I'm a little puzzled as to what semantics you would expect for a
snippet of code as above. Surely you wouldn't really expect to
disable bh for the duration of a call that might require an
arbitrary amount of time for swapping, etc?
Never mind, you've pointed out the issue I was missing.
However the commentary I made about exporting the precise soft/hard
IRQ management mechanism still applies, it really should be private to
a port and present only in arch asm-foo/ headers.
Later,
David S. Miller
davem@dm.cobaltmicro.com