**** BEGIN LOGGING AT Wed Jul 24 02:59:57 2019 Jul 24 03:00:13 But if we are sending data from ARM to PRU, that delay might not be good. Jul 24 03:10:14 understood, so in my current case, I need to right sample values to the PRU SRAM one-time only before PRU0 starts sending out pulses, so I guess both approaches don't make much of a difference Jul 24 03:13:47 zeekhuge: Is such a function that sends 'blocks' of data required in the daemon then? Jul 24 05:25:08 I wouldn't think so. I would guess that writing to PRU should not be delayed (logic being, to change PRU's IO behaviour faster), although writing to ARM can somewhat be delayed. Jul 24 05:25:26 pratimugale ^ Jul 24 07:21:16 I dont understand why sometime bb refuse to connect with ssh Jul 24 07:57:49 pratimugale: https://github.com/pratimugale/PRUSS-Bindings/tree/master/examples/firmware_examples/example3-pwm Jul 24 07:58:00 what is that archive-pwm-c ? Jul 24 07:58:11 is it like the archive of some old code ? Jul 24 07:58:27 If it is that, then it should be mentioned in the README their. Jul 24 08:00:04 pratimugale: what is this example about ? https://github.com/pratimugale/PRUSS-Bindings/tree/master/examples/firmware_examples/example2-rpmsg-pru1/firmware/PRU_Halt Jul 24 08:01:57 pratimugale: have you actually tested it ? https://github.com/pratimugale/PRUSS-Bindings/tree/master/examples/firmware_examples/example4-analog-wave-gen Jul 24 08:03:20 okay. just saw the video. Jul 24 08:27:24 > is it like the archive of some old code ? Jul 24 08:27:25 Yes it is, will mention it in the Readme Jul 24 09:01:45 It is the rpmsg loopback example, to get started with rpmsg using this Documentation https://github.com/pratimugale/PRUSS-Bindings/blob/master/Documentation/RPMsg.md. The Halt part was mentioned in TI's lab 5 but I don't think it's necessary Jul 24 10:00:50 julieng: jkridner cwicks Hello, as I said yesterday, I won't be able to participate in the meeting. Today, I need to visit a doctor. Nothing serious, it's just takes a lot of time. Jul 24 14:06:03 jkridner : I couldn't look into implementing the greybus iio protocol class as I was travelling yesterday for attending an event and had poor connectivity, I'll reach back tomorrow morning (in 12hrs and get back to work) Jul 24 14:12:40 abhishek_ zeekhuge I was able to add another channel to the waveform generator right now https://github.com/pratimugale/PRUSS-Bindings/tree/wip/examples/firmware_examples/example9-multichannel-waveform-gen Jul 24 14:13:34 It took me quite some time as I faced a lot of small errors (specially at boundary conditions) Jul 24 14:15:42 abhishek_ zeekhuge Output Videos: y=sin(x) and (y = x periodic waveform): https://www.youtube.com/watch?v=WIY_7yH4kKo; two sine waves out of phase by 180 degrees: https://www.youtube.com/watch?v=7AHE-pOzfVA Jul 24 14:16:58 can anyone told me after seeing the circuit that is it the correct way to get input from one beagle bone to another .see the pin no. p8_22,p8_23,p8_24 Jul 24 14:17:00 https://github.com/pranav083/pocket_beagle-work/blob/wip/6.shift_74hc299_bi/v2/shift_74hc299bi_bbbv2_bb.png Jul 24 14:19:15 https://github.com/pranav083/pocket_beagle-work/blob/wip/6.shift_74hc299_bi/v2/shift_74hc299bi_bbbv2_bb.png Jul 24 14:26:55 zeekhuge abhishek_ I interweaved the sampling values of both the waves together in the PRU SRAM; something like this -> https://pastebin.com/jgKeFj5e; otherwise the assembly program was getting very complex Jul 24 14:34:39 abhishek_: Could you please have a look at the code if that is how the 'channels' were to be made Jul 24 14:35:21 pratimugale: nice. Will look into code in some time. Good work Jul 24 14:38:51 zeekhuge: ok, thanks Jul 24 14:44:56 hello hendersa Jul 24 14:45:36 pranav_kumar[m]: Hello Pranav. Did you get your SSH issue sorted out? Jul 24 14:50:19 yseterday while i was testing the CALL, JMP,JAL command using push button program ,I have made switch circuit and and give switch output to bb p8_07 pin input , all the logic are at 3.3v only . After that my BB stop connecting to ssh . It still boots up Jul 24 14:52:31 yes the probl;em was that while one device is register through the ssh with the ip it generally didnot connect. So i have to unregister the device ip then trying again solve the problem Jul 24 14:55:08 hendersa: I had test that BBB with with the ethernet connction also . The circuit reference . https://github.com/pranav083/pocket_beagle-work/blob/wip/2.shift_74hc299/BB_Black/switch_bbb/switch_bbb_bb.png