Stored monitor layout not restored when an output reconnects during Muffin's own reconfiguration (hotplug misdetected, output left mirrored)

Aperta
#894 0 commenti 0 reazioni 0 assegnatari Vedi su GitHub

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
4/5
Tempo stimato
3-5 giorni
Idoneità per principianti
43/100
Tipo di issue
Bug
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
c
Ambito
desktop

Direzione di ricerca

Start with meta_monitor_manager_xrandr_handle_xevent() in src/backends/x11/meta-monitor-manager-xrandr.c and trace how RandR timestamps determine whether a change is treated as a hotplug. Compare that path with meta_monitor_manager_on_hotplug() and the stored monitor configuration lookup. Done means a quick disconnect and reconnect restores the saved three-monitor layout; the issue describes the setup and observable failure, but provides no test file or reproduction harness.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

Distribution

Linux Mint 22.3 (Cinnamon 6.6.9, X11)

Package version

muffin 6.6.3+zena (libmuffin0 6.6.3+zena)

Graphics hardware in use

NVIDIA GeForce RTX 2060, proprietary driver 595.99.02 (open kernel module). Two DisplayPort monitors and a 2012 Samsung TV on HDMI.

Frequency

Quite often (not always: it is a race). In one run 2 of 3 power cycles of the TV were fine; in the next, 4 hotplugs in a row hit it.

Bug description

When an output disconnects and reconnects within a couple of seconds, Muffin can skip the saved layout for the full set of monitors and leave the returning output mirrored with another monitor.

My TV keeps its HDMI output "connected" while switched off, but as it powers on it drops the output for about 2 seconds and brings it back. After that blip the saved 3-monitor layout is often not restored: the TV ends up at 0,0 on top of the primary monitor, and the Display settings show them as one mirrored pair.

Saved layout (~/.config/cinnamon-monitors.xml has it for exactly these three monitors):

HDMI-0  1920x1080+0+0
DP-5    1920x1080+0+1080
DP-3    1920x1080+1920+1080  (primary)

After the blip:

HDMI-0  1920x1080+0+0
DP-3    1920x1080+0+0        (primary)
DP-5    1920x1080+1920+0

In that state org.cinnamon.Muffin.DisplayConfig.GetCurrentState lists all three monitors with the same connector/vendor/product/serial as the saved configuration, and two logical monitors: (0,0) with DP-3 + HDMI-0, and (1920,0) with DP-5. So Muffin knows the monitors and has a stored configuration that matches them; it just never applies it.

What I think happens. meta_monitor_manager_xrandr_handle_xevent() in src/backends/x11/meta-monitor-manager-xrandr.c decides whether an event is a hotplug with:

is_hotplug = resources->timestamp < resources->configTimestamp;

In the broken state the X server reports:

timestamp       = 40948476   (last configuration change)
configTimestamp = 40948273   (last hotplug)

i.e. the last configuration change landed 203 ms after the last hotplug. That is consistent with this sequence:

  1. The TV disconnects. Muffin handles it as a hotplug and, having no stored configuration for the two remaining monitors, starts applying a linear one (DP-3 at 0,0, DP-5 at 1920,0).
  2. The TV reconnects (configTimestamp moves) before that configuration has been applied.
  3. Muffin's configuration lands (timestamp moves, now newer than configTimestamp).
  4. When the event for the reconnection is processed, timestamp < configTimestamp is false, so it is not treated as a hotplug: meta_monitor_manager_on_hotplug() is not called and the stored 3-monitor configuration is never looked up. The state is rebuilt from what X has, where the TV's CRTC is still at 0,0.

I have not instrumented Muffin itself; steps 1–3 are inferred from the two timestamps, the resulting layout and the code. The same comparison is still in master.

Steps to reproduce
  1. Three monitors with a saved non-linear layout, e.g. one output placed above the other two, so that removing it changes the positions of the rest.
  2. Make that output disconnect and reconnect within ~2 seconds (in my case: power on a TV that toggles HPD while it boots).
  3. Repeat a few times.
Expected behavior

After the output is back, the stored configuration for the full set of monitors is applied, as it is when the reconnection does not overlap with Muffin's own reconfiguration.

Additional information

Polling xrandr during one failure (local time, 0.4 s resolution):

18:47:03.4  DP-3 1920x1080+1920+1080 (primary) | DP-5 1920x1080+0+1080            <- HDMI-0 gone
18:47:05.2  HDMI-0 1920x1080+0+0 | DP-3 1920x1080+0+0 (primary) | DP-5 1920x1080+1920+0

Cinnamon logged Monitor Watch List updated: [0,1] at the disconnection and nothing afterwards (it logs [0,1,2] when the layout is restored correctly).

Reapplying the positions with xrandr --output … --pos … fixes it, and Muffin then reports three separate logical monitors again. As a workaround I run a small user service that listens for RandR ScreenChangeNotify events with a new config timestamp and, if two monitors end up at the same position shortly after, reapplies the layout from cinnamon-monitors.xml.

Possibly the same underlying problem as #639 (closed as stale).

Lingua principale
C
Stelle
249
Fork
133
Merge medio
17m
PR unite (30g)
1

Preparare l'ambiente

Questo progetto non fornisce container di sviluppo, Dockerfile né guida per i contributori, quindi l'ambiente è a tuo carico: parti dal suo README e consulta la nostra guida al primo contributo per i passaggi generali.

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di linuxmint/muffin

Tutte le issue di linuxmint/muffin

Issue simili

Altre issue su C

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.