*** Wetmelon has quit IRC | 01:00 | |
*** thaytan has joined #automotive | 01:25 | |
*** Newami has joined #automotive | 01:55 | |
*** Newami has quit IRC | 02:08 | |
*** vrubiolo has joined #automotive | 06:18 | |
*** vrubiolo has quit IRC | 06:41 | |
*** leon-anavi has joined #automotive | 07:06 | |
leon-anavi | hi | 07:10 |
---|---|---|
*** rpcme has quit IRC | 08:09 | |
*** vrubiolo has joined #automotive | 08:17 | |
*** vrubiolo has quit IRC | 08:22 | |
*** vrubiolo has joined #automotive | 08:38 | |
*** vrubiolo has quit IRC | 09:09 | |
*** eduardas has joined #automotive | 09:32 | |
*** vrubiolo has joined #automotive | 09:38 | |
*** Newami has joined #automotive | 09:56 | |
*** Newami has quit IRC | 10:02 | |
*** vrubiolo has quit IRC | 10:49 | |
*** vrubiolo has joined #automotive | 10:50 | |
cengiz_io | hello there! | 11:01 |
cengiz_io | hope you're having a great day | 11:01 |
*** vrubiolo has quit IRC | 11:02 | |
*** vrubiolo1 has joined #automotive | 11:02 | |
*** vrubiolo1 is now known as vrubiolo | 11:04 | |
*** adriano has joined #automotive | 12:08 | |
adriano | hello all, i have just built (bitbake agl-telematics-demo-platform) for raspberry pi 3, and now i try to use devtool to deploy some canges | 12:09 |
adriano | *changes | 12:10 |
adriano | how to setup wifi on raspberry pi? I tried adding a wpa_supplicant.conf, with the credentials | 12:10 |
adriano | # ip link set wlan0 upip: SIOCSIFFLAGS: Operation not possible due to RF-kill | 12:11 |
cengiz_io | adriano what does `ip a` command show for you | 12:28 |
cengiz_io | adriano wlan0 interface is available? | 12:28 |
adriano | raspberrypi3:~# ip a1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue qlen 1000 link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 scope host lo valid_lft forever preferred_lft forever inet6 ::1/128 scope host valid_lft forever preferred_lft forever2: wlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop | 12:29 |
adriano | qlen 1000 link/ether b8:27:eb:ef:21:df brd ff:ff:ff:ff:ff:ff3: can0: <NOARP,UP,LOWER_UP> mtu 72 qdisc noqueue qlen 1000 link/[280] | 12:29 |
cengiz_io | adriano what does `rfkill list all` show for you | 12:29 |
adriano | raspberrypi3:~# rfkill list all0: phy0: wlan Soft blocked: yes Hard blocked: no1: hci0: bluetooth Soft blocked: yes Hard blocked: no | 12:30 |
cengiz_io | rfkill unblock wifi | 12:32 |
cengiz_io | or whatever the identifier is | 12:32 |
cengiz_io | then you will need to figure out WHY it's blocked. | 12:33 |
cengiz_io | might be due to kernel module arguments | 12:33 |
cengiz_io | or the hackiest way would be executing rfkill unblock on every boot | 12:34 |
cengiz_io | which is absolutely terrible | 12:34 |
adriano | raspberrypi3:~# ifup wlan0Successfully initialized wpa_supplicantrun-parts: /etc/network/if-up.d: No such file or directory | 12:38 |
adriano | ok seems to me that the interface was disabled on purpose | 12:39 |
adriano | i am new to running agl, will look into it\ | 12:39 |
cengiz_io | good luck! | 12:39 |
adriano | thank you! | 12:43 |
adriano | agl should also work on Raspberry Pi 3a+, right? | 12:47 |
cengiz_io | I have no idea adriano. not experienced with raspberyy | 13:09 |
cengiz_io | can someone tell me why there are no Qt libs inside `agl-image-graphical-qt5` sample image? | 13:13 |
cengiz_io | what does it do for Qt then? | 13:13 |
cengiz_io | I'm on jellyfish 10.0.0 | 13:13 |
smurray | adriano: at this point AGL really only supports rpi 3 for the telematics and cluster demos | 13:15 |
cengiz_io | hello smurray | 13:17 |
smurray | cengiz_io: I'd need to experiment a bit to see why that is. As I mentioned previously, nothing tests that image | 13:20 |
