**** BEGIN LOGGING AT Wed Dec 19 03:00:02 2018 Dec 19 03:12:45 that would be a most definite maybe? Dec 19 03:14:45 Dang! Dec 19 03:16:00 Oh well. I can try something else. Dec 19 03:17:56 Hey GenTooMan: What are you doing w/ your BBB lately? Dec 19 03:47:28 set_ nothing not well and RL has me a bit busier than I would like. All the fun of redoing something from DEMO to functional is often more work than one would like. :D Dec 19 03:47:49 Got it. Dec 19 03:48:11 Oh well again. I guess this is what beams are made of. Yes beams. Dec 19 03:48:32 bbl! Up, up, and Otay! Dec 19 03:48:43 You might consider learning a bit about the PRU and use it for controlling multiple stepper motors simulatneously. Dec 19 03:48:59 Ok need sleep myself. Dec 19 09:49:29 hello? Dec 19 10:23:27 hello microdevel Dec 19 12:52:56 I'm looking to put ROS on a bbgw. Am I best off starting with Debian or Ubuntu ? Dec 19 12:54:32 Seems the bbgw is a bit quirky and the latest Debian works better than the image it was delivered with so I'd rather stay with Debian. But ROS almost -but not quite- installs on that whereas it has a repo for Ubuntu Dec 19 12:57:48 I'm newish to both ROS and beaglebone but have plenty of experience on desktop linux. I chose the bbgw for it's 4 USB ports (vs the bbb) and general reliability of bb over the pi (though I'm not sure if that's true of the bbgw) Dec 19 13:32:48 artag: the bbgw kinda sucks though because a whole bunch of its expansion i/o pins are occupied for the wifi/bluetooth chipset (completely unnecessarily, the beaglebone black wireless does not have this problem) Dec 19 13:33:30 I think debian also packages ros though? Dec 19 13:33:38 I wouldn't use ubuntu Dec 19 13:35:03 what do you mean by "ROS almost -but not quite- installs" ? Dec 19 13:40:25 yeah, I discovered late that the bbgw is a bit weird. But I don't actually need too much of the onboard stuff so it may be OK. Dec 19 13:40:27 if debian's ros packages fail to install, that would probably warrant a bug report to the package maintainer Dec 19 13:41:56 i haven't actually tried to install from debian's repos, the instructions said to use the ROS ones, but if there are some i'll try them. It's just that the ROS repos don't include armhf for debian - oinly for ubuntu Dec 19 13:42:24 so i tried from sources and started to trip over things Dec 19 13:42:32 debian packages ros for armhf Dec 19 13:42:37 i wasn't sure whether to persever with that or try ubuntu Dec 19 13:42:41 see apt-cache search -n '^ros-' Dec 19 13:43:04 will do. maybe that's why it's not in rtos repos. thanks Dec 19 13:43:15 s/rtos/ros/ Dec 19 20:28:11 Hello! How can I find out if pwmchip2 is available for use? I was helping this person but this person wants to know about this specific pwmchip. Dec 19 20:28:13 ... Dec 19 20:28:45 I looked on my board under /sys/class/pwm/ but chip2 was not available. Dec 19 20:33:04 is pwmchip2 taken up by one of these spidev1.0 - 2.1 muxed pins or do I need to just test the config-pin utility to find out if it is possible to make one of my pwmchips (pwmchip2) available? Dec 19 20:42:28 I also see that there are only seven pwmchips available instead of eight like the photo describes. Is it a fact that there are eight? Dec 19 20:48:47 on bbb.io/introduction, it states eight but I only found seven and on my BBBW is where I found that I only have six available. Dec 19 20:48:48 ... Dec 19 20:48:54 Does this sound correct? Dec 19 20:49:30 bbl...if you have time, please try to clarify my lack of shtick. Oh yea! Dec 19 21:16:27 what peripheral are you actually trying to use? "pwmchip2" doesn't mean anything to be, afaik the kernel just sequentially numbers pwm peripherals as it encounters them. if you