|
|
|
|
|
|
|
|
HP-UX Memory Management: White Paper > Chapter 1 MEMORY MANAGEMENT
OVERVIEW OF PHYSICAL AND VIRTUAL MEMORY |
|
The memory management system is designed to make memory resources available safely and efficiently among threads and processes:
The data and instructions of any process (a program in execution) or thread of execution within a process must be available to the CPU by residing in physical memory at the time of execution. To execute a process, the kernel creates a per-process virtual address space that is set up by the kernel; portions of the virtual space are mapped onto physical memory. Virtual memory allows the total size of user processes to exceed physical memory. Through "demand paging", HP-UX enables you to execute threads and processes by bringing virtual pages into main memory only as needed (that is, "on demand") and pushing out portions of a process's address space that have not been recently used. The term "memory management" refers to the rules that govern physical and virtual memory and allow for efficient sharing of the system's resources by user and system processes. The system uses a combination of pageout and deactivation to manage physical memory. Paging involves writing recently unreferenced pages from main memory to disk from time to time. A page is this smallest unit of physical memory that can be mapped to a virtual address with a given set of access attributes. On a loaded system, total unreferenced pages might be a large fraction of memory. Deactivation takes place if the system is unable to maintain a large enough free pool of physical memory. When an entire process is deactivated, the pages associated with the process can be written out to secondary storage, since they are no longer referenced. A deactivated process cannot run, and therefore, cannot reference its data. Secondary storage supplements physical memory. The memory management system monitors available memory and, when it is low, writes out pages of a process or thread to a secondary storage device called a swap device. The data is read from the swap device back into physical memory when it is needed for the process to execute. Pages are the smallest contiguous block of physical memory that can be allocated for storing data and code. Pages are also the smallest unit of memory protection. The page size of all HP-UX systems is four kilobytes. On a PA-RISC system, every page of physical memory is addressed by a physical page number (PPN), which is a software "reduction" of the physical page number from the physical address. Access to pages (and thus to the data they contain) are done through virtual addresses, except under specific circumstances.[1] When a program is compiled, the compiler generates virtual addresses for the code. Virtual addresses represent a location in memory. These virtual addresses must be mapped to physical addresses (locations of the physical pages in memory) for the compiled code to execute. User programs use virtual addresses only. The kernel and the hardware coordinate a mapping of these virtual and physical addresses for the CPU, called "address translation," to locate the process in memory. A PA-RISC virtual address consists of a space identifier (SID) and an offset.
Every process running on a PA-RISC processor shares a 48-bit (or larger, depending on HP-PA architecture version) global virtual address space with the kernel and with all other processes running on that machine. Although any process can create and attempt to read or write any virtual address, the kernel uses page granularity access control mechanisms to prevent unwanted interference between processes. When a virtual page is "paged" into physical memory, free physical pages are allocated to it from the free list. These pages may be randomly scattered throughout the memory depending on their usage history. Translations are needed to tell the processor where the virtual pages are loaded. The process of translating the virtual into physical address is called virtual address translation. Potentially the virtual address space can be much greater than the physical address space. The virtual memory system enables the CPU to execute programs much larger than the available physical memory and allows you run many more programs at a time than you could without a virtual memory system. For a process to execute, all the structures for data, text, and so on have to be set up. However, pages are not loaded in memory until they are "demanded" by a process -- hence the term, demand paging. Demand paging allows the various parts of a process to be brought into physical memory as the process needs them to execute. Only the working set of the process, not the entire process, need be in memory at one time. A translation is not established until the actual page is accessed. [1] When virtual translation must be turned off (the D and I bits are off), pages are accessed by their absolute addresses. |
|||||||||||||||||||||||||||||||||
|
||
|
|
|
|||||||||
|
|||||||||
