Bluetooth: btusb: Fix short read errors in btusb_qca_send_vendor_req()
If btusb_qca_send_vendor_req() gets a "short" read from a device, it
will accidentally treat that as a "real" read and populate the returned
value with some unknown and probably totally invalid data.
Fix this logic error up by calling usb_control_msg_recv() which
guarantees a "full" read happens, and then simplify the error checking
for when btusb_qca_send_vendor_req() is called.
Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: btmtk: Fix short read errors in btmtk_usb_reg_read()
If btmtk_usb_reg_read() gets a "short" read from a device, it will
accidentally treat that as a "real" read and populate the returned value
with some unknown and probably totally invalid data.
Fix this logic error up by calling usb_control_msg_recv() which
guarantees a "full" read happens, and then simplify the error checking
for when btmtk_usb_reg_read() is called (it's really just
btmtk_usb_id_get() that calls btmtk_usb_reg_read(), so fix up those
return sites.
Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: btmtk: Fix short read errors in btmtk_usb_uhw_reg_read()
If btmtk_usb_uhw_reg_read() gets a "short" read from a device, it will
accidentally treat that as a "real" read and populate the returned value
with some unknown and probably totally invalid data.
Fix this logic error up by calling usb_control_msg_recv() which
guarantees a "full" read happens, and then simplify the error checking
for when btmtk_usb_uhw_reg_read() is called.
Note, one caller of btmtk_usb_uhw_reg_read() does not check the return
value, but as we pre-initialize the return value as 0, an incorrect read
will not do anything wrong.
Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:23 +0000 (12:59 +0300)]
Bluetooth: hci_sync: remove unnecessary hci_conn_get in create_conn_sync
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: 12917f591cea ("Bluetooth: hci_conn: Fix null ptr deref in hci_abort_conn()") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:22 +0000 (12:59 +0300)]
Bluetooth: hci_sync: fix hci_conn_del() use in hci_le_create_conn_sync
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:21 +0000 (12:59 +0300)]
Bluetooth: hci_sync: hold conn in hci_past_sync() callback
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:20 +0000 (12:59 +0300)]
Bluetooth: hci_sync: hold conn in hci_connect_pa_sync() callback
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:19 +0000 (12:59 +0300)]
Bluetooth: hci_sync: hold conn in hci_connect_big_sync() callback
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:18 +0000 (12:59 +0300)]
Bluetooth: hci_sync: hold conn in hci_connect_acl/le_sync() callbacks
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 25 Jul 2026 09:59:17 +0000 (12:59 +0300)]
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") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Zijun Hu [Sat, 25 Jul 2026 08:54:40 +0000 (01:54 -0700)]
Bluetooth: btintel: Validate length before parsing diagnostics TLV
btintel_diagnostics() accesses tlv->val[0] without first validating
that the diagnostics VSE is long enough to contain that field, so
may cause reading data beyond the received frame.
Fix by validating the length before access.
Fixes: af395330abed ("Bluetooth: btintel: Add Intel devcoredump support") Signed-off-by: Zijun Hu <zijun.hu@oss.qualcomm.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
and also to races in iso_conn_add() vs. iso_conn_free().
Fix by adding spinlock hci_conn::proto_lock and using it to guard
hci_conn::iso_data.
Fixes: dc26097bdb86 ("Bluetooth: ISO: Use kref to track lifetime of iso_conn") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
The extra put for !sk in iso_conn_del() is currently required since
failing iso_chan_add() may leave iso_conn not associated with any sk.
Fix by having iso_pi(sk)->conn own refcount when non-NULL, so
iso_conn_del does not need to put it. Adjust the iso_conn_add()
refcounting so that conn is put if it does not get associated with an
sk.
Fixes: dc26097bdb86 ("Bluetooth: ISO: Use kref to track lifetime of iso_conn") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:32 +0000 (23:20 +0300)]
Bluetooth: ISO: ensure no dangling hcon references in iso_conn
After iso_conn_del(), ISO sockets should not dereference the hcon any
more. Currently, clearing iso_conn::hcon relies on iso_conn_del()
releasing the last reference to the iso_conn.
Simplify this by explicitly clearing conn->hcon in iso_conn_del(), to
avoid more complex reasoning on races about who holds the last
reference.
Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Fix the deadlock by removing the disable from the lock_sock sections.
Move the timer from iso_conn to iso_pinfo to decouple it from iso_conn
which may need to be freed in lock_sock section. Convert some of the
clear_timer to disable_timer.
Fixes: dc26097bdb86 ("Bluetooth: ISO: Use kref to track lifetime of iso_conn") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:30 +0000 (23:20 +0300)]
Bluetooth: ISO: fix leaking sk after socket release
iso_sock_kill() tests !sock_flag(sk, SOCK_ZAPPED) || sk->sk_socket ||
sock_flag(sk, SOCK_DEAD) for early return, but this is always true since
sock_orphan(sk) sets SOCK_DEAD, so the sk reference released by socket
always leaks, iso_sock_destruct is never called.
The socket reference also leaks when __iso_sock_close() does not set
SOCK_ZAPPED, since iso_conn_del() does not call iso_sock_kill() after
zapping.
Fix by replacing SOCK_DEAD by BT_SK_KILLED flag that is not used for
something else, and lock_sock to ensure iso_sock_kill() puts sk only
after socket release only once. Release and iso_conn_del may run
concurrently. Call iso_sock_kill() from iso_conn_del() to clean sk up
after zapping.
Remove call to iso_sock_kill() from iso_sock_close(), as it's generally
no-op there.
Fixes: ccf74f2390d6 ("Bluetooth: Add BTPROTO_ISO socket type") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:29 +0000 (23:20 +0300)]
Bluetooth: ISO: hold sk properly in iso_conn_ready
sk deref in iso_conn_ready must be done either under conn->lock, or
holding a refcount, to avoid concurrent close. conn->sk is currently
accessed without either:
[Task 1] [Task 2]
iso_sock_release
iso_conn_ready
sk = conn->sk
lock_sock(sk)
conn->sk = NULL
lock_sock(sk)
release_sock(sk)
iso_sock_kill(sk)
UAF on sk deref
Fix possible UAF by holding sk refcount in iso_conn_ready(). Also
recheck after lock_sock that the socket is still valid. Adjust locking
so conn->sk is cleared only under lock_sock.
Fixes: 27c24fda62b60 ("Bluetooth: switch to lock_sock in SCO") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:28 +0000 (23:20 +0300)]
Bluetooth: ISO: validate sockaddr_iso first in iso_sock_rebind_bis()
iso_sock_rebind_bis() updates socket iso_pi(sk)->bc_num_bis before
validating the BIS values, so it's possible to end up with bc_num_bis
inconsistent.
Assign to iso_pi(sk)->bc_num_bis only after validation.
Fixes: 80837140c1f2 ("Bluetooth: ISO: Allow binding a PA sync socket") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:27 +0000 (23:20 +0300)]
Bluetooth: ISO: fix timeout vs sync_timeout typo in check_bcast_qos
In iso.c check_bcast_qos(), missing bcast.timeout is not set to its
default value, and appears typoed as bcast.sync_timeout.
Fix the typo.
Fixes: b37cab587aa3 ("Bluetooth: ISO: Don't reject BT_ISO_QOS if parameters are unset") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:26 +0000 (23:20 +0300)]
Bluetooth: ISO: lock sk in iso_connect_ind
Accessing iso_pi(sk)->conn requires lock_sock, which is not taken in the
"ev3" part of iso_connect_ind. It may also be NULL if socket has
transitioned away from the LISTEN/CONNECT states before locking.
Fix by adding lock/release. Recheck hcon is valid after lock acquire
where needed.
Fixes: 168d9bf9c7f0 ("Bluetooth: ISO: Reassemble PA data for bcast sink") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:25 +0000 (23:20 +0300)]
Bluetooth: ISO: lock sk in iso_sock_getname
Accessing iso_pi(sk)->conn requires lock_sock, which is not held here.
Fix by adding the lock/release.
Fixes: 2df108c227b2 ("Bluetooth: ISO: Fix using BT_SK_PA_SYNC to detect BIS sockets") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Jul 2026 20:20:24 +0000 (23:20 +0300)]
Bluetooth: ISO: fix CONNECTED -> CLOSED transition on shutdown/release
Commit d57e506f6a1e ("Bluetooth: ISO: clear iso_data always when detaching conn from hcon")
merged a version of the UAF fix that breaks releasing connected
ISO sockets. Since hci_conn::iso_data is set to NULL, iso_chan_del() won't
be called when the hci_conn disconnects, and the ISO socket does not emit
POLLHUP correctly.
Fix by retaining full hci_conn <-> iso_conn association while in
BT_DISCONNECT state, so that local disconnect via shutdown() follows
similar ISO socket code path as remote disconnect. Use a separate flag
to track whether hci_conn_drop() is needed, instead of setting
iso_conn::hcon = NULL
In iso_sock_ready(), disallow disconnecting socket going BT_CONNECTED,
in case hcon connects while its drop is pending.
Fixes: d57e506f6a1e ("Bluetooth: ISO: clear iso_data always when detaching conn from hcon") Fixes: fbdc4bc47268 ("Bluetooth: ISO: Use defer setup to separate PA sync and BIG sync") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Chengfeng Ye [Thu, 23 Jul 2026 15:34:40 +0000 (23:34 +0800)]
Bluetooth: hci_sync: Fix advertising data UAFs
hci_find_adv_instance() returns an adv_info pointer that is valid only
while hdev->lock is held. The advertising command-sync paths perform
instance lookups without that lock and, in some cases, retain the pointer
while waiting for a controller response.
An advertising termination event can therefore interleave as follows:
BUG: KASAN: slab-use-after-free in hci_set_ext_scan_rsp_data_sync+0x2e1/0x300
Write of size 1 at addr ffff88810a45d21d by task kworker/u17:0/88
Workqueue: hci0 hci_cmd_sync_work
Call Trace:
hci_set_ext_scan_rsp_data_sync+0x2e1/0x300
hci_schedule_adv_instance_sync+0x390/0x4c0
hci_cmd_sync_work+0x173/0x300
Allocated by task 87:
hci_add_adv_instance+0x538/0xac0
add_advertising+0x885/0x1160
Freed by task 89:
kfree+0x131/0x3c0
hci_remove_adv_instance+0x1d8/0x3b0
hci_le_ext_adv_term_evt+0x17b/0x730
Protect the instance lookup and payload construction in the extended
advertising, scan response, and periodic advertising data paths. Snapshot
the advertising parameters under hdev->lock, but release the lock before
waiting for the controller.
Clear advertising-data dirty bits before issuing their commands and
restore them after a failure using a fresh lookup. Likewise, update the
reported transmit power through a fresh lookup after the parameter command
completes. No adv_info pointer then survives an HCI command wait.
