(In reply to comment #8)
> Depending on circumstances, you might find it in /var/log/messages, or if
> you can ssh in while the machine is "hung", just run dmesg, or if you can't,
> set up netconsole.

The /var/log/messages contains the same output to that of journalctl.
journalctl is the systemd way of working with the system logs.

dmesg also doesn't have much information about this crash. I will attach a
dmesg2.txt and a journalctl2.txt files in the next messages.

The nouveau Bugs wiki (http://nouveau.freedesktop.org/wiki/Bugs/) suggests to
add log_buf_len=1M into the kernel parameters. But I doubt it would help in my
case, would it? Maybe there is some other kernel/mesa parameter that will
change the log level to something more informative?