**** BEGIN LOGGING AT Wed Sep 19 03:00:00 2018 Sep 19 07:16:56 does someone knows how to setup overlayfs on pocketbeagle running latest IoT image? Sep 19 07:18:14 futesta: uhh, the same way you'd setup overlayfs on any other linux system Sep 19 07:19:04 zmatt: .-) ok thats fine info... trying... Sep 19 07:26:38 comming from raspPI and there is with the latest strech image a simple way, only to add a param to cmdline.txt on the /boot Sep 19 07:27:29 so my question was if there is quit a simple way on the debian IoT image on the pocketbeagle Sep 19 07:37:31 I don't have a pi so I'm not sure what you're referring to... filesystems are normally setup in /etc/fstab, not in a boot parameter Sep 19 07:39:46 Arrrrrrr moaning maties! Let's set sail! Sep 19 07:42:02 ah, it's that day of the year again Sep 19 07:42:14 aye Sep 19 08:18:58 hi Sep 19 08:21:13 so i have been playing around with the device tree for simple-soundcard but i can't get it to work. i can't get it to set theformat to i2s this is what i currently have: https://pastebin.com/AdqgNbzb Sep 19 08:22:25 the documentation explains i have to put the format attribute under the dai-link node I have done this but it is not recognized. https://elixir.bootlin.com/linux/v4.14.40/source/Documentation/devicetree/bindings/sound/simple-card.txt Sep 19 08:22:32 maybe i am missing something obvious here :) Sep 19 08:23:06 looks okay to me. mclk-fs should not be used however Sep 19 08:23:28 oh Sep 19 08:23:40 my driver looks for that value Sep 19 08:23:49 so that is why i included it. Sep 19 08:24:03 mclk-fs means the master clock frequency should be configured to a fixed multiple of the samplerate, regardless of what samplerate is used. since your master clock is a fixed-frequency oscillator, mclk-fs is not appropriate Sep 19 08:24:45 ok that is strange then :D i'll have to take a look why it uses it, in the driver code :) Sep 19 08:24:52 how is your driver even obtaining the value of that property? Sep 19 08:24:57 you'd have to do really weird stuff to locate it Sep 19 08:25:38 it is used internally by simple-audio-card Sep 19 08:26:39 and stored in its private data structure Sep 19 08:28:10 oh, and using the mclk-fs property also implicitly causes simple-audio-card to assume the cpu generates the master clock, rather than taking it as input Sep 19 08:28:48 which means that 1. you'll get a drive conflict on the master clock io 2. you'll get a garbage sample rate Sep 19 08:29:41 well now that i look at it more closely it extracts it from dai->codec and put's it in it's own data structure through snd_soc_codec_get_drvdata() function. and can then access it through the mclk attribute. . Sep 19 08:30:57 but the data structure in which mclk_fs is stored isn't even declared in any header Sep 19 08:31:21 it's declared in simple-audio-card.c, hence totally inaccessible from codec drivers Sep 19 08:31:27 it strange because when i set mclk-fs it gets filled with the value :) Sep 19 08:33:32 here is the function in question https://pastebin.com/0JHyKgjH at line 68 is where the value is used (priv->mclk) Sep 19 08:34:08 if you need to know the mclk/fs ratio, you should just calculate it from the mclk (as given to you by the codec's set_sysclk method) and fs themselves Sep 19 08:34:46 you're just reading a private field of your codec there and not showing where it's being set Sep 19 08:36:20 it is being set in the set_dai_sysclk function like so: https://pastebin.com/CQmq8D4S Sep 19 08:37:24 well there you go. beware that only clk_id 0 will ever be set when using simple-audio-card Sep 19 08:39:34 i had taken this code from a similair driver in the 4.14 source but had no idea that simple-soundcard would only set clk_id to zero :) Sep 19 08:39:37 interesting that you need that though, it seems the mainline tas571x and tas5720 drivers don't even implement set_sysclk Sep 19 08:40:02 wel i looked at the tas5086 code Sep 19 08:40:10 what are mclk and sclk ? Sep 19 08:41:57 simple-audio-card only passes clk_id==0 because it's very unusual to require anything else Sep 19 08:49:00 seems this codec uses "sclk" to refer