**** BEGIN LOGGING AT Thu Dec 31 02:59:57 2020 Dec 31 03:26:05 zmatt: I have set up a clean image and installed ti-sgx-ti335x-modules-4.19.94-ti-r42 and then the sgx-ti335x-userspace-1.14.3699939_1.1.deb, sgx-libgbm.deb and sgx-libgbm-dev.deb files using apt, dependencies were pulled in including libwayland-server0. I didn't see any errors, however, neither /usr/bin/pvrsrvctl nor /etc/init.d/rc.pvr exist, so if your driver works differently from the versions i've tried before, i'm not Dec 31 03:26:06 sure how to initialize it. Dec 31 03:32:39 is there a straightforward way you would recommend to attempt to test a demo like kmscube or glxgears with it? kmscube requires libdrm, and since there are a large number of libdrm versions in the rcn-ee.com repository, i'm unsure if i need a specific one for this driver or board. Dec 31 03:43:42 i tried kmscube with libdrm-dev since it pulls in most of the others and unfortunately i see PVR:(Error): Couldn't load WS module libpvrws_WAYLAND.so [0, ] both with -d /dev/dri/card0 and card1. in the previous debian installation, i attempted to get things to work by installing various different versions of packages and messing with the library files the way Rene said he did, but i don't think any of it helped. Dec 31 03:47:38 besides the library error, I actually see the exact same errors after that I did in the old installation (PVR:(Error): PVRSRVBridgeCall: Failed to access device... etc) Dec 31 04:01:09 i would guess upstream kmscube isn't compatible with anything here, since it wants gbm version 13.0 instead of 1.0.0 Dec 31 04:02:54 kmscube from debian repos simply tries to pull in mesa, so i don't think that's acceptable Dec 31 04:30:16 to make sure i know how to use kmscube to begin with i tried it on x86, works perfectly, excellent performance, correctly accesses GPU, even with dozens of mesa and X11 packages installed, all there was to it was closing my display manager and running it Dec 31 05:10:26 misternumberone: upstream kmscube indeed isn't compatible Dec 31 05:10:44 misternumberone: it should get initialized automatically using the udev rule Dec 31 05:10:59 oh! you might need to update initramfs .. I probably forgot the trigger for that since I don't use initramfs myself Dec 31 05:11:06 sudo update-initramfs -u Dec 31 05:11:52 misternumberone: try with this kmscube: https://github.com/mvduin/kmscube Dec 31 05:12:33 that's the one I tested with back then Dec 31 05:13:43 this was the main patch needed to make it compatible: https://github.com/mvduin/kmscube/commit/0c3357d541c4593e8852611f977f24cbeff8ba2c ... why does it have a hardcoded list of those? no idea :P Dec 31 05:14:00 and the pixel format Dec 31 05:15:03 I added support for different pixel formats, and then because I too was too lazy to add the appropriate autodetection of pixelformats supported by the device, I changed the hardcoded format to the RGB565 :P Dec 31 05:15:38 thanks that is very helpful, i'm trying those things now and will reboot Dec 31 05:16:41 so, I see my package installs /etc/systemd/system/sgx-init.service Dec 31 05:17:01 that's the service activated by the udev rule (/etc/udev/rules.d/sgx.rules) Dec 31 05:17:16 oh! and my package includes a kmscube! Dec 31 05:17:23 apparently Dec 31 05:26:02 (you can get the list of files from a .deb using dpkg-deb --contents) Dec 31 05:26:13 (may want to pipe the output through sort) Dec 31 05:36:35 ran # update-initramfs -u && reboot, output of # kmscube from your package (in path, /usr/bin/kmscube): https://termbin.com/c5jlr output of # ./kmscube from https://github.com/mvduin/kmscube after ./autogen && make: https://termbin.com/q5at Dec 31 05:37:12 sorry it takes so long to respond generally i have to wait for the beaglebone to load before every task Dec 31 05:37:52 okay weird, hmm Dec 31 05:38:33 systemctl status sgx-init Dec 31 05:39:12 thankfully i already started trying that one, https://termbin.com/e6uq Dec 31 05:39:38 says enabled, but unsure if it's doing what it should Dec 31 05:40:07 no, it didn't run.. try starting it manually Dec 31 05:40:48 maybe I misremember it being run automatically by the udev rule, lemme actually look at the service file and udev rule file Dec 31 05:42:18 no, I don't misremember Dec 31 05:42:32 no I did misremember Dec 31 05:43:12 job failed error code etc. output journalctl here :https://pastebin.com/raw/2PF0r3Zx Dec 31 05:43:18 so yeah I did design it to be pulled in as dependency by whatever needs 3d graphics... Dec 31 05:44:07 journalctl -x is kinda useless verbosity ;-) Dec 31 05:44:44 anything in kernel log as a result of trying to start sgx-init ? Dec 31 05:47:25 ./var/log/messages: https://termbin.com/1pukk last few lines: https://pastebin.com/raw/REAFWC1r Dec 31 05:47:46 memory allocation failure Dec 31 05:48:07 (specifically failure to allocate contiguous physical memory) Dec 31 05:48:12 though Dec 31 05:48:15 these aren't errors Dec 31 05:48:20 they're just warnings Dec 31 05:48:26 or more like debug messages :P Dec 31 05:48:46 oh, nice tracebacks, lol Dec 31 05:49:05 your wifi driver crashed? Dec 31 05:49:57 okay so no kernel messages as a result of trying to initialize, hmm Dec 31 05:50:56 ummm i may need to work on the wifi driver, for some reason it won't initialize right until a couple minutes after boot Dec 31 05:50:59 could very well be version incompatibility Dec 31 05:51:07 the sgx thing Dec 31 05:52:53 i've been thinking I should try to get the newest possible versions of the PVR SGX SDK for AM335x from TI's website , from here https://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/gfxsdk/latest/index_FDS.html or possibly elsewhere if there's a newer version Dec 31 05:53:52 that's an ancient obsolete sdk that won't work with current drivers Dec 31 05:54:16 however, yeah, I was confused by the multiple conflicting resources on the topic Dec 31 05:56:15 so, this is a message I wrote back in the day: https://www.mail-archive.com/beagleboard@googlegroups.com/msg43945.html Dec 31 05:56:55 the kernel stuff can be ignored Dec 31 05:57:04 since rcn has taken care of that with packages Dec 31 05:58:19 and I used