Fixes: cba6b758711c ("Bluetooth: hci_sync: Make use of hci_cmd_sync_queue set 2") Cc: stable@vger.kernel.org Suggested-by: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Signed-off-by: Chengfeng Ye <nicoyip.dev@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Jiale Yao [Wed, 22 Jul 2026 09:26:14 +0000 (17:26 +0800)]
Bluetooth: RFCOMM: validate skb length in rfcomm_recv_frame
rfcomm_recv_frame() casts skb->data to struct rfcomm_hdr and dereferences
hdr->addr and hdr->ctrl without validating skb->len first. A truncated
frame with skb->len less than the minimum header size causes an
out-of-bounds read of uninitialized memory. Additionally, a zero-length
frame causes skb->len-- to underflow to UINT_MAX, making
skb_tail_pointer() read far past the buffer.
Commit 23882b828c3c ("Bluetooth: RFCOMM: validate skb length in MCC
handlers") fixed the same class of missing-length-check bugs in the MCC
sub-handlers, but the top-level rfcomm_recv_frame() was left unfixed.
KMSAN reports:
BUG: KMSAN: uninit-value in rfcomm_run
...
Uninit was created at:
__alloc_skb+0x474/0xb60
vhci_write+0xe9/0x870
Fix this by rejecting frames smaller than sizeof(struct rfcomm_hdr) + 1
(the minimum frame must have a 3-byte header and a 1-byte FCS).
Signed-off-by: Jiale Yao <yaojiale02@163.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Jiale Yao [Thu, 23 Jul 2026 06:48:45 +0000 (14:48 +0800)]
Bluetooth: L2CAP: fix UAF in l2cap_le_connect_rsp
l2cap_le_connect_rsp() obtains a channel via
__l2cap_get_chan_by_ident() but neither holds a reference nor uses
l2cap_chan_hold_unless_zero() before locking and operating on it.
A concurrent l2cap_chan_del() triggered by a remote disconnect can
free the channel between the lookup and l2cap_chan_lock(), causing
a use-after-free.
The BR/EDR counterpart l2cap_connect_rsp() and the sibling handler
l2cap_le_command_rej() already use l2cap_chan_hold_unless_zero()
to safely hold a reference, but l2cap_le_connect_rsp() was left
unprotected.
Fix by adding l2cap_chan_hold_unless_zero() after the ident lookup
and l2cap_chan_put() on the exit path, consistent with other L2CAP
response handlers.
Fixes: f1496dee9cbd ("Bluetooth: Add initial code for LE L2CAP Connect Request") Assisted-by: Claude:deepseek-v4-pro Signed-off-by: Jiale Yao <yaojiale02@163.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
When hidp_get_raw_report() waits for a numbered report,
hidp_process_data() compares the expected report number with skb->data[0].
A connected HIDP peer can reply with only a DATA transaction header,
leaving the skb empty after the header is removed.
KMSAN reports an uninitialized-value use in hidp_session_run(), with the
value originating in __alloc_skb() through vhci_write(). The transaction
header checks remove the empty-frame reports, but this report remains until
the payload check is added.
The comparison can also consume a peer-controlled byte beyond the declared
L2CAP PDU. A DATA | FEATURE response followed by an extra 0x01 byte made
the current code accept that byte as report ID 1 and complete
HIDIOCGFEATURE with a zero-byte result. With this change the malformed
response is rejected with -EIO, while a subsequent valid response still
succeeds.
Require a payload byte before comparing a numbered report ID. Unnumbered
reports continue to accept an empty payload.
Fixes: 0ff1731a1ae5 ("HID: bt: Add support for hidraw HIDIOCGFEATURE and HIDIOCSFEATURE") Cc: stable@vger.kernel.org Signed-off-by: Sangho Lee <kudo3228@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Sangho Lee [Thu, 23 Jul 2026 03:28:06 +0000 (12:28 +0900)]
Bluetooth: HIDP: reject frames without a transaction header
hidp_recv_ctrl_frame() and hidp_recv_intr_frame() read skb->data[0]
before checking that the L2CAP SDU contains a transaction header. A
connected HIDP peer can send an empty basic-mode SDU and make both paths
use an uninitialized byte from skb tailroom.
KMSAN reports the use in hidp_session_run(), with the uninitialized value
originating in __alloc_skb() through vhci_write(). The control path
produces two reports and the interrupt path produces one.
The byte can also be controlled by a malformed lower-layer packet. If an
HCI ACL packet contains an L2CAP PDU with a declared zero-length payload
followed by an extra 0x15 byte, l2cap_recv_acldata() reduces skb->len to
the declared PDU length before dispatch. The current HIDP path nevertheless
consumes the extra byte as HIDP_TRANS_HID_CONTROL |
HIDP_CTRL_VIRTUAL_CABLE_UNPLUG and terminates the HIDP session. With this
change, the same packet is discarded and a subsequent feature report
request succeeds.
Pull the transaction header with skb_pull_data() and discard frames that
do not contain it.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable@vger.kernel.org Signed-off-by: Sangho Lee <kudo3228@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Zihan Xi [Thu, 23 Jul 2026 16:43:46 +0000 (00:43 +0800)]
Bluetooth: mgmt: fix pending command UAF in EIR updates
MGMT_OP_SET_LOCAL_NAME is handled asynchronously on powered controllers
and can run set_name_sync(). When the controller is BR/EDR capable,
set_name_sync() updates the local name and then rebuilds EIR data through
eir_create(). The EIR builder walks hdev->uuids, but the UUID list can
be changed and entries can be freed by MGMT_OP_ADD_UUID and
MGMT_OP_REMOVE_UUID.
pending_eir_or_class() is meant to serialize management commands that
can change EIR or the class of device, but it did not include
MGMT_OP_SET_LOCAL_NAME. In addition, it walked hdev->mgmt_pending
without hdev->mgmt_pending_lock even though pending commands are added
and removed under that mutex. A racing command completion can therefore
remove and free a pending command while pending_eir_or_class() is still
inspecting it, leading to a use-after-free in the pending-command list or
allowing a local name update to rebuild EIR while UUID entries are being
removed.
Take hdev->mgmt_pending_lock while scanning hdev->mgmt_pending and treat
MGMT_OP_SET_LOCAL_NAME as an EIR/class-affecting pending command on the
powered asynchronous path. Check for a conflicting pending command before
copying the new short name so a rejected SET_LOCAL_NAME request does not
modify hdev->short_name.
Fixes: 6fe26f694c82 ("Bluetooth: MGMT: Protect mgmt_pending list with its own lock") Cc: stable@vger.kernel.org Reported-by: Vega <vega@nebusec.ai> Assisted-by: Codex:gpt-5.4 Signed-off-by: Zihan Xi <zihanx@nebusec.ai> Signed-off-by: Ren Wei <enjou1224z@gmail.com> Reported-by: Vega <vega@nebusec.ai> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Zihan Xi [Tue, 21 Jul 2026 14:36:07 +0000 (22:36 +0800)]
Bluetooth: mgmt: fix UAF in pair command cancellation
The pairing completion and authentication failure callbacks look up the
pending MGMT_OP_PAIR_DEVICE command by walking hdev->mgmt_pending. The
lookup returned a command that was still linked on the shared pending list,
without keeping mgmt_pending_lock held for the later dereference and
removal.
A concurrent MGMT_OP_CANCEL_PAIR_DEVICE request can remove and free the
same pending command before the callback uses it. The reverse race is also
possible when cancel_pair_device() gets a command from pending_find() and a
callback removes it before the cancel path dereferences it. This can lead
to a use-after-free and a second list_del().
Make the pairing lookup helpers transfer ownership of the pending command
by removing it from hdev->mgmt_pending while holding mgmt_pending_lock.
The callbacks and cancel path then complete the command and free it
directly, so racing paths cannot find or free the same command again. Take
a temporary hci_conn reference in cancel_pair_device() because the command
completion drops the reference stored in the pending command.
Fixes: e9a416b5ce0c ("Bluetooth: Add mgmt_pair_device command") Cc: stable@vger.kernel.org Reported-by: Vega <vega@nebusec.ai> Assisted-by: Codex:gpt-5.4 Signed-off-by: Zihan Xi <xizh2024@lzu.edu.cn> Reviewed-by: Ren Wei <enjou1224z@gmail.com> Reported-by: Vega <vega@nebusec.ai> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Mon, 20 Jul 2026 14:53:33 +0000 (17:53 +0300)]
Bluetooth: ISO: clear iso_data always when detaching conn from hcon
When setting conn->hcon = NULL, also conn->hcon->iso_data = NULL is
necessary, otherwise later iso_conn_free() will UAF.
Fix clearing of iso_data in iso_sock_disconn()
Fixes KASAN: slab-use-after-free in iso_conn_hold_unless_zero on
iso_sock_release() followed by hci_abort_conn_sync().
Fixes: fbdc4bc47268 ("Bluetooth: ISO: Use defer setup to separate PA sync and BIG sync") Signed-off-by: Pauli Virtanen <pav@iki.fi> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Dawei Feng [Sun, 7 Jun 2026 14:57:06 +0000 (22:57 +0800)]
e1000: fix memory leak in e1000_probe()
In the e1000_probe() path, e1000_sw_init() allocates adapter->tx_ring and
adapter->rx_ring. If the subsequent CE4100-specific MDIO BAR mapping
fails, the error handling jumps past the ring cleanup code, leaking both
allocations.
Fix this leak by moving the err_mdio_ioremap label above the ring
deallocation logic. This guarantees the proper release of these resources
and prevents the memory leak.
The bug was first flagged by an experimental analysis tool we are
developing for kernel memory-management bugs while analyzing
v6.13-rc1. The tool is still under development and is not yet publicly
available. Manual inspection confirms that the bug is still
present in v7.1-rc6.
An x86_64 allyesconfig build showed no new warnings. As we do not have a
CE4100 reference platform to test with, no runtime testing was able to
be performed.
Fixes: 5377a4160bb65 ("e1000: Add support for the CE4100 reference platform") Cc: stable@vger.kernel.org Signed-off-by: Zilin Guan <zilin@seu.edu.cn> Signed-off-by: Dawei Feng <dawei.feng@seu.edu.cn> Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Matt Vollrath [Fri, 17 Apr 2026 03:34:52 +0000 (23:34 -0400)]
igbvf: Fix leak in TX DMA error cleanup
If an error is encountered while mapping TX buffers, the driver should
unmap any buffers already mapped for that skb.
Because count is incremented before each frag mapping, it will always
match the correct number of unmappings needed when dma_error is reached.
Decrementing count before the while loop in dma_error causes an
off-by-one error. If any mapping was successful before an unsuccessful
mapping, exactly one DMA mapping (the head) would leak.
This bug was introduced by a 2010 fix for an endless loop in dma_error.
All other affected drivers have already been fixed.
