pieq | Morning 'sktoppers! | 00:33 |
---|---|---|
pieq | duflu: hey! Yesterday I spotted something fishy with the latest Jammy daily build on a Intel+Nvidia GPU laptop: the nvidia drivers don't seem to be activated even after selected "3rd party drivers", and after activating them, the GNOME session used is X11. If selecting Wayland, the nvidia GPU is not available when starting programs and nvidia-settings become almost empty. | 01:49 |
pieq | duflu: do you know against what package I should file an `ubuntu-bug`? | 01:49 |
pieq | P.S. it's a device certified with 20.04... 😱 | 01:50 |
duflu | pieq, not sure. It depends whether we tried to install the Nvidia driver or not | 01:52 |
duflu | pieq, gnome-shell doesn't actually use secondary GPUs. So how are you judging not activated? | 01:54 |
pieq | duflu: in the "About" panel in GNOME Settings, if I'm using X11 session, I can see both Nvidia and Intel GPUs mentioned in the "Graphics" section. If I'm using Wayland, only Intel GPU shows up. | 02:07 |
sarnold | pieq: there might be something mentioned on https://bugs.launchpad.net/ubuntu/+source/egl-wayland/+bug/1935082 that might help explain the nvidia / wayland thing | 02:08 |
ubottu | Launchpad bug 1935082 in nvidia-graphics-drivers-495 (Ubuntu) "[MIR] egl-wayland" [Medium, In Progress] | 02:08 |
pieq | duflu: during install, I opened a terminal to check the install logs, and I can see a bunch of nvidia-* packages being downloaded, so I assume 3rd party driver checkbox did something. However, after install, when I go to "Additional drivers", I see that the driver being used is nouveau | 02:08 |
pieq | This laptop is using a Quadro nvidia GPU, which apparently is not compatible with nvidia-470, so I have to manually select nvidia-495. After doing so and rebooting in an X11 session, I can launch nvidia settings and see all the expected info | 02:09 |
duflu | pieq, ignoring the Settings app, it sounds like the only problem is the missing right-click menu option for launching on the secondary GPU in your Wayland session? | 02:10 |
duflu | pieq, although yeah 'nouveau' being active is a big problem. Please open a generic Ubuntu bug and include output from 'lspci -k' | 02:12 |
pieq | duflu: so `ubuntu-bug ubuntu`? | 02:17 |
pieq | I'm downloading the very last daily build just to make sure, and will file the bug afterwards. | 02:17 |
duflu | pieq, actually we want some graphics info so it might be fastest to start with 'ubuntu-bug gnome-shell' | 02:18 |
duflu | or xorg | 02:18 |
pieq | got it | 02:29 |
pieq | duflu: I guess activating WiFi during install will have an impact on the result. Should I activate it or not? | 02:42 |
duflu | pieq, while I don't myself, we do want a bug report only from the latest packages | 02:43 |
duflu | But if you get different results with or without wifi then that's important too | 02:43 |
pieq | duflu: ok, I'll start by activating it and see what happens, and then I might try again without, just for the sake of it. | 02:44 |
pieq | duflu: by the way, do you know exactly what packages are installedwhen users tick the "Install third-party software for graphics and Wi-Fi hardware and additional media formats"? | 02:45 |
pieq | My guess is that it installs nvidia proprietary drivers if needed, WiFi proprieratry drivers, and also non-free codecs | 02:46 |
pieq | but depending on who I ask, I get different answers :D | 02:46 |
duflu | pieq, your guess is what I believe | 02:47 |
duflu | In the distant past I've had trouble on Nvidia-only machines, or machines with Broadcom wifi that really need it. But it's also needed to get VAAPI acceleration out of the box | 02:48 |
