:55:47 jyothi[m]: at least describe the HW in the proposal... Mar 26 18:00:28 bbl Mar 26 18:02:08 Okay Mar 26 18:20:51 * hgonzalez[m] sent a long message: hgonzalez[m]_2018-03-26_18:20:51.txt Mar 26 18:52:28 ravikp7: can you run all the code through js-beautify and then do that every time before check-in? Mar 26 18:52:46 ravikp7: are you able to check across OS types for every release? can we create some process to do that to make sure we don't have regressions? Mar 26 18:53:24 * jkridner[m] googled for ideas and found https://github.com/ravikp7/electron-node-usb-testing :-) Mar 26 18:53:59 hmmm... https://github.com/resin-io/etcher/issues/788 Mar 26 19:01:34 jkridner: sure, will use js-beautify Mar 26 19:01:45 jm_donthineni: have you been able to get erik.welsh to review your proposal? Mar 26 19:02:04 jm_donthineni: can you ping directly as I'd want him to be engaged mentoring for this project. Mar 26 19:02:36 jkridner: okay Mar 26 19:03:36 jkridner: if you talking about testing electron app, no, I can't check for now :( Mar 26 19:05:28 ravikp7: can you come up with a strategy of some people you can ping to do testing or something that minimizes their time and maximizes the likelihood they will provide useful data and verify functionality? Mar 26 19:06:54 hgonzalez: has Abhishek__ been discussing the potential project with you? Is https://www.elinux.org/BeagleBoard/GSoC/Port_am335x-pru-package_to_remoteproc_-_GSOC_2018 the right link? Mar 26 19:08:25 jkridner: No, we didn't discuss it. I proposed 2 projects, that one and the PRU soft peripheral one Mar 26 19:08:54 jkridner: can u suggest me suitable AGC amplifier for bela mini? Mar 26 19:08:55 I can only see regular microphones connected to audio input on http://bela.io/ Mar 26 19:10:56 jm_donthineni: I cannot. Mar 26 19:12:16 jkridner: sorry, can't discuss now. I need sleep... Mar 26 19:13:29 jkridner: I can test it on all platforms win, linux and osx Mar 26 19:13:59 windows, linux on my laptop and osx on a hackintosh pc at home Mar 26 19:15:55 jkridner: I cannot find erik.welsh on the group. What is his IRC name? Mar 26 19:16:40 jkridner: will get back to you tommorow, want to discuss more about the proxy server, but not in a state to do it now Mar 26 19:18:20 k Mar 26 19:25:32 jm_donthineni: I think he's replied somewhere on https://bbb.io/gsocml Mar 26 19:25:46 jkridner: okay. Will ping him there. Thanks Mar 26 19:29:11 jkridner: I had already shared my proposal on the google group. Is there any other way I can contact him? Mar 26 20:03:00 did you see any e-mails from him to find his address to e-mail directly? Mar 26 20:19:29 No. I didn't find any. Mar 26 20:21:52 jkridner: Erik.Welsh@octavosystems.com Mar 26 20:21:52 I found this in your mail recipient list though. Can I mail to that address directly? Mar 26 20:37:16 yes, I know his email address. I was trying not to be rude by posting it to a publicly logged channel and to seek to understand your resolve. Mar 26 20:37:47 you can email that address directly. Mar 26 20:38:49 jkridner: Updated my proposal. Please have a look: https://elinux.org/BeagleBoard/GSoC/PyPRUSS Mar 26 22:32:29 <10hr remains... Mar 27 00:35:37 wow. no time left. :’( Mar 27 00:35:59 really that short? :’( Mar 27 00:36:06 Don't we have like 15 hrs till the deadline? Mar 27 00:36:34 I am still looking on the points mentioned here... Mar 27 00:48:58 ds2: jkridner: I am planning to use a keypad with alphabets and arrows for the project. Given that speech recognition is implemented, it isn't really necessary to have one but better to have one for better interaction. Should I go with this or just use a 4*4 keypad matrix with with basic buttons. Mar 27 00:51:43 jm_donthineni: a regular USB keyboard is a reasonable first start. Mar 27 00:52:42 keypad matrices are a solved problem. Mar 27 00:53:31 not worthy of a GSoC project. Mar 27 00:54:06 Jkridner: okay. Thanks:) Mar 27 00:54:29 does the bone have a KB matrix encoder like the BBC/BBX? Mar 27 01:10:02 I think pocketbeagle doesn't have one. It is only 4*4 matrix encoder right? It's not possible to access all alphabets on it Mar 27 01:10:07 ds2 : Mar 27 01:18:34 Do we have any time left ? Or the final proposals are done ? Mar 27 01:20:49 Hey isn't it 14 hrs? Mar 27 01:21:00 21:30 IST no? Mar 27 01:21:28 Sidharth: idk. What does your SoC dashboard say ? Mar 27 01:22:41 Well it said on the page 21:30 IST... didnt login in the page Mar 27 01:23:08 Was trying to collect more info... Mar 27 01:23:17 Login and we'll know for sure. (I am on mobile, can't really check) Mar 27 01:24:05 The SoC dashboard says 14.30 hrs more Mar 27 01:24:34 Cool ! Mar 27 01:26:49 Hey zeekhuge! Please tell me for how long are you here.. Will try to turn one more draft before I finalize it. Mar 27 01:29:47 muneeb17: man ! Much better now ! Mar 27 01:29:58 Sidharth: I am not sure. Mar 27 01:33:44 hgonzalez: there was some work done on a similar project in 2016, did you try to look into it ? Mar 27 01:35:28 muneeb17: BTW, one can simply use the resource_table file in the PRU code to write anything in the PRU ram while booting. In fact, the remoteproc driver does it for you. Mar 27 01:36:10 Also, you cab add some debug APIs. Look into the second blog in the PTP series 'zeekhug.me' Mar 27 01:36:15 * an Mar 27 01:36:24 *can Mar 27 01:36:26 Ah.. **** ENDING LOGGING AT Tue Mar 27 03:00:03 2018