**** BEGIN LOGGING AT Sat Mar 09 03:00:34 2019 Mar 09 03:00:39 more than enough even Mar 09 03:01:13 well the question would also be whether the interfaces you want are actually accessible of course in whatever subset the pocketbeagle has Mar 09 03:01:15 the latest talk about the bb ai .. that seemed interesting to me Mar 09 03:01:34 more that .. the BB has survived unlike many TI things Mar 09 03:01:49 the bb ai is a wholly different beast though Mar 09 03:02:02 It is supposed to have 4 PRUs! Mar 09 03:02:11 i like the idea if using my cortex-m knowledge instead of learning the pru Mar 09 03:02:20 DSP? Mar 09 03:02:21 pru is cool Mar 09 03:02:34 cortex-m is not a replacement for pru Mar 09 03:02:40 (nor vice versa) Mar 09 03:02:49 I found this odd board a while back w/ a m4 on it. Mar 09 03:02:58 because of the single cycle instructions of the pru? Mar 09 03:03:03 I figured I would learn one day. Mar 09 03:03:22 does that AI from BBB.io have a m4 on it, too? Mar 09 03:03:25 yeah, including some high-bandwidth data moves via the broadside interface Mar 09 03:03:32 do the cortex-m have access to shared memory and peripherals? Mar 09 03:03:34 and its direct I/O Mar 09 03:04:25 the cortex-m4 cores on the am572x ? I mean, yeah they can access everything in principle, but it's all via the L3 interconnect so you can expect very substantial latency and jitter Mar 09 03:04:34 ic Mar 09 03:04:45 they have no direct I/O like the PRU cores do Mar 09 03:05:00 hmm .. Mar 09 03:05:09 have you seen the announced stm32mp1? Mar 09 03:05:23 i would have thought that is how the am572x was setup Mar 09 03:05:24 the cortex-m4 cores on the am572x are typically used for things like managing video stuff Mar 09 03:05:38 I've glanced slightly at the stm32mp1 Mar 09 03:06:08 it looks neat Mar 09 03:06:29 not clearly better or worse than the am335x, mostly just different Mar 09 03:06:48 (not remotely comparable to the am572x) Mar 09 03:07:37 i've been using them for the last 5 years Mar 09 03:07:42 stm chips that is Mar 09 03:07:44 google did something too. Mar 09 03:07:54 I was about to say, stm32mp1 seemed extremely new Mar 09 03:07:55 that seems to be the nxp thing Mar 09 03:08:01 yeah you can't actually get it Mar 09 03:08:05 Oh. Mar 09 03:08:05 i think april Mar 09 03:08:18 the stm32 is april .. not sure about google Mar 09 03:08:40 I forget their name for the board but it was their make up and not NXP. Mar 09 03:08:57 coral Mar 09 03:09:05 Sounds familiar. Mar 09 03:09:21 I'm curious what the errata doc will look like half a year from now or so, especially since ST seems to have not much prior experience with SoCs of this size? Mar 09 03:09:40 this is their first 'A' cortex Mar 09 03:09:52 no, they had one before, but it quietly disappeared Mar 09 03:10:03 did they cheeky buggers Mar 09 03:10:11 cortex-A9 Mar 09 03:10:42 i don't remember seeing that one Mar 09 03:10:47 SPEAr1300 Mar 09 03:11:23 interesting .. shows up on there site with no part # Mar 09 03:11:46 2010 hmm Mar 09 03:12:05 heh i guess i was looking at msp430s then Mar 09 03:13:35 set_: btw, here's the class modified to support reversing a motor, just in case it's useful (it's a pretty trivial change): https://pastebin.com/R9RKt4hQ Mar 09 03:13:43 I've reversed motor3 and motor4 as example Mar 09 03:13:52 thanks for the info zmatt Mar 09 03:14:24 Rickta59: btw, since you want to play with pru, have I already plugged my py-uio library? ;) Mar 09 03:14:43 :) you probably have but not to me Mar 09 03:15:31 https://github.com/mvduin/py-uio#py-uio it's a python lib to directly interact with pruss (map shared memory, load executables, inspect and control core state, etc) Mar 09 03:15:42 includes a bunch of examples Mar 09 03:15:52 thanks Mar 09 03:16:09 * not a big python guy Mar 09 03:16:14 me neither Mar 09 03:16:32 this is my first python project of substantial size, and the more I use python the less I like it Mar 09 03:16:52 i used python once .. in 2002 wrote a exporter for blender 3d Mar 09 03:17:05 but the project started with the aim of making uio more accessible, and python is fairly accessible Mar 09 03:17:08 never wanted to use it again Mar 09 03:17:27 and it's nice to play with stuff in the repl Mar 09 03:18:02 i just can't get past the whitespace thing Mar 09 03:18:50 that doesn't bother me at all. in fact I've become