Control: retitle -1 gnome-shell: segmentation fault in mozjs rooting API on
resume from suspend
Control: tags -1 + unreproducible
On Sat, 12 Jul 2025 at 10:12:53 +0100, Luca Boccassi wrote:
Last week gnome-shell crashed on resume from suspend. I do not
remember anything particular about that session, and cannot reproduce
it either.
Marking as unreproducible accordingly, and adjusting the title to try to
avoid "me too" comments if someone is seeing an unrelated segfault
during resume that would be better reported separately.
#13 0x00007f5f261c71bb in ToggleQueue::handle_all_toggles
(this=0x7f5f262afe60
<ToggleQueue::get_default_unlocked()::the_singleton>,
handler=0x7f5f261ac2a0 <toggle_handler(ObjectInstance*,
ToggleQueue::Direction)>) at
./obj-x86_64-linux-gnu/../gi/toggle.cpp:69
#14 0x00007f5f261c722b in ToggleQueue::idle_handle_toggle
(data=<optimized out>) at ../gi/toggle.h:83
self = {<No data fields>}
#15 0x00007f5f2630c345 in ??? ()
#16 0x00058ea77800ead9 in ??? ()
#17 0x000055e6e0026ae0 in ??? ()
#18 0x000055e6f0cd3a00 in ??? ()
#19 0x0000000100000001 in ??? ()
#20 0x0000000000000000 in ??? ()
If you tell me which packages own the bottom of the stack there I can
re-decode the core.
If there's no information available in gdb, then I suspect it might be
JIT'd code generated by mozjs128, which we are unlikely to be able to
debug. I also can't see anything obviously useful in the quoted Journal
entries.
This might be a gjs or mozjs128 issue rather than gnome-shell itself,
but it's hard to tell either way.
smcv