to the bit clock. its rate, if you need to know it, should simply be calculated from the hw params Sep 19 08:52:39 it seems rather strange to want to know that though Sep 19 08:54:26 well when i look it seems that it doesn't seem to use it anywhere Sep 19 08:56:17 and mclk is used to build a map, to figure out what ratio to use. but then this value is never used either Sep 19 08:56:20 :) Sep 19 08:56:57 looking at the datasheet, it does seem like the codec wants to know both fs and the mclk/fs ratio Sep 19 08:59:18 you might need to put the clocks property also on the codec Sep 19 08:59:53 ah yea the code setting that register (that mapped value) is commented out. Sep 19 08:59:54 or moving the clock property up one level (to the dai) might also be supported it seems Sep 19 09:00:19 *clocks property Sep 19 09:09:00 so it seems that the mclkfs ratio is fixed at 512x in the code, it is set once at init but then not changed again. Sep 19 09:09:30 well then you can support only one samplerate Sep 19 09:10:05 which might be fine, as long as the codec driver reports that back correctly so userspace can know what samplerate to use Sep 19 09:10:26 anyway, I kinda need to go Sep 19 09:10:34 that is fine, we only use one samplerate :) Sep 19 09:10:41 KotH: you said you have some budget for equipment left. How about this? https://youtu.be/DXYje2B04xE Sep 19 09:10:43 zmatt: alright, you have been of great help! :) Sep 19 09:11:21 KotH: it may not be very practical, but then again, you said you're all research ;) Sep 19 09:12:07 KotH: I think it's about $1.3 million or so Sep 19 09:12:21 thinkfat: i think... i NEED this! Sep 19 09:12:58 thinkfat: only problem: it's about half of the money our group has for the next 5 years Sep 19 09:13:10 KotH: find a sponsor Sep 19 09:13:19 thinkfat: would you sponsor me? :) Sep 19 09:13:28 * thinkfat looks the other way... Sep 19 09:14:10 thinkfat: i'm doing science for the betterment of humankind through application of atomic clocks to chocolate production and consumation Sep 19 09:14:10 I think that sort of money is about more than half of what a normal person earns in a lifetime. Sep 19 09:14:43 thinkfat: a normal person makes that much money in 10-15 years... in a civilized country with proper chocolate ;) Sep 19 09:15:38 it's a really impressive beast. Sep 19 09:15:50 jupp Sep 19 09:16:10 shahriar's videos have also got a lot better since i last watched one of them Sep 19 09:16:55 if you're into RF stuff, he's sure the place to go on youtube Sep 19 09:18:24 not, like, shortwave RF, but ultra high frequency stuff Sep 19 09:18:37 anything below 1GHz is DC Sep 19 09:19:30 but that scope, man, that's really the closest to the "DC to Daylight" claim by a long shot Sep 19 09:19:47 KotH: where have I heard that before... Sep 19 09:20:04 thinkfat: probably from your HF professor :) Sep 19 09:20:29 KotH: nah. I never studied that. I dropped out of EE before it came to it Sep 19 09:21:14 pfff! Sep 19 09:21:19 you missed the good part! Sep 19 09:21:24 I guess so Sep 19 09:23:26 16GHz sytem clock.... ^^; Sep 19 09:25:04 ok.. from the way he explains things, he definitely talked to some keysight engineers Sep 19 09:27:31 yes. I think it's a concerted action, I'm sure he was under embargo until Keysight outputs their own announcement Sep 19 09:28:17 if you look at the publish timestamps of Keysights intro videos and Shahriar's, they're just one hour apart. Sep 19 09:29:39 it's also pretty interesting that of all the EE channels on youtube he's the one who gets not only a prototype for doing stuff but also gets actual boards for inspection Sep 19 09:32:00 but no wonder they're not sending it to Dave Jones, though ;) Sep 19 09:32:04 he tends to break stuff Sep 19 09:38:47 KotH: just look at the back of the DSP. It must be a nightmare to fan that out on the PCB. Sep 19 09:42:29 8 layers, impedance controlled I guess Sep 19 09:57:50 Hello i will display the temperature on a 7 segmentdisplay and this is my code with the error message : https://pastebin.com/9dxhqwZf Can somebody help ? Sep 19 09:59:51 zmatt: overlayfs now working... so