Commit graph

140,875 commits

Author SHA1 Message Date
Baul Lee
bc2dc66a66 vxlan: mdb: Fix use-after-free in vxlan_mdb_flush()
vxlan_mdb_flush() iterates over the MDB entries using
hlist_for_each_entry_safe(), which only tolerates the removal of the
current entry. Contrary to the comment above the loop, the removal of an
entry can trigger the removal of another entry.

Flushing the remotes of a (*, G) entry also removes the (S, G) entries
that were created for its source list, once they are left without
remotes:

vxlan_mdb_remotes_flush()
-> vxlan_mdb_remote_del()
   -> vxlan_mdb_remote_srcs_del()
      -> vxlan_mdb_remote_src_del()
         -> vxlan_mdb_remote_src_fwd_del()
            -> __vxlan_mdb_del()
               -> vxlan_mdb_entry_put()

Such an entry can be located after the (*, G) entry in the list, as
vxlan_mdb_entry_get() returns an existing entry without moving it to the
head of the list. This order is obtained by adding the (S, G) entry
before the (*, G) entry, the latter with NLM_F_REPLACE, as the addition
of the source otherwise fails with -EEXIST. The (S, G) entry is then the
entry saved by hlist_for_each_entry_safe() and it is freed while the
(*, G) entry is processed. The next iteration calls hlist_del() on it
again, writing LIST_POISON1 to LIST_POISON2 [1].

Besides device deletion, the flush is also reachable from RTM_DELMDB
with NLM_F_BULK.

Fix by re-reading the next entry after the remotes were flushed. The
current entry cannot be removed by this flush, as source lists can only
be configured on (*, G) entries and the removed entries are (S, G)
entries. It is therefore still linked and its next pointer reflects the
removals.

[1]
BUG: KASAN: wild-memory-access in vxlan_mdb_entry_put.part.0+0x328/0x588
Write of size 8 at addr dead000000000122 by task ip/327

CPU: 3 UID: 1000 PID: 327 Comm: ip Not tainted 7.2.0-rc7 #2 PREEMPT
Call trace:
 vxlan_mdb_entry_put.part.0+0x328/0x588
 vxlan_mdb_flush+0x1d8/0x25c
 vxlan_mdb_fini+0x8c/0x100
 vxlan_uninit+0x1c/0x7c
 unregister_netdevice_many_notify+0x954/0xd4c
 rtnl_dellink+0x210/0x530
 rtnetlink_rcv_msg+0x434/0x4d0
 netlink_rcv_skb+0xc4/0x204
 rtnetlink_rcv+0x18/0x24
 netlink_unicast+0x4b8/0x548
 netlink_sendmsg+0x29c/0x560
 ____sys_sendmsg+0x390/0x3ec
 ___sys_sendmsg+0x114/0x188
 __sys_sendmsg+0xf0/0x178
 __arm64_sys_sendmsg+0x48/0x60
 invoke_syscall.constprop.0+0x58/0x180
 el0_svc_common.constprop.0+0x74/0x140
 do_el0_svc+0x30/0x40
 el0_svc+0x38/0x98
 el0t_64_sync_handler+0xa0/0xe4
 el0t_64_sync+0x198/0x19c

Fixes: a3a48de5ea ("vxlan: mdb: Add MDB control path support")
Signed-off-by: Baul Lee <baul.lee@xbow.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260814153547.29567-1-baul.lee@xbow.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-18 11:58:02 +02:00
Ahmad Fatoum
fb58b6a696 net: dsa: realtek: use gpiod_set_value_cansleep for reset GPIO
rtl83xx_reset_assert() and rtl83xx_reset_deassert() are only called from
the probe path, which may sleep and is not timing-critical.  When the
reset GPIO is provided by a sleeping controller such as an I2C I/O
expander, gpiod_set_value() warns:

  WARNING: drivers/gpio/gpiolib.c:4030 at gpiod_set_value+0x44/0x80, CPU#1: kworker/u16:4/61
  Hardware name: B&O MAP CA33 Rev f (UNKNOWN) (DT)
  Workqueue: events_unbound deferred_probe_work_func
  pc : gpiod_set_value+0x44/0x80
  lr : rtl83xx_probe+0x1d8/0x3a0
  Call trace:
   gpiod_set_value+0x44/0x80 (P)
   rtl83xx_probe+0x1d8/0x3a0
   realtek_mdio_probe+0x24/0xa0
   mdio_probe+0x38/0x78
   really_probe+0xc4/0x3e0
   __driver_probe_device+0x15c/0x1b8
   driver_probe_device+0xb4/0x120
   __device_attach_driver+0xb8/0x1a0
   bus_for_each_drv+0x88/0xf0
   __device_attach+0xa0/0x1d8
   device_initial_probe+0x54/0x68
   bus_probe_device+0x38/0xa0
   deferred_probe_work_func+0xb8/0x120
   process_one_work+0x184/0x4e8
   worker_thread+0x188/0x308
   kthread+0x130/0x150
   ret_from_fork+0x10/0x20

Switch both helpers to gpiod_set_value_cansleep() so such a reset GPIO can
be used without triggering the warning.

The reset GPIO has been driven with the non-sleeping gpiod_set_value()
since the driver was added in v4.19.  The call has since been refactored
across several files - from realtek-smi.c / realtek-mdio.c into the common
rtl83xx.c module and then into the rtl83xx_reset_assert() and
rtl83xx_reset_deassert() helpers (both in v6.9).  This patch therefore
applies as-is only to kernels that carry those helpers (v6.9+); older
stable kernels need the same gpiod_set_value_cansleep() conversion at the
corresponding open-coded call sites.

Fixes: d8652956cf ("net: dsa: realtek-smi: Add Realtek SMI driver")
Cc: <stable@vger.kernel.org> # 6.9.x
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Co-developed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Alvin Šipraga <alvin.sipraga@analog.com>
Reviewed-by: Linus Walleij <linusw@kernel.org>
Reviewed-by: Luiz Angelo Daros de Luca <luizluca@gmail.com>
Link: https://patch.msgid.link/20260814110102.2362246-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 18:26:18 -07:00
Ruoyu Wang
7b196e27ad net: dsa: mv88e6xxx: Fix PCS link check on CMODE read error
mv88e6352_pcs_link_check() ignores errors returned by
port_get_cmode(). If the port status register read fails,
mv88e6352_port_get_cmode() returns without setting cmode. The link check
then compares an uninitialized value and may incorrectly treat the PCS
as active.

Save the return value and fail the link check after releasing the
register lock. marvell_c22_pcs_get_state() initializes the reported link
state to down before calling the check, so a read failure is handled
safely until a later poll succeeds.

This issue was found by a static analysis checker and confirmed by manual
source review.

Fixes: 8576455544 ("net: dsa: mv88e6xxx: convert 88e6352 to phylink_pcs")
Signed-off-by: Ruoyu Wang <ruoyuw560@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Link: https://patch.msgid.link/20260813153131.3952970-1-ruoyuw560@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 16:44:42 -07:00
Xiang Mei
984f831dda vxlan: vnifilter: enforce exact length of GROUP/GROUP6 attributes
The VXLAN VNI filter entry policy declares the GROUP/GROUP6 address
attributes as NLA_BINARY with only a maximum length, so validate_nla()
accepts a payload shorter than the address. The GROUP consumer reads it
with nla_get_in_addr(), an unconditional 4-byte load, so a short
attribute over-reads up to 3 bytes of uninitialised slab data, which are
stored into remote_ip and echoed back via RTM_GETTUNNEL, disclosing
kernel memory.

