veth: fix OOB txq access in veth_poll() with asymmetric queue counts
XDP redirect into a veth device (via bpf_redirect()) calls
veth_xdp_xmit(), which enqueues frames into the peer's ptr_ring using
smp_processor_id() % peer->real_num_rx_queues
as the ring index. With an asymmetric veth pair where the peer has
fewer TX queues than RX queues, that index can exceed
peer->real_num_tx_queues.
veth_poll() then resolves peer_txq for the ring via:
where queue_idx = rq->xdp_rxq.queue_index. When queue_idx exceeds
peer_dev->real_num_tx_queues this is an out-of-bounds (OOB) access
into the peer's netdev_queue array, triggering DEBUG_NET_WARN_ON_ONCE
in netdev_get_tx_queue().
The normal ndo_start_xmit path is not affected: the stack clamps
skb->queue_mapping via netdev_cap_txqueue() before invoking
ndo_start_xmit, so rxq in veth_xmit() never exceeds real_num_tx_queues.
Fix veth_poll() by clamping: only dereference peer_txq when queue_idx is
within bounds, otherwise set it to NULL. The out-of-range rings are fed
exclusively via XDP redirect (veth_xdp_xmit), never via ndo_start_xmit
(veth_xmit), so the peer txq was never stopped and there is nothing to
wake; NULL is the correct fallback.
Reported-by: Sashiko <sashiko-bot@kernel.org> Closes: https://lore.kernel.org/all/20260502071828.616C3C19425@smtp.kernel.org/ Fixes: dc82a33297fc ("veth: apply qdisc backpressure on full ptr_ring to reduce TX drops") Signed-off-by: Jesper Dangaard Brouer <hawk@kernel.org> Link: https://patch.msgid.link/20260505132159.241305-2-hawk@kernel.org Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Bobby Eshleman [Tue, 5 May 2026 01:42:11 +0000 (18:42 -0700)]
eth: fbnic: fix double-free of PCS on phylink creation failure
fbnic_phylink_create() stores the newly allocated PCS in fbn->pcs and
then calls phylink_create(). When phylink_create() fails, the error path
correctly destroys the PCS via xpcs_destroy_pcs(), but the caller,
fbnic_netdev_alloc(), responds by invoking fbnic_netdev_free() which
calls fbnic_phylink_destroy(). That function finds fbn->pcs non-NULL and
calls xpcs_destroy_pcs() a second time on the already-freed object,
triggering a refcount underflow use-after-free:
Instead of calling fbnic_phylink_destroy(), the prior initialization of
netdev should just be unrolled with free_netdev() and clearing
fbd->netdev.
Clearing fbd->netdev to NULL avoids UAF in init_failure_mode where
callers guard by checking !fbd->netdev, such as fbnic_mdio_read_pmd().
These callers remain active even after a failed probe, so fdb->netdev
still needs to be cleared.
In gmac_rx() (drivers/net/ethernet/cortina/gemini.c), when
gmac_get_queue_page() returns NULL for the second page of a multi-page
fragment, the driver logs an error and continues — but does not free the
partially assembled skb that was being assembled via napi_build_skb() /
napi_get_frags().
Free the in-progress partially assembled skb via napi_free_frags()
and increase the number of dropped frames appropriately
and assign the skb pointer NULL to make sure it is not lingering
around, matching the pattern already used elsewhere in the driver.
Fixes: 4d5ae32f5e1e ("net: ethernet: Add a driver for Gemini gigabit ethernet") Signed-off-by: Andreas Haarmann-Thiemann <eitschman@nebelreich.de> Signed-off-by: Linus Walleij <linusw@kernel.org> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260505-gemini-ethernet-fix-v2-1-997c31d06079@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
selftests: mptcp: pm: restrict 'unknown' check to pm_nl_ctl
When pm_netlink.sh is executed with '-i', 'ip mptcp' is used instead of
'pm_nl_ctl'. IPRoute2 doesn't support the 'unknown' flag, which has only
been added to 'pm_nl_ctl' for this specific check: to ensure that the
kernel ignores such unsupported flag.
No reason to add this flag to 'ip mptcp'. Then, this check should be
skipped when 'ip mptcp' is used.
Using '${?}' inside the if-statement to check the returned value from
the command that was evaluated as part of the if-statement is not
correct: here, '${?}' will be linked to the previous instruction, not
the one that is expected here (${cmd}).
Instead, simply mark the error, except if an error is expected. If
that's the case, 1 can be passed as the 4th argument of this helper.
Three checks from pm_netlink.sh expect an error.
While at it, improve the error message when the command unexpectedly
fails or succeeds.
Note that we could expect a specific returned value, but the checks
currently expecting an error can be used with 'ip mptcp' or 'pm_nl_ctl',
and these two tools don't return the same error code.
When looking at the maximum RTO amongst the subflows, inactive subflows
were taken into account: that includes stale ones, and the initial one
if it has been already been closed.
Unusable subflows are now simply skipped. Stale ones are used as an
alternative: if there are only stale ones, to take their maximum RTO and
avoid to eventually fallback to net.mptcp.add_addr_timeout, which is set
to 2 minutes by default.
Fixes: 30549eebc4d8 ("mptcp: make ADD_ADDR retransmission timeout adaptive") Cc: stable@vger.kernel.org Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20260505-net-mptcp-pm-fixes-7-1-rc3-v1-7-fca8091060a4@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When an ADD_ADDR needs to be retransmitted and another one has already
been prepared -- e.g. multiple ADD_ADDRs have been sent in a row and
need to be retransmitted later -- this additional retransmission will
need to wait.
In this case, the timer was reset to TCP_RTO_MAX / 8, which is ~15
seconds. This delay is unnecessary long: it should just be rescheduled
at the next opportunity, e.g. after the retransmission timeout.
Without this modification, some issues can be seen from time to time in
the selftests when multiple ADD_ADDRs are sent, and the host takes time
to process them, e.g. the "signal addresses, ADD_ADDR timeout" MPTCP
Join selftest, especially with a debug kernel config.
Note that on older kernels, 'timeout' is not available. It should be
enough to replace it by one second (HZ).
When an ADD_ADDR is retransmitted, the sk is held in sk_reset_timer(),
and released at the end.
If at that moment, it was the last reference being held, the sk would
not be freed. sock_put() should then be called instead of __sock_put().
But that's not enough: if it is the last reference, sock_put() will call
sk_free(), which will end up calling sk_stop_timer_sync() on the same
timer, and waiting indefinitely to finish. So it is needed to mark that
the timer is done at the end of the timer handler when it has not been
rescheduled, not to call sk_stop_timer_sync() on "itself".
mptcp: pm: ADD_ADDR rtx: always decrease sk refcount
When an ADD_ADDR is retransmitted, the sk is held in sk_reset_timer().
It should then be released in all cases at the end.
Some (unlikely) checks were returning directly instead of calling
sock_put() to decrease the refcount. Jump to a new 'exit' label to call
__sock_put() (which will become sock_put() in the next commit) to fix
this potential leak.
While at it, drop the '!msk' check which cannot happen because it is
never reset, and explicitly mark the remaining one as "unlikely".
This mptcp_pm_add_timer() helper is executed as a timer callback in
softirq context. To avoid any data races, the socket lock needs to be
held with bh_lock_sock().
If the socket is in use, retry again soon after, similar to what is done
with the keepalive timer.
ADD_ADDR can be sent for the ID 0, which corresponds to the local
address and port linked to the initial subflow.
Indeed, this address could be removed, and re-added later on, e.g. what
is done in the "delete re-add signal" MPTCP Join selftests. So no reason
to ignore it.
mptcp: pm: kernel: correctly retransmit ADD_ADDR ID 0
When adding the ADD_ADDR to the list, the address including the IP, port
and ID are copied. On the other hand, when the endpoint corresponds to
the one from the initial subflow, the ID is set to 0, as specified by
the MPTCP protocol.
The issue is that the ID was reset after having copied the ID in the
ADD_ADDR entry. So the retransmission was done, but using a different ID
than the initial one.
Eric Dumazet [Tue, 5 May 2026 13:32:33 +0000 (13:32 +0000)]
inetpeer: add a missing read_seqretry() in inet_getpeer()
When performing a lockless lookup over the inet_peer rbtree,
if a matching node is found, inet_getpeer() returns it immediately
without validating the seqlock sequence.
This missing check introduces a race condition:
Trigger Path: When a host receives an incoming fragmented IPv4 packet,
ip4_frag_init() (in net/ipv4/ip_fragment.c) calls inet_getpeer_v4()
to track the peer.
The Race: If the packet is from a new source IP, CPU A acquires the
write_seqlock, allocates a new inet_peer node (p), sets its IP address
(daddr), and links it to the rbtree (rb_link_node).
Uninitialized Access: Due to the lack of memory barriers between
rb_link_node and the initialization of the rest of the struct
(like refcount_set(&p->refcnt, 1)), CPU A can make the node visible
to readers before its refcnt is initialized.
This is especially true on weakly-ordered architectures like ARM64
where the CPU can reorder the memory stores.
Lockless Reader: Concurrently, CPU B processes a second fragmented packet
from the same source IP. CPU B does a lockless lookup, finds the newly
inserted node, and returns it immediately.
Use-After-Free (UAF): CPU B reads p->refcnt as uninitialized garbage
(left over from previous kmalloc-128/192 allocations).
If the garbage is > 0, refcount_inc_not_zero(&p->refcnt) succeeds.
CPU A then executes refcount_set(&p->refcnt, 1), overwriting CPU B's increment.
When CPU B finishes with the fragment queue, it calls inet_putpeer(),
which drops the refcount to 0 and frees the node via RCU.
The node is now freed but remains linked in the rbtree,
resulting in a Use-After-Free in the rbtree.
Fixes: b145425f269a ("inetpeer: remove AVL implementation in favor of RB tree") Reported-by: Damiano Melotti <melotti@google.com> Signed-off-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20260505133233.3039575-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
====================
netdevsim: psp: fix init and uninit bugs
This series has three fixes. The first is a straightforward NULL
pointer dereference that is reachable by creating and destroying some
vfs on a kernel with INET_PSP enabled.
The last two patches deal with nsim_psp_rereg_write(), which is a
debugfs handler that reregisters netdevsim's psp_dev without
aquiescing and disabling tx/rx processing. This was added to enable
some tests in psp.py where a psp device is unregistered while it still
referenced by tcp socket state.
There are two issues with this code:
1. Calls to nsim_psp_uninit() are not properly serialized
2. netdevsim's psp_dev refcount can be released while nsim_do_psp() is
reading from it.