the then-latest userspace libs with one patch: https://github.com/mvduin/omap5-sgx-ddk-um-linux/commit/2140ac17276536cd043a0d57955986c6af6369f9 Dec 31 05:58:31 three symlinks Dec 31 05:58:56 well, one symlink, but I included it for all targets Dec 31 06:03:18 this looks extremely similar to the directions I found here: https://elinux.org/BeagleBoneBlack/SGX_%2B_Qt_EGLFS_%2B_Weston#All_patches; one problem I had with this is I had difficulty compiling the special version of libgbm. this was also when i began having other library problems in general Dec 31 06:04:08 my branch of that libgbm from back then: https://github.com/mvduin/libgbm Dec 31 06:04:14 one difference is https://github.com/mvduin/libgbm which you recommend is a different uri from git://git.ti.com/glsdk/libgbm.git, they look similar, but i'll try it again with the different one Dec 31 06:05:03 the ti repo is upstream, my repo is a fork with some minor changes Dec 31 06:05:17 iirc I simplified it Dec 31 06:05:38 because this libgbm will only ever be used with the sgx libs (since upstream moved on) Dec 31 06:06:09 so it doesn't need to support anything else Dec 31 06:08:32 interesting that my package doesn't include that gbm_pvr.so symlink, hmm Dec 31 06:16:31 hmm Dec 31 06:21:51 okay wtf Dec 31 06:22:09 i should get this result after setting up the symlink right? file /usr/local/lib/gbm/gbm_pvr.so Dec 31 06:22:25 ./usr/local/lib/gbm/gbm_pvr.so: symbolic link to /usr/lib/arm-linux-gnueabihf/libpvrGBMWSEGL.so Dec 31 06:22:27 I seem to have repressed some memories of the horrors I've done back then Dec 31 06:24:08 this is using a custom-built libgbm that used /usr/local as prefix? Dec 31 06:24:42 this happened when i installed https://github.com/mvduin/libgbm Dec 31 06:25:12 if you didn't specify a prefix then, as always, the default prefix is /usr/local Dec 31 06:25:37 so, I'm still trying to reconstruct how my debian package works/worked Dec 31 06:25:53 I've found that it is actually this branch.. https://github.com/mvduin/omap5-sgx-ddk-um-linux/tree/ti-img-sgx/next ... however Dec 31 06:28:40 the libraries still compared different from the ones in the .deb, so I hexdumped them and compared the hexdumps Dec 31 06:29:47 it seems I replaced references to "libgbm.so.2" by "libgbm.so.1" ... o.O Dec 31 06:29:57 wtf did I do, and why Dec 31 06:30:24 umm well when i was trying to install based on the wiki directions, once I started getting .so file errors, there was a lot of that Dec 31 06:30:35 "libgbm.so.2 not found" Dec 31 06:32:36 huh, I found the commit by TI that bumped the library version to 2, yet my last commit changes it back to 1 Dec 31 06:32:39 why tho Dec 31 06:34:08 it's trivial enough to change it to 2 again as the userspace libs expect, just re-apply these two changes: https://github.com/mvduin/libgbm/commit/43cc786d0e5b8274fa93048c7f3bb8d6b62522db Dec 31 06:34:18 but it seems likely there's a story here Dec 31 06:34:48 oh, probably breakage of other programs that are linked against libgbm1 Dec 31 06:35:10 yeahhhh that's probably it Dec 31 06:38:47 i tried git checkout 43cc786d0e5b8274fa93048c7f3bb8d6b62522db now cat Makefile.am | grep LDFLAGS prints libgbm_la_LDFLAGS = -version-info 2:0:0 Dec 31 06:38:59 thats what it should be for now right? Dec 31 06:39:27 I mean, now you've moved to that commit and discarded everything after it Dec 31 06:39:49 I assume you're now on a detached HEAD (git will warn you of that) Dec 31 06:40:24 but yeah, changing the libgbm version to 2 was a bad idea Dec 31 06:41:53 I think it means you wouldn't be able to use any debian packages of libraries/programs that use opengl es (e.g. qt5) since they'd have a dependency on libgbm.so.1 Dec 31 06:42:39 so I guess I saw rolling back that change in libgbm and patching the userspace libs to depend on libgbm.so.1 instead of libgbm.so.2 as the lesser evil Dec 31 06:43:57 maybe I'll try to get sgx working again on my beaglebone when I get home in a few hours Dec 31 06:44:27 sorry i'm not very good at version control, i started over with fresh clone and edited the 1:0:0 to 2:0:0 manually, i'll try compiling and using this one now Dec 31 06:44:27 I feel like it's not hard if you know what you're doing, but it's kinda painful to figure out if you've never done it Dec 31 06:44:59 in both Makefile.am and configure.ac Dec 31 06:45:08 well you didn't need a fresh clone, you could have switched back using git checkout Dec 31 06:45:58 you could also use ti's libgbm instead of mine, I don't know if there's any real benefit to what I did Dec 31 06:46:06 or whether ti did anything in it since 2015 Dec 31 06:46:10 *2017 Dec 31 06:46:12 thanks i really appreciate you helping me with this since maybe you are one of very few people who have done this Dec 31 06:47:10 they actually did do stuff Dec 31 06:47:23 well when I used the libgbm here git://git.ti.com/glsdk/libgbm.git last time, i wasn't able to get that to work, but i could have been doing other things wrong as well Dec 31 06:47:58 they added a function, so presumably that function is requires by newer sgx libs Dec 31 06:48:25 so my 2017 one probably won't work without it Dec 31 06:49:52 ( https://git.ti.com/cgit/glsdk/libgbm/commit/?id=c5ddc6a37bb78ac753b317b17d890d1f7338dea6 ) Dec 31 06:53:08 oh, that's nice... that is an incompatible change in the API between libgbm and its plugin, i.e. libpvrGBMWSEGL.so Dec 31 06:54:31 because they didn't add it at the end of the struct Dec 31 06:54:34 idiots Dec 31 06:54:57 installing the ti version created libgbm.so.2 files and overwrote the other ones from yours in /usr/local/lib, so which ones should i try using first Dec 31 06:55:25 with what sgx libs? Dec 31 06:55:42 like, best bet is current sgx libs + current libgbm, both from ti Dec 31 06:58:37 okay, I have no idea what branch Dec 31 07:00:03 well i thought to try that first as well, but i didn't know how to do that so i just guessed. after you told me to try your packages, i wasn't sure if i had done it correctly before at all. Dec 31 07:00:21 i'm reconstructing the actual command i typed when i tried it, one moment Dec 31 07:00:41 my packages are