The privacy module registers timeout callbacks in the GLib main context.
The GeoClue callback uses sigc::mem_fun() with the Privacy object, but the
connection remains active when the module is destroyed during a bar reload.
If the timeout runs after destruction, it dereferences the freed Privacy
object and crashes Waybar:
#0 waybar::modules::privacy::Privacy::locationTimeout(bool)
#4 g_main_context_iteration()
#5 g_application_run()
Disconnect the GeoClue and visibility timeout connections in the Privacy
destructor so no callback can run after the module is gone.
The unpatched binary crashed after four SIGUSR2 reloads at 1.5-second
intervals. The patched binary survived ten reloads.
Fixes: eeb7bc702e ("Added GeoClue2 privacy item")
Assisted-by: Claude Opus + Cursor
Ipc::subscribe() assumes the next message on the event socket is the
reply to the IPC_SUBSCRIBE it just sent.
When a module subscribes more than once (sway/workspaces subscribes to
"workspace" and then to "window"), an event from the first subscription
can arrive before the reply to the second one. The payload check fails
and the thrown exception permanently disables the module for that bar.
This is easy to hit when bars are (re)created on output hotplug, since
sway emits a burst of workspace events at exactly that moment while
moving workspaces to the new output:
[warning] module sway/workspaces: Disabling module "sway/workspaces",
Unable to subscribe ipc event
Fix it by skipping over event messages (type high bit set) until the
subscribe reply is found, re-emitting them on signal_event so none are
lost.
Fixes: #5218
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Signed-off-by: Adrian Ratiu <adrian.ratiu@collabora.com>
Cache ALabel tooltip markup and provide it through query-tooltip instead of repeatedly setting GtkWidget's tooltip-markup property. This prevents frequently updating modules from restarting GTK's display-wide tooltip timer.
Keep the active tooltip updated directly so dynamic tooltip contents continue refreshing while the pointer remains stationary.
When min-brightness is set, a scroll step that would cross it was not
clamped — only scrolling while already at or below min was blocked.
With scroll-step: 5 and min-brightness: 1, scrolling from 5% would
pass the guard (5 > 1) and set brightness to 0%.
Fix by computing the post-step value before applying it: if it would
fall below min-brightness, set exactly to min-brightness instead.
Synchronize the cava module man page with the current implementation:
- Fix scdoc table syntax (correct header/body cell prefixes)
- Add missing Waybar-side options: format-icons, vertex_shader, fragment_shader
- Remove options not consumed by the module (data_format, raw_target)
- Clarify option scope: raw-only, GLSL-only, and cava-config-only settings
- Update descriptions to match backend behavior:
- ascii_range is derived from format-icons length
- bar_height is ignored by Waybar
- bar_delimiter is used for raw output
- Polish grammar, tighten structure, and fix example filenames
- Refresh STYLE and FRONTENDS sections to match the actual source
hardware.sh called smoke::stop between umockdev cases, but smoke::stop runs
`swaymsg exit` and tears down the shared compositor -- so only the first case
(backlight) had a display and the rest failed with "cannot open display".
Kill only waybar between cases; leave compositor teardown to the EXIT trap.
The state tier started actually running mpd in 74f0d33, which exposed a
deadlock in teardown: state.sh launches `mpd --no-daemon &` and never kills
it during the tier, and its cleanup() called smoke::stop *before* killing mpd.
smoke::stop ended in a bare `wait`, which reaps *every* background job of the
shell -- including the still-running mpd -- so it blocked until GitHub's 6h
job timeout. continue-on-error doesn't help: it catches failures, not hangs.
- lib.sh: smoke::stop now waits only on the PIDs it owns (waybar, compositor),
so an unrelated daemon left running by a tier can't deadlock teardown.
- state.sh: cleanup() tears down mpd/pulseaudio before smoke::stop.
- smoke.yml: add timeout-minutes: 20 so a future hang fails fast instead of
burning the default 6h runner budget.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
onEvent() runs on the Hyprland IPC listener thread and mutated the label's
style context (add/remove class) directly, racing the GTK main thread's
drawing and corrupting the heap (double free / corrupted double-linked list).
Follow the Submap pattern: onEvent only stores the layout under the mutex and
emits the dispatcher; update() swaps the CSS class on the main thread, tracking
the previously applied class.
