Skip to contentSkip to conversation

sctp: re-point retained control chunks on association migration

PATCH Superseded Jun Yang 2 participants 2 messages Aug 4 – Aug 5

Subsystems (3)

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Jun Yang

author
#0 lore raw

PATCH v2 1/1 sctp: re-point retained control chunks on association migration

From: Jun Yang <[email protected]>

sctp_control_set_owner_w() records the owning socket in a control chunk's skb->sk. sctp_sock_migrate() re-owns the association's DATA chunks via sctp_for_each_tx_datachunk(), but that walk keys off chunk->msg and so skips control chunks: any control chunk the association still holds (for example the saved stream-reset request asoc->strreset_chunk, the ASCONF request/ack lists, or asoc->addip_last_asconf) keeps pointing at the old socket after the association is moved to the new one.

Once the old socket is freed, a later retransmit reaches sctp_packet_transmit() -> skb_set_owner_w(head, chunk->skb->sk) and operates on the freed socket -- refcount_add() on its sk_wmem_alloc, then sk->sk_write_space() from sock_wfree() -- a use-after-free of struct sock.

Rename sctp_for_each_tx_datachunk() to sctp_for_each_tx_chunk() and walk the control chunks the association retains there as well, so migration re-owns them with the same clear/set bracketing already used for DATA chunks. sctp_set_owner_w_migrate() picks the right owner helper by testing chunk->msg, which is NULL for control chunks.

The per-chunk owner test that traverse_and_process() already applies is split out into sctp_process_tx_chunk() and reused for the control lists. A chunk can sit on two of them at once -- asoc->strreset_chunk and asoc->addip_last_asconf both stay queued on outqueue.control_chunk_list until they are flushed -- and the test keeps such a chunk from being cleared or re-owned twice, which would otherwise leak an shkey reference.

sctp_control_set_owner_w() re-reads chunk->shkey from asoc->shkey, so sctp_set_owner_w_migrate() releases the reference sctp_clear_owner_w() took by value instead of re-reading chunk->shkey, which would drop the wrong key if the active key changed while the chunk was queued.

Fixes d04adf1b3551 ("sctp: reset owner sk for data chunks on out queues when migrating a sock")
Assisted-by tencentos-corvus-ai:kimi-k3

---
This is based on David Lee's

[PATCH] sctp: hold shkey across socket migration
https://lore.kernel.org/netdev/[email protected]/

which adds sctp_set_owner_w_migrate()

v2:
- Rename sctp_for_each_tx_datachunk() to sctp_for_each_tx_chunk() and move the control-chunk traversal into it, rather than adding a separate sctp_for_each_tx_ctrlchunk() helper (Xin Long).
- Handle control chunks in sctp_set_owner_w_migrate() by testing chunk->msg, dropping the sctp_ctrl_set_owner_w() helper (Xin Long). Control chunks now go through the full clear/set bracketing instead of a bare skb->sk store, so sctp_control_set_owner_w() is no longer static.
- Factor the existing owner test out of traverse_and_process() into sctp_process_tx_chunk() so the control lists get it too.

v1: https://lore.kernel.org/netdev/[email protected]/

 include/net/sctp/sm.h    |  1 +
 net/sctp/sm_make_chunk.c |  2 +-
 net/sctp/socket.c        | 53 ++++++++++++++++++++++++++++++----------
 3 files changed, 42 insertions(+), 14 deletions(-)
include/net/sctp/sm.h +1 -0
index 3bfd261a53cc..76605d1ee839 100644
@@ -252,6 +252,7 @@ struct sctp_chunk *sctp_make_fwdtsn(const struct sctp_association *asoc,
 				    struct sctp_fwdtsn_skip *skiplist);
 struct sctp_chunk *sctp_make_auth(const struct sctp_association *asoc,
 				  __u16 key_id);
+void sctp_control_set_owner_w(struct sctp_chunk *chunk);
 struct sctp_chunk *sctp_make_strreset_req(const struct sctp_association *asoc,
 					  __u16 stream_num, __be16 *stream_list,
 					  bool out, bool in);
net/sctp/sm_make_chunk.c +1 -1
index 0ae30c3c8913..7684686798cf 100644
@@ -94,7 +94,7 @@ static void sctp_control_release_owner(struct sk_buff *skb)
 	}
 }
 
-static void sctp_control_set_owner_w(struct sctp_chunk *chunk)
+void sctp_control_set_owner_w(struct sctp_chunk *chunk)
 {
 	struct sctp_association *asoc = chunk->asoc;
 	struct sk_buff *skb = chunk->skb;
net/sctp/socket.c +40 -13

81 lines not on this page · raw message

Signature · 1 line
2.55.0
maintainer
#1 lore raw
Quoted 157 lines · Jun Yang

From: Jun Yang <[email protected]>

sctp_control_set_owner_w() records the owning socket in a control chunk's skb->sk. sctp_sock_migrate() re-owns the association's DATA chunks via sctp_for_each_tx_datachunk(), but that walk keys off chunk->msg and so skips control chunks: any control chunk the association still holds (for example the saved stream-reset request asoc->strreset_chunk, the ASCONF request/ack lists, or asoc->addip_last_asconf) keeps pointing at the old socket after the association is moved to the new one.

