**** BEGIN LOGGING AT Sat Oct 06 03:00:01 2018
Oct 06 03:22:25 <zmatt> why do you think that?
Oct 06 03:22:59 <zmatt> if you're uncertain about the modes, double-check the pin configuration using my show-pins utility: https://github.com/mvduin/bbb-pin-utils/#show-pins
Oct 06 03:24:54 <CoffeeBreakfast> I know your repos :D, and it says P9.42 104 Mode 4 (should be 6)
Oct 06 03:25:57 <CoffeeBreakfast> and P9.42 89 Mode 7 (should be 3)
Oct 06 03:27:02 <zmatt> my guess would be the latter is due to conflict with cape-universal.  you'll need to add   &ocp { P9_92_pinmux { status = "disabled";  }; };
Oct 06 03:27:39 <zmatt> (the P9_92 vs P9_42 is cape-universal's weird way of doing what I did with P9_42a vs P9_42b)
Oct 06 03:27:56 <zmatt> you should have seen an error about the conflict in the kernel log
Oct 06 03:29:08 <CoffeeBreakfast> dmesg: no complaints, and I'll try with P9_92
Oct 06 03:29:27 <zmatt> odd
Oct 06 03:35:12 <CoffeeBreakfast> no luck with P9_92_pinmux
Oct 06 03:37:13 <CoffeeBreakfast> the &pruss node is ok?
Oct 06 03:37:34 <zmatt> looks fine to me
Oct 06 03:38:08 <zmatt> (presumably the status = "okay"; is redundant since it's already being enabled by the uio-pruss overlay)
Oct 06 03:38:32 <zmatt> what exactly it show-pins saying for those two pins?
Oct 06 03:38:34 <zmatt> *is
Oct 06 03:39:17 <CoffeeBreakfast> P9.42                             89 fast rx down 7 gpio 0.07
Oct 06 03:39:28 <CoffeeBreakfast> P9.42                            104 fast rx down 4 mmc 0 wp
Oct 06 03:41:45 <CoffeeBreakfast> any key for looking into dmesg?
Oct 06 03:42:11 <zmatt> wait that's all they're showing?  so your pinmux is not being activated at all
Oct 06 03:42:38 <zmatt> and the mode 4 is just a mistake done by u-boot
Oct 06 03:42:40 <zmatt> most likely
Oct 06 03:42:50 <CoffeeBreakfast> you mean the entire output of show-pins=
Oct 06 03:42:53 <CoffeeBreakfast> ?
Oct 06 03:43:14 <zmatt> if the pins were claimed by anything in the kernel then there's be more at the end of each line
Oct 06 03:43:29 <zmatt> this is suggesting nothing in the kernel has configured these pins
Oct 06 03:43:33 <CoffeeBreakfast> oh
Oct 06 03:44:03 <CoffeeBreakfast> I see
Oct 06 03:45:16 <zmatt> maybe u-boot is attempting to apply your overlay *before* the uio-pruss overlay, or something like that
Oct 06 03:53:24 <CoffeeBreakfast> https://pastebin.com/K9RfxR4z
Oct 06 03:53:36 <CoffeeBreakfast> (u-boot)
Oct 06 04:02:47 <CoffeeBreakfast> yes, before
Oct 06 04:13:35 <CoffeeBreakfast> how you know?
Oct 06 08:31:49 <ayjay_t> huh i'm looking at using libinput as a building block for this but i just don't understand why everything is polling
Oct 06 13:23:49 <CoffeeBreakfast> The spin locks of the am335x are all used by the kernel?
Oct 06 13:28:03 <zmatt> you're referring to the spinlock peripheral of the SoC ?  it's completely unused
Oct 06 13:28:58 <CoffeeBreakfast> Yes, the peripheral.
Oct 06 13:33:40 <zmatt> not sure when you'd ever want to use it though
Oct 06 13:35:08 <CoffeeBreakfast> Could be useful in a circular-buffer between the ARM and the PRU?
Oct 06 13:35:26 <zmatt> why would you need a spinlock for that? o.O
Oct 06 13:35:44 <zmatt> ringbuffers are notable for not requiring any sort of lock (if written correctly)
Oct 06 13:36:27 <CoffeeBreakfast> I'm just reading the Doc for linux/circ_buf.h
Oct 06 13:37:37 <zmatt> see e.g. the stream.py example of py-uio
Oct 06 13:40:45 <zmatt> (for irq-based wakeup, just have the sender signal an irq whenever it posts the first message in a previously empty queue)
Oct 06 13:45:47 <zmatt> just make sure the write-pointer that's written by the sender is located in the same memory as the data itself, e.g. in stream.py both are in ddr memory
Oct 06 13:52:14 <CoffeeBreakfast> I'll take into account
Oct 06 13:53:54 <CoffeeBreakfast> Apart from that, in which case are the spinlock peripheral useful? have you used it before?
