Threat database

CVE-2026-12520 is a memory safety vulnerability in the Zephyr project's HL7800 cellular modem driver affecting versions from 2.4.0 up to but not including 4.4.2. The vulnerability involves improper handling of AT command response parsing, leading to a single-byte out-of-bounds write in most cases and a full stack buffer overflow in the +KCELLMEAS handler. This can cause kernel memory corruption, potentially resulting in a kernel crash or code execution in kernel context. Exploitation requires an attacker to control modem-emitted AT response lines longer than 128 bytes, which is a high complexity attack vector. The vulnerability has a CVSS score of 6.4 (medium severity).

The WNC-M14A2A LTE-M modem driver mishandles unsolicited %NOTIFYEV: events in on_cmd_socknotifyev() (drivers/modem/vendor_standalone/wncm14a2a.c). The response line is linearized into a fixed 40-byte stack buffer via net_buf_linearize(), which caps the copy at 39 bytes and returns out_len <= 39. The two quote-delimiter scanning loops, however, were bounded by len — the full CR/LF-delimited frame length returned by net_buf_findcrlf() — rather than by out_len. When a %NOTIFYEV: line longer than 39 bytes contains no " within the linearized region, the loop indices p1/p2 walk past value[39] and read adjacent stack memory until a stray quote byte is found or the index reaches len. The over-read string is then passed to strncmp()/atoi()/LOG_*, and if a quote byte is found out of bounds the subsequent value[p2] = '\0' performs a single-NUL out-of-bounds stack write at an attacker-influenced offset. The %NOTIFYEV: payload carries network-derived content (LTIME network time, SIB1 base-station system information, CSPS/RRCSTATE), so a rogue cellular base station, a malicious or compromised modem module, or RF manipulation that induces an over-long notify line reaches the defect without any application interaction; the handler runs automatically on the unsolicited event in the modem RX thread. The impact is out-of-bounds stack disclosure (into logs and parsing) and stack corruption that can crash the modem RX thread (denial of service). The write offset is only weakly controlled, so memory-safe code execution is not demonstrated. The fix bounds both scanning loops by out_len, keeping all accesses within the linearized buffer.

The flash_copy() system call is verified by z_vrfy_flash_copy() in drivers/flash/flash_util.c. On builds with CONFIG_USERSPACE enabled, this handler is the kernel-side trust boundary for a user-mode caller. Prior to the fix it validated only the output buffer (K_SYSCALL_MEMORY_WRITE) and passed the two struct device * arguments, src_dev and dst_dev, directly into the implementation without any object validation — unlike every sibling flash syscall, which guards its device pointer with K_SYSCALL_DRIVER_FLASH. A user-mode thread fully controls the values of src_dev/dst_dev and the contents of its own address space. The implementation z_impl_flash_copy() dereferences these pointers and calls through their driver-API function tables (e.g. api->get_parameters(dst_dev), flash_read(src_dev, ...), flash_write(dst_dev, ...)). By supplying a pointer to a forged struct device whose api table contains attacker-chosen function pointers, an unprivileged thread can cause the kernel to call arbitrary code in supervisor mode; passing any arbitrary or invalid address otherwise yields a kernel crash or out-of-bounds read. The result is a local privilege escalation out of the userspace sandbox (with kernel denial-of-service and information disclosure as lesser outcomes). The fix adds K_SYSCALL_DRIVER_FLASH(src_dev, read) and K_SYSCALL_DRIVER_FLASH(dst_dev, write) to z_vrfy_flash_copy(), which verify each device is a registered flash-driver kernel object the calling thread is permitted to use before any dereference, closing the path completely.

