UBUNTU-CVE-2024-26960
In the Linux kernel, the following vulnerability has been resolved: mm: swap: fix race between free_swap_and_cache() and swapoff() There was previously a theoretical window where swapoff() could run and teardown a swap_info_struct while a call to free_swap_and_cache() was running in another thread. This could cause, amongst other bad possibilities, swap_page_trans_huge_swapped() (called by free_swap_and_cache()) to access the freed memory for swap_map. This is a theoretical problem and I haven't been able to provoke it from a test case. But there has been agreement based on code review that this is possible (see link below). Fix it by using get_swap_device()/put_swap_device(), which will stall swapoff(). There was an extra check in _swap_info_get() to confirm that the swap entry was not free. This isn't present in get_swap_device() because it doesn't make sense in general due to the race between getting the reference and swapoff. So I've added an equivalent check directly in free_swap_and_cache(). Details of how to provoke one possible issue (thanks to David Hildenbrand for deriving this): --8<----- __swap_entry_free() might be the last user and result in "count == SWAP_HAS_CACHE". swapoff->try_to_unuse() will stop as soon as soon as si->inuse_pages==0. So the question is: could someone reclaim the folio and turn si->inuse_pages==0, before we completed swap_page_trans_huge_swapped(). Imagine the following: 2 MiB folio in the swapcache. Only 2 subpages are still references by swap entries. Process 1 still references subpage 0 via swap entry. Process 2 still references subpage 1 via swap entry. Process 1 quits. Calls free_swap_and_cache(). -> count == SWAP_HAS_CACHE [then, preempted in the hypervisor etc.] Process 2 quits. Calls free_swap_and_cache(). -> count == SWAP_HAS_CACHE Process 2 goes ahead, passes swap_page_trans_huge_swapped(), and calls __try_to_reclaim_swap(). __try_to_reclaim_swap()->folio_free_swap()->delete_from_swap_cache()-> put_swap_folio()->free_swap_slot()->swapcache_free_entries()-> swap_entry_free()->swap_range_free()-> ... WRITE_ONCE(si->inuse_pages, si->inuse_pages - nr_entries); What stops swapoff to succeed after process 2 reclaimed the swap cache but before process1 finished its call to swap_page_trans_huge_swapped()? --8<-----
02 / AFFECTED SOFTWARE
Affected packages
12 explicit affected versions
1 explicit affected versions
6 explicit affected versions
1 explicit affected versions
74 explicit affected versions
3 explicit affected versions
1 explicit affected versions
11 explicit affected versions
31 explicit affected versions
6 explicit affected versions
96 explicit affected versions
21 explicit affected versions
16 explicit affected versions
10 explicit affected versions
10 explicit affected versions
45 explicit affected versions
50 explicit affected versions
6 explicit affected versions
12 explicit affected versions
19 explicit affected versions
54 explicit affected versions
4 explicit affected versions
40 explicit affected versions
67 explicit affected versions
51 explicit affected versions
40 explicit affected versions
10 explicit affected versions
3 explicit affected versions
47 explicit affected versions
7 explicit affected versions
46 explicit affected versions
94 explicit affected versions
13 explicit affected versions
65 explicit affected versions
35 explicit affected versions
7 explicit affected versions
14 explicit affected versions
18 explicit affected versions
9 explicit affected versions
96 explicit affected versions
34 explicit affected versions
5 explicit affected versions
13 explicit affected versions
4 explicit affected versions
23 explicit affected versions
7 explicit affected versions
1 explicit affected versions
51 explicit affected versions
5 explicit affected versions
7 explicit affected versions
12 explicit affected versions
38 explicit affected versions
12 explicit affected versions
14 explicit affected versions
7 explicit affected versions
84 explicit affected versions
99 explicit affected versions
107 explicit affected versions
12 explicit affected versions
95 explicit affected versions
10 explicit affected versions
10 explicit affected versions
37 explicit affected versions
13 explicit affected versions
10 explicit affected versions
27 explicit affected versions
35 explicit affected versions
26 explicit affected versions
13 explicit affected versions
6 explicit affected versions
1 explicit affected versions
29 explicit affected versions
87 explicit affected versions
13 explicit affected versions
99 explicit affected versions
51 explicit affected versions
1 explicit affected versions
82 explicit affected versions
49 explicit affected versions
11 explicit affected versions
33 explicit affected versions
23 explicit affected versions
10 explicit affected versions
94 explicit affected versions
11 explicit affected versions
12 explicit affected versions
117 explicit affected versions
7 explicit affected versions
6 explicit affected versions
12 explicit affected versions
97 explicit affected versions
39 explicit affected versions
32 explicit affected versions
110 explicit affected versions
48 explicit affected versions
130 explicit affected versions
10 explicit affected versions
13 explicit affected versions
26 explicit affected versions
6 explicit affected versions
33 explicit affected versions
1 explicit affected versions
5 explicit affected versions
117 explicit affected versions
57 explicit affected versions
16 explicit affected versions
51 explicit affected versions
40 explicit affected versions
3 explicit affected versions
71 explicit affected versions
1 explicit affected versions
2 explicit affected versions
64 explicit affected versions
51 explicit affected versions
70 explicit affected versions