most likely incompatible with the way newer kernel drivers Dec 31 07:01:00 there weren't any library-related errors in what you experienced Dec 31 07:01:11 it loaded the library, which did a kernel call to the driver, which failed Dec 31 07:03:07 git clone -b ti-img-sgx/zeus/1.17.4948957 git://git.ti.com/graphics/omap5-sgx-ddk-um-linux.git Dec 31 07:03:55 however i think this might be a version for a different linux distributon called yocto, i don't know where to go for a debian version Dec 31 07:03:58 that seems to be the latest, but dunno if that's work-in-progress or something Dec 31 07:04:24 there are no separate libraries for different distros Dec 31 07:05:40 it's unfortunate rcn didn't include the driver version in the kernel module package Dec 31 07:06:37 well that's good, but i also noticed when i was trying this version that there are errors in dmesg if i don't try to make the pvrsrvkm module the same version as this Dec 31 07:07:31 what do you mean? I thought you used rcn's kernel module package rather than building one yourself? Dec 31 07:07:51 okay so the latest released processor sdk does use an 1.17 driver... so what branch does it use Dec 31 07:08:20 ti-img-sgx/thud/1.17.4948957 Dec 31 07:08:32 according to https://software-dl.ti.com/processor-sdk-linux/esd/AM335X/latest/exports/software_manifest.htm Dec 31 07:08:59 I wish they included the actual commit hash instead of just the branch Dec 31 07:11:57 well basically, in dmesg I see [drm] Initialized pvr 1.17.4948957 20110701 for 56000000.sgx on minor 1 Dec 31 07:12:37 and at certain points, i noticed that when the user space libraries stated 1.14 rather than 1.17, i was getting errors Dec 31 07:12:48 oh, so pvrsrvkm *did* get run? Dec 31 07:13:39 oh, i think the only thing that has appeared to work every time is the initial loading of pvrsrvkm, at boot Dec 31 07:13:50 did you maybe leave your rc file in addition to the systemd service my package installs? Dec 31 07:14:06 oh and the 1.17 branch of the userspace libs include libgbm Dec 31 07:14:19 .so.1 ... they backed out of the version bump, haha Dec 31 07:14:44 i don't think so, /etc/init.d/rc.pvr still doesn't exist Dec 31 07:15:48 anyway, it does sound like userspace libs and kernel need to be compatible Dec 31 07:16:01 so that rules out my .debs with rcn's kernel module .deb Dec 31 07:16:42 my suggestion would be the ti-img-sgx/thud/1.17.4948957 branch of TI's graphics/omap5-sgx-ddk-um-linux repo Dec 31 07:16:52 with rcn's kernel driver package Dec 31 07:17:14 and no libgbm since it's now bundled with omap5-sgx-ddk-um-linux Dec 31 07:18:50 installation to be done by manual copying :P Dec 31 07:19:17 so i should uninstall all of your packages right, and also remove all the libgbm files i think i installed, but leave ti-sgx-ti335x-modules-4.19.94-ti-r42 installed, then try to compile and install the repository from git clone -b ti-img-sgx/thud/1.17.4948957 git://git.ti.com/graphics/omap5-sgx-ddk-um-linux.git? Dec 31 07:19:46 yeah, remove my three packages Dec 31 07:19:56 and anything manually installed Dec 31 07:20:17 you could keep copies of my udev rule and service file Dec 31 07:20:25 they might come in handy Dec 31 07:20:49 my package is probably also a good guideline on what to put where Dec 31 07:21:07 I put a listing here: https://liktaanjeneus.nl/sgx/sgx-ti335x-userspace-1.14.3699939_1.1.ls-l.txt Dec 31 07:25:31 though maybe it would be better for the udev rule have a SYSTEMD_WANTS, though that's probably incompatible with how I wrote the service file... not sure, I'd need to check or test Dec 31 07:26:26 nah, how I did this is probably fine, just add an [Install] section to the service file to allow enabling it to run at boot Dec 31 07:27:13 dunno, I'd need to experiment, and I probably will Dec 31 07:29:30 ah, "thud" is the name of the yocto release Dec 31 07:29:36 hmm well i have noticed that the omap5-sgx-ddk-um-linux repostory seems to provide precompiled libraries only, which the makefile copies in, though it does include a libgbm as you said Dec 31 07:29:53 there's no source code available for these libraries Dec 31 07:29:55 i hope that will work correctly Dec 31 07:30:26 yeah I hope so too... like I said, just copy them to the appropriate place Dec 31 07:31:15 and make sure that any dependencies they need have been installed Dec 31 07:33:23 i also need /etc/powervr.ini right both you and other sources have been clear on that Dec 31 07:38:50 ¯\_(ツ)_/¯ Dec 31 07:39:13 probably I guess maybe? if its contents looks plausible Dec 31 07:39:31 if possible, getting it to use RGB565 format would be desirable Dec 31 07:40:18 and with the libgbm that comes with omap5-sgx-ddk-um-linux, do i still need the symlink from /usr/lib/gbm/gbm_pvr.so to something? (the omap-sgx-ddk-um-linux newest one from TI is installed now in /usr/lib). unfortunately there isn't actually any "libpvrDRMWSEGL.so" in this version. That part was particularly confusing for me before, and I didn't know what to do Dec 31 07:40:20 its dependencies appear to be: libdrm2 libexpat1 libwayland-client0 libwayland-server0 Dec 31 07:40:44 no, just assume they made sure it works as shipped Dec 31 07:41:37 and libraries need to go into /usr/lib/arm-linux-gnueabihf/ Dec 31 07:46:52 ok i think i installed everything the way you said, thanks, and there is /etc/init.d/rc.pvr again, so now I would use this as /etc/init.d/rc.pvr start to initialize the driver as before? I can't think of any reason to restart right now since I left the kernel modules unchanged Dec 31 07:48:24 I'd probably be paranoid and reboot anyway, though you could try it first and then reboot if it looks something is going wrong that's not a shared library isssue Dec 31 07:49:02 the ti-img-sgx/zeus/1.17.4948957 branch might also be worth trying btw, "zeus" is a more recent yocto release than "thud" Dec 31 07:50:14 I ran it and progress, i'm now at the same point i was able to get to without errors before reinstallation. I see PVR_K: UM DDK-(4948957) and KM DDK-(4948957) match. [ OK ] in dmesg as before. Dec 31 07:50:48 that was my thought too, so one difference is now i'm using the "thud" branch, where i was using "zeus" before Dec 31 07:50:51 okay, that sounds good? Dec 31 07:51:33 i suppose since this is the TI branch of the libraries, I should try to use the TI version of kmscube from here? git://git.ti.com/glsdk/kmscube.git Dec 31 07:51:46 you could try the included tests first Dec 31 07:53:01 hmm, which included tests? I know of this repository git://git.ti.com/graphics/img-pvr-sdk.git, but i didn't realize there was an alternative Dec 31 07:53:41 oh! Dec 31 07:53:46 didn't realize they didn't include the tests anymore Dec 31 07:54:01 then uhhh dunno lol Dec 31 07:54:10 kmscube I guess maybe Dec 31 07:58:23 ./kmscube -d /dev/dri/card0 output: ./kmscube -d /dev/dri/card0 Dec 31 07:58:32 https://termbin.com/na4ai sorry misclick Dec 31 07:59:07 I'm out of thoughts for now I think Dec 31 07:59:16 card1: https://termbin.com/fl8e Dec 31 07:59:47 well this has been a lot of progress i think because the error I see in kmscube IS actually different from before Dec 31 08:01:35 I don't see "PVR:(Error): PVRSRVBridgeCall: Failed to access device" anymore, so i'll try the other demos which were giving the same error previously Dec 31 08:33:14 ok i rebooted and tried again with no changes, output of $ /opt/img-powervr-sdk/Examples/Advanced/NullWS/OGLES2Water (same as root execpt root has XDG errors) https://termbin.com/pzdnh **** ENDING LOGGING AT Thu Dec 31 12:40:58 2020 **** BEGIN LOGGING AT Thu Dec 31 12:44:19 2020 Dec 31 19:47:38 Can BeagleBone-AI, equipped with EVEs and TIDL, be used for face detection and face identification ? Dec 31 19:48:23 Can BeagleBone-AI, equipped with EVEs and TIDL, be used for face detection and face identification ? Dec 31 19:48:43 impatient much? Dec 31 21:21:41 dug as far and wide as I can and Robert C Nelson says in this ancient post that EGL_NOT_INITIALIZED is caused by "the mesa libegl.so" being installed. this seems relevant because kmscube prints "EGL Vendor "Mesa Project"" and other mesa references even though i have been very careful never to install any "mesa" packages this time. Dec 31 21:21:50 The trouble is, all of these libraries I have I've taken directly out of the TI SGX SDK repository and installed in every location Robert mentions, so I don't know where else to get them. Dec 31 21:22:03 https://www.mail-archive.com/beagleboard@googlegroups.com/msg14451.html Dec 31 21:24:05 I'd like to compile the SGX SDK from source, the kernel module, and any other components from source, to try to make sure this is compatible with the current latest clean BBB image, but I've been told that for some reason compiling the entire toolchain from source is not allowed/possible Dec 31 21:36:43 I think installing a toolchain is allowed. Now, you will need this so-called toolchain to compile your source. It needs to be the same no matter what, i.e. .so files. So, if you are using debian Buster w/ libstdc++.so.6.25 the compilation machine needs to have that exact toolchain too. Take note of machine to machine. Dec 31 21:37:27 For instance... Dec 31 21:38:09 Forget it. Dec 31 21:38:13 I am wrong as usual. Dec 31 21:38:45 what libegl.so are you using on your dev. desktop? Dec 31 21:39:20 And what does debian Buster on the BBB use for libegl.so? libegl.so.x.x.xx? Dec 31 21:41:23 set_: I've been avoiding cross compilation and only using gcc directly within Debian on the BBB to compile the software components of this project Dec 31 21:41:48 however I am willing to set up a cross compilation environment if necessary to achieve my goals Dec 31 21:42:56 On the BBB I have specifically avoided using any libEGL.so except for the ones I've found in this repository: git://git.ti.com/graphics/omap5-sgx-ddk-um-linux.git Dec 31 21:43:28 at least for the current installation, I'll try any idea I possibly can Dec 31 21:45:05 Oh. Dec 31 21:45:39 I am just speaking here from learning about something. This something... Dec 31 21:46:04 I had to cross-compile w/ eclipse on Debian Buster to Debian Buster on the BBB. Dec 31 21:46:46 For one, I could only make this cross-comp. possible b/c of Debian Buster having the same libstdc++.so versioning as it is on the BBB. Dec 31 21:47:24 So, if you are using Ubuntu, GenToo, or whatever, I say switch to Debian and use Buster just like the BBB has on it. Dec 31 21:48:50 I have seen so far 4 people attest to accomplishing my goal, "executing OpenGLES 2.0 instructions on the BeagleBone Black's IT PowerVR SGX 530". This is now 4, with this person: https://wiki.thing-printer.com/index.php?title=SGX_SDK_5_Alpha. His implementation, though, is inaccessible, so not open source Dec 31 21:49:16 Okay. Dec 31 21:49:48 set_: I am willing to set up a Debian 10 x86 environment to try to replicate your cross-compilation process. I'll do this now Dec 31 21:50:08 So, it is possible. Yes. The BBB is nice in that fashion. I am just saying cross-compile from Debian Buster. Dec 31 21:50:39 Not from Windows or Ubuntu or Kubuntu or whatever else that may have different versioning of the .so file you need to install. Dec 31 21:51:51 Now, the instructions are long and not as straight forward b/c these instructions are for Ubuntu but Debian. I got some .so file to work for some complications for homing on a SDK. Dec 31 21:52:33 But yes. You need to get a debian workstation for compiling the source. Dec 31 21:55:19 So, let me get some instructions together in a pastebin paste and you can get the Debian Distro w/ Eclipse. Just get the lastest-greatest and make sure you are using the C/C++ version of Eclipse. Dec 31 21:55:32 Oh! Dec 31 21:55:52 Do not use the installer. Directly download Eclipse from their site online. Dec 31 22:01:41 downloading debian release from official, e.g. https://www.debian.org/distrib/, then I should download Eclipse from https://www.eclipse.org/downloads/download.php?file=/oomph/epp/2020-12/R/eclipse-inst-jre-linux64.tar.gz and select the components for C/C++ when installing it, right Dec 31 22:02:43 not sure what you exactly mean by "do not use the installer" Dec 31 22:02:53 Okay. One can just use the radio button online instead of the command line. The command line on