This is the directory
written in front
This article records my study career, learn a little and remember a little, and be ready to carry a bucket at any time.
1. Memory management
1. Segmentation
The virtual space of the memory space size required by the program is mapped to a certain physical address space.
Problem: The entire memory cannot be used efficiently, and it is easy to cause waste of memory (allocate physical memory for the program, and the program does not fully use the physical memory).
2. Pagination
Paging: 1. In order to solve the problems caused by segmentation, that is, the efficient use of memory; 1. For protection, the attributes and permissions of each page can be set individually.
Divide the memory into pages of fixed size, such as 1K or 4K (determined by the hardware), on this basis, the memory of the program can be further subdivided. Allocate part of the memory used by the program to physical memory, do not allocate actual physical memory to the part of memory that is not used yet, and allocate the actual physical address after subsequent use.
The VP0 and VP1 currently being used by the process are allocated to physical addresses PP2 and PP0, the other part of VP3 and VP2 are allocated to the disk, and the unused VP4, VP6 and VP7 are not allocated.
Now the translation of virtual address to physical address is done by dedicated hardware (MMU):
2. Thread management
3. Static library
1. Compile
The ac file is used for the demonstration of the following result steps
The documentation has been condensed as much as possible in order to simplify the display.
#define PI (111)
int main()
{
if(PI);
return 0;
}
1.1. Preprocessing
gcc -E a.c -i a.i
pi@NanoPi-NEO2:~/project/test$ cat b.i
# 1 "b.c"
# 1 "<built-in>"
# 1 "<command-line>"
# 31 "<command-line>"
# 1 "/usr/include/stdc-predef.h" 1 3 4
# 32 "<command-line>" 2
# 1 "b.c"
int main()
{
if((111));
return 0;
}
Expand the macro, replace the header file, remove the comment, add the line number, and keep the compilation command –>xx.i
1.2, compile
gcc -S a.i -i a.s
pi@NanoPi-NEO2:~/project/test$ cat b.s
.arch armv8-a
.file "b.c"
.text
.align 2
.global main
.type main, %function
main:
.LFB0:
.cfi_startproc
mov w0, 0
ret
.cfi_endproc
.LFE0:
.size main, .-main
.ident "GCC: (Ubuntu 9.3.0-17ubuntu1~20.04) 9.3.0"
.section .note.GNU-stack,"",@progbits
Generate assembly file –>xx.s
1.3, compilation
gcc -E a.c -i a.i
Generate machine code from assembly file –>xx.o
1.4. Links
gcc -E a.c -i a.i
Link together previous references (function variables) of the file.
A symbol table will be generated in different files, and all symbols (functions and variables) in the file are clearly indicated in the table, which is convenient for other files to reference.
–>xx.out
2. Compiler
3. Target file
.text
code snippet
.data
Initialized non-zero data segments (global variables, local static variables, allocated space occupying actual memory)
.bss
A data segment that is uninitialized or initialized to 0 (uninitialized may default to 0, there is no need to allocate space at this stage, so no space is allocated if it is empty, and only the symbol table is reserved) For example
:
#include"stdio.h"
#include"stdint.h"
uint16_t temp_1 = 222;
uint16_t temp_2;
int mian()
{
static uint16_t temp_3 = 111;
static uint16_t temp_4;
temp_1 = temp_3++;
temp_2 = temp_1++;
printf("this is test:%d\r\n",temp_1);
return 0;
}
Compile the above .c:
gcc -c main.c
Use objdump to view the information about the .o file generated by compiling the above code.
