> 2162031-libei-absolute-pointer-serialization.patch
I don't think the mutex or the emulating flag can be what stopped it,
though. The libei source and the RDP event queue's flush source are both
attached to the default main context, so they run on the same thread and
the mutex couldn't contend?
The backtrace looks to me like a ping is sent after the connection has
alread gone, and ei_disconnect sets its source to NULL. ei_send_message
then does a source_get_fd without a NULL check. g-r-d gets there because
its EI_EVENT_DISCONNECT case is an empty break, so it considers the
connection ready and keeps flushing input at a dead socket.
> This is independent of the separate NVIDIA GNOME Shell rendering crash
tracked in bug #2162030.
I don't think so, this crash from g-r-d needs the EIS server (GNOME
Shell) to die unexpectedly, which on your machine is bug #2162030.
For a boot where the crash happened, please run,
journalctl -b0 > journal.txt
so we can confirm GS dies just before g-r-d does.
I think the right fix belongs in the disconnect handling rather the
pointer path.
** Also affects: libei (Ubuntu)
Importance: Undecided
Status: New
** Summary changed:
- gnome-remote-desktop-daemon crashed with SIGSEGV in source_get_fd() from
ei_send_message() from ei_connection_request_sync() from ei_connection_sync()
from ei_ping()
+ gnome-remote-desktop-daemon crashes in libei when the EIS server disconnects
mid-session (ei_ping on NULL ei->source)
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2162031
Title:
gnome-remote-desktop-daemon crashes in libei when the EIS server
disconnects mid-session (ei_ping on NULL ei->source)
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/gnome-remote-desktop/+bug/2162031/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
