**** BEGIN LOGGING AT Wed Jun 18 02:59:59 2014 Jun 18 03:14:08 hi mranostay Jun 18 03:18:55 jkridner: how is the beaglepilot stuff coming along? Jun 18 03:19:59 I see lots of commits and vmayoral seems to be happy with the userspace driver stuff, but getting stuff into the kernel seems to be moving slowly. Jun 18 03:21:34 I can't say that any of the projects are really hitting it out of the park right now. I feel the second half is going to require some additional effort from most of the students. The tutorial thing never happened and not even I made it to the super-early Monday morning meeting. :( Jun 18 03:22:03 several of the students did make it, but I haven't seen more IRC activity because of it. Jun 18 03:22:22 really hard keeping this channel lively and engaged with the details. Jun 18 03:22:44 I'm going to be trying to run code from all of the projects in the next week or so. Jun 18 04:41:37 Morning Jun 18 04:44:46 Hi mranostay, jkridner Jun 18 04:45:30 hi Abhishek_ Jun 18 04:50:52 jkridner: likely a day more before BeagleLogic captures samples. Jun 18 04:51:25 putting the last pieces of the new PRU firmware and the kernel driver now Jun 18 06:06:45 morning ! Jun 18 06:07:04 hey Jun 18 06:08:31 * vvu starts to hate a bit public transportation Jun 18 06:11:08 the PRU firmware C port had a lot of bugs. Had. Jun 18 06:11:33 * vvu +1s the above statement Jun 18 06:30:48 are the USR leds broken out on the header pins? I'm looking for something to quickly test and hexdump Jun 18 06:31:11 (from /dev/beaglelogic :) ) Jun 18 07:03:07 * karki misses his platelets :( Jun 18 07:03:40 Abhishek_: yeah they are a GPIO Jun 18 07:05:51 * mranostay hides from joelagnel Jun 18 07:28:51 mranostay: looks like the LED pins aren't brought out on the headers, so might not be used for testing Jun 18 07:29:12 well not headers but it is a gpio Jun 18 07:29:42 depending on how well you solder you could red wire it one Jun 18 07:29:44 *to Jun 18 07:40:46 * Abhishek_ pulls out his hardware kit and soldering iron Jun 18 07:52:32 just stuck in a right angled IDC connector into the BBB connections and used some jumpers instead :) Jun 18 08:31:20 * Abhishek_ braces to connect the BeagleBone Black to a live I2S audio bus Jun 18 09:05:38 * Abhishek_ is currently testing BeagleLogic with this : http://www.theembeddedkitchen.net/stm32/an-expansion-board-for-the-stm32f4discovery/ Jun 18 09:17:27 captured my first 8 MB of samples :) Jun 18 09:19:25 good Jun 18 09:19:49 goodie Jun 18 09:19:51 print them and attach to your thesis report :) Jun 18 09:20:32 16-bit, 100MSa/s Jun 18 09:21:03 I2S? Jun 18 09:21:08 yep Jun 18 09:21:19 so you can recover the audio from that? Jun 18 09:21:25 I don't know but the audio output became noisy Jun 18 09:21:35 I might have to buffer the signals Jun 18 09:22:00 (remove a 200ohm series resistor between audio DAC and microcontroller) Jun 18 09:22:36 it seems sigrok can do that, lemme see if audio can be recovered Jun 18 09:39:49 vvu: on another channel: speaking of beagles, there's some dudes on the linux-usb mailing list who are apparently poking with using an android device connected via usb to a beaglebone to be the display, which sounds neat Jun 18 09:54:39 :) Jun 18 09:54:48 we let out the secret to early Jun 18 10:12:16 :) Jun 18 13:11:04 greetings guys! Jun 18 13:11:53 Abhishek_, av500: just catching up with your messages from some hours ago. Isn't that "using an android device connected via usb to a beaglebone to be the display" praveen's work? Jun 18 13:12:03 somebody hijacked it? Jun 18 13:14:21 yep :) http://thread.gmane.org/gmane.linux.usb.general/109974 Jun 18 13:40:32 vmayoral|pc: ^ Jun 18 13:40:55 Abhishek_: yes? Jun 18 13:41:33 that was the thread we were discussing Jun 18 13:53:27 Abhishek_: thanks :) Jun 18 14:23:30 jkridner: Good Morning, I send you a chat today on the morning 4AM your hour. I started to created new files on the github repo. This is necessary because after creating the tutorial, it will redirect to a page similar like this bone101/tutorials/{tutorial name} or {gist id }. Jun 18 14:26:33 Just to confirm that's the idea you have that after each tutorial with an address like this: bone101/tutorials/{tutorial name} or {gist id } Jun 18 14:33:52 jkridner: How about a video update today? Jun 18 14:48:34 Abhishek_: video update? Jun 18 14:49:02 yep, if I am able to upload it in time Jun 18 15:15:04 Abhishek_: that'd be great Jun 18 15:17:47 jkridner, likely I won't make today's meeting, have to interview someone Jun 18 15:18:02 joel_, ^^^ Jun 18 15:33:27 uploading :) Jun 18 15:34:12 quite slow though, ETA 40mins Jun 18 15:37:27 hi rseethamraju Jun 18 15:37:35 hey! Jun 18 15:38:00 how's your project going? Jun 18 15:38:09 hey Jun 18 15:39:30 good! Jun 18 15:39:41 how’s yours Jun 18 15:39:42 ? Jun 18 15:44:54 rseethamraju: I was a little delay, but this week i catch up Jun 18 15:45:07 extreme programming nigths :) Jun 18 15:45:58 hi vmayoral|pc Jun 18 15:46:07 DiegoTc: hi Jun 18 15:46:15 how are those cards going? Jun 18 15:46:16 vmayoral|pc: you're from Italy? Jun 18 15:46:40 DiegoTc: I live in Italy but I'm spanish Jun 18 15:46:57 planning a trip to Italy? Jun 18 15:46:58 futbol fan? Jun 18 15:47:39 I will like to, but there's one issue for