Oct 06 13:54:52 <zmatt> I haven't used it (other than a quick test to see how it works), I have never seen it be used for it, and I can't think of any use for it
Oct 06 13:55:08 <zmatt> *seen it be used for anything
Oct 06 13:55:38 <zmatt> I feel like in any situation you might want to use it, there are probably better solutions
Oct 06 15:20:26 <sanderz> Hi, I have jsut bought BBB and have tried to conenct to onboard server via 192.168.7.2. This cannot find server .... have tried using Windows10 and Ubuntu VM
Oct 06 15:20:40 <sanderz> Andy thoughts?
Oct 06 15:34:38 <tbr> sanderz: well, are the LEDs on and blinking?
Oct 06 15:35:26 <tbr> regardless it's a good idea to reflash it with the latest image available: http://beagleboard.org/latest-images
Oct 06 18:05:18 <ayjay_t> this is a lot
Oct 06 18:11:29 <CoffeeBreakfast> ayjay_t: Maybe you told me something in faceb.
Oct 06 19:20:34 <thinkfat_> zmatt: write ordering can be quite tricky with lock-free ringbuffers
Oct 06 19:21:24 <thinkfat_> zmatt: depending on the pou for the different observers..
Oct 06 19:21:29 <thinkfat_> PoU
Oct 06 19:23:35 <zmatt> for non-cacheable memory, all you need is that the write-pointer is located in the same target memory as the data, and if needed for the cpu a suitable memory barrier is used in the appropriate places
Oct 06 19:24:22 <zmatt> if you put the pointer and the data in different places then things get a lot more... interesting
Oct 06 19:27:10 <zmatt> I very much doubt a memory barrier is needed on the cortex-a8 though, since there's a control-dependency between reading the write-pointer and reading the data.  a compiler barrier should suffice.
Oct 06 19:28:16 <zmatt> if neon loads are used to read the data (or portability is needed), a memory barrier may be needed between it and updating the read-pointer.  otherwise the in-order architecture of the cortex-a8 again means a compiler barrier should suffice
Oct 06 19:39:52 <thinkfat> I don't know what glibc does with memcpy, I don't think it uses neon
Oct 06 19:39:59 <thinkfat> I know the kernel memcpy uses neon
Oct 06 19:40:08 <zmatt> pretty sure it does yes
Oct 06 19:40:09 <thinkfat> ah, no, doesn't
Oct 06 19:40:16 <zmatt> in userspace I mean
Oct 06 19:40:20 <zmatt> dunno about kernel
Oct 06 19:40:27 <thinkfat> kernel probably not, likely not
Oct 06 19:40:49 <zmatt> any sort of neon in the kernel is a bit of headache
Oct 06 19:40:59 <thinkfat> neon has lots of registers, if you allow kernel to clobber it ...
Oct 06 19:41:59 <zmatt> yeah you need to switch out the neon registers of the current userspace task (or more precisely the last userspace thread that used neon)
Oct 06 20:06:58 <thinkfat> tbr: https://www.aliexpress.com/item/110V-220V-700W-858D-ESD-Soldering-Station-LED-Digital-Solder-Iron-desoldering-station-BGA-Rework-Solder/32528750078.html?spm=2114.10010108.1000010.1.5b6d60a2tKEQS4&gps-id=pcDetailLeftTrendProduct&scm=1007.13438.100207.0&scm_id=1007.13438.100207.0&scm-url=1007.13438.100207.0&pvid=93b4fff0-90a7-4289-b343-ac5148a1b625
Oct 06 20:07:15 <thinkfat> tbr: didn't I just pay 50€ for basically the same thing?
Oct 06 20:07:53 <ayjay_t> CoffeeBreakfast: hmm?
Oct 06 20:08:00 <ayjay_t> no i just mean, there's always issues with the ethernet over usb
Oct 06 20:11:07 <CoffeeBreakfast> ayjay_t: I think you showed me py-uio in a fb group (Beaglebone Black) when I asked a tricky question there
Oct 06 20:11:07 <zmatt> I think it's always worked reliably for me, but I don't use it often
Oct 06 20:11:31 <zmatt> (usb ethernet)
Oct 06 20:14:17 <ayjay_t> ohhh CoffeeBreakfast i'm the admin there now lol
Oct 06 20:14:38 <CoffeeBreakfast> I noticed that :P
Oct 06 20:14:44 <ayjay_t> did i send you here too?