Zephyr's 6LoWPAN IP Header Compression (IPHC) uncompression code contains an out-of-bounds read in get_ihpc_inlined_size() (subsys/net/ip/6lo.c). The destination inline size is looked up in da_inline_size_table, which has 13 entries, using an index built from the M, DAC and DAM bits of the received IPHC dispatch word (iphc & NET_6LO_IPHC_DA_MASK, a 4-bit value of 0-15). The reserved combinations 13, 14 and 15 are not bounds-checked and read past the end of the table. The iphc word is taken directly from the received frame, and get_ihpc_inlined_size() is reached on every inbound 6LoWPAN frame via net_6lo_uncompress() from the 802.15.4 receive path (subsys/net/l2/ieee802154/ieee802154_6lo.c and ieee802154_6lo_fragment.c). An unauthenticated attacker on the radio/adjacent link can therefore craft a frame whose destination addressing-mode nibble selects an out-of-range index, with no privileges or user interaction. The out-of-bounds value becomes the computed inline_size, which then drives header reconstruction before the buffer-length check: it is used to dereference *(pkt->buffer->data + sizeof(iphc) + inline_size) and to compute a size_t diff that can underflow, leading to a further out-of-bounds read of the packet buffer and malformed uncompression. The practical impact is a radio-triggerable out-of-bounds read / denial-of-service on the receiver; the leaked byte is not returned to the attacker. The fix rejects any destination index beyond the table, aborting processing of the malformed frame.

The ARM PL011 UART driver in drivers/serial/uart_pl011.c fails to acknowledge receive error interrupts. On the PL011, the framing, parity, break, and overrun error interrupts (PL011_IMSC_ERROR_MASK) are cleared only by writing the interrupt-clear register UARTICR; reading the data register clears the RX interrupt and the per-byte RSR status but not the error interrupt status in MIS. The interrupt service routine pl011_isr() acknowledged only the CTS modem-status interrupt and never wrote icr for the error bits, so an asserted error interrupt remains pending after the ISR returns. When an application enables error-interrupt reporting via the public uart_irq_err_enable() API, an attacker who controls the serial peer can deterministically assert these error bits by injecting line errors on the RX line — a baud/stop-bit mismatch or mid-character break (framing/break error), a flipped parity bit (parity error), or FIFO flooding (overrun error). Because the error interrupt is never cleared, the interrupt line stays asserted and the CPU re-enters pl011_isr() immediately and indefinitely, producing an interrupt-storm livelock from which the core makes no forward progress. The impact is an availability-only denial of service (permanent hang), reachable from an external or removable UART peer. Exploitation is gated by configuration: the error interrupt is off by default and no in-tree subsystem enables it, so only applications that explicitly call uart_irq_err_enable() on a PL011-based, interrupt-driven port are affected. The fix makes pl011_isr() acknowledge the pending error bits via uart->icr, breaking the loop, and additionally clears the latched RSR status in pl011_err_check().

Zephyr's dynamic kernel-object disposal path unref_check() in kernel/userspace/userspace.c frees an object's storage (k_free(dyn->data)) once its reference count reaches zero, after running a per-object-type cleanup. The cleanup switch handled only K_OBJ_MSGQ and K_OBJ_STACK; there was no K_OBJ_TIMER case. A dynamically-allocated, initialized, and armed k_timer keeps its embedded struct _timeout dnode linked in the global timeout queue (_timeout_q), so freeing the timer storage without cancelling the timeout leaves a dangling node in that queue. When the timer next expires, the timeout machinery walks _timeout_q and invokes z_timer_expiration_handler() on the freed node, dereferencing and writing freed (and reusable) kernel heap in kernel/ISR context. This is a deterministic use-after-free that does not depend on SMP: the queued node is simply never unlinked at free time. The disposal is reachable from an unprivileged user thread under CONFIG_USERSPACE + CONFIG_DYNAMIC_OBJECTS: a thread that holds the last permission on such a timer drops it via the k_object_release() syscall (or by exiting, through k_thread_perms_all_clear()), and can arm the timer itself via the k_timer_start() syscall. The free and the expiration handler run at kernel privilege while the actor is a user thread, so the bug is a sandbox-escape memory-corruption primitive usable for privilege escalation. The fix adds k_timer_cleanup() (cancel the timeout and wait for any in-flight handler) and calls it for K_OBJ_TIMER before freeing.

