We know that there is handshaking between device driver and device controller but here question is that how operating system connects application request or we can say I/O request to set of network wires or to specific disk sector or we can say to hardware -operations.
To understand concept let us consider example which is as follows.
We are reading file from disk. The application we request for will refers to data by file name. Within disk, file system maps from file name through file-system directories to obtain space allocation for file. In MS-DOS, name of file maps to number that indicates as entry in file-access table, and that entry to table tells us that which disk blocks are allocated to file. In UNIX, name maps to inode number, and inode number contains information about space-allocation. But here question arises that how connection is made from file name to disk controller?
The method that is used by MS-DOS, is relatively simple operating system. The first part of MS-DOS file name, is preceding with colon, is string that identifies that there is specific hardware device.
The UNIX uses different method from MS-DOS. It represents device names in regular file-system name space. Unlike MS-DOS file name, which has colon separator, but UNIX path name has no clear separation of device portion. In fact, no part of path name is name of device. Unix has mount table that associates with prefixes of path names with specific hardware device names.
Modern operating systems gain significant flexibility from multiple stages of lookup tables in path between request and physical device stages controller. There is general mechanisms is which is used to pass request between application and drivers. Thus, without recompiling kernel, we can introduce new devices and drivers into computer. In fact, some operating system have the ability to load device drivers on demand. At the time of booting, system firstly probes hardware buses to determine what devices are present. It is then loaded to necessary drivers, accordingly I/O request.
The typical life cycle of blocking read request, is shown in the following figure. From figure, we can suggest that I/O operation requires many steps that together consume large number of CPU cycles.
- System call –
Whenever, any I/O request comes, process issues blocking read() system call to previously opened file descriptor of file. Basically, role of system-call code is to check parameters for correctness in kernel. If data we put in form of input is already available in buffer cache, data is going to returned to process, and in that case I/O request is completed.
- Alternative approach if input is not available –
If the data is not available in buffer cache then physical I/O must be performed. The process is removes from run queue and is placed on wait queue for device, and I/O request is scheduled. After scheduling, I/O subsystem sends request to device driver via subroutine call or in-kernel message but it depends upon operating system by which mode request will send.
- Role of Device driver –
After receiving the request, device driver have to receive data and it will receive data by allocating kernel buffer space and after receiving data it will schedules I/O. After all this, command will be given to device controller by writing into device-control registers.
- Role of Device Controller –
Now, device controller operates device hardware. Actually, data transfer is done by device hardware.
- Role of DMA controller –
After data transfer, driver may poll for status and data, or it may have set up DMA transfer into kernel memory. The transfer is managed by DMA controller. At last when transfers complete, it will generates interrupt.
- Role of interrupt handler –
The interrupt is send to correct interrupt handler through interrupt-vector table. It store any necessary data, signals device driver, and returns from interrupt.
- Completion of I/O request –
When, device driver receives signal. This signal determines that I/O request has completed and also determines request’s status, signals kernel I/O subsystem that request has been completed. After transferring data or return codes to address space kernel moves process from wait queue back to ready queue.
- Completion of System call –
When process moves to ready queue it means process is unblocked. When the process is assigned to CPU, it means process resumes execution at completion of system call.
Attention reader! Don’t stop learning now. Get hold of all the important CS Theory concepts for SDE interviews with the CS Theory Course at a student-friendly price and become industry ready.
- Hardware Protection and Type of Hardware Protection
- Hardware architecture (parallel computing)
- Interaction of a Program with Hardware
- Difference between Hardware and Software
- User View Vs Hardware View Vs System View of Operating System
- Difference between Hardware Interrupt and Software Interrupt
- Dual Mode operations in OS
- Externally Initiated Operations in Microprocessor
- Read and Write operations in Memory
- Operations on Processes
- Arithmetic Micro-operations in Registers
- Shift Micro-Operations in Computer Architecture
- move cmd command
- User mode and Kernel mode Switching
If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to email@example.com. See your article appearing on the GeeksforGeeks main page and help other Geeks.
Please Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.