khilman | note that `bitbake libappcontroller` seems to build/install fine, but that cmake dep still fails. | 00:02 |
---|---|---|
*** khilman has quit IRC | 01:22 | |
*** khilman has joined #automotive | 01:55 | |
mranostay | khilman: you seeing a build error on agl-service-audio-4a by chance? | 03:21 |
*** AlisonChaiken has joined #automotive | 04:11 | |
*** Tarnyko has joined #automotive | 04:33 | |
*** Tarnyko has left #automotive | 04:37 | |
*** OutBackDingo has joined #automotive | 05:17 | |
OutBackDingo | question support for AGL on raspberrypi 3B plus ? all i see support for is raspberripi3 | 05:52 |
*** thaytan has quit IRC | 07:02 | |
*** khilman has quit IRC | 07:11 | |
*** khilman has joined #automotive | 07:11 | |
*** kbroulik has joined #automotive | 07:15 | |
*** nayfe has joined #automotive | 07:15 | |
*** leon-anavi has joined #automotive | 07:57 | |
*** [AD]Turbo has joined #automotive | 08:01 | |
*** GuyElpotato has joined #automotive | 08:01 | |
*** [AD]Turbo has left #automotive | 08:01 | |
leon-anavi | hi | 08:02 |
*** psnsilva has joined #automotive | 08:23 | |
*** psnsilva has joined #automotive | 08:24 | |
*** fredw has joined #automotive | 08:24 | |
*** GuyElpotato has quit IRC | 08:42 | |
*** thaytan has joined #automotive | 08:51 | |
*** fredw has quit IRC | 08:58 | |
fury | OutBackDingo: it works on my 3B+ | 09:12 |
fury | Hi guys | 09:12 |
fury | How's it hangin? | 09:12 |
*** GuyElpotato has joined #automotive | 09:17 | |
dl9pf | khilman: there seems to be an issue with the m3ulcb and sstate-cache | 09:22 |
dl9pf | qemux86-64 does not need any emulation -> faster . qemuarm/qemuarm64 have some exec path with emulation. | 09:23 |
*** toscalix has joined #automotive | 09:31 | |
OutBackDingo | fury: they demo images they have for download work on ther 3B+, or you build your own | 10:22 |
fury | I built my own, but I tend not to stray very far from released demo images, so I would be surprised if the released demo image didn't work on a 3B+. Main consideration is the Pi doesn't have enough memory to run all the apps at full res | 10:24 |
fury | so it's set to 1280x720 by default | 10:25 |
fury | I will give the latest one a shot when I get in, I've been meaning to sync up anyway | 10:26 |
fury | I have a custom image that does landscape 800x480 I've been using for a while (flounder) | 10:26 |
dl9pf | OutBackDingo: the image should work on both types of the rpi3 . If not, please file a bug at jira.automotivelinux.org . Tnx. | 10:36 |
OutBackDingo | fury: i tried building for the pi based on the doc but im hitting No package 'appcontroller' found | 10:36 |
OutBackDingo | -- Checking for module 'appcontroller' | 10:37 |
OutBackDingo | | -- No package 'appcontroller' found | 10:37 |
OutBackDingo | | CMake Error | 10:37 |
OutBackDingo | oops... sorry | 10:37 |
fury | hmmm | 10:37 |
fury | which doc you looking at? | 10:37 |
OutBackDingo | https://wiki.automotivelinux.org/agl-distro/agl-raspberrypi | 10:38 |
OutBackDingo | building from master on Debian 9 | 10:38 |
OutBackDingo | fury: ^ | 10:38 |
fury | oof, slightly out of date | 10:38 |
OutBackDingo | fury: is there better? | 10:39 |
fury | docs.automotivelinux.org might be a better place to start, but that's also WIP (I'm trying to find some time to help, just got lots of other stuff I gotta do as well) | 10:39 |
dl9pf | OutBackDingo: are you on master ? | 10:39 |
dl9pf | then do a 'repo sync --force-sync -d' ... a fix for the build issue was just merged | 10:40 |
OutBackDingo | dl9pf: yes | 10:40 |
OutBackDingo | so, as a curiousity, looking for basic voice / navigation capbilities for prototyping a solution... any suggestions ? | 10:45 |
