**** BEGIN LOGGING AT Thu Jun 22 03:00:02 2017 Jun 22 04:01:43 zmatt - it's about 45 bits wide. Most are just inputs however. I can't see needing more than 30 bits back out, 12 of which would be the minimum at all times is what I mean. Jun 22 04:02:27 I did some research and can't find delays for the BBB IOs using PRUs. Jun 22 04:03:30 its a new ns from memory .. though I dont recall where from. Jun 22 04:04:19 Does anyone know what delay times are for bits or groups of bits? Lets say for a test if I just want to watch 16 IOs and mirror them on 16 other IOs. So a simple signal in on one pin and back out on another? Jun 22 04:05:51 I know with the Pi I need to deal with the scheduler. Which is inconsistant and slows things down as well. Doesn't seem this is the case with PRU which is perfect. Jun 22 04:06:14 no, the PRU runs pretty much free Jun 22 04:06:23 unless you stop it ofc :D Jun 22 04:08:59 when using pru's dedicated gpios the roundtrip latency (input -> output) was around 25ns iirc Jun 22 04:09:43 but there aren't 45 dedicated pru gpios Jun 22 04:09:46 can you get a 48-bit bus into the PRU ? ah Jun 22 04:10:04 wonder if you could mux it then .. Jun 22 04:10:40 zmatt: what does the "spreadsheet of wisdom" say? Jun 22 04:10:48 can you get 32? Jun 22 04:11:22 46 is 23 twice .. 48 would be 24. Jun 22 04:13:30 looks like 24 accessible (for both cores combined), or 26 if you're willing to give up eMMC Jun 22 04:14:30 I guess two more using edio Jun 22 04:14:37 if you got a fast switch, you could mux 48 in, and 12 out without too much hassle then... Jun 22 04:15:07 pretty hard-core mind Jun 22 04:15:24 well the 6502 -> pru thingy also used muxing Jun 22 04:15:38 ah what thingy was that? Jun 22 04:15:47 http://elinux.org/images/a/ac/What%27s_Old_Is_New-_A_6502-based_Remote_Processor.pdf Jun 22 04:17:08 LOL @ PRUSS - the ultimate bit-banger Jun 22 04:17:20 depending on the bus protocol you may also be able to overlap inputs and outputs, though only on the (bidirectional) edio pins Jun 22 04:19:08 "When in doubt your pinmux is wrong ... check and quadruple check .. then check a fifth time ... " LOL Jun 22 04:19:12 I like this guy :D Jun 22 04:19:49 zmatt: that is pretty cool Jun 22 04:19:59 I also liked "Why? Because we can." Jun 22 04:20:07 (Y) Jun 22 04:22:33 Wow, 25ns is perfect. The IO limit is a little tight, but I think I can make that work. I'll check out the 02 project. Close to my idea I bet. Thanks! Jun 22 04:22:37 Henry_: btw the "spreadsheet of wisdom" veremit mentioned is https://goo.gl/Jkcg0w :) Jun 22 04:23:02 :D Jun 22 04:23:09 P9/P8 tabs are most useful as quick reference, PRU tab is what I checked for this of course Jun 22 04:23:23 the Signals tab can also be handy at times Jun 22 04:24:08 NoteToSelf: I need to find (1) water valves to trigger from B* (2) a minimal install ISO with git ... [totally unrelated] Jun 22 04:24:25 zmatt: the whole spreadsheet is handy :P Jun 22 04:24:56 its "everything you need to know about BBB IOs but were afraid to ask" :P Jun 22 04:25:17 Henry_: btw although I have no particular love for the pi, I do feel compelled to point out that you can escape the linux kernel there too of course, by running code on the VideoCore processor Jun 22 04:27:12 of course that still ain't going to get you PRU-like crazy low latencies :) Jun 22 04:28:43 veremit: well, the last two tabs have dubious value :P Jun 22 04:28:48 zmatt: I wouldn't pick the Pi either .. Jun 22 04:28:54 zmatt: but but .. :P Jun 22 04:29:01 they're pretty though Jun 22 04:29:16 trying to "big up" your 'sheet man :P lol Jun 22 04:29:47 and it did reveal something kinda weird to me Jun 22 04:29:48 zmatt: I don't know if the VideoCore processor allows for faster GOIP access however. That's kind of what lead me here. I need something with 70ns or faster IO access. Jun 22 04:30:02 that's pretty tight Jun 22 04:30:38 Well, really I have a 350ns bus window. But I need to get data in, then process it, then back out again if needed. Jun 22 04:30:57 ah but that's still not the same thing Jun 22 04:31:04 So 150ns for IO and 150-200ns for processing I assumed. Jun 22 04:31:13 Hello beaglers, just bought a BBBW and I am having some trouble configuring the WiFi, I go in connmanctl and do a "enable wifi, scan wifi" but I get no results Jun 22 04:32:30 so, on the 1GHz superscalar cortex-a8 the processing would probably be a lot faster