making the trip. Money :( Jun 18 15:48:07 DiegoTc: Not so much into professional soccer. I am more a player than a watcher. Jun 18 15:49:06 vmayoral|pc: I guess that like a good español, you are with Spain on the world cup. Good Luck Today! Jun 18 15:50:34 DiegoTc: Thanks! Jun 18 15:50:54 DiegoTc: Aside from soccer, how is it going with your project? Jun 18 15:51:36 I was delay, last week, but I catch up really fast this week! Jun 18 15:51:53 The only issue is design right now, but we're working on that Jun 18 15:52:16 DiegoTc: did you send an update this week? Jun 18 15:52:23 jkridner: yes Jun 18 15:52:28 video in 20mins Jun 18 15:56:12 Hi gregkh Jun 18 15:56:20 Abhishek_: hi. Jun 18 15:56:28 * jkridner tries to quickly read the status reports ahead of the meeting (in 4 minutes) Jun 18 15:56:41 hi all..... Jun 18 15:57:36 <_av500_> hi Jun 18 15:58:28 ahoy Jun 18 15:59:23 oh hey, has anyone tried the eQEP driver? Jun 18 16:00:13 alexanderhiam: on my short to-do list, but I haven't made it past simply enabling it. Jun 18 16:00:31 it always reports a position of 0 for me... Jun 18 16:00:37 hello from my contract manufacturing house! Jun 18 16:00:55 Hey! Jun 18 16:01:06 hi VoltVisionSteve Jun 18 16:01:21 Hi VoltVisionSteve, did Oscar told you something? Jun 18 16:02:28 DiegoTc, I will ping him right now in skype to see where he is at. Did you get his email that he was intending to send? Jun 18 16:02:53 no I didn't you. Did you get it VoltVisionSteve ? Jun 18 16:03:03 hi all, let's get started by checking in on all the status reports from each of the students. Jun 18 16:03:20 posted mine Jun 18 16:04:13 DiegoTc, no I didnt. I am calling him now. Jun 18 16:04:18 jkridner: I believe rseethamraju is typing hers right now Jun 18 16:04:19 I have DiegoTc's latest at https://groups.google.com/d/msg/beagleboard-gsoc/mPa69fBjkmg/G-5GDNwN4FEJ Jun 18 16:04:32 ya sorry 2 min Jun 18 16:04:37 Here's mine: https://groups.google.com/forum/#!topic/beagleboard-gsoc/R0sCsWK-8JQ Jun 18 16:04:57 I have vmayoral|pc's latest at: https://groups.google.com/d/msg/beagleboard-gsoc/BISQ3fJj3P4/RCTt08Jp8KMJ Jun 18 16:05:02 Thanks Abhishek_ Jun 18 16:05:46 btw all, vmayoral|pc's format is the one desired: accomplishments, issues, plans. Jun 18 16:06:02 jkridner : sick the whole week. back to work from tomorrow :) Jun 18 16:06:17 anyone seen praveen? Jun 18 16:06:19 av500 ? Jun 18 16:06:31 karki: feeling better then? Jun 18 16:06:32 <_av500_> jkridner: yes, he is on irc at times and also commiting Jun 18 16:06:46 <_av500_> latest progress repot is missing, (angry) mail sent Jun 18 16:07:06 <_av500_> I will chase him to report status until the end of the week latest Jun 18 16:07:52 k. all, remember mid-terms are next week... this is a *really* bad time to be missing status reports! Jun 18 16:08:42 karki: sending a status report talking about the impact of your illness should be an assumed necessary task as you can manage. Jun 18 16:08:43 jkridner alexanderhiam : watched the video too. as for my health, fever is gone :D (but my platelet count is terribly low. If it falls lower, I'll be admitted to a hospital, but still should be able to code) Jun 18 16:09:17 done. https://groups.google.com/d/msg/beagleboard-gsoc/i4VLmPyWFAI/zAt_hzQqD9IJ Jun 18 16:09:25 Abhishek_: did you say there was a video for https://groups.google.com/d/msg/beagleboard-gsoc/R0sCsWK-8JQ/UkhaKaVd8LkJ as well? Jun 18 16:09:43 ah, uploading Jun 18 16:09:45 karki: glad to hear the fever is down, and who need platelets to write code anyways? ;) Jun 18 16:09:52 93% now Jun 18 16:09:54 alexanderhiam: ;-) Jun 18 16:10:01 <_av500_> use codelets Jun 18 16:10:13 :P Jun 18 16:11:09 rseethamraju: thanks, I see it now Jun 18 16:11:40 anyone seen disdi_? Jun 18 16:11:57 joel_, gregkh, disdi_: any reason no status report? Jun 18 16:12:14 last one I see is https://groups.google.com/d/msg/beagleboard-gsoc/cBc2Mf6LjTM/KO5bs14msHkJ Jun 18 16:12:32 jkridner : status report at the end of the meeting? I still have goals to discuss.... Jun 18 16:12:57 karki: better late than never, but deadline is before this meeting Jun 18 16:13:15 jkrinder: I am working on it Jun 18 16:13:16 karki: understand you've been very sick Jun 18 16:13:46 jkrinder: there are some discussions with joel before I tell my progress Jun 18 16:14:06 jkridner : okay I'll update in a 10 min :) but I had some tweaks in my TODO list based on the video I saw Jun 18 16:14:18 k, I'd like to whip around the room and discuss issues mentioned in the reports, but is there any other agenda topics? I know there was some issues with blown-up boards. Jun 18 16:14:34 * jkridner isn't seeing that mentioned in any of the status reports. Jun 18 16:15:07 let's go FIFO-style... vmayoral|pc: it seems you have a long list of issues. Jun 18 16:15:09 @All, BeagleLogic live: www.youtube.com/watch?v=UYRSvTC6VUI Jun 18 16:15:41 is the PREEMPT_RT discussion now being discussed on linux-omap or on another real Linux list? Jun 18 16:15:49 jkridner: yeap, i keep accumulating aspects non-strictly related to my scope Jun 18 16:16:45 vmayoral|pc: which of these are related to your scope and blocking? Jun 18 16:16:45 jkridner: following your recommendation i posted it there but i haven't seen much discussion. Still after