smurray | cengiz_io: If you're not adding your own package that had Qt bits as dependencies, you'll need to add packagegroup-agl-profile-graphical-qt5 to the image, I think | 13:21 |
cengiz_io | smurray ok. I don't want to bother you but my core problem is: I have a qt application that requires wayland-egl to run. so basically I need a compositor running (preferably by systemd at boot) with correct user as AGL conventions intended | 13:22 |
smurray | cengiz_io: look at the cluster demo, it does that | 13:22 |
cengiz_io | smurray previously on icefish, I've reached to a point where weston@display.service was failing because user "display" was not found in /etc/passwd. | 13:22 |
cengiz_io | I'll look right now | 13:23 |
cengiz_io | smurray meta-agl-demo repo or meta-agl-cluster-demo? | 13:25 |
smurray | cengiz_io: meta-agl-cluster-demo is simpler | 13:25 |
cengiz_io | ok it requires require agl-cluster-demo-qtcompositor.inc and so on | 13:26 |
smurray | cengiz_io: the display user is created by something in the core, iirc | 13:26 |
smurray | cengiz_io: no, the qtcompositor stuff is for running with the Qt compositor and not weston | 13:26 |
cengiz_io | oh | 13:27 |
cengiz_io | smurray this one then? https://gerrit.automotivelinux.org/gerrit/gitweb?p=AGL/meta-agl-cluster-demo.git;a=blob;f=recipes-platform/images/agl-cluster-demo-platform.bb;h=52d737b74e8146482e4c3a7c408f451cde72e4cd;hb=HEAD | 13:28 |
smurray | cengiz_io: the packagegroup is more useful | 13:29 |
cengiz_io | can I use that packagegroup without enabling agl-demo ? | 13:30 |
cengiz_io | don't think so | 13:30 |
smurray | cengiz_io: no | 13:30 |
smurray | cengiz_io: I meant to look at it to see what packages/packagegroups it's pulling in | 13:30 |
cengiz_io | ok | 13:31 |
smurray | cengiz_io: that was probably not a great idea, I forgot that we ended up with a separate profile for cluster | 13:31 |
cengiz_io | xD | 13:31 |
cengiz_io | smurray let me get this straight: even though I require a profile in my image, I need to explicitly install_append some packagegroups. | 13:32 |
smurray | cengiz_io: this stuff isn't perfect, I'd need to go back and look in the git history, we had to hack a bit at one point to make that qtcompositor demo that doesn't use weston work | 13:33 |
smurray | cengiz_io: and as I've said repeatedly, no one's tested using agl-image-graphical-qt5 directly that I can recall | 13:33 |
cengiz_io | ok thank you. I'll be trying harder. | 13:34 |
smurray | cengiz_io: if you just want to see the libraries add packagegroup-agl-profile-graphical-qt5. The other images manage to pull them in because their applications have the dependencies, I think | 13:34 |
cengiz_io | what about this? DISTRO_FEATURES_append = " agl-core-image-profile" | 13:35 |
smurray | that's not going to do much | 13:36 |
cengiz_io | for missing display user? | 13:36 |
smurray | cengiz_io: the display user is created by the agl-users package, you'll need to add that to your custom image, see what's done in meta-agl-demo or meta-agl-cluster-demo | 13:36 |
cengiz_io | is it that easy? omg | 13:36 |
smurray | took me a while to find that, haven't had coffee yet | 13:36 |
smurray | for historical reasons, that's not done in the core, as someone might want different users than the demo images use | 13:37 |
smurray | it does make that agl-image-graphical-qt5 image unusable by default, though. Like I said, not tested on its own | 13:37 |
cengiz_io | is there a list of agl-* packages that are essential? | 13:38 |
smurray | nope | 13:38 |
cengiz_io | by blood and pain you say | 13:38 |
