| 28 Jan 2023 |
K900 ⚡️ | As in I don't really contribute upstream | 18:24:03 |
Ulrik Strid | Okay 👌 | 18:24:59 |
K900 ⚡️ | kdesrc-build is probably what you want for that though | 18:25:25 |
| 2 Feb 2023 |
K900 ⚡️ | New Gear update: https://github.com/NixOS/nixpkgs/pull/214169 | 13:42:35 |
K900 ⚡️ | Any rounded corner enjoyers? https://github.com/NixOS/nixpkgs/pull/214220/files | 17:30:44 |
K900 ⚡️ | * Any rounded corner enjoyers? https://github.com/NixOS/nixpkgs/pull/214220 | 17:30:46 |
| 4 Feb 2023 |
K900 ⚡️ | So I finally figured out why KWallet isn't DBus activating for org.freedesktop.secrets | 18:37:17 |
K900 ⚡️ | The answer is because they just don't ship the unit files for it | 18:37:27 |
K900 ⚡️ | https://gitlab.com/K900/nix/-/blob/master/hacks/kwallet-fdo.nix | 18:37:30 |
K900 ⚡️ | If anyone wants to copy my homework until upstream figures it out | 18:37:38 |
@andreas.schraegle:helsinki-systems.de | some of you are using kde from unstable/master, right? because I was just trying to review https://github.com/NixOS/nixpkgs/pull/214520 and libsForQt5.kimageformats does not seem to contain any .so files anymore on master, whereas they're all there on 22.11 | 19:57:12 |
@andreas.schraegle:helsinki-systems.de | wait, nvm. was holding it wrong. | 19:58:21 |
@andreas.schraegle:helsinki-systems.de | * some of you are using kde from unstable/master, right? because I was just trying to review https://github.com/NixOS/nixpkgs/pull/214520 and libsForQt5.kimageformats does not seem to contain any .so files anymore on master, whereas they're all there on 22.11 | 19:59:23 |
| 6 Feb 2023 |
samueldr | (since it became on-topic) 5.27 will need work for aarch64 | 23:45:31 |
| 7 Feb 2023 |
samueldr | with gcc11 forced, kwin works, but same issues with displays; though unsure if I just missed the error in the firehose of errors, but there's those:
kwin_wayland_drm: Atomic commit failed! Permission denied
kwin_wayland_drm: Presentation failed! Permission denied
love to see errno creeping in
| 01:52:33 |
samueldr | from there, found a single useful result, which hinted at:
bool noAMS = qEnvironmentVariableIntValue("KWIN_DRM_NO_AMS", &isEnvVarSet)
which changes the errors to basically be those in a tight loop
kwin_core: Applying KScreen config failed!
kwin_wayland_drm: Setting vrr failed! Permission denied
kwin_wayland_drm: Setting vrr failed! Permission denied
kwin_wayland_drm: Setting vrr failed! Permission denied
kwin_wayland_drm: Setting vrr failed! Permission denied
| 01:53:45 |
samueldr | two tipmost changes are how I ended-up forcing gcc11, but ugh.... uuuuuugh... https://github.com/samueldr/nixpkgs/commits/wip/plasma-5.27-dont-look | 01:56:32 |
samueldr | since it's errno on DRM operations, I suspect the issue might be one of those two:
- kwin assuming too much about what it can do
- the DRM driver(?) exposes features it can't actually do
| 02:00:17 |
samueldr | and since VRR is new in 5.27 I believe, on 5.26 when disabling atomic mode setting
kwin_wayland_drm: Page flip failed: Permission denied
kwin_wayland_drm: Presentation failed! Permission denied
but not in a loop, just once
| 02:02:49 |
samueldr | * and since VRR is new in 5.27 I believe, on 5.26 when disabling atomic mode setting
kwin_wayland_drm: Page flip failed: Permission denied
kwin_wayland_drm: Presentation failed! Permission denied
but not in a loop, just ~~once~~ twice (when disconnecting)
| 02:03:12 |
samueldr | * and since VRR is new in 5.27 I believe, on 5.26 when disabling atomic mode setting
kwin_wayland_drm: Page flip failed: Permission denied
kwin_wayland_drm: Presentation failed! Permission denied
but not in a loop, just once twice (when disconnecting)
| 02:03:20 |
samueldr | hm... maybe not since I got it on 5.26 | 02:06:13 |
K900 ⚡️ | @samueldr huh I'm surprised it built with gcc11 | 05:52:34 |
K900 ⚡️ | Oh | 05:52:51 |
K900 ⚡️ | You added -mno-outline-atomics | 05:52:56 |
K900 ⚡️ | Yeah | 05:52:57 |
samueldr | ¯\_(ツ)_/¯ it was cheap to try | 05:57:31 |
samueldr | And you won't find me writing c++ code, especially "downgrading"(?) newer standard c++ to a previous one | 05:59:30 |
@andreas.schraegle:helsinki-systems.de | iirc there's been progress on finally getting aarch64-linux on a modern gcc for good. so no more libgcc from bootstrap (eventually). do you think that would solve this issue? | 10:52:22 |
K900 ⚡️ | Yes | 10:55:33 |