objdump -x -s -d main.o
The output file is as follows:
pi@NanoPi-NEO2:~/project/test$ objdump -x -s -d main.o
main.o: file format elf64-littleaarch64
main.o
architecture: aarch64, flags 0x00000011:
HAS_RELOC, HAS_SYMS
start address 0x0000000000000000
private flags = 0:
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00000088 0000000000000000 0000000000000000 00000040 2**2
CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
1 .data 00000004 0000000000000000 0000000000000000 000000c8 2**1
CONTENTS, ALLOC, LOAD, DATA
2 .bss 00000002 0000000000000000 0000000000000000 000000cc 2**1
ALLOC
3 .rodata 00000012 0000000000000000 0000000000000000 000000d0 2**3
CONTENTS, ALLOC, LOAD, READONLY, DATA
4 .comment 0000002b 0000000000000000 0000000000000000 000000e2 2**0
CONTENTS, READONLY
5 .note.GNU-stack 00000000 0000000000000000 0000000000000000 0000010d 2**0
CONTENTS, READONLY
6 .eh_frame 00000038 0000000000000000 0000000000000000 00000110 2**3
CONTENTS, ALLOC, LOAD, RELOC, READONLY, DATA
SYMBOL TABLE:
0000000000000000 l df *ABS* 0000000000000000 main.c
0000000000000000 l d .text 0000000000000000 .text
0000000000000000 l d .data 0000000000000000 .data
0000000000000000 l d .bss 0000000000000000 .bss
0000000000000000 l d .rodata 0000000000000000 .rodata
0000000000000002 l O .data 0000000000000002 temp_3.3838
0000000000000000 l O .bss 0000000000000002 temp_4.3839
0000000000000000 l d .note.GNU-stack 0000000000000000 .note.GNU-stack
0000000000000000 l d .eh_frame 0000000000000000 .eh_frame
0000000000000000 l d .comment 0000000000000000 .comment
0000000000000000 g O .data 0000000000000002 temp_1
0000000000000002 O *COM* 0000000000000002 temp_2
0000000000000000 g F .text 0000000000000088 mian
0000000000000000 *UND* 0000000000000000 printf
Contents of section .text:
0000 fd7bbfa9 fd030091 00000090 00000091 .{
..............
0010 00004079 01040011 223c0012 01000090 ..@y...."<......
0020 21000091 22000079 01000090 21000091 !..."..y....!...
0030 20000079 00000090 00000091 00004079 ..y..........@y
0040 01040011 223c0012 01000090 21000091 ...."<......!...
0050 22000079 01000090 210040f9 20000079 "..y....!.@. ..y
0060 00000090 00000091 00004079 e103002a ..........@y...*
0070 00000090 00000091 00000094 00008052 ...............R
0080 fd7bc1a8 c0035fd6 .{
...._.
Contents of section .data:
0000 de006f00 ..o.
Contents of section .rodata:
0000 74686973 20697320 74657374 3a25640d this is test:%d.
0010 0a00 ..
Contents of section .comment:
0000 00474343 3a202855 62756e74 7520392e .GCC: (Ubuntu 9.
0010 332e302d 31377562 756e7475 317e3230 3.0-17ubuntu1~20
0020 2e303429 20392e33 2e3000 .04) 9.3.0.
Contents of section .eh_frame:
0000 10000000 00000000 017a5200 04781e01 .........zR..x..
0010 1b0c1f00 20000000 18000000 00000000 .... ...........
0020 88000000 00410e10 9d029e01 60dedd0e .....A......`...
0030 00000000 00000000 ........
Disassembly of section .text:
0000000000000000 <mian>:
0: a9bf7bfd stp x29, x30, [sp, #-16]!
