CVE-2026-99012 in Linux
Summary
by MITRE • 10/23/2026
In the Linux kernel, the following vulnerability has been resolved:
Bluetooth: ISO: Serialize concurrent connect calls
iso_sock_connect() checks the socket state before taking the socket lock and drops the lock again before setting up a connection. Two callers can both pass the admission check while the socket is open or bound.
Caller A can copy its destination for route selection, then caller B can replace the socket destination before A binds or connects the CIS. A then uses B's destination with the route selected for its own request. B can also proceed after A attaches a connection and sets BT_CONNECT. For deferred BIS setup, B can bind a second BIS and overwrite iso_pi(sk)->conn, leaving the first connection's reference and conn->sk back-pointer stranded. Concurrent CIS connects can likewise replace the socket's connection. Later teardown only detaches the current connection, so a callback on the old connection can race with socket release and access a freed socket.
KASAN reported:
BUG: KASAN: slab-use-after-free in iso_sock_hold+0xf7/0x1b0 Call Trace: iso_sock_hold+0xf7/0x1b0 iso_conn_del+0x7b/0x1d0 iso_connect_cfm+0x186/0x16a0 hci_conn_failed+0x154/0x280 hci_abort_conn_sync+0x3dc/0x7d0 hci_cmd_sync_work+0x173/0x300
Allocated by task 121: sk_alloc+0x2b/0x6d0 bt_sock_alloc+0x29/0x370 iso_sock_alloc.constprop.0+0x19/0x300 iso_sock_create+0x94/0x100
Freed by task 121: kfree+0x121/0x3c0 __sk_destruct+0x42b/0x540 iso_sock_release+0x29d/0x340 __sock_release+0xa1/0x260
Check admission under the socket lock and mark the connect operation in progress before publishing its destination. Keep that flag set across route lookup and connection setup, rejecting another connect with -EBADFD before it can change the destination or attach a connection. Clear the flag after either helper returns, including on failure, so a failed setup can be retried.
A separate flag is needed because the socket lock must be released before acquiring the HCI device lock, while BT_CONNECT requires an attached connection. Keep the existing state transitions and the deferred CIS completion through iso_sock_recvmsg().
Reject an existing socket attachment in __iso_chan_add() before taking a new reference or publishing either pointer, so every caller preserves the connection association. Keep the same-socket, same-connection success case for deferred CIS setup. Reject an attached socket before BIS setup so a reusable BIS is not claimed before the attachment is rejected.
In CIS setup, reject a different HCI connection before iso_conn_add() and drop the per-attempt HCI hold directly. An unowned iso_conn may still carry another reference after detachment, so its destruction cannot be relied on to release that hold.
iso_listen_bis() creates a fresh PA HCI/ISO pair under the device lock, so the temporary iso_conn_put() frees the ISO candidate and drops its HCI hold on rejection. iso_conn_ready() uses a freshly allocated, unpublished child with no connection, making the socket-side rejection unreachable.
If you want to get the best quality for vulnerability data then you always have to consider VulDB.
Analysis
by VulDB Data Team • 10/23/2026
The Linux kernel Bluetooth subsystem contains a critical race condition vulnerability within the Isochronous Sockets (ISO) implementation that allows concurrent connect calls to corrupt internal state and lead to use-after-free memory errors. The root cause lies in iso_sock_connect(), which performs admission checks on the socket state before acquiring the necessary socket lock, subsequently releasing this lock prior to establishing a connection. This window of opportunity permits multiple threads or processes to pass initial validation checks simultaneously while the socket remains open or bound. When two callers proceed concurrently, one caller may copy its destination address for route selection only for another caller to overwrite that same destination before the first caller binds or connects the Connectionless Isochronous Stream (CIS). Consequently, the first caller inadvertently utilizes the second caller's destination with a routing table entry selected specifically for the second request. This misalignment creates an inconsistent state where connection parameters do not match the intended target, leading to unpredictable behavior and potential security breaches.
The operational impact of this vulnerability is severe, manifesting as kernel crashes due to use-after-free conditions detected by Kernel Address Sanitizer (KASAN). The race condition allows a second caller to attach a new connection or bind an additional Broadcast Isochronous Stream (BIS) while the first call is still in progress. This overwrites internal pointers such as iso_pi(sk)->conn, leaving references and back-pointers from the initial connection stranded without proper cleanup mechanisms. During teardown procedures, only the current active connection is detached, meaning callbacks associated with the old, now-orphaned connection can race against socket release operations. These stale callbacks attempt to access memory that has already been freed by another task, resulting in slab-use-after-free errors as seen in iso_sock_hold and related functions like iso_conn_del and hci_connect_cfm. This flaw effectively undermines the integrity of Bluetooth isochronous data streams and can be exploited for denial-of-service attacks or potentially more sophisticated exploitation vectors depending on memory layout conditions.
To mitigate this vulnerability, the kernel implementation has been updated to enforce strict serialization of connect operations by checking admission criteria under the protection of the socket lock. A new flag marks a connect operation as in progress before publishing any destination information, ensuring that subsequent attempts are rejected with an -EBADFD error code until the initial setup completes or fails. This flag remains set across critical phases including route lookup and connection establishment, preventing concurrent modifications to the socket's state. The logic also addresses specific edge cases such as deferred CIS setups by rejecting existing socket attachments in __iso_chan_add() before taking new references or publishing pointers. Furthermore, separate flags are utilized because the socket lock must be released prior to acquiring the HCI device lock, a constraint that necessitates careful management of state transitions and deferred completion handling through iso_sock_recvmsg().
The remediation strategy also involves rigorous validation during connection addition phases. In CIS setup scenarios, the code now rejects different HCI connections before calling iso_conn_add() and properly drops per-attempt HCI holds to prevent reference count leaks on unowned structures. For Broadcast Isochronous Streams, iso_listen_bis() creates fresh pairs under device locks with proper cleanup on rejection, while iso_conn_ready() ensures that freshly allocated children without active connections do not trigger unreachable socket-side rejections. These changes ensure that every caller preserves the correct connection association and that resource lifecycle management is robust against concurrent access patterns. By aligning state transitions with lock acquisition order and enforcing exclusive access during critical setup windows, the vulnerability eliminates the race condition window entirely. This fix adheres to best practices for concurrency control in kernel networking stacks, preventing memory corruption issues associated with CWE-362: Concurrent Execution using Shared Resource with Improper Synchronization (Race Condition). The mitigation also aligns with ATT&CK techniques related to resource manipulation and denial of service via exploitation of race conditions within system services.