Dave Huang wrote:
> On a NetBSD/mac68k 1.4_ALPHA system, April 10 sources, I was doing a
> "make build" when I got a "panic: closef: count < 0"
> I don't remember the exact stack trace, but ddb said curproc was "cc",
> and the trace had sys_exit() and exit1() in it.

this is the same panic i can get relatively quickly by running sysinst
under -current.  this is related to the mac68k interrupt/uvm/pool/pmap
problem that's currently being debugged (although if you can figure out
what's going on, please let someone know!!!!)
 
> When I tried to get a crash dump from the closef panic, I got a stack
> trace that looked better, but I was too tired to type it in :) Had to do
> with pmap_remove_mapping though. (And actually, 0xddc40 is
> pmap_remove_mapping + 0x14e: cmpal a3@(0x4),a4... "show reg" says that
> a3 is _proc0, a4 is _microtime. a3@(0x4) would try to deref a null
> pointer).

that's pretty close to what i can see.  it appears that something is
messing around with the file pool when it shouldn't do so...

later.

colin