than on the 200MHz unpipelined PRU Jun 22 04:32:40 but I/O latency would kill you Jun 22 04:32:53 Sure. So trade off. Jun 22 04:33:14 iirc around 150ns "ping time" Jun 22 04:33:24 is the data input/output continuous? Jun 22 04:33:42 zmatt: wonder if you can memory map it, to get better IO performance? Jun 22 04:33:59 veremit: uhh, all I/O is memory-mapped Jun 22 04:34:04 And on the Pi I don't think I can run OpenCL on the GPU. I kind of looked in to that and it's mostly closed source from what I can tell. Jun 22 04:34:04 oh nvm... Jun 22 04:34:41 Henry_: note that the "GPU" is actually the boot CPU, the ARM cores are technically auxiliary Jun 22 04:34:59 I do believe the Pi3 GPIOs however are on the GPUI data bus. So the should be quicker. Jun 22 04:35:43 afaik the three pis only differ in the ARM subsystem they stuck onto the side of the videocore4 SoC Jun 22 04:36:19 veremit: Data would not be continuous, but pretty close to it when under use. It would be idle most times however I believe. Jun 22 04:36:47 Henry_: have any protocol docs or such? Jun 22 04:37:49 Not really. I'm basically trying to use a SBC like the BBB as an ISA card - if that makes much sense. The idea being other interface cards can then be emulated in code. Jun 22 04:38:01 ah, ISA Jun 22 04:38:06 Using the built in BBB resources to connect to. Jun 22 04:39:23 Users can write their own apps. Like emulate a print card and when print it goes to a PDF, or the screen, or email.. etc. Up to user. Jun 22 04:39:50 "emulate ISA card" was the info I was looking for :) Jun 22 04:40:27 so definite similarities to the 6502 project Jun 22 04:41:07 kay... 24-bit address bus, 16-bit data bus, and assorted muck... hmm Jun 22 04:41:13 From what the PDF about the 02 project seems more like for testing? Jun 22 04:41:31 Well, 8bit Dx bus, but yes. Jun 22 04:41:54 And some other clock signals, IO/card signals... etc. Jun 22 04:42:23 the 6502 project was about making the 6502 run out of memory that was actually on-chip SRAM of the beaglebone Jun 22 04:42:35 So about 45 signals all together, and would like to base them on the system clock to keep things syncronised. Jun 22 04:42:57 PRU does support a parallel latch thingy Jun 22 04:43:10 "parallel latch thingy" ha! Jun 22 04:43:24 I'll have to read up on that. Jun 22 04:43:45 Like latching groups of bits at once? Jun 22 04:44:17 That's where the Pi killed me. Each IO was only able to be read one at a time. Jun 22 04:45:06 that sounds unlikely Jun 22 04:45:20 generally you always read 32 gpios at the same time or so Jun 22 04:45:35 depends on the python lib :) Jun 22 04:46:03 the parallel latch thing means the inputs get latched on rising edge of an external signal, which gives you more time to read them Jun 22 04:46:26 Ahh, ok. That makes sense. Jun 22 04:46:57 ah, I found timing diagrams... let's see what sort of deal this is Jun 22 04:47:24 Sounds like I need to do some testing. If I were to interface with an 8088 XT with ISA for example, seems like this could work then. Jun 22 04:47:38 I could see all the bus signals and emulate a card using the BBB. Jun 22 04:48:22 And even keep designs synchronous based on the CPU clock. Jun 22 04:48:37 Async gets messy quick. Jun 22 04:48:49 well the problem is that pruss obviously doesn't run on the bus clock Jun 22 04:49:41 The BBB bus clock or mine you mean? Jun 22 04:49:47 (ignoring the theoretical possibility of using a control loop to adjust the PLL output to PRUSS to sync onto the bus clock... *whistles innocently*) Jun 22 04:49:52 the isa bus clock Jun 22 04:50:36 Ok. I see. Jun 22 04:52:01 Would anyone recomend another SBC platform to do what I'm trying to acomplish? I just kind of found the BBB by accident. Jun 22 04:52:49 well, PRU cores are pretty much your best bet I think, short of going for FPGA Jun 22 04:53:49 Ok, I thought so too. I can go full FPGA but then it makes Open Sourcing the thing so much more complex for users. Jun 22 04:54:02 yeah, and ditto if you go for an obscure board Jun 22 04:54:09 the bbb is pretty tinker-friendly also Jun 22 04:54:37 Exactly. Here there's other projects users could load and base designs off of too. So that was my idea with the Pi as well. Jun 22 04:55:05 like, the bbx15 (pretending for a moment it were available yet) has twice as many PRU cores and lots more I/O, but you can't just stick some wires into its expansion headers