pieq | When I recommend devices to friends and family, I tend to avoid nvidia completely (either by using Intel-only GPUs for lighter loads, or AMD GPUs if needed), but I also know a lot of gamers and 3D artists as well as data scientists rely heavily on it (be it for specs or CUDA features). | 02:53 |
duflu | pieq, to be fair, I don't see bug reports from the Nvidia proprietary driver. But I do see a lot from 'amdgpu', which is open source and included in the kernel | 03:00 |
duflu | Although I am not subscribed to the Nvidia packages either | 03:00 |
pieq | duflu: from my experience working in Devices QA, I can tell you the integration of nvidia proprietary blobs is painful | 03:12 |
pieq | duflu: ok so the latest daily (20220126) seems to fix a bunch of stuff. nvidia-470 are installed and in use, nvidia-settings is working (and the PRIME profile selected is "On Demand", which is what we like as a default) | 03:12 |
pieq | however: (1) I don't see the right click option to start with the discrete GPU, and (2), it defaults to X11 session. Let me try logging in Wayland session... | 03:13 |
duflu | Defaulting to Xorg also indicates the Nvidia driver is in use | 03:15 |
pieq | oh! I thought Wayland worked with nvidia drivers now... | 03:23 |
pieq | duflu: so defaulting to xorg is the actual expected behavior when running with proprietary nvidia drivers? | 03:24 |
duflu | pieq, it's an indicator that Nvidia is installed. And it's probably sane because Nvidia's Wayland support is still relatively new. Even recently they rewrote it... I should check to see if we have that new path enabled in mutter 41. In fact that might be relevant here | 03:25 |
pieq | duflu: so should I still check stuff out on Wayland? | 03:28 |
pieq | side question regarding Wayland: in the past, I noticed Firefox required specific tricks to run as a Wayland native app instead of an XWayland app. Fedore had patched their launcher to include this. Do you know if there has been any improvement on that front to get some kind of autodetection depending on what environment is being used? | 03:29 |
pieq | crap. I switched from "On Demand" to "Performance" and back, and restarted the device, and now the session options I have are "Ubuntu" and "Ubuntu on Xorg" instead of "Ubuntu" and "Ubuntu on Wayland". If I select "Ubuntu" it boots into a Wayland session, so nvidia is not active (nothing in nvidia-settings), and if I boot in "Ubuntu on Xorg", gnome-shell crashes | 03:45 |
duflu | pieq, a bug report would help so I can see the hardware details. I'm lost trying to visualise things in IRC. | 04:05 |
duflu | pieq, Firefox has been Wayland-native for quite some time already. The way I used to check that is when it is absent from xwininfo and xprop | 04:07 |
pieq | duflu: https://twitter.com/FirefoxNightly/status/1485901776255692802 just saw that, so I guess it's not ready yet in stable | 04:08 |
duflu | Oh. On Ubuntu I thought it was native for some time already. Haven't verified recently | 04:09 |
pieq | duflu: another colleague is checking something on the device, I will file a bug later | 04:18 |
=== ahayzen_ is now known as ahayzen | ||
pieq | duflu: LP #1959362 | 06:11 |
ubottu | Launchpad bug 1959362 in gnome-shell (Ubuntu) "'Launch using dedicated graphics card' option absent (nvidia, proprietary drivers, on-demand PRIME mode)" [Undecided, New] https://launchpad.net/bugs/1959362 | 06:12 |
pieq | duflu: should I also file a bug upstream? | 06:12 |
duflu | pieq, not yet | 06:12 |
pieq | (and should I attach additional logs?) | 06:12 |
duflu | Don't know | 06:12 |
duflu | pieq, can you update the system? jammy is on gnome-shell 41 now | 06:14 |