A use-after-free exists in the Zephyr second-generation work queue (kernel/work.c) in the handling of delayable work timeouts. When a delayable work item's timeout has been dequeued and its handler work_timeout() is in flight (blocked acquiring the work-queue spinlock), a concurrent cancellation does not wait for that handler to finish. In unschedule_locked() the pre-fix code called z_abort_timeout(), which for an already-announcing record returns -EINVAL without removing it; cancel_async_locked() then observes the work as idle, so even k_work_cancel_delayable_sync() and k_work_flush_delayable() return without blocking on the in-flight handler. Because those are the APIs the kernel header documents as the safe way to cancel before freeing a k_work_delayable, a caller that frees the object immediately after a successful sync cancel can race the still-pending handler. work_timeout() subsequently dereferences the freed record: it reads to->dticks via z_is_timeout_handler_canceled() and, if the freed slot has been reused so the bail check fails, performs a read-modify-write of wp->flags (K_WORK_DELAYED_BIT) and submits work against a stale dw->queue pointer — a use-after-free read and write. The k_work API is kernel-mode only (no __syscall entry point), so this is a kernel-internal concurrency defect rather than a userspace privilege escalation. Triggering it requires an SMP build and a subsystem that schedules and then frees (or reschedules) a delayable work item in the narrow window while its timeout is announcing; an attacker able to influence the timing of such teardown (for example via connection churn driving subsystem timers) has a plausible but probabilistic path. The impact is kernel memory corruption or crash (denial of service). The fix makes unschedule_locked() wait, by spinning on z_try_abort_timeout() returning -EAGAIN while releasing and re-acquiring the work spinlock, until any in-flight handler completes before returning, and switches work_timeout() to atomic K_WORK_DELAYED_BIT ownership. This closes both the free-then-handler use-after-free and the related reschedule early-fire race.

The user-space system-call verifier z_vrfy_z_log_msg_static_create() in subsys/logging/log_msg.c was a pure pass-through: it forwarded the caller-supplied source, desc, package, and data arguments directly to the kernel-mode implementation z_impl_z_log_msg_static_create() without performing any of the mandatory K_SYSCALL_* checks. Because z_log_msg_static_create() is declared __syscall, under CONFIG_USERSPACE any unprivileged user-mode thread can invoke it directly with fully attacker-controlled arguments. The kernel-mode handler dereferences each of these untrusted values: frontend_runtime_filtering() reads through the source pointer as a struct log_source_dynamic_data, cbprintf_package_copy() reads desc.package_len bytes from the package pointer, and z_log_msg_finalize() performs a memcpy() of desc.data_len bytes from the data pointer. With no verification, a user thread can supply arbitrary kernel addresses and arbitrary lengths, and the kernel will read from them. The impact is a kernel-mode denial of service (the kernel faults dereferencing an attacker-chosen pointer) and, where a log backend output is observable to the attacker, disclosure of arbitrary kernel memory copied into the emitted log message — a confidentiality breach across the user/kernel boundary that the userspace sandbox is meant to enforce. The reads do not corrupt kernel memory, so there is no out-of-bounds write primitive. The fix adds the required validation to the verifier: it bounds desc.package_len against Z_LOG_MSG_MAX_PACKAGE, rejects non-NULL/length mismatches, and applies K_SYSCALL_MEMORY_READ() to package, data, and (when runtime filtering with a frontend is enabled) source, so any out-of-bounds or kernel pointer now raises K_OOPS instead of being honored.

The LoRaWAN Fragmented Data Block Transport service (subsys/lorawan/services/frag_transport.c) does not validate the fragment counter in a received DATA_FRAGMENT command before forwarding it to the configured decoder. In frag_transport_package_callback() the value frag_counter = hdr->frag_index_n & 0x3FFF is taken directly from the downlink payload and passed to the decoder, which derives an array index and flash offset as frag_counter - 1. DataFragment fragments are 1-indexed, so a frag_counter of 0 underflows that arithmetic. With the default Semtech/LoRaMAC-node decoder, this reaches FragDecoder.FragNbMissingIndex[fragCounter - 1] = 0; in FragDecoderProcess(), where fragCounter - 1 evaluates to -1 and writes a uint16_t zero out of bounds, just before the array and into the adjacent MatrixM2B recovery-matrix state of the static decoder object (CWE-787). A companion write derives a wild flash offset, but that path is rejected by the flash_area_write() bounds check. The in-tree low-memory decoder (frag_dec()) is not corrupted: its out-of-range bit-array and flash accesses are caught by sys_bitarray_ and flash_area_ bounds checks. The handler is the registered downlink callback for the fragmentation transport port, reachable whenever an active fragmentation session exists, so the triggering byte is attacker-influenceable LoRaWAN/FUOTA network input. Triggering it requires authenticated downlinks (LoRaWAN MAC session keys or a malicious/compromised network or FUOTA server) and an active fragmentation session. The impact is contained: corruption of decoder state and denial of the firmware-update (FUOTA) session rather than controllable memory corruption or code execution. The fix adds a transport-layer check that rejects frag_counter == 0, closing the defect for both decoder backends.