Once the old socket is freed, a later retransmit reaches sctp_packet_transmit() -> skb_set_owner_w(head, chunk->skb->sk) and operates on the freed socket -- refcount_add() on its sk_wmem_alloc, then sk->sk_write_space() from sock_wfree() -- a use-after-free of struct sock.

Rename sctp_for_each_tx_datachunk() to sctp_for_each_tx_chunk() and walk the control chunks the association retains there as well, so migration re-owns them with the same clear/set bracketing already used for DATA chunks. sctp_set_owner_w_migrate() picks the right owner helper by testing chunk->msg, which is NULL for control chunks.

The per-chunk owner test that traverse_and_process() already applies is split out into sctp_process_tx_chunk() and reused for the control lists. A chunk can sit on two of them at once -- asoc->strreset_chunk and asoc->addip_last_asconf both stay queued on outqueue.control_chunk_list until they are flushed -- and the test keeps such a chunk from being cleared or re-owned twice, which would otherwise leak an shkey reference.

sctp_control_set_owner_w() re-reads chunk->shkey from asoc->shkey, so sctp_set_owner_w_migrate() releases the reference sctp_clear_owner_w() took by value instead of re-reading chunk->shkey, which would drop the wrong key if the active key changed while the chunk was queued.

Fixes d04adf1b3551 ("sctp: reset owner sk for data chunks on out queues when migrating a sock")
Assisted-by tencentos-corvus-ai:kimi-k3

---
This is based on David Lee's

[PATCH] sctp: hold shkey across socket migration
https://lore.kernel.org/netdev/[email protected]/

which adds sctp_set_owner_w_migrate()

v2:
- Rename sctp_for_each_tx_datachunk() to sctp_for_each_tx_chunk() and move the control-chunk traversal into it, rather than adding a separate sctp_for_each_tx_ctrlchunk() helper (Xin Long).
- Handle control chunks in sctp_set_owner_w_migrate() by testing chunk->msg, dropping the sctp_ctrl_set_owner_w() helper (Xin Long). Control chunks now go through the full clear/set bracketing instead of a bare skb->sk store, so sctp_control_set_owner_w() is no longer static.
- Factor the existing owner test out of traverse_and_process() into sctp_process_tx_chunk() so the control lists get it too.

v1: https://lore.kernel.org/netdev/[email protected]/

 include/net/sctp/sm.h    |  1 +
 net/sctp/sm_make_chunk.c |  2 +-
 net/sctp/socket.c        | 53 ++++++++++++++++++++++++++++++----------
 3 files changed, 42 insertions(+), 14 deletions(-)
include/net/sctp/sm.h +1 -0
index 3bfd261a53cc..76605d1ee839 100644
@@ -252,6 +252,7 @@ struct sctp_chunk *sctp_make_fwdtsn(const struct sctp_association *asoc,
                                    struct sctp_fwdtsn_skip *skiplist);
 struct sctp_chunk *sctp_make_auth(const struct sctp_association *asoc,
                                  __u16 key_id);
+void sctp_control_set_owner_w(struct sctp_chunk *chunk);
 struct sctp_chunk *sctp_make_strreset_req(const struct sctp_association *asoc,
                                          __u16 stream_num, __be16 *stream_list,
                                          bool out, bool in);
net/sctp/sm_make_chunk.c +1 -1
index 0ae30c3c8913..7684686798cf 100644
@@ -94,7 +94,7 @@ static void sctp_control_release_owner(struct sk_buff *skb)
        }
 }
 