just remove the logic level converter from the circuit and keep the voltage level at 3.3v for the circuit Jul 24 14:57:46 however my system still shows that the bbb is connected to my laptop Jul 24 14:59:48 So, it looks like you've broken out those STROBE, RDWR, and ACK lines now. Jul 24 15:00:02 But no timeout on master if an ACK isn't received. Jul 24 15:00:54 What C code have you been using to load the firmware and communicate with the PRU via shared memory? Jul 24 15:01:13 I don't see any checked in with your .asm files in the WIP branch. Jul 24 15:01:33 * pranav_kumar[m] uploaded an image: Screenshot from 2019-07-24 20-26-13.png (424KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/BrxenCmjEgNubugYIjgIydNx > Jul 24 15:01:36 ok just see the screen shot Jul 24 15:02:09 you can see three eth connections as three bbb are connected right now Jul 24 15:02:58 so as you said then this ckt is fine to work with Jul 24 15:04:38 Aside from those lecture slides I gave you about buses and their control lines, have you done any research into the mechanics of buses to determine any potential problems with your current implementation? Jul 24 15:04:51 * pranav_kumar[m] uploaded an image: Screenshot from 2019-07-24 20-34-10.png (198KB) < https://matrix.org/_matrix/media/v1/download/matrix.org/DRfcPRbkjfqpOHrgRLkKllOQ > Jul 24 15:05:00 this will be more clear to you Jul 24 15:05:04 Any independent research on the problem you are solving? Jul 24 15:06:39 i had done with the setup just see the ckt is it right as . I am doing the testing Jul 24 15:07:00 asking about any potential problem in voltage Jul 24 15:30:26 gm all Jul 24 15:31:09 jkridner[m]: Morning, boss. Jul 24 15:31:35 lol Jul 24 15:31:55 really overwhelmed today, so my attention will be a bit limited. Jul 24 15:32:07 cwicks: we still getting together this evening? Jul 24 15:33:05 all evaluations are due, so please get them done today. Jul 24 15:33:14 ravikp7: you around? Jul 24 15:33:43 pranav_kumar: you here? Jul 24 15:34:36 yes , here , good evening all, sorry debugging the ckt and code Jul 24 15:41:46 pranav_kumar: I see some screenshots above, but I'm not sure what I'm looking at. Jul 24 15:42:08 pranav_kumar: can you bring us all up-to-date on your blockers and accomplishments? Jul 24 15:42:37 * jkridner[m] reads https://groups.google.com/forum/#!topic/beagleboard-gsoc/TnMInDz8n2A Jul 24 15:42:55 yes , just give me some time so that i can write all things up Jul 24 15:43:31 k Jul 24 15:50:32 jkridner: yes! venue change to a little closer to you and more comfortable for documentation hackathon - @welsh's house Jul 24 15:51:00 cwicks: cool Jul 24 15:51:44 pratimugale: maybe you can just start with explaining where you've been spending your time? Jul 24 15:52:10 oops, sorry pratimugale .... bad autocomplete Jul 24 15:52:19 I meant pranav_kumar Jul 24 15:53:06 jkridner: oh.. alright Jul 24 15:55:26 pranav_kumar: maybe it seems overwhelming. if you are writing something big up, maybe you can save it and put it in an e-mail? Jul 24 15:56:05 pranav_kumar: for now, I'd just like to generate some back-and-forth to know where your time and energy are directed. Jul 24 15:59:17 My msg from laptop are not able to get send Jul 24 15:59:24 i am a slow typer .sry for that jkridner Jul 24 15:59:24 ok . I will put all together in the mail jkridner Jul 24 15:59:26 i am spending my full time for the project except for the last 6 days when i was outside from the hostel jkridner .And worked on the suggstion of keep things updated as suggested by other mentors Jul 24 16:00:15 updating my mentors daily about my progress through the mail Jul 24 16:01:44 off-list or on? Jul 24 16:02:53 the bi directional commication part will get complete by today , i will upodate the video by tonight. Jul 24 16:03:13 means???? Jul 24 16:03:38 Off-list. pranav_kumar[m] has sent daily mail directly to his mentors for the past ten days or so for updates. Jul 24 16:03:48 with the beagleboard-gsoc google group in copy? Jul 24 16:04:04 I don't see a lot of back-and-forth on https://groups.google.com/forum/#!topic/beagleboard-gsoc/TnMInDz8n2A Jul 24 16:04:39 I know that's mostly for weeklies, but it does provide a way to go back and forth in more detail and draw in as much help as can come from the #beagle-gsoc community Jul 24 16:05:15 hendersa: that's cool/good. Jul 24 16:05:57 pranav_kumar: is the scope of the project a bit much to try to execute