The Bluetooth host GATT client function parse_read_std_char_desc() in subsys/bluetooth/host/gatt.c parses an ATT Read By Type Response received from a remote GATT server during BT_GATT_DISCOVER_STD_CHAR_DESC discovery. The per-entry stride rsp->len is taken directly from the peer's PDU, and the parse loop both tests its exit condition (length >= rsp->len) and advances (length -= rsp->len, pdu += rsp->len) using that value. The minimum value of rsp->len was never validated before the loop. A malicious or malfunctioning peer can reply with rsp->len = 0. Because length is unsigned and never decreases, the loop condition stays true forever and the read pointer never advances; as long as the body is at least a few bytes with a non-zero handle and a matching descriptor UUID, the host repeatedly re-parses the same bytes and invokes the discovery callback, never terminating. This hangs the Bluetooth host processing thread (CWE-835, loop with unreachable exit condition). The condition is reachable by any connected peer once the local device initiates standard-descriptor-value discovery; GATT discovery does not require bonding or encryption, so an unauthenticated adjacent attacker that the device connects to can trigger it. The impact is denial of service of the Bluetooth subsystem (and likely a watchdog reset on constrained targets); there is no memory disclosure or corruption. The fix adds a rsp->len < sizeof(struct bt_att_data) check before the loop, rejecting under-length responses so the stride is always non-zero and the loop terminates. The sibling parsers parse_include() and parse_characteristic() already validated rsp->len and are unaffected.

The Linkable Loadable Extensions (llext) subsystem mis-handles PLT/RELA relocation entries when linking a relocatable (partially-linked) ELF extension. In llext_link_plt() (subsys/llext/llext_link.c), the relocatable branch (tgt != NULL, the path used for Xtensa relocatable objects) computed the patch address as ext->mem[LLEXT_MEM_TEXT] - text.sh_offset + rela.r_offset + tgt->sh_offset and then performed the relocation write there without validating rela.r_offset. Its sibling shared/dynamic branch already rejected out-of-range offsets via llext_file_offset(). rela.r_offset is read directly from the ELF's RELA table, so a crafted entry with an offset larger than the target section makes the write land arbitrarily far outside the extension's text buffer. The result is an attacker-influenced out-of-bounds write (the location via r_offset, the written value being the resolved symbol address) performed in supervisor context at link time, before any extension code runs. The path is reached from llext_load() whenever an application loads an attacker-influenced ELF extension on Xtensa with writable storage; llext is documented to accept extensions of untrusted origin. Impact is supervisor-context memory corruption (integrity and availability loss, and a sandbox-boundary escape for user-mode extensions). Exploitation is gated by the Xtensa relocatable PLT path and writable storage, and turning the out-of-range write into a useful primitive is non-trivial. The fix adds a bound check rejecting any RELA entry whose r_offset >= tgt->sh_size, mirroring the existing validation in the shared branch.

