**** BEGIN LOGGING AT Sun Oct 28 02:59:59 2018 Oct 28 03:33:15 this guy is funny https://youtu.be/MzxLYGWQddA?t=15 Oct 28 04:02:51 "now with 0 support" Oct 28 04:02:54 hah, this cape looks pretty cool => http://www.ti.com/tool/DLPDLCR2000EVM Oct 28 14:30:34 how to interface camera and ultrasonic sensor with beaglebone black Oct 28 14:31:10 how to interface camera and ultrasonic sensor with beaglebone black Oct 28 16:50:05 I wonder if we should set a bounty for the quickest leave after asking a question Oct 28 16:50:29 but the time resolution from the IRC server is probably not good enough Oct 28 18:37:56 or rate questions according to vagueness :) Oct 28 18:39:25 hard to beat the occasional "HELP!!!???!!!!" and quit about 5s later Oct 28 18:42:35 " how to interface camera and ultrasonic sensor with beaglebone black" Oct 28 18:42:44 very carefully and with great thought. Oct 28 18:43:01 but I think that's a 2nd to yours lol Oct 28 18:45:25 there is a certain eclecticism to sid's statement, don't discount that Oct 28 18:47:24 it didn't sound like quick abbreviated prospeak to me though. Oct 28 18:47:56 the correct answer of course would have been "using wires" Oct 28 18:50:20 echo down > ../reactor1/controlrods doesn't work, HELP!!! Oct 28 18:51:24 root@blockIV.chernobyl:~# Oct 28 18:52:20 zmatt: that is realllly cool Oct 28 18:52:28 i wanna meet whoever di that Oct 28 18:52:46 you mean the cape? Oct 28 18:52:50 yeah Oct 28 18:52:51 i want one Oct 28 18:52:53 i might be it Oct 28 18:52:54 now Oct 28 18:52:59 also, IBM buying redhat? Oct 28 18:53:03 questions thoughts concerns? Oct 28 18:53:11 yeah I was considering getting one too Oct 28 18:53:20 "there goes the planet" Oct 28 18:53:47 Someone turn on the Stallman Signal Oct 28 18:53:54 * ayjay_t donates all of his money to the FSF Oct 28 18:54:10 * ayjay_t fights off junior sales people 2 mo out of underground trying to sell him RedHat Watson Oct 28 18:54:15 undergrad* Oct 28 18:55:36 isn't ibm a major linux user/contributor already? Oct 28 18:57:40 ibm's executive leadership and product development is heavily influenced and controlled by their enterprise sales and advertising departments Oct 28 18:58:15 it's also _cheap_ Oct 28 18:58:19 like, cheap as in low quality Oct 28 18:58:28 the profit per employee is about 50% that of redhat's Oct 28 18:59:14 ibm low quality? that's a statement that surprises me. I thought they were mainly into really expensive but high-end stuff? Oct 28 18:59:31 though to be honest I can't say I really keep track or anything Oct 28 18:59:31 yeah i phrased that incorrectly Oct 28 18:59:36 it is really high quality Oct 28 18:59:48 at least expensive Oct 28 19:00:30 and with expensive but really long-term support Oct 28 19:00:42 that's the associations I have with ibm anyway Oct 28 19:00:43 well i mean, that's both redhat and ibm's business models Oct 28 19:01:17 they are your engineer they're basically service companies that maintain a set of IP upon which they provide service Oct 28 19:01:23 trying to replace non-core parts of your infrastructure Oct 28 19:05:09 "Nobody ever got fired for buying IBM" Oct 28 19:07:32 well you can really do a lot with that much cash right? Oct 28 19:07:55 my office takes risks as long their cheap but if you're paying 10x you know you can move fast through a lot of iterations until you find a solution that works Oct 28 19:08:02 theyre* Oct 28 19:08:15 those soln might be tech, employees, vendors Oct 28 19:13:56 that's my _current_ take on how to spend money in engineering anyway Oct 28 19:40:50 its really interesting either way Oct 28 19:41:05 its a lot to unpack Oct 28 23:15:04 boo. C++17 introduced some sanity by giving many things an explicit order of operations, but g++ 8.2 still complains "operation may be undefined (-Wsequence-point)" Oct 29 01:01:46 now to get this alsa stuff working :) Oct 29 01:03:01 I think I'm going to give up on simple audio card for a bit because it doesn't give me much debugging info when it doesn't boot and I'm going to try the way suggested in the TI docs... I'm still getting snd-adau1977-adc ocp:sound: ASoC: CODEC DAI