Fixes: c1fa347f20f1 ("e1000/e1000e/igb/igbvf/ixgb/ixgbe: Fix tests of unsigned in *_tx_map()") Cc: stable@vger.kernel.org Assisted-by: Claude:claude-4-7-opus Signed-off-by: Matt Vollrath <tactii@gmail.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
David Carlier [Sun, 12 Jul 2026 13:22:42 +0000 (14:22 +0100)]
igc: remove napi_synchronize() in igc_down()
When an AF_XDP zero-copy application is killed abruptly, the XSK pool is
torn down but NAPI keeps polling. igc_clean_rx_irq_zc() then returns the
full budget on every poll, so napi_complete_done() never clears
NAPI_STATE_SCHED.
igc_down() calls napi_synchronize() before napi_disable(), so it spins
forever waiting for that bit and the interface never goes down. Drop the
napi_synchronize() and let napi_disable() do the job -- it sets
NAPI_STATE_DISABLE, which forces the stuck poll to complete. Reorder it
ahead of igc_set_queue_napi() so the NAPI mapping is cleared only after
polling has stopped, matching the recent igb fix b1e067240379.
Fixes: fc9df2a0b520 ("igc: Enable RX via AF_XDP zero-copy") Suggested-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> Cc: stable@vger.kernel.org Signed-off-by: David Carlier <devnexen@gmail.com> Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com> Tested-by: Moriya Kadosh <moriyax.kadosh@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Przemyslaw Korba [Wed, 20 May 2026 11:50:06 +0000 (13:50 +0200)]
ice: suppress DPLL errors during reset recovery
During reset recovery, the admin queue returns EBUSY which is expected
behavior. However, the DPLL subsystem was logging these as errors and
incrementing the error counter, potentially leading to unnecessary
warnings and even disabling the DPLL periodic worker if the threshold
was reached.
Suppress error logging and error counter increments when the admin
queue returns EBUSY, as this is expected during reset recovery and
not a real failure condition.
test case:
- ethtool --reset eth3 irq-shared dma-shared filter-shared offload-shared
mac-shared phy-shared ram-shared
- observe if dmesg EBUSY errors are gone
Fixes: d7999f5ea64b ("ice: implement dpll interface to control cgu") Signed-off-by: Przemyslaw Korba <przemyslaw.korba@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel) Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Dawei Feng [Tue, 16 Jun 2026 15:57:42 +0000 (23:57 +0800)]
ice: fix memory leak in ice_lbtest_prepare_rings()
ice_lbtest_prepare_rings() frees Rx rings only when
ice_vsi_start_all_rx_rings() fails. If ice_vsi_setup_rx_rings() fails
after allocating some descriptors, or if ice_vsi_cfg_lan() fails after
the Rx rings were prepared, the function reaches the Tx cleanup path
without releasing the initialized Rx resources.
Fix this by adding separate unwind paths for Rx setup failure and LAN
configuration failure. The Rx setup failure path releases the partially
prepared Rx rings before freeing Tx rings, while later failures first
undo the LAN Tx configuration and then release the Rx rings in reverse
setup order.
The bug was first flagged by an experimental analysis tool we are
developing for kernel memory-management bugs while analyzing
v6.13-rc1. The tool is still under development and is not yet publicly
available. Manual inspection confirms that the bug is still
present in v7.1-rc7.
An x86_64 allyesconfig build showed no new warnings. As we do not have an
Intel E800 Series adapter available to run the ethtool offline loopback
selftest, no runtime testing was able to be performed.
Fixes: 0e674aeb0b77 ("ice: Add handler for ethtool selftest") Cc: stable@vger.kernel.org Signed-off-by: Dawei Feng <dawei.feng@seu.edu.cn> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Dawid Osuchowski [Thu, 14 May 2026 16:35:55 +0000 (18:35 +0200)]
ice: fix VF interrupts cleanup
When a virtual function sends an IRQ map command, the PF will set up
interrupts according to that request. However, because these interrupts are
never reset, the next time Virtual Function initializes, the interrupts are
still enabled for a given VF, which leads to performance degradation in
certain cases due to interrupts being unexpectedly enabled and thus causing
interrupt floods.
Cc: stable@vger.kernel.org Fixes: 1071a8358a28 ("ice: Implement virtchnl commands for AVF support") Suggested-by: Vladimir Medvedkin <vladimir.medvedkin@intel.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Dawid Osuchowski <dawid.osuchowski@linux.intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Patryk Holda <patryk.holda@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Aaron Ma [Wed, 29 Apr 2026 03:48:49 +0000 (11:48 +0800)]
ice: wait for reset completion in ice_resume()
ice_resume() schedules an asynchronous PF reset and returns
immediately. The reset runs later in ice_service_task(). If
userspace tries to bring up the net device before the reset
finishes, ice_open() fails with -EBUSY:
ice_resume()
ice_schedule_reset() # sets ICE_PFR_REQ, returns
...
ice_open()
ice_is_reset_in_progress() # ICE_PFR_REQ still set, -EBUSY
...
ice_service_task()
ice_do_reset()
ice_rebuild() # clears ICE_PFR_REQ, too late
Reproduced on E800 series NICs during suspend/resume with irdma
enabled, where the aux device probe widens the race window.
ice 0000:81:00.0: can't open net device while reset is in progress
Add a best-effort wait (10s timeout, matching ice_devlink_info_get())
for the reset to complete before returning from ice_resume(). In
practice the reset completes in ~300ms.
Fixes: 769c500dcc1e ("ice: Add advanced power mgmt for WoL") Cc: stable@vger.kernel.org Reviewed-by: Kohei Enju <kohei@enjuk.jp> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com> Signed-off-by: Aaron Ma <aaron.ma@canonical.com> Tested-by: Alexander Nowlin <alexander.nowlin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
idpf: Fix mailbox IRQ name leak on request failure
idpf_mb_intr_req_irq() allocates the mailbox IRQ name before calling
request_irq(). On success, the name is released later through
kfree(free_irq()), but request_irq() failure returns without freeing it.
Free the allocated name on the request_irq() failure path.
Fixes: 4930fbf419a7 ("idpf: add core init and interrupt request") Signed-off-by: Yuho Choi <dbgh9129@gmail.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Samuel Salin <Samuel.salin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Joshua Hay [Tue, 30 Jun 2026 23:56:19 +0000 (16:56 -0700)]
idpf: adjust TxQ ring count minimum
Set the TxQ ring count minimum to 128 descriptors. Any lower than this,
and the queue will stall and trigger Tx timeouts in flow based
scheduling mode. This is because next_to_clean might never be updated.
In flow based scheduling mode, next_to_clean is only updated after a
descriptor completion is processed, i.e. after the RE bit is set in the
last descriptor of a Tx packet. This will never happen with a ring size
of 64 and an IDPF_TX_SPLITQ_RE_MIN_GAP of 64. No matter what the value
of last_re is initialized/set to, the calculated gap will be at most 63
and never trigger the RE bit.
Even a ring size of 96 does not solve this. Because of how infrequent
next_to_clean is updated and how small the ring is, IDPF_DESC_UNUSED
will be much smaller on average. This increases the chance the queue
will be stopped because a multi-descriptor packet, e.g. a large LSO
packet, does not see enough resources on the ring. In this case, the
queue will trigger the stop logic. The queue permanently stalls because
there is no chance for a descriptor completion to update next_to_clean
since it is dependent on a packet being sent.
Fixes: 5f417d551324 ("idpf: replace flow scheduling buffer ring with buffer pool") Signed-off-by: Joshua Hay <joshua.a.hay@intel.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Samuel Salin <Samuel.salin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
idpf: bound interrupt-vector register fill to the allocated array
idpf_get_reg_intr_vecs() fills the caller-allocated reg_vals[] array from
the VIRTCHNL2_OP_ALLOC_VECTORS reply in adapter->req_vec_chunks, bounding
its inner loop only by the per-chunk num_vectors. The array is sized
separately: idpf_intr_reg_init() allocates
kzalloc_objs(struct idpf_vec_regs, total_vecs) from
caps.num_allocated_vectors and only checks the returned count after the
fill. The sum of per-chunk num_vectors is never reconciled against
total_vecs, so a reply with a small num_allocated_vectors but chunks
summing higher writes past the end of reg_vals[].
Impact: a control plane (a PF or hypervisor device model) that returns a
VIRTCHNL2_OP_ALLOC_VECTORS reply whose per-chunk num_vectors sum exceeds
num_allocated_vectors writes struct idpf_vec_regs entries past the end of
the reg_vals kmalloc allocation (KASAN slab-out-of-bounds write).
Bound the fill loop to the array capacity passed in by the callers,
mirroring the sibling idpf_vport_get_q_reg(). The existing
num_regs < num_vecs check then rejects an undersized reply without the
out-of-bounds write happening first.
Fixes: d4d558718266 ("idpf: initialize interrupts and enable vport") Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Samuel Salin <Samuel.salin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Stanislaw Pal [Mon, 27 Jul 2026 16:32:16 +0000 (18:32 +0200)]
spi: spi-qpic-snand: write the feature value before executing SET_FEATURE
qcom_spi_send_cmdaddr() programs NAND_FLASH_CMD/NAND_EXEC_CMD and submits
the descriptors, which makes the controller execute the command
immediately. For SPINAND_SET_FEATURE the value to be written is only
placed into NAND_FLASH_FEATURES afterwards, by qcom_spi_io_op(), in a
second submission - so the chip is programmed with whatever that register
happened to hold from a previous operation, and the intended value is only
applied by the *next* SET_FEATURE.
Measured on a TP-Link Archer AX55 v1 (IPQ5018, ESMT F50L1G41LB): writing
0x40 to the configuration register (0xb0) leaves the chip at 0x00, and the
subsequent write of 0x00 leaves it at 0x40 - every write lands one
operation late.
This stayed unnoticed until v6.18 added SPI-NAND OTP support together
with OTP entries for ESMT chips. spinand_otp_rw() enables OTP mode,
reads, and disables it again, and mtd_otp_nvmem_add() does this during
MTD registration. With the off-by-one, the "disable" write actually
applies the previously requested value, so CFG_OTP_ENABLE ends up set:
the chip stays in OTP mode, every subsequent array read returns the OTP
area instead of the array (UBI reports an empty device) and all writes
fail with -EIO because the OTP area is write protected. On this board
that makes the whole flash unusable and the device unbootable.
Write the feature value into NAND_FLASH_FEATURES as part of the same
transaction, before NAND_EXEC_CMD. While at it, copy only the bytes the
operation actually carries - the previous code dereferenced a 4-byte
pointer on a one-byte buffer (spinand->scratchbuf).
With this patch the flash contents read back bit-identical to a
known-good dump of the same board taken under the vendor firmware
(md5-verified across partitions), and writes work.
