Providing Storage for Spin Locks and Protected Data

As part of device start-up, a driver must allocate resident storage for any spin-lock-protected data or resources and for corresponding spin locks in one of the following places:

  • The device extension of a device object that the driver sets up by calling IoCreateDevice

  • The controller extension of a controller object that the driver sets up by calling IoCreateController

  • Nonpaged, system-space memory that the driver obtains by calling ExAllocatePoolWithTag

Attempting to access pageable data while holding a spin lock causes a fatal page fault if that page is not present. Referencing a spin lock that is invalid (because it was stored in pageable memory and is currently paged out) also causes a fatal page fault.

A driver must provide the storage for each of the following kinds of executive spin lock it might use:

  • Any spin lock that the driver explicitly acquires and releases using any of the Ke*Xxx* spin lock routines.

  • Any spin lock used as a parameter to any of the ExInterlocked*Xxx* routines.

While a driver can make calls to the ExInterlocked*Xxx* routines from its ISR or SynchCritSection routines, it cannot use any of the Ke*Xxx* routines to acquire and release spin locks at any IRQL greater than DISPATCH_LEVEL. Consequently, any driver that reuses a spin lock between calls to the KeXxxSpinLock and ExInterlocked*Xxx* routines must make every call while running at IRQL <= DISPATCH_LEVEL.

A driver can pass the same spin lock to ExInterlockedInsertHeadList as it does to another ExInterlocked*Xxx* routine, as long as both routines use the spin lock at the same IRQL. For more information about how spin lock usage affects performance, see Using Spin Locks: An Example.

In addition to the storage for its executive spin locks, a device driver must provide the storage for another spin lock to be associated with its interrupt objects if it has a multivector ISR or more than one ISR.

Send comments about this topic to Microsoft