Switch both entries to NLA_POLICY_EXACT_LEN() so the validator rejects
any GROUP/GROUP6 that is not exactly 4 / 16 bytes; a valid address is
always sent at full width.

Fixes: f9c4bb0b24 ("vxlan: vni filtering support on collect metadata device")
Reported-by: Weiming Shi <bestswngs@gmail.com>
Signed-off-by: Xiang Mei <xmei5@asu.edu>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260812215341.763123-1-xmei5@asu.edu
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 15:57:07 -07:00
Xuanqiang Luo
1cb9663789 net: ravb: serialize PTP clock teardown
ravb_ptp_interrupt() can race with ravb_ptp_stop() and pass the clock to
ptp_clock_event() while ptp_clock_unregister() is freeing it. This can
lead to a use-after-free.

Use READ_ONCE() and WRITE_ONCE() for lockless access to the clock pointer.
Atomically detach it with xchg() before disabling PTP interrupts, then
synchronize all IRQs which can invoke ravb_ptp_interrupt() before
unregistering the detached clock.

A handler which read the old pointer completes before the clock is
unregistered, while later handlers read NULL and skip the event.

Fixes: a0d2f20650 ("Renesas Ethernet AVB PTP clock driver")
Cc: stable@vger.kernel.org
Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
Link: https://patch.msgid.link/20260811103733.62599-3-xuanqiang.luo@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 15:43:17 -07:00
Xuanqiang Luo
1f77af0aaf net: ravb: avoid dereferencing an invalid PTP clock
The PTP clock is unavailable before the first open, so querying its
index can dereference a NULL pointer. Registration failures can also
leave an error pointer in priv->ptp.clock.

Cache the PHC index separately and report -1 while no clock is
registered. Normalize registration errors to NULL and preserve the
static timestamping capabilities.

Fixes: a0d2f20650 ("Renesas Ethernet AVB PTP clock driver")
Cc: stable@vger.kernel.org
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
Link: https://patch.msgid.link/20260811103733.62599-2-xuanqiang.luo@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 15:43:17 -07:00
Anton Protopopov
d09c98a6da virtio_net: Fix resize of the RX ring
When a AF_XDP socket is attached, the virtnet_rx_resize
should resize the rq->xsk_buffs XSK buffer array. Otherwise,
when the size grows, the virtnet_rx_resume() causes a write
past the end of the array. This is easily reproducable with

    ethtool -G ens3 rx 32
    ./xdpsock -i eth0 -q 0 -r -z &
    ethtool -G eth0 rx 256

Fixes: e9f3962441 ("virtio_net: xsk: rx: support fill with xsk buffer")
Signed-off-by: Anton Protopopov <a.s.protopopov@gmail.com>
Link: https://patch.msgid.link/20260810120728.47445-1-a.s.protopopov@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 14:20:34 -07:00
Qingfang Deng
92c1bf630a pppox: drain queued packets on channel handoff
PPPIOCGCHAN both returns the channel index and marks a PPPOX socket as
bound to generic PPP, despite its getter semantic. Packets received
before that transition are queued on sk_receive_queue, but a bound
socket is no longer readable. Such packets therefore remain queued until
the socket is destroyed.

After marking a socket bound, wait for receive paths that observed the
old state to finish queueing packets, and then drain the queue into
generic PPP.

Fixes: 1da177e4c3 ("Linux-2.6.12-rc2")
Signed-off-by: Qingfang Deng <qingfang.deng@linux.dev>
Link: https://patch.msgid.link/20260811035314.302878-1-qingfang.deng@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 14:00:30 -07:00
Vladimir Oltean
4f1d06cf8a net: dsa: b53: fix error propagation from b53_fdb_dump()
The blamed commit replaced "return ret" statements in b53_fdb_dump()
with "break;" which jumps to the mutex_unlock() -> return 0 section.

This is notably problematic because it swallows errors from the
b53_fdb_copy() -> cb() path, and this will result in FDB dump truncation
when the netlink skb overflows - see commit 21b52fed92 ("net: dsa:
sja1105: fix broken backpressure in .port_fdb_dump").

Let's go back to "return ret". We don't need to preinitialize "ret" with
0, because the "do {} while" block guarantees we cannot reach the end of
the function without at least once calling b53_arl_search_wait(), which
will have initialized ret to some valid value.

Fixes: f7eb4a1c08 ("net: dsa: b53: serialize access to the ARL table")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://patch.msgid.link/20260812201121.2012356-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 13:48:16 -07:00
Jakub Kicinski
1e3d24fc42 Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue
Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2026-08-11 (ice, idpf)

For ice:
Przemyslaw Korba adds a fallback path to utilize sideband queue when the
low-latency PHY timer writes time out, likely, due to NVM updates or EMP
resets.

Petr Oros restores call to ice_clear_dflt_vsi() for the default VSI in
ice_vsi_release() to ensure no stale rules are left in the device which
can cause various issues.

Robert Malz breaks NVM reads to occur under separate lock requests as
large combined NVM reads can cause the NVM lock to be held longer than
maximum allowed time and be reclaimed by firmware.

For idpf:
Willem de Bruijn fixes possible endian issue of descriptor by adding
cpu_to_le32() call.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  idpf: add missing cpu_to_le32 in idpf_tx_splitq_build_flow_desc
  ice: acquire NVM lock around each flash read
  ice: clear the default forwarding VSI rule when releasing a VSI
  ice: fall back to SBQ when LL PHY timer interface times out
====================

Link: https://patch.msgid.link/20260812000918.220714-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 13:31:00 -07:00
Kyle Zeng
d5d4a7b538 vxlan: keep the last remote linked during FDB flush
A non-nexthop FDB entry is expected to have at least one remote while it
remains reachable through the FDB hash table. A filtered bulk flush
violates this invariant when every remote matches: It unlinks the last
remote in vxlan_fdb_dst_destroy() and only afterwards tells vxlan_flush()
to destroy the parent FDB entry.

An RCU reader can find the parent during this interval.
first_remote_rcu() then applies list_entry_rcu() to the empty list head,
producing an invalid remote pointer that the receive learning path can
read from and write to.

When a matching remote is the sole remaining remote, leave it linked and
ask the caller to destroy the entire FDB entry. vxlan_fdb_destroy() keeps
the remote attached while sending the deletion notification and removing
the parent from the lookup structures.

Fixes: c499fccb71 ("vxlan: vxlan_core: Support FDB flushing by destination VNI")
Cc: stable@vger.kernel.org
Signed-off-by: Kyle Zeng <kylebot@openai.com>
Co-developed-by: David Lee <david.lee@trailofbits.com>
Signed-off-by: David Lee <david.lee@trailofbits.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260810144115.821654-1-david.lee@trailofbits.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 13:21:33 -07:00
Prabu Thayalan
5da6ec6f06 ionic: fix completion descriptor access with 2x desc size
The old ionic_rx_service() and ionic_tx_service() used array
indexing to access completion descriptors:

    comp = &((struct ionic_rxq_comp *)cq->base)[cq->tail_idx];

This assumes the stride is sizeof(struct ionic_rxq_comp) = 16 bytes.
However, when the IONIC_Q_F_2X_CQ_DESC flag is set, the actual
completion descriptor size is 32 bytes (2 * sizeof(comp)), and the
completion itself is located at the end of that 32-byte slot. Array
indexing with a 16-byte stride would access the wrong offset.

