> That reminds me....  assuming the Xserver process doesn't have any
> pages out on disk, maybe there should be an option on disk-full
> workstations to handle mouse interrupts at the same priority level as
> disk interrupts (if they're not already) so that the user won't
> perceive such apparent hangs quite so easily.

(a) I'm not sure that will help, because the X server has to run, and
if the machine is so flooded with disk interrupts as to lock out mouse
interrupts, userland has no hope.  (That also isn't the scenario
described upthread; in that one, the machine is not flooded with
interrupts; it's just that processes that try to touch the disk find
themselves stuck behind a 20-30 second queue of disk requests.
Userland that doesn't try to touch the disk can run fine.  And,
similarly, other interrupts, such as mouse interrupts, are fine.)

(b) I'm not convinced it's good to have the mouse responsive if the
machine is so swamped that interrupt priority makes a difference.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B