cross-comp. is not something I am really knowledgeable about right now. Dec 31 22:02:54 Oh. Dec 31 22:03:05 I thought Debian had an installer in their repos. Dec 31 22:05:07 Hey. Are you using the server or the entire distro? Dec 31 22:06:53 set_: I was planning to install debian with Xorg and execute Eclipse to display GUI in Xorg, should I worry about specific desktop environment, or do I need to specifically make sure I have an installation with no graphical environment, the way I do for the BBB? Dec 31 22:07:12 No. Dec 31 22:07:25 Use the graphical display for Eclipse, so far. Dec 31 22:07:53 Debian >>> Graphical && Eclipse >>> Graphical Dec 31 22:11:47 debian.org Dec 31 22:12:04 That is the site. It just shows the download radio button on the front page. Dec 31 22:12:09 It already picks your arch. Dec 31 22:12:27 Let me go and find the eclipse site. Dec 31 22:12:50 oh thanks yes I am installing the debian environment, i didn't already have one available so it will just take a minute to launch it Dec 31 22:13:03 This is the eclipse site: https://www.eclipse.org/downloads/ Dec 31 22:13:12 Okay. No issue. Debian takes an age to install. Dec 31 22:20:44 https://pastebin.com/AtVb0Hpv is the start. Update/Upgrade and some libs. and the jre is for running eclipse eventhough I think eclipse now comes prebundled w/ jre. I am pretty sure. Dec 31 22:29:23 ok in the eclipse installer do I choose "for C/C++ Developers" or "for Embedded C/C++ Developers"? only asking since I am pretty sure the "ES" in the "OpenGLES 2.0" API I am trying to use, does stand for "Embedded System" Dec 31 22:29:46 C/C++ Dec 31 22:30:17 You are only cross compiling the libxxxx.so onto the BBB for use, right? Dec 31 22:30:45 So, you can use libegl.so? Dec 31 22:31:44 well I need some kind of libEGL.so and libGLESv2.so I am sure, to execute kmscube Dec 31 22:31:53 for developers. Dec 31 22:31:58 Okay. Dec 31 22:32:02 So yea. Dec 31 22:32:22 Okay. Dec 31 22:32:28 I'm trying to set executing kmscube successfully, or some other extremely basic demo, as a goal to confirm I can access the SGX530 Dec 31 22:32:58 I've installed everything you said and now I see eclipse Dec 31 22:33:25 Okay. Although I am not granting you can use the SGX530, even though I know it is possible, I am only here to help you get libegl.so and libGLESv2.so on your BBB. Dec 31 22:34:12 Did you see the paste? Dec 31 22:34:36 ok well I hope we can create some that will work, yes i installed those packages also Dec 31 22:34:43 Okay. Dec 31 22:34:50 Now... Dec 31 22:35:27 go to install new software. Dec 31 22:35:32 Oops. Dec 31 22:35:43 Let me do this instruction in the paste again. Dec 31 22:39:27 https://pastebin.com/WcG09Lp5 is the second paste. Dec 31 22:40:07 Hey misternumberone: Do you know Eclipse well? Dec 31 22:41:26 ok done. I have used eclipse before for Java development. I wouldn't say it's my favorite IDE but it has a lot of features so that is good. Dec 31 22:41:43 It will restart. This is okay. Please let me know after the restart is done. Dec 31 22:41:52 yes it restarted Dec 31 22:41:55 Nice. Dec 31 22:41:57 Okay. Dec 31 22:42:12 This may be simpler than once thought. Dec 31 22:43:20 Window -> Preferences -> "expand" C/C++ and then "expand" Build -> Global Tools Paths Dec 31 22:43:48 Default toolchain to Linaro ARMv7 Linux GNU EABI HF Dec 31 22:44:23 ok Dec 31 22:44:39 Now, do you already have the sdk downloaded? Dec 31 22:45:15 in .tar format or whatever? Dec 31 22:46:41 umm is that one of these? https://software-dl.ti.com/processor-sdk-linux/esd/AM335X/latest/index_FDS.html Dec 31 22:46:50 Hmm. Dec 31 22:46:53 See. Dec 31 22:47:13 I thought you needed to put the .so file on the BBB. I thought you knew about your SDK. Dec 31 22:47:24 Oh? Dec 31 22:47:41 You are trying to put the entire Linux SDK from TI on your BBB? Dec 31 22:48:14 I thought you were just trying to put libegl.so and libGLESv2.so on your BBB so you can alter your SGX to perform. Dec 31 22:49:31 Well specifically I think I need just the SGX SDK in order to access the GPU. I thought that was in this git repository git://git.ti.com/graphics/omap5-sgx-ddk-um-linux.git, but nothing I've tried with files from there has worked Dec 31 22:49:58 Okay... Dec 31 22:50:10 Let me go to TI's git real quickly. Dec 31 22:50:57 we need to find that file or get it w/ wget in .tar format, i.e. not w/ git. Dec 31 22:51:17 Aw hell. I am plugged in. let me try. Dec 31 22:52:11 I thought the format would be in .tar format. This was my mistake. Dec 31 22:52:27 Dang it. Please hold. Dec 31 22:54:03 Okay. Dec 31 22:54:11 So, they have the .tar files. Dec 31 22:54:30 Do you need the um-linux.git or linux.git file? Dec 31 22:55:15 Here is one: https://git.ti.com/cgit/graphics/omap5-sgx-ddk-linux/ Dec 31 22:55:24 well, I think the um stands for user mode, as opposed to the kernel space portion of the SGX530 driver which is a kernel module Dec 31 22:55:25 That is the actual site. Dec 31 22:55:33 Okay. Dec 31 22:56:12 So, first use wget to get the required .tar file from that page online that is listed but try the um-linux version instead. Dec 31 22:56:21 user mode! Dec 31 22:58:33 Dang it. Dec 31 22:58:40 user mode does not have any .tar files. Dec 31 22:58:44 They are outdated too. Dec 31 22:58:45 so on the site you sent I see the .tar.gz file, but on https://git.ti.com/cgit/graphics/omap5-sgx-ddk-um-linux/ I don't see the tar files Dec 31 22:59:00 Right. Dec 31 22:59:15 Forget about um-linux for now. They are outdated, era '15. Dec 31 22:59:23 instead I see a list of all the branches from the git repository, and also i've never been sure which of those was correct Dec 31 22:59:34 ok Dec 31 22:59:46 Try the .tar.gz version, most up to date. Dec 31 23:00:48 ok i have it, should I extract this Dec 31 23:00:49 Everything is older...hmm. Dec 31 23:00:53 No. NOt yet. Dec 31 23:01:13 Is Eclipse still open? Dec 31 23:01:17 yes Dec 31 23:01:51 Okay. In some dir, extract it. Dec 31 23:02:32 done, I see eurasia_km Dec 31 23:02:42 Now, in eclipse, File -> OpenProjects from File System! Dec 31 23:03:28 Make sure that "Search for Nested Projects" and "Detect and