the conversations with panto and ds2 (and the results with vanilla) we decided to focus there for now Jun 18 16:16:56 can you annotate blocking ones in your report? Jun 18 16:17:18 jkridner: at the time none. They are all out of the gsoc scope Jun 18 16:17:20 i will do so Jun 18 16:17:39 jkridner: this week, as planned i started the review of the ROS-BeaglePilot integration Jun 18 16:18:10 jkridner: can take a look at what has been done so far http://erlerobotics.gitbooks.io/erlerobot/en/mavlinkros.html Jun 18 16:18:24 vmayoral|pc: k, let's move on to DiegoTc then and come back if we have time if issues are non-blocking. Jun 18 16:18:52 DiegoTc: you didn't list off issues. are you blocked waiting on VoltVisionSteve's Guatemala team? Jun 18 16:19:04 one minute. Jun 18 16:19:38 jkridner: yes & no Jun 18 16:19:59 jkridner, We had a great meeting on Monday and there are at least 2 designers on the task of helping him....I am pushing them for an update and progress right now. Jun 18 16:20:01 vmayoral|pc: maybe it is loading slow for me, but I'm not finding install/run steps to reproduce your work. Jun 18 16:20:28 I continue working with design. I didn't push the changes to the repo of git pages. But I have been working on a new design it's pushe on the branch nodejs https://github.com/DiegoTc/bone101 Jun 18 16:20:50 jkridner: you have to review the subsections of the link i facilitated, sorry, my bad (e.g.: http://erlerobotics.gitbooks.io/erlerobot/en/mavlink_ros.html) Jun 18 16:21:21 jkridner: that link is sort of like a summary of the ROS packages analyzed Jun 18 16:21:34 * vmayoral|pc is linking subsections with hiperlinks Jun 18 16:22:38 DiegoTc: are we aligned on not making changes to the server except to perhaps add an oauth2 service? Are you still trying to put the authentication and secret into the server? Jun 18 16:22:47 nop Jun 18 16:23:02 why not? Jun 18 16:25:39 jkridner: I'm using the example you told me Jun 18 16:26:03 https://oauth.io/ Jun 18 16:26:11 all in the client side, no server side Jun 18 16:26:22 oh, so we ARE aligned then, yes? Jun 18 16:26:38 I see, you were saying no to my 2nd question... Jun 18 16:26:46 my bad for asking 2 questions. Jun 18 16:26:51 * jkridner should know better Jun 18 16:27:07 You're correct with what you said Jun 18 16:27:09 vmayoral|pc, could you point me to the preempt_rt thread? thanks Jun 18 16:27:22 jkridner: question about https://oauth.io/ Jun 18 16:27:33 https://github.com/DiegoTc/bone101/tree/nodejs/Support/GSOC doesn't currently reflect this. Jun 18 16:27:43 DiegoTc: you still need to push your latest version, right? Jun 18 16:27:56 correct Jun 18 16:28:04 the gist you posted was created from a client-side only app? Jun 18 16:28:11 correct Jun 18 16:28:17 just client side Jun 18 16:28:25 vmayoral|pc: joel_ +1... I haven't seen the thread Jun 18 16:28:44 jkridner, joel_: sure, let me fetch it Jun 18 16:28:47 DiegoTc: why the yes and no? Jun 18 16:28:53 jkridner: https://github.com/DiegoTc/bone101/blob/master/Support/index11.txt this one also. https://github.com/DiegoTc/bone101/blob/master/Support/index11.txt Jun 18 16:29:56 karki: thanks for the posted update Jun 18 16:30:05 DiegoTc: are you blocked or no? Jun 18 16:30:10 that was for the design, yes I am a little blocked because it doesn't depends on me. No I will continue working on that. I will not stay waiting Jun 18 16:30:43 jkridner, joel_: the conversation started https://groups.google.com/forum/#!topic/beaglepilot/7DKcdm0AEPo Jun 18 16:30:59 vmayoral|pc: that isn't a linux list Jun 18 16:31:01 jkridner: wondering if I'm being clear of what I'm trying to tell you Jun 18 16:31:02 jkridner, joel_: and later it was moved to drones-discuss group (audience matters) https://groups.google.com/forum/#!searchin/drones-discuss/PREEMPT/drones-discuss/7DKcdm0AEPo/gL31uMu9vkAJ Jun 18 16:31:19 jkriner: mmm you refer to the linux mailing list Jun 18 16:31:38 jkriner: i sent a message direcly to the list, let me search that out Jun 18 16:31:44 DiegoTc: I'm trying to understand if your able to stay productive and what you'll be doing while waiting on the Guatemala team Jun 18 16:31:56 your = you are Jun 18 16:32:10 * jkridner hates that typo Jun 18 16:32:21 The answer is yes. A lot of work to do. Jun 18 16:32:28 good. Jun 18 16:32:30 I can;t depend on them Jun 18 16:32:47 jkridner: question. Bone101 will have a limit of 1000 users monthly Jun 18 16:33:01 DiegoTc: putting the page into git and sharing seems to be the next task, right? Jun 18 16:33:14 DiegoTc: with oauth.io, yeah, that'd seem to be the case. Jun 18 16:33:35 I can look at starting an oauth service with the same API on beagleboard.org. Jun 18 16:33:44 ...the guatemalans are just helping with "pretty" graphic design...Diego is making the functionality work. Jun 18 16:33:47 vmayoral|pc, thomas gleixner added this "debug_objects" framework to the kernel, its useful for tracking object corruptions in the kernel Jun 18 16:34:05 at first thought, I thought you were see some real-time or performance related issues Jun 18 16:34:21 having a service that simply does oauth seems like the right approach to me rather than building it in as a requirement of the server hosting the tutorials. Jun 18 16:35:08 VoltVisionSteve: understood. I just wanted confirmation from DiegoTc that he feels comfortable progressing on that while waiting on the design input. Jun 18 16:35:24 oger Jun 18 16:35:27 roger Jun 18 16:35:44 vmayoral|pc: all of the linux lists are archived. Jun 18 16:36:09 vmayoral|pc, so the Xenomai patched kernel works fine for you? It appeared from the thread you were settling for that Jun 18 16:36:31 jkridner: found the mail. It was bumped. I resent it to you Jun 18 16:36:38 Abhishek_: watched your video. thanks. Jun 18 16:37:28 jkridner: I was planning on working on the edit the card and adding the page to git will be my plan Jun 18 16:37:41 I don't see sharing with high importance Jun 18 16:37:47 right now Jun 18 16:38:10 Abhishek_: you didn't list any issues. Does that mean you are having good progress and nothing is slowing you down? Jun 18 16:38:41 jkridner: you finish with me? Jun 18 16:38:43 jkridner: is that group down? I have a mail from you confirming the e-mail of the list linux-omap@vger.kernel.org. Weird :( Jun 18 16:38:54 DiegoTc: k. I'll be looking at the status of the software you've developed on Monday for the mid-term evaluation. Please have some code to show progress checked in by then. Jun 18 16:39:03 joel_: Xenomai didn't suffer from the issues with the capemgr Jun 18 16:39:04 DiegoTc: yeah, moving on to Abhishek_ Jun 18 16:39:14 joel_: but it might require lots of work on the driver side Jun 18 16:39:16 DiegoTc: feeling you aren't blocked. Jun 18 16:39:27 jkridner: things are going fine, the code needs review on some issues like correct use of locks Jun 18 16:39:44 joel_: thereby we discarded it. Furthermore, exploiting Xenomai properly would require to code for the "xenomai kernel" Jun 18 16:39:45 Abhishek_: who are you looking to provide that review? Jun 18 16:40:05 vmayoral|pc, will be interesting to just preempt-rt patch a kernel and load random cape, I may try it as some point, this week is a bit crazy Jun 18 16:40:26 did it and proved that the capemgr crashes Jun 18 16:40:36 vmayoral|pc, so you do have some advantages to using the regular kernel and would like to, correct? or are you moving on and ok with xenomai Jun 18 16:40:38 * vvu pops into the scene Jun 18 16:40:59 vvu have you seen praveen? Jun 18 16:41:00 vmayoral|pc, ok so it is easily reproducible, that's good. Jun 18 16:41:15 spoke with him yesterday Jun 18 16:41:16 vvu: please help get praveen to submit timely status reports. Jun 18 16:41:23 joel_: moving on with vanilla (and also with vanilla+PREEMPT) for now Jun 18 16:41:31 jkridner: when i see him online i will, his internet connection is really bad as I know Jun 18 16:41:36 vvu: av500 has sent multiple requests that have gone unresponded. Jun 18 16:41:38 joel_: anyhow i set up a FS with the 4 kernels (and modules) Jun 18 16:41:40 and that is why he is not online from time to time Jun 18 16:41:47 joel_: so i can test it in all of them Jun 18 16:41:48 jkridner: That was intended for everyone. I just wanted to show how the kernel part of BeagleLogic works and how it can be used and tested right now by anyone. Interfaces will be added later, but this is the core I wanted to show. Jun 18 16:42:02 jkridner: i kept track of what he is doing but reports are goodie Jun 18 16:42:04 * joel_ hacked capemgr at one point and made it capable of static linking, that eliminated ~40% of the code :D Jun 18 16:42:19 jkridner: oops, I meant to say panto and mranostay Jun 18 16:42:21 Abhishek_: who in particular? what is being done to make sure you get the review you need? Jun 18 16:42:38 Abhishek_: ah, k. panto, mranostay: are you able to review the locks? Jun 18 16:42:47 joel_: probably hacking the capemgr and figuring out what's the issue with the PREEMPT_RT patches would be the best choice Jun 18 16:42:57 vmayoral|pc, wait. vanilla is the one crashing and xenomai works correct? I'm confused Jun 18 16:43:03 joel_: unfortunately... timing issues with GSOC Jun 18 16:43:48 joel_: no, the one crashing is RT_PREEMPT. The ones working are vanilla, vanilla+PREEMPT (PREEMPT option when compiling the kernel, no patches applied) and Xenomai Jun 18 16:44:05 ah ok! ok fine, I assumed you rt_preempt patch it in all cases, cool. Jun 18 16:44:27 joel_: if you haven't seen it, i believe you might find this report interesting http://erlerobot.com/blog/beaglepilot-cyclictests-different-kernels/ Jun 18 16:44:31 panto, mranostay: ping. Jun 18 16:44:37 pong Jun 18 16:44:48 mranostay: are you able to look at the locks in Abhishek_'s code? Jun 18 16:45:00 * jkridner is trying to figure out if it is a reasonable expectation. Jun 18 16:45:14 sure one second. Abhishek_ have a link? Jun 18 16:45:35 vmayoral|pc, thanks for pointing to the report, it will be an interesting read.. I'll let you know when I get a chance to if I find anything Jun 18 16:46:11 Abhishek_: please provide ASAP so we can confirm no blocking issues and move on to rseethamraju. Jun 18 16:46:18 joel_: np. Let me know if you wish to reproduce anything and you have issues Jun 18 16:46:30 vmayoral|pc, cool, will do. Jun 18 16:46:38 Abhishek_: or at least say that you'll post momentarily and that there are no blocking issues. Jun 18 16:46:39 mranostay: https://github.com/abhishek-kakkar/BeagleLogic/blob/master/beaglelogic-kernel-driver/beaglelogic.c