3.3 Review question: io devices and file systems

Why does fixed partitioning suffer from internal fragmentation whereas dynamic partitioning suffers from external fragmentation? when is compaction needed?

      Fixed partitioning suffers from internal fragmentation because some processes may use less memory than the fixed partition size. Dynamic partitioning suffers from external fragmentation because the memory "holes" left between two partitions may be too small for another process to use. Compaction is used in a system with dynamic partitioning to place all of the partitions in a contiguous range of memory. The effect of this is to combine all the small holes into one larger hole.

What improvements does paging make relative to fixed partitioning?

With paging, a process is divided into a large number of small, fixed size pages. These pages are then placed into frames of main memory, each of which is the size of a page. The main improvement over fixed partitioning is that the pages of the process do not need to be contiguous in main memory. In addition, internal fragmentation is reduced because there is only fragmentation on the last page of the process and the pages are small. The cost of these improvements is the overhead of needing a page table for each process.

How is a logical address translated into a physical address on a system that uses pure paging? how is the translation done for a system that uses pure segmentation?

In a system that uses pure paging, the logical address is divided into a page number and an offset. The page number is used as an index into the page table for the running process; the corresponding entry contains the frame number that the page is mapped to. The offset is added to the frame number to form a physical address.

      In a system that uses pure segmentation, the logical address is divided into a segment number and an offset. The segment number is used as an index into the segment table for the running process; the corresponding entry contains a segment base and length. The segment base is the beginning of the segment of memory that the process is mapped into and the length gives the length of this segment. If the offset is greater than the segment length, a segmentation error occurs. Otherwise, the offset is added to the segment base to form a physical address.

Why is locality of reference so important for a virtual memory system?

Locality is important because a virtual memory system keeps only a small subset of a process image in memory at any given time. If locality holds, then many memory references will refer to this subset and the process will be able to run normally. However, if locality doesn't hold, then memory references will not map to a valid physical address. When this happens, a memory fault occurs and the missing data will need to be loaded from secondary storage. The process will run slowly because secondary storage is much slower than main memory.

When does a page fault occur? describe what the operating system does to handle a page fault.

A page fault occurs when a reference is mapped to a page that is not currently in main memory. When this happens, the process is suspended and the operating system takes control. If main memory is full, then a page must be replaced. Once an empty frame is found, the operating system schedules an I/O operation to bring in the page. After the page is brought in, and the page table updated, the process can be started again.