pieq | duflu: I'll check with my Taipei colleagues | 06:14 |
pieq | duflu: updating now. I thought such a recent image would have everything up to date... I was wrong :) | 06:15 |
duflu | pieq, it was literally only published 24 hours ago | 06:15 |
duflu | pieq, please also add output from 'lspci -k' and 'journalctl -b0' | 06:15 |
pieq | ok | 06:18 |
pieq | duflu: aaaaand the option appears | 06:23 |
duflu | in 41? :) | 06:26 |
duflu | pieq, also relevant: nvidia-495's gbm backend support is enabled only in mutter 41.2 onward. I'm not sure if that has made multi-GPU support better or worse (less tested) though | 06:30 |
duflu | Oh it's using nvidia-470 | 06:31 |
jibel | good morning all | 06:34 |
duflu | Morning jibel | 06:35 |
pieq | yes in 41.3 duflu | 06:38 |
duflu | My work here is done | 06:38 |
pieq | :) | 06:39 |
jibel | Hi duflu | 06:39 |
pieq | salut jibel ! | 06:40 |
jibel | Salut pieq | 06:47 |
didrocks | good morning | 06:52 |
didrocks | hey jibel, pieq | 06:52 |
pieq | salut didrocks ! Ça va ? | 06:53 |
duflu | Morning didrocks | 06:53 |
oSoMoN | good morning desktoppers, happy Friday! | 06:54 |
didrocks | salut pieq, ça va, et toi ? | 06:54 |
didrocks | hey duflu, salut oSoMoN | 06:54 |
didrocks | bon vendredi à toi aussi :) | 06:55 |
pieq | didrocks: ça va ! Lunar New Year holidays start tonight! \o/ | 06:56 |
oSoMoN | salut didrocks & pieq | 06:57 |
oSoMoN | hey duflu | 06:57 |
duflu | 🐯 | 06:57 |
duflu | (year of the tiger) | 06:58 |
pieq | indeed! | 06:58 |
didrocks | :) | 06:59 |
oSoMoN | I'm not fully awake yet, I initially read "eye of the tiger" :) | 07:03 |
pieq | :D | 07:03 |
jibel | Salut didrocks | 07:32 |
jibel | et oSoMoN | 07:32 |
oSoMoN | salut jibel | 07:33 |
jpnurmi | good morning | 07:59 |
duflu | Hi jpnurmi | 07:59 |
jpnurmi | hey duflu | 08:11 |
seb128 | goood morning desktopers | 08:33 |
oSoMoN | salut seb128 | 08:35 |
oSoMoN | good morning jpnurmi | 08:36 |
jpnurmi | good morning seb128 oSoMoN | 08:36 |
didrocks | salut seb128 | 08:40 |
seb128 | lut oSoMoN, didrocks, hey jpnurmi, how are you on this friday? | 08:40 |
seb128 | didrocks, dunno if you saw my backlog ping on wednesday evening that v4l2-relayd was back in the NEW queue for review with the issues addressed? | 08:41 |
duflu | Hi seb128 | 08:41 |
jpnurmi | good morning didrocks | 08:41 |
oSoMoN | seb128, a bit tired, but good otherwise. How are you yourself? | 08:41 |
seb128 | hey duflu, how are you? ready for the weekend? | 08:41 |
seb128 | oSoMoN, a bit tired as well but alright :) | 08:41 |
didrocks | hey jpnurmi | 08:41 |
duflu | seb128, yeah I guess. Working on a hard problem that won't be fixed this week. | 08:41 |
didrocks | seb128: good good :) oh I haven’t see your ping on wednesday evening, let me look | 08:42 |
seb128 | didrocks, thanks! | 08:42 |
jpnurmi | feeling good. our puppy is finally reaching the age that one can start doing longer kickbike trips with him. we did 5km yesterday and he really loves it. that guy was born to pull :) | 08:44 |
seb128 | duflu, I didn't check reports yet, no surprise with the gnome-shell 41 landing? | 08:44 |
duflu | seb128, 41 is good. I'm still in a made rush to finish 42 features | 08:44 |
duflu | mad rush even | 08:45 |
seb128 | when is GNOME feature freeze? I think beta is end of february | 08:45 |
seb128 | it's not like that the shell maintainers were respecting the freezes usually though... | 08:46 |
duflu | Around then | 08:47 |
duflu | seb128, arguably not a feature, just performance. So code freeze is March. Either way, if I miss GNOME-42 but have triple buffering ready around then I'll be strongly suggesting it as a patch. Lots of people have been using it as their daily for a year or so already and I am happy to support it | 08:51 |