Use pointer arithmetic that accounts for the actual descriptor size
from cq->desc_size:

    comp = cq->base +
           cq->desc_size * cq->tail_idx +
           cq->desc_size - sizeof(*comp);

This correctly calculates the completion location regardless of
descriptor size. For the common case where desc_size equals
sizeof(*comp), use array indexing in a likely() fast path to avoid
performance regression.

Fixes: 65e548f6b0 ("ionic: remove the cq_info to save more memory")
Signed-off-by: Prabu Thayalan <prabu.ponrajthayalan@amd.com>
Signed-off-by: Eric Joyner <eric.joyner@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260811195039.1315045-3-eric.joyner@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 12:58:46 -07:00
Eric Joyner
621f44af8c ionic: add missing dma_rmb() after the completion publish check
Each completion service routine tests a device-written publish flag and
then reads the rest of the descriptor with nothing ordering those loads.
A control dependency does not order loads, so a weakly ordered CPU may
satisfy the payload reads from a cache line state observed before the
flag became valid. Add the barrier to all four completion paths.

Fixes: 1d062b7b6f ("ionic: Add basic adminq support")
Fixes: 0f3154e6bc ("ionic: Add Tx and Rx handling")
Fixes: 77ceb68e29 ("ionic: Add notifyq support")
Signed-off-by: Eric Joyner <eric.joyner@amd.com>
Reviewed-by: Brett Creeley <brett.creeley@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260811195039.1315045-2-eric.joyner@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 12:58:46 -07:00
Runyu Xiao
43b0213529 net: ibm: emac: mal: fix NAPI locking
Since commit 413f0271f3 ("net: protect NAPI enablement with
netdev_lock()"), napi_enable() and napi_disable() take netdev_lock().

mal_register_commac() and mal_unregister_commac() call these helpers
while holding mal->lock with interrupts disabled. In the unregister
path, napi_disable() may also wait for polling to finish, while the poll
completion path takes mal->lock.

Take netdev_lock() before mal->lock, use the locked NAPI helpers, and
drop mal->lock before napi_disable_locked().

Fixes: 413f0271f3 ("net: protect NAPI enablement with netdev_lock()")
Cc: stable@vger.kernel.org
Signed-off-by: Runyu Xiao <runyu.xiao@seu.edu.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260811070813.377573-1-runyu.xiao@seu.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 12:47:07 -07:00
Fan Gong
d0c2bed692 hinic3: Fix skb linearization mismatch and drop skb when skb_checksum_help() failed
Previously, hinic3_send_one_skb() cached the skb fragment count before
calling hinic3_tx_offload(). If hinic3_tx_csum() falls back to
skb_checksum_help() for unsupported tunnel packets, the skb may be
linearized. Continuing to build the TX descriptor with the stale
fragment count leads to a descriptor mismatch, which can trigger
out-of-bounds DMA reads or IOMMU faults.

Furthermore, the old code ignored the return value of skb_checksum_help(),
transmitting corrupted packets with incomplete checksums upon failure.

Fix this by:
1. Moving the hinic3_tx_offload() call before calculating 'num_sge' to
   ensure the correct fragment count is used if the SKB is linearized.
2. Propagating skb_checksum_help() errors and returning
   HINIC3_TX_OFFLOAD_INVALID to properly drop the skb.

Fixes: 17fcb3dc12 ("hinic3: module initialization and tx/rx logic")
Co-developed-by: Teng Peisen <tengpeisen@huawei.com>
Signed-off-by: Teng Peisen <tengpeisen@huawei.com>
Co-developed-by: Wu Di <wudi234@huawei.com>
Signed-off-by: Wu Di <wudi234@huawei.com>
Signed-off-by: Fan Gong <gongfan1@huawei.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/78d8c61cab588240948eaddcb437d59add9f77ae.1786448013.git.tengpeisen@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 11:28:56 -07:00
Fan Ye
3c8b26ebf5 net: thunderbolt: Mark the connection down when bringing it up fails
Every failure path in tbnet_connected_work() undoes its own work and
returns without clearing login_sent, so the connection still looks
established. The next tbnet_tear_down() therefore takes its main branch
and repeats a teardown that already happened: it stops rings that are
already stopped, which is a dev_WARN() and fatal under panic_on_warn,
and it releases net->remote_transmit_path even on the HopID mismatch
path, where this connection never owned that id, silently freeing one
that someone else is still using.

Clear login_sent on those paths. That is enough for tbnet_tear_down() to
leave the unwound state alone, and login_received has to stay set: it
records that the peer has logged in and carries the transmit path it gave
us, which nothing on this side can make the peer send again. Two things
change beyond keeping the teardown out of the way: the logout request in
that block is no longer sent, and the peer's next login request now
re-queues our login work rather than connected_work, giving the
connection a fresh login instead of a retry on stale state.

Fixes: e69b6c02b4 ("net: Add support for networking over Thunderbolt cable")
Cc: <stable@vger.kernel.org> # 5.13+
Signed-off-by: Fan Ye <fy15309206903@gmail.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260811-b4-tbnet-hopid-v3-2-9e75d1b51331@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 11:14:05 -07:00
Fan Ye
2f1463554d net: thunderbolt: Release the Rx HopID that was handed out on mismatch
tb_xdomain_alloc_in_hopid() passes the wanted HopID to ida_alloc_range()
as the lower bound, so a taken id is not an error there: the allocator
returns the next free one above it. tbnet_connected_work() asks for the
peer's transmit path, treats any other id as a failure and returns
without releasing what it got, so that allocation stays live for the rest
of the XDomain connection with nothing left holding a reference to it.

Release the id when it is not the one we asked for, the same way the
error unwind at the end of the function releases the expected one.

Fixes: 180b068942 ("thunderbolt: Allow multiple DMA tunnels over a single XDomain connection")
Cc: stable@vger.kernel.org
Signed-off-by: Fan Ye <fy15309206903@gmail.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260811-b4-tbnet-hopid-v3-1-9e75d1b51331@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 11:14:05 -07:00
Karl Mehltretter
4b92a3710d octeontx2-af: initialize lmac_bmap in rvu_mcs_set_lmac_bmap()
rvu_mcs_set_lmac_bmap() declares lmac_bmap without initializing it and
only sets bits for valid lmacs with set_bit(), which ORs into the word
without clearing it first. Bits for invalid or skipped ports keep
whatever was on the stack, and the garbage is stored into
mcs->hw->lmac_bmap.

Initialize lmac_bmap to 0 so only valid lmacs are marked.

Found with Clang's -Wconditional-uninitialized.

Fixes: ca7f49ff88 ("octeontx2-af: cn10k: Introduce driver for macsec block.")
Signed-off-by: Karl Mehltretter <kmehltretter@gmail.com>
Reviewed-by: Ratheesh Kannoth <rkannoth@marvell.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260812060730.6181-1-kmehltretter@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-17 11:06:24 -07:00
Asim Viladi Oglu Manizada
447c930394 net: tun: bound receive headroom
tun_get_user() uses tun->align both as skb headroom and when choosing how
much packet data to keep linear. OVS can propagate an oversized headroom
request from another port to TUN or TAP.

When align is larger than the usable space in a one-page skb head,
SKB_MAX_HEAD(align) underflows and the result becomes negative when stored
in good_linear. That value later wraps when assigned to the size_t linear
variable, and tun_alloc_skb() can place skb->data outside the allocated
head.

Bound the headroom stored by TUN to the one-page skb-head budget and the
largest non-sentinel 16-bit skb header offset. Leave one linear byte for
raw TUN and a complete Ethernet header for TAP, including NET_IP_ALIGN.

