**** BEGIN LOGGING AT Thu Jun 09 02:59:58 2016 Jun 09 07:05:37 Hey Jun 09 07:05:52 Wormo: there ? Jun 09 07:08:02 Okay .. Do not ask to ask, just ask. Jun 09 07:10:02 So what will happen if two drivers register for the same device ? Jun 09 07:12:11 this rpmsg_pru driver is the key i guess Jun 09 07:12:13 https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.4.11-ti-r29/drivers/rpmsg/rpmsg_pru.c#L292 Jun 09 07:13:12 and it registers for the pru rpmsg device, that i believe is a virtio device and it then gets probed. Jun 09 07:32:44 ooops .. Jun 09 07:32:56 jic23: ^^ Jun 09 07:33:17 Morning - the beep just made me jump as I forgot I left this open. Jun 09 07:33:43 not caught up with your report yet this morning. Jun 09 07:33:56 Good morning :) Jun 09 07:34:16 (or woken up yet either ;) Jun 09 07:34:19 Coffee in hand. Jun 09 07:35:35 can you please help with the query ? :) Jun 09 07:35:46 jic23: ^ Jun 09 07:37:03 hmm. Not sure without digging. Jun 09 07:37:22 A given device itself can only have one driver, but that's not say there aren't 'client' drivers hanging off that. Jun 09 07:38:06 the client drivers use APIs right ? declared by the device drivers ? Jun 09 07:38:22 My limited understanding matches yours. The pru front end provides services to the rproc (subsystem) which in turn provides some services on to the virtio stuff in rpmesg Jun 09 07:38:28 Yeah, indeed on the API front. Jun 09 07:38:51 hang on call coming in. Jun 09 07:40:16 So, to communicate with the prus we need the handle to the rpmsg device, which gets passed to the rpmsg_pru driver, as in its match_table and then the driver takes hold of it. It further does not define any APIs. Jun 09 07:40:42 all this with reference to this driver : https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.4.11-ti-r29/drivers/rpmsg/rpmsg_pru.c Jun 09 07:47:40 It just adds device files : https://github.com/RobertCNelson/linux-stable-rcn-ee/blob/4.4.11-ti-r29/drivers/rpmsg/rpmsg_pru.c#L233 Jun 09 07:47:50 and no APIs Jun 09 07:51:15 * ZeekHuge has woke up just half an hour ago, and is on laptop since then. Will be back in a while .. Jun 09 07:51:34 okay .. i have mobile too .. Jun 09 08:01:09 Time to add some then ZeekHuge :) Jun 09 08:01:16 APIs that is. Jun 09 08:03:06 May well be worth pulling TI into the loop on this. Go for all the recent patch authors from their tree. Jun 09 08:03:13 http://git.ti.com/gitweb/?p=rpmsg/remoteproc.git;a=summary Jun 09 08:03:36 (in particular beyond Suman Anna, make sure you add Andrew F. Davis. Jun 09 08:04:06 He's a regular IIO contributor - as he put it in an email a few weeks back. If it's odd it tends to land on his desk ;) Jun 09 08:04:47 We are talking in kernel API, so it is not like they are set in stone. We can and probably will change them lots in the future. Jun 09 08:06:06 Also compare that tree 4.4 branch with what rcn is using. Might be a lag and there are at least some commits in there 'hinting' that they are heading towards in kernel interfaces. Jun 09 08:08:37 Okay so I will be contributing to rcn's tree. Right ? And not TI' s tree. ? Jun 09 08:08:49 Is that open to contribute ? Jun 09 08:09:26 That = Ti's tree Jun 09 08:09:56 jic23: ^ Jun 09 08:10:41 Not sure on the TI tree, but they do ultimately take patches from rcn's tree at times I believe. Jun 09 08:10:57 Anyhow, it's all GPL, nothing stops TI from taking your patches ;) Jun 09 08:11:10 Okay. Jun 09 08:11:16 