Jun 22 04:55:55 Ok, think I have some answers for what I was looking for. Thanks everyone! Jun 22 04:56:08 BBX15? Let me check that out also... Jun 22 04:56:27 you can find more info on PRU's I/O in the AM335x Technical Reference Manual Jun 22 04:57:07 http://elinux.org/Beagleboard:BeagleBoard-X15 Jun 22 04:58:03 unfortunately it seems to be a bit cursed :/ eternal delay hell Jun 22 04:58:57 Yeah, so it seems. And the price is going to be well above my goal I think. Jun 22 04:59:16 3-5 years things will be different. For now I'm in Pi3 hell. Jun 22 04:59:47 * zmatt is the proud owner of no pi whatsoever :D Jun 22 05:00:43 Anyone here freelance by chance? I have some proof of concept testing I'd like to do and may need some help. Jun 22 05:02:26 zmatt: you should get a Pi-Zero-W just cos. I did. Jun 22 05:02:30 in fact I now have two Jun 22 05:07:30 bcm2835... back to armv6 ? no thanks :) Jun 22 05:08:55 eh w/e :P Jun 22 05:09:05 but yeah its the shitty cpu :/ Jun 22 08:42:26 when will debian 9 be available for BBB Jun 22 10:06:28 I have beaglebone black board. I am getting" error writing to output file - write (28: No space left on device) beaglebone black" Jun 22 12:48:44 have anyone worked with Uniflash for BBB? Jun 22 13:33:58 nextloop: it already has been available for quite some time Jun 22 14:05:35 were - Okay, so that means the drive is full Jun 22 14:32:32 oooh just got a pi myself Jun 22 14:32:48 its probably just going to collect dust tho Jun 22 14:32:51 also got a new bbb Jun 22 14:42:07 o/ got a new bb with wifi. tried to set it up to connect to my home network, but it's configuration is ... special :D Jun 22 14:42:15 any good docs on wifi ? Jun 22 14:43:35 e.g. how do I disable the AP in /etc/default/bb-wl18xx Jun 22 15:36:12 * zmatt is not familiar with the wifi setup on current images unfortunately Jun 22 15:41:07 hi Jun 22 15:42:21 I have problem connecting BBB to PC Jun 22 16:44:45 Hey all, looking for support in writing Linux executables in C/C++ that uses GPIO interrupts on the Beaglebone Black. Windows hosted development. Have Code Composer Studio. Looking for what libraries I need. Is StarterWare wholly sufficient? Jun 22 16:48:01 Looks like StarterWare is only for baremetal. Library suggestions for configuring GPIO interrupts without wiping the distro it shipped with> Jun 22 16:48:06 ack Jun 22 16:48:23 you want latest debian on the BBB Jun 22 16:49:14 are you sure you want to run userspace executables and not drivers? Jun 22 16:50:27 Think so. Need to be able to run a second operation over ssh to deconfigure interrupts. Only active in one mode of operation. Jun 22 16:51:06 there should be decent ways to do that. e.g. through sysfs Jun 22 16:51:45 I'll check it out. Thank you so much for your help! Only done this sort of work on baremetal before. Jun 22 16:52:06 https://www.linux.com/learn/beaglebone-black-how-get-interrupts-through-linux-gpio Jun 22 16:52:24 yeah, it's quite different from that Jun 22 16:52:58 btw, I hope you don't have hard real time constraints? the convenience of a full blown OS comes with not being cycle accurate Jun 22 16:53:11 Actually already had that link open! Wasn't clear on what libraries were being used for that C++ code. Just glib2 Jun 22 16:53:12 ? Jun 22 16:53:14 especially for a random process Jun 22 16:54:21 Not hugely restrictive time contraints. So long as the GPIO catches the edge withing several milliseconds I'm happy. Jun 22 16:55:24 Hello, do I need to change the device tree overlay on the beagle bone blue in order to make the can transceiver to work? Jun 22 16:55:27 that should be fine, yeah. Jun 22 16:55:45 Not expecting rapid-fire events. There's some possibility of unexpected events but I will be manipulating the events according to my own needs. The real-time feel of the ISR is to help separate out erroneous events from real. Jun 22 16:56:00 Sweet thanks! Jun 22 16:56:22 I haven't done IRQ handling in userspace, so I'm not so great at it Jun 22 16:56:41 the bottom line is to once understand how you interact with the sysfs virtual files Jun 22 16:56:54 once you catch that you can implement it in anything Jun 22 16:59:12 https://www.kernel.org/doc/Documentation/gpio/sysfs.txt Jun 22 16:59:38 If the pin can be configured as interrupt-generating interrupt Jun 22 16:59:38 and if it has been configured to generate interrupts (see the Jun 22 16:59:38 description