Message-ID: [ref] (raw)

There are two bugs in rose_rt_device_down() that can lead to
use-after-free:

1. The loop bound `t->count` is modified within the loop, which can
   cause the loop to terminate early and miss some entries.

2. When removing an entry from the neighbour array, the subsequent entries
   are moved up to fill the gap, but the loop index `i` is still
   incremented, causing the next entry to be skipped.

For example, if a node has three neighbours (A, B, A) and A is being
removed:
- 1st iteration (i=0): A is removed, array becomes (B, A, A), count=2
- 2nd iteration (i=1): We now check A instead of B, skipping B entirely
- 3rd iteration (i=2): Loop terminates early due to count=2

This leaves the second A in the array with count=2, but the rose_neigh
structure has been freed. Accessing code assumes that the first `count`
entries are valid pointers, causing a use-after-free when it accesses
the dangling pointer.

(Resending because I forgot to cite the patch, please ignore the former 
reply from me. Sorry for messing up.)

The example ([Senario2] below) in the commit message was incorrect. 
Correctly, UAF will happen in the [Senario1] below.

Let me clarify those senarios.

When the entries to be removed (A) are consecutive, the second A is not 
checked, leading to UAF.
[Senario1]
    (A, A, B) with count=3
    i=0: 
         (A, A, B) -> (A, B) with count=2
          ^ checked
    i=1: 
         (A, B) -> (A, B) with count=2
             ^ checked (B, not A!)
    i=2: (doesn't occur because i < count is false)
    ===> A remains with count=2 although A was freed, so UAF will happen.


When the entries to be removed (A) are not consecutive, all A entries are 
removed luckily.
[Senario2]
    (A, B, A) with count=3
    i=0: 
         (A, B, A) -> (B, A) with count=2
          ^ checked
    i=1: 
         (B, A) -> (B) with count=1
             ^ checked (A, not B)
    i=2: (doesn't occur because i < count is false)
    ===> No A remains. No UAF in this case.

Although, even in the senario2, the fundamental issue remains 
because B is never checked.
The fix addresses issues by preventing unintended skips.

Please let me know if I'm overlooking something or my understanding is 
incorrect. 
Thanks!