====================
Daniel Zahka [Tue, 5 May 2026 10:42:25 +0000 (03:42 -0700)]
netdevsim: psp: rcu protect psp_dev reference
There are two issues with the way psp_dev is used in nsim_do_psp():
1. There is no check for IS_ERR() on the peers psp_dev, before
dereferencing.
2. The refcount on this psp_dev can be dropped by
nsim_psp_rereg_write()
To fix this, we can make netdevsim's reference to its psp_dev an rcu
reference, and then nsim_do_psp() can read the fields it needs from an
rcu critical section.
Fixes: f857478d6206 ("netdevsim: a basic test PSP implementation") Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20260505-psd-rcu-v1-3-a8f69ec1ab96@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Daniel Zahka [Tue, 5 May 2026 10:42:24 +0000 (03:42 -0700)]
netdevsim: psp: serialize calls to nsim_psp_uninit()
The debugfs write handler, nsim_psp_rereg_write(), can race against
nsim_destroy() and against itself, causing nsim_psp_uninit() to run
more than once concurrently. Two complementary changes serialize all
callers:
1. Delete the psp_rereg debugfs file from nsim_psp_uninit() before
doing the actual teardown. debugfs_remove() drains any in-flight
writers and prevents new ones from starting.
2. Add a mutex around the body of nsim_psp_rereg_write() so that two
concurrent userspace writers cannot both enter the teardown path
at once.
The teardown work itself is moved into a new __nsim_psp_uninit() that
the rereg handler calls under the mutex, while the public
nsim_psp_uninit() wraps it with the debugfs_remove()/mutex_destroy()
pair so nsim_destroy() doesn't have to know about the psp internals.
Fixes: f857478d6206 ("netdevsim: a basic test PSP implementation") Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20260505-psd-rcu-v1-2-a8f69ec1ab96@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Daniel Zahka [Tue, 5 May 2026 10:42:23 +0000 (03:42 -0700)]
netdevsim: psp: only call nsim_psp_uninit() on PFs
VFs go through nsim_init_netdevsim_vf() which never calls
nsim_psp_init(), so ns->psp.dev stays NULL. nsim_psp_uninit() guards
with !IS_ERR(ns->psp.dev), so destroying a VF reaches
psp_dev_unregister(NULL) and dereferences NULL on the first
mutex_lock(&psd->lock):
Fixes: f857478d6206 ("netdevsim: a basic test PSP implementation") Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20260505-psd-rcu-v1-1-a8f69ec1ab96@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
1. Fix an IPv6 encapsulation error path that leaked route references
when UDPv6 ESP decapsulation resolved to an error route.
From Yilin Zhu.
2. Fix AH with ESN on async crypto paths by accounting for the extra
high-order sequence number when reconstructing the temporary
authentication layout in the completion callbacks.
From Michael Bomarito.
3. Fix XFRM output so it does not overwrite already-correct inner header
pointers when a tunnel layer such as VXLAN has already saved them.
The fix comes with new selftests. From Cosmin Ratiu.
4. Add the missing native payload size entry for XFRM_MSG_MAPPING in the
compat translation path. From Ruijie Li.
5. Harden __xfrm_state_delete() against repeated or inconsistent unhashing
of state list nodes by keying the removal on actual list membership and
using delete-and-init helpers. From Michal Kosiorek.
6. Prevent ESP from decrypting shared splice-backed skb fragments in place
by marking UDP splice frags as shared and forcing copy-on-write in ESP
input when needed. From Kuan-Ting Chen.
* tag 'ipsec-2026-05-05' of git://git.kernel.org/pub/scm/linux/kernel/git/klassert/ipsec:
xfrm: esp: avoid in-place decrypt on shared skb frags
xfrm: defensively unhash xfrm_state lists in __xfrm_state_delete
xfrm: provide message size for XFRM_MSG_MAPPING
xfrm: Don't clobber inner headers when already set
tools/selftests: Add a VXLAN+IPsec traffic test
tools/selftests: Use a sensible timeout value for iperf3 client
xfrm: ah: account for ESN high bits in async callbacks
ipv6: xfrm6: release dst on error in xfrm6_rcv_encap()
====================
Jakub Kicinski [Wed, 6 May 2026 23:10:02 +0000 (16:10 -0700)]
Merge tag 'ovpn-net-20260504' of https://github.com/OpenVPN/ovpn-net-next
Antonio Quartulli says:
====================
Includes changes:
* ensure MAC header offset is reset before delivering packet
* ensure gro_cells_receive() and dstats_dev_add() are called
with BH disabled
* reduce ping count in selftest to ensure it completes within
timeout
* tag 'ovpn-net-20260504' of https://github.com/OpenVPN/ovpn-net-next:
selftests: ovpn: reduce ping count in test.sh
ovpn: ensure packet delivery happens with BH disabled
ovpn: reset MAC header before passing skb up
====================
Bluetooth: HIDP: serialise l2cap_unregister_user via hidp_session_sem
Commit dbf666e4fc9b ("Bluetooth: HIDP: Fix possible UAF") made
hidp_session_remove() drop the L2CAP reference and set
session->conn = NULL once the session is considered removed, and
added a bare if (session->conn) guard around the kthread-exit
l2cap_unregister_user() call in hidp_session_thread(). The sibling
ioctl site in hidp_connection_del() still reads session->conn
unlocked and unguarded, and the kthread-exit guard itself is a
lockless double-read.
hidp_session_find() drops hidp_session_sem before returning, so
hidp_session_remove() can null session->conn between the lookup and
the call in hidp_connection_del(). Worse, since commit 752a6c9596dd
("Bluetooth: L2CAP: Fix use-after-free in l2cap_unregister_user")
takes mutex_lock(&conn->lock) inside l2cap_unregister_user(), a
stale non-NULL snapshot also UAFs on conn->lock. v1 only added an
if (session->conn) guard at the ioctl site, which doesn't address
either race; Luiz suggested snapshotting session->conn under the
sem and clearing it before the call.
Taking hidp_session_sem across l2cap_unregister_user() would be
wrong: l2cap_conn_del() already establishes the lock order
conn->lock -> hidp_session_sem
via l2cap_unregister_all_users() -> user->remove ==
hidp_session_remove(), so taking hidp_session_sem before conn->lock
would AB/BA deadlock.
Factor a helper hidp_session_unregister_conn() that under
down_write(&hidp_session_sem) snapshots session->conn and clears
the member, then outside the sem calls l2cap_unregister_user() and
l2cap_conn_put() on the snapshot. Call it from both
hidp_connection_del() and hidp_session_thread()'s exit path. At
most one consumer wins the write-sem; later callers observe
session->conn == NULL and skip the unregister and put, so the
reference hidp_session_new() took via l2cap_conn_get() is consumed
exactly once. session_free() already tolerates a NULL session->conn.
Fixes: dbf666e4fc9b ("Bluetooth: HIDP: Fix possible UAF") Suggested-by: Luiz Augusto von Dentz <luiz.dentz@gmail.com> Link: https://lore.kernel.org/all/20260422011437.176643-1-michael.bommarito@gmail.com/ Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
sizeof(conn->num_bis) is wrong - it would make sense to either use
conn->num_bis (before setting that to 0) or sizeof(conn->bis).
Fix it by using sizeof(conn->bis), the least intrusive change.
Luckily, nothing actually depends on this memset() working properly:
Nothing seems to ever read from conn->bis beyond conn->num_bis, and when
conn->num_bis is increased, the corresponding elements of conn->bis are
initialized. So I think this line could also just be removed.
This is a purely theoretical fix and should have no impact on actual
behavior.
Fixes: 42ecf1947135 ("Bluetooth: ISO: Do not emit LE BIG Create Sync if previous is pending") Signed-off-by: Jann Horn <jannh@google.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: RFCOMM: pull credit byte with skb_pull_data()
rfcomm_recv_data() treats the first payload byte as a credit field when
the UIH frame carries PF and credit-based flow control is enabled.
After the header has been stripped, the PF/CFC path consumes that byte
with a direct skb->data dereference followed by skb_pull(). A malformed
short frame can reach this path without a byte available.
Use skb_pull_data() so the length check and pull happen together before
the returned credit byte is consumed.
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
virtbt_rx_handle() reads the leading pkt_type byte from the RX skb
and forwards the remainder to hci_recv_frame() for every
event/ACL/SCO/ISO type, without checking that the remaining payload
is at least the fixed HCI header for that type.
After the preceding patch bounds the backend-supplied used.len to
[1, VIRTBT_RX_BUF_SIZE], a one-byte completion still reaches
hci_recv_frame() with skb->len already pulled to 0. If the byte
happened to be HCI_ACLDATA_PKT, the ACL-vs-ISO classification
fast-path in hci_dev_classify_pkt_type() dereferences
hci_acl_hdr(skb)->handle whenever the HCI device has an active
CIS_LINK, BIS_LINK, or PA_LINK connection, reading two bytes of
uninitialized RX-buffer data. The same hazard exists for every
packet type the driver accepts because none of the switch cases in
virtbt_rx_handle() check skb->len against the per-type minimum HCI
header size before handing the frame to the core.
After stripping pkt_type, require skb->len to cover the fixed
header size for the selected type (event 2, ACL 4, SCO 3, ISO 4)
before calling hci_recv_frame(); drop ratelimited otherwise.
Unknown pkt_type values still take the original kfree_skb() default
path.
Use bt_dev_err_ratelimited() because both the length and pkt_type
values come from an untrusted backend that can otherwise flood the
kernel log.
Fixes: 160fbcf3bfb9 ("Bluetooth: virtio_bt: Use skb_put to set length") Cc: stable@vger.kernel.org Cc: Soenke Huster <soenke.huster@eknoes.de> Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: virtio_bt: clamp rx length before skb_put
virtbt_rx_work() calls skb_put(skb, len) where len comes directly
from virtqueue_get_buf() with no validation against the buffer we
posted to the device. The RX skb is allocated in virtbt_add_inbuf()
and exposed to virtio as exactly 1000 bytes via sg_init_one().
Checking len against skb_tailroom(skb) is not sufficient because
alloc_skb() can leave more tailroom than the 1000 bytes actually
handed to the device. A malicious or buggy backend can therefore
report used.len between 1001 and skb_tailroom(skb), causing skb_put()
to include uninitialized kernel heap bytes that were never written by
the device.
The same path also accepts len == 0, in which case skb_put(skb, 0)
leaves the skb empty but virtbt_rx_handle() still reads the pkt_type
byte from skb->data, consuming uninitialized memory.
Define VIRTBT_RX_BUF_SIZE once and reuse it in alloc_skb() and
sg_init_one(), and gate virtbt_rx_work() on that same constant so
the bound checked matches the buffer actually exposed to the device.
Reject used.len == 0 in the same gate so an empty completion can
no longer reach virtbt_rx_handle().