of "edge"), you can poll(2) on that file and Jun 22 16:59:38 poll(2) will return whenever the interrupt was triggered. If Jun 22 16:59:45 that's the relevant bit Jun 22 17:00:07 bumblebee: I guess that should have you all set up? Jun 22 17:00:51 Enough to run with for now, thank you, I really appreciate it! Jun 22 17:01:32 no problem, glad to help Jun 22 17:03:22 knicklux: I believe you can disable that in connman.conf by setting the Tether option to off Jun 22 17:09:27 TetheringTechnologies and PersistentTetheringMode Jun 22 17:10:13 I'm pretty sure all I did was set PersistentTetheringMode to false Jun 22 17:41:25 bumblebee: note that the am335x also has peripherals that can do pulse-counting, or edge-detection with timestamping Jun 22 18:17:42 freemor, already disabled connman. actually you can just change the value in the config of bb-wl* Jun 22 18:18:17 i'm having some trouble still cause the wifi is not available during boot Jun 22 18:18:39 wlan0 shows a few seconds until the last systemd target was reached Jun 22 18:18:55 what might be the cause? where is the driver implemented? Jun 22 18:48:06 Hi All , can i create a sdcard with windows? Jun 22 18:49:48 Yes, you can use etcher Jun 22 18:50:36 CCC12: what is the equivalent of sudo dd if=./u-boot/MLO of=${DISK} count=1 seek=1 bs=128k? Jun 22 18:53:27 cyberguy:I have no idea, I just happened to read about etcher on the bb starting guide. Jun 22 18:53:52 It looks pretty simple to use Jun 22 18:54:15 yeah, but i want to change the uboot, i compiled it on my linuxvm, how do i put it back to the sdcard Jun 22 18:57:23 Use etcher to burn the image to a SD card of yours. Jun 22 19:02:02 yeah i can do this, but i need to change the uboot inside the image Jun 22 19:02:10 how do i do that...? Jun 22 19:02:58 hi.. Jun 22 19:33:36 anyone have a clue on how to write uboot to sdcard with windows? Jun 22 19:37:55 there's guides on the beagle.org site I thought .. or even on elinux .. and about a thousand other places. There's a utility you run, but most are using linux and just 'dd' it :p Jun 22 19:38:08 there*are :( SPaG fail :/ Jun 22 19:50:03 crap, couldn't find anything , i'll think i'll go with cygwin Jun 22 19:50:52 ah with uboot you're probably on your own there I think. You need a rawrite utility Jun 22 19:51:10 I thought there was one .. but I've only ever used it once if so :D Jun 22 19:52:23 cygwin is my best frind Jun 22 19:52:48 cygwin is a reasonable compromise, but I always found it a complete b*** to build packages within it. Jun 22 20:00:53 Does anyone have experience getting socketcan running on begalbone blue? Jun 22 20:49:37 CCC12: how do you flash the BBB ? with uniflash or with sdcard? Jun 23 02:36:11 Hey Guys! Does anybody know if the BBB's io memory is accessable in barebones enviroment without doing any special initalization? Also anybody know where I could find documentation of said memory map layout? Jun 23 02:43:31 Hey Guys! Does anybody know if the BBB's io memory is accessable in barebones environment without doing any special initalization? Also anybody know where I could find documentation of said memory map layout? Jun 23 02:44:13 Herald: 1. yes Jun 23 02:44:26 although doing some basic mmu setup is highly recommended for performance Jun 23 02:44:48 Awesome!!! Jun 23 02:45:41 Thanks! Any idea on the mem map documentation? Jun 23 02:46:05 I have a tiny example in assembly that doesn't bother with the mmu setup, just toggles leds in response to pressing the S2 button (irq-driven): https://github.com/mvduin/bbb-asm-demo Jun 23 02:46:17 Cool! Jun 23 02:46:23 same as everything else you need to know, the AM335x Technical Reference Manual Jun 23 02:47:06 I also have a spreadsheet for the memory maps of the AM335x ("subarctic") and the related DM814x ("centaurus"): https://goo.gl/UHF2Fy Jun 23 02:47:51 Perfect! Thank you once again :) Jun 23 02:49:01 for mmu setup, this may be useful -> https://community.arm.com/processors/b/blog/posts/cortex-a8-translation-table-init-example Jun 23 02:50:09 Another browser favorite! Looks like I am all set and more! Jun 23 02:50:45 this is the start.S for my own C/C++ baremetal code, the only assembly source file I use (the rest just being C++ with small bits of inline asm): https://liktaanjeneus.nl/start.S.html Jun 23 02:51:04 d'oh and he's gone Jun 23 02:51:20 -.- **** ENDING LOGGING AT Fri Jun 23 03:00:04 2017