3 explicit affected versions
1 explicit affected versions
12 explicit affected versions
37 explicit affected versions
27 explicit affected versions
23 explicit affected versions
8 explicit affected versions
12 explicit affected versions
5 explicit affected versions
8 explicit affected versions
1 explicit affected versions
2 explicit affected versions
13 explicit affected versions
86 explicit affected versions
13 explicit affected versions
51 explicit affected versions
7 explicit affected versions
72 explicit affected versions
43 explicit affected versions
10 explicit affected versions
12 explicit affected versions
12 explicit affected versions
86 explicit affected versions
44 explicit affected versions
16 explicit affected versions
13 explicit affected versions
26 explicit affected versions
37 explicit affected versions
37 explicit affected versions
6 explicit affected versions
14 explicit affected versions
38 explicit affected versions
40 explicit affected versions
9 explicit affected versions
40 explicit affected versions
11 explicit affected versions
48 explicit affected versions
7 explicit affected versions
21 explicit affected versions
58 explicit affected versions
121 explicit affected versions
100 explicit affected versions
4 explicit affected versions
1 explicit affected versions
46 explicit affected versions
93 explicit affected versions
13 explicit affected versions
43 explicit affected versions
1 explicit affected versions
52 explicit affected versions
144 explicit affected versions
10 explicit affected versions
8 explicit affected versions
14 explicit affected versions
87 explicit affected versions
4 explicit affected versions
76 explicit affected versions
11 explicit affected versions
1 explicit affected versions
47 explicit affected versions
03 / CONNECTIONS
Connected vulnerabilities
04 / EVIDENCE
Source records
In the Linux kernel, the following vulnerability has been resolved: mm: swap: fix race between free_swap_and_cache() and swapoff() There was previously a theoretical window where swapoff() could run and teardown a swap_info_struct while a call to free_swap_and_cache() was running in another thread. This could cause, amongst other bad possibilities, swap_page_trans_huge_swapped() (called by free_swap_and_cache()) to access the freed memory for swap_map. This is a theoretical problem and I haven't been able to provoke it from a test case. But there has been agreement based on code review that this is possible (see link below). Fix it by using get_swap_device()/put_swap_device(), which will stall swapoff(). There was an extra check in _swap_info_get() to confirm that the swap entry was not free. This isn't present in get_swap_device() because it doesn't make sense in general due to the race between getting the reference and swapoff. So I've added an equivalent check directly in free_swap_and_cache(). Details of how to provoke one possible issue (thanks to David Hildenbrand for deriving this): --8<----- __swap_entry_free() might be the last user and result in "count == SWAP_HAS_CACHE". swapoff->try_to_unuse() will stop as soon as soon as si->inuse_pages==0. So the question is: could someone reclaim the folio and turn si->inuse_pages==0, before we completed swap_page_trans_huge_swapped(). Imagine the following: 2 MiB folio in the swapcache. Only 2 subpages are still references by swap entries. Process 1 still references subpage 0 via swap entry. Process 2 still references subpage 1 via swap entry. Process 1 quits. Calls free_swap_and_cache(). -> count == SWAP_HAS_CACHE [then, preempted in the hypervisor etc.] Process 2 quits. Calls free_swap_and_cache(). -> count == SWAP_HAS_CACHE Process 2 goes ahead, passes swap_page_trans_huge_swapped(), and calls __try_to_reclaim_swap(). __try_to_reclaim_swap()->folio_free_swap()->delete_from_swap_cache()-> put_swap_folio()->free_swap_slot()->swapcache_free_entries()-> swap_entry_free()->swap_range_free()-> ... WRITE_ONCE(si->inuse_pages, si->inuse_pages - nr_entries); What stops swapoff to succeed after process 2 reclaimed the swap cache but before process1 finished its call to swap_page_trans_huge_swapped()? --8<-----
05 / REFERENCES
Further evidence
- https://git.kernel.org/linus/82b1c07a0af603e3c47b906c8e991dc96f01688e
- https://git.kernel.org/stable/c/0f98f6d2fb5fad00f8299b84b85b6bc1b6d7d19a
- https://git.kernel.org/stable/c/1ede7f1d7eed1738d1b9333fd1e152ccb450b86a
- https://git.kernel.org/stable/c/2da5568ee222ce0541bfe446a07998f92ed1643e
- https://git.kernel.org/stable/c/363d17e7f7907c8e27a9e86968af0eaa2301787b
- https://git.kernel.org/stable/c/3ce4c4c653e4e478ecb15d3c88e690f12cbf6b39
- https://git.kernel.org/stable/c/82b1c07a0af603e3c47b906c8e991dc96f01688e
- https://git.kernel.org/stable/c/d85c11c97ecf92d47a4b29e3faca714dc1f18d0d
- https://ubuntu.com/security/CVE-2024-26960
- https://ubuntu.com/security/notices/USN-6816-1
- https://ubuntu.com/security/notices/USN-6817-1
- https://ubuntu.com/security/notices/USN-6817-2
- https://ubuntu.com/security/notices/USN-6817-3
- https://ubuntu.com/security/notices/USN-6878-1
- https://ubuntu.com/security/notices/USN-6898-1
- https://ubuntu.com/security/notices/USN-6898-2
- https://ubuntu.com/security/notices/USN-6898-3
- https://ubuntu.com/security/notices/USN-6898-4
- https://ubuntu.com/security/notices/USN-6917-1
- https://ubuntu.com/security/notices/USN-6919-1
- https://ubuntu.com/security/notices/USN-6927-1
- https://ubuntu.com/security/notices/USN-7019-1
- https://ubuntu.com/security/notices/USN-7069-1
- https://ubuntu.com/security/notices/USN-7069-2
- https://ubuntu.com/security/notices/USN-7073-1
- https://ubuntu.com/security/notices/USN-7073-2
- https://ubuntu.com/security/notices/USN-7119-1
- https://www.cve.org/CVERecord?id=CVE-2024-26960