Repository navigation
Conversation
stretch_dot() arms a GIVE_UP_SECONDS (10 s) GLib.Timeout whose callback
calls wrapper.get_data("launch_dot"). The wrapper and the dot were
captured with `unowned` -- raw, non-owning pointers. When the dock item is
rebuilt or destroyed inside that window (dock rebuild, app-grid change,
session teardown) the timer fires on freed memory and the shell dies with
SIGSEGV in g_object_get_data ->
g_type_check_instance_is_fundamentally_a.
Confirmed live on O6N (arm64, NCZ-OS 26.10): the core of singularity-desktop
pid 2613 has main-thread frame singularity-desktop+0x152644 called from the
main-loop source dispatch; disassembling the shipped (stripped) binary at
that offset against the pinned sources maps it exactly to this lambda.
Fix: capture owned references instead of unowned so the objects stay valid
until the timer fires; the closure releases them when the source is
removed. Behavior is otherwise unchanged (the dot is still cleaned up on
the same condition).
Assisted-by: opencode:deepseek-flash
AI-Scope: Root-caused the SIGSEGV from the coredumpctl backtrace plus objdump of the stripped binary, and wrote the owned-capture fix.
perlowja
added a commit
to ncz-os/cix-installer
that referenced
this pull request
Oct 13, 2026
The pinned shell (f9c93296) takes down the whole session with SIGSEGV when a dock item is destroyed inside the 10 s launch-feedback give-up window: DockLaunchFeedback.stretch_dot() captured the item wrapper and its dot with `unowned` (non-owning pointers) and the timer callback then calls g_object_get_data() on freed memory. Live on O6N arm64: core pid 2613, main-thread frame singularity-desktop+0x152644 -> g_object_get_data -> g_type_check_instance_is_fundamentally_a; the wrapper then escalates to Safe Mode and the operator cannot get a desktop. Carry upstream PR singularityos-lab/singularity-shell#37 as assets/singularity/patches/singularity-shell/0006-dock-launch-feedback-owned-timer-captures.patch so the next arm64 payload stops crash-looping. Verified with build/singularity-pins.sh check-patches against the pinned commit.
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
1 participant
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.
Generative AI assistance
AI assistance: disclosed
What
DockLaunchFeedback.stretch_dot()arms aGIVE_UP_SECONDS(10 s)GLib.Timeoutwhose callback calls
wrapper.get_data("launch_dot"). Thewrapperdock item andthe
bindot were captured withunowned— raw, non-owning pointers. If the dockitem is rebuilt or destroyed inside that 10 s window (dock rebuild, app-grid change,
session teardown), the timer fires on freed memory.
Symptom
Hard SIGSEGV in the shell main thread:
Observed live on NCZ-OS 26.10 (arm64):
coredumpctlcore pid 2613, booting a freshsession, ~4 min after login. The offset maps exactly to this callback against the
pinned sources.
Fix
Capture owned references instead of
unownedso both objects stay valid until thetimer fires; the closure releases them when the source is removed. No behavior
change beyond not crashing — the dot is still cleaned up on the same condition.
Assisted-by: opencode:deepseek-flash