in favor of significant whitespace over curly-brace (even though I primarily code in curly-brace languages) Mar 09 03:19:54 sane code is properly indented already anyway, and as a human it's what you look at to determine the block structures of code anyway. braces are just redundant information that give an opportunity to create a mismatch between what the human sees and what the compiler sees Mar 09 03:20:01 * looking at your examples Mar 09 03:20:07 you have an x15? Mar 09 03:20:10 yes Mar 09 03:20:17 what do you think of that? Mar 09 03:20:33 seems like the all dancing all singing bb Mar 09 03:21:20 I mean, it has a ton of stuff, and the AM572x SoC is a beast of course with its many cores and subsystems Mar 09 03:21:42 the board is quite inconvenient however for hardware experimentation. you can't just stick wires into it like you can with the beaglebone Mar 09 03:21:59 i guess that is what the bb ai is all about Mar 09 03:22:04 yep Mar 09 03:22:15 seems to use a chip that isn't available yet though Mar 09 03:22:19 the 5729 Mar 09 03:23:06 my x15 has one too... I suspect every x15 so far does, or at least the first batch... dunno if it's guaranteed or not Mar 09 03:23:34 I already confirmed the EVEs work on my x15 before I even knew they were supposed to Mar 09 03:25:09 haven't played with them yet though Mar 09 03:26:19 I hope the py-uio examples make some sense... I really ought to work on some actual documentation :P Mar 09 03:28:11 i'll check it out once i have reworking bb Mar 09 03:29:23 the latest console image should probably just work on it. if whatever is currently on it is really ancient you may need to hold down the S2 button when booting from sd card to bypass the stone-age bootloader on eMMC ;) Mar 09 03:29:43 zmatt: Here is the pastebin example. This is the exact error: https://pastebin.com/rSQL2vKt. Mar 09 03:29:44 ... Mar 09 03:29:51 i'll probably just get an sdcard going Mar 09 03:29:54 If you get bored, please look at that idea. Mar 09 03:30:22 set_: what does my show-pins utility show for that pin? Mar 09 03:30:30 Oh. Please hold. Mar 09 03:30:41 mvduin? Mar 09 03:30:53 Rickta59: ah you were looking into the nodejs usb flasher thing? I haven't tried this one yet myself Mar 09 03:31:00 set_: https://github.com/mvduin/bbb-pin-utils/#show-pins Mar 09 03:31:03 Okay. Mar 09 03:31:28 ? nodejs usb flasher? Mar 09 03:31:43 been a long time i don't actually remember what i did to get going Mar 09 03:31:49 i thought i copied a disk image Mar 09 03:31:50 Rickta59: I linked to earlier? Mar 09 03:32:07 https://github.com/ravikp7/node-beagle-boot Mar 09 03:32:33 ah ok .. i missed that first time around Mar 09 03:32:38 thanks Mar 09 03:32:46 using an sd card is probably easiest though, if you have one you can borrow to use to reflash the bbb Mar 09 03:33:11 i have a small laptop usb drive .. which is my eventual goal Mar 09 03:33:28 the bbb can't boot from an usb drive Mar 09 03:33:49 but it can bounce from a small sd card that then uses the usb drive as root no? Mar 09 03:33:53 or from emmc Mar 09 03:33:59 sure Mar 09 03:34:09 * that is how i have the rpi going Mar 09 03:34:21 should be easier having u-boot Mar 09 03:34:37 personally I generally avoid the usb interface when I can... although I think it works better nowadays than it used to Mar 09 03:35:11 why not just use run from eMMC ? Mar 09 03:35:27 it is only 2gb ... i'm hoping i didn't bork it up Mar 09 03:35:35 i think it probably still has angstrom on there Mar 09 03:35:43 P8_14 rx down 7 low >> Mar 09 03:36:01 set_: please copy the actual output, not a few random words Mar 09 03:36:08 or better yet, pastebin the output Mar 09 03:36:18 Okay. Mar 09 03:36:30 Rickta59: 2GB is plenty of space for many purposes Mar 09 03:36:43 have you done any bare metal bb? Mar 09 03:36:55 I have Mar 09 03:37:25 seems like it would be fun to have a 600MHz cortex to muck about with Mar 09 03:37:56 yeah, if you want to play with a baremetal cortex-A, the beaglebone would definitely be a good choice. Mar 09 03:38:26 the cortex-A8 is still pretty sane, and ROM initializes the SoC to a usable state already for you Mar 09 03:38:35 i was wondering about that Mar 09 03:38:54 you used the CCS stuff? Mar 09 03:39:11 there somethingware thing? Mar 09 03:39:14 their Mar 09 03:39:54 my baremetal codebase (which I've mostly