Skip to contentSkip to conversation

Bluetooth: hci_conn: hold conn references in hci_sync tasks

PATCH Applied Pauli Virtanen 3 participants 10 messages Jul 25 – Jul 27

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

v5 cover Bluetooth: hci_conn: hold conn references in hci_sync tasks

Have hci_sync tasks hold reference to hci_conn pointer they want to use later.

Avoids UAFs and passing potentially reused (possible even if very unlikely) pointers to hci_conn_valid().

hci_conn_del() dequeues running works for the same connection, but this has no effect if the work is already started running, which is the race condition here.

v5:
- resend
- no need to check hci_conn_valid() in create_big_complete, the bit is safe to clear also if hci_conn_del() has run

v4:
- Check !conn in hci_connect_big_sync() first.
It's probably bug in iso.c that it may call this with NULL, but probably better fixed separately.

v3:
- resending some rebased parts from
https://lore.kernel.org/linux-bluetooth/[email protected]/
https://lore.kernel.org/linux-bluetooth/[email protected]/

Pauli Virtanen (7):
Bluetooth: hci_conn: hold conn reference in abort_conn_sync()
Bluetooth: hci_sync: hold conn in hci_connect_acl/le_sync() callbacks
Bluetooth: hci_sync: hold conn in hci_connect_big_sync() callback
Bluetooth: hci_sync: hold conn in hci_connect_pa_sync() callback
Bluetooth: hci_sync: hold conn in hci_past_sync() callback
Bluetooth: hci_sync: fix hci_conn_del() use in hci_le_create_conn_sync
Bluetooth: hci_sync: remove unnecessary hci_conn_get in
create_conn_sync

 net/bluetooth/hci_conn.c | 12 +++++-
 net/bluetooth/hci_sync.c | 85 +++++++++++++++++++++++++---------------
 2 files changed, 65 insertions(+), 32 deletions(-)
Signature · 1 line
2.55.0
#1 lore raw

PATCH v5 1/7 Bluetooth: hci_conn: hold conn reference in abort_conn_sync()

There is theoretical UAF if the conn is freed while the hci_sync task is running.

Hold refcount to avoid that.

Fixes 227a0cdf4a02 ("Bluetooth: MGMT: Fix not generating command complete for MGMT_OP_DISCONNECT")

---

Notes:
v5:
- no change
v4:
- no change
v3:
- split to multiple patches per different Fixes:

 net/bluetooth/hci_conn.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)
net/bluetooth/hci_conn.c +11 -1
index 1966cd153d97..6036ff66d8d9 100644
@@ -3163,6 +3163,13 @@ static int abort_conn_sync(struct hci_dev *hdev, void *data)
 	return hci_abort_conn_sync(hdev, conn, conn->abort_reason);
 }
 
+static void abort_conn_destroy(struct hci_dev *hdev, void *data, int err)
+{
+	struct hci_conn *conn = data;
+
+	hci_conn_put(conn);
+}
+
 int hci_abort_conn(struct hci_conn *conn, u8 reason)
 {
 	struct hci_dev *hdev = conn->hdev;
@@ -3188,7 +3195,10 @@ int hci_abort_conn(struct hci_conn *conn, u8 reason)
 	 * as a result to MGMT_OP_DISCONNECT/MGMT_OP_UNPAIR which does
 	 * already queue its callback on cmd_sync_work.
 	 */
-	err = hci_cmd_sync_run_once(hdev, abort_conn_sync, conn, NULL);
+	err = hci_cmd_sync_run_once(hdev, abort_conn_sync, hci_conn_get(conn),
+				    abort_conn_destroy);
+	if (err)
+		hci_conn_put(conn);
 	return (err == -EEXIST) ? 0 : err;
 }
 
Signature · 1 line
2.55.0
#2 lore raw

This is automated email and please do not reply to this email!

