On Feb 4, Jukka Marin wrote
> I think my window manager had something to  do with it.. when I switched
> to the console screen in pcvt, it worked immediately.

You need to be in text mode to enter the debugger. The debugger don't
know how to switch from graphic to text (it's wideo-board dependant, and
thus only the X server know how to do it). Also, the X server has
complete control of the keyboard, and ctrl-alt-esc isn't catched when under
X.

> I see.  Well, all my serial ports were functioning normally.  Ethernet was
> alive and well.  As far as I can tell, only disk access fails (I haven't
> tried to access the IDE disk while the problem exists - I'm not sure if I
> can do that without causing an access to the SCSI disk (swap, system) at
> the same time (shells do funny things sometimes).

Hum, from the spl man page, it seems that splsoftnet is at a lower
priotity than splbio. So if the problem comes from a splbio without splx,
then the network would be hung as well. Can someone confirm this ?

The problem can also come from a dead lock somewhere in the vfs or
disk driver. It would be really really usefull to know on which resources the
hung processes are wainting for. A kernel core dump would help for this.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--