used for small tests) is mostly C++ with small bits of inline asm, and only one asm file for the basic initialization and exception vectors: https://liktaanjeneus.nl/start.S.html Mar 09 03:39:59 https://pastebin.com/fKYr83Lu Mar 09 03:40:07 starterware is a steaming pile of pig manure Mar 09 03:40:08 thanks Mar 09 03:40:34 manure! Mar 09 03:40:47 set_: there's a cape on this beaglebone? proto cape? Mar 09 03:40:51 what were you using for a jtag device? Mar 09 03:40:53 MotorCape. Mar 09 03:41:11 ok, the overlay uses weird names for the pinmux nodes then Mar 09 03:41:15 Rickta59: xds100v2 Mar 09 03:41:39 i think that is on one of the cc1310 boards i have Mar 09 03:41:42 of the three low-cost jtag debuggers I've used, it is by far the best Mar 09 03:43:02 set_: oh right, that dumbass overlay. I remember previously recommending you to disable it Mar 09 03:43:13 What overlay? Mar 09 03:43:21 I do not remember. Mar 09 03:43:23 try uncommenting in /boot/uEnv.txt: disable_uboot_overlay_addr0=1 Mar 09 03:43:28 Oh. Mar 09 03:43:31 assuming the cape is on slot 0 Mar 09 03:43:35 (eeprom) Mar 09 03:43:35 Okay. Mar 09 03:43:38 Aw. Mar 09 03:43:40 the overlay for the motor cape Mar 09 03:43:45 Alright. Mar 09 03:43:49 it declares the motor direction pins to be... leds Mar 09 03:44:00 Hmm. Okay. No issue. Mar 09 03:44:14 note that not being able to use config-pin on these pins would be normal when an overlay is used Mar 09 03:44:17 that's not a problem Mar 09 03:44:22 Oh. Mar 09 03:44:25 the problem is you can't use the motor direction pins as gpios Mar 09 03:44:32 Aw! Mar 09 03:44:37 oh actually Mar 09 03:44:45 not sure, that might actually work on current kernels Mar 09 03:44:58 P8.14, 16, 18, 26 are my pins for GPIO. Mar 09 03:45:20 like, why were you trying to use config-pin anyway? Mar 09 03:45:30 does the python script have problems with the gpios? Mar 09 03:45:35 To test the config-pin utility. Mar 09 03:45:42 No. My LEDs light up. Mar 09 03:46:10 okay yeah. rcn patched the kernel to simply ignore gpio conflicts Mar 09 03:46:19 I can, in theory b/c of the lit up LEDs, use all four motors easily but there are issues. Mar 09 03:46:28 so the fact the overlay declares them to be leds ends up not actually being a problem Mar 09 03:46:33 Okay. Mar 09 03:46:37 So, that is not the issue. Mar 09 03:46:44 I am clueless. Mar 09 03:46:48 I mean, the leds just indicate the pwm outputs are enabled right? Mar 09 03:46:53 Right. Mar 09 03:47:03 the gpios are for direction. can the motors reverse direction? does that work? Mar 09 03:47:26 your script tested both forward and reverse iirc Mar 09 03:47:42 Yea. Mar 09 03:47:53 actually, if it were unable to export the gpios, the script would just fail entirely anyway Mar 09 03:48:00 so yeah there's no actual problem Mar 09 03:48:11 not being able to use config-pin on those pins is Working As Designed Mar 09 03:48:17 Okay. Mar 09 03:48:52 since those pins are for the cape, and their function is automatically configured for it (by the overlay that's loaded automatically) Mar 09 03:48:52 So, what do you think I should do about the motors and the directions? Just wire them differently as a starter? Mar 09 03:49:01 Oh. Right. Mar 09 03:49:19 I gave you an updated version of my script with support for declaring motors to be reverse-direction Mar 09 03:49:35 Yea. A while ago. Mar 09 03:49:38 I still use it. Mar 09 03:49:42 37 min ago Mar 09 03:49:49 What? I did not see it. Mar 09 03:50:06 then scroll back 37 minutes Mar 09 03:50:11 Alright. Mar 09 03:51:42 I checked. I do not see it. Mar 09 03:51:48 Too many black characters. Mar 09 03:52:05 I checked in the logs. Mar 09 03:52:15 They have the times in the logs. Mar 09 03:52:46 04:13 < zmatt> set_: btw, here's the class modified to support reversing a motor, just in case it's useful (it's a pretty trivial change): https://pastebin.com/R9RKt4hQ Mar 09 03:52:49 04:13 < zmatt> I've reversed motor3 and motor4 as example Mar 09 03:53:26 Oh. I found it. Mar 09 03:53:36 It was when you and the other person were discussing python. Mar 09 03:55:20 Rickta59: oh I also made this silliness years ago, in response to someone asking for a pure assembly example (for some reason): https://github.com/mvduin/bbb-asm-demo Mar 09 03:55:42 of course there's no good reason to actually do so much in asm Mar 09 03:56:38 heh, I