Fixes: 7304d1909080 ("spi: spi-qpic: add driver for QCOM SPI NAND flash Interface") Cc: stable@vger.kernel.org Signed-off-by: Stanislaw Pal <kuncy7@gmail.com> Reviewed-by: Md Sadre Alam <md.alam@oss.qualcomm.com> Link: https://patch.msgid.link/20260727163216.109938-1-kuncy7@gmail.com Signed-off-by: Mark Brown <broonie@kernel.org>
Richard Zhu [Tue, 28 Jul 2026 03:51:59 +0000 (11:51 +0800)]
PCI: imx6: Fix i.MX6Q/DL boot hang caused by improper PHY power sequencing
commit 610fa91d9863 ("PCI: imx6: Assert PERST# before enabling regulators")
introduced a boot hang on i.MX6Q/DL variants by reordering
imx_pcie_host_init() to call imx6q_pcie_enable_ref_clk() (which powered up
the PHY) before imx6q_pcie_core_reset() (which powered it back down).
1. imx_pcie_assert_core_reset() - power down PHY (set TEST_PD), set
REF_CLK_EN
2. imx_pcie_clk_enable() - power up PHY (clear TEST_PD), set REF_CLK_EN
3. Link training starts with PHY powered up (TEST_PD cleared)
4. Link training succeeds
1. imx_pcie_clk_enable() - power up PHY (clear TEST_PD), set REF_CLK_EN
2. imx_pcie_assert_core_reset() - power down PHY (set TEST_PD), set
REF_CLK_EN
3. imx_pcie_deassert_core_reset() - does nothing
4. Link training starts with PHY powered down (TEST_PD set)
5. Link training fails and boot hangs when PHY register accesses hang
To fix this:
- Remove TEST_PD PHY power control from imx6q_pcie_enable_ref_clk()
- Remove REF_CLK_EN control from imx6q_pcie_core_reset()
- Add TEST_PD PHY power control to imx6qp_pcie_core_reset(), which
previously relied on imx6q_pcie_enable_ref_clk() to power up the PHY by
clearing TEST_PD
- Clear TEST_PD to power on PHY in imx_pcie_deassert_core_reset()
These changes together ensure the correct sequence:
1. REF_CLK_EN set in clk_enable() (TEST_PD untouched)
2. TEST_PD set in assert_core_reset() (PHY power off)
3. TEST_PD cleared in deassert_core_reset() (PHY power on)
4. Link training starts with proper PHY state
The i.MX6Q/DL PCIe PHY requires approximately 120us between TEST_PD
de-assertion and link training start. Add usleep_range(200, 500) in
imx6q_pcie_core_reset() after clearing TEST_PD to satisfy this requirement.
Add explicit imx_pcie_assert_core_reset() calls in error paths and
host_exit() to ensure no power leak.
Fixes: 610fa91d9863 ("PCI: imx6: Assert PERST# before enabling regulators") Reported-by: Leonardo Costa <leoreis.costa@gmail.com> Closes: https://lore.kernel.org/lkml/20260629143439.361560-1-leoreis.costa@gmail.com/ Signed-off-by: Richard Zhu <hongxing.zhu@nxp.com> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Tested-by: Leonardo Costa <leonardo.costa@toradex.com> Reviewed-by: Frank Li <Frank.Li@nxp.com> Link: https://patch.msgid.link/20260728035159.2702021-1-hongxing.zhu@oss.nxp.com
Robert Abrahamse [Tue, 28 Jul 2026 14:03:14 +0000 (16:03 +0200)]
ALSA: usb-audio: Add quirk for Corsair Virtuoso (later revision)
Add USB mixer mapping quirk for later revisions of the Corsair Virtuoso
headset with USB IDs 0x1b1c:0x0a43 (wired) and 0x1b1c:0x0a44
(wireless). These devices exhibit the same mixer label collision as
earlier Virtuoso variants: all controls are labelled "Headset", causing
applications like PulseAudio to move the sidetone control instead of
the main playback volume.
snd_pcm_drain() on a linked stream parks an on-stack wait entry on the
drained peer's runtime->sleep, and after schedule_timeout() removes it
only if that peer is still found in the caller's group. If group
membership changes during the wait and the sleep ends by signal or
timeout (so autoremove_wake_function() does not run), finish_wait() is
skipped and snd_pcm_drain() returns with the entry still queued on that
stream's sleep list; a later wake_up() then walks a freed stack frame.
This is reachable by unlinking either the drained or the draining stream.
Unlike the close path (snd_pcm_drop() -> snd_pcm_post_stop()),
snd_pcm_unlink() never wakes the sleep queues. Wake every group member
under the group lock before the membership change, so a linked drainer is
released and drops its entry while the streams are still grouped.
The window was opened when snd_pcm_link_rwsem stopped being held across
the wait and the removal became conditional on group membership (see
Fixes). The later switch to finish_wait() kept that conditional removal,
so the signal/timeout case remained.
Jinu Kim [Tue, 21 Jul 2026 10:35:12 +0000 (19:35 +0900)]
KVM: x86/mmu: Check all address spaces before skipping unsync
mmu_try_to_unsync_pages() skips the shadow-page lookup when the
supplied memslot allows a hugepage, because a shadow page would disallow
hugepages. But hugepage metadata is per-address-space while shadow pages
are shared across all address spaces. With SMM, the other address space
can therefore have a shadow page even when the supplied memslot allows a
hugepage.
Check the corresponding memslot in the other address space before
taking the fast path. Skip the shadow-page lookup only when all address
spaces allow a hugepage.
Fixes: b3ae3ceb5569 ("KVM: x86/mmu: KVM: x86/mmu: Skip unsync when large pages are allowed") Assisted-by: Codex:GPT-5 Signed-off-by: Jinu Kim <kimjw04271234@gmail.com>
[invert direction of the conditional. - Paolo]
Message-ID: <20260721103512.2136240-3-kimjw04271234@gmail.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Jinu Kim [Tue, 21 Jul 2026 10:35:11 +0000 (19:35 +0900)]
KVM: x86/mmu: Check write tracking in all address spaces
kvm_gfn_is_write_tracked() checks only the supplied memslot, but page
tracking is per-address-space and shadow pages are shared across all
address spaces. With SMM, a GFN can therefore be write-tracked in one
address space and appear untracked through the other.
Check the supplied slot first, then the slot for the other address space.
This ensures all callers honor write tracking regardless of the active
address space. In particular, it prevents mmu_try_to_unsync_pages() from
marking an upper-level shadow page unsync and eventually triggering the
BUG in pte_list_remove().
Fixes: 699023e23965 ("KVM: x86: add SMM to the MMU role, support SMRAM address space") Assisted-by: Codex:GPT-5 Signed-off-by: Jinu Kim <kimjw04271234@gmail.com>
Message-ID: <20260721103512.2136240-2-kimjw04271234@gmail.com>
[invert direction of the conditional. - Paolo] Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
hwmon: (pmbus) Fix return value from pmbus_update_byte_data()
pmbus_update_byte_data() is supposed to return a negative error code or 0.
However, if no change is made to the register, it actually returns the
register value. This can result in problems if the calling code explicitly
expects to see an error code or 0.
Fix it to return 0 on success or the error code as expected.
Fixes: 11c119986f270 ("hwmon: (pmbus) add helpers for byte write and read modify write") Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Cancel (and flush) the I/O APIC's delayed EOI handling work during the
"pre VM destroy" phase, before vCPUs are destroyed, as processing the EOI
broadcast will inject another IRQ if the line is asserted, i.e. will try
to deliver an IRQ to the target vCPU(s). Canceling the work after vCPUs
are destroyed leads to UAF if the delayed work is processed after vCPUs are
destroyed.
BUG: KASAN: slab-use-after-free in __kvm_irq_delivery_to_apic_fast+0x9bf/0xa20 arch/x86/kvm/lapic.c:1250
Read of size 8 at addr ffff8880499abea0 by task kworker/1:2/1218
Note, the VM is unreachable once kvm_destroy_vm() starts, and scheduling
new work via kvm_ioapic_send_eoi() can only be done via KVM_RUN, i.e.
requires a live vCPU.
Alternatively, KVM could simply destroy the I/O APIC during the "pre" phase
of VM destruction, but that gets more than a bit sketchy as KVM expects the
I/O APIC to exist if ioapic_in_kernel() is true, and nested virtualization
in particular has a bad habit of touching VM-scope state during vCPU
destruction. E.g. attempting to free the PIC during the pre phase would
lead to a NULL pointer dereference in kvm_cpu_has_extint(), and it's not
hard to imagine the I/O APIC having a similar flaw.
Fixes: 17bcd7144263 ("KVM: x86: Free vCPUs before freeing VM state") Reported-by: <zdi-disclosures@trendmicro.com> Reported-by: Zhong Wang <wangzhong.c0ss4ck@bytedance.com> Reported-by: Xuanqing Shi <shixuanqing.11@bytedance.com> Cc: stable@vger.kernel.org Signed-off-by: Weiming Shi <bestswngs@gmail.com> Co-developed-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260727171718.543491-1-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Tue, 21 Jul 2026 16:31:49 +0000 (18:31 +0200)]
KVM: VMX: add memory clobber to asm for VMX instructions
VMCLEAR/VMREAD/VMWRITE/VMPTRLD access the internal VMCS cache, which
is not visible to the compiler; without a memory clobber, the compiler
can reorder them in troublesome ways because "asm volatile" and "asm goto"
only protect against removal of the asm. For example, placing a VMWRITE
before the corresponding VMCS pointer is loaded can lead to corruption.
While none of this has been observed, it is better to prevent than cure.
Likewise, INVEPT and INVVPID access the TLB and, even though in their
case the effect is only visible to the next VMLAUNCH/VMRESUME, it is
technically correct to add the clobber there too. So avoid any urge to
special case them, and simply hardcode "memory" into the clobber list
of vmx_asm1() and vmx_asm2(). __vmcs_readl() open-codes its own asm,
so add the clobber there as well.
Paolo Bonzini [Tue, 28 Jul 2026 15:43:23 +0000 (17:43 +0200)]
Merge tag 'kvmarm-fixes-7.2-3' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD
KVM/arm64 fixes for 7.2, take #3
- Fix a tiny buglet when propagating the deactivation of an interrupt
from a nested guest, which happened to trigger a gold plated CPU bug
on a particular implementation
- Fix a race between LPI unmapping and mapping, resulting in leaked
LPIs
- Make LPI mapping more robust on memory allocation failure
- Fix the handling of the EL2 tracing clock being disabled
- A couple of Sashiko-driven fixes for corner cases in the EL2 tracing
code
- Add missing sysreg tracepoint for the EL2 code
- Tidy-up the mutual exclusion of guest-memfd and MTE
- Update Fuad's email address to point to @linux.dev
Carlo Caione [Mon, 27 Jul 2026 08:36:59 +0000 (10:36 +0200)]
of/address: Fix NULL bus dereference in of_pci_range_parser_one()
The bus matching rework made of_match_bus() return NULL for nodes with
ranges/dma-ranges but no local #address-cells. parser_init() stored that
NULL bus, and the range iterator later dereferenced it.
