MEMORY MANAGEMENT

9/3/2012

Agenda
Introduction to Solaris History of solaris Solaris memory architecture. Backing store VM system. Solaris memory management. Swapping Demand paging.

9/3/2012

Introduction to solaris
Solaris is a Unix operating system originally developed by Sun Microsystems. Solaris is known for its scalability especially on SPARC systems and for originating many innovative features such as DTrace, ZFS and Time Slider. Solaris supports SPARC-based and x-86 based workstation and servers from Sun and other vendors. Solaris has a reputation for being well-suited to symmetric multiprocessing, supporting a large number of CPUs. Programmed in C. 9/3/2012 Its source model is Mixed open source/closed

History of solaris
In earlier days SunOS is implemented in uniprocessor workstation. In 1980s distributed and network-based computing became popular, so they went to multiprocessor system to speedup. In 1987, AT&T bell and Sun joined together for a project and developed a new OS by merging the existing OS.(SunOS,BSD,SVR3,Xenix). That new OS is named as System V Release 4 (SVR4) and then its name changed as Solaris 2. Latest version is Solaris 11.

9/3/2012

Solaris Memory Architecture

Physical memory is divided into fixed-sized pieces called pages. The size of the page varies from platform to platform. The common size is 8 Kbytes. Each page is associated with a file and offset. The file and offset identify the backing store for the page.

9/3/2012

Backing store

Backing store is the location to which the physical page contents will be migrated when that page is need to be taken for another process. The pages are migrated to slower medium like disk and that is called swap space.
The location where the pages are migrated is called page-out. When again that pages are needed by the process, the location from which the pages are read back again is called page-in.
9/3/2012 6

Virtual memory System

Virtual memory system is the core for a Solaris system.

Why have a Virtual memory System?

It presents simple memory programming model. It allows process to see linear range of bytes, regardless of fragmentation of the real memory. It gives programming model with a larger size than available physical storage.

9/3/2012

Jobs of VM System

The job of VM system is to keep the most frequently referenced portions of memory in primary storage. When RAM shortage comes, VM is required to free the RAM by transfering infrequently used memory out to the backing store.

By doing so,the VM system optimizes the performance.

Process Memory Allocation

A process will have a linear virtual address space. The linear Virtual address space of a process is divided into segments like

Executable-text : Executable instructions in binary form Executable-Data : Initialized variables in the executable Heap space : Memory allocated by malloc() Process Stack and so on

9/3/2012

Process memory allocation

9/3/2012

10

Solaris virtual to physical memory management

9/3/2012

11

Virtual memory management unit(MMU)

Solaris kernel breaks the process into segments and segments into pages. The hardware MMU maps those pages into physical memory by using a platform-specific set of translation tables. Each entry in the table has the physical address of the page of the RAM,so that memory access can be converted on-the-fly in hardware.

9/3/2012

12

Shared Mapped File

9/3/2012

13

Solaris Memory Management

Two basic types of memory management manage the allocation and migration of physical pages of memory to and from swap space :

1.Swapping 2.Demand paging

The VM system uses a global paging model that implements a single global policy to manage the allocation of memory between the processes.
9/3/2012 14

Swapping

The swapping algorithm uses user process as the granularity for managing memory. If there is a shortage of memory, then all the pages of the least active process will be swapped out to the swap device, freeing memory for other processes. Then the corresponding flag in the process table is set to indicate that this process has been swapped out.

9/3/2012

15

Memory Scheduler

The memory scheduler is launched at boot time and does nothing. If the memory is consistently less, it starts looking for processes that it can completely swap out. If shortage is minimal, then soft swap takes place. Otherwise hard-swap.

9/3/2012

16

Soft Swapping

If the process have been inactive for atleast maxslp(default 20 seconds) seconds, then memory sceduler swaps out all the pages for that process.

9/3/2012

17

Hard Swapping

Hard swapping takes place when all of the following are true :

Atleast two processes are on the run queue, waiting for CPU. The average free memory over 30 seconds is consistently less than minimum. Excessive paging (page-out + page-in is high).