Use bt_dev_err_ratelimited() because the length value comes from an
untrusted backend that can otherwise flood the kernel log.
Same class of bug as commit c04db81cd028 ("net/9p: Fix buffer
overflow in USB transport layer"), which hardened the USB 9p
transport against unchecked device-reported length.
Fixes: 160fbcf3bfb9 ("Bluetooth: virtio_bt: Use skb_put to set length") Cc: stable@vger.kernel.org Cc: Soenke Huster <soenke.huster@eknoes.de> Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: btmtk: validate WMT event SKB length before struct access
btmtk_usb_hci_wmt_sync() casts the WMT event response SKB data to
struct btmtk_hci_wmt_evt (7 bytes) and struct btmtk_hci_wmt_evt_funcc
(9 bytes) without first checking that the SKB contains enough data.
A short firmware response causes out-of-bounds reads from SKB tailroom.
Use skb_pull_data() to validate and advance past the base WMT event
header. For the FUNC_CTRL case, pull the additional status field bytes
before accessing them.
Fixes: d019930b0049 ("Bluetooth: btmtk: move btusb_mtk_hci_wmt_sync to btmtk.c") Cc: stable@vger.kernel.org Signed-off-by: Tristan Madani <tristan@talencesecurity.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: ISO: Fix data-race on iso_pi(sk) in socket and HCI event paths
Several iso_pi(sk) fields (qos, qos_user_set, bc_sid, base, base_len,
sync_handle, bc_num_bis) are written under lock_sock in
iso_sock_setsockopt() and iso_sock_bind(), but read and written under
hci_dev_lock only in two other paths:
- iso_connect_bis() / iso_connect_cis(), invoked from connect(2),
read qos/base/bc_sid and reset qos to default_qos on the
qos_user_set validation failure -- all without lock_sock.
- iso_connect_ind(), invoked from hci_rx_work, writes sync_handle,
bc_sid, qos.bcast.encryption, bc_num_bis, base and base_len on
PA_SYNC_ESTABLISHED / PAST_RECEIVED / BIG_INFO_ADV_REPORT /
PER_ADV_REPORT events. The BIG_INFO handler additionally passes
&iso_pi(sk)->qos together with sync_handle / bc_num_bis / bc_bis
to hci_conn_big_create_sync() while setsockopt may be mutating
them.
Acquire lock_sock around the affected accesses in both paths.
The locking order hci_dev_lock -> lock_sock matches the existing
iso_conn_big_sync() precedent, whose comment documents the same
requirement for hci_conn_big_create_sync(). The HCI connect/bind
helpers do not wait for command completion -- they enqueue work via
hci_cmd_sync_queue{,_once}() / hci_le_create_cis_pending() and
return -- so the added hold time is comparable to iso_conn_big_sync().
KCSAN report:
BUG: KCSAN: data-race in iso_connect_cis / iso_sock_setsockopt
read to 0xffffa3ae8ce3cdc8 of 1 bytes by task 335 on cpu 0:
iso_connect_cis+0x49f/0xa20
iso_sock_connect+0x60e/0xb40
__sys_connect_file+0xbd/0xe0
__sys_connect+0xe0/0x110
__x64_sys_connect+0x40/0x50
x64_sys_call+0xcad/0x1c60
do_syscall_64+0x133/0x590
entry_SYSCALL_64_after_hwframe+0x77/0x7f
write to 0xffffa3ae8ce3cdc8 of 60 bytes by task 334 on cpu 1:
iso_sock_setsockopt+0x69a/0x930
do_sock_setsockopt+0xc3/0x170
__sys_setsockopt+0xd1/0x130
__x64_sys_setsockopt+0x64/0x80
x64_sys_call+0x1547/0x1c60
do_syscall_64+0x133/0x590
entry_SYSCALL_64_after_hwframe+0x77/0x7f
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 UID: 0 PID: 334 Comm: iso_setup_race Not tainted 7.0.0-10949-g8541d8f725c6 #44 PREEMPT(lazy)
The iso_connect_ind() races were found by inspection.
Fixes: ccf74f2390d6 ("Bluetooth: Add BTPROTO_ISO socket type") Signed-off-by: SeungJu Cheon <suunj1331@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: ISO: Fix data-race on dst in iso_sock_connect()
iso_sock_connect() copies the destination address into
iso_pi(sk)->dst under lock_sock, then releases the lock and reads
it back with bacmp() to decide between the CIS and BIS connect
paths:
if (bacmp(&iso_pi(sk)->dst, BDADDR_ANY)) // <- no lock held
This read after release_sock() races with any concurrent write to
iso_pi(sk)->dst on the same socket.
Fix by reading the destination address directly from the local
sockaddr argument (sa->iso_bdaddr) instead of iso_pi(sk)->dst.
Since sa is a function-local argument, reading it requires no
locking and avoids the race.
This patch addresses only the bacmp() race in iso_sock_connect();
other unprotected iso_pi(sk) accesses are fixed separately in the
next patch.
KCSAN report:
BUG: KCSAN: data-race in memcmp+0x39/0xb0
race at unknown origin, with read to 0xffff8f96ea66dde3 of 1 bytes by task 549 on cpu 1:
memcmp+0x39/0xb0
iso_sock_connect+0x275/0xb40
__sys_connect_file+0xbd/0xe0
__sys_connect+0xe0/0x110
__x64_sys_connect+0x40/0x50
x64_sys_call+0xcad/0x1c60
do_syscall_64+0x133/0x590
entry_SYSCALL_64_after_hwframe+0x77/0x7f
value changed: 0x00 -> 0xee
Reported by Kernel Concurrency Sanitizer on:
CPU: 1 UID: 0 PID: 549 Comm: iso_race_combin Not tainted 7.0.0-08391-g1d51b370a0f8 #40 PREEMPT(lazy)
Fixes: ccf74f2390d6 ("Bluetooth: Add BTPROTO_ISO socket type") Signed-off-by: SeungJu Cheon <suunj1331@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: hci_uart: Fix NULL deref in recv callbacks when priv is uninitialized
When a fault is injected during hci_uart line discipline setup, the
proto open() callback may fail leaving hu->priv as NULL. A subsequent
TIOCSTI ioctl can trigger the recv() callback before priv is
initialized, causing a NULL pointer dereference.
Fix all four affected HCI UART protocol drivers by adding a NULL check
on hu->priv at the start of their recv() callbacks: h4, h5, ath and
bcsp.
Reported-by: syzbot+ff30eeab8e07b37d524e@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=ff30eeab8e07b37d524e Signed-off-by: Aurelien DESBRIERES <aurelien@hackers.camp> Assisted-by: Claude:claude-sonnet-4-6 Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: btintel_pcie: treat boot stage bit 12 as warning
CSR boot stage register bit 12 is documented as a device warning,
not a fatal error. Rename the bit definition accordingly and stop
including it in btintel_pcie_in_error().
This keeps warning-only boot stage values from being classified as
errors while preserving abort-handler state as the actual error
condition.
Fixes: 190377500fde ("Bluetooth: btintel_pcie: Dump debug registers on error") Signed-off-by: Kiran K <kiran.k@intel.com> Signed-off-by: Sai Teja Aluvala <aluvala.sai.teja@intel.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sat, 18 Apr 2026 15:41:12 +0000 (18:41 +0300)]
Bluetooth: SCO: hold sk properly in sco_conn_ready
sk deref in sco_conn_ready must be done either under conn->lock, or
holding a refcount, to avoid concurrent close. conn->sk and parent sk is
currently accessed without either, and without checking parent->sk_state:
[Task 1] [Task 2]
sco_sock_release
sco_conn_ready
sk = conn->sk
lock_sock(sk)
conn->sk = NULL
lock_sock(sk)
release_sock(sk)
sco_sock_kill(sk)
UAF on sk deref
and similarly for access to sco_get_sock_listen() return value.
Fix possible UAF by holding sk refcount in sco_conn_ready() and making
sco_get_sock_listen() increase refcount. Also recheck after lock_sock
that the socket is still valid. Adjust conn->sk locking so it's
protected also by lock_sock() of the associated socket if any.
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>
l2cap_conn_ready() [hdev->lock via hci_cb_list_lock]
...
mutex_lock(&conn->lock)
This is a classic AB/BA deadlock which lockdep reports as a circular
locking dependency when connecting a BLE MIDI keyboard (Carry-On FC-49).
Fix this by making hci_le_conn_update() defer the HCI command through
hci_cmd_sync_queue() so it no longer needs to take hdev->lock in the
caller context. The sync callback uses __hci_cmd_sync_status_sk() to
wait for the HCI_EV_LE_CONN_UPDATE_COMPLETE event, then updates the
stored connection parameters (hci_conn_params) and notifies userspace
(mgmt_new_conn_param) only after the controller has confirmed the update.
A reference on hci_conn is held via hci_conn_get()/hci_conn_put() for
the lifetime of the queued work to prevent use-after-free, and
hci_conn_valid() is checked before proceeding in case the connection was
removed while the work was pending. The hci_dev_lock is held across
hci_conn_valid() and all conn field accesses to prevent a concurrent
disconnect from invalidating the connection mid-use.
Fixes: f044eb0524a0 ("Bluetooth: Store latency and supervision timeout in connection params") Signed-off-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Dudu Lu [Wed, 15 Apr 2026 10:43:55 +0000 (18:43 +0800)]
Bluetooth: l2cap: fix MPS check in l2cap_ecred_reconf_req
The L2CAP specification states that if more than one channel is being
reconfigured, the MPS shall not be decreased. The current check has
two issues:
1) The comparison uses >= (greater-than-or-equal), which incorrectly
rejects reconfiguration requests where the MPS stays the same.
Since the spec says MPS "shall be greater than or equal to the
current MPS", only a strict decrease (remote_mps > mps) should be
rejected. Keeping the same MPS is valid.
2) The multi-channel guard uses `&& i` (loop index) to approximate
"more than one channel", but this incorrectly allows MPS decrease
for the first channel (i==0) even when multiple channels are being
reconfigured. Replace with `&& num_scid > 1` which correctly
checks whether the request covers more than one channel.
Fixes: 7accb1c4321a ("Bluetooth: L2CAP: Fix invalid response to L2CAP_ECRED_RECONF_REQ") Signed-off-by: Dudu Lu <phx0fer@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Dudu Lu [Wed, 15 Apr 2026 09:39:53 +0000 (17:39 +0800)]
Bluetooth: bnep: fix incorrect length parsing in bnep_rx_frame() extension handling
In bnep_rx_frame(), the BNEP_FILTER_NET_TYPE_SET and
BNEP_FILTER_MULTI_ADDR_SET extension header parsing has two bugs:
1) The 2-byte length field is read with *(u16 *)(skb->data + 1), which
performs a native-endian read. The BNEP protocol specifies this field
in big-endian (network byte order), and the same file correctly uses
get_unaligned_be16() for the identical fields in
bnep_ctrl_set_netfilter() and bnep_ctrl_set_mcfilter().
