* [PATCH 1/2] Bluetooth: SMP: Use pairing state for the debug key check
2026-11-21 6:57 [PATCH 0/2] Bluetooth: SMP: Avoid stale pairing key reads Cen Zhang
@ 2026-11-21 6:57 ` Cen Zhang
2026-11-21 6:57 ` [PATCH 2/2] Bluetooth: SMP: Use pairing data for IRK notifications Cen Zhang
1 sibling, 0 replies; 3+ messages in thread
From: Cen Zhang @ 2026-11-21 6:57 UTC (permalink / raw)
To: marcel, luiz.dentz
Cc: linux-bluetooth, linux-kernel, baijiaju1990, jjzuming, zzzccc427
A management LTK reload can reclaim the key saved in smp->ltk after
notification and before the successful pairing destroys its context.
load_long_term_keys() holds hdev->lock, but SMP receive processing runs
after the ACL receive path drops that lock:
SMP receive Management / RCU
save smp->ltk
notify the key
clear the LTK list
reclaim the saved key
smp_chan_destroy()
read smp->ltk->type
The destructor reads the key type even for an ordinary successful
pairing which needs no key removal. Derive the debug-key decision from
the pairing's Secure Connections and debug-key flags instead. Requiring
both flags preserves legacy fallback behavior without reading the
registry entry just to decide whether it is a debug key.
The excerpt below retains the headline and stack traces.
KASAN report as below:
BUG: KASAN: slab-use-after-free in smp_chan_destroy.isra.0+0xafd/0xe80
Read of size 1 at addr ffff88810c140928 by task kworker/u17:0/501
Workqueue: hci0 hci_rx_work
Call Trace:
<TASK>
dump_stack_lvl+0x93/0xd0
print_report+0xce/0x630
? smp_chan_destroy.isra.0+0xafd/0xe80
? srso_alias_return_thunk+0x5/0xfbef5
? __virt_addr_valid+0x20d/0x410
? smp_chan_destroy.isra.0+0xafd/0xe80
kasan_report+0xe0/0x110
? smp_chan_destroy.isra.0+0xafd/0xe80
smp_chan_destroy.isra.0+0xafd/0xe80
smp_distribute_keys+0xc57/0x29a0
? __pfx_smp_distribute_keys+0x10/0x10
? kasan_save_track+0x14/0x30
? srso_alias_return_thunk+0x5/0xfbef5
? __kasan_kmalloc+0xaa/0xb0
? smp_recv_cb+0x2feb/0x8970
? srso_alias_return_thunk+0x5/0xfbef5
? __kmalloc_cache_noprof+0x251/0x610
? smp_recv_cb+0x2feb/0x8970
? smp_recv_cb+0x30de/0x8970
? srso_alias_return_thunk+0x5/0xfbef5
smp_recv_cb+0x30de/0x8970
? srso_alias_return_thunk+0x5/0xfbef5
? srso_alias_return_thunk+0x5/0xfbef5
? __pfx_smp_recv_cb+0x10/0x10
? __pfx___might_resched+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? find_held_lock+0x2b/0x80
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_lock+0x182c/0x2590
? srso_alias_return_thunk+0x5/0xfbef5
? l2cap_recv_frame+0x9ef/0x8830
? srso_alias_return_thunk+0x5/0xfbef5
? _raw_spin_unlock_irqrestore+0x27/0x60
? srso_alias_return_thunk+0x5/0xfbef5
? try_to_wake_up+0xf98/0x1b80
? __pfx___mutex_lock+0x10/0x10
? __pfx_try_to_wake_up+0x10/0x10
? __lock_acquire+0x466/0x2260
? __pfx_l2cap_chan_hold_unless_zero+0x10/0x10
? l2cap_recv_frame+0x13cc/0x8830
? __pfx_smp_recv_cb+0x10/0x10
l2cap_recv_frame+0x13cc/0x8830
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_lock+0x182c/0x2590
? __pfx_l2cap_recv_frame+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? find_held_lock+0x2b/0x80
? l2cap_recv_acldata+0x1eb/0x10e0
? __pfx___mutex_lock+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_unlock_slowpath+0x147/0x920
? __pfx_l2cap_conn_hold_unless_zero+0x10/0x10
? __pfx_hci_conn_enter_active_mode+0x10/0x10
? l2cap_recv_acldata+0x13d/0x10e0