quit simple, example given in uEnv.txt, damn ;-) Sep 19 11:13:22 zmatt: it works :) i got audio out! Sep 19 11:31:51 OMG! Sep 19 11:40:27 Duality: hurray :) Sep 19 11:58:37 Can somebody help me finding the right library for the 7 segment display ? Sep 19 11:58:41 I dont know how Sep 19 14:22:58 is there a easy way to disable the mass storage device when connecting over USB to the PB unless editing the /opt/scripts/boot/am335x_evm.sh? Sep 19 14:23:35 i want to disable all the usb action when connecting via this Sep 19 14:46:28 m Sep 19 15:51:06 Duality: btw earlier I said moving the 'clocks' property up one level to the dai-link might work... it doesn't, I misunderstood the kernel code Sep 19 15:52:01 rather, the fallback is to the device that has the digital audio link (i.e. mcasp or the codec) Sep 19 16:06:03 i.e. the three options to declare the sysclk (in order of preference) are https://pastebin.com/NNiYJqHF Sep 19 18:28:29 I use beaglebone and my circuit accumulate a lot of statics. I suspect that come from the beagle bone. is there any way I can solve it? Sep 19 18:34:06 I am a beginner for beagle bone. I use beagle bone and my circuit accumulates a lot of statics. I suspect that come from the beagle bone. is there any way I can solve it? Sep 19 18:35:11 statics as in ESD or? Sep 19 18:37:39 I really not sure. i use 1.8V for anolog pin(pin 39 and pin 40) then statics accumulates alot in there Sep 19 18:44:39 you mean noise? Sep 19 19:10:31 what do you mean by statics? Sep 19 20:05:06 tbr: so, raspi 3b+ consumes 1.5W for doing essentially nothing. Sep 19 20:05:23 tbr: bbbw consumes 500mW for the same task ;) Sep 19 20:05:28 or, non-task Sep 19 20:05:39 tbr: and that Sep 19 20:06:06 tbr: number of 1.5W is only after you enable runtime pm for the USB IP on the Pi Sep 19 20:06:20 tbr: right out of reboot it consume 2.5W! Sep 19 20:06:23 can you believe that Sep 19 20:06:32 doing _nothing_! Sep 19 20:11:42 thinkfat: i will do some power measurments at the PB tomorrow ;-) Sep 19 20:12:36 dg7sgq: PB? Sep 19 20:12:51 pocketbeagle Sep 19 20:12:53 oh Sep 19 20:12:55 :) Sep 19 20:12:56 yes Sep 19 20:13:01 I have that one here, too Sep 19 20:13:02 get one last week Sep 19 20:13:10 plan to replace some Pi projets Sep 19 20:13:44 I just replaced one Sep 19 20:13:48 the PB have a big balls... Sep 19 20:13:54 bad joke Sep 19 20:13:56 pi just could not stand the heat in our garden hut Sep 19 20:14:27 coffee: nice! .-) Sep 19 20:15:12 that bga balls are huge Sep 19 20:15:34 but what a pity that the eth phy isn't on the headers ;( Sep 19 20:16:28 plan to use a WIZ or ENC Sep 19 20:16:45 coffee: have probs with shorts? Sep 19 20:18:43 acronyms? Sep 19 20:19:34 no... :-) shorts on the balls! Sep 19 20:20:35 i mean Sep 19 20:20:36 nope Sep 19 20:23:39 there is a lot inside, that need huge balls... Sep 19 20:25:22 because it is a system in package Sep 19 20:31:55 Anyway, it is so cute! Sep 19 20:33:36 jepp Sep 19 20:42:19 me, I just tried to fill the cart while ordering parts from digikey to get free shipping. they had the pocket beagle in stock, problem solved. :) Sep 19 20:46:36 If I ordered a chair from digikey, it would come inside a blue bag, love it. Sep 19 20:56:58 dg7sgq: time to just get an osd335x and do some deadbug soldering to reach the ethernet interface? ;) Sep 19 20:58:40 uuuuh Sep 19 20:59:26 with the size of the balls, and a good microscope, why not? Sep 19 21:00:30 microscope, for 1mm pitch? Sep 19 21:00:39 or 0.8 for the -sm version (iirc) Sep 19 21:00:47 I guess it might help Sep 19 21:01:07 jepp, but the osd335x is very pricy, PB is lower priced at digikey eg Sep 19 21:01:18 kinda weird Sep 19 21:01:46 I mean, I get how it happens, but it still feels kinda weird Sep 19 21:04:16 testing the speed of the WIZ and ENC at SPI in defualt config delivers aprox 4Mbit/s, thanks enough today for IoT Sep 19 21:04:55 WIZ? ENC? Sep 19 21:05:14 are the wiz drivers in the official image? Sep 19 21:05:35 zmatt: meaning the ENC28J60 and WIZ-chip für SPI ethernet Sep 19 21:06:18 ah Sep 19 21:06:20 CoffeeBreakfast: just activating the cape overlay