adau1977-hifi not registered Oct 29 01:03:12 but all the modules load right Oct 29 01:04:01 huh you're still having problems with simple-audio-card? how? Oct 29 01:06:42 can I see the last version of the DT you had with simple-audio-card? (I'm not going to dig into the legacy stuff so if you insist on going that route I can't help you) Oct 29 01:16:49 I'll just paste bin it Oct 29 01:17:36 https://paste.debian.net/1049469/ Oct 29 01:20:56 is there a specific reason you're overriding the configuration for the i2c2 bus and pins? Oct 29 01:21:08 zmatt: I can paste my boot logs too with simple-audio-card but I can't see much in the logs Oct 29 01:21:54 no, i just copied it because I did for i2c1 Oct 29 01:22:06 though I'm using the default pins for i2c2. Oct 29 01:22:20 yeah, i2c2 is enabled by default since it's the i2c bus also used for cape autodetection Oct 29 01:22:40 it shouldn't hurt enabling it explicitly though, I was just wondering Oct 29 01:22:42 yeah, I'm not using any of the cape stuff / auto detectiojn Oct 29 01:23:15 well you are, unless you built a custom u-boot Oct 29 01:23:37 oh I meant a cape with eeprom etc Oct 29 01:23:55 i.e. i don't need to support any other capes Oct 29 01:24:02 I understand you're not taking advantage of it, but the bus is still being configured and auto-probing is still being done :) Oct 29 01:24:09 yeah Oct 29 01:26:14 hmm, didn't your previous version use a different osc than the beaglebone's integrated 24.576 MHz osc ? Oct 29 01:26:51 not on purpose :) Oct 29 01:27:41 I don't have an external clock, happy with the codec or mcasp clocking Oct 29 01:28:50 okay, so the beaglebone's audio osc is providing the codec with a 24.576 MHz master clock (on P9.25), and the codec provides the audio bit clock (P9.31), data (P9.30), and frame sync (P9.29) ? Oct 29 01:29:08 yes Oct 29 01:29:30 data is 4 channels TDM Oct 29 01:29:50 on a single serialiser / pin Oct 29 01:33:35 and P9.30 is mcasp0_axr0 which I assume is the first serializer. I've set the serial-dir to <2 0 0 0> Just because I don't see any examples with a single enabled one but I assume it would be fine too Oct 29 01:33:46 yes that's fine Oct 29 01:34:15 having #sound-dai-cells in simple-audio-card,codec is bogus btw, although it probably shouoldn't cause problems Oct 29 01:34:29 so you said you had boot failure with this? what did the serial console say? Oct 29 01:34:40 https://paste.debian.net/1049472/ Oct 29 01:35:38 so failed on fdt_overlay_apply(): FDT_ERR_NOTFOUND looks like the issue, but it does find the file so I assume that's all the info I get as far as parsing errors Oct 29 01:35:39 what on earth is going on there Oct 29 01:36:15 no idea :) Oct 29 01:36:49 okay so this has nothing to do with simple-audio-card or indeed anything going on in the DT or in the kernel, but u-boot is borking up big time Oct 29 01:37:17 ? If I disable simple-audio-card everything is working for me. Oct 29 01:37:18 oh or I guess it's probably in the DT overlay Oct 29 01:37:31 FDT_ERR_NOTFOUND probably means an unresolved reference or something? Oct 29 01:37:36 yeah Oct 29 01:37:39 I assume so Oct 29 01:37:41 so... let's hunt for the typo Oct 29 01:37:56 or first, lemme check what that error means Oct 29 01:38:33 this is one of the reasons I don't use overlays... when using a monolithic dts, the device tree compiler will simply fail at compile-time if there are any unresolved references Oct 29 01:38:48 ahh okay Oct 29 01:39:19 whereas dtc can't verify any references when compiling an overlay Oct 29 01:39:40 and apparently u-boot currently reports errors via a boot failure ;) Oct 29 01:39:54 interesting ) Oct 29 01:39:56 :) Oct 29 01:40:25 thus why I was contemplating the just do it as a kernel module method, so GCC can tell me things :) Oct 29 01:48:34 the codec is presumably powered from the beaglebone's 3.3V ? Oct 29 01:48:50 yeah, and I can see it over I2C Oct 29 01:49:08 it ends up in sys correctly when I use the non-simple audio card thing Oct 29 01:49:21 (the way analog suggests) Oct 29 01:49:50 oh I see the mistake already, or at least a mistake Oct 29 01:50:03 could be one