In systems with multiple SCSI host adapters or networking cards, it seems
particularly likely that unforseen problems could arise. From what I've seen,
it is generally considered a good thing for the interrupt handler to leave
interrupts disabled only during the initial phase of acquiring status
information, and then enable interrupts while SCSI completion processing is
handled. But while good for interrupt latency, this strategy could easily lead
to cases of stack overflow under heavy load.
So how am I to know whether my interrupt handler (along with the SCSI
completion processing) is using a "reasonable" amount of space?
A single page seems too small to guarantee safety here.
Leonard