Jun 18 16:46:43 Abhishek_: thanks! Jun 18 16:46:52 joel_: I stored the kernels, configs and modules at https://github.com/erlerobot/kernels. Could be useful Jun 18 16:47:38 rseethamraju: how far behind are you with having these exams? Jun 18 16:48:08 I’m supposed to be done with the camera by 23rd Jun 18 16:48:14 alexanderhiam: are you comfortable rseethamraju has been able to make good progress? Jun 18 16:48:15 joel_: (they are meant to be used wit BB-boards). For the BBB kernels should work but u-boot related aspects should be worked out Jun 18 16:48:15 sure Jun 18 16:48:29 and I did SPI before that’s actually scheduled for later Jun 18 16:48:30 Abhishek_: so are you halting the PRU when you do the writes its memory? Jun 18 16:48:38 mranostay: yes Jun 18 16:48:54 oh what is the question about the locks in the driver? Jun 18 16:49:03 vmayoral|pc, gotcha, I may just use my own config to begin with. Jun 18 16:49:06 jkridner: yeah, she'll be right back on schedule by midterm Jun 18 16:49:14 joel_:actually that's an interesting point, @jkridner, i tried BB-working kernels in a BBB and it didn't work. I was expecting it to work out of the box. Can we get an explanation why the difference? Jun 18 16:49:33 rseethamraju: and getting stuff soldered down won't slow you down too much? Jun 18 16:49:54 no because that’s actually preponed Jun 18 16:49:57 vmayoral|pc, did you change the dtb? bbb and bb have different dt Jun 18 16:49:58 jkridner: we also moved things around a bit by adding an eQEP library to the list Jun 18 16:50:18 but I’ll get it done romorrow I don;t have a soldering machine at home so Jun 18 16:50:21 vmayoral|pc: the hardware difference? you using old, unpatched kernels on BBB? that could kill boards! there is a voltage rail that is different on the board. Jun 18 16:50:36 jkridner, pong Jun 18 16:50:40 ya eQEP wasn’t there on the schdule in the first place Jun 18 16:50:49 panto, mranostay: Suitable placements of spinlocks in IRQs where the buffer states are being changed , wait queues to wake up reads after buffer is ready. Jun 18 16:51:04 and then I tried fixing the stepper motor library that still doesn’t work Jun 18 16:51:12 they work in the video but I understand there are cases wherein it may lock. Jun 18 16:51:31 jkridner, joel_: good to know. I didn't even change the dtb's. Jun 18 16:51:34 koen didn't like publishing the patches for the 3.2 kernel to work on BBB, but it is rather trivial. Jun 18 16:51:36 and in beaglelogic_f_read Jun 18 16:51:41 panto: aren't spinlocks in a interrupt handler a bad idea? Jun 18 16:51:50 * mranostay has a meeting to go to.. bb in 30 Jun 18 16:51:52 very Jun 18 16:51:53 panto: just looking at a review of Abhishek_'s code. Jun 18 16:52:01 jkridner: I did the C optimisation before coding started. So by finishing the camera library I’ll be right on track Jun 18 16:52:22 well, not per-se, but you need to be extra careful, and it's bad form Jun 18 16:52:25 however Jun 18 16:52:30 if you use a threaded irq no problems Jun 18 16:52:37 rseethamraju: thanks. will move on as I don't hear any blocking issues. Jun 18 16:53:06 karki: any comments on what you saw in the video or other issues? Jun 18 16:53:06 ya nothing much. alexanderhiam helps a lot. Jun 18 16:53:50 panto, what other option if you need to use locks in non-threaded irq :) Jun 18 16:54:18 karki: you still around? Jun 18 16:54:38 jkridner : ya, couple of points Jun 18 16:54:51 panto: or if locks are really needed based on the situation and design, please look at the code and advise Jun 18 16:55:03 I didn't see you bring them up as issues in the status report. what do you need clarified? Jun 18 16:55:39 jkridner : for one, there will be a stream interface (read socket) which will expose the botspeak interpreter Jun 18 16:55:56 read? Jun 18 16:56:04 for output from the interpreter? Jun 18 16:56:12 not read/write socket? Jun 18 16:56:23 there I go again asking multiple questions at a time. Jun 18 16:56:25 jkridner : I meant read it as socket Jun 18 16:56:29 ;) Jun 18 16:56:36 bi-directional? Jun 18 16:56:41 yes Jun 18 16:57:09 will you implement the commands to start, end and run a script as well as immediately interpret a command? Jun 18 16:57:43 once you answer that, will there be interpretation in the driver or all on the PRU? Jun 18 16:57:43 I still don't get the start and end script thing..... Jun 18 16:57:48 :( Jun 18 16:57:58 I don't get why we need that Jun 18 16:58:18 chris' interpreter on Arduino and javascript runs commands immediately..... Jun 18 16:59:13 so, if you send "SET DIO[13],1", that IO is immediately set high. Jun 18 16:59:47 later folks Jun 18 17:00:10 heading back to work.... Jun 18 17:00:28 jkridner : and about the interpreter, the Botspeak server cum interpreter is the interface the client finally gets. He won't have to use a python based API! the end user just talks to a socket! Jun 18 17:00:38 See https://sites.google.com/site/botspeak/the-language/lingo Jun 18 17:01:09 karki: so, if you send a 'SCRIPT' command, the mode of the interpreter switches from executing script commands immediately to saving them off into memory. Jun 18 17:02:13 The 'ENDSCRIPT' command terminates the mode, ending the script loading into memory. Jun 