[kdb] [PATCH] KDB: commandeer vector 0xfe for KDB_VECTOR]]

[Keith Owens]

The more I look at this code, the more I wonder why we have a special
interrupt number for KDB_VECTOR.  It made sense when KDB did not use
NMI to hit recalcitrant cpus, in those days KDB_VECTOR needed to be a
reasonably high value.  But with the current code, KDB_VECTOR does not
need a high priority and it can be any value.  More to the point, I
think that it can be dynamically allocated using create_irq() and
friends.

If changing KDB_VECTOR to a dynamic irq works, then we could also
change KDBENTER_VECTOR to a dynamic irq number.  That would require a
little more work, any use of KDB_ENTER() in the kernel would have to be
recorded and patched at startup to use the assigned irq.  No big deal,
we know how to do that.

At the moment I do not have any test hardware to play with, so I cannot
check if KDB can use dynamic irqs.

[resend from correct address]