Kernel space and user space 2: content of the system calling user space in the kernel space

How to operate User Files in Linux kernel: Use get_fs () and set_fs (kernel_ds) to call the system call to change permissions, and then use the Kernel File Operation Function to access user space.

(1) macro Introduction
The source code is defined in include/ASM/uaccess. h as follows:

    #define MAKE_MM_SEG(s) ((mm_segment_t) { (s) })     #define KERNEL_DS MAKE_MM_SEG(0xFFFFFFFF)     #define USER_DS MAKE_MM_SEG(PAGE_OFFSET)     #define get_ds() (KERNEL_DS)     #define get_fs() (current->addr_limit)     #define set_fs(x) (current->addr_limit = (x)) 

And its annotations are also clear:/** the FS value determines whether argument validity checking shoshould be completed MED or not. If get_fs () = user_ds, checking is already med,
Get_fs () = kernel_ds, checking is bypassed. for historical reasons, these macros are grossly misnamed .*/

Therefore, we can see that the FS value is a flag for checking the parameters. When the system call parameters are required to come from the user space and use the system call in the kernel, set_fs (get_ds () changes the user space limit, that is, expands the user space range, therefore, you can use the parameters in the kernel.

(2) function example: Create a new text in a module and write content to the text.

#include <linux/kernel.h>#include <linux/module.h>#include <linux/init.h>#include <linux/fs.h>#include <linux/string.h>#include <linux/mm.h>#include <linux/syscalls.h>#include <asm/unistd.h>#include <asm/uaccess.h>#define MY_FILE "LogFile"//#define MY_FILE "/root/LogFile"asmlinkage long sys_open(const char __user *filename,int flags, int mode);char buf[128];struct file *file = NULL; static int __init init(void){        mm_segment_t old_fs;        int ret = 0;        printk("<0>""Hello, I'm the module that intends to write messages to file./n");        if(file == NULL)                file = filp_open(MY_FILE, O_RDWR | O_APPEND | O_CREAT, 0644);        //                 file = sys_open(MY_FILE, O_RDWR | O_APPEND | O_CREAT, 0644);        if (IS_ERR(file)) {                printk("<0>""error occured while opening file %s, exiting.../n", MY_FILE);                return 0;        }        memset(buf,0x61,128);        sprintf(buf,"%s", "The Messages is xjkdljflk.");        old_fs = get_fs();        printk("<0>""old_fs is %lx/n",old_fs.seg);        set_fs(KERNEL_DS);        old_fs = get_fs();        printk("<0>""old_fs is %lx/n",old_fs.seg);        printk("<0>""KERNEL_DS is %lx/n",KERNEL_DS);        ret = file->f_op->write(file, (char *)buf, sizeof(buf), &file->f_pos);        if(ret < 0)                printk("<0>""file write failed/n");        set_fs(old_fs);        return 0;}static void __exit fini(void){        if(file != NULL)                filp_close(file, NULL);}module_init(init);module_exit(fini);MODULE_LICENSE("GPL");

The makefile used is as follows:

ifeq ($(KERNELRELEASE),)KERNELDIR := /usr/src/linux-headers-2.6.32-28-genericPWD := $(shell pwd)modules: $(MAKE) -C $(KERNELDIR) M=$(PWD) modulesmodules_install: $(MAKE) -C $(KERNELDIR) M=$(PWD) modules_installclean: rm -rf *.o *~ core .depend .*.cmd *.ko *.mod.c .PHONY:modules modules_install cleanelseobj-m := call.oendif

After compilation, execute insmod and rmmod to view the desired result.

(3) process description

Use the flip_open function to open the file and obtain the pointer FP of struct file. Use the pointer FP for corresponding operations. For example, you can use FP-> f_ops-> Read to read the file, and finally use the filp_close () function to close the file.

In short: to use the system call in the kernel space, the parameter address passed to-> write () is the address of the kernel space, on user_ds (user_ds ~ Kernel_ds). If no other processing is performed, in the write () function, the address is deemed to have exceeded the user_ds range. Therefore, the address is regarded as a "deliberate destruction" of the user space ", further execution is not allowed. To solve this problem, set_fs (kernel_ds) expands the space limit that can be accessed to kernel_ds, so that the system can be called smoothly in the kernel!