? srso_alias_return_thunk+0x5/0xfbef5
l2cap_recv_acldata+0xe9f/0x10e0
? srso_alias_return_thunk+0x5/0xfbef5
hci_rx_work+0x5a6/0xf00
? srso_alias_return_thunk+0x5/0xfbef5
process_one_work+0x908/0x19c0
? __pfx_process_one_work+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? lock_is_held_type+0x8f/0x100
? srso_alias_return_thunk+0x5/0xfbef5
worker_thread+0x65c/0xe40
? __pfx_worker_thread+0x10/0x10
kthread+0x34f/0x460
? srso_alias_return_thunk+0x5/0xfbef5
? __pfx_kthread+0x10/0x10
ret_from_fork+0x659/0x940
? __pfx_ret_from_fork+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? __switch_to+0x74f/0xf80
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1a/0x30
</TASK>
Allocated by task 502:
kasan_save_stack+0x33/0x60
kasan_save_track+0x14/0x30
__kasan_kmalloc+0xaa/0xb0
__kmalloc_cache_noprof+0x251/0x610
hci_add_ltk+0x1f5/0x3a0
sc_add_ltk+0x174/0x350
smp_recv_cb+0x62e8/0x8970
l2cap_recv_frame+0x13cc/0x8830
l2cap_recv_acldata+0xe9f/0x10e0
hci_rx_work+0x5a6/0xf00
process_one_work+0x908/0x19c0
worker_thread+0x65c/0xe40
kthread+0x34f/0x460
ret_from_fork+0x659/0x940
ret_from_fork_asm+0x1a/0x30
Freed by task 520:
kasan_save_stack+0x33/0x60
kasan_save_track+0x14/0x30
kasan_save_free_info+0x3b/0x60
__kasan_slab_free+0x5f/0x80
__rcu_free_sheaf_prepare+0x65/0x2a0
rcu_free_sheaf+0x1f/0xe0
rcu_core+0x661/0x1d10
handle_softirqs+0x201/0x930
__irq_exit_rcu+0x110/0x1e0
irq_exit_rcu+0xe/0x20
sysvec_apic_timer_interrupt+0x37/0x80
asm_sysvec_apic_timer_interrupt+0x1a/0x20
Fixes: 923e24143efe ("Bluetooth: Fix SMP debug key handling")
Signed-off-by: Cen Zhang <zzzccc427@gmail.com>
---
This patch was written with assistance from Codex (OpenAI).
net/bluetooth/smp.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/bluetooth/smp.c b/net/bluetooth/smp.c
index f8f86b8f73f05..ead1b8e3e361a 100644
--- a/net/bluetooth/smp.c
+++ b/net/bluetooth/smp.c
@@ -745,7 +745,8 @@ static void smp_chan_destroy(struct l2cap_conn *conn)
/* Ensure that we don't leave any debug key around if debug key
* support hasn't been explicitly enabled.
*/
- if (smp->ltk && smp->ltk->type == SMP_LTK_P256_DEBUG &&
+ if (smp->ltk && test_bit(SMP_FLAG_SC, &smp->flags) &&
+ test_bit(SMP_FLAG_DEBUG_KEY, &smp->flags) &&
!hci_dev_test_flag(hcon->hdev, HCI_KEEP_DEBUG_KEYS)) {
list_del_rcu(&smp->ltk->list);
kfree_rcu(smp->ltk, rcu);
^ permalink raw reply [flat|nested] 3+ messages in thread* [PATCH 2/2] Bluetooth: SMP: Use pairing data for IRK notifications
2026-11-21 6:57 [PATCH 0/2] Bluetooth: SMP: Avoid stale pairing key reads Cen Zhang
2026-11-21 6:57 ` [PATCH 1/2] Bluetooth: SMP: Use pairing state for the debug key check Cen Zhang
@ 2026-11-21 6:57 ` Cen Zhang
1 sibling, 0 replies; 3+ messages in thread
From: Cen Zhang @ 2026-11-21 6:57 UTC (permalink / raw)
To: marcel, luiz.dentz
Cc: linux-bluetooth, linux-kernel, baijiaju1990, jjzuming, zzzccc427
The IRK saved in smp->remote_irk can be reclaimed by Load IRKs before
pairing completes. The management device lock does not serialize with
the SMP receive callback, and the finder's RCU section has already ended:
SMP pairing Management / RCU
save IRK A in remote_irk
clear A and reclaim it
allocate B at A's address
clear B and reclaim it
smp_notify_keys()
mgmt_new_irk(remote_irk)
The notification can read freed storage or report a replacement key.
