Process address space

Scene Introduction

When we learn various languages , You usually hear that the memory is divided into the following areas ：

We do an experiment with the following code ： At the beginning , We print val The value and address of , Because the child process did not modify the data , So the parent and child processes share a copy of the code , So at first , The values we print are the same . But when cnt==3 when , Subprocess modified g_val Value , So a write time copy is sent , At this point, the parent-child process prints g_val The value and address of should be different . But will the result be what we imagined ？

Found after running results ：g_val After modification , The printed values are different , But the address is the same ！

Why is the parent-child process g_val The value of is the same , The address is different ？ Because the space used by language programs is not physical memory , It's the process address space ！ If it is the same physical memory , It is impossible to have the same space , But different values .

Process address space

Processes do not use physical memory directly , Instead, use the process address space , Then map to the physical memory through the page table . When we create a process , In addition to creating pcb, Also create a structure — Process address space variables (mm_struct). The purpose of creating the process address space is to make the process think that it owns the operating system .
So the logic of the process using space is this ：

So in the first code , Father son process g_val The address is the same for different values of , Because g_val The address in the process address space of the parent-child process is the same , But the parent-child process g_val The addresses mapped to the physical space must be different .

After understanding what the process address space is , So why do you need a process address space ？

1. It is not safe to manipulate physical memory directly because it is not safe . If the program illegally accesses or destroys data , Will seriously threaten computer security . So the process address space is introduced to separate the program from the physical memory , Let the operation of physical memory be left to the operating system .
2. Divide the concepts of memory application and memory use in time , Through virtual address space , To mask the process of applying for memory at the bottom , Achieve process read-write memory and OS Perform memory management operations , Separate the software . for example ： When we apply 1000 A space of bytes , If you don't use this space immediately （ That is, do not use read-write operations ）, This space is not immediately available to you in physical memory . Because if you apply for space , But don't use , Other processes cannot use this space , That's a waste of space . therefore , When we apply for space , Only one will be opened in the process address space “ Virtual space ”, Tell you that this space has been applied for ; When you actually use this space by reading and writing , Will open up this space on the physical memory for you , And write the data to the physical space through the write time copy .
3. With an address space ,cpu You can go to a fixed location in the address space to read code and data , Then map through the page table , Find the real code and data ; With address space and page table mapping , When the disk loads data into physical memory , Can be loaded to any location in physical memory , Reduce the pressure of memory management （cpu To read data, just go to a fixed location , Find the data and give it to the operating system , Look through the address space and page table ）. If there is no address space ,cpu Read the code and data of each process , You have to go to different physical spaces , increase cpu burden .