2) The length is multiplied by 2, but unlike BNEP_SETUP_CONN_REQ where
the length byte counts UUID pairs (requiring * 2 for two UUIDs per
entry), the filter extension length field already represents the total
data size in bytes. This is confirmed by bnep_ctrl_set_netfilter()
which reads the same field as a byte count and divides by 4 to get
the number of filter entries.
The bogus * 2 means skb_pull advances twice as far as it should,
either dropping valid data from the next header or causing the pull
to fail entirely when the doubled length exceeds the remaining skb.
Fix by splitting the pull into two steps: first use skb_pull_data() to
safely pull and validate the 3-byte fixed header (ctrl type + length),
then pull the variable-length data using the properly decoded length.
Fixes: bf8b9a9cb77b ("Bluetooth: bnep: Add support to extended headers of control frames") Signed-off-by: Dudu Lu <phx0fer@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Bluetooth: hci_event: Fix OOB read and infinite loop in hci_le_create_big_complete_evt
hci_le_create_big_complete_evt() iterates over BT_BOUND connections for
a BIG handle using a while loop, accessing ev->bis_handle[i++] on each
iteration. However, there is no check that i stays within ev->num_bis
before the array access.
When a controller sends a LE_Create_BIG_Complete event with fewer
bis_handle entries than there are BT_BOUND connections for that BIG,
or with num_bis=0, the loop reads beyond the valid bis_handle[] flex
array into adjacent heap memory. Since the out-of-bounds values
typically exceed HCI_CONN_HANDLE_MAX (0x0EFF), hci_conn_set_handle()
rejects them and the connection remains in BT_BOUND state. The same
connection is then found again by hci_conn_hash_lookup_big_state(),
creating an infinite loop with hci_dev_lock held.
Fix this by terminating the BIG if in case not all BIS could be setup
properly.
Fixes: a0bfde167b50 ("Bluetooth: ISO: Add support for connecting multiple BISes") Cc: stable@vger.kernel.org Signed-off-by: ZhiTao Ou <hkbinbinbin@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
David Carlier [Sun, 12 Apr 2026 20:29:16 +0000 (21:29 +0100)]
Bluetooth: hci_conn: fix potential UAF in create_big_sync
Add hci_conn_valid() check in create_big_sync() to detect stale
connections before proceeding with BIG creation. Handle the
resulting -ECANCELED in create_big_complete() and re-validate the
connection under hci_dev_lock() before dereferencing, matching the
pattern used by create_le_conn_complete() and create_pa_complete().
Keep the hci_conn object alive across the async boundary by taking
a reference via hci_conn_get() when queueing create_big_sync(), and
dropping it in the completion callback. The refcount and the lock
are complementary: the refcount keeps the object allocated, while
hci_dev_lock() serializes hci_conn_hash_del()'s list_del_rcu() on
hdev->conn_hash, as required by hci_conn_del().
hci_conn_put() is called outside hci_dev_unlock() so the final put
(which resolves to kfree() via bt_link_release) does not run under
hdev->lock, though the release path would be safe either way.
Without this, create_big_complete() would unconditionally
dereference the conn pointer on error, causing a use-after-free
via hci_connect_cfm() and hci_conn_del().
Fixes: eca0ae4aea66 ("Bluetooth: Add initial implementation of BIS connections") Cc: stable@vger.kernel.org Co-developed-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com> Signed-off-by: David Carlier <devnexen@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Sun, 12 Apr 2026 18:47:42 +0000 (21:47 +0300)]
Bluetooth: SCO: fix sleeping under spinlock in sco_conn_ready
sco_conn_ready calls sleeping functions under conn->lock spinlock.
The critical section can be reduced: conn->hcon is modified only with
hdev->lock held. It is guaranteed to be held in sco_conn_ready, so
conn->lock is not needed to guard it.
Move taking conn->lock after lock_sock(parent). This also follows the
lock ordering lock_sock() > conn->lock elsewhere in the file.
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>
Jakub Kicinski [Wed, 6 May 2026 14:29:31 +0000 (07:29 -0700)]
Merge tag 'wireless-2026-05-06' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless
Johannes Berg says:
====================
Quite a number of fixes now:
- mac80211
- remove HT NSS validation to work with broken APs
(with a kunit fix now)
- remove 'static' that could cause races
- check station link lookup before further processing
- fix use-after-free due to delete in list iteration
- remove AP station on assoc failures to fix crashes
- ath12k
- fix OF node refcount imbalance
- fix queue flush ("REO update") in MLO
- fix RCU assert
- ath12k:
- fix Kconfig with POWER_SEQUENCING
- fix WMI buffer leaks on error conditions
- don't use uninitialized stack data when processing RSSI events
- fix logic for determining the peer ID in the RX path
- ath5k: fix a potential stack buffer overwrite
- rsi: fix thread lifetime race
- brcmfmac: fix potential UAF
- nl80211:
- stricter permissions/checks for PMK and netns
- fix netlink policy vs. code type confusion
- cw1200: revert a broken locking change
- various fixes to not trust values from firmware
* tag 'wireless-2026-05-06' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless: (25 commits)
wifi: nl80211: re-check wiphy netns in nl80211_prepare_wdev_dump() continuation
wifi: nl80211: require CAP_NET_ADMIN over the target netns in SET_WIPHY_NETNS
wifi: nl80211: fix NL80211_PMSR_FTM_REQ_ATTR_FTMS_PER_BURST usage
wifi: mac80211: remove station if connection prep fails
wifi: mac80211: use safe list iteration in radar detect work
wifi: libertas: notify firmware load wait on disconnect
wifi: ath5k: do not access array OOB
wifi: ath12k: fix peer_id usage in normal RX path
wifi: ath12k: initialize RSSI dBm conversion event state
wifi: ath12k: fix leak in some ath12k_wmi_xxx() functions
wifi: cw1200: Revert "Fix locking in error paths"
wifi: mac80211: tests: mark HT check strict
wifi: rsi: fix kthread lifetime race between self-exit and external-stop
wifi: mac80211: drop stray 'static' from fast-RX rx_result
wifi: mac80211: check ieee80211_rx_data_set_link return in pubsta MLO path
wifi: nl80211: require admin perm on SET_PMK / DEL_PMK
wifi: libertas: fix integer underflow in process_cmdrequest()
wifi: b43legacy: enforce bounds check on firmware key index in RX path
wifi: b43: enforce bounds check on firmware key index in b43_rx()
wifi: brcmfmac: Fix potential use-after-free issue when stopping watchdog task
...
====================
Maoyi Xie [Wed, 6 May 2026 06:48:54 +0000 (14:48 +0800)]
wifi: nl80211: re-check wiphy netns in nl80211_prepare_wdev_dump() continuation
NL80211_CMD_GET_SCAN is implemented as a multi-call dumpit. The first
invocation of nl80211_prepare_wdev_dump() validates the requested wdev
against the caller's netns via __cfg80211_wdev_from_attrs(). Subsequent
invocations look up the same wiphy by its global index and do not check
that the wiphy is still in the caller's netns.
Add the same filter to the continuation path. If the wiphy's netns no
longer matches the caller's, return -ENODEV and the netlink dump
machinery terminates the walk cleanly.
Maoyi Xie [Wed, 6 May 2026 06:48:53 +0000 (14:48 +0800)]
wifi: nl80211: require CAP_NET_ADMIN over the target netns in SET_WIPHY_NETNS
NL80211_CMD_SET_WIPHY_NETNS dispatches with GENL_UNS_ADMIN_PERM, which
verifies that the caller has CAP_NET_ADMIN for the source netns. It
doesn't verify that the caller has CAP_NET_ADMIN over the target netns
selected by NL80211_ATTR_NETNS_FD or NL80211_ATTR_PID.
This diverges from the convention enforced in
net/core/rtnetlink.c::rtnl_get_net_ns_capable():
/* For now, the caller is required to have CAP_NET_ADMIN in
* the user namespace owning the target net ns.
*/
if (!sk_ns_capable(sk, net->user_ns, CAP_NET_ADMIN))
return ERR_PTR(-EACCES);
A user with CAP_NET_ADMIN in their own user namespace can therefore
push a wiphy into an arbitrary netns (including init_net) over which
they have no privilege.
Mirror the rtnetlink convention by requiring CAP_NET_ADMIN in the
target netns before calling cfg80211_switch_netns().
This is documented as a u8 and has a policy of NLA_U8, but uses
nla_get_u32() which means it's completely broken on big-endian.
Fix it to use nla_get_u8().
Johannes Berg [Tue, 5 May 2026 13:15:34 +0000 (15:15 +0200)]
wifi: mac80211: remove station if connection prep fails
If connection preparation fails for MLO connections, then the
interface is completely reset to non-MLD. In this case, we must
not keep the station since it's related to the link of the vif
being removed. Delete an existing station. Any "new_sta" is
already being removed, so that doesn't need changes.
This fixes a use-after-free/double-free in debugfs if that's
enabled, because a vif going from MLD (and to MLD, but that's
not relevant here) recreates its entire debugfs.
Jason Xing [Sat, 2 May 2026 20:07:22 +0000 (23:07 +0300)]
xsk: fix u64 descriptor address truncation on 32-bit architectures
In copy mode TX, xsk_skb_destructor_set_addr() stores the 64-bit
descriptor address into skb_shinfo(skb)->destructor_arg (void *) via a
uintptr_t cast:
On 32-bit architectures uintptr_t is 32 bits, so the upper 32 bits of
the descriptor address are silently dropped. In XDP_ZEROCOPY unaligned
mode the chunk offset is encoded in bits 48-63 of the descriptor
address (XSK_UNALIGNED_BUF_OFFSET_SHIFT = 48), meaning the offset is
lost entirely. The completion queue then returns a truncated address to
userspace, making buffer recycling impossible.
Fix this by handling the 32-bit case directly in
xsk_skb_destructor_set_addr(): when !CONFIG_64BIT, allocate an
xsk_addrs struct (the same path already used for multi-descriptor
SKBs) to store the full u64 address. The existing tagged-pointer logic
in xsk_skb_destructor_is_addr() stays unchanged: slab pointers returned
from kmem_cache_zalloc() are always word-aligned and therefore have
bit 0 clear, which correctly identifies them as a struct pointer
rather than an inline tagged address on every architecture.
Factor the shared kmem_cache_zalloc + destructor_arg assignment into
__xsk_addrs_alloc() and add a wrapper xsk_addrs_alloc() that handles
the inline-to-list upgrade (is_addr check + get_addr + num_descs = 1).