rcn is just lifting their tree for this stuff I think... Jun 09 08:12:18 Also of interest is that there is at least one pru using driver navigating round all the restrictions of rpmesg. Jun 09 08:12:43 Not sure it's a good idea to copy it, but might give some interesting reading. It's the pru_ethernet driver. Jun 09 08:15:09 deosnt' support this particular processor, but the rpu side of things is the same. Jun 09 08:15:23 looks likely it directly maps the pru memories. Jun 09 08:16:29 Also, interestingly looks like they are not making much effort to distribute the source for it. Comes with the relevant dev kits.... Jun 09 08:17:47 Anyhow, frame an email to suman anna and Andrew Davis, then perhaps send + mentors a draft to glance over. Need these guys help ideally so worth a bit of care on the first email about this particular issue. Jun 09 08:18:03 send me + mentors that is. Jun 09 08:18:25 Or if it looks like more the TI way, post on relevant forums. Jun 09 08:19:58 Actually we are talking more about a client interface for rpmesg than one for rproc. Don't have to care about TI that much ;) Jun 09 08:20:05 Beyond keeping them in the loop. Jun 09 08:22:09 The maintainer looks to be a freelancer - pull requests go from him to Linus rather than through TI :) Jun 09 08:24:27 https://www.kernel.org/doc/Documentation/rpmsg.txt includes a client driver example. Jun 09 08:29:11 so this is an amusing dance (i.e. total mess). Jun 09 08:30:04 Anyhow, the simple answer is you replace that rpmsg_pru.c driver with whatever you like. Jun 09 08:32:19 magic is in the name the pru kicks back. Right now it is "rpmsg-pru". Jun 09 08:32:40 Chase where that comes from. It's supposed to be from the firmware, but relevant code may not yet exist... Jun 09 08:32:56 Anyhow, back later, should drop in on the day job :) Jun 09 08:33:25 ZeekHuge, very handy to work this stuff out and draw some nice diagrams for a blog post etc... Jun 09 08:35:48 Sorry jic23 , was having lunch .. Jun 09 08:35:58 Yep, will try to figure it out Jun 09 08:36:02 not to worry! have to eat sometime ;) Jun 09 08:36:19 See if the magic rpmsg-pru name is coming from somewhere in the firmware, because it should be ;) Jun 09 08:36:27 Might be in a linker script or similar. Jun 09 08:36:45 Anyhow, that's the magic you need to make it use your driver rather than the the ti userspace one. Jun 09 08:36:56 Actually a fairly elegant solution ;) Jun 09 08:38:37 yep it is from the fw https://github.com/ZeekHuge/BeagleScope/blob/master/examples/pru_pin_state_reader/PRU_PIN_STATE_READER/main.c#L69 Jun 09 08:38:53 Yeah :) Jun 09 08:38:54 cool, that was quick ;) Jun 09 08:39:07 Anyhow, next steps. Jun 09 08:39:23 Take a copy of the userspace rpmsg driver from ti, change that name and change the firmware name then make sure it still talks. Jun 09 08:39:37 If it does, you have your own in kernel driver to hook up to other stuff. Jun 09 08:39:43 :) Jun 09 08:39:53 Yes ! its getting interesting ! Jun 09 08:39:57 :) Jun 09 08:40:20 + usefully I'm not sure that driver will have any ties to the underlying TI stuff so could be upstreamed separately as rpmesg-parallel or something like that. Jun 09 08:41:30 I'd really be tempted whilst waiting for your incoming ADC board to see if you can bring beaglelogic up to date with current standard interfaces and see how quick it can run. Jun 09 08:41:41 Very useful side output of the project. Jun 09 08:42:02 