Also pull the raw-TUN protocol byte and the TAP Ethernet header before
accessing them, so these checks remain safe for nonlinear skbs supplied by
other allocation paths.

Fixes: eaea34b23c ("net/tun: implement ndo_set_rx_headroom")
Cc: stable@vger.kernel.org
Signed-off-by: Asim Viladi Oglu Manizada <manizada@pm.me>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260812012139.2134643-1-manizada@pm.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-13 17:35:08 -07:00
Mikhail Gavrilov
3aa1dcaa4f Revert "wifi: mt76: Disable napi when removing device"
This reverts commit 13b7e6a96a.

That commit made mt76_dma_cleanup() disable every RX NAPI instance before
deleting it, to silence WARNs in __netif_napi_del_locked() and
page_pool_disable_direct_recycling() seen when unloading mt7915e with an
MT7916.

On mt7921e and mt7925e the same instances are already disabled earlier,
in mt7921e_unregister_device() and mt7925e_unregister_device(), which
only afterwards call mt792x_dma_cleanup() -> mt76_dma_cleanup().  Each
instance is therefore disabled twice, and napi_disable() is not
idempotent: on return it leaves NAPIF_STATE_SCHED and NAPIF_STATE_NPSVC
set, so the second call spins in usleep_range() forever, waiting for bits
that nobody will clear.

mt7921_pci_shutdown() and mt7925_pci_shutdown() reuse the remove path, so
this is hit on every reboot, poweroff and module unload.  It is silent:
the stuck task keeps sleeping and rescheduling, so neither the hung task
detector nor the lockup detectors fire, and the last line on the console
is "systemd-shutdown[1]: Rebooting."

  task:modprobe        state:D stack:25720 pid:7954  tgid:7954
  Call Trace:
   <TASK>
   __schedule+0x11b8/0x26d0
   schedule+0xe7/0x2f0
   schedule_hrtimeout_range_clock+0x218/0x330
   usleep_range_state+0x133/0x1b0
   napi_disable_locked+0x37d/0x5f0
   napi_disable+0x43/0x80
   mt76_dma_cleanup+0x2b4/0x860 [mt76]
   mt7921_pci_remove+0x17f/0x350 [mt7921e]
   pci_device_remove+0xb6/0x1e0
   device_release_driver_internal+0x38d/0x540
   driver_detach+0xd0/0x1b0
   bus_remove_driver+0x127/0x2d0
   pci_unregister_driver+0x2a/0x280
   __do_sys_delete_module+0x36a/0x5b0
   do_syscall_64+0x11c/0x6d0
   entry_SYSCALL_64_after_hwframe+0x76/0x7e
   </TASK>

Dropping the two driver-side loops instead was tried and rejected: with
them gone, the RX poll can reach mt76_token_release() via
PKT_TYPE_TXRX_NOTIFY and mt7921_mac_tx_free() while
mt76_connac2_tx_token_put() is running idr_destroy(&dev->token) outside
token_lock, which is a use-after-free rather than a hang [1].

Revert for now, so that reboot, poweroff and module unload work again.
The WARNs on mt7915e are a less severe problem than an unbootable
machine, and fixing them belongs in the drivers that delete the NAPI
instances, where each one can pick a point that is safe for its own
teardown order, rather than in the shared mt76_dma_cleanup().

[ This is the "landing soonish" known regression fix mentioned in the
  previous networking merge commit       - Linus ]

Reported-by: Bert Karwatzki <spasswolf@web.de>
Closes: https://lore.kernel.org/all/20260724151419.26014-1-spasswolf@web.de/
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=221818
Link: https://lore.kernel.org/all/20260730050428.GA73812@sol/ [1]
Signed-off-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com>
Acked-by: Nicolas Cavallari <nicolas.cavallari@green-communications.fr>
Fixes: 13b7e6a96a ("wifi: mt76: Disable napi when removing device")
Tested-by: Devin Wittmayer <lucid_duck@justthetip.ca>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2026-08-13 09:00:55 -07:00
Jakub Kicinski
2bb155e921 Included fixes:
* release key slot crypto transforms from a workqueue rather than an RCU
   callback, because crypto_free_aead() may sleep with async or hardware
   implementations
 * run all deferred ovpn work on a module-owned workqueue and drain it on
   module exit, so no work item can still be executing module text after
   the module is unloaded
 * finish crypto callback cleanup (key slot release and leftover skb)
   before dropping the peer reference that gates netdev unregistration
   and module removal
 * avoid dereferencing a NULL key slot when userspace asks to kill a key
   that is not installed on the peer
 -----BEGIN PGP SIGNATURE-----
 
 iJEEABYIADkWIQQr0db7q+Rc7Zog28Fc8QQzwdnOtwUCanjnnRsUgAAAAAAEAA5t
 YW51MiwyLjUrMS4xMiwyLDIACgkQXPEEM8HZzre4UwEAqIw3zlYNPzJeNRq4GB0b
 is5pBp/ZWiWIpJjdNSHmtpgA/1/1oBil7ZTdM0KhUCcCR1LHIXLb2sJOFUfhzowc
 QO4K
 =qG8X
 -----END PGP SIGNATURE-----

Merge tag 'ovpn-net-20260809' of https://github.com/OpenVPN/ovpn-net-next

Antonio Quartulli says:

====================
Included fixes:

* release key slot crypto transforms from a workqueue rather than an RCU
  callback, because crypto_free_aead() may sleep with async or hardware
  implementations
* run all deferred ovpn work on a module-owned workqueue and drain it on
  module exit, so no work item can still be executing module text after
  the module is unloaded
* finish crypto callback cleanup (key slot release and leftover skb)
  before dropping the peer reference that gates netdev unregistration
  and module removal
* avoid dereferencing a NULL key slot when userspace asks to kill a key
  that is not installed on the peer

* tag 'ovpn-net-20260809' of https://github.com/OpenVPN/ovpn-net-next:
  ovpn: defer key slot crypto freeing to workqueue
  ovpn: run deferred work on a module-owned workqueue
  ovpn: finish crypto callback cleanup before peer release
  ovpn: fix NULL dereference when killing missing key
====================

Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260809212142.2249027-1-antonio@openvpn.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-12 18:09:18 -07:00
Siddharth Vadapalli
36a05d2820 net: ethernet: ti: am65-cpsw-nuss: Fix port_id extraction from SRC TAG
On the packet reception path, the ID of the MAC Port on which the packet
was received, is embedded in the RX DMA Descriptor's metadata. The ID is
extracted using the helper function cppi5_desc_get_tags_ids() which fills
in the 16-bit Source Tag into the 'port_id' variable. However, it is only
the lower 8-bits of the 16-bit Source Tag that represent the MAC Port ID,
while the upper 8-bits are Hardware-Reserved and carry an arbitrary value.
With the existing logic, sporadic kernel crash is observed due to the
subsequent driver code accessing out-of-bound memory because of an invalid
port_id.

Hence, fix the port_id extraction logic to use only the lower 8-bits of the
Source Tag as the MAC Port ID.

