**** BEGIN LOGGING AT Mon Feb 20 03:00:01 2017 Feb 20 11:30:32 Hello all. Feb 20 11:31:01 Hey tokencolour ! Feb 20 11:31:49 I want to participate in GSoC this year through BeagleBoard.org. Feb 20 11:32:08 I went through http://elinux.org/BeagleBoard/GSoC/Ideas. Feb 20 11:32:51 And ? what idea seems interesting to you ? Feb 20 11:33:04 and I am interested in the idea BeagleBone/Beagle board PRU DMA Feb 20 11:33:29 Do you have a BBB ? some experience with PRUs ? Feb 20 11:33:54 and about DMA ? Feb 20 11:34:49 I am unable to find some references in relation to this. Feb 20 11:35:06 like about the PRUs ? or about what DMA is ? Feb 20 11:35:38 I did run a few examples with the PRU. Feb 20 11:36:03 and I have studied what DMA is in my microprocessor classes. Feb 20 11:36:19 cool ! Feb 20 11:36:21 but I am unable to find a starting point. Feb 20 11:37:04 so, the point is about sharing data between the main processor (A8) and the PRUs. Feb 20 11:37:17 since there is no DMA support as of now. Feb 20 11:37:32 what we mostly do is : Feb 20 11:37:43 use 1 PRU to manage the I/O Feb 20 11:38:08 and use the other PRU to do all the data transfer between PRUSS and the A8 Feb 20 11:38:35 that project aims at developing DMA support to get data from the PRUSS Feb 20 11:38:58 so that we can free the second PRU Feb 20 11:38:59 and hence allow developers to use both PRUs to manage I/O or other processing. Feb 20 11:40:16 Thanks zeekhuge for the explanation. Feb 20 11:41:04 However, I am stuck as to how to start with the planning required to start with the proposal Feb 20 11:41:28 I need to visualize what exactly I'll need to do. Feb 20 11:42:04 Also, I'm not very familiar with linux kernel programming. Feb 20 11:42:10 well, ds2 will be able to guide more on that, and he'll be available (if) only after like 8-9 PM (IST) Feb 20 11:42:35 tokencolour: you familiar with linux based os in general right ? Feb 20 11:43:03 yes. I use it daily. Feb 20 11:43:08 Cool then . Feb 20 11:43:37 About the project, what I can imagin is about exploring how the current methods actually work Feb 20 11:43:52 then, diving into linux kernel code and see how DMA works there .. Feb 20 11:44:09 you definitely need to learn kernel coding for that Feb 20 11:44:21 there are numerous resources for that. Feb 20 11:46:03 Should I see how linux kernel does DMA and think how I'll make it work for the PRU? Feb 20 11:46:45 I think so. Feb 20 11:46:54 Which resource for learning kernel coding do you recommend? Feb 20 11:47:24 ldd3 Feb 20 11:48:46 Thank you zeekhuge. I'll go through it. Feb 20 15:40:51 m_w: hi, I managed to compile kernel with spi slave by Geerts. Config file was Beaglebone with some recommended changes that attach slave support, Feb 20 15:45:55 Now i will try add slave support to OMAP. Unfortunately I don't have a lot of time but it doesn't require many modifications Feb 20 16:55:00 pmezydlo: awesome Feb 20 17:06:54 Soon I will have to begin to develop my project brief Feb 20 17:08:02 m_w: Do you think there will be FPGA project this year? Feb 20 17:08:36 I hope so Feb 20 17:09:20 I would like to do project based on FPGA and beaglebone Feb 20 17:09:37 anything specific? Feb 20 17:10:37 I think about ice40 cape for BBB Feb 20 17:11:23 and using pru I want to program FPGA Feb 20 17:11:45 the ice40 use SPI to program actually Feb 20 17:11:56 so you don't need the PRU Feb 20 17:12:24 and I think that develop bridge between fpga and arm on pru its good idea Feb 20 17:13:27 any advantage over mmio? Feb 20 17:15:21 probably no advantage