Oct 06 20:15:30 <ayjay_t> i've been arguing with facebook about a post i just made, a picture of two beaglebone blacks. zuckerburg says its spam, i say it is, in fact, not spam. it is #beagle related. i wonder if people can still see it?
Oct 06 20:15:38 <ayjay_t> err two pocketbeagles
Oct 06 20:17:14 <CoffeeBreakfast> the link works, but no miniatures in the post
Oct 06 20:18:26 <ayjay_t> :-O facebook is so annoying
Oct 06 20:19:13 <ayjay_t> i'm hoping people post pictures of their projects. i always like beaglebone pictures
Oct 06 20:21:24 <CoffeeBreakfast> That Hakko brings me confidence
Oct 06 20:22:19 <ayjay_t> i still use it, although my lab has upgraded from a college dorm room
Oct 06 20:22:22 <ayjay_t> err college apartment
Oct 06 20:22:32 <ayjay_t> i actually liked my chinese soldering station more for a while but it eventually broke
Oct 06 20:25:40 <thinkfat> ayjay_t: https://plus.google.com/+MatthiasWelwarsky/posts/5qkhvSfFs3q
Oct 06 20:25:51 <thinkfat> ayjay_t: beaglebone wireless project picture for you ;)
Oct 06 20:41:06 <zmatt> picture of a bbb while I was doing pmic-related tests quite a while ago... https://liktaanjeneus.nl/bbb-pmic-testing.jpg  It kinda looks like it's on an intensive-care unit or something ;)
Oct 06 20:51:22 <CoffeeBreakfast> poor little BBB
Oct 06 20:54:02 <ayjay_t> those are pretty interesting
Oct 06 21:04:42 <thinkfat> ayjay_t: mine is pretty simple though. basically a shift register driven by some gpios, which in turn drives some logic-level triacs
Oct 06 21:05:33 <thinkfat> ayjay_t: what I spent most time on was the switch-mode regulator that powers the board...
Oct 06 21:17:41 <thinkfat> ayjay_t: and this is where it is now: https://plus.google.com/photos/photo/114942752636092445410/6602596261190274418
Oct 06 22:50:44 <CoffeeBreakfast> Somebody can explain the option -b of pasm?
Oct 06 22:51:20 <zmatt> it selects binary output format
Oct 06 22:53:18 <zmatt> if you just run pasm without any arguments you get usage info
Oct 06 22:53:51 <CoffeeBreakfast> because my pru code doesn't work without that flag
Oct 06 22:54:17 <zmatt> ehm, without that flag it doesn't produce a .bin output file
Oct 06 22:54:49 <zmatt> you should pass the appropriate flag for the output format you want
Oct 06 22:55:04 <CoffeeBreakfast> what is the output without that flag?
Oct 06 22:55:23 <zmatt> no idea
Oct 06 22:55:43 <CoffeeBreakfast> besides the fact that I'm a moron...
Oct 06 22:55:49 <zmatt> Note: Using default output '-c'
Oct 06 22:55:52 <zmatt> is what pasm says
Oct 06 22:59:54 <CoffeeBreakfast> by default, makes a C include file...
Oct 06 23:02:20 <CoffeeBreakfast> The next program I'm gonna do, will be a mandelbrot plotter... and the default output file will be in MIDI format, with a lot of drums
Oct 06 23:02:34 <zmatt> sounds good
Oct 06 23:03:01 <CoffeeBreakfast> lol
Oct 06 23:06:24 <zmatt> I am still curious how crazy-deep mandelbrot zooms (like https://youtu.be/dSA7OZHdaoA ) work... like, how they avoid numerical instability, and how they locate an embedded mandelbrot figure at the appropriate zoom level to end the video at (as seems to be tradition among deep mandelbrot zooms)
Oct 06 23:14:00 <CoffeeBreakfast> good question
Oct 06 23:22:09 <zmatt> unfortunately I don't actually care *quite* enough to really dig into it
Oct 06 23:30:39 <zmatt> although I guess it helps that "Every neighborhood of a point on the boundary of the Mandelbrot set contains infinitely many embedded copies of the Mandelbrot Set."
**** ENDING LOGGING AT Sun Oct 07 02:59:59 2018