of many :) Oct 29 01:50:44 yes, which is why I'll finish what I'm doing (which is converting your overlay to dtsi format so I can compile the base dt with your definitions added and get errors for all missing references) Oct 29 01:51:13 ahh and just have a single one with no overlays? Oct 29 01:52:43 you can still use overlays if you prefer... if you write it in dtsi format and use my overlay-utils to convert it into an overlay then you can easily go between using overlays vs using a monolithic dts Oct 29 01:53:19 if it makes it so the compiler can check everything that sounds nice to me :) Oct 29 01:54:00 if you're making a custom product based on the beaglebone then using a monolithic dts is probably worth considering Oct 29 01:54:08 yeah I am. Oct 29 01:55:04 though longer term contemplating what the chip will be. It's vexing the module the pocket beagle uses costs more than a pocket beagle... Oct 29 01:56:24 looks like this was the only error... here's the bane of your existence apparently: Oct 29 01:56:27 ERROR (phandle_references): Reference to non-existent node or label "i2c2_pins" Oct 29 01:56:51 ahh okay Oct 29 01:57:05 they are called bb_i2c2_pins... Oct 29 01:57:41 I'd simply call the label i2c2_pins, but that's just my preference for not using pointless verbosity Oct 29 01:57:56 like... if not the bb's i2c2 pins... whose? Oct 29 01:58:03 but I probably don't need that section anyway, given it'll be defined already Oct 29 01:58:12 yeah, I used the name seeed used. :) Oct 29 01:58:18 wouldn't be my choice Oct 29 01:59:17 well I suppose it might be nice to configure the pins to fixed function (rather than allowing it to be runtime-configurable), in which case you should also declare the pinmux block. Oct 29 02:00:14 like I do here? Oct 29 02:03:49 so simple card module gets loaded now and things boot Oct 29 02:04:21 btw your tx/rx-num-evt values in &mcasp0 seem really low Oct 29 02:05:37 they are the size of the fifo buffers? Oct 29 02:06:38 the ctag card and the seeed card both set them to 1 Oct 29 02:06:39 basically it means the dma controller will be asked to perform a transfer (from fifo to memory) every N samples Oct 29 02:07:09 weird. I'd set it to 32 Oct 29 02:07:31 mmm... I wonder if it actually means 1 block or something? Oct 29 02:07:39 where block is defined elsewhere Oct 29 02:08:23 bb-audi sets it to 1 as well Oct 29 02:08:23 no, it's in words (i.e. samples) Oct 29 02:08:32 maybe they all copied each other :) Oct 29 02:08:53 BB-HDMI-TDA998x-00A0 uses 32 Oct 29 02:09:36 the soudns like it makes more sense Oct 29 02:09:40 at the very least I'd recommend using a multiple of 8 Oct 29 02:09:49 otherwise what do you have a buffer for. Oct 29 02:10:05 yup, <1> is pretty much "don't use the fifo" Oct 29 02:12:39 oh actually <1> doesn't even make sense Oct 29 02:12:51 no wait, it does Oct 29 02:13:18 I thought for a moment it needed to be multiple of the number of slots, but that's nonsense... it just needs to be a multiple of the number of serializers, which is 1 in your case Oct 29 02:18:34 wait, but you're using a beaglebone green, not black? Oct 29 02:18:47 I noticed the M-BB-BBG-00A0.dtbo in your log Oct 29 02:18:47 yeah Oct 29 02:18:50 green Oct 29 02:18:52 does it even have an audio osc? Oct 29 02:19:07 considering the BBB added that for the sake of hdmi, which the BBG doesn't have Oct 29 02:19:34 you mean the mcasp0 or an actual crystal? Oct 29 02:19:52 I mean the actual audio oscillator Oct 29 02:19:57 24.576 MHz Oct 29 02:20:13 okay, it seems it does have it Oct 29 02:20:29 yeah, thought it did Oct 29 02:21:29 though it can generate 24 internally? in the AM3358 Oct 29 02:22:11 thought I remembered reading something along those lines. Oct 29 02:22:18 yeah it can provide a 24 MHz clock. not generally considered useful for audio purposes since it's not a multiple of any plausible samplerate Oct 29 02:23:07 https://pastebin.com/raw/hNVMzFTU anyway, this is what I have after a bit of cleanup ;) Oct 29 02:23:24 not sure why I did that, but here you have it in case it's of any use Oct 29 02:24:30 neat, thanks zmatt Oct 29 02:25:12 that seemed