against without more time coming up to speed? Jul 24 16:06:28 pranav_kumar: do we need more on-boarding tasks? like, are you going through all of the examples on PRU Cookboock (bbb.io/pru) Jul 24 16:06:48 i had started when mentors suggest to update daily things , istead of uppdating the things in bundle Jul 24 16:07:17 k. Jul 24 16:09:00 the assembly part is handle by me as it is bit clear for , except for the furher progress i need some asssistance from the mentors Jul 24 16:11:36 pranav_kumar: do you need more time to build up the required skills? is there still any possibility of hitting the milestones? Jul 24 16:12:11 no , I am on the testing phase of Codes , i had done with example parts till after the week and half after the second evaluation. The thing is that i got the real IC a bit late that make my project vulnerable. of the timming. Jul 24 16:15:02 Hi all Jul 24 16:15:09 jkridner: Hi Jul 24 16:15:12 pranav_kumar: are you able to spend all the extra time to catch up now? Jul 24 16:15:16 * sry that first evaluation Jul 24 16:15:21 howdy vaishnav98_ Jul 24 16:15:51 regarding the interfacing part and communication part . NO. But , Getting it ready for till the user space of the project , I can hit the milestones if i get some timely assistance Jul 24 16:16:22 sorry embden vaishnav98_ pratimugale : I never got around to calling roll Jul 24 16:16:52 jkridner: here Jul 24 16:17:01 gm Jul 24 16:17:42 Well, technically the meeting doesn't even start for another 10 minutes or so. Jul 24 16:17:47 i am spendind my full time on the project , except for the fact that my college has been started and not attending it . So that i can get the project on time as much as i can Jul 24 16:18:38 jkridner: great, I couldn't spend time on the project for the last two days as I was travelling to attend an event now I am returning back and will reach back in a few hours, could you give me a little more detailed overview of how the greybus iio protocol should work? Jul 24 16:20:41 vaishnav98_: I cannot. There are a number of presentations out there. I think a lot of the knowledge can come from working on a few kernel drivers independent of greybus. Jul 24 16:21:45 jkridner: like how it should deal with additional platform data like (details of an i2c mux device in the mpu9dof click) Jul 24 16:21:53 vaishnav98_: I noted that at least irq was part of the SPI platform data, so if that's all that is needed it can be satisfied in a generic way. Jul 24 16:23:11 it may be needed to come up with some kind of 'click' device to get a mapping between click header pins and host drivers. Jul 24 16:24:03 the same-ol-problem of coming up with overlays to support the devices against that pinout might be necessary, but it seems you might also be able to push for providing that data in-kernel and avoid overlays. Jul 24 16:24:50 that is, hard-code them (which is frowned upon in the ARM SoC Linux world today), but have it be OK as this is for a bus definition and not an SoC definition. Jul 24 16:25:09 er, guess I should say mikroBus rather than Click. Jul 24 16:25:39 * jkridner[m] just realized my computer clock was still on East-coast time.... Jul 24 16:25:52 didn't even realize we weren't full-on into meeting time. Jul 24 16:25:53 * jkridner[m] blushes Jul 24 16:26:19 jkridner: sure , I'll try to go through them, current most of the time I am spending on reading articles still couldn't figure out a generic way, for i2c devices I found that i2c_board_info contains an entry for of_node struct so can I search for the corresponding greybus node from the dt and attach it to the device? Jul 24 16:27:44 the timing of that event seems unfortunate... right around GSoC evaluations. What time have you spent on GSoC to now and what time will you spend forward? Is there any way to keep the original scope by building your expertise with experimentation (takes a lot of time), rather than being fed the details by your mentors? Jul 24 16:32:10 jkridner: I currently don't have a click /device which only has a interrupt only as extra platform data requirement , can you recommend a way to start with this break into simpler tasks without requiring a simpler device Jul 24 16:35:10 hmmm.... what if you worked up a driver that combined all the interfaces on mikroBus as a userspace rapid prototyping exercise and breadboarded some tests? Jul 24 16:35:38 for example, start with a gpio driver.... Jul 24 16:36:18 I think you've already seen