18 17:02:30 The 'RUN' command then executes the script. Jun 18 17:02:49 This is a 'streaming' interface, where commands are interpreted upon getting \n characters. Jun 18 17:03:21 Rather than a 'shared memory' interface, where the host processor is responsible for filling the memory with the script and telling the slave processor to run. Jun 18 17:03:41 The usefulness comes from several areas... Jun 18 17:04:10 you get the ability to send commands directly to the interpreter and see the output on the pins, making initial wire-up debug simple. Jun 18 17:04:30 you further get the ability to "message" to the interpreter in a fairly simple and generic way. Jun 18 17:04:45 I'm taking my leave everyone. Keep in touch! Jun 18 17:05:24 The 'SET' command, for example, can update digital/analog outputs, but can also update variables in the interpreter's memory. Jun 18 17:05:57 So, say your script is producing a pulse-width signal to a servo motor.... Jun 18 17:06:16 a variable could be used to store a timer value. Jun 18 17:07:14 during the running of the script, a 'SET' command could come in and change the variable value, adjusting the duty-cycle of the pulse-width signal and therefore the position of the servo motor. Jun 18 17:07:20 karki: does that help at all? Jun 18 17:07:28 :) Jun 18 17:07:39 but that gets me to my next doubt Jun 18 17:07:50 karki: that is all why I wrote the pseudo code I did in the beginning.... Jun 18 17:08:12 such that the command buffer is checked within the inner loop of the PRU code. Jun 18 17:08:50 karki: and that doubt is? Jun 18 17:09:17 jkridner : once again, the client shouldn't get bothered that the code is getting executed on the 'PRU' per say right? (well, this isn't my doubt, but anyway...) Jun 18 17:09:38 all he knows is that he is streaming botspeak code Jun 18 17:09:46 vs in the driver or in some other service that is listening to a socket? Jun 18 17:09:48 to an endpoint on the BB Jun 18 17:10:13 karki: each command sent should technically have a return to userspace Jun 18 17:10:24 according to the botspeak definition Jun 18 17:10:54 jkridner : as the user who writes the botspeak code Jun 18 17:11:10 alexanderhiam, karki: it doesn't matter where it gets interpreted as the response can be asynchronous over the socket. Jun 18 17:11:30 I don't see any reason not to send it all the way down to userspace though Jun 18 17:11:46 I'll just send it to the socket, and assume that that will do my job and return me the value Jun 18 17:11:46 alexanderhiam: "down to userspace"? Jun 18 17:12:32 i.e. the return result of each command should be piped back through that socket Jun 18 17:13:04 the PRU needs to always be able to accept new individual commands as well as be running a 'script'. The 'script' could be code that has been compiled into native code or there could be a real interpreter on the PRU, either solution is valid. Jun 18 17:13:21 jkridner: so if I have a userspace function that writes a command to the socket, it should then read and return the response from the interpreter Jun 18 17:13:22 panto are you here Jun 18 17:13:36 alexanderhiam: for sure, every immediate command should send back a response over the socket. Jun 18 17:14:28 just pointing it out since it's another difference to the current design Jun 18 17:15:20 hmm.. Jun 18 17:15:32 karki: I think that's what you meant by "all he knows is he's streaming botspeak code" Jun 18 17:15:43 mranostay: no, the PRU is polling for downcalls Jun 18 17:15:55 PRU1 is halted Jun 18 17:16:00 karki: he should also know the response from each command Jun 18 17:16:07 or she ;) Jun 18 17:16:22 yes Jun 18 17:16:30 yes, the socket will return a value to him/her/it ;) Jun 18 17:16:48 karki: is the socket a kernel driver? Jun 18 17:16:55 panto I wanted all of us to be on the same page Jun 18 17:17:25 mranostay: Sec, I'm giving you a link to the latest PRU firmware Jun 18 17:17:28 jkridner : nope, a userspace interface, a notmal network socket Jun 18 17:18:11 karki: is the PRU able to accept new commands and provide a response while it is running a script? Jun 18 17:18:44 jkridner : not as of now, though allowances have been made for that Jun 18 17:19:05 karki: k, you see what it is required, right? Jun 18 17:20:09 yes Jun 18 17:20:18 karki: when will you implement it? Jun 18 17:20:57 jkridner : there are a *lot* of features ! It's hard to decide which ones to do first :( Jun 18 17:21:00 mranostay: All latest PRU-side firmware here: https://github.com/abhishek-kakkar/BeagleLogic/tree/prufw/beaglelogic-firmware Jun 18 17:21:23 karki: you should have a project plan you are executing and this should be included. Jun 18 17:22:01 karki: what will be sent from the driver to the PRU to tell it what to do? Jun 18 17:22:28 jkridner : what do you mean? Jun 18 17:22:45 jkridner : the same thing that is being done now Jun 18 17:22:49 karki: not sure how to be more clear. what are you sending? Jun 18 17:23:05 just downcalls Jun 18 17:23:06 karki: what is being done now? Jun 18 17:23:14 what is your definition of "downcalls"? Jun 18 17:24:17 jkridner : Instructions are loaded in shared memory, a downcall is issued. (like the ones in panto's code ) Jun 18 17:25:06 the PRU knows where the find the instruction and