Dear submitter,

Thank you for submitting the patches to the linux bluetooth mailing list. This is a CI test results with your patch series:
PW Link:https://patchwork.kernel.org/project/bluetooth/list/?series=1134345

---Test result---

Test Summary:

CheckPatch                    PASS      4.97 seconds
VerifyFixes                   PASS      0.13 seconds
VerifySignedoff               PASS      0.13 seconds
GitLint                       FAIL      2.26 seconds
SubjectPrefix                 PASS      1.85 seconds
BuildKernel                   PASS      26.83 seconds
CheckAllWarning               PASS      29.51 seconds
CheckSparse                   PASS      28.39 seconds
BuildKernel32                 PASS      25.84 seconds
CheckKernelLLVM               SKIP      0.00 seconds
TestRunnerSetup               PASS      491.84 seconds
TestRunner_l2cap-tester       PASS      59.36 seconds
TestRunner_iso-tester         FAIL      90.00 seconds
TestRunner_bnep-tester        PASS      19.20 seconds
TestRunner_mgmt-tester        FAIL      217.48 seconds
TestRunner_rfcomm-tester      PASS      25.19 seconds
TestRunner_sco-tester         PASS      31.40 seconds
TestRunner_ioctl-tester       PASS      26.32 seconds
TestRunner_mesh-tester        FAIL      25.89 seconds
TestRunner_smp-tester         PASS      23.55 seconds
TestRunner_userchan-tester    PASS      20.01 seconds
TestRunner_6lowpan-tester     PASS      23.23 seconds
IncrementalBuild              PASS      43.48 seconds

Details
##############################
Test: GitLint - FAIL
Desc: Run gitlint
Output:
[v5,2/7] Bluetooth: hci_sync: hold conn in hci_connect_acl/le_sync() callbacks

18: B2 Line has trailing whitespace: " "
[v5,7/7] Bluetooth: hci_sync: remove unnecessary hci_conn_get in create_conn_sync

1: T1 Title exceeds max length (81>80): "[v5,7/7] Bluetooth: hci_sync: remove unnecessary hci_conn_get in create_conn_sync"
##############################
Test: CheckKernelLLVM - SKIP
Desc: Build kernel with LLVM + context analysis
Output:
Clang not found
##############################
Test: TestRunner_iso-tester - FAIL
Desc: Run iso-tester with test-runner
Output:
Total: 141, Passed: 134 (95.0%), Failed: 7, Not Run: 0

Failed Test Cases

ISO Disconnect - Success                             Timed out    2.653 seconds
ISO Reconnect - Success                              Timed out    1.995 seconds
ISO Reconnect Send and Receive #16 - Success         Timed out    1.995 seconds
ISO Reconnect AC 6(i) - Success                      Timed out    2.632 seconds
ISO Reconnect AC 6(ii) - Success                     Timed out    2.004 seconds
ISO Broadcaster Reconnect - Success                  Timed out    1.971 seconds
ISO Broadcaster Receiver Defer Reconnect - Success   Timed out    2.417 seconds

##############################
Test: TestRunner_mgmt-tester - FAIL
Desc: Run mgmt-tester with test-runner Output:
Total: 494, Passed: 489 (99.0%), Failed: 1, Not Run: 4

Failed Test Cases
Read Exp Feature - Success Failed 0.249 seconds
##############################
Test: TestRunner_mesh-tester - FAIL
Desc: Run mesh-tester with test-runner Output:
Total: 10, Passed: 8 (80.0%), Failed: 2, Not Run: 0

Failed Test Cases

Mesh - Send cancel - 1                               Timed out    2.648 seconds
Mesh - Send cancel - 2                               Timed out    1.988 seconds

https://github.com/bluez/bluetooth-next/pull/493

---
Regards,
Linux Bluetooth

#3 lore raw

PATCH v5 2/7 Bluetooth: hci_sync: hold conn in hci_connect_acl/le_sync() callbacks

In reply to Pauli Virtanen

There is theoretical UAF if the conn is freed while the hci_sync task is running.