The userspace syscall verifiers z_vrfy_zsock_sendmsg() and z_vrfy_zsock_recvmsg() in subsys/net/lib/sockets/sockets.c snapshot the caller-supplied struct net_msghdr into a kernel-side copy with k_usermode_from_copy(), but then re-read the still-live user struct for subsequent decisions. The kernel iovec shadow buffer is sized from one read of msg->msg_iovlen, while the population loop is bounded by a second, live read of the same field. Because msg points into ordinary user memory, a cooperating second thread in the same memory domain can inflate msg->msg_iovlen in the window between the sizing read and the loop test (a classic double-fetch / TOCTOU). The population loop then iterates past the number of net_iovec slots actually allocated, writing attacker-influenced iov_base/iov_len values beyond the end of the kernel-heap shadow buffer. The recvmsg verifier has the same defect on both its inbound and result write-back loops. The code is reachable from an unprivileged user thread whenever CONFIG_USERSPACE is enabled and the zsock_sendmsg/zsock_recvmsg syscalls are available. A successful race corrupts kernel-managed heap memory across the user-to-kernel privilege boundary, yielding a local privilege-escalation primitive or, at minimum, a kernel-fault denial of service. The fix copies the header once and derives every size, bound, and gate from the snapshot, copying each iovec entry atomically so its base and length can no longer be raced apart.

The PSA Protected Storage credential backend (subsys/net/lib/tls_credentials/tls_credentials_trusted.c) declared its credential-store mutex as a plain zero-filled static struct k_mutex credential_lock; and never called k_mutex_init() on it. A statically zero-filled k_mutex has an uninitialized wait queue (its dlist head/tail are NULL instead of the self-referential sentinels that k_mutex_init/K_MUTEX_DEFINE install). The uncontended lock path does not touch the wait queue, so the defect is latent and serialized use behaves correctly. When two execution contexts contend on the lock, k_mutex_lock() pends the blocking thread on the wait queue via z_pend_curr(), which calls sys_dlist_append() on the zeroed list and dereferences a NULL tail pointer (tail->next = node), faulting the kernel. The lock is held during TLS handshake credential loading and by all credential add/get/delete operations, so a deployment performing concurrent TLS handshakes (for example a server handling multiple simultaneous connections from a remote peer) or a credential-management operation concurrent with a handshake can trigger the dereference. The impact is a denial of service: a deterministic kernel panic / device reset on the first contention. There is no memory corruption beyond the NULL dereference and no confidentiality or integrity impact; mutual exclusion on the fast path remains correct. Exposure is limited to builds with CONFIG_TLS_CREDENTIALS_BACKEND_PROTECTED_STORAGE enabled (PSA Protected Storage / TF-M platforms); the default volatile RAM backend initializes its lock correctly and is unaffected. The fix initializes the mutex statically with K_MUTEX_DEFINE(credential_lock), providing a valid wait queue so the contended path no longer touches a NULL list.

The Intel ALH digital-audio-interface driver function dai_alh_get_properties() in drivers/dai/intel/alh/alh.c used a caller-supplied int stream_id with no range validation. The value indexes the fixed-size static const uint8_t alh_handshake_map[64] array and scales a FIFO register address, so an out-of-range stream_id produces an out-of-bounds read of one byte at an attacker-chosen signed offset from the array. That byte is written into prop->dma_hs_id and the resulting struct dai_properties is copied back to the caller, leaking it. dai_get_properties_copy() is a Zephyr __syscall, and its verifier z_vrfy_dai_get_properties_copy() (drivers/dai/dai_handlers.c) validates only the device-object permission and the destination buffer, not stream_id. A user-mode thread that has been granted access to the ALH DAI device object can therefore call the syscall with an arbitrary stream_id, crossing the userspace/kernel sandbox boundary. The impact is a one-byte-per-call arbitrary-offset kernel information disclosure (and leakage of a computed kernel address via fifo_address); a stream_id that resolves to an unmapped page faults in kernel context, giving a local denial of service. Exploitation requires CONFIG_USERSPACE and device access, making this a local, moderate-severity issue. The fix rejects negative and too-large stream_id values up front and returns NULL, which the copy wrapper maps to -ENOENT.