The three former open-coded kmem_cache_zalloc call sites now reduce to
a single call each.
Propagate the -ENOMEM from xsk_skb_destructor_set_addr() through
xsk_skb_init_misc() so the caller can clean up the skb via kfree_skb()
before skb->destructor is installed.
The overhead is one extra kmem_cache_zalloc per first descriptor on
32-bit only; 64-bit builds are completely unchanged.
Closes: https://lore.kernel.org/all/20260419045824.D9E5EC2BCAF@smtp.kernel.org/ Fixes: 0ebc27a4c67d ("xsk: avoid data corruption on cq descriptor number") Signed-off-by: Jason Xing <kernelxing@tencent.com> Acked-by: Stanislav Fomichev <sdf@fomichev.me> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-9-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:21 +0000 (23:07 +0300)]
xsk: fix xsk_addrs slab leak on multi-buffer error path
When xsk_build_skb() / xsk_build_skb_zerocopy() sees the first
continuation descriptor, it promotes destructor_arg from an inlined
address to a freshly allocated xsk_addrs (num_descs = 1). The counter
is bumped to >= 2 only at the very end of a successful build (by calling
xsk_inc_num_desc()).
If the build fails in between (e.g. alloc_page() returns NULL with
-EAGAIN, or the MAX_SKB_FRAGS overflow hits), we jump to free_err, skip
calling xsk_inc_num_desc() to increment num_descs and leave the half-built
skb attached to xs->skb for the app to retry. The skb now has
1) destructor_arg = a real xsk_addrs pointer,
2) num_descs = 1
If the app never retries and just close()s the socket, xsk_release()
calls xsk_drop_skb() -> xsk_consume_skb(), which decides whether to
free xsk_addrs by testing num_descs > 1:
if (unlikely(num_descs > 1))
kmem_cache_free(xsk_tx_generic_cache, destructor_arg);
Because num_descs is exactly 1 the branch is skipped and the
xsk_addrs object is leaked to the xsk_tx_generic_cache slab.
Fix it by directly testing if destructor_arg is still addr. Or else it
is modified and used to store the newly allocated memory from
xsk_tx_generic_cache regardless of increment of num_desc, which we
need to handle.
Closes: https://lore.kernel.org/all/20260419045824.D9E5EC2BCAF@smtp.kernel.org/ Fixes: 0ebc27a4c67d ("xsk: avoid data corruption on cq descriptor number") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-8-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:20 +0000 (23:07 +0300)]
xsk: avoid skb leak in XDP_TX_METADATA case
Fix it by explicitly adding kfree_skb() before returning back to its
caller.
How to reproduce it in virtio_net:
1. the current skb is the first one (which means no frag and xs->skb is
NULL) and users enable metadata feature.
2. xsk_skb_metadata() returns a error code.
3. the caller xsk_build_skb() clears skb by using 'skb = NULL;'.
4. there is no chance to free this skb anymore.
Closes: https://lore.kernel.org/all/20260415085204.3F87AC19424@smtp.kernel.org/ Fixes: 30c3055f9c0d ("xsk: wrap generic metadata handling onto separate function") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-7-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:19 +0000 (23:07 +0300)]
xsk: prevent CQ desync when freeing half-built skbs in xsk_build_skb()
Once xsk_skb_init_misc() has been called on an skb, its destructor is
set to xsk_destruct_skb(), which submits the descriptor address(es) to
the completion queue and advances the CQ producer. If such an skb is
subsequently freed via kfree_skb() along an error path - before the
skb has ever been handed to the driver - the destructor still runs and
submits a bogus, half-initialized address to the CQ.
Postpone the init phase when we believe the allocation of first frag is
successfully completed. Before this init, skb can be safely freed by
kfree_skb().
Closes: https://lore.kernel.org/all/20260419045822.843BFC2BCAF@smtp.kernel.org/ Fixes: c30d084960cf ("xsk: avoid overwriting skb fields for multi-buffer traffic") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-6-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:18 +0000 (23:07 +0300)]
xsk: fix use-after-free of xs->skb in xsk_build_skb() free_err path
When xsk_build_skb() processes multi-buffer packets in copy mode, the
first descriptor stores data into the skb linear area without adding
any frags, so nr_frags stays at 0. The caller then sets xs->skb = skb
to accumulate subsequent descriptors.
If a continuation descriptor fails (e.g. alloc_page returns NULL with
-EAGAIN), we jump to free_err where the condition:
if (skb && !skb_shinfo(skb)->nr_frags)
kfree_skb(skb);
evaluates to true because nr_frags is still 0 (the first descriptor
used the linear area, not frags). This frees the skb while xs->skb
still points to it, creating a dangling pointer. On the next transmit
attempt or socket close, xs->skb is dereferenced, causing a
use-after-free or double-free.
Fix by using a !xs->skb check to handle first frag situation, ensuring
we only free skbs that were freshly allocated in this call
(xs->skb is NULL) and never free an in-progress multi-buffer skb that
the caller still references.
Closes: https://lore.kernel.org/all/20260415082654.21026-4-kerneljasonxing@gmail.com/ Fixes: 6b9c129c2f93 ("xsk: remove @first_frag from xsk_build_skb()") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-5-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:17 +0000 (23:07 +0300)]
xsk: handle NULL dereference of the skb without frags issue
When a first descriptor (xs->skb == NULL) triggers -EOVERFLOW in
xsk_build_skb_zerocopy() (e.g., MAX_SKB_FRAGS exceeded), the
free_err -EOVERFLOW handler unconditionally dereferences xs->skb
via xsk_inc_num_desc(xs->skb) and xsk_drop_skb(xs->skb), causing
a NULL pointer dereference.
Fix this by guarding the existing xsk_inc_num_desc()/xsk_drop_skb()
calls with an xs->skb check (for the continuation case), and add
an else branch for the first-descriptor case that manually cancels
the one reserved CQ slot and increments invalid_descs by one to
account for the single invalid descriptor.
Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-4-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:16 +0000 (23:07 +0300)]
xsk: free the skb when hitting the upper bound MAX_SKB_FRAGS
Fix it by explicitly adding kfree_skb() before returning back to its
caller.
How to reproduce it in virtio_net:
1. the current skb is the first one (which means xs->skb is NULL) and
hit the limit MAX_SKB_FRAGS.
2. xsk_build_skb_zerocopy() returns -EOVERFLOW.
3. the caller xsk_build_skb() clears skb by using 'skb = NULL;'. This
is why bug can be triggered.
4. there is no chance to free this skb anymore.
Note that if in this case the xs->skb is not NULL, xsk_build_skb() will
call xsk_drop_skb(xs->skb) to do the right thing.
Fixes: cf24f5a5feea ("xsk: add support for AF_XDP multi-buffer on Tx path") Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Link: https://patch.msgid.link/20260502200722.53960-3-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jason Xing [Sat, 2 May 2026 20:07:15 +0000 (23:07 +0300)]
xsk: reject sw-csum UMEM binding to IFF_TX_SKB_NO_LINEAR devices
skb_checksum_help() is a common helper that writes the folded
16-bit checksum back via skb->data + csum_start + csum_offset,
i.e. it relies on the skb's linear head and fails (with WARN_ONCE
and -EINVAL) when skb_headlen() is 0.
AF_XDP generic xmit takes two very different paths depending on the
netdev. Drivers that advertise IFF_TX_SKB_NO_LINEAR (e.g. virtio_net)
skip the "copy payload into a linear head" step on purpose as a
performance optimisation: xsk_build_skb_zerocopy() only attaches UMEM
pages as frags and never calls skb_put(), so skb_headlen() stays 0
for the whole skb. For these skbs there is simply no linear area for
skb_checksum_help() to write the csum into - the sw-csum fallback is
structurally inapplicable.
The patch tries to catch this and reject the combination with error at
setup time. Rejecting at bind() converts this silent per-packet failure
into a synchronous, actionable -EOPNOTSUPP at setup time. HW csum and
launch_time metadata on IFF_TX_SKB_NO_LINEAR drivers are unaffected
because they do not call skb_checksum_help().
Without the patch, every descriptor carrying 'XDP_TX_METADATA |
XDP_TXMD_FLAGS_CHECKSUM' produces:
1) a WARN_ONCE "offset (N) >= skb_headlen() (0)" from skb_checksum_help(),
2) sendmsg() returning -EINVAL without consuming the descriptor
(invalid_descs is not incremented),
3) a wedged TX ring: __xsk_generic_xmit() does not advance the
consumer on non-EOVERFLOW errors, so the next sendmsg() re-reads
the same descriptor and re-hits the same WARN until the socket
is closed.
Closes: https://lore.kernel.org/all/20260419045822.843BFC2BCAF@smtp.kernel.org/#t Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Acked-by: Stanislav Fomichev <sdf@fomichev.me> Signed-off-by: Jason Xing <kernelxing@tencent.com> Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com> Fixes: 30c3055f9c0d ("xsk: wrap generic metadata handling onto separate function") Link: https://patch.msgid.link/20260502200722.53960-2-kerneljasonxing@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jakub Kicinski [Wed, 6 May 2026 02:13:12 +0000 (19:13 -0700)]
Merge branch 'net-mlx5-fixes-for-socket-direct'
Tariq Toukan says:
====================
net/mlx5: Fixes for Socket-Direct
This series fixes several race conditions and bugs in the mlx5
Socket-Direct (SD) single netdev flow.
Patch 1 serializes mlx5_sd_init()/mlx5_sd_cleanup() with
mlx5_devcom_comp_lock() and tracks the SD group state on the primary
device, preventing concurrent or duplicate bring-up/tear-down.
Patch 2 fixes the debugfs "multi-pf" directory being stored on the
calling device's sd struct instead of the primary's, which caused
memory leaks and recreation errors when cleanup ran from a different PF.
Patch 3 fixes a race where a secondary PF could access the primary's
auxiliary device after it had been unbound, by holding the primary's
device lock while operating on its auxiliary device.
Patch 4 fixes missing cleanup on ETH probe errors. The analogous gap on
the resume path requires introducing sd_suspend/resume APIs that only
destroy FW resources and is left for a follow-up series.
====================
Shay Drory [Mon, 4 May 2026 18:02:06 +0000 (21:02 +0300)]
net/mlx5e: SD, Fix race condition in secondary device probe/remove
When utilizing Socket-Direct single netdev functionality the driver
resolves the actual auxiliary device using mlx5_sd_get_adev(). However,
the current implementation returns the primary ETH auxiliary device
without holding the device lock, leading to a potential race condition
where the ETH device could be unbound or removed concurrently during
probe, suspend, resume, or remove operations.[1]
Fix this by introducing mlx5_sd_put_adev() and updating
mlx5_sd_get_adev() so that secondaries devices would get a ref and
acquire the device lock of the returned auxiliary device. After the lock
is acquired, a second devcom check is needed[2].