except relief CPU Feb 20 17:16:05 well the PRU needs to be accessed via mmio as well Feb 20 17:16:07 ICE40 have support from kernel? Feb 20 17:16:24 there is a patch out on the mailing lists Feb 20 17:16:37 not sure if it was merged yet Feb 20 17:17:01 and is it mapped in memory? Feb 20 17:17:52 I have to look at what is this support Feb 20 17:17:55 it is the programming interface only Feb 20 17:18:21 it is not too much Feb 20 17:18:34 interactionwith the fpga varies from platform to platform Feb 20 17:18:59 some use peripheral interfaces like SPI,I2C or I2S Feb 20 17:19:10 some use MMIO Feb 20 17:20:14 the FPGA support in the kernel is primarily programming and standard memory map interface drivers Feb 20 17:20:27 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/fpga Feb 20 17:21:05 fpga-mgr is the programming interface which can also load mmio drivers Feb 20 17:22:00 https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/Documentation/fpga/fpga-mgr.txt Feb 20 17:23:54 check this out: https://fosdem.org/2017/schedule/event/floss_high_level_synth/ Feb 20 17:24:28 ok i will have a look Feb 20 17:27:24 if I make a ice40 cape, I bet they would allow a project to add support for it Feb 20 17:28:20 they don't like the hardware to be the focus of the project Feb 20 17:33:13 Yes this is obvious but I don't like this. We should think about how to support this. Feb 20 17:35:00 shouldn't be a problem if we get the hardware ready in advance Feb 20 17:36:16 while I am at ELC, I can get started on the hardware Feb 20 17:36:34 leaving for Portland today Feb 20 17:37:08 ok I will let you know when I prepare better about this idea Feb 20 17:37:15 I will post the design to github so you can colloborate Feb 20 17:39:26 Can you give me an already started kicad project? I can finish it Feb 20 17:40:48 maybe, you want in on the hardware design side? Feb 20 17:41:58 yes I want Feb 20 17:42:18 if I create a github you can fork it and do pull requests Feb 20 17:42:38 then I can mentor you through the process Feb 20 17:42:38 yes this is good solutions Feb 20 17:42:40 thanks Feb 20 17:43:15 pcb design is an interesting case Feb 20 17:43:24 done any substantial PCB design before? Feb 20 17:45:41 yes I created some (5 pcs) pcb Feb 20 17:46:01 what was the design? Feb 20 17:46:41 mostly adapter to gold pins Feb 20 17:47:06 well this will be a bit more complicated than that :D Feb 20 17:47:07 one beaglebone cape Feb 20 17:47:42 one pcb for AVR Feb 20 17:47:55 I think that we should do something similar to the logi-bone but with an FPGA Feb 20 17:48:06 how much layer do you use Feb 20 17:48:11 http://valentfx.com/logi-bone/ Feb 20 17:48:23 this would be at least 4 layers Feb 20 17:50:13 until now I used eagle Feb 20 17:50:34 the free version of eagle won't cut it Feb 20 17:50:52 but kicad I used only once Feb 20 17:51:09 I use kicad so you might want to start doing some tutorials Feb 20 17:51:52 design a simple board and post it so that I can review Feb 20 17:52:00 so you can get used to the process Feb 20 17:52:00 yes this is similar but tutorials is good options Feb 20 17:52:56 I will prepare simple cape Feb 20 17:53:42 probably for 4 uarts ports on bb Feb 20 17:54:00 sure that should be straight forward Feb 20 17:54:24 RS232? Feb 20 17:54:41 yes -+ 12 to TTL Feb 20 17:55:29 okay please post the design and I will review it as you go Feb 20 17:57:06 yes ok I have to go Feb 20 17:57:11 later Feb 20 17:57:28 it's time for dinner Feb 20 17:57:30 bye Feb 20 22:01:57 hi! I'm a GSoC 2017 aspirant and I'd like