First instance, do the same chrdev as abishek did then we can look at upstreaming path for that. Jun 09 08:42:21 and kick TI into upstreaming the rproc stuff at the same time ;) Jun 09 08:42:36 Anyhow, busy day for you I think ;) Jun 09 08:42:51 really back later, or I'll get nothing done today on the whole getting paid thing :) Jun 09 08:43:23 This is way more interesting than looking at noise sources on some x-ray detectors - waiting for new toys (wafer scale cameras) that are stuck in customs. Jun 09 08:43:43 Those are fun, but testing them is really dull.. Jun 09 12:09:14 Hi Kiran4399, how's it going? Jun 09 12:09:43 good.. alexhiam suggested me to take a break from sensor kernel testing and start making pru apis.. Jun 09 12:09:53 alexhiam: meanwhile you know what I just realized that.. Jun 09 12:10:05 oops.. sorry. Jun 09 12:10:33 jic23: ... the sensors are 9250 and bmp280 in the robotics cape rev D Jun 09 12:10:58 jic23: I was testing with 6050 and bmp180 Jun 09 12:11:01 hmm. unlikely to cause the issues you were seeing I think... Jun 09 12:11:32 that error -121 was particularly odd as that's one obscure code. Jun 09 12:13:39 jic23: so for the 6500 and 280.. what changes do you want me to do? Jun 09 12:14:04 should just be device tree updates to match. 280 is certainly supported out the box. Jun 09 12:16:17 hmm. actually those two parts are fairly different - still would have expected you to just get garbage output as there is no sanity check on a chip id or anything in th e driver. Jun 09 12:17:13 6500 should be out of the box support as well by the look of it... Jun 09 12:17:38 though 9150 might be closer.. not sure. Jun 09 12:19:00 so.. curently the mpu kernel driver suppots 6050, 6500 and 9150.. Jun 09 12:19:13 I think I should test with the 6500.. Jun 09 12:19:31 yeah - though the limitation is mostly a question of what developers had to hand to test. Once you've tested it we'll add the 9250 as well (you can do the patch ;) Jun 09 12:20:51 jic23: did you try?? yesterday you told me that you had some breakouts. Jun 09 12:21:55 yeah, but not wired up to a beagle bone unfortunately. Will be the weekend at least before I can bodge one together. mw has a cape for the 96 boards with one on, but not sure he ever ran on a beaglebone either. Jun 09 12:22:16 mw? Jun 09 12:22:28 Michael Welling (mentor) Jun 09 12:22:30 oh.. you mean m_w Jun 09 12:22:45 oh yeah...m_w Jun 09 12:22:56 everyone should have nice distinctive IDs like us two ;) Jun 09 12:23:04 yeah.. Jun 09 12:23:08 What is your name? Jun 09 12:23:20 and which timezone do you generally operate? Jun 09 12:25:52 Jonathan Cameron, UK so GMT + 1 Jun 09 12:26:21 jic23 was my university ID a while back and as it's never taken anywhere I use it for most things :) Jun 09 12:26:33 haha.. same with me.. Jun 09 12:26:39 jic23: When do you think the latest patchset for mpu will come? Jun 09 12:26:42 and who is making it? Jun 09 12:27:26 Not sure, Lenoard (Crestez Dan Leonard) Jun 09 12:27:47 Depends when he gets back to it. At any given time he tends to have half a dozen sets at various stages of review. Jun 09 12:28:17 The one I posted a tree for the other day will be pretty close to final if you want to try that. Jun 09 12:28:34 Though not worth it until the hardware is talking to you. Jun 09 12:29:00 Need to slap someone with a wet fish. Lot of separate handling for different IMUs in that driver, but in reality all the config