Fixes: 93a7653031 ("net: ethernet: ti: introduce am65x/j721e gigabit eth subsystem driver")
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Chintan Vankar <c-vankar@ti.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260807111738.2055900-1-s-vadapalli@ti.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-12 17:36:19 -07:00
Jordan Rhee
3992ced109 gve: fix NULL dereference due to missing ptp adjfine
Fix NULL dereference due to missing implementation of adjfine, which can
be triggered from usermode as follows:

sudo ./testptp -d /dev/ptp0 -f 0
[  551.943697] BUG: kernel NULL pointer dereference, address: 0000000000000000
[...]
[  552.061946] Call Trace:
[  552.064487]  <TASK>
[  552.066681]  ptp_clock_adjtime+0x1c0/0x2c0
[  552.070874]  ? get_clock_desc+0x6b/0xb0
[  552.074825]  pc_clock_adjtime+0x78/0xc0
[  552.078755]  __do_sys_clock_adjtime+0x85/0x110
[  552.083293]  do_syscall_64+0xea/0x610

Cc: stable@vger.kernel.org
Fixes: acd1638052 ("gve: Add initial PTP device support")
Signed-off-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Harshitha Ramamurthy <hramamurthy@google.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Link: https://patch.msgid.link/20260807224315.234152-3-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-11 18:26:04 -07:00
Jordan Rhee
6bf14575c6 gve: fix zero-length skb frag with header-split
When header split is enabled and a header-only packet is
received such as a pure TCP ACK, GVE will indicate an
RX SKB with a zero-length fragment. If this SKB is then
hairpinned and sent back out, the GVE TX path will emit
a zero-length descriptor. Hardware considers this
an illegal descriptor and stops the queue, causing a
TX timeout and interface reset.

Fix it by not adding the zero-length skb frag.

Cc: stable@vger.kernel.org
Fixes: 5e37d8254e ("gve: Add header split data path")
Suggested-by: Praveen Kaligineedi <pkaligineedi@google.com>
Co-developed-by: Ziwei Xiao <ziweixiao@google.com>
Signed-off-by: Ziwei Xiao <ziweixiao@google.com>
Signed-off-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Harshitha Ramamurthy <hramamurthy@google.com>
Link: https://patch.msgid.link/20260807224315.234152-2-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-11 18:26:04 -07:00
Willem de Bruijn
b802a8c1ca idpf: add missing cpu_to_le32 in idpf_tx_splitq_build_flow_desc
idpf_tx_splitq_build_flow_desc performs a 32-bit store to &cmd_dtype
to set the 8-bit cmd_dtype and zero the adjacent 3-byte timestamp
field in a single operation.

Descriptors are in little endian. Add missing cpu_to_le32 and cast to
__le32 to ensure the fields are written correctly also on big endian
platforms.

Fixes: 1a49cf814f ("idpf: add Tx timestamp flows")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Jason Xing <kerneljasonxing@gmail.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-08-11 13:22:57 -07:00
Robert Malz
df88d6f1ed ice: acquire NVM lock around each flash read
FW caps the NVM read lock at a maximum of 3000ms regardless of the timeout
requested via ice_acquire_nvm(). ice_read_flat_nvm() splits a read into
multiple ice_aq_read_nvm() commands, one per 4KB sector, all issued under a
single lock taken by the caller. Reading a large region can exceed 3000ms,
so FW reclaims the lock mid-read and the remaining commands might fail.

Move the lock acquire/release into ice_read_flat_nvm() so it brackets each
individual ice_aq_read_nvm() command, ensuring the lock is never held
across more than one FW read.

ice_release_nvm() issues its own AQ command and overwrites
hw->adminq.sq_last_status, which some callers inspect after a failed read.
Add an optional read_aq_err output parameter to ice_read_flat_nvm() to
capture the failing read's AQ error before the release; callers that need
it (ice_discover_flash_size() and the ethtool/devlink log paths) use it
instead of sq_last_status, others pass NULL.

Callers that previously took the lock around ice_read_flat_nvm(),
ice_read_sr_word() or ice_read_flash_module() now call them without it.
The now-redundant per-block locking in ice_devlink_nvm_snapshot() is
dropped. ice_read_sr_word() is now a thin wrapper, so ice_read_sr_word_aq()
is folded into it.

Fixes: e94509906d ("ice: create function to read a section of the NVM and Shadow RAM")
Signed-off-by: Robert Malz <robert.malz@canonical.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Marcin Szycik <marcin.szycik@linux.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>
2026-08-11 13:22:57 -07:00
Petr Oros
d04287e27b ice: clear the default forwarding VSI rule when releasing a VSI
When a VSI is configured as the switch's default forwarding VSI
(ICE_SW_LKUP_DFLT) and is then torn down, the rule is left behind in
the switch. ice_vsi_release() no longer removes it, and the SR-IOV VF
free path (ice_free_vfs() -> ice_free_vf_res() -> ice_vf_vsi_release()
-> ice_vsi_release()) does not disable promiscuous mode either, which
only happens on VF reset in ice_vf_clear_all_promisc_modes().

A trusted VF that enters unicast promiscuous mode becomes the default
forwarding VSI (this is the default mode, when the PF does not have VF
true-promiscuous mode enabled). If the VFs are then destroyed without
the VF first leaving promiscuous mode, the ICE_SW_LKUP_DFLT rule for
the now-freed VSI is leaked. When VFs are recreated, a VSI reuses the
freed hw_vsi_id. If it is assigned a different VSI handle than the
leaked rule holds, ice_set_dflt_vsi() does not recognize it as
already-default, and ice_add_update_vsi_list() folds the dangling
(freed) handle into a VSI list, which the firmware rejects. The VSI
handle assigned on re-creation varies, so the failure is intermittent
rather than every cycle.

Reproduce by repeatedly running the cycle below on the two ports of the
same card, where $VF0 and $VF1 are the netdevs of vf 15 once they
appear. The VF must be brought up so iavf actually pushes the unicast
promiscuous request, and the rule must settle before the VFs are torn
down again:

  echo 16 > /sys/class/net/$PF0/device/sriov_numvfs
  echo 16 > /sys/class/net/$PF1/device/sriov_numvfs
  ip link set $PF0 vf 15 trust on
  ip link set $PF1 vf 15 trust on
  ip link set $VF0 up
  ip link set $VF1 up
  ip link set $VF0 promisc on
  ip link set $VF1 promisc on
  sleep 1
  echo 0 > /sys/class/net/$PF0/device/sriov_numvfs
  echo 0 > /sys/class/net/$PF1/device/sriov_numvfs

Within a few cycles the ice PF and iavf VF log:

  Failed to set VSI 25 as the default forwarding VSI, error -22
  Turning on/off promiscuous mode for VF 63 failed, error: -22
  PF returned error -53 (IAVF_ERR_ADMIN_QUEUE_ERROR) to our request 14

This cleanup used to live in ice_vsi_release() but was dropped by the
referenced refactor. Restore it. Clear the default forwarding VSI rule
in ice_vsi_release() when this VSI owns it, which covers every teardown
path.

Fixes: 6624e780a5 ("ice: split ice_vsi_setup into smaller functions")
Signed-off-by: Petr Oros <poros@redhat.com>
Reviewed-by: Marcin Szycik <marcin.szycik@linux.intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-08-11 13:22:57 -07:00
Przemyslaw Korba
6959fbdc94 ice: fall back to SBQ when LL PHY timer interface times out
The low-latency (LL) PHY timer interface relies on a tight, atomic poll
of the PF_SB_ATQBAL register with a 2ms timeout. After an NVM update /
EMPR, FW may need significantly longer than 2ms to start responding to
ATQBAL commands. The first PHY adjust or incval write issued by
ice_ptp_rebuild_owner() fails with -ETIMEDOUT.

Fix this by falling back to the existing SBQ-based PHY register write
path when LL times out. This makes sure PTP is initialized when FW takes
longer than expected to come back online.