OutBackDingo | as i highly doubt this is going to work out of the box :) | 10:45 |
fury | That's a good question! I'd love to know too. I'm a little rusty. Last time I tried the navigation demo app it was the one where I had to download the maps separately and add to the SD card after flashing, was just a mock-up for UK or Japan (fixed location with the ability to simulate drives). Don't know if that's still the latest nav demo | 10:56 |
fury | I think the cluster demo uses the Qt mapviewer demo which looks a little smoother but may require an internet connection to download maps on the go | 10:57 |
fury | I haven't tried any voice stuff tho | 10:57 |
OutBackDingo | fury: can you decipher this ? https://pastebin.com/ykHkYKih | 10:59 |
fury | OutBackDingo: might be nothing, I've seen those when I build but it still works. I think what it means is it won't work with the standalone SDK as configured | 11:17 |
fury | but when done through bitbake it's fine | 11:17 |
fury | These demo apps typically aren't built separately anyway | 11:17 |
fury | Someone working on a new one would configure it with the right app template | 11:18 |
dl9pf | OutBackDingo: there is an voice demo which can currently use alexa as backend. | 11:30 |
*** khouloud has joined #automotive | 11:55 | |
khouloud | > dl9pf Hello, this patch was merged but the code didn't that's why we are not receiving any test email report | 12:19 |
khouloud | https://gerrit.automotivelinux.org/gerrit/#/c/21082/ | 12:19 |
dl9pf | khouloud: I did https://gerrit.automotivelinux.org/gerrit/#/c/21234/ but (!) : | 12:21 |
dl9pf | lavacli -i $lab results $job_id > test_plan.raw | 12:21 |
dl9pf | if [ $? -eq 0 ]; then | 12:21 |
dl9pf | this will fail b/c when $? is not 0 jenkins will always about (e.g. run test script under 'set -e' ) | 12:21 |
dl9pf | TDLR 'set -e' means abort bash script as soon as one command fails | 12:22 |
dl9pf | the effect is that we actually never reach the if at all ... bash will fail and the job will fail already when lavacli returns -1 | 12:23 |
dl9pf | workaround: set +e before and set -e after that section. | 12:24 |
khouloud | > dl9pf oky will do that | 12:24 |
dl9pf | better we need to make sure we don't rely on the return value where possible. | 12:24 |
dl9pf | note: set +e is only a workaround | 12:24 |
montjoie | dl9pf: relying or not on the return code will not change the fact that lavacli will return a non 0 value and so will be aborted by set +e | 12:32 |
montjoie | but we could add some || true and check only the content | 12:33 |
montjoie | oups didnt read the right code | 12:35 |
dl9pf | yes, || true is an option in that case | 12:35 |
dl9pf | but if something does return -1 and we do || true , the $? is 0 always. so we need to change the way we detect it. | 12:36 |
montjoie | so you want to not crash is lavacli results fail, right ? | 12:37 |
*** claneys has joined #automotive | 12:47 | |
dl9pf | montjoie: khouloud: the question: is do we need to fail the *whole* job if we cannot retrieve the result or not at this point. E.g. can we retry/recover/go on with less info. | 13:01 |
dl9pf | the problem is that failing the jenkins job means we fail the CI run as well. | 13:01 |
montjoie | in that case do we care about the inability of getting results ? | 13:02 |
montjoie | I rewrite: does we want to be aware/alerted | 13:03 |
montjoie | adding || true remove that possibility | 13:03 |
montjoie | oups thinking just after write, perhaps we could check the output before grepping the testplan | 13:04 |
dl9pf | that is the question now. so here are 2 cases as example | 13:05 |
dl9pf | a) if the lava job has failures, we will want to fail CI | 13:05 |