stuff is the same for all current parts.... Jun 09 12:29:18 They should have split it only when the configs differed, not on the basis they would some day do so. Jun 09 12:29:23 Just makes it harder to read. Jun 09 12:29:37 So upshot is that any supported part will be handled the same way. Jun 09 12:29:41 jic23: did you have a look at the android driver? Jun 09 12:29:58 for mpu devices.. Jun 09 12:30:40 Nope - have enough trouble keeping up with the drivers I get directly so rarely go looking further. Android drivers tend to be particularly hacky as well. No quality control whatsoever! Jun 09 12:31:02 Though often they aren't a bad starting point to know how a device works. Jun 09 12:32:01 jic23: one quick question Jun 09 12:32:13 Hi ZeekHuge. Jun 09 12:32:25 How did you know that the string "rpmsg-pru" is being sent by fw ? Jun 09 12:32:36 I was searching for it since tomorrow Jun 09 12:32:41 but in rprco drivers Jun 09 12:32:55 jic23: I mean this https://android.googlesource.com/kernel/msm/+/android-msm-hammerhead-3.4-kk-fr2/drivers/staging/iio/imu/inv_mpu/README Jun 09 12:34:10 according to the remoteproc.txt, this module is responsible to create virtio devices, so i thought it is somehow probing the rpmsg-pru Jun 09 12:38:29 ZeekHuge, err, can't remember - was at least a few hours ago ;) Jun 09 12:39:54 ahh , okay. Jun 09 12:41:03 Found a reference to it somewhere - right now I can't even find the userspace interface driver... Jun 09 12:41:12 I think you posted a link to that. Where was it? Jun 09 12:42:41 Kiran4399 - yeah on the android driver - we did get some stuff heading that way but threw it back at them... Jun 09 12:43:00 The reason being it needs a custom driver part for every device hanging off the back of the imu. Jun 09 12:43:14 So ultimately an extra copy of every driver than anyone ever uses like that. Jun 09 12:43:42 No way we were going to take that level of duplication - the discussions on that ultimately lead to what Leonard is doing now. Jun 09 12:44:15 In between we had an i2c bypass mode which is actually what Invensense were recommending at the time. Jun 09 12:49:51 jic23: yeah, actually it clearly written http://git.ti.com/pru-software-support-package/pru-software-support-package/blobs/master/examples/am335x/PRU_RPMsg_Echo_Interrupt0/main.c#line62 Jun 09 12:50:04 But I somehow missed that .. Jun 09 12:50:16 Thanks to you then :) Jun 09 12:51:15 But I think it still the remotperoc driver that creates the channel with this name. Jun 09 12:53:08 jic23: there? Jun 09 12:55:15 kiran4399 yup. Jun 09 12:55:35 jic23: which was the patch you were refereing? Jun 09 12:56:05 ZeekHuge - the drive reads from the firmware an indeed creates the channel, but if you change it in the firmware it should result in a different driver loading. Jun 09 12:56:28 kiran4399, lost - where were we? Jun 09 12:57:14 jic23: the patchset which is near to the final version.. Jun 09 12:57:17 for mpu devices Jun 09 12:58:38 https://git.kernel.org/cgit/linux/kernel/git/jic23/iio.git/log/?h=experimental-beaglefun Jun 09 12:58:54 Though I think I accidentally did that on top of the rt tree rather than the normal version... Jun 09 12:58:59 (of rcn's tree). Jun 09 12:59:26 Also scrap the top patch before using it as that'll probably prevent the beaglebone booting - was a bodge to allow me to do a test x86 build. Jun 09 12:59:37 jic23: want me to go for it?? BTW what are the issues with bbb 