have the latest version of the bb-customizations package then you'll find more useful symlinks in /dev/pwm/ Dec 19 21:16:35 *mean anything to me Dec 19 21:17:41 Okay. Thank you. Dec 19 21:18:11 I will check later on in the day. /dev/pwm/ Dec 19 21:19:31 also where do you get "eight pwmchips" .. there are either five or nine, depending on whether timer4-7 are also used for pwm (I'm not sure whether they're setup for that by default) Dec 19 21:19:52 Oh. Hold. I will get the link. Dec 19 21:20:31 https://beagleboard.org/Support/bone101 and it is further down the page under PWM and Timers. Dec 19 21:20:52 3 ehrpwm peripherals and 2 ecap peripherals (ecap0 and ecap2, since ecap1's output is not on the expansion headers) Dec 19 21:21:15 I see nothing there about pwmchips.. (note: pwmchip is not the same thing as pwm output) Dec 19 21:21:24 Oh. Dec 19 21:21:42 I got you now. the chip avaible is not the same as a pin. Okay. Dec 19 21:22:03 the term "pwmchip" is a bit awkward too since of course they're not actual chips but just peripherals Dec 19 21:22:20 the ehrpwm peripherals have two pwm outputs each Dec 19 21:22:31 Okay. Dec 19 21:23:06 which yields 8 outputs (when excluding the four timer pins): 6 ehrpwm + 2 ecap Dec 19 21:23:20 Two pins per "chip" which is actually a ehrpwm peripheral. Dec 19 21:23:22 Aw! Dec 19 21:23:25 Okay. Dec 19 21:23:30 correct Dec 19 21:24:02 ecap2 does seem to conflict with hdmi audio, so that will need to be disabled to be able to use ecap2 Dec 19 21:25:18 Okay. zmatt: Can I capture this screen to teach this person about what it is exactly that I misunderstood? Dec 19 21:25:25 I can leave out the name parts if necessary. Dec 19 21:25:35 irc logs are public Dec 19 21:25:39 Okay. Dec 19 21:26:15 So...dang. I will start my BBBW now. I am going to look into /dev/mem/ quickly. Dec 19 21:26:34 sorry. Dec 19 21:26:38 /dev/pwm Dec 19 21:26:48 Yea. Dec 19 21:28:58 the udev rules that create those symlinks are in version 1.20181120 of bb-customizations Dec 19 21:29:06 (which is the latest version) Dec 19 21:29:48 Okay. I just updated my BBBW. So, they would have been installed if I did not already have them, right? Dec 19 21:30:48 correct Dec 19 21:31:32 Okay. I had this fellow, I guess it is a fellow, update his board w/ bbb.io/latest-images. Dec 19 21:31:55 Once that person updates and upgrades, that person should have them too. This is good news. I was trying to get on the same page w/ this fellow. Dec 19 21:32:04 the latest image is 2018-10-07 and therefore can't include the latest version of this package Dec 19 21:32:16 Oh. Dec 19 21:32:21 so it'll need to be updated Dec 19 21:32:25 Aw! Dec 19 21:32:40 w/ apt update && apt upgrade? Dec 19 21:33:00 note that of course it's not required to be able to use pwm outputs, people have lived without it for a long time... I just have no idea how they ensured they used the right one :P Dec 19 21:33:14 Blah! Dec 19 21:33:26 while the symlinks make it really easy Dec 19 21:33:32 Oh...got it. Dec 19 21:33:42 I will let this fellow know. Dec 19 21:39:57 the important bit to be able to use all 8 pwm outputs is simply disabling hdmi audio in /boot/uEnv.txt Dec 19 21:40:34 Right...that is what I just told this fellow. Dec 19 21:40:55 I am sending in the info. now on the beagleboard.org groups section. Dec 19 21:41:02 in principle there could be four more pwm outputs (timer4-7). I'm actually not sure why those aren't set up by default Dec 19 21:41:31 That confused me at first. But, they were highlighted. I just figured I could set them up (if necessary). Dec 19 21:42:14 also... Dec 19 21:42:55 When you stated that those "symlinks" at /dev/pwm/ were available, are there supposed some type of info. listed in