output: add -o to use only a named output - #523
Open
quinnjr wants to merge 1 commit into
Open
Conversation
Cage's existing multi-monitor control (-m extend|last) cannot target a specific monitor: `last` picks whichever output happens to be enumerated last, which is not something the user chooses. For kiosk and greeter setups it is often necessary to pin the session to one known connector. Add -o <name>, matching the wlroots output name (DP-1, HDMI-A-1, ...). -o is an admission filter applied before a cg_output is allocated, while -m is an arrangement policy over outputs already admitted, so -o takes precedence and an advisory is logged when both are given. The check sits after the non_desktop branch so DRM leasing of non-desktop outputs (VR headsets) is unaffected. Excluded outputs are explicitly disabled rather than merely skipped. An output may arrive already enabled -- the DRM backend does this, headless does not -- and leaving one enabled without ever rendering to it means it keeps scanning out whatever framebuffer it held when cage took DRM master. On a greeter that leaves the previous session's last frame on screen next to the login prompt. Never adding excluded outputs to server->outputs also keeps them out of the output-management heads, which is what keeps wlr_output->data non-NULL for every head reachable in output_config_apply(). If no connected output matches, cage exits non-zero instead of running with an empty output list. Without this a typo produces a live compositor rendering nowhere, forever, with no error: the existing self-terminate path in output_destroy() is gated on was_nested_output and on an output having been registered first, so neither fires. A supervisor such as greetd reads the exit status, not the log, so failing loudly is what lets it fall back to a text login. Reaching that exit revealed that the listener-teardown block only runs on the normal path, so any goto end after those listeners are registered trips an assertion in wlr_output_layout_destroy(). Add a teardown: label so this exit unwinds the same way a normal shutdown does. The earlier goto end sites are left alone: they run before the links are initialised, where wl_list_remove would be invalid.
Member
|
We'd like to move away from CLI flags for output configuration, see #466. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR adds a
-o <name>flag that pins Cage to a single output, matched against the wlroots connector name (DP-1,HDMI-A-1, and so on).-m lastalready restricts Cage to one monitor, but which one you get depends on enumeration order, so there's currently no way to say which screen you actually want. That matters for kiosks and greeters that need to come up on a specific panel.Excluded outputs get switched off rather than just skipped. The DRM backend hands them over already enabled, and if Cage never renders to one it will sit there scanning out whatever was in its framebuffer when Cage took DRM master, so on a greeter the previous session's last frame stays up next to the login prompt.
If no connected output matches the name, Cage now exits non-zero instead of running with an empty output list. The self-terminate path in
output_destroy()is gated onwas_nested_outputand on an output having been registered first, so it can't catch this case, and the result today is a live compositor rendering nowhere with one INFO line to show for it.Reaching that new exit turned up an assertion in
wlr_output_layout_destroy(), because the listener teardown block only runs on the normal path. I added ateardown:label to unwind the same way. The earliergoto endsites are left alone since they run before those links are initialised.Tested on the headless backend (
WLR_BACKENDS=headless WLR_HEADLESS_OUTPUTS=2) for the matching, non-matching, empty-name, and-owith-m lastcombinations, and on a real DRM session with two DisplayPort monitors, where the selected output came up as expected and the excluded one went dark. Hotplug is untested.