Comment 4 for bug 917668

While it may feel like an Xorg freeze, actually from your logs I'm not seeing anything that indicates X is complicit. Retargeting to the kernel.

sock_aio_read() sounds like it might be happening in some asynchronous I/O. However, the dmesg you posted only starts at 19199 sec, so might be missing some pertinent information. I would suggest reproducing it again and collecting dmesg quickly after the problem starts, so it show the period both before and after the fault started.