Fixes#4665
tooltip-format-enumerate-connected[-battery] only received the device_* args,
so {status}, {num_connections} and {controller_*} threw fmt "argument not
found". Pass those args to the enumerate fmt::format as well.
Fixes#4384
loadConfig() called exit(EXIT_FAILURE) when the cava config failed to load or
no input source was available, killing the whole bar. Throw std::runtime_error
instead: the factory/bar catch it at construction and disable only the cava
module. The read_thread_ also calls loadConfig() at runtime, so contain the
throw there too, logging instead of terminating.
Fixes#4456
config_["hide-vacant"].asBool() was called from the river status listeners.
When the option is given as a string ("true") jsoncpp's asBool() throws
"Value is not convertible to bool", and that exception unwinding through
libwayland's C dispatch aborts the process. Parse it once in the constructor
into a bool member (accepting the string form) and read the cached value.
Fixes#4078
The state tier started actually running mpd in 74f0d33, which exposed a
deadlock in teardown: state.sh launches `mpd --no-daemon &` and never kills
it during the tier, and its cleanup() called smoke::stop *before* killing mpd.
smoke::stop ended in a bare `wait`, which reaps *every* background job of the
shell -- including the still-running mpd -- so it blocked until GitHub's 6h
job timeout. continue-on-error doesn't help: it catches failures, not hangs.
- lib.sh: smoke::stop now waits only on the PIDs it owns (waybar, compositor),
so an unrelated daemon left running by a tier can't deadlock teardown.
- state.sh: cleanup() tears down mpd/pulseaudio before smoke::stop.
- smoke.yml: add timeout-minutes: 20 so a future hang fails fast instead of
burning the default 6h runner budget.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The first run went green but the best-effort tiers silently didn't exercise
anything:
- state.sh: the mpd `audio_output { ... }` block was on one line, which mpd
rejects ("Unknown tokens after '{'"), so mpd never started and the mpd
sub-tier (the #5183 empty-queue repro) skipped. Split the block across lines.
Also move teardown into a cleanup() with `|| true` so a dead-pid kill under
`set -e` no longer makes the step exit 1.
- hardware.sh: umockdev-run's LD_PRELOAD lands ahead of the linked-in ASan
runtime, so ASan aborted before main() ("ASan runtime does not come first")
and waybar never started -- the backlight/slider (#5179) path wasn't tested.
Set verify_asan_link_order=0 so the instrumented binary runs under the preload.
- leakcheck.sh: abort_on_error=1 turned LSan's exit report into a core dump;
force abort_on_error=0 for this report-only tier.
- lib.sh: assert_clean now also flags "ASan runtime does not come first" so a
future preload/link-order regression fails loudly instead of passing.
The smoke test only exercised steady-state rendering and killed waybar with
SIGTERM, so whole crash classes were invisible: exit-time use-after-free
(#5182), module state transitions (#5183) and hardware-backed modules (#5179).
Add tiers that hit those paths under ASan (+ _GLIBCXX_ASSERTIONS):
- lib.sh: assert_clean_exit (SIGINT teardown -> checks segfault/abort + ASan
report emitted during destruction), output hotplug helpers, signal/reload,
WAYBAR_WRAP hook, opt-in leak detection.
- lifecycle.sh (gating): clean exit, runtime output hotplug (Bar/module
destroy), toggle/reload churn, fast-interval teardown race (#5182 class).
- fuzz.sh (gating): pathological custom-backend output (empty, nonzero exit,
invalid JSON, huge, non-UTF8, empty format).
- coverage.sh (gating): every Factory module must be classified for smoke
coverage; a new unlisted module fails the job (#5179 slipped through).
- modules.sh: also render the whole matrix inside a group; SIGINT teardown per
module instead of SIGTERM.
- state.sh (best-effort): real mpd driven through an empty queue (#5183) +
stop/clear; pulseaudio + slider on a null sink.
- hardware.sh (best-effort): backlight, backlight/slider (#5179) and battery
via umockdev.
- leakcheck.sh (report-only): clean-exit run under LSan.
Members are destroyed in reverse declaration order, so modules_all_ (and
the modules it owns) are gone before the GtkWindow. Tearing down a mapped
window emits `unmap`, whose handler runs toggleSuspend() over the already
freed modules. Disconnect the map/unmap handlers in ~Bar first.
Fixes#5182
Alexis Rouillard