smurray | well, the usual recommendation is to start with the demo platform image and pare it down | 13:38 |
smurray | starting directly from the plain profile images isn't for beginners | 13:39 |
cengiz_io | AGL/meta-agl-demo since you said AGL/meta-agl-cluster-demo has a different profile. | 13:39 |
smurray | I've actually recommended we nuke them in the upcoming layer reorg, since they're not tested | 13:39 |
smurray | right, that's another historical artifact I would have avoided if possible, but could not, sadly | 13:39 |
cengiz_io | great. printing out all of AGL/meta-agl-demo files *jk* | 13:40 |
smurray | if you add agl-users to the image you have now, you'll be good, I think, no need to restart | 13:40 |
smurray | I'll make a note, that particular requirement will need to change for the upcoming layer reorg, the display user should probably be created elsewhere | 13:41 |
cengiz_io | noted. | 13:42 |
cengiz_io | one last question before you shoot me: what do you suggest BBFILE_PRIORITY_meta-localdev ? | 13:42 |
cengiz_io | it's 20 now | 13:42 |
smurray | I've no idea, tbh, kind of depends on whether you think you'll need to over-ride some things | 13:45 |
smurray | and note that's sort of intended for temporary local development, if you're going further, you'd likely take that layer and add it to the manifest in a local fork of AGL-repo | 13:46 |
smurray | and probably move it off of the meta-localdev name | 13:47 |
*** adriano has quit IRC | 13:56 | |
cengiz_io | smurray got it. I was following https://events19.linuxfoundation.org/wp-content/uploads/2018/07/ALS_jsmoeller_july_2019.pdf | 14:03 |
cengiz_io | Creating Your Own Project on-top of AGLFrom minimal image to own project. How to reuse AGL in your own company.Jan-Simon Möller, jsmoeller@linuxfoundation.org | 14:03 |
smurray | cengiz_io: so I was slightly wrong re agl-users, you'll actually want to pull in packagegroup-agl-core-security (which includes agl-users) if you're going to use the application framework | 14:04 |
smurray | cengiz_io: I can't remember why things are like that :/ | 14:05 |
cengiz_io | smurray no idea what application framework is but sure. all I want is weston to be running at boot | 14:05 |
cengiz_io | I can perfectly come up with my own systemd units but why reinvent? | 14:05 |
smurray | cengiz_io: heh, okay, then just agl-users is likely fine for now, that'll get the display user | 14:05 |
*** cengiz_io has quit IRC | 14:10 | |
*** cengiz_io has joined #automotive | 14:11 | |
*** cengiz_io has quit IRC | 14:12 | |
*** cengiz_io has joined #automotive | 14:13 | |
*** adriano has joined #automotive | 14:21 | |
*** rpcme has joined #automotive | 14:32 | |
dl9pf | probably need to redo that slidedeck ... a lot has changed | 14:33 |
cengiz_io | hello dl9pf. I'm following that project-manifest method. works great so far. thanks | 14:35 |
rpcme | smurray - it is the agl-demo-platform target | 14:35 |
smurray | rpcme: what services are failing? | 14:36 |
rpcme | I have to turn the equipment on. need to work on gps this morning | 14:38 |
rpcme | give me a few minutes please | 14:38 |
dl9pf | rpcme: uae pi4. | 14:38 |
dl9pf | pi3 is too weak for the agl-demo-platform now | 14:39 |
dl9pf | tests only done on pi4 | 14:39 |
dl9pf | cengiz_io: yes, that will do fine. | 14:40 |
smurray | dl9pf: that was adriano | 14:40 |
rpcme | also, if someone has good documentation pointers on configuring SMACK for apps working backwards from access requirements, I would love to see it because right now I am using the virtual sledgehammer to grant system access which doesn't smell good at all | 14:40 |