In addition, update The callers to pair the get operation with the new
put operation, ensuring the lock is held while the auxiliary device is
being operated on and released afterwards.
The "primary" designation is determined once in sd_register(). It's set
before devcom is marked ready, and it never changes after that.
In Addition, The primary path never locks a secondary: When the primary
device invoke mlx5_sd_get_adev(), it sees dev == primary and returns.
no additional lock is taken.
Therefore lock ordering is always: secondary_lock -> primary_lock. The
reverse never happens, so ABBA deadlock is impossible.
Shay Drory [Mon, 4 May 2026 18:02:05 +0000 (21:02 +0300)]
net/mlx5e: SD, Fix missing cleanup on probe error
When _mlx5e_probe() fails, the preceding successful mlx5_sd_init() is
not undone. Auxiliary bus probe failure skips binding, so mlx5e_remove()
is never called for that adev and the matching mlx5_sd_cleanup() never
runs - leaking the per-dev SD struct.
Call mlx5_sd_cleanup() on the probe error path to balance
mlx5_sd_init().
A similar gap exists on the resume path: mlx5_sd_init() and
mlx5_sd_cleanup() are currently bundled with both probe/remove and
suspend/resume, even though only the FW alias state actually needs to
follow the suspend/resume lifecycle - the sd struct allocation and
devcom membership are software state that should track the full bound
lifetime. As a result, a failed resume can leave a still-bound device
with sd == NULL, which mlx5_sd_get_adev() can't distinguish from a
non-SD device. Fixing this requires sd_suspend/resume APIs which will
only destroy FW resources and is left for a follow-up series.
Fixes: 381978d28317 ("net/mlx5e: Create single netdev per SD group") Signed-off-by: Shay Drory <shayd@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260504180206.268568-4-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Shay Drory [Mon, 4 May 2026 18:02:04 +0000 (21:02 +0300)]
net/mlx5: SD, Keep multi-pf debugfs entries on primary
mlx5_sd_init() creates the "multi-pf" debugfs directory under the
primary device debugfs root, but stored the dentry in the calling
device's sd struct. When sd_cleanup() run on a different PF,
this leads to using the wrong sd->dfs for removing entries, which
results in memory leak and an error in when re-creating the SD.[1]
Fix it by explicitly storing the debugfs dentry in the primary
device sd struct and use it for all per-group files.
[1]
debugfs: 'multi-pf' already exists in '0000:08:00.1'
Shay Drory [Mon, 4 May 2026 18:02:03 +0000 (21:02 +0300)]
net/mlx5: SD: Serialize init/cleanup
mlx5_sd_init() / mlx5_sd_cleanup() may run from multiple PFs in the same
Socket-Direct group. This can cause the SD bring-up/tear-down sequence
to be executed more than once or interleaved across PFs.
Protect SD init/cleanup with mlx5_devcom_comp_lock() and track the SD
group state on the primary device. Skip init if the primary is already
UP, and skip cleanup unless the primary is UP.
The state check on cleanup is needed because sd_register() drops the
devcom comp lock between marking the comp ready and assigning
primary_dev on each peer. A concurrent cleanup that acquires the lock
in this window would observe devcom_is_ready==true while primary_dev
is still NULL (causing mlx5_sd_get_primary() to return NULL) or while
the FW alias setup performed by mlx5_sd_init()'s body has not yet run
(causing sd_cmd_unset_primary() to dereference a NULL tx_ft). Gate the
cleanup body on primary_sd->state == MLX5_SD_STATE_UP, which is set
only at the very end of mlx5_sd_init() under the same comp lock - so
observing UP guarantees primary_dev, secondaries[], tx_ft, and dfs are
all populated. Also bail explicitly if mlx5_sd_get_primary() returns
NULL, in case state is checked on a peer whose primary_dev hasn't been
assigned yet.
In addition, move mlx5_devcom_comp_set_ready(false) from sd_unregister()
into the cleanup's locked section, including the !primary and
state != UP early-exit paths, so the device cannot unregister and free
its struct mlx5_sd while devcom is still marked ready. A concurrent
init acquiring the devcom lock will now observe devcom is no longer
ready and bail out immediately.
Fixes: 381978d28317 ("net/mlx5e: Create single netdev per SD group") Signed-off-by: Shay Drory <shayd@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260504180206.268568-2-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The reason for failure is the existence of TX keys, which are removed by
the PSP dev unregistration happening in:
profile->cleanup() -> mlx5e_psp_unregister() -> mlx5e_psp_cleanup()
-> psp_dev_unregister()
...but this isn't invoked in the devlink reload flow, only when changing
the NIC profile (e.g. when transitioning to switchdev mode) or on dev
teardown.
Move PSP device registration into mlx5e_nic_enable(), and unregistration
into the corresponding mlx5e_nic_disable(). These functions are called
during netdev attach/detach after RX & TX are set up.
This ensures that the keys will be gone by the time the PD is destroyed.
Cosmin Ratiu [Mon, 4 May 2026 18:10:58 +0000 (21:10 +0300)]
net/mlx5e: psp: Fix invalid access on PSP dev registration fail
priv->psp->psp is initialized with the PSP device as returned by
psp_dev_create(). This could also return an error, in which case a
future psp_dev_unregister() will result in unpleasantness.
Avoid that by using a local variable and only saving the PSP device when
registration succeeds.
In case psp_dev_create() fails, priv->psp and steering structs are left
in place, but they will be inert. The unchecked access of priv->psp in
mlx5e_psp_offload_handle_rx_skb() won't happen because without a PSP
device, there can be no SAs added and therefore no packets will be
successfully decrypted and be handed off to the SW handler.
Pavitra Jha [Fri, 1 May 2026 11:07:12 +0000 (07:07 -0400)]
net: wwan: t7xx: validate port_count against message length in t7xx_port_enum_msg_handler
t7xx_port_enum_msg_handler() uses the modem-supplied port_count field as
a loop bound over port_msg->data[] without checking that the message buffer
contains sufficient data. A modem sending port_count=65535 in a 12-byte
buffer triggers a slab-out-of-bounds read of up to 262140 bytes.
Add a sizeof(*port_msg) check before accessing the port message header
fields to guard against undersized messages.
Add a struct_size() check after extracting port_count and before the loop.
In t7xx_parse_host_rt_data(), guard the rt_feature header read with a
remaining-buffer check before accessing data_len, validate feat_data_len
against the actual remaining buffer to prevent OOB reads and signed
integer overflow on offset.
Pass msg_len from both call sites: skb->len at the DPMAIF path after
skb_pull(), and the validated feat_data_len at the handshake path.
Eric Dumazet [Mon, 4 May 2026 16:38:42 +0000 (16:38 +0000)]
net/sched: sch_fq_codel: annotate data-races from fq_codel_dump_class_stats()
fq_codel_dump_class_stats() acquires qdisc spinlock only when requested
to follow flow->head chain.
As we did in sch_cake recently, add the missing READ_ONCE()/WRITE_ONCE()
annotations.
Fixes: edb09eb17ed8 ("net: sched: do not acquire qdisc spinlock in qdisc/class stats dump") Signed-off-by: Eric Dumazet <edumazet@google.com> Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com> Link: https://patch.msgid.link/20260504163842.1162001-1-edumazet@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
I added missing Fixes tag. The original description:
Since commit 041ee6f3727a ("kthread: Rely on HK_TYPE_DOMAIN for preferred
affinity management"), the HK_TYPE_KTHREAD housekeeping cpumask may no
longer be correct in showing the actual CPU affinity of kthreads that
have no predefined CPU affinity. As the ipvs networking code is still
using HK_TYPE_KTHREAD, we need to make HK_TYPE_KTHREAD reflect the
reality.
This patch series makes HK_TYPE_KTHREAD an alias of HK_TYPE_DOMAIN
and uses RCU to protect access to the HK_TYPE_KTHREAD housekeeping
cpumask.
Julian plans to post a nf-next patch to limit the connections by using
"conn_max" sysctl. With Simon Horman, they agreed that this is an old
problem that we do not have a limit of connections and it is not a
stopper for this patchset.
* tag 'nf-26-05-05' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
sched/isolation: Make HK_TYPE_KTHREAD an alias of HK_TYPE_DOMAIN
ipvs: Guard access of HK_TYPE_KTHREAD cpumask with RCU
ipvs: fix shift-out-of-bounds in ip_vs_rht_desired_size
ipvs: fix races around est_mutex and est_cpulist
ipvs: do not leak dest after get from dest trash
ipvs: fix the spin_lock usage for RT build
ipvs: fix races around the conn_lfactor and svc_lfactor sysctl vars
ipvs: fixes for the new ip_vs_status info
====================
Pavan Chebbi [Mon, 4 May 2026 08:36:11 +0000 (14:06 +0530)]
bnxt_en: Use absolute target ns from ptp_clock_request
There is no need to calculate the target PHC cycles required
to make phase adjustment on the PPS OUT signal. This is because
the application supplies absolute n_sec value in the future and
is already the actual desired target value.
Remove the unnecessary code.
Fixes: 9e518f25802c ("bnxt_en: 1PPS functions to configure TSIO pins") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Cc: Richard Cochran <richardcochran@gmail.com> Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev> Tested-by: Vadim Fedorenko <vadim.fedorenko@linux.dev> Link: https://patch.msgid.link/20260504083611.1383776-5-pavan.chebbi@broadcom.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Kalesh AP [Mon, 4 May 2026 08:36:10 +0000 (14:06 +0530)]
bnxt_en: Check return value of bnxt_hwrm_vnic_cfg
When the bnxt RDMA driver is loaded, it calls bnxt_register_dev().
As part of this, driver sends HWRM_VNIC_CFG firmware command
to configure the VNIC to operate in dual VNIC mode. Currently
the driver ignores the result of this firmware command. The RDMA
driver must know the result since it affects its functioning.
Check return value of call to bnxt_hwrm_vnic_cfg() in
bnxt_register_dev() and return failure on error.
Fixes: a588e4580a7e ("bnxt_en: Add interface to support RDMA driver.") Reviewed-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Link: https://patch.msgid.link/20260504083611.1383776-4-pavan.chebbi@broadcom.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Michael Chan [Mon, 4 May 2026 08:36:09 +0000 (14:06 +0530)]
bnxt_en: Set bp->max_tpa according to what the FW supports
Fix the logic to set bp->max_tpa no higher than what the FW supports.
On P5 chips, some older FW sets max_tpa very low so we override it to
prevent performance regressions with the older FW.