seb128 | duflu, ack | 08:54 |
ackk | oSoMoN, hi, is the fix for https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/1849478 available in any snap build? | 08:54 |
ubottu | Launchpad bug 1849478 in chromium-browser (Ubuntu) "[snap] (experimental) pipewire support not available" [Wishlist, Fix Committed] | 08:54 |
=== ackk is now known as ack | ||
duflu | Hmm | 08:54 |
seb128 | didrocks, thanks! | 08:57 |
oSoMoN | ack, yes it is available in all the channels, but there are reports that it's not always fully working. If it's not working for you, please comment on the bug report with details on your host system | 08:58 |
didrocks | seb128: yw! checking binNEW now | 08:58 |
RikMills | morning | 08:59 |
RikMills | oSoMoN: may I quickly ask if someone has been looking at the 'native host connector' not working issue? | 08:59 |
oSoMoN | RikMills, yes, this is https://github.com/flatpak/xdg-desktop-portal/issues/655, and it is being worked on | 09:00 |
ubottu | Issue 655 in flatpak/xdg-desktop-portal "NativeMessaging portal for sandboxed browsers" [Open] | 09:00 |
RikMills | cool. thanks! | 09:01 |
ack | oSoMoN, the last comment on there is mine :) | 09:01 |
seb128 | j_amesh is actively working on it, he said he would get ready to push by eow | 09:01 |
RikMills | oSoMoN: so will that just work for plasma-browser-integration, or should someone from KDE etc chip in? | 09:04 |
ack | oSoMoN, LMK if I can provide more info on that | 09:06 |
oSoMoN | RikMills, the first step is to implement this new portal, and then to modify browsers to use it. When this is done, plasma-browser-integration and other extensions using native host messaging should just work. | 09:07 |
RikMills | ok. great :) | 09:08 |
seb128 | oSoMoN, do we expect having any UI prompting as part of the workflow? | 09:31 |
oSoMoN | seb128, I was initially envisioning some kind of prompt, but there was some pushback when this was discussed on the xdg-desktop-portal issue. Not sure what james_h chose to implement. | 09:32 |
seb128 | oSoMoN, alright, I was mostly asking because if there is an UI then kde might need to do some work on their own portal | 09:33 |
oSoMoN | true | 09:33 |
seb128 | RikMills, ^ tbd | 09:33 |
RikMills | the kde portal is already past feature freeze for the version we can have in 22.04 :( | 09:35 |
RikMills | i.e. the beta of it is already in jammy | 09:35 |
* RikMills ponders backporting/git-snapshots and FFEs | 09:39 | |
seb128 | RikMills, let's wait first to see if there is an UI or not | 09:41 |
RikMills | seb128: oh I know, :) | 09:46 |
slingamn | i got a pull request merged into the upstream system-config-printer: https://github.com/OpenPrinting/system-config-printer/pull/247 | 10:03 |
ubottu | Pull 247 in OpenPrinting/system-config-printer "remove python3-requests" [Merged] | 10:03 |
slingamn | i'm trying to determine the best way to get it to land in ubuntu 22.04 | 10:03 |
slingamn | should i ask upstream about whether they plan to include it in a tagged release? | 10:03 |
schopin | slingamn: hi again :) would this package be one of the last blockers for -request removal or just another step in the right direction? | 10:05 |
slingamn | i took a look at https://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.jammy/rdepends/requests/python3-requests | 10:05 |
slingamn | it's a little challenging to read but if i understand correctly, it will remove python3-requests from the default desktop image | 10:06 |
slingamn | it will still ship in the default server image as a dependency of cloud-init, but cloud-init (and its dependencies) can be removed | 10:06 |