The later connection-address update also follows the saved pointer.
Build a temporary IRK from the identity address, key and RPA already
owned by the pairing context, and use that data for both the event and
the connection-address update. No registry pointer is dereferenced by
these notification operations.
The excerpt below retains the headline and stack traces.
Use a copy of the IRK for the later notification, as commit
9ff737119177 ("Bluetooth: hci_sync: Copy the resolving-list IRK under RCU")
does for the resolving list. The pairing context supplies the
identity data and key used here.
KASAN report as below:
BUG: KASAN: slab-use-after-free in mgmt_new_irk+0x296/0x2b0
Read of size 6 at addr ffff888110c0b8a0 by task kworker/u17:0/495
Workqueue: hci1 hci_rx_work
Call Trace:
<TASK>
dump_stack_lvl+0x93/0xd0
print_report+0xce/0x630
? mgmt_new_irk+0x296/0x2b0
? srso_alias_return_thunk+0x5/0xfbef5
? __virt_addr_valid+0x20d/0x410
? mgmt_new_irk+0x296/0x2b0
kasan_report+0xe0/0x110
? mgmt_new_irk+0x296/0x2b0
mgmt_new_irk+0x296/0x2b0
? __pfx_mgmt_new_irk+0x10/0x10
? __pfx_pmbd_probe_hit_cookie+0x10/0x10
smp_distribute_keys+0x54b/0x28a0
? __pfx_smp_distribute_keys+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? kasan_save_track+0x14/0x30
? srso_alias_return_thunk+0x5/0xfbef5
? __kasan_kmalloc+0xaa/0xb0
? smp_recv_cb+0x4d31/0x8de0
? srso_alias_return_thunk+0x5/0xfbef5
? __kmalloc_cache_noprof+0x251/0x610
? smp_recv_cb+0x4d31/0x8de0
? smp_recv_cb+0x4e24/0x8de0
? srso_alias_return_thunk+0x5/0xfbef5
smp_recv_cb+0x4e24/0x8de0
? __pfx_smp_recv_cb+0x10/0x10
? __pfx___might_resched+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? stack_depot_save_flags+0x29/0x9a0
? srso_alias_return_thunk+0x5/0xfbef5
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_lock+0x182c/0x2590
? kasan_save_stack+0x42/0x60
? l2cap_recv_frame+0x9ef/0x8830
? __kasan_slab_free+0x5f/0x80
? kmem_cache_free+0x242/0x580
? srso_alias_return_thunk+0x5/0xfbef5
? find_held_lock+0x2b/0x80
? l2cap_recv_frame+0xaef/0x8830
? __pfx___mutex_lock+0x10/0x10
? __lock_acquire+0x466/0x2260
? __pfx_l2cap_chan_hold_unless_zero+0x10/0x10
? l2cap_recv_frame+0x13cc/0x8830
? __pfx_smp_recv_cb+0x10/0x10
l2cap_recv_frame+0x13cc/0x8830
? srso_alias_return_thunk+0x5/0xfbef5
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_lock+0x182c/0x2590
? srso_alias_return_thunk+0x5/0xfbef5
? __pfx_l2cap_recv_frame+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? find_held_lock+0x2b/0x80
? l2cap_recv_acldata+0x1eb/0x10e0
? __pfx___mutex_lock+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? __mutex_unlock_slowpath+0x147/0x920
? __pfx_l2cap_conn_hold_unless_zero+0x10/0x10
? __pfx_hci_conn_enter_active_mode+0x10/0x10
? l2cap_recv_acldata+0x13d/0x10e0
? srso_alias_return_thunk+0x5/0xfbef5