configure project natures" is highlighted. Dec 31 23:03:43 IMport the source! Dec 31 23:03:57 Click done/finish/or whatever... Dec 31 23:04:15 ok i see the source tree in project explorer Dec 31 23:04:18 Nice. Dec 31 23:05:08 Okay. So, go to the head of the dir. in the project exp. and right click. Dec 31 23:05:38 Now, go to Build-Configurations and set DebugBeagle. Dec 31 23:05:57 I am not completely sure what this entails since that co. did this idea. Dec 31 23:06:10 i.e. the DebugBeagle. Dec 31 23:06:44 Now, right-click the same heading and Build or Build Project. Dec 31 23:07:10 Did that work? Dec 31 23:07:33 I don't see Build-Configurations in any right click menus on the top folder or other folders Dec 31 23:07:59 Oh. Dec 31 23:08:00 Hmm. Dec 31 23:08:23 Hmm. Dec 31 23:08:27 Wait... Dec 31 23:09:10 Okay. Dec 31 23:09:32 So, our makefile for building needs to be completed. Dec 31 23:09:50 Did their makefile show up in your project exp? Dec 31 23:10:20 Anywhere in the project exp? Dec 31 23:10:48 I see severale Makefiles and also .mk files Dec 31 23:11:37 to be specific build/linux2/common/omap4430_linux/Makefile and build/linux2/common/omap4430_linux/omap5430_linux/Makefile Dec 31 23:11:39 Okay. Dang it. I see my mistake. It seems this co. I described in the pastebin.com paste made their makefile specifically for the use b/c of the BBB. So... Dec 31 23:11:59 This means. More, more, more. Dec 31 23:12:22 I will need more time and I will have to do it myself now to checkout the files to test them. Dec 31 23:12:40 I am not sure what makefiles you are looking at right now. Dec 31 23:12:58 I thought they would have had a particular build env. Makefile for the BBB. Dec 31 23:13:06 That may be too much to ask from TI. Dec 31 23:13:44 Since...beagleboard.org is a separate entity from TI.com, I can see this happening. Dec 31 23:13:45 So... Dec 31 23:13:47 i screenshotted some of what I see https://i.imgur.com/YzfFzi7.png Dec 31 23:13:50 Nice. Dec 31 23:14:41 so, omap4430-linux is where i think you need to right-click. Dec 31 23:14:59 And then look for the build configurations option. Dec 31 23:16:00 I could not see the rest of the file name in the post. Dec 31 23:16:23 https://i.imgur.com/D7plgmk.png Dec 31 23:17:11 I am guessing now. Let me try it soon. Dec 31 23:17:51 its ok, here is when i right click on the file itself https://i.imgur.com/ikPdbf0.png Dec 31 23:18:06 i am willing to try anything to make what I want work Dec 31 23:18:24 Okay. Dec 31 23:18:37 So, eurasiacon is a name you chose? Dec 31 23:19:04 no umm so i think that was already there Dec 31 23:19:11 Hmm. Dec 31 23:19:33 Did you right click that section? Dec 31 23:19:45 I will install it now and try. Dec 31 23:20:14 yes, i see i think the same thing in every right click menu within project explorer Dec 31 23:20:30 Dang it. Dec 31 23:21:14 Try to build it but I hope you have a powerful system. Dec 31 23:21:23 the path is omap5-sgx-ddk-linux-glsdk_7.04.00.03/eurasia_km/eurasiacon, etc Dec 31 23:21:46 Oh. I got you. Dec 31 23:22:22 Okay. I am going to leave and come back in another name w/ info. on a Debian Distro. Is that okay? Dec 31 23:22:43 I am going to repeat some steps and try something else too. Dec 31 23:22:45 well under eclipse Project/Build Build All and Build Project are gray Dec 31 23:22:52 yes thats great Dec 31 23:23:35 Let me go and I will be right back. Dec 31 23:39:28 Okay. SOrry. Dec 31 23:39:36 It is me...set_. Dec 31 23:39:55 I am setting everything up now. Dec 31 23:40:28 cool Dec 31 23:42:43 Okay...this is the first part of the hyperlink: https://git.ti.com/cgit/ Dec 31 23:42:50 What is the rest of it again? Dec 31 23:43:59 https://git.ti.com/cgit/graphics/omap5-sgx-ddk-linux/ Dec 31 23:45:08 Thank you. Dec 31 23:45:21 I will brb. I need to set up eclipse again for some reason. Dec 31 23:49:05 Hey... Dec 31 23:49:21 Did you pick the eclipse for developers or the embedded one? Dec 31 23:49:47 i chose C/C++ for developers not the embedded one Dec 31 23:50:01 Okay. Thank you. Jan 01 00:02:50 Okay. Cough. So, I must have been mistaken. The teknic.com co. must have put together some makefiles for their specific sdk but it is not accessible outside of their sdk builds. So... Jan 01 00:03:23 Maybe trying the build w/ another type of makefile may prove valuable. Jan 01 00:03:35 I am sorry. It is a large lib. and I am unfamiliar w/ it for now. Jan 01 00:03:42 I will keep trying. Jan 01 00:04:15 They have a cc-check.sh file that may be able to be run under some type of Linux machine. Jan 01 00:04:19 Have you checked that out yet? Jan 01 00:05:43 ./cc-check.sh: usage: ./cc-check.sh [--64] --cc CC --out OUT [cflag] Jan 01 00:07:19 Oh. Okay. Also, there is a Makefile-template. Jan 01 00:07:24 I may be able to use it. Jan 01 00:08:39 ${cross_make} Beagle && ${cross_make} Linux x86 <<< This is what I am trying to make work right now w/ the Eclipse build and libegl.so right now. Jan 01 00:13:53 misternumberone: Have you ever cross-compiled before today? Jan 01 00:15:08 set_: yes but never for an ARM device, only for a completely different unrelated CPU Jan 01 00:15:14 I do not know everything, i.e please excuse me. But! Jan 01 00:17:16 look at the makefile-template. Jan 01 00:17:45 Supposedly, that is the highest level makefile we need. Jan 01 00:18:02 I say we b/c I am interested in making this work too. NO offense. Jan 01 00:19:21 ok I see Makefile.template under kbuild, is there a different one called makefile-template? Jan 01 00:19:38 sorry i'm in a lot of chats at once Jan 01 00:19:43 Oh. Jan 01 00:19:52 thanks for helping with this i have really struggled with it Jan 01 00:20:01 Okay. No issue. You are right. it is a . and not a - Jan 01 00:20:44 We still need the kernel I think. Jan 01 00:21:01 So, 32-bit Linux kernel, maybe? Jan 01 00:21:50 oh no I installed a 64-bit release of Debian Jan 01 00:21:56 I am reading the makefile...No. Jan 01 00:21:59 It is not an issue. Jan 01 00:22:07 64 bit is not big deal. Jan 01 00:22:35 I am reading the makefile. It seems we may be able to put the file on the BBB w/ kbuild. Jan 01 00:22:42 I am not too familiar w/ kbuild. Jan 01 00:22:52 so, maybe sftp and /kbuild? Jan 01 00:24:28 i have ssh server on beaglebone and i can ssh there but you mean I should set up ftp server? Jan 01 00:26:43 No. Jan 01 00:26:45 Hey! Jan 01 00:26:49 I found an install file! Jan 01 00:26:55 It gives directions! Jan 01 00:27:23 Look toward the bottom of the window where all the files are located. Jan 01 00:27:41 They have a file called and I quote, "INSTALL." Jan 01 00:27:42 Ha. Jan 01 00:29:46 Hey, you may not even need to cross-compile. It will be a task of knowing what has transpired in the last five years and what has remained the same. Jan 01 00:30:12 It seems make all, make clean, and others are available for use. Jan 01 00:30:39 But then...it says "THESE THINGS MUST BE MET" and I am not sure why. Jan 01 00:32:48 looks great but it says the $DISCIMAGE and $KERNELDIR should be set, which means i need access to the filesystem and kernel currently on the BBB right? so i would guess it's easier to compile there instead of cross compile? Jan 01 00:33:00 I saw that idea. Jan 01 00:33:26 I am not sure. I think the beagleboard.org people would have to pay you a favor. Jan 01 00:33:35 or at least the kernel headers, maybe i could install and then copy those from bbb, then cross compile and install into a temp folder, and copy the installation back? Jan 01 00:34:08 Seems odd but maybe. Trial and error is one way to learn. "I KNOW!" Jan 01 00:34:14 Ha. Jan 01 00:34:35 This is weird for me. I thought it was a .so file we were trying to put on the BBB. Jan 01 00:35:04 I think that the Imag. Tech. people stopped producing SGX source in '15. Jan 01 00:35:18 It may still work but so much has changed. Jan 01 00:35:18 well that is essentially exactly what i was trying before, just that didn't work Jan 01 00:35:43 Okay. So, I must have repeated you at some point. Jan 01 00:36:08 I am going to put this source on my BBB and test it now. Jan 01 00:37:08 i have to go for a while but i'm staying logged in and i will also try to compile this directly on Debian on BBB by downloading this source tar there Jan 01 00:37:20 once i come back Jan 01 00:37:27 That is what I am trying now. I will let you know what happens. Jan 01 00:37:53 beagleboard.org may have made those libs. available in their build. Jan 01 01:03:53 But...cc must still happen to get the sdk onboard the BBB. Jan 01 01:04:05 I just tried w/ scp and sftp. NO go so far. Jan 01 01:32:57 Okay, okay. Jan 01 01:33:36 I got it. You have to listen to the commands, make, and then use kbuild (Maybe) to put the items on the board. BBB! Jan 01 01:34:50 has the 4.14.x image been built w/ CONFIG_PREEMPT=y? Jan 01 01:38:54 ok i'm back, actually the kernel I see is 4.19.94-ti-r42 since I was instructed to clean flash the latest image from beagleboard.org. however I see CONFIG_PREEMPT=y in /boot/config-4.19.94-ti-r42 Jan 01 01:39:15 Nice. Jan 01 01:39:19 Now... Jan 01 01:39:53 Hmm. Jan 01 01:40:05 I am working things out in my head but dead ends are everywhere. Jan 01 01:40:27 back to reading the two main files. Jan 01 01:41:29 I tried the make command already. Install dos2unix if you have not done so already. Jan 01 01:42:28 I need to cc a 32-bit Linux kernel I guess. Jan 01 01:42:35 Dang it or... Jan 01 01:42:40 usually if i need dos2unix, the magic spell is to execute it on everything in the entire cloned repository, is that necessary here? Jan 01 01:42:44 I just need to export the path to that kernel. Jan 01 01:42:53 No. Jan 01 01:43:19 I installed it on my dev. desktop and it works. I get past that point in error accumulation. Jan 01 01:45:15 misternumberone: Where is the kernel on the dev. desktop located? Jan 01 01:47:49 the x86 kernel currently running or the kernel for bbb? the kernel header files on the x86 debian are located in /usr/src - after the debian kernel headers package is installed, which it isn't atm but i can do that if that's what you mean Jan 01 01:48:56 x86. Jan 01 01:49:39 I think I need to account for the toolchain, kernel dir, and some other stuff before I can move forward. Jan 01 01:49:59 well to be precise, currently im pretty sure the kernel i see when i run uname -r is actually stored in /boot Jan 01 01:50:21 Right. Okay. Jan 01 01:50:25 I was just making sure. Jan 01 01:50:38 I am going to go w/ it and type in vmlinuz. Jan 01 01:50:41 In /boot. Jan 01 01:51:42 yes that's the kernel but i think it's a binary executable format file, unsure of how this process works though Jan 01 01:52:25 So, we need to export the kernel, filesytem, and toolchain. Jan 01 01:52:50 That is what I think we need so far for my build to take place w/ make. Jan 01 01:53:05 In the INSTALL File, there are a couple pointers but I need to research the rest of it. Jan 01 01:54:39 https://wiki.debian.org/FileSystem will help. Jan 01 01:54:47 I found hints! Jan 01 01:56:46 the INSTALL file says "to build for OMAP5430/DRA7xx, change to the appropriate target directory" Jan 01 01:57:14 to run make command I think I need to do something like that since those are the only places I've found any files named "Makefile" Jan 01 01:57:28 however i'm unsure which of those to use, i don't know what those are Jan 01 01:59:19 this website from TI suggests those might be different models of ARM CPUs Jan 01 01:59:23 https://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/bios/sysbios/6_45_03_32/exports/bios_6_45_03_32/docs/cdoc/ti/sysbios/family/doc-files/devices.html Jan 01 01:59:36 however, i don't have any of those, I only have AM335x Jan 01 02:00:54 RIght. Okay. Jan 01 02:01:00 Okay. Jan 01 02:01:50 Have you ever exported the filesystem, a kernel, and/or the toolchain? Jan 01 02:02:57 It is as simple as >>> export PATH=$PATH:/path/to/toolchain <<< That! Jan 01 02:03:11 Now, remember that toolchain we used? Jan 01 02:03:58 i was about to try building without cross compilation since i'm unsure what the path to toolchain would be on the x86 debian system. I don't know where Eclipse has installed those files you told me to set earlier. Jan 01 02:04:14 Okay. Jan 01 02:04:24 You read my mind. I will look for them. Jan 01 02:04:31 Try whatever. Jan 01 