to be the easiest way to specify the regulator as I don't actually have it powering down anyway Oct 29 02:25:25 where did you get those channels-{capture,playback}-override properties from? they don't seem to exist in kernel 4.14 Oct 29 02:25:27 no power management for the audio path Oct 29 02:25:53 the seeed example Oct 29 02:25:54 I understand, my comment just meant: surely the 3.3V supply you're using (from the beaglebone) already has an existing node somewhere in DT ? Oct 29 02:26:09 but I was too lazy to actually dig for it Oct 29 02:26:21 ahh, yeah. I was also too lazy :) Oct 29 02:27:41 I guess it would be &vmmcsd_fixed .. although it's a bit dumb they named it that Oct 29 02:27:52 but the 3.3v on the expansion header is the same one feeding eMMC and the SD card slot Oct 29 02:28:04 ahh cool Oct 29 02:32:49 btw, note that with these declarations the audio osc will be shut down when the audio path is idle. if this confuses the codec (which I think would be a bug in the codec driver, but that doesn't mean it won't happen of course) then you may instead want to use the fixed-clock as source and add a declaration like this: https://pastebin.com/raw/5qte7HdE Oct 29 02:34:23 ahh cool I did see mention of that in some of the examples Oct 29 02:34:29 and those "override" properties definitely do not exist in 4.14 (neither 4.14-bone nor 4.14-ti) Oct 29 02:35:02 seems to detect the 4 inputs correctly so not needed either way Oct 29 02:35:16 it normally shouldn't be indeed Oct 29 02:42:21 acording to arecord things are actually happening :) Oct 29 02:44:57 zmatt thanks again for all your help getting this up :) Oct 29 02:45:44 also, I won't claim to really understand what it all does, but as far as I understand this is how you can explain to alsa what's actually connected to the adc's inputs: https://pastebin.com/raw/yhRZ6HJ1 Oct 29 02:46:14 (AIN1-AIN4 reference widgets declared by the adau1977 driver) Oct 29 02:46:44 so that'll just give you prettier labels if you want right? Oct 29 02:46:54 or does the widget type actually do something? Oct 29 02:47:15 alsamixer already gives me all the controls correctly Oct 29 02:47:26 I think the type is probably just informative and for user interfaces Oct 29 02:48:40 as for what it does... they are called "DAPM widgets" where DAPM stands for Dynamic Audio Power Management, so presumably widget routing can have tangible effects. I don't know whether merely declaring plugs attached to the inputs will have any consequence other than labeling though Oct 29 02:49:04 this is one of the parts where my understanding of asoc is pretty foggy Oct 29 02:49:05 yeah, i assume if you've got jack roles defined for hot plugging Oct 29 02:49:41 you can set it so that jacks do things on detecting connections Oct 29 02:49:52 I don't need anything snazzy like that though Oct 29 02:49:59 at the very least this would tell alsa that it's four mono inputs, rather than e.g. two stereo inputs Oct 29 02:50:53 see am57xx-beagle-x15-common.dtsi for an example of one stereo line output and one stereo line input Oct 29 02:51:03 not sure if that'll make much difference in my code but useful to know. Oct 29 02:51:09 probably not Oct 29 02:51:22 I don't have any widgets or routes myself either Oct 29 02:51:36 so this is more of an "just so you know this is a thing that exists and does something" Oct 29 02:52:01 current issue is that when I run: arecord -vvv -c1 -f dat -D plughw:0,0 /dev/null Oct 29 02:52:15 the metering either gives me on or off... Oct 29 02:53:08 Max peak (12000 samples): 0x00007fff #################### 99% Oct 29 02:53:50 or 0% nothing in between, even with very low input signals... so maybe my analog bit is broken or maybe the code bit is a bit confused Oct 29 02:57:32 hmm, I don't know if micbias will get automatically enable if there's no dapm route for it Oct 29 02:57:35 *enabled Oct 29 02:57:49 oh... that would be problematic Oct 29 02:59:36 again, something I'm foggy about Oct 29 02:59:42 but I have seen MICBIAS in dapm routes Oct 29 02:59:55 and the adau1977 defines a MICBIAS widget **** ENDING LOGGING AT Mon Oct 29 03:00:00 2018