starts to execute them after the downcall Jun 18 17:25:15 is done Jun 18 17:25:46 karki: I'm not sure that model fits the logic of the botspeak interpreter... Jun 18 17:25:48 * jkridner looks at https://github.com/deepakkarki/pruspeak/blob/master/src/pru-firmware/4_firmware/pru_firmware_4.c Jun 18 17:26:10 so, it is an interpreter, but it is using its own bytecode. Jun 18 17:26:17 yes Jun 18 17:26:23 1:1 mapping Jun 18 17:26:28 of instructions Jun 18 17:26:32 * jkridner wonders why to not at least use tinyspeak or something else already defined as the bytecode. Jun 18 17:26:41 it's internal Jun 18 17:26:48 so it doesn't matter Jun 18 17:26:50 * jkridner is fine with having less logic on the PRU for string parsing. Jun 18 17:27:02 no string parsing in the PRU Jun 18 17:27:10 you'll run out of code space before you know it Jun 18 17:27:11 panto: +1 Jun 18 17:27:22 the PRU is best for IOops Jun 18 17:27:32 tinyspeak is a bytecode already defined for botspeak. Jun 18 17:28:18 karki: ah, I guess the driver could handle the SCRIPT/ENDSCRIPT commands, in which case it wouldn't signal the PRU until it got the ENDSCRIPT command Jun 18 17:28:30 tinyspeak is defined at http://botspeak.org/?page_id=158 Jun 18 17:28:41 jkridner : as an alternative, here I have handled the arguments and pack it in integer width Jun 18 17:29:35 also, remember that implementations can be looked at to help figure out how to structure your code. Jun 18 17:30:35 specifically https://github.com/tuftsceeo/botspeak/blob/master/Devices/Arduino/VM/BotSpeak_Arduino_URAPI/BotSpeak_Arduino_URAPI.ino Jun 18 17:30:40 jkridner : tiny speak is an alternative, its weaker, I just defined a clear 1:1 mapping. tinyspeak needlessly will be limiting me. I did have a look at it Jun 18 17:31:06 karki: not sure why your mapping is better than theirs. Jun 18 17:31:21 cause theirs is not a mapping Jun 18 17:31:31 its a smaller language Jun 18 17:31:40 for 8/16 bits Jun 18 17:31:46 I have 32 bits Jun 18 17:31:53 so why limit my self Jun 18 17:32:44 + it is internal, to the world, it is still botspeak :) Jun 18 17:33:18 karki: you mean they have an 8-bit mapping and you want to do a 32-bit mapping? Jun 18 17:33:46 their 8-bit mapping works nicely over serial and is "almost" human parsable. Jun 18 17:34:02 the way I've done it is different, placing the arguments etc Jun 18 17:34:12 8 bits for the command is correct Jun 18 17:34:37 but each inst + args take 32 bits *always* Jun 18 17:34:47 karki: can you write out an equivalent specification as http://botspeak.org/?page_id=158 so that we know what you are interpreting? Jun 18 17:34:52 which simplifies things a lot Jun 18 17:35:02 yes, that I'll do. Jun 18 17:35:23 it seems with their definition, some command+args can take more than 32 bits. Jun 18 17:35:36 it will be interesting to see if/how you avoid that. Jun 18 17:36:46 my preference on order of implementation is to complete the DIO instruction, then move to all of the immediate interpretation issues before moving on to the other instructions. Jun 18 17:37:05 panto, alexanderhiam: comments on the order of implementation? Jun 18 17:37:22 karki: of course, I'd want to see the spec written up first. Jun 18 17:37:31 wow, we are SOOOOOO far over time. :( Jun 18 17:38:11 * jkridner is surprised no one suggested moving this conversation off-line before they decided to drop. Jun 18 17:38:17 jkridner, karki, panto: I say focus on getting the command streaming working now, just using what commands are currently implemented Jun 18 17:38:29 but I wonder why we chose the PRU as interpreter platform over the BBB itself, not that anyone is trying to do HiFi things with BotSpeak........ Jun 18 17:38:41 guess we are just lucky this was the last project to send in a status report. :-) Jun 18 17:39:20 ;) Jun 18 17:40:20 alexanderhiam, I agree Jun 18 17:40:26 get something going first Jun 18 17:41:00 karki: once that's working you can start adding the script logic, then you'll know for sure if the current model works Jun 18 17:41:48 there's some script logic already there, assuming the script is located at a specific place in memory. Jun 18 17:42:08 alexanderhiam : command streaming? through the socket? Jun 18 17:42:13 the immediate execution does not seem to be there. guess some downcall needs to be implemented for that. Jun 18 17:42:24 karki: yeah Jun 18 17:43:08 jkridner : what happened to the board at TI b'lore? Jun 18 17:43:38 karki: thought you'd know or would have bugged me sooner in the meeting. Jun 18 17:43:38 jkridner: well with the current model the immediate command is just a one line script without a GOTO 0 Jun 18 17:44:14 karki: team in Bangalore was being unresponsive. Jun 18 17:44:28 :( Jun 18 17:44:36 alexanderhiam: well, that can't work if you already have a running script.... Jun 18 17:44:49 adding a new downcall to do immediate interpretation, however, seems like it would work easily. Jun 18 17:45:14 jkridner: true, I guess there'd just need to be another dedicated memory location Jun 18 17:45:33 I was actually more worried about implementing PWM this week....... Jun 18 17:45:34 downcall events are polled after every command execution.... this looks like the pseudo-code I had originally. Jun 18 17:45:51 alexanderhiam: right. Jun 