the files? Dec 19 21:43:33 sorry. *supposed to be some type of info. listed in the files? Dec 19 21:43:40 what do you mean? Dec 19 21:44:21 I think I am not understanding you right now. You stated symlink. In those symlink files, am I supposed to be able to see any characters? Dec 19 21:44:45 Or are they just empty files w/ hard or soft symlinks? Dec 19 21:45:20 what you're saying/asking is so confused I'm not even sure how to respond Dec 19 21:45:47 Okay. Please hold on b/c you are correct. confused, I am. Dec 19 21:45:59 I need to be able to state this in simpler terms. Dec 19 21:46:38 forget it. I am mind numbing you w/ my lack of knowledge. Dec 19 21:46:39 ... Dec 19 21:46:50 I cannot describe what it is that i do not know as of now. Dec 19 21:46:54 that directory will contain one symlink per pwm output, e.g.: https://pastebin.com/raw/rB5if3v5 (the "ecap1" symlink would be missing on a beaglebone, as would the "ecap2" symlink if hdmi audio is enabled) Dec 19 21:46:55 sorry. Dec 19 21:47:19 each symlink points to the pwm output's directory in sysfs, which contains the attributes through which the pwm output is controlled Dec 19 21:47:30 I got you now. Dec 19 21:47:52 I figured I could write those files and use my favorite editor to change things w/in those files. My bad. Dec 19 21:48:12 using a text editor in sysfs is generally a bad idea Dec 19 21:48:12 ls -l shows me the links like you stated. Dec 19 21:48:24 Okay. noted. Dec 19 21:50:33 Sorry for that extraordinary repsonse and quetioning. Dec 19 21:50:44 *questioning. Dec 19 21:50:46 Yikes. Dec 19 23:23:27 zmatt: recall I was asking about configuring the RAM of a beagleboard-x15 to 4GB. You suggested I use "mem" kernel parameters. The changes to the kernel device tree code to accomplish the same is noted in https://pastebin.com/6Lq15pD7 Dec 19 23:24:09 I suggested the "mem" kernel parameter could be used as temporary workaround until you found the correct way to pass the relevant info Dec 19 23:25:32 normally u-boot injects this information into the DT because u-boot already needs to know the ram configuration before it can possibly load the DT, so putting the information in the DT as stored on disk would be redundant Dec 19 23:25:33 yes, it took a little bit of digging to figure it out. Dec 19 23:26:11 so doing it as shown in your pastebin makes no sense to me Dec 19 23:26:36 u-boot appears to only see 2GB. Dec 19 23:26:42 but it works. Dec 19 23:28:51 if u-boot still has the wrong idea about the amount of ram, but you override its knowledge of the memory configuration using manually written DT entries, then that feels as much like a workaround as using the "mem" kernel parameter does Dec 19 23:29:11 of course saying "hey, it works" is valid too Dec 19 23:32:54 Now that I know what to look for, I can test this in the u-boot code. The relevant file is am57xx-beagle-x15-common.dtsi. Dec 19 23:33:14 https://elixir.bootlin.com/u-boot/latest/source/arch/arm/lib/bootm-fdt.c#L34 Dec 19 23:33:28 this is the u-boot code that writes information into the /memory DT node Dec 19 23:34:31 it looks like any existing stuff in /memory will be overwritten by u-boot Dec 19 23:34:55 but I guess it won't delete the second node if it thinks there's only one memory region Dec 19 23:36:16 oh actually it looks like it will never create more than one node Dec 19 23:36:55 it just puts all banks into the reg property of one node Dec 19 23:39:02 When i specified the extra ram in the single node of the kernel DT, it was ignored. The second node worked. Dec 19 23:39:17 that's because u-boot was overwriting it Dec 19 23:39:24 with its knowledge of the memory layout Dec 19 23:39:42 I'm reverting my change to the kernel and re-compiling