Fixes: 79632e9ba386 ("bnxt_en: Expand bnxt_tpa_info struct to support 57500 chips.") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Reviewed-by: Colin Winegarden <colin.winegarden@broadcom.com> Reviewed-by: Rukhsana Ansari <rukhsana.ansari@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Link: https://patch.msgid.link/20260504083611.1383776-3-pavan.chebbi@broadcom.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Michael Chan [Mon, 4 May 2026 08:36:08 +0000 (14:06 +0530)]
bnxt_en: Delay for 5 seconds after AER DPC for all chips
The FW on all chips is requiring a 5-second delay after Downstream
Port Containment (DPC) AER. The previously added 900 msec delay was
not long enough in all cases because the chip's CRS (Configuration
Request Retry Status) mechanism is not always reliable.
Fixes: d5ab32e9b02d ("bnxt_en: Add delay to handle Downstream Port Containment (DPC) AER") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Michael Chan <michael.chan@broadcom.com> Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Link: https://patch.msgid.link/20260504083611.1383776-2-pavan.chebbi@broadcom.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Alyssa Ross [Sun, 3 May 2026 19:25:16 +0000 (21:25 +0200)]
ipv6: default IPV6_SIT to m
This basically defaulted to m until recently, since IPV6 defaulted to
m. Since IPV6 was changed to a boolean with a default of y, IPV6_SIT
started defaulting to built-in as well. This results in a surprise
sit0 device by default for defconfig (and defconfig-derived config)
users at boot. For me, this broke an (admittedly non-robust) script.
Preserve the behaviour of most configs by avoiding building this
module, that's probably overall seldom used compared to IPv6 as a
whole, into the kernel.
Fixes: 309b905deee59 ("ipv6: convert CONFIG_IPV6 to built-in only and clean up Kconfigs") Signed-off-by: Alyssa Ross <hi@alyssa.is> Reviewed-by: Fernando Fernandez Mancera <fmancera@suse.de> Link: https://patch.msgid.link/20260503192515.290900-2-hi@alyssa.is Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Benjamin Berg [Tue, 5 May 2026 13:15:40 +0000 (15:15 +0200)]
wifi: mac80211: use safe list iteration in radar detect work
The call to ieee80211_dfs_cac_cancel can cause the iterated chanctx to
be freed and removed from the list. Guard against this to avoid a
slab-use-after-free error.
Johannes Berg [Tue, 5 May 2026 15:52:32 +0000 (17:52 +0200)]
Merge tag 'ath-current-20260505' of git://git.kernel.org/pub/scm/linux/kernel/git/ath/ath
Jeff Johnson says:
==================
ath.git update for v7.1-rc3
Fix an ath5k potential stack buffer overwrite.
Fix several issues in ath12k:
- WMI buffer leaks on error conditions
- use of uninitialized stack data when processing RSSI events
- incorrect logic for determining the peer ID in the RX path
==================
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Dipayaan Roy [Fri, 1 May 2026 02:47:12 +0000 (19:47 -0700)]
net: mana: Fix crash from unvalidated SHM offset read from BAR0 during FLR
During Function Level Reset recovery, the MANA driver reads
hardware BAR0 registers that may temporarily contain garbage values.
The SHM (Shared Memory) offset read from GDMA_REG_SHM_OFFSET is used
to compute gc->shm_base, which is later dereferenced via readl() in
mana_smc_poll_register(). If the hardware returns an unaligned or
out-of-range value, the driver must not blindly use it, as this would
propagate the hardware error into a kernel crash.
The following crash was observed on an arm64 Hyper-V guest running
kernel 6.17.0-3013-azure during VF reset recovery triggered by HWC
timeout.
From the crash signature x20 = ffff8000a200001b, this address
ends in 0x1b which is not 4-byte aligned, so the 'ldr w1, [x20]'
instruction (readl) triggers the arm64 alignment fault (FSC = 0x21).
The root cause is in mana_gd_init_vf_regs(), which computes:
The offset is used without any validation. The same problem exists
in mana_gd_init_pf_regs() for sriov_base_off and sriov_shm_off.
Fix this by validating all offsets before use:
- VF: check shm_off is within BAR0, properly aligned to 4 bytes
(readl requirement), and leaves room for the full 256-bit
(32-byte) SMC aperture.
- PF: check sriov_base_off is within BAR0, aligned to 8 bytes
(readq requirement), and leaves room to safely read the
sriov_shm_off register at sriov_base_off + GDMA_PF_REG_SHM_OFF.
Then check sriov_shm_off leaves room for the full SMC aperture.
All arithmetic uses subtraction rather than addition to avoid
integer overflow on garbage values.
Define SMC_APERTURE_SIZE (32 bytes, derived from the 256-bit aperture
width)
Return -EPROTO on invalid values. The existing recovery path in
mana_serv_reset() already handles -EPROTO by falling through to PCI
device rescan, giving the hardware another chance to present valid
register values after reset.
Nan Li [Fri, 1 May 2026 01:08:44 +0000 (09:08 +0800)]
net/rds: handle zerocopy send cleanup before the message is queued
A zerocopy send can fail after user pages have been pinned but before
the message is attached to the sending socket.
The purge path currently infers zerocopy state from rm->m_rs, so an
unqueued message can be cleaned up as if it owned normal payload pages.
However, zerocopy ownership is really determined by the presence of
op_mmp_znotifier, regardless of whether the message has reached the
socket queue.
Capture op_mmp_znotifier up front in rds_message_purge() and use it as
the cleanup discriminator. If the message is already associated with a
socket, keep the existing completion path. Otherwise, drop the pinned
page accounting directly and release the notifier before putting the
payload pages.
This keeps early send failure cleanup consistent with the zerocopy
lifetime rules without changing the normal queued completion path.
Fixes: 0cebaccef3ac ("rds: zerocopy Tx support.") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Co-developed-by: Xiao Liu <lx24@stu.ynu.edu.cn> Signed-off-by: Xiao Liu <lx24@stu.ynu.edu.cn> Signed-off-by: Nan Li <tonanli66@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Reviewed-by: Allison Henderson <achender@kernel.org> Link: https://patch.msgid.link/d2ea98a6313d5467bac00f7c9fef8c7acddb9258.1777550074.git.tonanli66@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
openvswitch: fix self-deadlock on release of tunnel vports
Two patches - the fix for the actual bug and the selftest that reproduces it.
I missed the self-deadlock in the original patch that introduced the issue,
because testing required code modification in the ovs-vswitchd to force it to
use legacy tunnel ports. I thought I made the change correctly, but apparently
something went wrong and the tests were run with the standard LWT infra instead.
The selftest added in this patch set will at least prevent this kind of mistakes
in the future.
I mentioned, however, that these tunnel vports are legacy and not actually used
by ovs-vswitchd. RTM_NEWLINK + COLLECT_METADATA is used in conjunction with the
standard OVS_VPORT_TYPE_NETDEV instead since 2017. The code to use the legacy
tunnels still exists in ovs-vswitchd however, but only as a fallback for older
kernels and we're planning to remove it in the next release. I'll be sending an
RFC to remove support for these legacy tunnel types from the kernel, as they
serve no real purpose today and only increase the uAPI surface for CVEs, but
we need to fix the known bugs for stable versions.
selftests: openvswitch: add tests for tunnel vport refcounting
There were a few issues found with the tunnel vport types around the
vport destruction code. Add some basic tests, so at least we know that
they can be properly added and removed without obvious issues.
The test creates OVS datapath, adds a non-LWT tunnel port, makes sure
they are created, and then removes the datapath and waits for all the
ports to be gone.
The dpctl script had a few bugs in the none-lwt tunnel creation code,
so fixing them as well to make the testing possible:
- The type of the --lwt option changed in order to properly disable it.
- Removed byte order conversion for the port numbers, as the value
supposed to be in the host order.
- Added missing 'gre' choice for the tunnel type.
openvswitch: vport: fix self-deadlock on release of tunnel ports
vports are used concurrently and protected by RCU, so netdev_put()
must happen after the RCU grace period. So, either in an RCU call or
after the synchronize_net(). The rtnl_delete_link() must happen under
RTNL and so can't be executed in RCU context. Calling synchronize_net()
while holding RTNL is not a good idea for performance and system
stability under load in general, so calling netdev_put() in RCU call
is the right solution here.
However,
when the device is deleted, rtnl_unlock() will call netdev_run_todo()
and block until all the references are gone. In the current code this
means that we never reach the call_rcu() and the vport is never freed
and the reference is never released, causing a self-deadlock on device
removal.
Fix that by moving the rcu_call() before the rtnl_unlock(), so the
scheduled RCU callback will be executed when synchronize_net() is
called from the rtnl_unlock()->netdev_run_todo() while the RTNL itself
is already released.
openvswitch: vport: fix race between tunnel creation and linking
When a tunnel vport is created it first creates the tunnel device, e.g.,
with geneve_dev_create_fb(), then it calls ovs_netdev_link() to take a
reference and link it to the device that represents openvswitch datapath.
The creation of the device is happening under RTNL, but then RTNL is
released and re-acquired to find the device by name. It is technically
possible for the tunnel device to be re-named or deleted within that
window while RTNL is not held, and some other device created in its
place. This will cause a non-tunnel device to be referenced in the
vport and tunnel-specific functions used on it, e.g. vxlan_get_options()
that directly casts the private netdev data into a struct vxlan_dev
causing an invalid memory access:
BUG: KASAN: slab-use-after-free in vxlan_get_options+0x323/0x3a0
vxlan_get_options+0x323/0x3a0
ovs_vport_cmd_new+0x6e3/0xd30
Fix that by taking a reference to the just created device before
releasing RTNL. This ensures that the device in the vport is always
the one that was just created. The search by name is only needed
for a standard vport-netdev that links pre-existing devices, so that
functionality and device type checks are moved to netdev_create().
It is also awkward that ovs_netdev_link() takes ownership of the vport
and destroys it on failure. It doesn't know the type of the port it is
dealing with, so we need to pass down the indicator that it's a tunnel,
so the link can be properly deleted on failure.
It's possible to refactor the logic to make the ovs_netdev_link() do
only the linking part and let the callers perform a proper destruction,
but it will be much more code for each legacy tunnel port type, so it
is not worth it for the bug fix.
Fixes: 614732eaa12d ("openvswitch: Use regular VXLAN net_device device") Reported-by: Yuan Tan <tanyuan98@outlook.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Reported-by: Yang Yang <n05ec@lzu.edu.cn> Signed-off-by: Ilya Maximets <i.maximets@ovn.org> Acked-by: Eelco Chaudron <echaudro@redhat.com> Link: https://patch.msgid.link/20260430213349.407991-1-i.maximets@ovn.org Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
net: mana: Fix mana_destroy_rxq() cleanup for partial RXQ init
When mana_create_rxq() fails partway through initialization (e.g. the
hardware rejects the WQ object creation), the error path calls
mana_destroy_rxq() to tear down a partially-initialized RXQ.