Hold refcount to avoid that.

Fixes 881559af5f5c ("Bluetooth: hci_sync: Attempt to dequeue connection attempt")

---

Notes:
v5:
- no change
v4:
- no change
v3:
- split to multiple patches per different Fixes:

hci_conn_get() was added inside hci_le_create_conn_sync() in commit 76c2d047410ba, but it is too late to do there as the hci_conn_get() itself may be UAF.

 net/bluetooth/hci_sync.c | 32 ++++++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 8 deletions(-)
net/bluetooth/hci_sync.c +24 -8
index d5554fbf3129..83efca415eaf 100644
@@ -7128,12 +7128,23 @@ static int hci_acl_create_conn_sync(struct hci_dev *hdev, void *data)
 	return err;
 }
 
+static void hci_acl_create_conn_sync_complete(struct hci_dev *hdev, void *data,
+					      int err)
+{
+	struct hci_conn *conn = data;
+
+	hci_conn_put(conn);
+}
+
 int hci_connect_acl_sync(struct hci_dev *hdev, struct hci_conn *conn)
 {
 	int err;
 
-	err = hci_cmd_sync_queue_once(hdev, hci_acl_create_conn_sync, conn,
-				      NULL);
+	err = hci_cmd_sync_queue_once(hdev, hci_acl_create_conn_sync,
+				      hci_conn_get(conn),
+				      hci_acl_create_conn_sync_complete);
+	if (err)
+		hci_conn_put(conn);
 	return (err == -EEXIST) ? 0 : err;
 }
 
@@ -7144,36 +7155,41 @@ static void create_le_conn_complete(struct hci_dev *hdev, void *data, int err)
 	bt_dev_dbg(hdev, "err %d", err);
 
 	if (err == -ECANCELED)
-		return;
+		goto done;
 
 	hci_dev_lock(hdev);
 
 	if (!hci_conn_valid(hdev, conn))
-		goto done;
+		goto unlock;
 
 	if (!err) {
 		hci_connect_le_scan_cleanup(conn, 0x00);
-		goto done;
+		goto unlock;
 	}
 
 	/* Check if connection is still pending */
 	if (conn != hci_lookup_le_connect(hdev))
-		goto done;
+		goto unlock;
 
 	/* Flush to make sure we send create conn cancel command if needed */
 	flush_delayed_work(&conn->le_conn_timeout);
 	hci_conn_failed(conn, bt_status(err));
 
-done:
+unlock:
 	hci_dev_unlock(hdev);
+done:
+	hci_conn_put(conn);
 }
 
 int hci_connect_le_sync(struct hci_dev *hdev, struct hci_conn *conn)
 {
 	int err;
 
-	err = hci_cmd_sync_queue_once(hdev, hci_le_create_conn_sync, conn,
+	err = hci_cmd_sync_queue_once(hdev, hci_le_create_conn_sync,
+				      hci_conn_get(conn),
 				      create_le_conn_complete);
+	if (err)
+		hci_conn_put(conn);
 	return (err == -EEXIST) ? 0 : err;
 }
 
Signature · 1 line
2.55.0
#4 lore raw

PATCH v5 3/7 Bluetooth: hci_sync: hold conn in hci_connect_big_sync() callback

In reply to Pauli Virtanen

There is theoretical UAF if the conn is freed while the hci_sync task is running.

Hold refcount to avoid that. Handle NULL hcon, return 0 + do nothing to match the previous behavior.