02:08:03 I got further. Jan 01 02:08:27 I put in the kerneldir=/boot/vmlinuz.x.x.x and i got further down the line. Jan 01 02:11:18 no matter whether I use /boot/vmlinuz... or /usr/src/linux-headers... I see "$(KERNELDIR)/vmlinux does not exist." so this says it expects a file called "vmlinux" , which I don't have on either the x86 system or bbb, as far as I can tell Jan 01 02:14:35 even despite that, the build does attempt to proceed. The actual error is fatal error: asm/system.h: No such file or directory, meaning it also expects this file Jan 01 02:15:32 Me too. Jan 01 02:15:52 i couldn't find a list of build dependencies for this anywhere, but z matt was somehow able to quickly find a list of runtime dependencies for the precompiled version of the user space SDK when presented with it Jan 01 02:15:56 I have it but I am getting a Makefile error. Jan 01 02:16:08 Oh. Jan 01 02:16:09 Nice. Jan 01 02:16:24 so maybe there's some way to do that easily for the build dependencies.. i don't know it though Jan 01 02:17:34 I exported everything on the dev. desktop already and went into the file I need to be in to use make from the Makefile but the compilation error tells me this: grep: /boot/vmlinuz-4.19.0-13-amd64/Makefile: Not a directory Jan 01 02:18:35 There is not a Makefile in that dir. I know! But why should I make an export of the filesystem, arch, KERNEL, and other exports. None of them have any makefiles in them associated w this build. Jan 01 02:18:44 Odd. Jan 01 02:19:13 Old instructions. Jan 01 02:20:27 I will try the git repo. and try on the BBB. again. Sheesh. This is taking forever. Jan 01 02:22:39 from the very start, when I learned the information that there are no emmc images for beaglebone black on beaglebone.org that have access to the IT PowerVR SGX530 device inside the AM3358 SoC, the only related package in the rcn-ee.com repository is the kernel space driver, and everything else must be installed by hand or written from scratch, Jan 01 02:23:32 I seriously wondered if the device is actually supported on modern linux by TI or anyone else Jan 01 02:25:32 Oh. Jan 01 02:25:43 There are eMMC images available. Jan 01 02:25:51 beagleboard.org/latest-images Jan 01 02:26:14 '15 is the last update to the driver source. Jan 01 02:26:27 I am trying to compile now on the BBB. Oops. Jan 01 02:28:27 I got the same error on the BBB and the CC on my dev. desktop. I am doing something incorrectly. Jan 01 02:28:59 There needs to be a makefile in the vmlinuz kernel dir. Jan 01 02:29:15 At least, that is what my machines are telling me from errors. Jan 01 02:31:17 hmmm I'm reading closely that page again and there is something I missed. Jan 01 02:31:21 "Stretch ImgTec (with Imagination Technologies 3D developer kit) for BeagleBone via microSD card" Jan 01 02:31:30 debian 9 Jan 01 02:32:14 could this mean the full driver, as it turns out, only works in debian 9, and I might find the actual SDK and demos I desire already in fact preinstalled this whole time, in a different, alternative beaglebone black image? Jan 01 02:33:18 Oh. Yea. '15. See, it is completely done from what I understood. Jan 01 02:33:31 I cross-compiled it by building it. Jan 01 02:33:39 Firstly, the builds were all done. Jan 01 02:33:48 But, it is not a normal file. Jan 01 02:33:58 So, I could not put it directly on the BBB. Jan 01 02:34:20 But...the git clone worked. And all the files are there. Jan 01 02:34:50 But, building it w/ make does not work which makes me think something is up. But...who knows? It may work w/ the source you provide. Jan 01 02:34:53 Try it! Jan 01 02:35:20 Or the examples? There may be some examples. Jan 01 02:38:57 If stddef.h is on the BBB, I think you can use the examples. I have not found any yet. Jan 01 02:40:25 I found the main.c program. Jan 01 02:40:34 It seems that it is done. Jan 01 02:40:39 source away! Jan 01 02:40:59 all I need is an example of the SGX530 actually executing OpenGLES 2.0 instructions. From there I can clone the emmc to make a snapshot, and begin porting the real software I want directly into whichever environment that is. I will flash this image, the one I overlooked this whole time: https://debian.beagleboard.org/images/bone-debian-9.12-imgtec-armhf-2020-04-06-4gb.img.xz and search for any evidence of this. Jan 01 02:42:05 Oh...is that image on the main page. Odd. Jan 01 02:42:12 I just remembered that you are right. Jan 01 02:42:14 Nice. Jan 01 02:42:22 how were you able to compile the omap5-sgx-ddk-linux though? did you get resulting libraries? Jan 01 02:42:29 Yea. Jan 01 02:42:32 I got all the builds. Jan 01 02:42:39 make Jan 01 02:42:49 Let me see. Jan 01 02:42:52 Please hold. Jan 01 02:43:32 I think since both libs. are five to six years old, the um-linux and the other one, either may work for you. Jan 01 02:43:44 Hmm. Jan 01 02:43:52 I should try to get both to test things out. Jan 01 02:46:51 Maybe um-linux has the examples. Jan 01 02:48:12 z matt told me he thinks there are no tests in any of the repositories i've discussed with you Jan 01 02:48:33 Oh. Jan 01 02:48:34 Okay. Jan 01 02:48:37 I am finding out now. Jan 01 02:48:40 but i have been trying to test with a few demos including those from this repository Jan 01 02:48:42 git://git.ti.com/graphics/img-pvr-sdk.git Jan 01 02:48:49 Oho. Okay. Jan 01 02:48:51 Nice. Jan 01 02:49:36 Sometimes it is fun to go back to use older equipment. The BBB is not exhausted nor de funct. So, going back is not an issue to me since it is still relevant. Jan 01 02:51:38 on the um-linux git repo, there is a file you can open, .doc file, that has info. I guess. Jan 01 02:51:54 I need to get it on my dev. 'puter first to view it. Jan 01 02:53:55 Dang it. Jan 01 02:58:01 if it turns out the device works with the debian 9 image, that will be awesome. it takes a long time for the image to flash, i'm going to be elsewhere while i wait for it. if I see evidence of SGX530 activation by the installed "Imagination Technologies 3D developer kit", which may or may not consist in part or whole of the SDK we have been attempting to install, **** ENDING LOGGING AT Fri Jan 01 03:00:13 2021