18 17:45:52 space for a one-liner and a flag that the interpreter can look at between each script command Jun 18 17:46:12 karki, panto: so, we are using shared memory for all of this rather than a circular buffer? Jun 18 17:48:14 karki: I did some thinking before about how to handle timer events by putting instruction execution on a timer event. if you can't swallow that idea, then at least 'wait' need to generate some blocking flag against a timer for the event loop rather than doing a spin. Jun 18 17:48:14 shared memory is easier for that kind of application Jun 18 17:48:26 panto: k. Jun 18 17:48:27 you're not 'streaming' something Jun 18 17:48:47 you're issuing a command (and either arrange to check for it's completion, or wait right on the spot) Jun 18 17:49:01 panto: yeah, not per this implementation. the references were streaming over a serial port, but I think we are moving all of that streaming logic into userspace. Jun 18 17:49:11 right Jun 18 17:50:48 * karki has to scram Jun 18 17:51:56 * jkridner slams gavel. Jun 18 17:52:00 run forest run Jun 18 17:52:09 thank you all very much for time today. Jun 18 17:53:18 jkridner: this was a long long meeting Jun 18 17:55:05 jkridner: Based on my video you saw today, do you have any suggestions or any aspect you would like me to consider as I move forward? Jun 18 17:55:59 Abhishek_: link, bitte :) Jun 18 17:57:04 vvu: http://www.youtube.com/watch?v=UYRSvTC6VUI Jun 18 18:00:17 kernel drivers here, kernel drivers there...kernel drivers everywhere Jun 18 18:00:36 yep, 5/7 GSoC projects this time on kernel drivers Jun 18 18:00:46 * vvu really thanks OS Lab course from university :) Jun 18 18:00:49 utilize kernel drivers in some or the other manner Jun 18 18:01:12 I went from zero to kernel in two weeks :) Jun 18 18:01:33 no worries Abhishek_ i was the same Jun 18 18:02:21 Abhishek_: can't wait for the sigrok bindings. not sure what/how you were doing the visualzations. Jun 18 18:02:47 jkridner: I actually ran sigrok-cli directly on the binary data Jun 18 18:03:02 gathered straight from /dev/beaglelogic Jun 18 18:03:33 and then I copied that file to my PC and ran it in PulseView, the frontend for sigrok Jun 18 18:04:19 Abhishek_: now at internship they gave me a super heavy oscilloscope...why don't u have the beaglelogic ready so i can cheat a bit ? Jun 18 18:05:48 vvu: The code is at least in workable state now if you want to compile it and run Jun 18 18:06:55 do u have README ? Jun 18 18:07:25 yep, I should build one. But usually make is all you would need Jun 18 18:07:48 make the kernel modules, the PRU firmware, the device tree and install them Jun 18 18:08:04 which pins are used ? Jun 18 18:08:55 vvu: http://www.theembeddedkitchen.net/wp/wp-content/uploads/2014/05/BeagleLogic-FinalSlides.pdf Jun 18 18:09:21 don't promise a test this week but next one for sure Jun 18 18:09:54 slide 7 in there Jun 18 18:10:39 oh Abhishek_ opening a pdf with only images kills my slow chromebook Jun 18 18:10:52 oops Jun 18 18:10:56 * vvu likes ASCII since chromebook :) Jun 18 18:11:24 it kills my 2048x1536 android tab as well :P Jun 18 18:11:50 too bad the processor isn't match for the soo goood display Jun 18 18:12:33 hack it and put a bigger one Jun 18 18:12:41 just for the sake of hacking Jun 18 18:14:23 there've been a lot of kickstarters of people hacking iPad retina screens as a secondary PC monitor Jun 18 18:14:58 Wonder if the usb framebuffer driver could help me do that in a non-invasive and clean manner Jun 18 18:15:21 that thingie is kinda old Jun 18 18:15:52 i'm really curious how we are gonna interface it nicely Jun 18 18:16:28 bulk transfers aren't enough? Jun 18 18:16:30 if we manage maybe we can have some hexdumping on the android side of the frames Jun 18 18:16:43 is enough but also you need to do the logic on the android part Jun 18 18:16:46 and that may be killer Jun 18 18:17:11 and we all know how java likes to eat memory a lot Jun 18 18:19:35 indeed, android runs so slow on these tabs, they could do lot better with just linux, isn't it? Jun 18 18:19:55 they did a lot of hacks the guys from google to optimize it Jun 18 18:20:15 now the problem is that samsung and all the vendors just pur all their fancy UIs Jun 18 18:20:30 for my on my nexus 7 old model everything is great, keep it at minimum Jun 18 18:21:33 I flashed my mum's samsung N7000 with CyanogenMod, works quite snappily as compared to Samsung TouchWiz bloat Jun 18 18:21:52 told you Jun 18 18:22:04 no fancy stuff...no me gusta for me at least Jun 18 18:23:13 I have a MTK6589 vanilla android 4.2 phone, beat the original N7000 by a huge margin in terms of UX Jun 18 18:23:34 and lasts 5-6 days on a single charge with no cellular, just WiFi Jun 18 18:24:03 ohooo Jun 18 18:24:07 that is majestic Jun 18 18:24:20 truly exquisite i can say, i live one day with my n5 Jun 18 18:25:55 (although I have to say that it is standby most of the time listening to IRC messages) Jun 18 18:27:24 k, time to leave now. Good night everyone Jun 18 18:28:29 gn Abhishek_ ! Jun 19 00:07:53 jkridner: av500: Sorry. sorry. I missed the meeting. Jun 19 00:08:08 I was suffering with high fever. 102F :/ Jun 19 00:08:20 I sent the status mail. **** ENDING LOGGING AT Thu Jun 19 02:59:59 2014