4.4 kernel Jun 09 12:59:38 ? Jun 09 12:59:59 jic23: Yep, I completely got that, I am just trying to make a call-structure in my head of all this, with rpmsg and fw and then remoteproc Jun 09 13:00:41 Kiran4399 - it should 'work' without those patches - get that going first. only apply the last few bits in that tree once you have the accel and gyro working. They should allow you to bring up the magnetometer as well. Jun 09 13:00:59 The'll hit mainline sometime this cycle most likely and get picked up by rcn from there in around 5 months... Jun 09 13:01:03 They'll. Jun 09 13:01:26 ZeekHuge - sure. Chasing code fun and games ;) Jun 09 13:02:33 Really want to see a blog post when you have it figured. Might even (ultimately) make a nice more formal article at a later date. Jun 09 13:03:07 This sort of thing is wide abstract interest in the community as well as for people who want to use the PRUs. Jun 09 13:04:57 alexhiam: Hi! Jun 09 13:05:07 *laughs* You missed him ;) Jun 09 13:05:30 alexhiam: hi! Jun 09 13:05:38 jic23: I guess not this time. Jun 09 13:05:42 :-P Jun 09 13:06:04 ahoy Jun 09 13:11:08 kiran4399: hmm, maybe not just the 6500 vs 6050 issue, looks like a wrong whoami value is just a warning: https://github.com/beagleboard/linux/blob/4.4/drivers/iio/imu/inv_mpu6050/inv_mpu_core.c#L776 Jun 09 13:13:30 alexhiam: I think I'll test the kernel with mpu-6050 today.. will update you in about 3 hours.. Jun 09 13:13:51 alexhiam: which breakout did you order? Jun 09 13:14:22 http://www.ebay.com/itm/272210139103?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT Jun 09 13:14:39 oh.. Jun 09 13:14:56 alexhiam: how many days does it take for ebay to deliver a product to you house? Jun 09 13:15:34 kiran4399: depends on the seller. They're in New York, which is pretty close Jun 09 13:15:59 looks like Monday Jun 09 15:21:52 how goes it? Jun 09 16:20:50 alexhiam: Piece of cake!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Jun 09 16:21:04 alexhiam: linux rocks! Jun 09 16:21:25 alexhiam: after 1 week of crying!! Jun 09 16:21:52 alexhiam: got the mpu6050 kernel driver detected! Jun 09 16:22:53 woo hoo! Jun 09 16:23:02 http://pastebin.com/ymBRebb2 Jun 09 16:24:26 so what was the magic? Jun 09 16:24:36 No idea.. :-P Jun 09 16:24:47 oh dear... Jun 09 16:24:48 simple instantiation.. I did not even use the overlay.. Jun 09 16:25:11 interesting... Which overlay? Jun 09 16:25:26 jic23: how can I get the data btw? Jun 09 16:25:49 jic23: I created overlay for mpu 6050, 6500 and bmp280.. I am saying I did not even use that.. Jun 09 16:26:09 hmm. Wonder if something odd was causing pin muxing to stop that bus working... Jun 09 16:26:14 driver running now then? Jun 09 16:26:30 how can I get the data? Jun 09 16:26:44 if so cat /sys/bus/iio/devices/iio\:device0/in_accel_x_raw Jun 09 16:26:55 or something like that... Jun 09 16:27:32 to do more fun things you'll need to compile either the generic_buffer tool in the kernel source tree, or mess around with a standard library like analog devices' libiio Jun 09 16:27:55 (gives you access to data at high rates - but it's in binary rather than pretty printed) Jun 09 16:36:11 or write up something to parse it yourself :D Jun 09 16:37:16 Hurray!!! working even with the overlay!!!! Yipeee!!! Jun 09 16:37:26 jic23, m_w, alexhiam: ^^ Jun 09 16:37:38 good Jun 09 16:37:46 first contribution to beagleboard.org !! :-P Jun 09 16:39:31 http://pastebin.com/2srK3uR4 