4: 910003fd mov x29, sp
8: 90000000 adrp x0, 0 <mian>
8: R_AARCH64_ADR_PREL_PG_HI21 .data+0x2
c: 91000000 add x0, x0, #0x0
c: R_AARCH64_ADD_ABS_LO12_NC .data+0x2
10: 79400000 ldrh w0, [x0]
14: 11000401 add w1, w0, #0x1
18: 12003c22 and w2, w1, #0xffff
1c: 90000001 adrp x1, 0 <mian>
1c: R_AARCH64_ADR_PREL_PG_HI21 .data+0x2
20: 91000021 add x1, x1, #0x0
20: R_AARCH64_ADD_ABS_LO12_NC .data+0x2
24: 79000022 strh w2, [x1]
28: 90000001 adrp x1, 0 <mian>
28: R_AARCH64_ADR_PREL_PG_HI21 temp_1
2c: 91000021 add x1, x1, #0x0
2c: R_AARCH64_ADD_ABS_LO12_NC temp_1
30: 79000020 strh w0, [x1]
34: 90000000 adrp x0, 0 <mian>
34: R_AARCH64_ADR_PREL_PG_HI21 temp_1
38: 91000000 add x0, x0, #0x0
38: R_AARCH64_ADD_ABS_LO12_NC temp_1
3c: 79400000 ldrh w0, [x0]
40: 11000401 add w1, w0, #0x1
44: 12003c22 and w2, w1, #0xffff
48: 90000001 adrp x1, 0 <mian>
48: R_AARCH64_ADR_PREL_PG_HI21 temp_1
4c: 91000021 add x1, x1, #0x0
4c: R_AARCH64_ADD_ABS_LO12_NC temp_1
50: 79000022 strh w2, [x1]
54: 90000001 adrp x1, 2 <mian+0x2>
54: R_AARCH64_ADR_GOT_PAGE temp_2
58: f9400021 ldr x1, [x1]
58: R_AARCH64_LD64_GOT_LO12_NC temp_2
5c: 79000020 strh w0, [x1]
60: 90000000 adrp x0, 0 <mian>
60: R_AARCH64_ADR_PREL_PG_HI21 temp_1
64: 91000000 add x0, x0, #0x0
64: R_AARCH64_ADD_ABS_LO12_NC temp_1
68: 79400000 ldrh w0, [x0]
6c: 2a0003e1 mov w1, w0
70: 90000000 adrp x0, 0 <mian>
70: R_AARCH64_ADR_PREL_PG_HI21 .rodata
74: 91000000 add x0, x0, #0x0
74: R_AARCH64_ADD_ABS_LO12_NC .rodata
78: 94000000 bl 0 <printf>
78: R_AARCH64_CALL26 printf
7c: 52800000 mov w0, #0x0 // #0
80: a8c17bfd ldp x29, x30, [sp], #16
84: d65f03c0 ret
The .data segment is initialized with non-zero data, temp_1(de) and temp_3(6f). However, temp_2 and temp_4 have not allocated space and are stored in bss.
attribute
attribute ((section("dame"))) , add this before the function or variable, which means to place the function or variable in the name section. As follows, add adamepart
#include"stdio.h"
#include"stdint.h"
__attribute__((section(".demo"))) uint8_t tttt;
int mian()
{
printf("this is test\r\n");
return 0;
}
pi@NanoPi-NEO2:~/project/test$ objdump -x -s -d main_1.o
main_1.o: file format elf64-littleaarch64
main_1.o
architecture: aarch64, flags 0x00000011:
HAS_RELOC, HAS_SYMS
start address 0x0000000000000000
private flags = 0:
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00000020 0000000000000000 0000000000000000 00000040 2**2
CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
1 .data 00000000 0000000000000000 0000000000000000 00000060 2**0
CONTENTS, ALLOC, LOAD, DATA
2 .bss 00000000 0000000000000000 0000000000000000 00000060 2**0
ALLOC
3 .demo 00000001 0000000000000000 0000000000000000 00000060 2**0
CONTENTS, ALLOC, LOAD, DATA
4 .rodata 0000000e 0000000000000000 0000000000000000 00000068 2**3
CONTENTS, ALLOC, LOAD, READONLY, DATA
5 .comment 0000002b 0000000000000000 0000000000000000 00000076 2**0
CONTENTS, READONLY
6 .note.GNU-stack 00000000 0000000000000000 0000000000000000 000000a1 2**0
CONTENTS, READONLY
7 .eh_frame 00000038 0000000000000000 0000000000000000 000000a8 2**3
CONTENTS, ALLOC, LOAD, RELOC, READONLY, DATA
3.1. Symbols
View the symbol table of the file:
Create a new main_1.c file
#include"stdio.h"
#include"stdint.h"
__attribute__((section(".demo"))) uint8_t tttt;
uint8_t temp_1 = 0;
uint8_t temp_2 = 0;
void fun(uint8_t test)
{
printf("this is test:%d\r\n",test);
}
int mian()
{
temp_1 = temp_2++;
fun(temp_1);
return 0;
}
Compile and view the file content:
pi@NanoPi-NEO2:~/project/test$ readelf -s main_1.o
Symbol table '.symtab' contains 21 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