Reject such nodes in parser_init(), leaving an explicit empty
iterator for callers that ignore the init return, and make
of_dma_get_max_cpu_address() honour the init failure so a rejected node
cannot clamp the DMA limit.
Carlos Song [Mon, 25 May 2026 03:04:00 +0000 (11:04 +0800)]
i2c: imx: mark I2C adapter when hardware is powered down
On some i.MX platforms, certain I2C client drivers keep a periodic
workqueue which continues to trigger I2C transfers.
During system suspend/resume, there exists a time window between:
- suspend_noirq and the system entering suspend
- the system starting to resume and resume_noirq
In this window, the I2C controller resources such as clock and pinctrl
may already be disabled or not yet restored.
If a workqueue triggers an I2C transfer in this period, the driver
attempts to access I2C registers while the hardware resources are
unavailable, which may lead to system hang.
Mark the I2C adapter as suspended during noirq suspend and block new
transfers until resume, ensuring that I2C transfers are only issued
when hardware resources are available.
Fixes: 358025ac091e ("i2c: imx: make controller available until system suspend_noirq() and from resume_noirq()") Signed-off-by: Carlos Song <carlos.song@nxp.com> Cc: <stable@vger.kernel.org> # v6.14+ Reviewed-by: Frank Li <Frank.Li@nxp.com> Acked-by: Oleksij Rempel <o.rempel@pengutronix.de> Signed-off-by: Andi Shyti <andi.shyti@kernel.org> Link: https://lore.kernel.org/r/20260525030400.3182911-1-carlos.song@oss.nxp.com
tracing/fprobe: Roll back on enable_trace_fprobe() failure
enable_trace_fprobe() sets the file link or the TP_FLAG_PROFILE flag and
then registers each trace_fprobe in the probe list. If
__register_trace_fprobe() fails partway through, the function returns
immediately without unregistering the trace_fprobes it already registered
or undoing the file link / flag it set, leaving the event half-enabled and
leaking the registered fprobe(s).
enable_trace_kprobe() already handles this with a rollback path. Do the
same for fprobe: on failure, unregister all probes and clear the file link
or profile flag.
Link: https://lore.kernel.org/all/20260724064208.480030-1-raushan.jhon@gmail.com/ Fixes: 334e5519c375 ("tracing/probes: Add fprobe events for tracing function entry and exit.") Cc: stable@vger.kernel.org Signed-off-by: Raushan Patel <raushan.jhon@gmail.com> Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Merge tag 'for-7.2-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"Zoned mode:
- fix assertion and handle case of finished zone and truncated extent
- fix zone metadata write pointer on actual zone reset
- fix deadlock caused metadata writeback and transaction commit
- fix return value reuse leading to confusion about chunk
reservations
raid56 scrub:
- fix tracking of sector checksums when there are not checksums found
- fix inverted logic when submitting parity read bio
mount/remount fixes:
- fix leaking 'remount in progress' state which can break other
operations to work (qgroup rescan, autodefrag, reclaim)
- adjust using global block reserve after read-only mount when using
rescue= option
- handle missing raid stripe tree when mounted with 'ignorebadroots'
Misc:
- fix -Wmaybe-uninitialized warning in GET_CSUMS ioctl"
* tag 'for-7.2-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: raid56: fix scrub read assembly submitting no reads
btrfs: zoned: skip fully truncated ordered extents at zone finish
btrfs: initialize 'args' to avoid compiler warning in btrfs_ioctl_get_csums()
btrfs: zoned: fix missing chunk metadata reservation
btrfs: raid56: fix an incorrect csum skip during scrub
btrfs: report missing raid stripe tree root during lookup
btrfs: skip global block reserve accounting for rescue mounts
btrfs: zoned: reset meta_write_pointer on zone reset
btrfs: zoned: fix deadlock between metadata writeback and transaction commit
btrfs: fix leaking BTRFS_FS_STATE_REMOUNTING flag
tracing/probes: Reject $arg0 in meta argument expansion
traceprobe_expand_meta_args() parses $argN with simple_strtoul() and
calls sprint_nth_btf_arg(n - 1, ...). For $arg0, n is 0 so the index is
-1. Because ctx->nr_params is signed, the "idx >= nr_params" guard in
sprint_nth_btf_arg() does not catch the negative index, and
ctx->params[-1].name_off is read out of bounds.
The normal per-argument path (parse_probe_vars()) already rejects
$arg0 via its argument-number check, but meta-argument expansion runs
before per-argument parsing and substitutes the value first, bypassing
that check.
Reject $arg0 explicitly during expansion.
Link: https://lore.kernel.org/all/20260724054435.146279-1-raushan.jhon@gmail.com/ Fixes: 18b1e870a496 ("tracing/probes: Add $arg* meta argument for all function args") Cc: stable@vger.kernel.org Signed-off-by: Raushan Patel <raushan.jhon@gmail.com> Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
net: ethernet: mtk_eth_soc: pass eth to mtk_handle_irq_rx in poll_controller
mtk_handle_irq_rx expects a struct mtk_eth * (matching the request_irq
cookie), but mtk_poll_controller incorrectly passed the net_device *.
Calling ndo_poll_controller with CONFIG_NET_POLL_CONTROLLER enabled
would then crash.
Zhao Li [Thu, 23 Jul 2026 01:09:28 +0000 (09:09 +0800)]
wifi: mac80211: validate individual TWT params before driver setup
ieee80211_process_rx_twt_action() only partially validates a received
S1G TWT setup frame before queueing it.
An individual agreement can therefore reach ieee80211_s1g_rx_twt_setup()
with twt->length too short for the full struct ieee80211_twt_params.
The individual path passes twt to drv_add_twt_setup(). Both the tracepoint
and the driver callback consume the complete parameters block, not merely
req_type. Do not pass a short individual agreement to the driver.
Broadcast agreements remain unchanged because they are rejected locally
after accessing only req_type.
Fixes: f5a4c24e689f ("mac80211: introduce individual TWT support in AP mode") Assisted-by: Codex:gpt-5 Assisted-by: Claude:opus-4.8 Signed-off-by: Zhao Li <enderaoelyther@gmail.com> Link: https://patch.msgid.link/20260723010928.76551-1-enderaoelyther@gmail.com
[edit commit message to not overclaim lack of validation nor
understate driver impact] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Zhao Li [Thu, 23 Jul 2026 20:22:23 +0000 (04:22 +0800)]
wifi: cfg80211: publish PMSR request before starting the driver
nl80211_pmsr_start() assigns the request cookie, calls the driver's
->start_pmsr() callback, and only then adds the request to
wdev->pmsr_list, without holding pmsr_lock for the addition.
mac80211_hwsim saves the request in its start callback and returns. Since
nl80211 uses parallel_ops, an immediate REPORT_PMSR can then run before
nl80211_pmsr_start() reaches its post-start list_add_tail(). hwsim also
dispatches reports from its virtio receive workqueue. Completion removes
the request from wdev->pmsr_list under pmsr_lock and frees it.
Thus completion can precede publication, race the unlocked list mutation,
or free the request before nl80211_pmsr_start() reads req->cookie for the
netlink reply.
Add the request to wdev->pmsr_list under pmsr_lock before calling the
driver, and use a cookie value saved before the call so the request is not
dereferenced after a successful start. On an error return the driver has
not retained or completed the request, so remove it from the list under the
lock and free it.
Zhao Li [Tue, 28 Jul 2026 11:53:25 +0000 (19:53 +0800)]
wifi: mwifiex: use the subframe length when parsing A-MSDU TDLS frames
mwifiex_11n_dispatch_amsdu_pkt() splits an A-MSDU with
ieee80211_amsdu_to_8023s() and walks the resulting subframes. For each
subframe it passes the subframe data pointer to
mwifiex_process_tdls_action_frame(), but pairs it with skb->len, the
length of the A-MSDU parent, instead of rx_skb->len:
The parent is not a valid description of that buffer, and may not be
valid memory at all. ieee80211_amsdu_to_8023s() ends with
if (!reuse_skb)
dev_kfree_skb(skb);
and it only sets reuse_skb when the parent is linear, is not a
head_frag, and is being consumed as the *last* subframe. So when the
parent does not qualify for reuse it has already been freed, and the
read of skb->len is a use-after-free. When it is reused, skb->len is
the length of the last subframe, applied to every earlier subframe,
which over-states the buffer whenever an earlier subframe is shorter.
The callee cannot absorb a wrong length, because it derives its own
ceiling from the value it is given. Each frame type computes
ies_len = len - sizeof(struct ethhdr) - TDLS_*_FIX_LEN;
and the element walk is then bounded entirely against that ceiling,
so a too-large len moves end past the end of the subframe and the walk
reads and copies beyond it. The A-MSDU layout is chosen by the sender,
which makes the difference between the last subframe and a shorter
earlier one remotely selectable. Reaching this requires TDLS support in
firmware and the TDLS ethertype on the subframe.
The other caller, mwifiex_process_rx_packet(), is correct: it passes a
pointer and a length that describe the same region of the RX buffer.
Pass rx_skb->len, the length of the subframe actually being parsed.
Fixes: 776f742040ca ("mwifiex: fix AMPDU not setup on TDLS link problem") Assisted-by: Codex:gpt-5.6-sol Assisted-by: Kimi:K3 Cc: stable@vger.kernel.org Signed-off-by: Zhao Li <enderaoelyther@gmail.com> Link: https://patch.msgid.link/20260728115325.19128-1-enderaoelyther@gmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
wifi: cfg80211: validate IEs in cfg80211_wext_siwgenie()
The KASAN allocation trace shows that a malformed IE buffer is
stored via SIOCSIWGENIE (cfg80211_wext_siwgenie()) without any
validation. The crash trace shows that a subsequent SIOCSIWESSID
triggers a connection attempt which calls cfg80211_sme_get_conn_ies()
to process the stored IE buffer, causing:
- An out-of-bounds read in skip_ie() which reads ies[pos+1]
(the length byte) past the end of the 1-byte buffer.
- An integer underflow in the memcpy size argument when offs
returned by ieee80211_ie_split() exceeds ies_len, causing
unsigned subtraction to wrap to SIZE_MAX and triggering a
fortify panic.
Fix this by validating the IE buffer in cfg80211_wext_siwgenie()
before storing it.
if (start_txq)
ieee80211_agg_start_txq(sta, tid, false);
if (send_delba)
ieee80211_send_delba(..., tid_tx->ndp);
That read is not covered by an RCU read-side critical section, and it runs
in preemptible process context: both callers hold the wiphy mutex, reaching
it either from the ieee80211_ba_session_work() wiphy work or from
ieee80211_sta_tear_down_BA_sessions() during station teardown.
