Swap-Space Management

About Swap-Space Management To illustrate the methods used to manage the swap-space, we will now follow the evolution of swapping and paging on GNU / Linux. As will be discussed fully in Chapter 7, GNU / Linux to start with the implemen tation of copying the entire process of swapping between the contiguous disk area (not cut) and memory. UNIX evolved into a combination of swapping and paging with the availability of hardware for paging. In 4.3BSD, swap-space allocated to the process when a process starts. Provided enough space to accommodate the program, also known as the pages of text (text pages) or text segment and data segment of the process. Allocation takes place early in this way generally prevents a process to run out of swap-space while the process was done. When the process began, the text in it on-page of the file system. The pages (pages) will be placed in the swap if necessary, and read back from there, so the file system will be accessed once for each text page. The pages of the data segment is read from the file system, or created (if not before), and placed in the swap space and in-page back if necessary. One example of optimization (for example, when two users use the same editor) is a process that is identical with the text page split pages (pages), either in the memory like any in the swap-space. Swap the two maps for each process used by the kernel to track swap-space use. Fixed amount of text segments, then the swap space is allocated for 512K per piece (chunks), except for the last piece, which keep the rest of the pages (pages) earlier, with increases (increments) of 1K. Swap maps of the data segment is more complicated, because a large segment of the data can be made at any time. Size of the map itself remains, but the store survived a swap addresses for the blocks, the amount varies. Suppose there are index i, blah-blah-blah, with a maximum of 2 megabytes. The data structure is shown in Figure 13.8. (Minimum and maxi mum Large blocks vary, and may be changed when rebooting the system.) When a process tries to increase its data segment exceeds the allocated blocks in the swap, the operating system allocates another block, the first two times as large. This scheme caused her minor processes using small blocks. It also minimizes fragmentation. Blocks of a large process can be found quickly, and keep a small swap maps. On Solaris 1 (SunOS 4), the author makes changes to the method of standard UNIX me to improve my efficiency and to reflect changes in technology. When a process is running, hala man-pages (pages) of the text segment is brought back from the file system diak process in main memory, and discarded when it was decided to in-pageout. A would be more efficient to reread a page (page) of the tem sis file rather than putting it in the swap-space and re-read it from there. Even more changes to the Solaris 2. The biggest change there is Solaris 2 allocates swap-space only when a page (page) was forced out of memory, rather than when the page (page) da ri virtual memory was first made. These changes provide a better form on modern computers, have already made a lot more memory than the computers with the old system, and more rarely do paging.