slingamn | for example i have a 22.10 server with no cloud-init and no requests | 10:06 |
schopin | for context, python3-requests is a 50kB deb, ~250kB unpacked | 10:08 |
slingamn | it has a few more transitive dependencies, in particular python3-chardet and python3-idna | 10:08 |
schopin | but I don't know how pervasive its dependencies are. | 10:08 |
slingamn | it's not really about the file size for me, it's about memory consumption (it takes another 10 MB per process to import requests and its transitive dependencies) and also that i don't like that this large, messy package has become part of the critical infrastructure for desktop linux | 10:09 |
schopin | fwiw I think it would be a nice feature | 10:12 |
slingamn | ah, thanks :-) | 10:14 |
KGB-1 | deja-dup signed tags 4f67643 Sebastien Bacher ubuntu/42.8-2ubuntu2 * deja-dup Debian release 42.8-2ubuntu2 * https://salsa.debian.org/gnome-team/deja-dup/tags/ubuntu/42.8-2ubuntu2 | 10:18 |
KGB-1 | deja-dup ubuntu/master b1782c3 Sebastien Bacher debian/ changelog patches/series patches/git_newglib_compat.patch * backends: fix path parsing with glib 2.71+ * https://salsa.debian.org/gnome-team/deja-dup/-/commit/b1782c3 | 10:18 |
ubottu | Commit b1782c3 in gnome-team/deja-dup "backends: fix path parsing with glib 2.71+" | 10:18 |
KGB-1 | deja-dup ubuntu/master 2cd137d Sebastien Bacher debian/changelog * upload to ubuntu * https://salsa.debian.org/gnome-team/deja-dup/-/commit/2cd137d | 10:18 |
ubottu | Commit 2cd137d in gnome-team/deja-dup "upload to ubuntu" | 10:18 |
seb128 | didrocks, in fact the gnome portal is already listed on https://people.canonical.com/~ubuntu-archive/component-mismatches-proposed.html , does it require anything more than promotion now? | 10:51 |
seb128 | in the promotion to main section | 10:51 |
seb128 | slingamn, you should use https://bugs.launchpad.net/ubuntu/+source/system-config-printer/+filebug to request the change to be backported to the Ubuntu package before the LTS | 11:36 |
jbicha | seb128: hi, are you going to do an ubuntu-meta package upload for the GNOME portal addition? | 12:01 |
seb128 | jbicha, hey, I'm never clear on whether the tasks being updated is enough for the installer but we should probably do an ubuntu-desktop upload yes, I've mostly tried to be lazy and hoped that someone would do a refresh for some other changes | 12:20 |
jbicha | in this case, the GNOME portal isn't in the latest live image 🤷 | 12:23 |
seb128 | right, then we probably need an ubuntu-desktop upload | 12:27 |
seb128 | you maybe want to do that at some point? I'm unsure I will be able to squeeze that in today | 12:27 |
seb128 | otherway it will wait for monday, I need to check anyway way the pending amd64 images failed CI yesterday and today | 12:27 |
seb128 | there has been a new ubiquity upload on wednesday, I wonder if the installer is currently broken | 12:28 |
didrocks | seb128: no, feel free to promote it + change the MIR bug status | 13:00 |
seb128 | didrocks, 👍 | 13:01 |
axino | hello desktop team, is HDR display support planned for 22.04 ? | 18:13 |
slingamn | seb128: thanks! | 18:33 |
slingamn | https://bugs.launchpad.net/ubuntu/+source/system-config-printer/+bug/1959440 opened it here | 18:44 |
ubottu | Launchpad bug 1959440 in system-config-printer (Ubuntu) "backport GH #247 for Ubuntu 22.04 release to break python3-requests dependency" [Undecided, New] | 18:44 |
seb128 | axino, hey, that's a question you probably want to ask duflu on monday in an Australia compatible time of the day | 20:25 |
axino | seb128: OK thanks | 20:25 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!