Sure... that's probably true... my knowledge of Linux SCSI device
drivers is limited to hacking the ability to recognize a PCI card. One
question though... why does the Mylex DAC960 driver acquire the
io_request spinlock? They seem to be similar functions to those in the
BusLogic driver that don't have locks. Oh well... just my ignorance.
There are three separate sets of locking functions defined in the DAC960
driver, which is not a SCSI driver so the rules are somewhat different.
Generalized access to the DAC960 controller requires a lock since the caller
will not already hold the lock. Request functions already hold the lock and so
cannot reacquire it. The interrupt holder must again acquire the lock.
These abstractions may not be quite right, but they will make it a lot easier
to divorce the DAC960 driver from the general io_request_lock when that becomes
desirable.
Leonard
/*
DAC960_AcquireControllerLock acquires exclusive access to Controller.
*/
static inline
void DAC960_AcquireControllerLock(DAC960_Controller_T *Controller,
ProcessorFlags_T *ProcessorFlags)
{
spin_lock_irqsave(&io_request_lock, *ProcessorFlags);
}
/*
DAC960_AcquireControllerLockRF acquires exclusive access to Controller,
but is only called from the request function with the io_request_lock held.
*/
static inline
void DAC960_AcquireControllerLockRF(DAC960_Controller_T *Controller,
ProcessorFlags_T *ProcessorFlags)
{
}
/*
DAC960_AcquireControllerLockIH acquires exclusive access to Controller,
but is only called from the interrupt handler.
*/
static inline
void DAC960_AcquireControllerLockIH(DAC960_Controller_T *Controller,
ProcessorFlags_T *ProcessorFlags)
{
spin_lock_irqsave(&io_request_lock, *ProcessorFlags);
}
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/