Fixes 024421cf3992 ("Bluetooth: hci_conn: Fix not setting timeout for BIG Create Sync")

---

Notes:
v5:
- clear_bit() without hci_conn_valid check, we hold reference v4:
- check for conn == NULL before hci_conn_get(), since it appears iso.c may end up calling this with NULL
v3:
- split to multiple patches per different Fixes:
- hold RCU instead of hdev->lock

 net/bluetooth/hci_sync.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
net/bluetooth/hci_sync.c +12 -4
index 83efca415eaf..5610ef00fd0f 100644
@@ -7498,10 +7498,12 @@ static void create_big_complete(struct hci_dev *hdev, void *data, int err)
 	bt_dev_dbg(hdev, "err %d", err);
 
 	if (err == -ECANCELED)
-		return;
+		goto done;
 
-	if (hci_conn_valid(hdev, conn))
-		clear_bit(HCI_CONN_CREATE_BIG_SYNC, &conn->flags);
+	clear_bit(HCI_CONN_CREATE_BIG_SYNC, &conn->flags);
+
+done:
+	hci_conn_put(conn);
 }
 
 static int hci_le_big_create_sync(struct hci_dev *hdev, void *data)
@@ -7553,8 +7555,14 @@ int hci_connect_big_sync(struct hci_dev *hdev, struct hci_conn *conn)
 {
 	int err;
 
-	err = hci_cmd_sync_queue_once(hdev, hci_le_big_create_sync, conn,
+	if (!conn)
+		return 0;
+
+	err = hci_cmd_sync_queue_once(hdev, hci_le_big_create_sync,
+				      hci_conn_get(conn),
 				      create_big_complete);
+	if (err)
+		hci_conn_put(conn);
 	return (err == -EEXIST) ? 0 : err;
 }
 
Signature · 1 line
2.55.0
#5 lore raw

PATCH v5 4/7 Bluetooth: hci_sync: hold conn in hci_connect_pa_sync() callback

In reply to Pauli Virtanen

There is theoretical UAF if the conn is freed while the hci_sync task is running.

Hold refcount to avoid that.

Fixes 6d0417e4e1cf ("Bluetooth: hci_conn: Fix not setting conn_timeout for Broadcast Receiver")

---

Notes:
v5:
- no change
v4:
- no change
v3:
- split to multiple patches per different Fixes:

 net/bluetooth/hci_sync.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
net/bluetooth/hci_sync.c +7 -2
index 5610ef00fd0f..86c865306d21 100644
@@ -7312,7 +7312,7 @@ static void create_pa_complete(struct hci_dev *hdev, void *data, int err)
 	bt_dev_dbg(hdev, "err %d", err);
 
 	if (err == -ECANCELED)
-		return;
+		goto done;
 
 	hci_dev_lock(hdev);
 
@@ -7336,6 +7336,8 @@ static void create_pa_complete(struct hci_dev *hdev, void *data, int err)
 
 unlock:
 	hci_dev_unlock(hdev);
+done:
+	hci_conn_put(conn);
 }
 
 static int hci_le_past_params_sync(struct hci_dev *hdev, struct hci_conn *conn,
@@ -7486,8 +7488,11 @@ int hci_connect_pa_sync(struct hci_dev *hdev, struct hci_conn *conn)
 {
 	int err;
 
-	err = hci_cmd_sync_queue_once(hdev, hci_le_pa_create_sync, conn,
+	err = hci_cmd_sync_queue_once(hdev, hci_le_pa_create_sync,
+				      hci_conn_get(conn),
 				      create_pa_complete);
+	if (err)
+		hci_conn_put(conn);
 	return (err == -EEXIST) ? 0 : err;
 }
 
Signature · 1 line
2.55.0
#6 lore raw

PATCH v5 5/7 Bluetooth: hci_sync: hold conn in hci_past_sync() callback

In reply to Pauli Virtanen

Avoids giving freed pointers to hci_conn_valid(), which kmalloc may have reused.

Hold refcount to avoid that.

