> Sound like different issues then. If I understand correctly, the > "freezes" that still happen resolve themselves after some time? Yes. Usually just shy of a minute, sometimes over five. Load has shot up to >>50 during that time, so there is certainly some io wait involved. > X is involved in the vt switch (when you press C-A-F1, X gets SIGUSR1, This was new to me. Perhaps I should try chvt, then? > Do you get anything in the X or kernel log when these freezes happen? Nothing out of the ordinary, that's why it is so hard to figure out what happens. > What if you set /sys/module/drm/parameters/debug to 6? If the machine Done. What am I expecting? I guess these are benign: [9402544.148312] [drm:intel_crtc_cursor_set], [9402544.148316] [drm:intel_crtc_cursor_set], cursor off [9402550.044010] [drm:intel_gpu_idle_timer], idle timer fired, downclocking > is reachable over the network, can you connect over ssh and attach gdb > to the X process to get a stack trace? I will try to. If it is a longer one, it will be no problem, but on a shorter one I will have to be quick. Cheers, -Juha -- ----------------------------------------------- | Juha Jäykkä, juhaj@iki.fi | | http://www.maths.leeds.ac.uk/~juhaj | -----------------------------------------------
Attachment:
signature.asc
Description: This is a digitally signed message part.