The USB device-side CDC NCM class control-to-host handler usbd_cdc_ncm_cth in subsys/usb/device_next/class/usbd_cdc_ncm.c builds a fixed-size response for the GET_NTB_PARAMETERS (28-byte struct ntb_parameters) and GET_NTB_INPUT_SIZE (8-byte struct ntb_input_size) class requests and copies the whole structure into the control DATA IN buffer with net_buf_add_mem(buf, ..., sizeof(...)), ignoring the host-supplied wLength. The control DATA IN buffer is allocated by the USB stack with a capacity of exactly wLength bytes (usbd_ep_ctrl_data_in_alloc -> udc_ctrl_data_alloc -> net_buf_alloc_len(&udc_ep_pool, wLength); no round-up is applied for the IN endpoint). Because net_buf_add_mem/net_buf_simple_add only bounds the copy with an __ASSERT_NO_MSG, which is compiled out in production builds, a host that issues one of these standard CDC NCM control requests with a wLength smaller than the response structure (e.g. wLength = 1) causes the handler to memcpy up to 27 bytes past the end of the allocated pool buffer. The request fields come straight from the USB SETUP packet, so any host (or USB interposer) the Zephyr device enumerates against can trigger the overflow with no authentication once an image built with the device_next USB stack and the CDC NCM class is connected. The out-of-bounds write corrupts adjacent allocations and metadata in the shared udc_ep_pool, primarily causing memory corruption and denial of service of the USB stack; the overflow length is bounded (<= 27 bytes) and the written content is fixed device constants, and the bug reads nothing back so there is no information disclosure. The fix clamps the copy with MIN(sizeof(...), setup->wLength), matching the existing CDC ACM handler.

The USB DFU class implementation in Zephyr's new (experimental) device_next USB device stack contains a NULL pointer dereference in handle_download() (subsys/usb/device_next/class/usbd_dfu.c). The handler computes MIN(setup->wLength, buf->len) and passes buf->data to the image write callback without checking that the buf net_buf pointer is non-NULL. The handler is reached over the USB control endpoint, driven by the USB host. For a DFU_DNLOAD (download) request with no Data OUT stage — notably the zero-length terminating download that the DFU protocol uses to end a firmware transfer — the USB core invokes the class handler with a NULL buffer. After the device has been advanced to the DFU_DNLOAD_IDLE state (by sending one valid download block and a GET_STATUS), a zero-length DFU_DNLOAD reaches handle_download() with buf == NULL, dereferencing it. The result is a NULL+offset read that triggers a fatal CPU fault, i.e. a denial of service (device crash/reset). The attacker is whatever controls the USB host the device is attached to; DFU download support must be enabled with a registered image. There is no memory corruption or information disclosure — impact is limited to availability. The fix adds an explicit if (buf != NULL) guard so the callback receives a zero-length, NULL-data transfer instead of crashing.

The Realtek BEE Bluetooth HCI driver's send callback, bt_hci_bee_send() in drivers/bluetooth/hci/hci_bee.c, violated the bt_hci_driver_api buffer-ownership contract. That contract requires the driver to consume (unref) the transmit net_buf only on success; on an error return the host caller retains ownership and unrefs the buffer itself. The pre-fix code routed all error paths through a shared cleanup label that unconditionally called net_buf_unref(buf) before returning the error code. Because the host TX paths (in subsys/bluetooth/host/hci_core.c) unref the buffer again after send() returns an error, the buffer is freed twice: the driver returns it to its net_buf pool and the host then unrefs the already-freed buffer, corrupting the shared pool / underflowing the reference count (CWE-415). The same error branch additionally dereferenced buf->len inside a LOG_ERR call after the buffer had already been unref'd, a read of freed memory (CWE-416) that is compiled in at the default error log level. The failing edges are reached when the controller's host-to-controller buffer allocation fails or the controller send fails (resource-exhaustion / IO conditions). A remote Bluetooth peer can push the device toward these conditions indirectly by driving heavy host transmit activity, at which point the double-free corrupts the host net_buf pool and most likely crashes the device, with residual potential for further memory corruption. The impact is confined to builds using this specific Realtek BEE HCI driver. The fix returns early from each error path without unreffing and unrefs the buffer only on the success path, restoring the ownership contract and eliminating both the double-free and the use-after-free read.