Softirqs can run in that window too, both from the local_bh_enable() that
ends ieee80211_agg_start_txq() and from any interrupt exit, so the RCU
callback can free tid_tx before the read.
Driving the function from a test module with the grace period forced into
that window, KASAN reports the read, and the free arrives on the ordinary
RCU softirq path:
BUG: KASAN: slab-use-after-free in ieee80211_stop_tx_ba_cb+0x3cd/0x400
Read of size 1 at addr ffff888002b9f52e by task kworker/0:1/10
[...]
Freed by task 57:
__kasan_slab_free+0x47/0x70
__rcu_free_sheaf_prepare+0x70/0x250
rcu_free_sheaf_nobarn+0x18/0x40
rcu_core+0x426/0x1310
handle_softirqs+0x144/0x590
__irq_exit_rcu+0xea/0x150
irq_exit_rcu+0x9/0x20
sysvec_apic_timer_interrupt+0x6b/0x80
asm_sysvec_apic_timer_interrupt+0x1a/0x20
send_delba is only set when tx_stop is set, which happens for
AGG_STOP_LOCAL_REQUEST alone, so this is reached on local teardown -
session idle timeout, PTK rekey, suspend, HW reconfig - and not from a
peer's DELBA.
Read ndp into a local before the session is freed, while sta->lock is still
held. tid_tx->ndp has a single writer, in
ieee80211_tx_ba_session_handle_start(), which cannot run concurrently here:
both paths are serialised by the wiphy mutex, and the session is already
marked HT_AGG_STATE_STOPPING at this point. tid_tx->ndp is also the only
tid_tx dereference left after ieee80211_remove_tid_tx() in this function.
Fixes: 98acd4c1d9f7 ("wifi: mac80211: add support for NDP ADDBA/DELBA for S1G") Assisted-by: Codex:gpt-5.6-sol Assisted-by: Kimi:K3 Cc: stable@vger.kernel.org Signed-off-by: Zhao Li <enderaoelyther@gmail.com> Link: https://patch.msgid.link/20260728112156.96822-1-enderaoelyther@gmail.com
[move/change the comment a bit to be more general not just on ndp,
initialize ndp directly] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
drm/vc4: Zero the tile state data array before each BIN job
The binner BO is a single 16MB buffer split into 512KB slots that are
handed out to jobs at submission time and recycled as jobs complete,
without ever being cleared. Each slot holds the job's Tile State Data
Array (TSDA) at its start, followed by the tile allocation pool.
While the tile allocation pool is only walked by the render thread
through branches the binner generated during the current job, the
TSDA is the PTB's own per-tile bookkeeping and is consumed by the
hardware itself. Although the kernel sets the "Auto-initialise Tile
State Data Array" flag in the tile binning mode configuration, the
PTB demonstrably still acts on stale tile state left by the slot's
previous user: the binner ends up creating invalid command streams
with invalid primitive streams and branches, which can cause GPU hangs
as observed in [1][2].
Zero the TSDA when the job's binning slot is configured. This clears
48 bytes per tile (~24KB for a 1080p frame) in the submission path, and
guarantees the PTB never sees another job's tile state.
The tile count is only checked for being non-zero today, so the 8-bit
fields it comes from can describe a tile state array almost six times
larger than the slot it has to live in. Bound it before the slot is
handed out, since such size decides how much of the slot is left for
the tile alloc pool.
drm/vc4: Supply the overflow slot size in BPOS, not the whole bin BO size
vc4_overflow_mem_work() points BPOA at a 512KB slot inside the 16MB
binner BO, but writes the size of the whole BO to BPOS. On every binner
out-of-memory event the PTB is therefore authorized to write tile lists
across all the other slots (which may hold the tile state, tile alloc and
overflow memory of in-flight jobs) and, for any slot but the first, past
the end of the binner BO into unrelated CMA memory.
Since CMA pages are recycled into page cache and user allocations, this
is arbitrary memory corruption by GPU DMA. In practice it shows up as GPU
hangs with corrupted control list pointers, userspace heap corruption, a
GPU that stays permanently wedged after the first hang, and occasional
full system crashes, whenever a job overflows the initial binner slot.
The bug dates back to the conversion from a dedicated overflow BO (where
writing the full BO size was correct) to the slotted binner BO.
Fixes: 553c942f8b2c ("drm/vc4: Allow using more than 256MB of CMA memory.") Cc: stable@vger.kernel.org Assisted-by: Claude:claude-opus-4.8 Signed-off-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com> Reviewed-by: Maรญra Canal <mcanal@igalia.com> Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> Link: https://patch.msgid.link/20260727-vc4-bin-oom-fixes-v2-1-0d8a5eddc7c9@igalia.com Signed-off-by: Maรญra Canal <mcanal@igalia.com>
Eric Dumazet [Fri, 24 Jul 2026 09:11:37 +0000 (09:11 +0000)]
net: udp_tunnel: fix memory leak in udp_tunnel_nic_unregister()
syzbot reported a memory leak [1] in the UDP tunnel NIC offload code.
When device registration fails (e.g. in register_netdevice()), netdev core
unwinds by sending a single NETDEV_UNREGISTER notification. If work was queued
during NETDEV_REGISTER (utn->work_pending is set), udp_tunnel_nic_unregister()
returns early:
if (utn->work_pending)
return;
Because failed registrations do not enter netdev_wait_allrefs_any(), no
subsequent NETDEV_UNREGISTER rebroadcast will ever occur. As a result, the
struct udp_tunnel_nic allocated in udp_tunnel_nic_alloc() is leaked
permanently.
Fix this by removing the early return. Instead, synchronously cancel any
pending work with cancel_delayed_work_sync() before freeing @utn.
To be able to call cancel_delayed_work_sync() while holding RTNL (the work also
needs RTNL), switch udp_tunnel_nic_device_sync_work() to rtnl_trylock(). If RTNL
is contended, requeue the work with a 1 jiffy delay (via queue_delayed_work())
to prevent high CPU contention while waiting for RTNL lock.
The utn->work_pending bookkeeping is no longer needed and is removed, as
the workqueue core already tracks the pending/running state of the work.
Fixes: cc4e3835eff4 ("udp_tunnel: add central NIC RX port offload infrastructure") Reported-by: syzbot+eca845fb8c18dd6b44c1@syzkaller.appspotmail.com Closes: https://lore.kernel.org/netdev/6a632b15.dde6c935.cf6c8.0011.GAE@google.com/T/#u Signed-off-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20260724091137.1792543-1-edumazet@google.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Xuanqiang Luo [Thu, 23 Jul 2026 06:04:45 +0000 (14:04 +0800)]
bpf: lwt: Fix dst reference leak on reroute failure
bpf_lwt_xmit_reroute() obtains a referenced dst from the route
lookup. When skb_cow_head() fails before that dst is installed on the
skb, the error path only frees the skb. The skb still owns its previous
dst, so the newly looked up dst reference is leaked.
Release the new dst reference before freeing the skb on this error
path.
Xuanqiang Luo [Thu, 23 Jul 2026 10:54:54 +0000 (18:54 +0800)]
net/smc: fix socket use-after-free during link group termination
__smc_lgr_terminate() drops conns_lock after finding a connection in
lgr->conns_all, but before taking a reference on its socket. The connection
is embedded in the socket, and its registration reference protects it only
while the connection remains in the tree.
A concurrent close can unregister the connection and drop that reference,
freeing the socket before the termination worker reaches sock_hold().
The race is reachable when close overlaps link group termination.
Local stress testing reproduced the use-after-free and KASAN reported:
BUG: KASAN: slab-use-after-free in __smc_lgr_terminate.part.0 [smc]
Write of size 4 by task kworker/3:3
Workqueue: events smc_lgr_terminate_work [smc]
__smc_lgr_terminate.part.0 [smc]
The socket was allocated by smc_create(), freed through
slab_free_after_rcu_debug(), and was followed by:
refcount_t: addition on 0; use-after-free.
__smc_lgr_terminate.part.0 [smc]
Take the socket reference while conns_lock still protects the tree entry.
The unregister path then cannot drop the last reference until termination
has finished using the socket.
net/sched: cls_u32: validate offshift to prevent shift-out-of-bounds
u32_change() copies the user-provided tc_u32_sel.offshift (unsigned char,
0-255) into the kernel knode object without bounds validation. When a
packet later hits u32_classify() with TC_U32_VAROFFSET set, it evaluates
`ntohs(offmask & *data) >> offshift` where the left operand is a 16-bit
value promoted to a 32-bit int. Any offshift >= 32 is undefined behavior
per C11 6.5.7p3, triggerable by an unprivileged user via user/network
namespaces.
UBSAN: shift-out-of-bounds in net/sched/cls_u32.c:236:43
shift exponent 32 is too large for 32-bit type int
Fix this by rejecting offshift >= 16 during filter creation in
u32_change().
Yehyeong Lee [Thu, 23 Jul 2026 01:08:29 +0000 (10:08 +0900)]
net: mpls: initialize rtm_tos in mpls_getroute()
mpls_getroute() builds the RTM_NEWROUTE reply to an RTM_GETROUTE
request by filling a struct rtmsg allocated from an skb whose data
area is not zeroed (alloc_skb(NLMSG_GOODSIZE, ...)). It sets every
field of the header except rtm_tos:
struct rtmsg has no padding, so the one uninitialised byte rtm_tos
(offset 3) is copied straight to user space on recvmsg(), leaking a
byte of uninitialised heap memory. This is in contrast to
mpls_dump_route(), which fills the very same header and does set
rtm_tos = 0.
Initialize rtm_tos to 0, matching mpls_dump_route().
Reproduced with KMSAN by adding an MPLS route and issuing a
non-RTM_F_FIB_MATCH RTM_GETROUTE for its label:
BUG: KMSAN: kernel-infoleak in _copy_to_iter+0x36c/0x33f0
_copy_to_iter+0x36c/0x33f0
__skb_datagram_iter+0x196/0x12c0
skb_copy_datagram_iter+0x5b/0x210
netlink_recvmsg+0x37b/0xef0
...
Uninit was created at:
__alloc_skb+0x8ca/0x10e0
mpls_getroute+0x1280/0x3a40
rtnetlink_rcv_msg+0x1138/0x15a0
...
Byte 19 of 64 is uninitialized
Xuanqiang Luo [Wed, 22 Jul 2026 08:38:58 +0000 (16:38 +0800)]
fou: Fix use-after-free in fou_create()
fou_create() publishes struct fou through sk_user_data before adding the
new FOU port to the per-netns list. If fou_add_to_port_list() fails,
the error path frees fou while it is still reachable through
sk_user_data. A concurrent receive can then dereference the freed
object in fou_from_sock().
This ordering issue was previously noted in the linked discussion.