-static void sctp_control_set_owner_w(struct sctp_chunk *chunk)
+void sctp_control_set_owner_w(struct sctp_chunk *chunk)
 {
        struct sctp_association *asoc = chunk->asoc;
        struct sk_buff *skb = chunk->skb;
net/sctp/socket.c +29 -11
index 4a08023d52aa..d09b9f139070 100644
@@ -155,9 +155,24 @@ static void sctp_clear_owner_w(struct sctp_chunk *chunk)
 
 static void sctp_set_owner_w_migrate(struct sctp_chunk *chunk)
 {
-       sctp_set_owner_w(chunk);
-       if (chunk->shkey)
-               sctp_auth_shkey_release(chunk->shkey);
+       struct sctp_shared_key *shkey = chunk->shkey;
+
+       if (chunk->msg)
+               sctp_set_owner_w(chunk);
+       else
+               sctp_control_set_owner_w(chunk);
+
+       if (shkey)
+               sctp_auth_shkey_release(shkey);
+}
+
+static void sctp_process_tx_chunk(struct sctp_association *asoc,
+                                 struct sctp_chunk *chunk, bool clear,
+                                 void (*cb)(struct sctp_chunk *))
+{
+       if ((clear && asoc->base.sk == chunk->skb->sk) ||
+           (!clear && asoc->base.sk != chunk->skb->sk))
+               cb(chunk);
 }
 
 #define traverse_and_process() \
@@ -165,17 +180,14 @@ do {                              \
        msg = chunk->msg;       \
        if (msg == prev_msg)    \
                continue;       \
-       list_for_each_entry(c, &msg->chunks, frag_list) {       \
-               if ((clear && asoc->base.sk == c->skb->sk) ||   \
-                   (!clear && asoc->base.sk != c->skb->sk))    \
-                       cb(c);  \
-       }                       \
+       list_for_each_entry(c, &msg->chunks, frag_list) \
+               sctp_process_tx_chunk(asoc, c, clear, cb);      \
        prev_msg = msg;         \
 } while (0)
 
-static void sctp_for_each_tx_datachunk(struct sctp_association *asoc,
-                                      bool clear,
-                                      void (*cb)(struct sctp_chunk *))
+static void sctp_for_each_tx_chunk(struct sctp_association *asoc,
+                                  bool clear,
+                                  void (*cb)(struct sctp_chunk *))
 
 {
        struct sctp_datamsg *msg, *prev_msg = NULL;
@@ -198,6 +210,21 @@ static void sctp_for_each_tx_datachunk(struct sctp_association *asoc,
 
        list_for_each_entry(chunk, &q->out_chunk_list, list)
                traverse_and_process();
+
+       list_for_each_entry(chunk, &q->control_chunk_list, list)
+               sctp_process_tx_chunk(asoc, chunk, clear, cb);
+
+       list_for_each_entry(chunk, &asoc->asconf_ack_list, transmitted_list)
+               sctp_process_tx_chunk(asoc, chunk, clear, cb);

net/sctp/socket.c static void sctp_for_each_tx_datachunk(struct sctp_association *asoc, · line 228

Quoted 10 lines
net/sctp/socket.c +9 -0
+
+       list_for_each_entry(chunk, &asoc->addip_chunk_list, list)
+               sctp_process_tx_chunk(asoc, chunk, clear, cb);
+
+       if (asoc->strreset_chunk)
+               sctp_process_tx_chunk(asoc, asoc->strreset_chunk, clear, cb);
+
+       if (asoc->addip_last_asconf)
+               sctp_process_tx_chunk(asoc, asoc->addip_last_asconf, clear, cb);
 }

You can just use cb(...) here, no need to extract sctp_process_tx_chunk().

These checks of clear and asoc->base.sk were introduced by:

5c3e82fe1596 ("sctp: fix refcount bug in sctp_wfree")

The issue only existed on DATA chunks.

Also, please hold this patch until upstream applies the dependence:

https://lore.kernel.org/netdev/[email protected]/

Otherwise, your patch will fail on these sashiko reviews as "Failed To Apply".

Thanks.

Quoted 17 lines
net/sctp/socket.c +2 -3
 static void sctp_for_each_rx_skb(struct sctp_association *asoc, struct sock *sk,
@@ -9640,9 +9667,9 @@ static int sctp_sock_migrate(struct sock *oldsk, struct sock *newsk,
         * paths won't try to lock it and then oldsk.
         */
        lock_sock_nested(newsk, SINGLE_DEPTH_NESTING);
-       sctp_for_each_tx_datachunk(assoc, true, sctp_clear_owner_w);
+       sctp_for_each_tx_chunk(assoc, true, sctp_clear_owner_w);
        sctp_assoc_migrate(assoc, newsk);
-       sctp_for_each_tx_datachunk(assoc, false, sctp_set_owner_w_migrate);
+       sctp_for_each_tx_chunk(assoc, false, sctp_set_owner_w_migrate);
 
        /* If the association on the newsk is already closed before accept()
         * is called, set RCV_SHUTDOWN flag.
--

2.55.0