Fixes d3413703d5f8 ("Bluetooth: ISO: Add support to bind to trigger PAST")

---

Notes:
v5:
- no change
v4:
- no change
v3:
- split to multiple patches per different Fixes:

 net/bluetooth/hci_sync.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
net/bluetooth/hci_sync.c +8 -3
index 86c865306d21..63d0b89a88f3 100644
@@ -7582,6 +7582,8 @@ static void past_complete(struct hci_dev *hdev, void *data, int err)
 
 	bt_dev_dbg(hdev, "err %d", err);
 
+	hci_conn_put(past->conn);
+	hci_conn_put(past->le);
 	kfree(past);
 }
 
@@ -7646,8 +7648,8 @@ int hci_past_sync(struct hci_conn *conn, struct hci_conn *le)
 	if (!data)
 		return -ENOMEM;
 
-	data->conn = conn;
-	data->le = le;
+	data->conn = hci_conn_get(conn);
+	data->le = hci_conn_get(le);
 
 	if (conn->role == HCI_ROLE_MASTER)
 		err = hci_cmd_sync_queue_once(conn->hdev,
@@ -7657,8 +7659,11 @@ int hci_past_sync(struct hci_conn *conn, struct hci_conn *le)
 		err = hci_cmd_sync_queue_once(conn->hdev, hci_le_past_sync,
 					      data, past_complete);
 
-	if (err)
+	if (err) {
+		hci_conn_put(data->conn);
+		hci_conn_put(data->le);
 		kfree(data);
+	}
 
 	return (err == -EEXIST) ? 0 : err;
 }
Signature · 1 line
2.55.0
#7 lore raw

PATCH v5 6/7 Bluetooth: hci_sync: fix hci_conn_del() use in hci_le_create_conn_sync

In reply to Pauli Virtanen

hci_conn_del() caller must hold hdev->lock, check the conn was not concurrently deleted, and usually inform socket the conn is going to be deleted.

Use hci_abort_conn_sync() instead of calling hci_conn_del() without locks etc.

Fixes 8e8b92ee60de5 ("Bluetooth: hci_sync: Add hci_le_create_conn_sync")

---

Notes:
v5:
- no change
v4:
- no change
v3:
- use hci_abort_conn_sync instead of lock + hci_conn_valid + hci_conn_del

 net/bluetooth/hci_sync.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
net/bluetooth/hci_sync.c +3 -1
index 63d0b89a88f3..0f76221db2bd 100644
@@ -6733,7 +6733,9 @@ static int hci_le_create_conn_sync(struct hci_dev *hdev, void *data)
 		if (hci_dev_test_flag(hdev, HCI_LE_SCAN) &&
 		    hdev->le_scan_type == LE_SCAN_ACTIVE &&
 		    !hci_dev_test_flag(hdev, HCI_LE_SIMULTANEOUS_ROLES)) {
-			hci_conn_del(conn);
+			conn->state = BT_OPEN;
+			hci_abort_conn_sync(hdev, conn,
+					    HCI_ERROR_REJ_LIMITED_RESOURCES);
 			hci_conn_put(conn);
 			return -EBUSY;
 		}
Signature · 1 line
2.55.0
#8 lore raw

PATCH v5 7/7 Bluetooth: hci_sync: remove unnecessary hci_conn_get in create_conn_sync

In reply to Pauli Virtanen

hci_conn_get() without already held reference is data race against concurrent deletion.

In previous patches, the refcount has been changed to be taken before starting the hci_sync task, so remove these extra get() + put() as they are not needed.

Fixes 76c2d047410ba ("Bluetooth: hci_conn: Fix null ptr deref in hci_abort_conn()")

---

Notes:
v5:
- no change
v4:
- no change

 net/bluetooth/hci_sync.c | 13 -------------
 1 file changed, 13 deletions(-)
net/bluetooth/hci_sync.c +0 -13
index 0f76221db2bd..7779d9d1663a 100644
@@ -6717,11 +6717,6 @@ static int hci_le_create_conn_sync(struct hci_dev *hdev, void *data)
 
 	bt_dev_dbg(hdev, "conn %p", conn);
 