On the Zephyr ARM port, enabling the hardware FPU (CONFIG_FPU) forces the "Floating point ABI" choice, which defaults to CONFIG_FP_HARDABI. Both FP_HARDABI and FP_SOFTABI permit the compiler to emit hardware FP instructions in any function, even code that never uses floating-point types. However, the callee-saved FP registers (s16-s31 / d8-d15) are only saved and restored across a context switch when CONFIG_FPU_SHARING is enabled (arch/arm/core/cortex_m/swap_helper.S and arch/arm/core/cortex_a_r/swap_helper.S), and prior to this fix selecting an ABI did not enable FPU register sharing, which defaults off. In a build that enables the FPU with the default ABI but leaves CONFIG_FPU_SHARING disabled, the kernel preserves no callee-saved FP register state across thread switches. The documented precondition for this "unshared" mode — that only a single thread ever executes FP instructions — is silently violated because the compiler may generate FP instructions in every thread. Under CONFIG_USERSPACE, where threads are mutually isolated, this becomes an information-disclosure boundary crossing: a victim thread can leave secret-derived values in s16-s31, and a co-resident unprivileged thread can read those registers directly (FP register access is not privilege-gated), recovering data left behind by another thread. Without userspace the same defect causes cross-thread FP state corruption (a correctness fault). The leak is bounded to the 16 callee-saved single-precision registers and is opportunistic, so impact is low. The fix makes FP_HARDABI and FP_SOFTABI select CONFIG_FPU_SHARING and tags every thread with K_FP_REGS at creation, so callee-saved FP state is always preserved across context switches whenever the compiler may emit FP instructions.

The Bluetooth HCI driver for Bouffalo Lab on-chip BLE controllers (BL60x/BL70x/BL61x), bt_bflb_send() in drivers/bluetooth/hci/hci_bflb.c, violates the bt_hci_driver_api.send() buffer-ownership contract. That contract (documented at include/zephyr/drivers/bluetooth.h) requires the buffer reference to be consumed only on success; on error the caller still owns the reference and unrefs it. The driver instead routed all error paths through a shared label that unconditionally called net_buf_unref(buf) before returning the error code, consuming the buffer on failure as well. When send() returns an error, the host TX path (send_buf() in subsys/bluetooth/host/conn.c) unrefs the same buffer again, believing it still owns it. This double-unref over-decrements the net_buf reference count. Because the buffer is a TX fragment whose destroy callback also decrements its still-queued parent buffer, the parent is freed prematurely while reachable on the connection TX queue, producing a use-after-free and corruption of the shared net_buf pool rather than a benign leak. The error conditions are on the host-to-controller transmit path (controller send failure, or an unsupported H:4 packet type), so they are not driven directly by attacker-supplied radio bytes; a remote/adjacent peer can influence them only indirectly, e.g. by inducing controller TX failures under heavy link load. The consequence when reached is BLE-stack denial of service (crash / pool corruption) with possible further memory corruption, bounded to devices using one of these Bouffalo Lab on-chip controllers.

The UpdateHub management subsystem (subsys/mgmt/updatehub/updatehub.c) drives every update operation through a single file-scope ctx structure that holds the CoAP block context, payload buffer, status code, socket, and a one-element poll-fd array fds[1]. Access to ctx was not serialized, and prepare_fds() wrote ctx.fds[ctx.nfds] and incremented ctx.nfds with no bounds check. Two independent paths mutate ctx concurrently: the background autohandler running on the system workqueue, and user-triggered operations reached through the updatehub run shell command, direct API calls, or — since the operations are exposed as syscalls — userspace threads. When a second flow enters prepare_fds() while ctx.nfds is already 1, the write lands one element past the array; by struct layout it overlaps the adjacent ctx.sock/ctx.nfds members. More broadly, the unsynchronized sharing lets two flows interleave connection setup and teardown, double-closing a socket descriptor or scribbling the shared buffers. The result is corruption of the update subsystem's internal state and denial of service of the firmware-update path; the out-of-bounds write is contained within the ctx structure and there is no demonstrated path to memory outside it or to code execution. Triggering requires a local actor able to invoke update operations (or, with CONFIG_USERSPACE, an unprivileged userspace thread) and to win a timing race against the background handler; remote peers cannot control the race timing. The fix serializes the entry points with a mutex and adds a bounds check to prepare_fds().