Comment # 11 on bug 811452 from Egbert Eich

The Xserver never blocks SIGTERM. The only case where this signal is not
delivered is when the Xserver is hanging in some kernel thread. However,
looking at the Xserver log, this doesn't seem to be the problem anyway: the
signal got delivered as the input devices are unloaded. 
What one can also see is that the last line that should show the return value
of the Xserver ie:
   (II) Server terminated successfully (0). Closing log file.
is missing.
This indicates that the server is still waiting for something to shut down, ie
a driver for instance.
A prerequisite to fix anything would be to have an idea about the location of
the hang.
Since Joshi was able to issue a "kill 4292" the system seems to be fully
accessible.
SO if this happens again, it would be good to attach a gdb to check in a
backtrace where exactly we are busy waiting.

And indeed, killing the Xserver with signal 9 can leave the console unrestored.
This begs the question if there is any message that still needs to be
displayed.

Still I would much rather like to see this issue getting resolved:

- Can one trigger it on an installed system as well?
- The report here was for 13.1 - has this been observed on 13.2 or later as
well?

(Further note: Chipset used in Gen3: Q35 G).