that you can load a gpio driver using what is already there, but what if you did it in a "mikroBus" kernel module (ie., driver).... Jul 24 16:36:30 so, a driver loads another driver.... Jul 24 16:36:55 then, you can get a bit of a feel for what it takes to consume another driver in the kernel itself. Jul 24 16:37:15 you could add SPIDEV to your driver, then I2CDEV.... Jul 24 16:37:44 simply making your driver consume all those interfaces. Jul 24 16:38:34 functionally, it would just solve the problem of mapping a mikroBus pin to a physical pin on the board. Jul 24 16:39:09 pranav_kumar: are you working on an answer to jkridner question: " What time have you spent on GSoC to now and what time will you spend forward? Is there any way to keep the original scope by building your expertise with experimentation (takes a lot of time), rather than being fed the details by your mentors?" Jul 24 16:40:00 cwicks: I encourage pranav_kumar to also follow-up on those questions by e-mail as it seems his IRC fingers don't fly too fast. Jul 24 16:40:16 *encouraged Jul 24 16:40:19 Yes @cwicks:matrix.org Jul 24 16:40:47 jkridner: pranav_kumar got it - thank you Jul 24 16:41:10 trying to play catch up in the meeting 😊 Jul 24 16:41:47 vaishnav98_: did you have a chance to provide your update? Jul 24 16:43:30 jkridner : great , I think it would be better to split up into tasks like these to finally achieve clickboard support , trying to bring in support for a particular click mostly doesn't seem to have worked out well, but I can confirm that the additional platform data seems to be the only blocking thing as when I hardcode them in the greybus spi_device_config call , Jul 24 16:43:31 the click seems to work well Jul 24 16:48:19 jkridner cwicks ravikp7: last week I tried to bring in device tree support for greybus for discovering greybus nodes from the device tree: I was able to parse the platform data for a particular click device from the device tree and cast it to the particular platform_data structure for the corresponding click (if this structure is known already), then passing the data Jul 24 16:48:20 to the spi_new_device call successfully instantiates the device, but this doesn't sem to be generic method as this platform_data struct for each click board device is very different from each other Jul 24 16:50:11 jkridner cwicks ravikp7: also for the last the last two days I was out of station and was having poor connectivity , I am travelling back to home now and will resume the work in a few hours Jul 24 16:53:35 vaishnav98_: casting platform data structures sounds risky. Jul 24 16:54:23 vaishnav98_: rather than change the greybus driver itself, give it this shot to create a driver that loads that driver. Jul 24 16:55:09 mdp: let me know if the idea of creating a 'mikroBus' driver that consumes the greybus driver is crazy. Jul 24 16:55:35 jkridner: also I think I have support for most of the clicks(which I currently has acces to) figured out after solving the platform data issue except the ultrasonic 2 click which uses the pga460 ultrasonic signal processor, I wasn't able to find a Linux driver for the same does it exist Jul 24 16:57:39 jkridner: I got a lot of kernel panics if I did something wrong, I understand that it's not a good practice but I was able to use it for testing device instantiation Jul 24 16:58:24 jkridner: sure I'll work on this suggestion and will update you soon Jul 24 17:00:01 looks like you might need to write it. ;-) Jul 24 17:00:12 jkridner cwicks: sorry for the delay in providing the updates , I was typing it from a moving bus Jul 24 17:00:24 cwicks: you driving the meeting from here? Jul 24 17:00:41 yes I can jkridner Jul 24 17:01:48 Thanks for the extra effort from a moving bus! vaishnav98_ Jul 24 17:02:13 Did I see that embden_new was at a dr appointment during the meeting today? Jul 24 17:02:38 jkridner: sure I'll try to do it after figuring out this platform data issue as per your suggestion, I have done some research on the device communication Jul 24 17:02:47 yes Jul 24 17:02:47 cwicks: :-) Jul 24 17:02:55 embden_new can't make it this week per his email. Jul 24 17:02:59 Any mentors have an update on his blocker from the weekly report "I'm trying to add the irq number translation function for the crossbar (it is not working correctly for now)" Jul 24 17:05:00 Hi pratimugale are you ready to make your update? Jul 24 17:05:14 cwicks: hi Jul 24 17:05:18 yes Jul 24 