even made a release for it: https://github.com/mvduin/bbb-asm-demo/releases Mar 09 03:56:42 zmatt: I thought this program could already move in reverse w/ the - sign? Mar 09 03:57:39 set_: yes. the point of this is that you can change the default for a motor, i.e. swap the behaviour of positive vs negative speed for individual motors Mar 09 03:57:50 effectively swapping the leads of the motor, except in software Mar 09 03:58:04 Okay. Wozzers! Mar 09 03:58:21 I will try it. Mar 09 03:58:52 so if you feel like a motor is going the "wrong way" when you set it to 100, just add reverse=True to the declaration of the motor (see the modified declarations of motor3 and motor4 as example) Mar 09 03:59:05 Okay. Mar 09 04:00:51 Yea. I saw it. Mar 09 04:00:57 Thank you. Mar 09 04:16:06 set_: btw, if you're impressed with the two dual-core pruss instances (4 pru cores total) of the am572x, the new AM65xx has three quad-core PRUSS instances (12 pru cores total) ;) Mar 09 19:52:08 I think I broke my BB :-( Whenever I apply power it only flashes the power led and pressing the reset/power buttons doesn't really do anything. Also, have tried leaving it powered off for a couple days, no difference. Any suggestions? Mar 09 19:52:48 mine did this too Mar 09 19:52:49 so I replaced it Mar 09 19:53:06 maybe you shorted its pins with your aluminium computer case, or did a wrong electrical connection Mar 09 19:53:23 Yeah, happened when I was messing with serial connectivity and that's all I can assume. Mar 09 19:53:39 makes a sad cereal :-( Mar 09 19:54:20 :( Mar 09 19:54:44 do you remember the connections you made ? Mar 09 19:54:54 yeah, i was pretty confident it was right Mar 09 19:54:59 so that you avoid doing the same thing when you get a new one Mar 09 19:55:11 but maybe it's not the cause, it could be a metal table, junk electronics on the table, whatever Mar 09 19:55:26 yeah exactly, I think that's more likey the culprit honestly Mar 09 19:55:45 i wonder if I can send it in to get repaired... Mar 09 19:55:50 I don't think so Mar 09 19:55:53 but you can try Mar 09 19:56:18 yeah, i'd assume thats the case but, i can always ask! Mar 09 20:48:48 is it ok to run debian jessie with the 4.19 kernel? Mar 10 00:48:22 Rickta59: .... WHY Mar 10 01:01:57 cereal: serial connectivity to what? Mar 10 01:03:04 cereal: be sure to never expose the expansion header pins to more than 3.3V (or 1.8V in case of the ADC pins). also never apply any voltage to expansion header pins while the beaglebone is unpowered Mar 10 01:03:57 (generally the latter is achieved by powering whatever you connect to the expansion headers from the 3.3v supplied by the beaglebone itself) Mar 10 01:57:40 If under /dev/ it does spidev1.0, 1.1, ..., does that mean that SPI is configured and ready to transmit information? Mar 10 02:10:36 (nope) Mar 10 02:19:01 What's an easy way to check if my SPI is configured? Mar 10 02:27:39 Anthony_: I'm guessing you missed the stuff I said to you yesterday (was it yesterday? I think so) Mar 10 02:28:19 to use spi (and most other things) on the latest image you don't need to mess with /boot/uEnv.txt at all, but you do need to configure the pins to spi mode using the 'config-pin' utility Mar 10 02:29:31 if you want to use spi1 rather than spi0 you'll also have to disable hdmi audio by uncommenting the relevant line in /boot/uEnv.txt since it occupies some pins needed for spi1 (config-pin would fail for those pins otherwise) Mar 10 02:32:17 as alternative to using config-pin you can also configure an overlay in /boot/uEnv.txt to setup spi pins, e.g. uboot_overlay_addr4=/lib/firmware/BB-SPIDEV1-00A0.dts (or BB-SPIDEV1A1-00A0.dts or BB-SPIDEV0-00A0.dts depending on which pins you want to use). disabling hdmi audio would still be required for both spi1 options Mar 10 02:32:42 you can double-check your current pin configuration e.g. using my show-pins utility: https://github.com/mvduin/bbb-pin-utils/#show-pins Mar 10 02:36:51 the fact that /dev/spidev* (and many other kinds of devices) show up regardless is a consequence of the universal overlay. it enables basically everything and lets you change pinmux at runtime (using config-pin). the device drivers are not aware of the fact their devices may not actually be connected to any external pins via pinmux **** ENDING LOGGING AT Sun Mar 10 02:59:56 2019