Linux Fifth Week study summary

Bérénice Angremy

I. Summary of knowledge points (i) Add time and Time-asm commands to Menuos

• 1. Update the menu code to the latest version
• 2.test.c in main function, add Menuconfig ()
• 3. Add a corresponding two functions, time and TIMEASM functions
• 4.make rootfs Automatic Compilation script

(ii) Call the kernel function using the GDB debug tracking system Sys_time

• After you set a breakpoint for the system call Systime that handles the time function, time is executed in Menuos. The discovery system stopped at Systime. Continue to press N to step into the schedule function.
• After the Sys_time returns to the assembly code processing, GDB cannot continue tracing.
• If the breakpoint is set at Syscall (entry32. S), and after entering C, it is found that it will not stop at Sys_call (because this is a system call function rather than a normal function).

(iii) process of system invocation in kernel code

1. Mechanism and initialization of system calls in kernel code

• Time is important throughout the system call process.

• In the case of System_call, the int 0x80 directive and the Systemcall are linked by an interrupt vector, and the API and the corresponding SYS are linked by the system call number.

• User state, the system calls XYZ () using int 0x80, which corresponds to calling System_call.

• The process on the right (assembly code) is very important to match by the system call number.

2. Initialization of the system invocation mechanism

trap_initThere is a function inside the function set_system_trap_gate that involves the interrupt vector of the system call SYSCALL_VECTOR and the assembly code entry system_call , once the int 0x80,cpu directly jumps to system_call execute.

3. Simplified, easy-to-understand system_call pseudo-code

The Systemcall is located at entry (Systemcall), and the other interrupts are treated similarly.

• Save_all: Save Site

• call *sys_call_table(,%eax,4)Call the system dispatch processing function, EAX is the system call number, is the actual system scheduler

• sys_call_table: System call Dispatch table

• syscall_after_all: Save return value

If there is sys_exit_work , then enter sys_exit_work : There will be a process scheduling timing.
work_pending -> work_notifysigTo handle the signal.

• Possible call schedule: Process dispatch code.
• May jump to Restore_all and resume the scene.

If none sys_exit_work , the recovery is performed restore_all and the user state is returned.

Interrupt_return <=> iret, end.

4. Simply browse system_call to the main code between the Iret

• 1.save_all: Save Site
• 2.syscall_call: Called the system call handler function
• 3.restore_all: Recovering the scene (because the system calls the handler function is also a special "interrupt")
• 4.syscallexitwork: As described in 3.
• 5.INTERRUPT RETURN: That is iret, the system call ends here

Second, the experiment: Analysis System_call Interrupt processing process one) use GDB trace to parse a system call kernel function (last week selected system call)--getpid

• 1. Execute RM MENU-RF First, forcibly delete the original menu folder, and update the menu code to the latest version using the git command.

• 2. Adding C functions, assembly functions in TEST.c

• 3.make Rootfs, enter help to see the addition of the commands we added earlier in QEMU.

System_call to Iret process flowchart

Linux Fifth Week study summary