in the uEnv.txt Sep 19 21:06:45 neat. Sep 19 21:07:14 4 Mbit/s sounds a lot lower than what should be possible from a hw perspective... I guess the kernel drivers just aren't terribly efficient Sep 19 21:09:15 zmatt: sure, but works out of the box and enough for IoT; btw i think the phy of the osd335x is mutch faster ;) Sep 19 21:09:28 still, it would be nice to have routed the ethernet, even underneath the SoP, as unpopulated smt Sep 19 21:09:46 CoffeeBreakfast: ack Sep 19 21:09:59 I suspect the pb layout is already very cramped though Sep 19 21:11:17 there is nothing special on the PB board, just do a own one, but the OSD is 37Eur at digikey :( Sep 19 22:19:53 Hey. Are the labels on the Cape mislabeled? Sep 19 22:19:59 Sorry. The MotorCape? Sep 19 22:21:09 For the Power in and GND? Sep 19 22:21:44 bbl Sep 19 23:04:37 Hey! Sep 19 23:04:55 The Co. that produced the Cape told me something. Sep 19 23:05:27 They say I should power the BBBW via 5v barrel jack instead of using the USB connection to power the board. Sep 19 23:05:32 I am off to test it. Sep 19 23:06:13 ahoy set_ Sep 19 23:08:00 Coffee! Sep 19 23:08:49 The MotorCape needs the BBB to be powered via 5v barrel jack instead of by USB. Now, I need to set up a .service file for running on boot. Sep 19 23:09:12 set_: oh lol it indeed uses vdd_5v instead of sys_5v for its 5v logic supply Sep 19 23:09:14 that's... weird Sep 19 23:09:21 I know. Sep 19 23:09:33 I am upset over this fact but if it is this way, so be it. Sep 19 23:10:23 Is there a way to change this? Sep 19 23:10:37 I have trouble understanding why they did that, sys_5v should have no problem supplying the current the driver needs on it logic supply Sep 19 23:11:26 I understand that fact. This is why I am upset. I tried hours and hours of software thinking I was incorrect. Little later...I was wrong. Way wrong. Sep 19 23:11:34 What about the other Capes that just came out? Sep 19 23:11:41 Same story? Sep 19 23:11:47 I'd need to check the schematics Sep 19 23:12:03 Do not worry. I will look later. Sep 19 23:12:53 sys_5v is behind a reseteable fuse right? Sep 19 23:13:32 CoffeeBreakfast: sys_5v is the output of the pmic's 5v power switch, which also selects between the two 5v inputs Sep 19 23:13:53 it's used for everything that needs a 5v supply Sep 19 23:15:08 (it's also the input supply for all LDOs and switching power supplies) Sep 19 23:19:52 zmatt: Where do I put my .service files again? Sep 19 23:20:31 etc / systemd / system? Sep 19 23:20:36 yes Sep 19 23:20:45 Okay. Awesome! Sep 19 23:21:12 you already have fully working software that's ready to be run as service? Sep 19 23:21:19 Yes. Sep 19 23:21:23 For starters. Sep 19 23:21:47 I just still want my motor to turn and the LED to signify this happening. Sep 19 23:22:02 So, off to the .service file concept. Sep 20 00:03:40 Dang it. I want to allow everyone right now to forgive me for my pursuit of the MotorCape, its vdd_5v supply, and my lack of experience. Sep 20 00:03:51 So, yes! I will accept my own apology. Sep 20 00:03:53 It works! Sep 20 00:03:58 I repeat. IT WORKS! Sep 20 00:11:03 no need to apologize in this case, the vdd_5v thing is really really counterintuitive and deserves a very obvious notice in a place where a potential buyer can't miss it Sep 20 00:14:28 ... Sep 20 00:16:01 My mistake. I even harassed the people at Mouser.com. I just sent in a brief explanation of the issue and what I thought about the set up. Sep 20 00:16:24 I let them know that I was mistaken and 100% wrong. Sep 20 00:16:47 I hate me sometimes. Sep 20 00:39:55 Now, onto posting something outrageous w/ software and the MotorCape! Sep 20 00:40:28 Do not worry. I will explain the idea about what you said zmatt, i.e. "vdd_5v and not sys_5v." Sep 20 01:35:31 Otay. I got me a new post on that hackster.io site. Sep 20 01:35:37 MotorCape! Sep 20 01:36:08 zmatt: Do you want credit for that hunk of software you gave me? Sep 20 01:36:26 I told the people that there was a person on #beagle on Freenode. Sep 20 01:36:30 Is that adequate? Sep 20 01:36:42 brb **** ENDING LOGGING AT Thu Sep 20 03:00:01 2018