-	/* Hold a reference so conn stays valid for the HCI_CONN_CREATE
-	 * clear_bit() at done.
-	 */
-	hci_conn_get(conn);
-
 	clear_bit(HCI_CONN_SCANNING, &conn->flags);
 	conn->state = BT_CONNECT;
 
@@ -6736,7 +6731,6 @@ static int hci_le_create_conn_sync(struct hci_dev *hdev, void *data)
 			conn->state = BT_OPEN;
 			hci_abort_conn_sync(hdev, conn,
 					    HCI_ERROR_REJ_LIMITED_RESOURCES);
-			hci_conn_put(conn);
 			return -EBUSY;
 		}
 
@@ -6834,7 +6828,6 @@ static int hci_le_create_conn_sync(struct hci_dev *hdev, void *data)
 
 	/* Re-enable advertising after the connection attempt is finished. */
 	hci_resume_advertising_sync(hdev);
-	hci_conn_put(conn);
 	return err;
 }
 
@@ -7109,11 +7102,6 @@ static int hci_acl_create_conn_sync(struct hci_dev *hdev, void *data)
 	else
 		cp.role_switch = 0x00;
 
-	/* Hold a reference so conn stays valid for the HCI_CONN_CREATE
-	 * clear_bit() below.
-	 */
-	hci_conn_get(conn);
-
 	/* Mark create connection in flight so hci_cancel_connect_sync() can
 	 * cancel it while blocking on the connection complete event.
 	 */
@@ -7125,7 +7113,6 @@ static int hci_acl_create_conn_sync(struct hci_dev *hdev, void *data)
 				       conn->conn_timeout, NULL);
 
 	clear_bit(HCI_CONN_CREATE, &conn->flags);
-	hci_conn_put(conn);
 
 	return err;
 }
Signature · 1 line
2.55.0
#9 lore raw
In reply to Pauli Virtanen
This series was applied to bluetooth/bluetooth-next.git (master) by Luiz Augusto von Dentz <[email protected]>:

Hello:

This series was applied to bluetooth/bluetooth-next.git (master) by Luiz Augusto von Dentz <[email protected]>:

Quoted 11 lines · On Sat, 25 Jul 2026 12:59:16 +0300 you wrote:

Have hci_sync tasks hold reference to hci_conn pointer they want to use later.

Avoids UAFs and passing potentially reused (possible even if very unlikely) pointers to hci_conn_valid().

hci_conn_del() dequeues running works for the same connection, but this has no effect if the work is already started running, which is the race condition here.

[...]

Here is the summary with links:
- [v5,1/7] Bluetooth: hci_conn: hold conn reference in abort_conn_sync()

    https://git.kernel.org/bluetooth/bluetooth-next/c/3bf1394e8868
  - [v5,2/7] Bluetooth: hci_sync: hold conn in hci_connect_acl/le_sync() callbacks
    https://git.kernel.org/bluetooth/bluetooth-next/c/69a694ebe97f
  - [v5,3/7] Bluetooth: hci_sync: hold conn in hci_connect_big_sync() callback
    https://git.kernel.org/bluetooth/bluetooth-next/c/b1196afe58a9
  - [v5,4/7] Bluetooth: hci_sync: hold conn in hci_connect_pa_sync() callback
    https://git.kernel.org/bluetooth/bluetooth-next/c/cddfec3618d0
  - [v5,5/7] Bluetooth: hci_sync: hold conn in hci_past_sync() callback
    https://git.kernel.org/bluetooth/bluetooth-next/c/3c5f5c6be878
  - [v5,6/7] Bluetooth: hci_sync: fix hci_conn_del() use in hci_le_create_conn_sync
    https://git.kernel.org/bluetooth/bluetooth-next/c/423e5fc465eb
  - [v5,7/7] Bluetooth: hci_sync: remove unnecessary hci_conn_get in create_conn_sync
    https://git.kernel.org/bluetooth/bluetooth-next/c/73b6871b261f

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html