dl9pf | ok, need coffee. | 14:40 |
smurray | rpcme: you'll need to elaborate, do you mean for access to hardware devices? | 14:40 |
rpcme | smurray: filesystem. More specifically, this is what I did to overcome the startup issue. sed -i 's/System _ -----l/System _ -wx--l/g' /etc/smack/accesses.d/default-access-domains | 14:48 |
*** d__ep__th has joined #automotive | 14:49 | |
smurray | rpcme: when you say "apps", what do you mean, exactly? That definitely should not be required for application framework apps installed from a widget file | 14:49 |
smurray | rpcme: except if they access stuff outside their install directory | 14:49 |
rpcme | smurray: Installing and running AWS IoT Greengrass. | 14:49 |
smurray | rpcme: okay, so some system daemon or the like. If it's running as root, it'd probably be okay | 14:51 |
smurray | rpcme: if it's not, you'll need to create a SMACK configuration for it, I think | 14:51 |
rpcme | the daemon is running as root, but the cgroups are running as specific user/group | 14:52 |
rpcme | I get the feeling the access permission trouble with that then | 14:52 |
smurray | cgroups? | 14:52 |
rpcme | yes | 14:52 |
smurray | it creates some and runs stuff in them? | 14:52 |
rpcme | yes | 14:52 |
smurray | you'll need to investigate how to come up with a SMACK configuration for that | 14:53 |
rpcme | yes, I need to make it specific. I don't like the sledgehammer. | 14:54 |
rpcme | and i need to find a way to automate that for AGL or yocto w SMACK specifically and add it to the public recipe ... that will be interesting | 14:54 |
rpcme | Here is the list of errors and audit failures from the boot on the renesas H3 board https://gist.github.com/rpcme/683a1ba6bade20e077e7efbd6c647c09 | 14:55 |
rpcme | I have a task to go through them one by one but if anything stands out would be grateful to hear it. Need to finish up some gps work now. | 14:56 |
smurray | unicens you can ignore, it fails if the h/w isn't present, which is the case for pretty much everyone | 14:56 |
smurray | Is that with or without that default-access-domains change? | 14:57 |
smurray | some services will fail if you don't have the h/w, like the unicens, I wouldn't expect stuff like homescreen to fail | 14:58 |
*** adriano has quit IRC | 15:00 | |
d__ep__th | testing from weechat | 15:01 |
d__ep__th | works | 15:01 |
*** adriano has joined #automotive | 15:15 | |
*** d__ep__th has quit IRC | 15:16 | |
*** adriano has joined #automotive | 15:16 | |
*** adriano has quit IRC | 15:18 | |
rpcme | smurray: it was failing like that prior to the change | 15:22 |
rpcme | also, the audit failures (I believe) didn't happen until I enabled the wireless network dongle | 15:23 |
smurray | I definitely can't guarantee that there are no failures during a normal boot up | 15:27 |
smurray | that's not been a big concern in the demo images | 15:27 |
smurray | if you're not getting homescreen started, that would be a real problem | 15:28 |
smurray | I'd need the full journal to maybe able to get an idea of what's happened | 15:28 |
*** d__ep__th has joined #automotive | 15:34 | |
rpcme | Ok. it might be some time. Will come back when I have it organized. | 15:38 |
*** adriano has joined #automotive | 15:46 | |
*** d__ep__th has quit IRC | 16:05 | |
*** rpcme has quit IRC | 16:45 | |
*** leon-anavi has quit IRC | 16:57 | |
*** leon-anavi has joined #automotive | 16:57 | |
*** adriano has quit IRC | 17:37 | |
*** eduardas has quit IRC | 20:11 | |
*** leon-anavi has quit IRC | 20:42 | |
*** vrubiolo has quit IRC | 22:00 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!