u-boot with modified memory node. it should take a few minutes. Dec 19 23:56:44 It didn't work. I tried appending the extra range in the memory@0 node, and also tried creating a 2nd memory@1 node. U-boot reports 2GB when it boots and the kernel only sees 2GB after it boots. Dec 20 00:03:23 what did you change in u-boot to fix its understanding of the memory present? Dec 20 00:03:51 it looks like CONFIG_NR_DRAM_BANKS needs to be 2, and the ranges themselves need to be configured somewhere Dec 20 00:04:44 re-load https://pastebin.com/6Lq15pD7 Dec 20 00:06:43 u-boot doesn't read or interpret the dtsi, it *overwrites* the memory information you put there with its own understanding of the memory layout Dec 20 00:08:33 that's what fdt_fixup_memory_banks() in u-boot is doing Dec 20 00:24:48 Oh, I had it backwards. I just broke the partition of the uSD I was using. I'll start over tomorrow. Thank you for your help. Dec 20 02:13:54 are the gfx drivers opensource? do they support gles2/3 and egl? Dec 20 02:18:08 georgios: the kernel driver is open source, the userspace libs aren't. egl and gles1/2 supported, gles3 isn't. getting the libs to work is also somewhat nontrivial currently (they're not shipped by default) Dec 20 02:18:40 the libs also do not support X11, only fullscreen (drm/gbm) and wayland Dec 20 02:18:57 that is something Dec 20 02:20:19 also, while the kernel driver is open source, it isn't mainline (nor is there any hope it could be mainlined in its current form. as is not uncommon for "vendor" drivers, merely looking at the code makes you want to gauge your eyes out) Dec 20 02:21:36 would an embedded gfx subsystem be welcome? Dec 20 02:21:44 what do you mean? Dec 20 02:22:29 note btw that the gpu and the display controller are completely separate things. the latter has a mainline driver, and is what's used for actually getting video output. the gpu is solely useful for 3d rendering Dec 20 02:22:32 porting a graphic library Dec 20 02:25:24 what is the current status if rpi and graphics? Dec 20 02:26:16 you might have better luck with that question in an rpi channel... I don't have any myself, nor track its development Dec 20 02:27:40 the rpi is a complicated story anyway, with its "gpu" (the videocore processor) actually being the boot cpu Dec 20 02:32:22 =O Dec 20 02:32:40 i am looking at banana pi Dec 20 02:32:48 mali 420 gpu Dec 20 02:33:17 specfiles needed Dec 20 02:37:11 well,imo even closed source drivers are welcome at this point Dec 20 02:38:58 i am interested in graphics for minix3 basicaly, and the arm port of it is developed on a beaglebone Dec 20 02:41:45 is beagle suitable for this? Dec 20 02:42:38 ehm, that's a really vague question Dec 20 02:43:37 the beaglebone makes sense as a target for an arm port of another OS since the hardware is open, the SoC is very well documented and easy to initialize Dec 20 02:44:22 is the gpu documented? are driver blobs avaiable Dec 20 02:44:29 sorry i am too tired Dec 20 02:44:43 the hardest part about getting basic graphics (framebuffer output) working would be configuring the poorly documented HDMI framer (from NXP) working Dec 20 02:45:29 i saw that only one board has a fb Dec 20 02:46:22 if it doesn't have a working framebuffer yet on the beaglebone then that would presumably be your first priority, not the gpu Dec 20 02:46:55 ok Dec 20 02:47:04 unless you're content to render 3d graphics into a piece of memory without being able to display it :P Dec 20 02:47:30 well thanks for the valuable information Dec 20 02:47:38 regardless, getting the gpu working would be a huge and complex project Dec 20 02:47:43 i must now go to bed Dec 20 02:47:58 good night **** ENDING LOGGING AT Thu Dec 20 03:00:01 2018