23 Jul
2011
23 Jul
'11
10:43 a.m.
On Sat, Jul 23, 2011 at 11:50:30AM +0200, Arnd Bergmann wrote:
Yes, that makes sense. There are also cases where a mutex should really be a spinlock (which is by definition not interruptible), or vice versa. I don't know if this is one of them.
We would be using spinlocks except the underlying buses sleep waiting for the hardware to complete the transfer - the operations are quick enough from a user perspective but at the very low level we want the CPU to be able to go off and do other things while they're happening.