Link address in ARM (why ARM needs relocation)

    Before considering why relocation is necessary, we must first understand the following points:
    
    1. Definition of link address: My own understanding is that it is the memory address where the program is defined to start execution.
    
    2. When ARM starts from nandflash, the CPU automatically copies the first 8K code in nandflash to address 0 of 6410 (i.e. Stepping Stone), and then executes the program from address 0.
    
    3. During the execution of the program, the link address is used to access global variables. Global variables are generally stored at the end of the program.
    
    When the program exceeds 8K, if relocation is not performed, an access error similar to the following will occur, which will cause the program to fail to execute normally.
    
    Suppose we define the link address of the program as 0x50000000. When the program is executed from address 0 of 6410 (i.e. Stepping Stone), if a global variable is to be accessed, the disassembled code is as follows:
    500000a4 e59f303c ldr r0, [pc, #60] ;r0 = [0x500000e8] =0x00000200
    500000a8 e5933000 ldr r0, [r0] ;r0 = [0x00000200]
    500000ac ......
    ......
    500000e8 00000200 .......
    
    00000200 :
    200: 00000000 ......    
    

    We can see that the value of r0 is obtained from the address 0x500000e8, but the address 0x500000e8 cannot be found in Stepping Stone, so the program will fail. Therefore, after the program is copied to Stepping Stone by CPU, we need to write a program to initialize nandflash and memory, and then use position-independent code to copy the program from nandflash to memory, and then jump to other functions to continue execution. This is ARM relocation.

*************************************************** *************************************************** *************************************************** *************************************

The above is purely my personal opinion. If there are any mistakes, please correct me!