1: 0000000000000000 0 FILE LOCAL DEFAULT ABS main_1.c
2: 0000000000000000 0 SECTION LOCAL DEFAULT 1
3: 0000000000000000 0 SECTION LOCAL DEFAULT 3
4: 0000000000000000 0 SECTION LOCAL DEFAULT 4
5: 0000000000000000 0 SECTION LOCAL DEFAULT 5
6: 0000000000000000 0 NOTYPE LOCAL DEFAULT 5 $d
7: 0000000000000000 0 NOTYPE LOCAL DEFAULT 4 $d
8: 0000000000000000 0 SECTION LOCAL DEFAULT 6
9: 0000000000000000 0 NOTYPE LOCAL DEFAULT 6 $d
10: 0000000000000000 0 NOTYPE LOCAL DEFAULT 1 $x
11: 0000000000000000 0 SECTION LOCAL DEFAULT 8
12: 0000000000000014 0 NOTYPE LOCAL DEFAULT 9 $d
13: 0000000000000000 0 SECTION LOCAL DEFAULT 9
14: 0000000000000000 0 SECTION LOCAL DEFAULT 7
15: 0000000000000000 1 OBJECT GLOBAL DEFAULT 5 tttt
16: 0000000000000000 1 OBJECT GLOBAL DEFAULT 4 temp_1
17: 0000000000000001 1 OBJECT GLOBAL DEFAULT 4 temp_2
18: 0000000000000000 44 FUNC GLOBAL DEFAULT 1 fun
19: 0000000000000000 0 NOTYPE GLOBAL DEFAULT UND printf
20: 000000000000002c 80 FUNC GLOBAL DEFAULT 1 mian
The above file has fun and printf functions. The fun function can find the corresponding definition, but printf cannot find it. The previous ndx is UND (undefined).
3.2. Compatible with C language – extern C
extern "C" {
int func(int);
int var;
}
Why do you do this?
After compiling in C++, the function name or variable name will be repackaged and modified, that is, the symbol _ ZN3fun3barE will be generated after fun is compiled .
In C, _ fun or fun will be generated , depending on the support of the compiler.
After using extern C, C++ will install the files in brackets in the compiled format of C language to generate a symbol table.
/*d.cpp*/
#include<stdio.h>
#define PI (111)
extern "C" {
void funci(){
;}}
static void func(float){
;}
void ffunc(int){
;}
int main()
{
if(PI);
return 0;
}
Compile and debug to view:
pi@NanoPi-NEO2:~/project/test$ g++ -c d.cpp
pi@NanoPi-NEO2:~/project/test$ readelf -s d.o
Symbol table '.symtab' contains 14 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
1: 0000000000000000 0 FILE LOCAL DEFAULT ABS d.cpp
2: 0000000000000000 0 SECTION LOCAL DEFAULT 1
3: 0000000000000000 0 SECTION LOCAL DEFAULT 2
4: 0000000000000000 0 SECTION LOCAL DEFAULT 3
5: 0000000000000000 0 NOTYPE LOCAL DEFAULT 1 $x
6: 0000000000000008 20 FUNC LOCAL DEFAULT 1 _ZL4funcf
7: 0000000000000000 0 SECTION LOCAL DEFAULT 5
8: 0000000000000014 0 NOTYPE LOCAL DEFAULT 6 $d
9: 0000000000000000 0 SECTION LOCAL DEFAULT 6
10: 0000000000000000 0 SECTION LOCAL DEFAULT 4
11: 0000000000000000 8 FUNC GLOBAL DEFAULT 1 func
12: 000000000000001c 20 FUNC GLOBAL DEFAULT 1 _Z5ffunci
13: 0000000000000030 8 FUNC GLOBAL DEFAULT 1 main
func does not use the C++ symbol naming method. fun is named using C++ symbols.
_Z5ffunci:
_Z: fixed character
5: function name has 5 characters
i: int type (f: float, v: void...)
4. Link – ld
According to the above, each file will generate a symbol table, such as ao and bo two files, how to link ao and bo together to generate an executable file?
There are two methods of linking:
1. Combine files in order according to the order of the files, so that each file will have repeated .text and .data. . . etc.
2. Combine all segments with the same attribute, that is, text together and data together. (mainstream solution)
example:
#include"stdio.h"
#include"stdint.h"
void fun(uint8_t test)
{
;
}
int test()
{
fun(1);
return 0;
}
#include"stdio.h"
#include"stdint.h"
int main()
{
test();
return 0;
}
gcc -c main.c main_1.c
ld main.o main_1.o -e main -o ab
-e mainIndicates that the main function is used as the program entry, and the default program entry of the ld linker is _start.