17:05:41 pranav_kumar: also, please complete your evaluation. pratimugale vaishnav98_ embden thank you for completing your evaluations. Jul 24 17:06:19 ok, will do it by tonight Jul 24 17:07:19 * pratimugale[m] sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/wfOVnzBjXLyKlNlnDUqOAFfK > Jul 24 17:07:37 no update - he's stuck debugging why the virtual interrupt controller he's adding to Xen is not working Jul 24 17:12:14 abhishek_ zeekhuge: Could you please guide me on how to proceed after completing this example; i.e. if more examples are to be added, I was thinking of FSK modulation, and something based on UART. Jul 24 17:13:29 pratimugale: i would suggest to work on documenting the examples in a more friendly way. Jul 24 17:13:31 ds2: cwicks[m]: The latest update on the IRC was yesterday. He was looking at why Linux is not detecting the crossbar. Jul 24 17:14:16 pratimugale: I made a hack pruss_uio_shmem driver: https://github.com/beagleboard/BeagleBoard-DeviceTrees/pull/3 Jul 24 17:14:35 ds2: ^^ Jul 24 17:14:36 thanks julieng Jul 24 17:14:51 zeekhugeabhishek_ ^^^ Jul 24 17:15:00 it needs a lot of love. Jul 24 17:15:01 thanks ds2 Jul 24 17:15:21 rpmsg is great and all, but most people still want to use shared memory communications.... Jul 24 17:15:23 julieng: isn't that the same as it not working ? or does the Xen UART fully work now and it is just linux? Jul 24 17:15:38 but, the uio driver conflicts with the rproc driver.... Jul 24 17:15:44 so, you have to use root access and /dev/mem to do the memory mapping. Jul 24 17:15:57 so, I just removed a bunch of blockers from the uio driver so it can run at the same time. Jul 24 17:17:06 ds2: IIRC the UART is now fully working in Xen. Now he is looking at Dom0. Jul 24 17:17:14 pranav_kumar: Is there an update you'd like to make in the meeting today - or are you just sending it via email? Jul 24 17:17:38 julieng: oh nice... didn't catch that first part Jul 24 17:18:17 I will send all via email , thanks for the call @cwicks:matrix.org Jul 24 17:19:55 rpmsg doesn't fit all the PRUSS use cases Jul 24 17:20:15 a kernel driver is better Jul 24 17:20:43 i.e been playing with the idea of a PRUSS based I2S interface Jul 24 17:21:47 ds2 : Isn't it mostly because for most of the cases one has to use pru1 for data read-write and other for the IO? Jul 24 17:22:29 So, if we had a "working" EDMA, things would have been better. Jul 24 17:23:33 zeekhuge[m]: yes... it was one of the driving reasons for the DMA project Jul 24 17:23:34 jkridner: I will have to study more about it, I'm not very familiar with .dts and the codebase at the moment Jul 24 17:24:49 zeekhuge[m]: there is a lot of parallels with the old NE2000 ethernet chip (dumb P/O vs other cards of that era that supported DMA,etc) Jul 24 17:26:46 Yeah. EDMA project is kind of outdated now. Not sure the amount of effort that will be required to revamp it. Jul 24 17:30:14 zeekhuge: you had suggested making a kernel driver, could you please explain what it would be needed for and how to go about developing it Jul 24 17:31:46 pratimugale: it would be a very trivial exercise. The default driver uses some "rpmsg30" As the key. I was just thinking to keep a dedicated driver with "pruss-api" Or something relevant as the key. Jul 24 17:32:32 wonder what would be involved in having a RTrpmsg variant Jul 24 17:34:42 pratimugale : by "key" I mean the device id thats required to probe the driver Jul 24 17:34:48 Oh.. I see, will get on it after the current work Jul 24 17:34:53 https://github.com/ZeekHuge/BeagleScope/blob/master/examples/kernel_examples/n-blinky/module/rpmsg_pru_parallel_example.c#L235 Jul 24 17:34:59 The device id. Jul 24 17:35:59 pratimugale: the corresponding firmware needs the same channel name Jul 24 17:35:59 https://github.com/ZeekHuge/BeagleScope/blob/master/examples/kernel_examples/n-blinky/firmware/main_pru1.c#L65 Jul 24 17:36:35 This is how the kernel probes the right driver Jul 24 17:36:37 By the channel name Jul 24 17:38:10 zeekhuge: Understood Jul 24 17:39:34 Will look into it and get back to you by tomorrow evening, I have a test tomo for which I will have to prepare for the rest of the night Jul 24 17:40:17 pratimugale: tests, like midterms? Jul 24 17:40:36 So for next 2 days or what? Jul 24 17:40:36 no, just a single one Jul 24 17:40:37 small Jul 24 17:40:53 Okay. Jul 24 17:41:03 Good luck. Jul 24 17:41:24 zeekhuge: Thank you :) **** ENDING LOGGING AT Thu Jul 25 02:59:57 2019