weird remote debug problem

[Chuck Tuffli]

I'm trying to track down what I think is a bug in bus_dmamem_alloc()
using kgdb over a serial line and am having problems with break
points.

/dev/cuaa0 is the comconsole
/dev/cuaa1 is the debug port

The kernel is DragonFly_Stable as of a day ago with the following
diffs from GENERIC
+options        SMP
+options        APIC_IO
+options        BREAK_TO_DEBUGGER
+options        GDB_REMOTE_CHAT
-device         ahd
+device         sio1    at isa? port IO_COM2 flags 0x80 irq 3

After loading my driver, sending a BREAK over the comconsole drops me
into gdb. I set a break point on bus_dmamem_alloc and continue. Then,
I send an ioctl to the driver to trigger memory allocation and think
that the break point has triggered because the below shows in the
comconsole

CPU1 stopping CPUs: 0x00000001
 stopped

but gdb hasn't triggered. Has anyone seen this before? Any ideas? Tnx!

-- 
Chuck Tuffli
Agilent Technologies