dl9pf | b) if it is for the email only, we might retry, leave a warning but allow CI to exit successfully | 13:05 |
montjoie | yes we can definitively add || true and verify the file was not empty | 13:05 |
dl9pf | a1) lava failures should only be lava-test failures , lava infra failures we want to notify but not fail CI (atm to not spam developers with -1 due to infra) | 13:06 |
OutBackDingo | fury: ok ive got it built now... agl-demo-platform-raspberrypi3-20190502115937.rootfs.wic.xz and agl-demo-platform-raspberrypi3-20190502115937.rootfs.tar.xz are there i see no sdimg .... | 13:10 |
OutBackDingo | ahhh xzcat [output image] | sudo dd of=[sdcard device] bs=4M status=progress | 13:10 |
OutBackDingo | wic it is :) | 13:11 |
OutBackDingo | fury: ok.. touchscreen overscan ... ughhh ... so i guess i need to edit boot/config.txt ? | 13:32 |
fury | OutBackDingo: seems like it yes, though that's odd, i didn't have to do that to mine | 13:33 |
fury | what touchscreen you using? | 13:33 |
OutBackDingo | emmm.... works fine with most others | 13:36 |
OutBackDingo | fury: its a usb touchscreen | 13:36 |
OutBackDingo | well USB powered DVI | 13:36 |
*** fredw has joined #automotive | 13:39 | |
OutBackDingo | fury: ok u-boot on screen .... pffffft just sitting there | 13:42 |
OutBackDingo | its not booting | 13:45 |
GuyElpotato | use the command boot when you are on the u-boot prompt | 14:11 |
*** fredw has quit IRC | 14:52 | |
*** kbroulik has quit IRC | 15:07 | |
fury | odd. haven't seen it do that before. what were the options you passed in to aglsetup.sh? | 15:16 |
*** clement has joined #automotive | 15:18 | |
fury | last one i used was source meta-agl/scripts/aglsetup.sh -m raspberrypi3 -b build-pi agl-audio-4a-framework agl-devel agl-appfw-smack agl-demo | 15:18 |
fury | i recall having to add agl-netboot or something like that to get the minnowboard to work | 15:19 |
fury | but not the pi | 15:19 |
*** khouloud has quit IRC | 15:31 | |
OutBackDingo | fury: are we sure this works on 3B+ | 15:44 |
khilman | mranostay: yes, build error on agl-service-audio-4a (sorrygot disconected, so didn't see if you explained in more detail) | 15:54 |
khilman | is this a known issues? It's happening in ci/jenkins and also locally for me | 15:55 |
*** AlisonChaiken has quit IRC | 16:16 | |
fury | OutBackDingo: reasonably sure...I've been using it on a 3B+ for a few months. leon-anavi any ideas? | 16:18 |
fury | I still haven't gotten my build machine set back up for latest master yet, though, so it isn't out of the question for something to have broken with it on master... | 16:21 |
fury | trying to get the image stuff built into a GitLab CI process so it's all automatic the next time I have to rebuild that machine | 16:21 |
fury | or I could just give in and set up Jenkins like the AGL CIAT :) | 16:24 |
fury | but I likes my one tool to rule them all! | 16:25 |
fury | GitLab needs a bit more work before it's capable of doing a full build from source thing like building AGL or AOSP | 16:26 |
fury | not that it can't (any shell script can be run), it's just not got specific support for workspaces so the question of caching work and getting the artifacts out the other end is not yet answered I guess | 16:27 |
leon-anavi | fury, I've built rpi3b/b+ image of AGL master recently. It built and booted as expected. | 17:03 |
OutBackDingo | leon-anavi: well both the one i built and the one i downloaded agl-demo-platform-crosssdk-raspberrypi3-20190501032140.rootfs.wic.xz both hang at u-boot just sitting at the u-boot prompt | 17:08 |
khilman | dl9pf: another ci-management patch seems to have gone missing: https://gerrit.automotivelinux.org/gerrit/#/c/21062/ | 17:08 |