Steps to reproduce:
./nvmupdate64e -if devlink -f
Update E810 card with nvmupdate64e, and observe dmesg errors:
  Failed to write PHC increment value, status -110
  PTP reset failed, error: -110 (-ETIMEDOUT)

Fixes: ef9a64c072 ("ice: implement low latency PHY timer updates")
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>
2026-08-11 13:22:56 -07:00
Jiawen Wu
5f3a13e0bb net: ngbe: fix NULL pointer dereference in non-MSI-X interrupt enabling
In non-MSI-X mode (such as legacy INTx or single MSI), wx->msix_entry is
not allocated or initialized. Calling NGBE_INTR_MISC(wx) dereferences
wx->msix_entry->entry, leading to a NULL pointer dereference crash.

This issue was introduced by fixing the IRQ vector when the number of
VFs is 7. Fix the issue by explicitly checking `pdev->msix_enabled` to
determine the correct vector index.

Additionally, as a side fix, set the interrupt mask to BIT(0) for the
non-MSI-X fallback. In MSI/INTx mode, the MISC and queue interrupts
share vector 0, and the WX_PX_MISC_IVAR register is only valid in the
MSI-X case. Thus, BIT(0) is the correct mask for the miscellaneous cause
when MSI-X is disabled.

Fixes: 4174c0c331 ("net: ngbe: specify IRQ vector when the number of VFs is 7")
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/B2693E9A8BFAD110+20260807062214.410838-1-jiawenwu@trustnetic.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-11 13:25:03 +02:00
Wei Fang
cbb35cbe8d net: tap: fix wrong transport_header when sending VLAN-tagged frame
In tap_get_user_xdp(), when processing a VLAN-tagged frame (e.g.
ETH_P_8021Q), skb_set_network_header() is called first to advance
network_header past the VLAN tag to the inner protocol header.
skb_probe_transport_header() is then called with skb->protocol still
set to ETH_P_8021Q, while nhoff (derived from skb_network_offset())
already points past the VLAN tag to the inner protocol header.

In __skb_flow_dissect(), proto is initialized to ETH_P_8021Q and nhoff
points past the VLAN tag. When the dissector hits case ETH_P_8021Q, it
reads a struct vlan_hdr at the current nhoff via __skb_header_pointer(),
but that offset contains the inner protocol header (e.g. an IP header).
The bytes are misinterpreted as a VLAN header, yielding a garbage
encapsulated EtherType that matches no known protocol. The dissector
returns false, so skb_probe_transport_header() never calls
skb_set_transport_header(), leaving transport_header at its uninitialized
sentinel value (~0U).

Move skb_set_network_header() to after skb_probe_transport_header(). At
the time skb_probe_transport_header() is called, network_header still
points to the VLAN header (offset ETH_HLEN), so nhoff is correct and the
flow dissector can parse the VLAN header, extract the inner EtherType,
and advance nhoff to the inner protocol header, allowing transport_header
to be set correctly.

Fixes: 8c76e77f90 ("tap: call skb_probe_transport_header after setting skb->dev")
Assisted-by: WChat:claude-opus-4-8
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260807063405.688780-3-wei.fang@oss.nxp.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-11 13:14:00 +02:00
Baul Lee
b37971686e vxlan: do not arm the ageing timer on a device that is down
vxlan_changelink() arms vxlan->age_timer whenever the requested ageing
interval differs from the configured one:

	if (conf.age_interval != vxlan->cfg.age_interval)
		mod_timer(&vxlan->age_timer, jiffies);

There is no netif_running() test, so the timer is armed even on a device
that was never brought up.  The only synchronous cancel in the driver is
the timer_delete_sync() in vxlan_stop(), which is .ndo_stop.
netif_close_many() drops devices without IFF_UP before
__dev_close_many() runs, so that cancel is skipped for such a device.

vxlan_setup() sets dev->needs_free_netdev = true and age_timer is a
member of struct vxlan_dev, so free_netdev() releases the allocation the
timer lives in while it is still queued on a timer_base.
expire_timers() unlinks the entry before it loads timer->function, so
the timer core writes through the freed object's list pointers:

  BUG: KASAN: slab-use-after-free in __run_timers+0x208/0x654
  Write of size 8 at addr ffff00001adace68 by task true/192
   __asan_store8+0x84/0xac
   __run_timers+0x208/0x654
   run_timer_softirq+0x154/0x18c
  Allocated by task 189:
   alloc_netdev_mqs+0x64/0x720
   rtnl_create_link+0x4ac/0x520
   rtnl_newlink+0x758/0xd00
  Freed by task 191:
   netdev_release+0x40/0x58
   netdev_run_todo+0x4a4/0x8c0
   rtnl_dellink+0x200/0x4e8

The rtnl operations involved are netns-scoped, so an unprivileged user
can perform them in a new user and network namespace.

Arming the timer on a down device never had an effect: vxlan_cleanup()
returns early on !netif_running(), and vxlan_open() arms the timer for
any non-zero interval once the device is brought up.  Add the missing
test.

Discovered by XBOW, triaged by Baul Lee <baul.lee@xbow.com>

Fixes: 40051c4dca ("vxlan: Allow changing ageing time")
Cc: stable@vger.kernel.org
Signed-off-by: Baul Lee <baul.lee@xbow.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260809111829.78834-1-baul.lee@xbow.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-11 11:42:52 +02:00
Koichiro Den
d2121faf13 NTB: ntb_netdev: Preserve RX queue depth on allocation failure
ntb_netdev_rx_handler() hands the received skb to the network stack
before allocating its replacement. If the allocation fails, nothing is
reposted. Every failure therefore takes one buffer out of the RX queue
while the interface remains up, and enough failures eventually stall
reception.

A retry path could refill the queue later, but ntb_netdev has none.
Allocate the replacement first instead. If that fails, drop the packet
and repost the same skb. This keeps the queue full and lets packet
delivery resume as soon as memory is available again.

Fixes: 548c237c0a ("net: Add support for NTB virtual ethernet device")
Cc: stable@vger.kernel.org
Signed-off-by: Koichiro Den <den@valinux.co.jp>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Link: https://patch.msgid.link/20260806032537.3526498-1-den@valinux.co.jp
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-11 11:04:55 +02:00
Oleksij Rempel
202fef9bbb net: phy: realtek: fix EEE advertisement write on the internal PHY MMD path
In rtlgen_write_mmd(), the MDIO_AN_EEE_ADV case swaps the arguments to
rtlgen_write_vend2(): it passes the MMD register number as the OCP address
and the OCP address constant as the value. The caller's value is discarded
and the write lands on the wrong register, so the EEE advertisement cannot
be configured on the affected PHYs.

Mirror rtlgen_read_mmd() and write the value to RTL_MDIO_AN_EEE_ADV.

Fixes: da681ed73f ("net: phy: realtek: improve mmd register access for internal PHY's")
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Nicolai Buchwitz <nb@tipi-net.de>
Link: https://patch.msgid.link/20260806134716.3511821-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 16:05:05 -07:00
Jonas Köppeler
60db47f02b veth: fix queue index used to wake the peer txq in veth_poll
veth_poll() derives the index of the peer TX queue to wake from
rq->xdp_rxq.queue_index. That field is only initialized by
xdp_rxq_info_reg() in veth_enable_xdp_range(), which runs only when an
XDP program is attached. On the plain GRO/NAPI path
(veth_napi_enable_range()) xdp_rxq_info_reg() is never called, so
queue_index stays 0 for every queue, as priv->rq is zero-allocated.

