The I/O request packet (IRP) is where the I/O system stores information it needs to process an I/O request. When a thread calls an I/O API, the I/O manager constructs an IRP to represent the operation as it progresses through the I/O system. If possible, the I/O manager allocates IRPs from one of three per-processor IRP nonpaged look-aside lists: the small-IRP look-aside list stores IRPs with one stack location (IRP stack locations are described shortly), the medium-IRP look-aside list contains IRPs with 4 stack locations (which can also be used for IRPs that require only 2 or 3 stack locations), and the large-IRP look-aside list contains IRPs with more than 4 stack locations—by default, the system stores IRPs with 10 stack locations on the large-IRP look-aside list, but once per minute the system adjusts the number of stack locations allocated and can increase it up to a maximum of 20, based on how many stack locations have been recently required. Additionally, these lists are backed by global look-aside lists as well, allowing efficient cross-CPU IRP flow. If an IRP requires more stack locations than are contained in the IRPs on the large-IRP look-aside list, the I/O manager allocates IRPs from nonpaged pool. After allocating and initializing an IRP, the I/O manager stores a pointer to the caller’s file object in the IRP.
Note
If defined, the DWORD registry value HKLM\System\CurrentControlSet\Session Manager\I/O System\LargeIrpStackLocations specifies how many stack locations are contained in IRPs stored on the large-IRP look-aside list.
Figure 8-9 shows a sample I/O request that demonstrates the relationship between an IRP and the file, device, and driver objects described in the preceding sections. Although this example shows an I/O request to a single-layered device driver, most I/O operations aren’t this direct; they involve one or more layered drivers. (This case will be shown later in this section.)
IRP Stack Locations
An IRP consists of two parts: a fixed header (often referred to as the IRP’s
Most drivers specify dispatch routines to handle only a subset of possible major function codes, including create (open), read, write, device I/O control, power, Plug and Play, system control (for WMI commands), cleanup, and close. (See the following experiment for a complete listing of major function codes.) File system drivers are an example of a driver type that often fills in most or all of its dispatch entry points with functions. In contrast, a driver for a simple USB device would probably fill in only the routines needed for open, close, read, write, and sending I/O control codes. The I/O manager sets any dispatch entry points that a driver doesn’t fill to point to its own
EXPERIMENT: Looking at Driver Dispatch Routines
Вильям Л Саймон , Вильям Саймон , Наталья Владимировна Макеева , Нора Робертс , Юрий Викторович Щербатых
Зарубежная компьютерная, околокомпьютерная литература / ОС и Сети, интернет / Короткие любовные романы / Психология / Прочая справочная литература / Образование и наука / Книги по IT / Словари и Энциклопедии