9/3/2012

18

Pros and Cons of Swapping

Pros: 1.This is the inexpensive way to conserve memory. 2. Easy implementation. Cons: 1. It dramatically affects a processs performance.

So the swapping is used only as a last resort when the system is desperate for memory.
9/3/2012 19

Demand paging

The Demand paging model uses a page as the granularity for memory management. Pages of memory are allocated on demand.

When memory is first referenced, a page fault occurs and memory is allocated one page at a time.
The page scanner and the virtual memory page fault mechanism are the core of the demand paged memory management.
9/3/2012 20

Page Scanner

The page scanner is a kernel thread, which is awakened when the amount of memory on the free-page list falls below a system threshhold, typically 1/64 th of total physical memory. Scanner (pageout_scanner) tracks pages by reading the state of the hardware bits in MMU MMU bits maintain state of referenced and written (dirty). Uses a twohanded clock algorithm to determine eviction.
9/3/2012 21

Two Handed Clock Algorithm

Both hands rotate clock-wise. The front hand clears the referenced and modified bit of each page. The trailing back hand then inspects the referenced and modified bits some time later.
9/3/2012 22

Scan Rate
The scan rate changes during the course of the page scanners operation Scanners scans at slowscan when memory is below lotsfree and then increases to fastscan when memory has fallen below minfree threshold

Slowscan is set to 100 pages by default. Fastscan is set to physicalmemory/2 capped at 8192 pages
9/3/2012 23

Page Faults
When do Page Faults occur?

MMU-generated exceptions (trap) tell the operating system when a memory access cannot continue without the kernels intervention.

Three major types of memory-related hardware exceptions can occur: Major page faults Minor page faults Protection faults

9/3/2012

24

Advantages of Demand paging

Loading pages of memory on demand dramatically lowers the memory footprint

And startup time of the process.

9/3/2012

25

Memory sharing
Multiple users processes can share memory

Multiple processes can sharing program binaries and application data. The Solaris kernel introduced dynamically linked libraries.

9/3/2012

26

Memory protection
A users process must not be able access the memory of another process. A program fault in one program could cause another program (or the entire operating system) to fail. The protection is implemented by using protection modes read, write, execute and boundary checking.

9/3/2012

27

Kernel Virtual Memory Layout

The kernel uses virtual memory and MMU like the process. The kernel uses top 256 Mbytes or 512 Mbytes in common virtual address space. Most of the kernel memory are not pageable. This characteristics avoids the deadlocks. The kernel cannot rely on the global paging.

9/3/2012

28

Kernel Memory Allocation

Kernel memory is allocated at different levels. Page allocator It allocates unmapped pages from the free list to the kernel address space. Solaris uses Resource map allocator to allocate the free memory to the kernel. Resource map allocator uses first-fit algorithm.

9/3/2012

29

Kernel Memory Slab Allocator

Solaris provides a general-purpose memory allocator that provides arbitrarily sized memory allocations.

We use the slab allocator for memory requests that are: Smaller than a page size Not an even multiple of a page size Frequently going to be allocated and freed, so would otherwise fragment the kernel map

9/3/2012

30

Why Slab Allocator?

The reasons for introducing the slab allocator were as follows:

The SVR4 allocator was slow to satisfy allocation requests. Significant fragmentation problems arose with use of the SVR4 allocator. The allocator footprint was large, wasting a lot of memory. With no clean interfaces for memory allocation, code was duplicated in many places.

9/3/2012

31

9/3/2012

32

The slab allocator uses the term object to describe a single memory allocation unit. Cache to refer to a pool of like objects. Slab to refer to a group of objects that reside within the cache. Slab allocator solves many of the fragmentation issues by grouping different sized memory objects. Many cache will be activate at once in the solaris kernel.

9/3/2012

33

References
Book:

SOLARIS Internals,Core Kernel

Website: www.sun.com

9/3/2012

34

THANK YOU.

9/3/2012

35