四级页表结构
现在的64位Linux系统中,并没有使用全部的64位地址空间,而是使用其低48位,高16位并没有使用.
其中
39至47这9位用于索引PGD(page global directory),其中读取的值是PUD(page upper directory)的地址
30至38这9位用于索引PUD以获取PMD(page middle directory)的地址
21至29这9位用于索引PMD以获取PTE(the lowest level page table)的地址
12至20这9位用于索引PTE以获数据所在物理page frame的地址
最后的12位是page frame 的偏移,用于定位具体的数据所在地址
图形表示如图:
注:
原来的三级页表为:PGD-->PMD-->PTE
二级页表为:PGD(PGD)-->PTE
获取系统的page size
$ getconf PAGE_SIZE
获取当前内核使用多少级的页表结构
通过编译内核的config文件,搜索CONFIG_PGTABLE_LEVELS,既可以知道使用的是多少级的页表结构
内核地址空间中物理地址与虚拟地址的转换
#include <asm/io.h>
static inline void *phys_to_virt(phys_addr_t address);//__pa
static inline phys_addr_t virt_to_phys(volatile void *address)__va
读取任意程序内存实现
#include <linux/init.h>
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
#include <linux/sched.h>
#include <linux/sched/signal.h>
#include <linux/mm_types.h>
#include <asm/pgtable.h>
static ulong data_addr = 0;
static int data_len = 0;
static int target_pid = 0;
module_param(data_addr,ulong,S_IRUGO);
module_param(data_len,int,S_IRUGO);
module_param(target_pid,int,S_IRUGO);
static int __init main_init(void)
{
struct task_struct *task;
char *vaddr;
int retval = 0;
int i = 0;
pgd_t *pgd = NULL;
pud_t *pud = NULL;
pmd_t *pmd = NULL;
pte_t *pte = NULL;
unsigned long paddr = 0;
unsigned long page_addr = 0;
unsigned long page_offset = 0;
if(data_addr == 0 || data_len == 0 || target_pid == 0){
printk("insmod main <data_addr=?> <data len=?> <target_pid=?>\n");
return 0;
}
printk("data_addr:0x%lX, data_len:%d, target_pid:%d\n",data_addr,data_len,target_pid);
for_each_process(task){
if(task->pid == target_pid)
{
printk("find task:%s\n",task->comm);
retval = 1;
break;
}
}
if(retval == 0){
printk("not find task\n");
return -1;
}
pgd = pgd_offset(task->mm,data_addr);
if(pgd_none(*pgd)){
printk("not mapped in pgd\n");
return -1;
}
pud = pud_offset((p4d_t*)pgd,data_addr);
if(pud_none(*pud)){
printk("not mapped in pud\n");
return -1;
}
pmd = pmd_offset(pud,data_addr);
if(pmd_none(*pmd)){
printk("not mapped in pmd\n");
return -1;
}
pte = pte_offset_kernel(pmd,data_addr);
if(pte_none(*pte)){
printk("not mapped in pte\n");
return -1;
}
page_addr = pte_val(*pte) & PTE_PFN_MASK;
page_offset = (data_addr) & (0xFFF);
paddr = page_addr | page_offset;
vaddr = __va(paddr);
for(i = 0;i<data_len;i++){
printk("0x%X('%c')\n",vaddr[i],vaddr[i]);
}
return 0;
}
static void __exit main_exit(void)
{
}
module_init(main_init);
module_exit(main_exit);
MODULE_LICENSE("GPL");
makefile:
obj-m := main.o
all:
make -C /usr/src/linux-headers-$(shell uname -r) M=$(shell pwd) modules
clean:
make -C /usr/src/linux-headers-$(shell uname -r) M=$(shell pwd) clean
参考资料
How The Kernel Manages Your Memory:
https://manybutfinite.com/post/how-the-kernel-manages-your-memory/
Five-level page tables:
https://lwn.net/Articles/717293/
Page Table Management:
https://www.kernel.org/doc/gorman/html/understand/understand006.html
Page table
https://github.com/lorenzo-stoakes/linux-vm-notes/blob/master/sections/page-tables.md
Virtual memory map
https://www.kernel.org/doc/Documentation/x86/x86_64/mm.txt
Linux 内存管理分析
https://www.jianshu.com/p/fc719d1cfbc2
Linux中的虚拟地址转换物理地址实现