So in a multi-queue setup with GRO enabled and no XDP program attached,
every NAPI instance looks at the peer's TX queue 0. If veth_xmit() stops
peer TX queue 1 because the ptr_ring is full (NETDEV_TX_BUSY), nothing
ever wakes it again: the poller draining queue 1 wakes queue 0 instead.
veth implements no ndo_tx_timeout, so the netdev watchdog does not kick
in either, and the queue stays stopped indefinitely.

Derive the index from the position of the rq within priv->rq instead,
which is correct regardless of whether XDP was ever enabled.

Scripts to reproduce the stall are available at
https://github.com/netoptimizer/veth-backpressure-performance-testing

Fixes: dc82a33297 ("veth: apply qdisc backpressure on full ptr_ring to reduce TX drops")
Signed-off-by: Jonas Köppeler <j.koeppeler@tu-berlin.de>
Tested-by: Jesper Dangaard Brouer <hawk@kernel.org>
Acked-by: Jesper Dangaard Brouer <hawk@kernel.org>
Link: https://patch.msgid.link/20260806-veth-fix-poll-queue-idx-v1-1-c5357fb7573d@tu-berlin.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:57:47 -07:00
Eric Dumazet
cef51860be macvlan: inherit needed_headroom and needed_tailroom from lowerdev
macvlan devices inherit hard_header_len from lowerdev during macvlan_init(),
but leave needed_headroom and needed_tailroom set to 0.

When the underlying lowerdev requires extra headroom or tailroom for
headers/trailers (e.g. macsec, ipsec, wireguard, tunnels, or veth with rx
headroom), upper layers calculating packet headroom and tailroom fail to
reserve sufficient space.

This can result in reallocation overhead, skb headroom underflows, or KASAN
slab-use-after-free crashes when dev_hard_header() / macvlan_hard_header()
prepends header data or when lower devices append tailroom.

Fix this by:
1. Inheriting needed_headroom and needed_tailroom from lowerdev in macvlan_init().
2. Propagating needed_headroom and needed_tailroom updates to attached macvlans
   in macvlan_device_event() when receiving NETDEV_FEAT_CHANGE events.

Fixes: b863ceb7dd ("[NET]: Add macvlan driver")
Reported-by: Tangxin Xie <xietangxin@h-partners.com>
Closes: https://lore.kernel.org/netdev/CANn89i+1EW-sFNK8xoq98gMbPCeLS7e=+rs9gHfLg5Wj+4x0sw@mail.gmail.com/T/#m16adf0ff972cbfd8066c3a8e656e75eaeb12d021
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Hangbin Liu <liuhangbin@kylinos.cn>
Link: https://patch.msgid.link/20260806141938.287660-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:21:41 -07:00
Eric Dumazet
e16e960d55 ipvlan: inherit needed_headroom and needed_tailroom from phy_dev
ipvlan devices inherit hard_header_len from phy_dev during ipvlan_init(),
but leave needed_headroom and needed_tailroom set to 0.

When the underlying phy_dev (or stacked lower device) requires extra headroom
or tailroom for headers/trailers (e.g. macsec, ipsec, wireguard, tunnels, or
veth with rx headroom), upper layers calculating packet headroom and tailroom
fail to reserve sufficient space.

This can result in reallocation overhead, skb headroom underflows, or KASAN
slab-use-after-free crashes when dev_hard_header() / ipvlan_hard_header()
prepends header data or when lower devices append tailroom.

Fix this by:
1. Inheriting needed_headroom and needed_tailroom from phy_dev in ipvlan_init().
2. Propagating needed_headroom and needed_tailroom updates to attached ipvlans
   in ipvlan_device_event() when receiving NETDEV_FEAT_CHANGE events.

Fixes: 2ad7bf3638 ("ipvlan: Initial check-in of the IPVLAN driver.")
Reported-by: syzbot+1f9fd0f4b601cf88d6e6@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6a720a21.40259c87.584f4.04bb.GAE@google.com/T/#u
Reported-by: Tangxin Xie <xietangxin@h-partners.com>
Closes: https://lore.kernel.org/netdev/CANn89i+1EW-sFNK8xoq98gMbPCeLS7e=+rs9gHfLg5Wj+4x0sw@mail.gmail.com/T/#mcc6307f115e500df23ea2980d5669fe95f20b6b4
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Hangbin Liu <liuhangbin@kylinos.cn>
Link: https://patch.msgid.link/20260806103857.115541-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:20:06 -07:00
Jakub Kicinski
51e96fa31f eth: bnxt: avoid deadlock when canceling IRQ affinity notifier
Unregistering IRQ affinity notifiers waits for the callback synchronously.
bnxt takes the netdev instance lock in the notifier (to restart the queue)
and cancels the work under the same lock. This may obviously deadlock.

Move the restart to the async service task. The queue restart isn't
super time sensitive. Store the new TPH tag, schedule the task.
Safely canceling the service task is already ironed out.

In bnxt_request_irq() the order of registering notifier, affinity and
initial TPH programming has to be inverted. I think it was racy
previously since user may trigger an update as soon as notifier
is installed.

There's a small known gap - if pcie_tph_get_cpu_st() fails at init
and the target tag is 0 we may miss programming the entry.
This does not seem worth fixing, the code has skip-on-failure
all over the place, anyway.

Fixes: c214410c47 ("bnxt_en: Add TPH support in BNXT driver")
Tested-by: Vishvambar Panth S <vishvambar.panth-s@broadcom.com>
Link: https://patch.msgid.link/20260803193135.2030368-5-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:15:06 -07:00
Jakub Kicinski
0fd562c7b9 eth: bnxt: decrease indent in bnxt_request_irq()
bnxt_request_irq() has unnecessary level of indentation.
Use continue instead. No need to re-fetch NUMA node for
each IRQ, move to the function level.

No functional changes.

Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260803193135.2030368-4-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:15:06 -07:00
Jakub Kicinski
4b5cb58a44 eth: bnxt: keep the aRFS rmap updated when TPH is enabled
The TPH support must have broken aRFS in bnxt. IRQ can only have one
notifier, so installing the TPH notifier is overriding the one implicitly
installed by irq_cpu_rmap_add().

Make sure we call cpu_rmap_update() from the TPH notifier.

We need to be careful with the ordering and not free the rmap
until we unregistered the notifier. Note that moving the rmap
freeing after the early return in bnxt_free_irq() is fine -
there's no path that could leave rmap with irq_tbl being NULL.

Fixes: c214410c47 ("bnxt_en: Add TPH support in BNXT driver")
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20260803193135.2030368-3-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:15:06 -07:00
Jakub Kicinski
c703f62c30 eth: bnxt: cancel IRQ notifier before freeing affinity mask
bnxt_irq_affinity_notify() copies into irq->cpu_mask.
Cancel the notifier before freeing irq->cpu_mask.

Fixes: c214410c47 ("bnxt_en: Add TPH support in BNXT driver")
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20260803193135.2030368-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-10 15:15:05 -07:00
Ralf Lici
2da3dfa1dd ovpn: defer key slot crypto freeing to workqueue
Key slots are released through a kref and the existing release path
frees the AEAD transforms from an RCU callback. That is not safe for all
crypto implementations: crypto_free_aead can sleep, for example when an
async or hardware implementation has teardown work to complete.

