kernel crash in kdb_enter

Hi 

I got the following in my /var/log/messsage:

Jul 26 14:04:41 numeri kernel: Fatal trap 3: breakpoint instruction fault while in kernel mode
Jul 26 14:04:41 numeri kernel: instruction pointer      = 0x8:0xc04de1c4
Jul 26 14:04:41 numeri kernel: stack pointer            = 0x10:0xde3c9304
Jul 26 14:04:41 numeri kernel: frame pointer            = 0x10:0xde3c9304
Jul 26 14:04:41 numeri kernel: code segment             = base 0x0, limit 0xfffff, type 0x1b
Jul 26 14:04:41 numeri kernel: = DPL 0, pres 1, def32 1, gran 1
Jul 26 14:04:41 numeri kernel: processor eflags = IOPL = 0
Jul 26 14:04:41 numeri kernel: current process          = 676 (mozilla-bin)
Jul 26 14:04:41 numeri kernel: trap number              = 3
Jul 26 14:04:41 numeri kernel: panic: breakpoint instruction fault
Jul 26 14:04:41 numeri kernel: KDB: enter: panic


I got a gassilion of them with different stack/frame pointers but all with the 
same instruction pointer. According to nm this instruction pointer is in 
kdb_enter:

k_at_numeri ~ >nm -n /boot/kernel/kernel | grep c04de1
c04de114 T kdb_backtrace
c04de144 T kdb_dbbe_select
c04de198 T kdb_enter
c04de1c8 T kdb_init

It seems likt the trap was repeated until it finally rebooted itself. 
Unfortunatly I did not get a kernel dump since it just reboot without
trying to dump.

This is with a kernel built from source from 20040719. I'm rebuilding with 
current source from now and hopefully its fixed. 

The kernel includes the following options:
options         KDB                     # Enable kernel debugger support.
options         KDB_TRACE
options         DDB                     # Support DDB.
options         DDB_NUMSYM
options         GDB                     # Support remote GDB.
options         INVARIANTS              # Enable calls of extra sanity checking
options         INVARIANT_SUPPORT       # Extra sanity checks of internal structures, required by INVARIANTS
options         WITNESS                 # Enable checks to detect deadlocks and cycles
options         WITNESS_SKIPSPIN        # Don't run witness on spinlocks for speed
options         ACPI_DEBUG

but I do not have it set up for remote debugging at the moment.

Also of interest, this is with ULE as scheduler and I did NOT comment out 
PREEMPTION since I was hoping to get a dump if it crashed and analys it 
after the fact :-(

I do not know if there is any more info I can provide but feel free to ask. 


/Johan K

-- 
Johan Karlsson		mailto:johan_at_FreeBSD.org