Jun 09 16:41:39 kiran4399: is the iio registration showing up? Jun 09 16:42:03 where? Jun 09 16:42:04 m_w: Jun 09 16:42:13 /sys/buss/iio Jun 09 16:42:31 can you forward the source for the overlay? Jun 09 16:43:26 m_w: Yeah.. it is showing.. I can even cat the readings!! Oh my!! :-D Jun 09 16:43:38 excellent Jun 09 16:43:42 yeah.. Jun 09 16:44:45 I think the mpu9250 on the cape is dead.. now should I use the same cape? Jun 09 16:44:50 m_w: ds2: Jun 09 16:45:31 lemme check the schematic Jun 09 16:45:40 you have a link? Jun 09 16:45:41 http://pastebin.com/S739VGEm Jun 09 16:45:55 kinran4399 - I guess this is with the breakout then! (missed that detail) Jun 09 16:46:02 kiran4399^^ Jun 09 16:46:19 m_w: link for what? Jun 09 16:47:05 nevermind I found it Jun 09 16:47:08 https://github.com/StrawsonDesign/Robotics_Cape_Documentation/blob/master/SD-101D%20Robotics%20Cape%20Schematic.pdf Jun 09 16:49:27 nothing crazy there Jun 09 16:57:05 Beyond cramming it on one page ;) Jun 09 16:57:38 actually I'm being silly - I'm impressed they laid it out like that. Jun 09 16:58:07 (been looking at some old school clpd schematic diags today...) Jun 09 16:58:17 nice Jun 09 16:58:22 no f**ing clue why these don't work... Jun 09 16:58:28 :D Jun 09 16:58:35 Anyhow hour late heading home. Bye all. Jun 09 16:58:45 bye! Jun 09 16:59:32 hmmm what's the Z value? Jun 09 16:59:51 z value? Jun 09 17:00:09 cat in_accel_z_raw Jun 09 17:00:43 7364 Jun 09 17:00:58 ah... board must not be level Jun 09 17:01:47 ds2: thats right.. some idiot from my lab took the soldering iron.. so I had to put something between the board and the pins to keep them connected and intact.. Jun 09 17:03:23 what'sthe Y value? Jun 09 17:03:29 trying to sanitize your data Jun 09 17:03:58 -240 Jun 09 17:04:05 :-/ Jun 09 17:04:07 2 exams and 1 project in one day but now I can start working Jun 09 17:04:17 that seems a little off Jun 09 17:04:26 most be an a pretty bad offset Jun 09 17:04:52 either that or you must be very high in elevation :D Jun 09 17:05:26 gravity vector is only about 8.4m/s^2 Jun 09 17:05:39 or prehaps you are rising in a balloon :D Jun 09 17:06:25 ds2: I really am on earth!! Jun 09 17:25:51 * Abhishek_ checks in Jun 09 17:30:54 did we square away the ADC board for ZeekHuge? Jun 09 17:31:10 I did not get any replies to my email Jun 09 17:31:11 yes ! Jun 09 17:31:25 jkridner: ordered it ! Jun 09 17:31:43 okay that would have been good to tell me :-/ Jun 09 17:32:55 ahh .. sorry m_w . I .. missed that. Jun 09 17:33:14 well no big deal Jun 09 17:33:52 Right now, a single unit will go to ZeekHuge as nothing else was specified in the spreadsheet. Jun 09 17:33:52 I guess 2, since ds2 has it in his line items. Jun 09 17:34:02 this is what he said ^^ Jun 09 17:34:20 okay Jun 09 17:34:45 I guess I need to stay on irc more :) Jun 09 17:34:55 ZeekHuge: how do you plan to generate the clock for the ADC (not supply it)? Jun 09 17:36:26 Abhishek_ using NPN in common emitter mode. Jun 09 17:36:36 and thus buffer it Jun 09 17:37:12 so that 50ohm termination is what the thing to care about. Jun 09 17:37:22 and therefore doing this buffering. Jun 09 17:38:48 btw, I have made quite reasonable sense of rpmsg by now, and just trying to put it all on paper, and than make a module. If I am able to stay awake the night today, it will probably be done by morning Jun 09 17:39:18 why all of the sleep deprivation? Jun 09 17:39:26 ZeekHuge: