**** BEGIN LOGGING AT Tue Nov 20 02:59:58 2018 Nov 20 03:31:10 Crofton: he has a point but too soon. The real fix is here https://github.com/meta-qt5/meta-qt5/pull/127 Nov 20 08:53:17 hi. could somebody help me understand how do gpio pins within linux correspond to the actual gpio pins on my module? i have gpiochip0, gpiochip32 [...] gpiochip128 in /sys/class/gpio/, so sysfs gpio support seems to be there Nov 20 08:54:24 and there's a comprehensive hardware manual for the module, i just can't understand how do i transform something like "GPIO1_3" to linux gpio number Nov 20 08:57:33 waterscale: i think this is a pretty neat explanation: https://xilinx-wiki.atlassian.net/wiki/spaces/A/pages/18842398/Linux+GPIO+Driver#LinuxGPIODriver-UsingtheGPIODriverfromaUserSpaceApplication Nov 20 08:58:37 ah no sorry, wrong paragraph. see https://xilinx-wiki.atlassian.net/wiki/spaces/A/pages/18842398/Linux+GPIO+Driver#LinuxGPIODriver-UsingGPIOwithSysFs Nov 20 08:59:34 also see the linked https://lkml.org/lkml/2014/7/7/390 Nov 20 08:59:47 in a nutshell: it depends a bit on the specific SoC in use Nov 20 09:00:46 LetoThe2nd: "it depends" was the answer i feared most lol. thanks for the links, gonna try those Nov 20 09:01:49 waterscale: i can only tell you (IIRC) what applies for my HW (atmel sama5d3): its gpio bank * 32 + pin number, so for gpio1_3 it would be "3". echoing that into export does the trick Nov 20 09:02:59 that's the bruteforce approach i'm keeping for later :) Nov 20 09:03:33 why? Nov 20 09:04:36 it's not urgent, i'd rather try to understand the underlying concept first, and then after i fail in a couple hours i'll resolve to trying to guess the correct gpio Nov 20 09:04:46 it's a neverending cycle Nov 20 09:05:59 no, i mean, its not a bruteforce approach. its the usual process to use a gpio. you echo something into export to get access to the pin Nov 20 09:14:27 that much i understand, i just don't get how it actually maps within the system yet. gpiochip{0,32,64,96,128}/label gives me something that's probably directly out of the device tree, stuff like "209c000.gpio" Nov 20 09:15:12 waterscale: if you look into the gpiochip directories, then at label, you pretty much get exactly the mmio address. Nov 20 10:51:18 koen, hi Nov 20 10:51:28 I asked long ago about meta-kodi Nov 20 10:52:05 do you know about issues with mips in the last kodi versions? Nov 20 11:12:15 btw I see p8platform: update to 2.1.0.1 in meta-oe Nov 20 12:38:50 ant_work: haven't tried it on mips, only aarch64 and x86_64 Nov 20 13:43:34 koen, ok, I will just try with qemumips Nov 20 13:44:40 it is for the STB vuplus duo2, I think there might be issues with the gles driver Nov 20 13:45:21 that crazy dutch people at OpenPLi have an unfinished 17.6 Nov 20 13:45:59 (they are now rebasing to rocko) Nov 20 18:40:37 khem: I've fixed known_hosts on abaco to rsync to milla, will see after the build is finished if it works now Nov 20 18:40:45 halstead: ^ Nov 20 18:42:41 JaMa, Sounds good. I was reading over that thread earlier today. This is the next logical fix. Nov 20 18:43:05 halstead: what was the other fix? Nov 20 18:45:59 JaMa: thx Nov 20 18:52:22 khem: are you using rpi3 for something serious? I have seen a lot of interesting patches from you to meta-raspberrypi recently Nov 20 19:20:00 JaMa: yes Nov 20 19:21:05 This is default reference for RDK now Nov 20 19:30:13 khem: aarch64 as well were you using vc4graphics for arm? Nov 20 19:31:42 khem: I'm asking because with webOS OSE we have strange combination, 32bit arm + vc4graphics + userland (without egl, removed with patch, but otherwise similar to your new userland-nogl) for multimedia Nov 20 19:32:53 khem: this combo doesn't work well with newer kernels (so I'm tempted to drop vc4graphics completely, which leads to some issues with qtbase) and for aarch64 our media guys didn't figure out how to replace userland binaries at all Nov 20 19:33:34 JaMa: vc4graphics is the way to go and not drop it Nov 20 19:34:08 eventually we want everything to work well with mesa/vc4graphics Nov 20 19:34:53 we are optimizing RDK wayland compositor (westeros) to work well with this combo Nov 20 19:35:13 so far gstreamer-omx is helpting with omx path Nov 20 19:35:41 eventually when we have v4l2 for rpi then it would be great we can hopefully drop omx completely Nov 20 19:36:08 I want to merge userland-nogl and userland recipes into 1 as well Nov 20 19:36:21 eventually yes, but fkms doesn't seem to be ready for prime time and vc4-kms-v3d isn't meant to play nice together with userland for multimedia Nov 20 19:36:26 and operate based on vc4graphics DISTRO_FEATURE Nov 20 19:36:57 fkms is a stepping stone Nov 20 19:38:20 https://github.com/Comcast/rdk-on-raspberrypi says to BBMASK whole recipes-qt, is there something public which already shows how you're using Qt in RDK with rpi? Nov 20 19:38:47 I've seen that qtbase bbappend in meta-raspberrypi, just wondering about the other bits Nov 20 19:43:42 JaMa: yes, that is best version Nov 20 19:43:47 we dont use qt Nov 20 19:44:17 but wayland+westros+spark+wpe Nov 20 19:44:28 spark is another native rendering engine **** ENDING LOGGING AT Wed Nov 21 02:59:58 2018