The failure is reachable when local port 0 is requested. Each socket
binds to a different ephemeral port, but fou_cfg_cmp() compares the
requested port 0 and reports -EALREADY once an entry already exists.
Release the tunnel socket before freeing fou so sk_user_data is cleared
first, and defer reclamation with kfree_rcu() to protect concurrent RCU
readers. This matches the lifetime handling in fou_release().
i2c: designware: defer probe if child GpioInt controllers are not bound
I2C controllers may have child devices with GpioInt resources that
depend on GPIO controllers being fully initialized. If the I2C
controller probes and enumerates children before the referenced GPIO
controller has completed probe, GPIO interrupts may not be properly
configured, leading to device failures.
On Lenovo Yoga 7 14AGP11, the WACF2200 touchscreen (child of
AMDI0010:02) has a GpioInt resource pointing to GPIO 157 on the
pinctrl-amd controller (AMDI0030:00). When i2c-designware probes
AMDI0010:02 before pinctrl-amd finishes initializing, I2C transactions
fail with lost arbitration errors:
0.285952 amd_gpio_probe: registering gpiochip <- GPIO chip visible
0.287121 amd_gpio_probe: requesting parent IRQ <- probe still running
0.301454 AMDI0010:02 dw_i2c_plat_probe: start <- races here
2.348157 lost arbitration
Add a dependency check that walks ACPI child devices and defers probe
until any referenced GPIO controller is bound.
Fixes: 3812a9e84265 ("pinctrl-amd: enable IRQ for WACF2200 touchscreen on Lenovo Yoga 7 14AGP11") Closes: https://bugzilla.kernel.org/show_bug.cgi?id=221494 Suggested-by: Mario Limonciello <mario.limonciello@amd.com> Suggested-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Hardik Prakash <hardikprakash.official@gmail.com> Assisted-by: Claude:claude-sonnet-5 Assisted-by: DeepSeek:deepseek-v4-pro Cc: <stable@vger.kernel.org> # v7.1+ Acked-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com> Reviewed-by: Andy Shevchenko <andriy.shevchenko@intel.com> Signed-off-by: Andi Shyti <andi.shyti@kernel.org> Link: https://lore.kernel.org/r/20260718054330.8975-2-hardikprakash.official@gmail.com
powerpc/ps3: Fix map failure path in dma_ioc0_map_pages()
If lv1_put_iopte() fails in dma_ioc0_map_pages(), the error path
decrements iopage but keeps using the failed mapping's offset. As a
result, it repeatedly tries to invalidate the failed IOPTE slot and
leaves the already installed IOPTEs valid.
Recompute offset and invalidate the installed IOPTEs instead.
powerpc: Fix exit_flags field placement in pt_regs for ptrace
Commit d7a6797e0bc1 ("powerpc: add exit_flags field in pt_regs") added
the exit_flags field to struct pt_regs to pass internal exit control
flags (e.g. _TIF_RESTOREALL) from syscall_exit_prepare() to the
low-level assembly exit path.
However, the field was placed in a way that was visible to userspace
tools such as strace via PTRACE_GETREGS, or caused a struct layout or
size regression observable through ptrace. The field is purely
kernel-internal and must not be exposed beyond the user_pt_regs
boundary.
Move exit_flags into struct thread_info where it is only accessible to
the kernel, and keep it out of the ptrace-visible register window
entirely.
Fixes: d7a6797e0bc1 ("powerpc: add exit_flags field in pt_regs") Reported-by: Dmitry V. Levin <ldv@strace.io> Closes: https://lore.kernel.org/all/20260722070155.GA11808@strace.io/ Signed-off-by: Mukesh Kumar Chaurasiya (IBM) <mkchauras@gmail.com> Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com> Link: https://patch.msgid.link/20260723194809.4046600-1-mkchauras@gmail.com
powerpc/970: fix nap return address corruption on async interrupt exit
On PowerMac G5 (PPC970, CONFIG_PPC_970_NAP) the system panics shortly
after boot with symptoms including instruction fetch faults, kernel data
access faults, and stack corruption, predominantly on SMP and always
somewhere inside softirq processing.
The PPC970 idle path works by setting _TLF_NAPPING in the current
thread's local flags before entering the MSR_POW nap loop. When any
async interrupt wakes the CPU, nap_adjust_return() is expected to detect
_TLF_NAPPING, clear it, and rewrite regs->NIP to power4_idle_nap_return
so that the interrupt returns cleanly to the caller of power4_idle_nap()
rather than back into the nap spin loop.
DEFINE_INTERRUPT_HANDLER_ASYNC generates the following sequence:
irq_enter_rcu();
____func(regs); /* timer_interrupt / do_IRQ body */
irq_exit_rcu(); /* softirqs run here, irqs re-enabled */
arch_interrupt_async_exit_prepare(regs); /* nap_adjust_return was here */
irqentry_exit(regs, state);
irq_exit_rcu() calls invoke_softirq() -> do_softirq_own_stack(), which
runs softirqs with hardware interrupts re-enabled. A nested async
interrupt can therefore arrive while _TLF_NAPPING is still set. That
nested interrupt reaches nap_adjust_return() in its own
arch_interrupt_async_exit_prepare() call, finds _TLF_NAPPING set, and
redirects *its own* regs->NIP to power4_idle_nap_return. Returning via
that blr with an unrelated LR on the softirq stack jumps to a garbage
address, causing the observed crashes.
The comment that previously lived in arch_interrupt_async_exit_prepare()
even described this exact hazard ("must come before irq_exit()"), but
nap_adjust_return() was placed after irq_exit_rcu() in the macro, so
the protection was never effective.
Fix this by calling nap_adjust_return() inside DEFINE_INTERRUPT_HANDLER_ASYNC
immediately before irq_exit_rcu(), ensuring _TLF_NAPPING is cleared and
regs->NIP is adjusted before any code that can re-enable interrupts or
invoke softirqs runs. Move the explanatory comment into
nap_adjust_return() itself and remove it from arch_interrupt_async_exit_prepare().
powerpc/pseries: Skip vpa_init() for boot cpu in smp_setup_cpu()
During pSeries_setup_arch(), VPA for boot-cpu is first to be
initialized. However later in the boot, smp_setup_cpu() is called for
setting up VPA on boot and secondary cpus that were brought online. This
results in vpa_init() being called twice for boot-cpu and three redundant
H_REGISTER_VPA hcalls being made to the hypervisor.
Fix this by adding an extra condition in smp_set_cpu() to call vpa_init()
only on non boot-cpus.
powerpc/pseries: Ensure vpa,slb_shadow & dtl are unregistered during crash
Currently pseries_kexec_cpu_down() skips unregistering vpa, slb_shadow and
dtl areas during a crash and kexec shutdown path. It was done to avoid
doing an HCALL while crashing. However recently Anushree reported that
during kernel crash while the kdump kernel was coming up, Hypervisor
reported invalid values for 'vpa.yield_count' while it dispatching L2-KVM
Guest vcpus. The error manifested as debug build Hypervisor assert
triggering to indicate possible VPA corruption.
Looking at the kexec cpu offline path it was discovered that during crash
kernel doesn't unregister the VPA/SLB-Shadow/DTL area with
Hypervisor. Instead it re-allocates and re-registers these areas
for cpus during boot. During kexec boot the previously allocated areas
can get overwritten with new content without hypervisor knowledge. This
creates a small window where while kexec kernel boots and the L2-VCPUs are
being dispatched, Hypervisor may try to read/write to a wrong memory area
which previously belonged to older VPA.
Fix this possible race and memory corruption by updating
pseries_kexec_cpu_down() to also unregister vpa,slb_shadow & dtl areas
during a kernel crash.
This drive is a Samsung 870 EVO SSD in good SMART standing, and I wasn't
aware of any reason it should be taking so long to standby. The issue is
intermittent, but I observed it sometimes taking 7 seconds to manually
standby. I assume this was interruption of background maintenance after
a power outage.
As a desktop user, I would prefer to wait the extra 2 seconds at suspend
to let the drive finish its business rather than drop the rails from
under it.
The change from 30 to 5 seconds was implicit when switching suspend
from START STOP UNIT to an internal command with no timeout table entry.
No reason was stated for the change.
Fixes: aa3998dbeb3a ("ata: libata-scsi: Disable scsi device manage_system_start_stop") Cc: stable@vger.kernel.org Signed-off-by: Matt Vollrath <tactii@gmail.com> Assisted-by: Claude:claude-5-fable Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Randy Dunlap [Sat, 25 Jul 2026 01:52:09 +0000 (18:52 -0700)]
ata: libata: avoid kernel-doc warnings
Modify comments to prevent kernel-doc warnings:
- use "/*" for a non-kernel-doc comment
- add a Returns: section for ata_id_major_version()
Warning: include/linux/ata.h:770 Cannot find identifier on line:
*
Warning: include/linux/ata.h:782 function parameter 'id' not described in 'ata_id_sct_data_tables'
Warning: include/linux/ata.h:782 expecting prototype for Word(). Prototype was for ata_id_sct_data_tables() instead
Warning: include/linux/ata.h:820 No description found for return value of 'ata_id_major_version'
Signed-off-by: Randy Dunlap <rdunlap@infradead.org> Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Frank Sorenson [Sat, 25 Jul 2026 21:04:44 +0000 (21:04 +0000)]
cifs: add fscache_resize_cookie() to cifs_setsize()
Several code paths update the VFS inode size by calling
netfs_resize_file() and cifs_setsize(), but omit the corresponding
fscache_resize_cookie() call, leaving the fscache cookie out of sync
with the actual file size:
- cifs_file_set_size() in inode.c: server-side truncation via setattr
- cifs_do_truncate() in file.c: truncates to zero on O_TRUNC open
- smb2_duplicate_extents() in smb2ops.c: file clone extending EOF
- smb3_simple_falloc() in smb2ops.c: two branches that extend EOF
via write-range and SMB2_set_eof respectively
Since every caller of cifs_setsize() must resize the fscache cookie,
add the call to cifs_setsize() itself, consistent with how
truncate_pagecache() is already consolidated there.
Fixes: 70431bfd825d ("cifs: Support fscache indexing rewrite") Fixes: 93a43155127f ("cifs: Fix missing set of remote_i_size") Fixes: 110fee6b9bb5 ("smb: client: fix missing timestamp updates with O_TRUNC") Fixes: 7a06d3b816d7 ("smb/client: emulate small EOF-extending mode 0 fallocate ranges") Cc: stable@vger.kernel.org Cc: David Howells <dhowells@redhat.com> Cc: Paulo Alcantara <pc@manguebit.org> Cc: Huiwen He <hehuiwen@kylinos.cn> Signed-off-by: Frank Sorenson <sorenson@redhat.com> Reviewed-by: Paulo Alcantara <pc@manguebit.org> Signed-off-by: Steve French <stfrench@microsoft.com>
Eric Joyner [Thu, 23 Jul 2026 04:13:42 +0000 (21:13 -0700)]
ethtool: Embed FEC hist ranges as buffer in struct
When a driver's .get_fec_stats() handler is called and the driver
supports FEC histogram stats, the driver supplies the histogram bin
ranges via a pointer. This pointer is assigned while under the netdev
ops lock in fec_prepare_data(), but the actual data is only read after
the lock is released; so this allows the driver to change the ranges
(e.g. from another .get_fec_stats() call) while the current call chain
is reading them in fec_fill_reply().