what's your github/blog? Jun 09 17:40:03 ds2: blog under development . github : https://github.com/ZeekHuge/BeagleScope Jun 09 17:40:19 It will be done by morning . I blv Jun 09 17:40:54 m_w: need to get the project back in hands, I blv we are lagging behind a lot. Jun 09 17:42:21 well get some coffee or tea :) Jun 09 17:43:12 yeah . That is must, I have them :) Jun 09 17:43:37 already 5 cups today. Jun 09 17:43:47 back to work. Jun 09 17:44:47 kiran4399: awesome! Jun 09 17:45:29 idk who's assembling the robotics cape. That could be a production prototype, quite possible there's a bridge on the imu pins Jun 09 17:48:02 they should be testing them Jun 09 17:48:24 they shouldn't bridge that easily Jun 09 17:48:40 I have done 0.4mm (MPU6500) QFNs by hand and haven't had bridging issues Jun 09 17:48:44 opens are far more likely on QFN Jun 09 17:49:01 good point Jun 09 17:49:09 no issues with opens either Jun 09 17:49:14 due to too much solder on the bottom pad usually Jun 09 17:49:19 QFNs are pretty easy Jun 09 17:49:23 I just mean that could be an untested prototype Jun 09 17:49:29 IIRC - the MPU's do not recommend using the bottom pad Jun 09 17:49:33 the cape's not for sale yet Jun 09 17:49:56 and if you do, the pad design determines if you will are screwed or not Jun 09 17:50:57 the paste mask can be problematic if machine assembled Jun 09 17:51:34 I used hotair rework to populated the MPU-9250 on the robomezzi Jun 09 17:51:45 to "populate" Jun 09 17:52:14 m_w: why did you choose the MPU-9x series? Jun 09 17:53:20 not sure Jun 09 17:53:33 vs MPU-6 + better external mag Jun 09 17:54:15 the magnetometer is not good in the MPU-9250? Jun 09 17:54:28 there are better ones Jun 09 17:54:35 I haven't really gotten to trying to access it yet Jun 09 17:54:43 I am using SPI Jun 09 17:54:46 the 9's are literally mag + 6 Jun 09 17:55:43 I like the integration but not at the cost of accuracy Jun 09 17:56:06 the robomezzi is up for a spin and I can change it to whatever Jun 09 17:56:29 that's assuming you have algs to take advantage of a better mag Jun 09 17:56:55 true Jun 09 17:57:29 the problem is robomezzi is a side project and I don't get enough time to work on the software side Jun 09 17:58:22 people want to make it into a product but no one is willing to fund it directly Jun 09 17:58:35 so it poke at it here and there Jun 09 17:59:07 update the kernel on the 96boards is a bit cumbersome Jun 09 17:59:22 updating Jun 09 18:05:12 M_w how about crowd funding robomezzi? Jun 09 18:05:55 Guess not big numbers to tidy up and productise it? Jun 09 18:06:54 Not that I have ever done anything similar :) Jun 09 18:07:43 Do a lot of similar stuff at work but pcb stuff is never the expensive bit. Jun 09 18:09:00 crowd funding is a lot of work Jun 09 18:09:46 evaluating productizing a mini-Bone myself Jun 09 18:10:34 Hey jic23 Jun 09 18:11:08 ZeekHuge: I know the emitter follower part you use to drive the clock Jun 09 18:11:33 The question is, where do you get the clock from to drive the emitter follower? Jun 09 18:12:06 PRU itself Jun 09 18:12:29 we can later try external clock ? Jun 09 18:12:34 Abhishek_: ^ Jun 09 18:12:38 Are you able to generate a clock signal from the PRUs? Jun 09 18:12:45 Have you tried to do so yet? Jun 09 18:13:33 clock on PRU ie toggling PRU pin ? Jun 09 18:13:50 Yup, at the frequency you want to. Jun 09 18:14:36 not with a variable frequency, just simple examples : https://github.com/ZeekHuge/BeagleScope/tree/master/examples Jun 09 18:14:48 Do you have a scope? Jun 09 18:15:08 or a time machine ;) Jun 09 18:15:10 Yep, it in my college lab and i can access that. Jun 09 18:15:23 ds2: LOL Jun 09 18:15:37 though that will have to remain there . Jun 09 18:15:47 if you got a time machine, speed foward and borrow a BeagleScope :D Jun 09 18:16:06 Haha ! Jun 09 18:17:31 ZeekHuge: I want you to make sure you can generate a clock at the correct frequency and voltage level before the ADC arrives Jun 09 18:17:46 ahh and, ds2 .. Abhishek_ has instructed me to go with rpmsg for now and will see later if we can make it better. So rpmsg for now. Jun 09 18:19:08 So build the emitter follower on a breadboard (or solder it, if that's your thing), connect it to the PRUs and then examine the output on a scope so that you can see the clock and that the voltage levels do not exceed 3V at any point Jun 09 18:20:21 Okay, so as soon as i have a lkm talking to pru, probably by tomorrow. I will be on it . I think 2nd from now . will that be okay .. Jun 09 18:20:30 ? Jun 09 18:20:47 Abhishek_: ^ Jun 09 18:21:08 Before you even plug in the ADC board to the BeagleBoard, I want you to make sure that the clocking circuit is working correctly Jun 09 18:21:40 Take your time until the board arrives and figure out what works best for you. Jun 09 18:21:55 okay, got it . will do it :) Jun 09 18:22:36 ZeekHuge: okay Jun 09 18:23:03 ZeekHuge: post the schematic if you can Jun 09 18:23:49 ds2: How do we try the DMA approach? When I was working on BeagleLogic the two PRU approach worked so well that I didn't bother looking at how DMA could help here. Jun 09 18:24:24 the one for data i/o is here : https://github.com/ZeekHuge/BeagleScope/blob/master/docs/BBB-ADC.pdf Jun 09 18:24:29 ds2: ^ Jun 09 18:24:31 Abhishek_: ^ Jun 09 18:26:16 Abhishek_: EDMA against the DRAM Jun 09 18:26:29 probally have to generate a EDMA event for every N'th byte Jun 09 18:26:38 ds2: Abhishek_ that looks okay ? Jun 09 18:27:11 ZeekHuge: You using PRU0 for the clock? Jun 09 18:27:23 ZeekHuge: what about the clock circuit? Jun 09 18:27:46 what's that signal going to P9? Jun 09 18:28:36 oh that ! Jun 09 18:28:49 it 16th bit .. for parallel capture Jun 09 18:29:20 it routed to P9 Jun 09 18:30:04 ds2: since it is yet to be tested ... I didnt add it. Jun 09 18:30:09 I see. Jun 09 18:30:22 I will put once I test it. Jun 09 18:31:20 I guess the schematic *looks* fine. Jun 09 18:46:40 jic23: I have had a few people try to convince me to crowdfund Jun 09 18:46:54 but I am not an actor so it would be bad Jun 09 18:51:29 M_w can't say I blame you. Jun 09 18:52:25 perhaps I can write up a script and use puppets :) Jun 09 18:52:54 or maybe use a robot as the narrator Jun 09 18:53:30 that would get a few backers just on the nerd factor Jun 09 18:55:45 m_w: Use fiverr Jun 09 18:56:07 whats that? Jun 09 18:56:27 Freelance workers available starting 4$ Jun 09 18:56:32 *5$ Jun 09 18:57:32 Abhishek: you done a crowdfund before? Jun 09 18:58:48 I feel it needs blood and sweat, and is a day job in itself. Jun 09 18:58:58 No, I haven't Jun 09 18:59:21 Plus, long wait times Jun 09 19:00:03 I rather not overextend myself, the primary contract pays the bills Jun 09 20:00:30 clear Jun 09 20:00:34 oops Jun 09 21:20:02 Visaoni: what's your github link? Jun 09 22:02:23 hi jkridner Jun 10 00:21:25 9 **** ENDING LOGGING AT Fri Jun 10 02:59:59 2016