This exposed multiple issues in mana_destroy_rxq() path, as it assumed
the RXQ was always fully initialized, leading to multiple issues:
1. xdp_rxq_info_unreg() was called on an unregistered xdp_rxq,
triggering a WARN_ON ("Driver BUG") in net/core/xdp.c.
2. mana_destroy_wq_obj() was called with INVALID_MANA_HANDLE,
sending a bogus destroy command to the hardware.
3. mana_deinit_cq() was called twice — once inside mana_destroy_rxq()
and again in mana_create_rxq()'s error path — causing a
use-after-free since mana_destroy_rxq() frees the rxq first.
This was observed during ethtool ring parameter changes when the
hardware returned an error creating the RXQ. This series makes
mana_destroy_rxq() safe to call at any stage of RXQ initialization
by guarding each teardown step, and removes the redundant cleanup
in mana_create_rxq().
====================
Dipayaan Roy [Thu, 30 Apr 2026 03:57:54 +0000 (20:57 -0700)]
net: mana: remove double CQ cleanup in mana_create_rxq error path
In mana_create_rxq(), the error cleanup path calls mana_destroy_rxq()
followed by mana_deinit_cq(). This is incorrect for two reasons:
1. mana_destroy_rxq() already calls mana_deinit_cq() internally,
so the CQ's GDMA queue is destroyed twice.
2. mana_destroy_rxq() frees the rxq via kfree(rxq) before returning.
The subsequent mana_deinit_cq(apc, cq) then operates on freed memory
since cq points to &rxq->rx_cq, which is embedded in the
already-freed rxq structure — a use-after-free.
Remove the redundant mana_deinit_cq() call from the error path since
mana_destroy_rxq() already handles CQ cleanup. mana_deinit_cq() is
itself safe for an uninitialized CQ as it checks for a NULL gdma_cq
before proceeding.
Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)") Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com> Signed-off-by: Dipayaan Roy <dipayanroy@linux.microsoft.com> Reviewed-by: Aditya Garg <gargaditya@linux.microsoft.com> Link: https://patch.msgid.link/20260430035935.1859220-4-dipayanroy@linux.microsoft.com Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Dipayaan Roy [Thu, 30 Apr 2026 03:57:53 +0000 (20:57 -0700)]
net: mana: Skip WQ object destruction for uninitialized RXQ
In mana_destroy_rxq(), mana_destroy_wq_obj() is called unconditionally
even when the WQ object was never created (rxobj is still
INVALID_MANA_HANDLE). When mana_create_rxq() fails before
mana_create_wq_obj() succeeds, the error path calls mana_destroy_rxq()
which sends a bogus destroy command to the hardware:
mana 7870:00:00.0: HWC: Failed hw_channel req: 0x1d
mana 7870:00:00.0: Failed to send mana message: -71, 0x1d
mana 7870:00:00.0 eth7: Failed to destroy WQ object: -71
Guard mana_destroy_wq_obj() with an INVALID_MANA_HANDLE check so that
mana_destroy_rxq() is safe to call at any stage of RXQ initialization.
Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)") Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com> Signed-off-by: Dipayaan Roy <dipayanroy@linux.microsoft.com> Link: https://patch.msgid.link/20260430035935.1859220-3-dipayanroy@linux.microsoft.com Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Dipayaan Roy [Thu, 30 Apr 2026 03:57:52 +0000 (20:57 -0700)]
net: mana: check xdp_rxq registration before unreg in mana_destroy_rxq()
When mana_create_rxq() fails at mana_create_wq_obj() or any step before
xdp_rxq_info_reg() is called, the error path jumps to `out:` which calls
mana_destroy_rxq(). mana_destroy_rxq() unconditionally calls
xdp_rxq_info_unreg() on xilinx xdp_rxq that was never registered,
triggering a WARN_ON in net/core/xdp.c:
Guard the xdp_rxq_info_unreg() call with xdp_rxq_info_is_reg() so that
mana_destroy_rxq() is safe to call regardless of how far initialization
progressed.
Fixes: ed5356b53f07 ("net: mana: Add XDP support") Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com> Signed-off-by: Dipayaan Roy <dipayanroy@linux.microsoft.com> Link: https://patch.msgid.link/20260430035935.1859220-2-dipayanroy@linux.microsoft.com Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Jakov Novak [Mon, 4 May 2026 16:23:57 +0000 (18:23 +0200)]
wifi: libertas: notify firmware load wait on disconnect
Currently, when the firmware is not fully loaded and if_usb_disconnect
is called, if_usb_prog_firmware gets stuck waiting for
cardp->surprise_removed or cardp->fwdnldover while lbs_remove_card
also waits for the firmware loading to be completed, which never happens.
This caused the reported syzbot bug. To address this, the wake_up
function call can be added in the if_usb_disconnect function which notifies
the if_usb_prog_firmware thread and resolves the firmware loading.
Kuan-Ting Chen [Mon, 4 May 2026 15:27:12 +0000 (23:27 +0800)]
xfrm: esp: avoid in-place decrypt on shared skb frags
MSG_SPLICE_PAGES can attach pages from a pipe directly to an skb. TCP
marks such skbs with SKBFL_SHARED_FRAG after skb_splice_from_iter(),
so later paths that may modify packet data can first make a private
copy. The IPv4/IPv6 datagram append paths did not set this flag when
splicing pages into UDP skbs.
That leaves an ESP-in-UDP packet made from shared pipe pages looking
like an ordinary uncloned nonlinear skb. ESP input then takes the no-COW
fast path for uncloned skbs without a frag_list and decrypts in place
over data that is not owned privately by the skb.
Mark IPv4/IPv6 datagram splice frags with SKBFL_SHARED_FRAG, matching
TCP. Also make ESP input fall back to skb_cow_data() when the flag is
present, so ESP does not decrypt externally backed frags in place.
Private nonlinear skb frags still use the existing fast path.
This intentionally does not change ESP output. In esp_output_head(),
the path that appends the ESP trailer to existing skb tailroom without
calling skb_cow_data() is not reachable for nonlinear skbs:
skb_tailroom() returns zero when skb->data_len is nonzero, while ESP
tailen is positive. Thus ESP output will either use the separate
destination-frag path or fall back to skb_cow_data().
Daniel Golle [Sat, 2 May 2026 10:55:02 +0000 (11:55 +0100)]
net: dsa: mt7530: fix .get_stats64 sleeping in atomic context
The .get_stats64 callback runs in atomic context, but on
MDIO-connected switches every register read acquires the MDIO bus
mutex, which can sleep:
[ 12.645973] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:609
[ 12.654442] in_atomic(): 0, irqs_disabled(): 0, non_block: 0, pid: 759, name: grep
[ 12.663377] preempt_count: 0, expected: 0
[ 12.667410] RCU nest depth: 1, expected: 0
[ 12.671511] INFO: lockdep is turned off.
[ 12.675441] CPU: 0 UID: 0 PID: 759 Comm: grep Tainted: G S W 7.0.0+ #0 PREEMPT
[ 12.675453] Tainted: [S]=CPU_OUT_OF_SPEC, [W]=WARN
[ 12.675456] Hardware name: Bananapi BPI-R64 (DT)
[ 12.675459] Call trace:
[ 12.675462] show_stack+0x14/0x1c (C)
[ 12.675477] dump_stack_lvl+0x68/0x8c
[ 12.675487] dump_stack+0x14/0x1c
[ 12.675495] __might_resched+0x14c/0x220
[ 12.675504] __might_sleep+0x44/0x80
[ 12.675511] __mutex_lock+0x50/0xb10
[ 12.675523] mutex_lock_nested+0x20/0x30
[ 12.675532] mt7530_get_stats64+0x40/0x2ac
[ 12.675542] dsa_user_get_stats64+0x2c/0x40
[ 12.675553] dev_get_stats+0x44/0x1e0
[ 12.675564] dev_seq_printf_stats+0x24/0xe0
[ 12.675575] dev_seq_show+0x14/0x3c
[ 12.675583] seq_read_iter+0x37c/0x480
[ 12.675595] seq_read+0xd0/0xec
[ 12.675605] proc_reg_read+0x94/0xe4
[ 12.675615] vfs_read+0x98/0x29c
[ 12.675625] ksys_read+0x54/0xdc
[ 12.675633] __arm64_sys_read+0x18/0x20
[ 12.675642] invoke_syscall.constprop.0+0x54/0xec
[ 12.675653] do_el0_svc+0x3c/0xb4
[ 12.675662] el0_svc+0x38/0x200
[ 12.675670] el0t_64_sync_handler+0x98/0xdc
[ 12.675679] el0t_64_sync+0x158/0x15c
For MDIO-connected switches, poll MIB counters asynchronously using a
delayed workqueue every second and let .get_stats64 return the cached
values under a spinlock. A mod_delayed_work() call on each read
triggers an immediate refresh so counters stay responsive when queried
more frequently.
MMIO-connected switches (MT7988, EN7581, AN7583) are not affected
because their regmap does not sleep, so they continue to read MIB
counters directly in .get_stats64.
David Carlier [Sat, 2 May 2026 14:19:45 +0000 (15:19 +0100)]
psp: strip variable-length PSP header in psp_dev_rcv()
psp_dev_rcv() unconditionally removes a fixed PSP_ENCAP_HLEN, even
when psph->hdrlen indicates that the PSP header carries optional
fields. A frame whose PSP header advertises a non-zero VC or any
extension would therefore be silently mis-decapsulated: option bytes
would spill into the inner packet head and downstream parsing would
fail on a corrupted skb.
Compute the full PSP header length from psph->hdrlen, pull the
optional bytes into the linear region, and strip the whole header
when decapsulating. Optional fields (VC, ...) are still ignored,
just discarded with the rest of the header instead of leaking.
crypt_offset and the VIRT flag are intentionally not validated here
- callers know their device's PSP implementation and can decide.
Both in-tree callers gate on hardware-validated PSP, so this is a
correctness fix rather than a reachable corruption path under
current configurations.
Fixes: 0eddb8023cee ("psp: provide decapsulation and receive helper for drivers") Reviewed-by: Willem de Bruijn <willemb@google.com> Reviewed-by: Daniel Zahka <daniel.zahka@gmail.com> Cc: stable@vger.kernel.org Signed-off-by: David Carlier <devnexen@gmail.com> Link: https://patch.msgid.link/20260502141945.14484-1-devnexen@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
projects / thirdparty / kernel / linux.git / log