Fix this by adding an ethtool core-owned buffer, ranges_buf, to struct
ethtool_fec_hist. Drivers whose ranges are built dynamically (currently
just mlx5) fill ranges_buf and then point the existing ranges pointer at
it, giving ethtool a consistent copy that stays valid after the netdev
ops lock is dropped and later in fec_fill_reply(). Drivers whose ranges
are compile-time constants (bnxt, netdevsim) are unaffected by the
potential race and keep setting the existing ranges pointer to their
constant array, without making copies.
Yun Lu [Tue, 21 Jul 2026 02:38:36 +0000 (10:38 +0800)]
rtase: fix double free of multi-frag skb on DMA map failure
In rtase_start_xmit(), when the head buffer DMA mapping fails after
rtase_xmit_frags() has mapped all fragments, the error path clears
the fragment descriptors with rtase_tx_clear_range(), which frees
the skb through the last-frag slot and accounts tx_dropped. Control
then falls through to the common error label, which frees the same
skb a second time and counts it again.
Return right after clearing the fragments when the skb owns frags;
the no-frag case still drops through and frees the head skb once.
Fixes: d6e882b89fdf ("rtase: Implement .ndo_start_xmit function") Signed-off-by: Yun Lu <luyun@kylinos.cn> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Reviewed-by: Justin Lai <justinlai0215@realtek.com> Link: https://patch.msgid.link/20260721023836.6691-1-luyun_611@163.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Gate the SIOCDEVPRIVATE ioctl commands SIOC_QETH_ADP_SET_SNMP_CONTROL,
SIOC_QETH_GET_CARD_TYPE and SIOC_QETH_QUERY_OAT with CAP_NET_ADMIN
capable check to ensure unprivileged users cannot invoke them.
Fixes: 18787eeebd71 ("qeth: use ndo_siocdevprivate") Cc: stable@vger.kernel.org Suggested-by: Christian Borntraeger <borntraeger@linux.ibm.com> Reviewed-by: Christian Borntraeger <borntraeger@linux.ibm.com> Reviewed-by: Alexandra Winter <wintera@linux.ibm.com> Signed-off-by: Aswin Karuvally <aswin@linux.ibm.com> Link: https://patch.msgid.link/20260723140050.762991-1-aswin@linux.ibm.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
hwmon: (adt7470) Fix PWM auto temp state array and bounds check
In pwm_auto_temp_store(), the parsed user input was missing bounds
checks, allowing values > 0xF to overflow into the adjacent channel's
bits. Furthermore, the value was being incorrectly written to the
pwm_automatic state array instead of pwm_auto_temp.
Fix this by rejecting values > 0xF with -EINVAL, and assigning the
value to the correct array only after a successful I2C write.
Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/all/20260727034932.0B7C41F000E9@smtp.kernel.org/#t Fixes: 6f9703d0be16 ("hwmon: add support for adt7470") Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com> Link: https://lore.kernel.org/r/20260727-adt7470_fixes-v2-8-598e38a46ba6@gmail.com Signed-off-by: Guenter Roeck <linux@roeck-us.net>
hwmon: (adt7470) Fix divide-by-zero TOCTOU crash in fan speed read
If the fan data becomes 0 between the FAN_DATA_VALID() check and the
FAN_PERIOD_TO_RPM() conversion, it will result in a divide-by-zero crash
due to a race with a concurrent update of the cached fan value.
adt7470_pwm_read() currently ignores failures returned by
pwm1_freq_get(). If the register read fails, the negative error code is
returned through *val while the function itself reports success,
potentially exposing a negative PWM frequency through sysfs.
Fix this by using the cached PWM frequency maintained by the driver,
eliminating the register access from the read path.
Apart from the corrected error propagation and using the cached value,
no functional change is intended.
hwmon: (adt7470) Fix swapped PWM3 and PWM4 auto mode masks
The ADT7470_PWM3_AUTO_MASK and ADT7470_PWM4_AUTO_MASK macros are
currently defined with swapped bit values.
According to Table 22 of the ADT7470 datasheet, the Fan Control Mode
Configuration for register 0x69 follows the exact same bit position
layout as register 0x68:
- 0x68 Bit[7] corresponds to BHVR1 (PWM1) -> 0x80
- 0x68 Bit[6] corresponds to BHVR2 (PWM2) -> 0x40
- 0x69 Bit[7] corresponds to BHVR3 (PWM3) -> 0x80
- 0x69 Bit[6] corresponds to BHVR4 (PWM4) -> 0x40
Consequently, PWM3 should use mask 0x80 and PWM4 should use 0x40.
This typo did not cause any functional bugs because these specific
macros are never referenced in the driver code. Instead, the driver
correctly applies the configuration by relying on the modulo parity of
the channel index (e.g., `channel % 2`) to selectively apply either
ADT7470_PWM1_AUTO_MASK (0x80) or ADT7470_PWM2_AUTO_MASK (0x40).
Since the bit layout is identical between the two configuration
registers, the hardware is currently configured correctly.
Fix the macro definitions to reflect the datasheet accurately and
prevent future bugs or confusion during code review and refactoring.
As this is a purely cosmetic fix with no functional impact, a backport
to stable kernels is not necessary.
hwmon: (adt7470) Fix temperature alarm logic in hwmon_temp_read()
During the conversion the alarm callback started interpreting the
channel index as an alarm bitmask, resulting in incorrect alarm
reporting. Compute the proper alarm bit instead.
hwmon: (adt7470) Fix busy-loop and I2C flooding in update thread
When userspace configures 'auto_update_interval' to 0 via sysfs, the
background kthread executes schedule_timeout_interruptible(0), which
returns immediately.
If 'num_temp_sensors' is concurrently or previously set to 0, the
msleep_interruptible() delay inside adt7470_read_temperatures() also
becomes 0. This combination forces the background thread into a tight,
unbounded busy-loop, hogging the CPU and flooding the I2C bus with a
continuous stream of transactions.
Fix this vulnerability by raising the lower limit of the clamp_val in
auto_update_interval_store() from 0 to 500 milliseconds. This guarantees
a reasonable minimum sleep window between sensor updates, protecting the
system from intentional or accidental I2C bus denial of service.
Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/r/20260716213252.EACA71F000E9@smtp.kernel.org Fixes: 89fac11cb3e7 ("adt7470: make automatic fan control really work") Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com> Link: https://lore.kernel.org/r/20260727-adt7470_fixes-v2-3-598e38a46ba6@gmail.com Signed-off-by: Guenter Roeck <linux@roeck-us.net>
hwmon: (adt7470) Fix cache updated before hardware write on I2C error
adt7470_temp_write() and adt7470_pwm_write() update the driver's
cached values (temp_min, temp_max, pwm_input, pwm_enable) before issuing
the corresponding regmap_write(), and never check whether the write
succeeded before committing that update. If the I2C transaction fails,
the function correctly propagates the error to the caller, but the cache
silently keeps the new value, which was never actually applied to the
hardware. Subsequent reads then report a value that does not match the
device state.
Reorder both write paths to update the cache only after a successful
regmap_write(), so the cache always reflects what was actually
written to the hardware.
hwmon: (adt7470) Fix fans stuck in manual mode on I2C errors
During adt7470_read_temperatures(), the driver temporarily switches
the PWM channels to manual mode, performs the temperature collection,
and then restores the original configuration registers.
However, if an I2C transaction fails at any point after entering manual
mode, the function aborts and returns immediately. This leaves the
configuration registers un-restored, permanently trapping the fans in
manual mode.
Introduce a recovery path to ensure that the original PWM configuration
registers are always restored, even when intermediate I2C operations
fail.
Reported-by: sashiko-bot@kernel.org Closes: https://lore.kernel.org/r/20260716213252.EACA71F000E9@smtp.kernel.org Fixes: ef67959c4253 ("hwmon: (adt7470) Convert to use regmap") Signed-off-by: Luiz Angelo Daros de Luca <luizluca@gmail.com> Link: https://lore.kernel.org/r/20260727-adt7470_fixes-v2-1-598e38a46ba6@gmail.com Signed-off-by: Guenter Roeck <linux@roeck-us.net>
nv_remove() frees the per-CPU txrx_stats before unregister_netdev().
Until unregister completes, ndo_get_stats64, the NAPI/xmit data path,
and nv_close()/drain may still access txrx_stats, leading to a
use-after-free.
Free the stats only after unregister_netdev().
Fixes: f4b633b911fd ("forcedeth: use per cpu to collect xmit/recv statistics") Signed-off-by: Chenguang Zhao <zhaochenguang@kylinos.cn> Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20260723092637.2135095-1-chenguang.zhao@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Frank Sorenson [Fri, 24 Jul 2026 16:30:36 +0000 (11:30 -0500)]
cifs: fix time_last_write stamp placement in setattr/truncate paths
cifs_file_set_size() calls cifs_setsize() on success, which calls
i_size_write(), updating i_size to the new value. The subsequent
check attrs->ia_size != i_size_read() in both cifs_setattr_unix()
and cifs_setattr_nounix() therefore always evaluates false after a
successful cifs_file_set_size(), making the smp_store_release() of
time_last_write dead code. The truncate path was unprotected against
stale readdir size updates.
Move the stamp to before the cifs_file_set_size() RPC call, guarded
by attrs->ia_size != i_size_read() to exclude no-op same-size
ftruncate(2) calls from stamping time_last_write unnecessarily.
On the error path the stamp remains rather than being restored:
restoring a stale snapshot (prev_tlw) could silently erase a
concurrent _cifsFileInfo_put() close stamp if that close arrived
between the READ_ONCE and the smp_store_release. readdir is
suppressed until the stamp expires, which extends beyond one acregmax
if the caller retries failed truncations. stat() is unaffected: the
cifs_revalidate_dentry_attr() path calls cifs_fattr_to_inode() with
from_readdir=false, which bypasses the time_last_write check in
is_size_safe_to_change() entirely and always writes the authoritative
QUERY_INFO result to i_size.
Remove the now-unreachable stamp from the dead block in both functions.
Fixes: e8a8d54c2d50 ("cifs: prevent readdir from changing file size due to stale directory metadata") Signed-off-by: Frank Sorenson <sorenson@redhat.com> Reviewed-by: Paulo Alcantara (Red Hat) <pc@manguebit.org> Signed-off-by: Steve French <stfrench@microsoft.com>
projects / thirdparty / linux.git / log