khilman | shows merged, but is not in the repo | 17:08 |
khilman | that's what's preventing the email reports from being requested | 17:09 |
OutBackDingo | leon-anavi: does anythign require editing in cmdline.txt or config.txt ? | 17:09 |
OutBackDingo | leon-anavi: so if you have an image i can test :) | 17:10 |
OutBackDingo | fury: pretty confident they broke something | 17:11 |
fury | OutBackDingo: i'm downloading right now to try it, the only thing that isn't fast about this machine is its internet connection :X | 17:18 |
OutBackDingo | i downloading this one from a month ago to test it also https://download.automotivelinux.org/AGL/release/guppy/7.0.1/raspberrypi3/deploy/images/raspberrypi3/agl-demo-platform-crosssdk-raspberrypi3-20190412122811.rootfs.wic.xz | 17:19 |
leon-anavi | OutBackDingo, which rpi model are you using? | 17:19 |
fury | 3B+ | 17:20 |
OutBackDingo | leon-anavi: 3B+ | 17:20 |
leon-anavi | can you file in a JIRA and assign it to me? | 17:20 |
leon-anavi | please specify which exact release of AGL are you using and if possible attach logs from the serial output | 17:21 |
OutBackDingo | leon-anavi: im on master as of today also | 17:21 |
OutBackDingo | itll be in the morning after i get a serial on it, right now ive got a 7" LCD on it | 17:21 |
OutBackDingo | ill see if the previous image from a montyh ago works in about 5 minutes burning it now | 17:22 |
fury | i would give my left arm for a serial cable that attaches right to the header without me having to figure out which pin is which. i got spoiled by the TTL-232R-3V3 cable for the minnowboard... | 17:22 |
leon-anavi | OutBackDingo, ok. I have to leave for a dinner (it's 20:22 local time :) ). Pls submit the JIRA issue when you can. | 17:23 |
OutBackDingo | leon-anavi: will do | 17:23 |
leon-anavi | OutBackDingo, pls list all peripherals that you have attached to the rpi3b+ | 17:24 |
leon-anavi | last time I booted the image on rpi3b with HDMI monitor and it seems that your setup is different | 17:24 |
OutBackDingo | leon-anavi: nothing just the HDMI display / usb power for it | 17:24 |
leon-anavi | hm, ok | 17:24 |
*** leon-anavi has quit IRC | 17:34 | |
fury | 3224 of 6821. halfway there, living on a prayer | 17:54 |
fury | i probably should've just copied my old downloads folder over to my new drive | 17:54 |
fury | bet most of these are already cached in there | 17:55 |
fury | but then again, most of that folder is probably also obsoleted downloads | 17:55 |
OutBackDingo | fury: i just confirmed raspbian boots fine on the board, AGL hangs at u-boot, let me know how you make out | 17:57 |
*** ToxicGumbo-work has quit IRC | 18:26 | |
fury | almost there. flashing sd card | 19:40 |
fury | weirdly enough my build of it is 367.9 mb but the full extracted sd card image is 2.3 gigs, the one i'm downloading from the AGL snapshot releases is 661 mb | 19:41 |
fury | oh, crosssdk, that's why | 19:41 |
fury | saves a few hundred megs to not include the crosssdk part | 19:41 |
fury | OutBackDingo: mine is booting up, so i guess now the question is why'd it not boot up for you? can you show me what xzcat line looked like? | 19:51 |
*** toscalix has quit IRC | 19:51 | |
fury | OutBackDingo: here's mine, for reference: https://pastebin.com/vYv1xx4J | 19:57 |
fury | my site.conf just points sstate cache and downloads to the right dir and tells bitbake to use 40 threads | 19:58 |
fury | nothing too fancy | 19:58 |
fury | my game server beat me, but bloodman just killed my vibe | 20:45 |
fury | aaaaand wrong channel | 20:45 |
*** khem has quit IRC | 21:35 | |
*** khem has joined #automotive | 21:36 | |
banana | o: | 22:07 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!