l2cap_recv_acldata+0xe9f/0x10e0
? srso_alias_return_thunk+0x5/0xfbef5
hci_rx_work+0x5a6/0xf00
? srso_alias_return_thunk+0x5/0xfbef5
process_one_work+0x908/0x19c0
? __pfx_process_one_work+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? lock_is_held_type+0x8f/0x100
? srso_alias_return_thunk+0x5/0xfbef5
worker_thread+0x65c/0xe40
? __pfx_worker_thread+0x10/0x10
kthread+0x34f/0x460
? srso_alias_return_thunk+0x5/0xfbef5
? __pfx_kthread+0x10/0x10
ret_from_fork+0x659/0x940
? __pfx_ret_from_fork+0x10/0x10
? srso_alias_return_thunk+0x5/0xfbef5
? __switch_to+0x74f/0xf80
? __pfx_kthread+0x10/0x10
ret_from_fork_asm+0x1a/0x30
</TASK>
Allocated by task 802:
kasan_save_stack+0x33/0x60
kasan_save_track+0x14/0x30
__kasan_kmalloc+0xaa/0xb0
__kmalloc_cache_noprof+0x251/0x610
hci_add_irk+0x13e/0x2d0
load_irks+0x5bc/0x7d0
hci_sock_sendmsg+0x128b/0x22e0
__sys_sendto+0x425/0x470
__x64_sys_sendto+0xe5/0x1c0
do_syscall_64+0x115/0x6a0
entry_SYSCALL_64_after_hwframe+0x77/0x7f
Freed by task 802:
kasan_save_stack+0x33/0x60
kasan_save_track+0x14/0x30
kasan_save_free_info+0x3b/0x60
__kasan_slab_free+0x5f/0x80
__rcu_free_sheaf_prepare+0x65/0x2a0
rcu_free_sheaf+0x1f/0xe0
rcu_core+0x661/0x1d10
handle_softirqs+0x201/0x930
__irq_exit_rcu+0x110/0x1e0
irq_exit_rcu+0xe/0x20
sysvec_apic_timer_interrupt+0x6c/0x80
asm_sysvec_apic_timer_interrupt+0x1a/0x20
Fixes: 95fbac8a8e45 ("Bluetooth: Add support for sending New IRK event")
Signed-off-by: Cen Zhang <zzzccc427@gmail.com>
---
This patch was written with assistance from Codex (OpenAI).
net/bluetooth/smp.c | 18 +++++++++++++++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/net/bluetooth/smp.c b/net/bluetooth/smp.c
index ead1b8e3e361a..c758f7c927af6 100644
--- a/net/bluetooth/smp.c
+++ b/net/bluetooth/smp.c
@@ -1045,15 +1045,27 @@ static void smp_notify_keys(struct l2cap_conn *conn)
}
if (smp->remote_irk) {
- mgmt_new_irk(hdev, smp->remote_irk, persistent);
+ struct smp_irk irk = {
+ .addr_type = smp->id_addr_type,
+ };
+
+ /* The IRK list may have been replaced during pairing.
+ * Report the pairing data instead of the saved list entry.
+ */
+ bacpy(&irk.bdaddr, &smp->id_addr);
+ memcpy(irk.val, smp->irk, sizeof(irk.val));
+ if (hci_bdaddr_is_rpa(&hcon->dst, hcon->dst_type))
+ bacpy(&irk.rpa, &hcon->dst);
+
+ mgmt_new_irk(hdev, &irk, persistent);
/* Now that user space can be considered to know the
* identity address track the connection based on it
* from now on (assuming this is an LE link).
*/
if (hcon->type == LE_LINK) {
- bacpy(&hcon->dst, &smp->remote_irk->bdaddr);
- hcon->dst_type = smp->remote_irk->addr_type;
+ bacpy(&hcon->dst, &smp->id_addr);
+ hcon->dst_type = smp->id_addr_type;
/* Use a short delay to make sure the new address is
* propagated _before_ the channels.
*/
^ permalink raw reply [flat|nested] 3+ messages in thread