mm/ksm: fix possible UAF of stable_node

The commit 2c653d0ee2ae ("ksm: introduce ksm_max_page_sharing per page
deduplication limit") introduced a possible failure case in the
stable_tree_insert(), where we may free the new allocated stable_node_dup
if we fail to prepare the missing chain node.

Then that kfolio return and unlock with a freed stable_node set...  And
any MM activities can come in to access kfolio->mapping, so UAF.

Fix it by moving folio_set_stable_node() to the end after stable_node
is inserted successfully.

Link: https://lkml.kernel.org/r/20240513-b4-ksm-stable-node-uaf-v1-1-f687de76f452@linux.dev
Fixes: 2c653d0ee2ae ("ksm: introduce ksm_max_page_sharing per page deduplication limit")
Signed-off-by: Chengming Zhou <chengming.zhou@linux.dev>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Stefan Roesch <shr@devkernel.io>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

diff --git a/mm/ksm.c b/mm/ksm.c
index f5138f43f0d27abfd866a31faa70a553635a9c75..452ac8346e6e280bc6d782ef609d1f2e53be617b 100644 (file)
--- a/mm/ksm.c
+++ b/mm/ksm.c
@@ -2153,7 +2153,6 @@ again:
 
        INIT_HLIST_HEAD(&stable_node_dup->hlist);
        stable_node_dup->kpfn = kpfn;
-       folio_set_stable_node(kfolio, stable_node_dup);
        stable_node_dup->rmap_hlist_len = 0;
        DO_NUMA(stable_node_dup->nid = nid);
        if (!need_chain) {
@@ -2172,6 +2171,8 @@ again:
                stable_node_chain_add_dup(stable_node_dup, stable_node);
        }
 
+       folio_set_stable_node(kfolio, stable_node_dup);
+
        return stable_node_dup;
 }