-o abIndicates that the link output file name is ab, and the default is a.out.
All the same attribute pairs in the ab file generated after execution will correspond together.
pi@NanoPi-NEO2:~/project/test$ objdump -h main.o
main.o: file format elf64-littleaarch64
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00000018 0000000000000000 0000000000000000 00000040 2**2
CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
1 .data 00000000 0000000000000000 0000000000000000 00000058 2**0
CONTENTS, ALLOC, LOAD, DATA
2 .bss 00000000 0000000000000000 0000000000000000 00000058 2**0
ALLOC
pi@NanoPi-NEO2:~/project/test$ objdump -h main_1.o
main_1.o: file format elf64-littleaarch64
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 00000030 0000000000000000 0000000000000000 00000040 2**2
CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
1 .data 00000000 0000000000000000 0000000000000000 00000070 2**0
CONTENTS, ALLOC, LOAD, DATA
2 .bss 00000000 0000000000000000 0000000000000000 00000070 2**0
ALLOC
pi@NanoPi-NEO2:~/project/test$ objdump -h ab
ab: file format elf64-littleaarch64
Sections:
Idx Name Size VMA LMA File off Algn
0 .text 000000c4 0000000000400120 0000000000400120 00000120 2**2
CONTENTS, ALLOC, LOAD, READONLY, CODE
1 .eh_frame 00000060 00000000004001e8 00000000004001e8 000001e8 2**3
CONTENTS, ALLOC, LOAD, READONLY, DATA
2 .got 00000010 0000000000410fd8 0000000000410fd8 00000fd8 2**3
CONTENTS, ALLOC, LOAD, DATA
3 .got.plt 00000018 0000000000410fe8 0000000000410fe8 00000fe8 2**3
CONTENTS, ALLOC, LOAD, DATA
4 .data 00000004 0000000000411000 0000000000411000 00001000 2**1
CONTENTS, ALLOC, LOAD, DATA
5 .bss 0000000c 0000000000411004 0000000000411004 00001004 2**1
ALLOC
6 .comment 0000002a 0000000000000000 0000000000000000 00001004 2**0
CONTENTS, READONLY
View the generated file symbol table:
pi@NanoPi-NEO2:~/project/test$ readelf -s ab
Symbol table '.symtab' contains 20 entries:
Num: Value Size Type Bind Vis Ndx Name
0: 0000000000000000 0 NOTYPE LOCAL DEFAULT UND
1: 00000000004000b0 0 SECTION LOCAL DEFAULT 1
2: 00000000004000f8 0 SECTION LOCAL DEFAULT 2
3: 0000000000000000 0 SECTION LOCAL DEFAULT 3
4: 0000000000000000 0 FILE LOCAL DEFAULT ABS main.c
5: 00000000004000b0 0 NOTYPE LOCAL DEFAULT 1 $x
6: 000000000040010c 0 NOTYPE LOCAL DEFAULT 2 $d
7: 0000000000000000 0 FILE LOCAL DEFAULT ABS main_1.c
8: 00000000004000c8 0 NOTYPE LOCAL DEFAULT 1 $x
9: 0000000000400130 0 NOTYPE LOCAL DEFAULT 2 $d
10: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 _bss_end__
11: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 __bss_start__
12: 00000000004000c8 20 FUNC GLOBAL DEFAULT 1 fun
13: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 __bss_end__
14: 00000000004000dc 28 FUNC GLOBAL DEFAULT 1 test
15: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 __bss_start
16: 00000000004000b0 24 FUNC GLOBAL DEFAULT 1 main
17: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 __end__
18: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 _edata
19: 0000000000410fe8 0 NOTYPE GLOBAL DEFAULT 2 _end
Each function corresponds to a unique address. Symbols that cannot be found in the compilation of a single file will be temporarily replaced with a fake address, and the real address will not be found and replaced until linking.
For example: before the link, if main.c cannot find the test function, it will set the address of the test function to 0, and the
linked file will fill in the corresponding address.