to work on this project: BeagleBone synchronous data collection [http://elinux.org/BeagleBoard/GSoC/Ideas#BeagleBone_synchronous_data_collection\ Feb 20 22:04:28 i noted the possible mentor is Hunyue Yau Feb 20 22:05:13 Could I get his contact info so that I can get more details on the project? Feb 20 22:05:57 anandkp92: what do you want to know? Feb 20 22:06:15 I'd like to know what the project is about, and what the requirements are Feb 20 22:06:43 the project is building the equiv of BeagleLogic but for synchronous acquisition Feb 20 22:07:14 requirements are - 1) Understanding what those terms are; 2) And probally be able to at least build the BeagleLogic stuff Feb 20 22:07:59 rather then contacting me directly, please email the Beagle GSoC list Feb 20 22:08:09 ok I'm looking at the beaglelogic wiki Feb 20 22:08:09 https://github.com/abhishek-kakkar/BeagleLogic/wiki Feb 20 22:08:19 I assume this is a good starting point? Feb 20 22:08:23 do you know what it means to be synchronous? Feb 20 22:08:27 yes Feb 20 22:08:28 yes, that is a good starting point Feb 20 22:08:50 okay - the first question for you is - in what ways is the Beaglelogic stuff NOT synchronous Feb 20 22:09:29 I dont know the beagle logic part :| Feb 20 22:10:07 okay - have you used a logic analyzer before? Feb 20 22:10:31 or what may be helpful for me is - what are you studying in school right now? Feb 20 22:10:31 nope Feb 20 22:10:43 i'm in the us. carnegie mellon university Feb 20 22:10:56 what's your major? Feb 20 22:11:10 energy sciences, but I specialize in computer engineering Feb 20 22:11:24 i have a bachelors in computer science too Feb 20 22:11:31 okay, that gives a rough background Feb 20 22:12:07 in this case, you might want to learn what a logic analyzer is - not sure if you are allowed to, but it won't hurt to look at equipment in the lab at school Feb 20 22:12:31 i think i can. Feb 20 22:12:33 what you need to look for is the types of signals you hook up to the logic analyzer Feb 20 22:12:53 pay attention to clocks and any config things relating to it Feb 20 22:13:03 this would give you a background Feb 20 22:14:00 ok definitely. thanks! Feb 20 22:16:03 Btw, nice to meet you. My name is Anand Prakash Feb 20 22:16:04 :) Feb 20 22:17:16 same here... If you have other q's...ask... I may not respond immediately but I usually see stuff here Feb 20 22:17:56 ok, if you dont mind could you tell me about this project too BeagleBone/Beagle board PRU DMA? Feb 20 22:18:23 dma is the direct memory access? Feb 20 22:18:59 sure - Feb 20 22:19:36 all PRU code I am aware of transfer things between the PRU and the main processor by the action of either the A8 or the PRU Feb 20 22:19:42 ok Feb 20 22:19:50 the onboard DMA controller (EDMA) can do the work. Feb 20 22:20:04 the goal of that project is to figure out how to get the EDMA to do it and write sample code to show it doing it Feb 20 22:20:30 i.e. the beaglelogic code xfers things by using one of the PRUs to xfer data. this burns that PRU (there are only 2 per PRUSS) Feb 20 22:20:46 i see. Feb 20 22:20:55 moving it to the EDMA can free it up. Conversely, some setups use the ARM side to suck data from the PRUSS Feb 20 22:21:04 it burns cpu cycles. using EDMA can simplify it. Feb 20 22:21:15 so that's the project in a nuthshell Feb 20 22:21:29 yeah cause dma doesn't require cpu intervention until the xfer is complete right? Feb 20 22:21:49 yep Feb 20 22:25:15 cool, thanks! i'll check these projects and get back to you. **** ENDING LOGGING AT Tue Feb 21 03:00:00 2017