Use queue_rcu_work for key-slot release. This keeps the RCU grace period
needed by lockless key-slot readers, but runs the actual crypto teardown
from workqueue context where sleeping is allowed. Once the rcu_work
callback runs, pre-existing RCU readers are gone, and the final kref put
already proves that no transform user remains, so the worker can release
the AEAD transforms and free the slot directly.

The previous patch drains ovpn_wq during module exit, so queued key-slot
teardown work cannot outlive module text.

Fixes: 8534731dbf ("ovpn: implement packet processing")
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
2026-08-09 22:47:57 +02:00
Ralf Lici
e9714db804 ovpn: run deferred work on a module-owned workqueue
ovpn queues several work items whose callbacks execute module text.
These works currently run on the global system workqueues, so module
exit has no driver-owned drain point that guarantees the callbacks have
fully returned before the module text can be freed.

Object references protect the objects used by the callbacks, but they do
not prove that a workqueue function has returned. In particular, a
worker can drop the final reference that unblocks device teardown while
it is still executing ovpn code.

Add a module-owned workqueue and queue all ovpn work items on it. During
module exit, unregister rtnl and netlink first, flush the workqueue so
ordinary ovpn workers finish, run the final RCU barrier, and destroy the
workqueue last. This keeps the workqueue available for cleanup work
queued from RCU callbacks, while ensuring no ovpn work item can outlive
the module text.

The per-device delayed keepalive work remains explicitly disabled during
netdev teardown (disable_delayed_work_sync in ndo_uninit), since
flush_workqueue does not flush delayed work that is still only pending
on its timer.

Fixes: 3ecfd9349f ("ovpn: implement keepalive mechanism")
Fixes: 11851cbd60 ("ovpn: implement TCP transport")
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
2026-08-09 22:47:57 +02:00
Sun Jian
cb6379feaa veth: fix skb length accounting after XDP frag adjustment
veth exposes non-linear skb fragments through an xdp_buff. If an XDP
program adjusts the fragment area, veth_xdp_rcv_skb() copies
xdp_frags_size back to skb->data_len but leaves skb->len containing the
old fragment contribution.

After a fragment shrink, this makes skb_headlen() larger than the actual
linear area. In the reproduced UDP receive path, __skb_datagram_iter()
copied 1024 bytes past the actual linear tail to userspace, starting at
struct skb_shared_info. The copied bytes included the affected skb's
nr_frags, xdp_frags_size, and a kernel pointer from
skb_shinfo(skb)->frags[0]. Real packet data was displaced by the same
amount and truncated at the end.

Subtract the old data_len before replacing it and add the new data_len
afterwards, keeping skb->len and skb->data_len synchronized.

Additionally, bpf_xdp_pull_data() can advance data_end while leaving
frags present. The skb is then still non-linear, so the old
__skb_put(skb, off) triggers SKB_LINEAR_ASSERT().

Use skb_set_tail_pointer() and update skb->len explicitly instead,
following bpf_prog_run_generic_xdp(). Unlike __skb_put(),
skb_set_tail_pointer() does not require a linear skb.

A 60000-byte UDP datagram on a veth pair with MTU 64000 was shortened by
1024 bytes from its fragment area. Before the fix, all 10 runs produced
corrupted payloads. After the fix, all 10 runs matched the expected
payload exactly. A forced-tailroom reproducer also exercises
bpf_xdp_pull_data() with frags still present; the old code triggers
SKB_LINEAR_ASSERT(), while this fix passes 10/10 runs.

Fixes: 718a18a0c8 ("veth: Rework veth_xdp_rcv_skb in order to accept non-linear skb")
Cc: stable@vger.kernel.org
Reported-by: Mohsin Bashir <mohsin.bashr@gmail.com>
Link: https://lore.kernel.org/bpf/80687d9c-9c27-494c-b3f2-efd0230b1895@gmail.com/
Suggested-by: Lorenzo Bianconi <lorenzo@kernel.org>
Acked-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Sun Jian <sun.jian.kdev@gmail.com>
Link: https://patch.msgid.link/20260804054040.613675-3-sun.jian.kdev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-07 17:01:05 -07:00
Ralf Lici
9e163917a8 ovpn: finish crypto callback cleanup before peer release
Crypto completion callbacks hold both key-slot and peer references. The
peer reference pins the netdev, and dropping the last peer reference can
let netdev unregistration and module removal make progress.

Do not release that peer reference before the callback has finished its
own cleanup. If ovpn_crypto_key_slot_put runs after ovpn_peer_put, it can
schedule an RCU callback backed by module text after ovpn_cleanup
rcu_barrier has already run. The TX error path also freed the remaining
skb after ovpn_peer_put, leaving callback cleanup outside the peer/netdev
lifetime window.

Release the key slot and free any remaining skb first, then drop the peer
reference as the last callback action.

Fixes: 8534731dbf ("ovpn: implement packet processing")
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
2026-08-07 02:12:13 +02:00
Ralf Lici
41d44ac7a6 ovpn: fix NULL dereference when killing missing key
ovpn_crypto_kill_key assumes both crypto slots are populated and
dereferences each slot before checking it. That is not guaranteed: a
peer can have only one installed key, and the kill path may be asked to
remove a key that is not present.

Read each slot once while holding the crypto state lock, check for NULL
before looking at key_id, and only replace the slot that actually
matches.

Fixes: 89d3c0e461 ("ovpn: kill key and notify userspace in case of IV exhaustion")
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
2026-08-07 02:12:13 +02:00
Philipp David
b0ce5fd9fa igc: fix netdev not re-attached after resume if interface is down
__igc_resume() calls netif_device_attach() only inside the
netif_running() branch, so an interface that was down during suspend
is never re-attached on resume. It then stays in the not-present state
that __igc_shutdown() set via netif_device_detach(): ethtool reports
ENODEV and every attempt to bring the interface up fails the
netif_device_present() check in __dev_open() with -ENODEV, silently,
since __igc_resume() returns 0. Only reloading the driver recovers the
device.

This is easy to hit in practice because NetworkManager brings managed
interfaces down before sleep unless Wake-on-LAN is configured, making
the adapter unusable after every suspend/resume cycle with WoL
disabled.

Re-attach the netdev on every successful resume, as igb and e1000e do.

Fixes: 6f31d6b643 ("igc: Refactor runtime power management flow")
Cc: stable@vger.kernel.org
Signed-off-by: Philipp David <pd-lkml@3b.pm>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20260804222205.1580328-11-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-06 10:21:28 -07:00
Bhargava Marreddy
bfec39ff14 bnge: Fix resource leak in bnge_init_nic() error path
If bnge_init_chip() fails, bnge_init_nic() jumps to err_free_ring_grps
and returns immediately, skipping cleanup for RX ring pair buffers.

Remove the early return so execution falls through to
err_free_rx_ring_pair_bufs to properly free resources on error.

Fixes: 23df6aebf8 ("bng_en: Allocate stat contexts")
Signed-off-by: Bhargava Marreddy <bhargava.marreddy@broadcom.com>
Reviewed-by: Dharmender Garg <dharmender.garg@broadcom.com>
Reviewed-by: Rajashekar Hudumula <rajashekar.hudumula@broadcom.com>
Link: https://patch.msgid.link/20260805094022.15487-1-bhargava.marreddy@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-06 09:26:22 -07:00
Doruk Tan Ozturk
fde39b8a52 net: usb: ipheth: fix carrier_work UAF on disconnect
ipheth_sndbulk_callback() re-arms the carrier-check work on any
non-zero URB status:

	else
		schedule_delayed_work(&dev->carrier_work, 0);

Nothing ties that to the interface being up, so the work can be armed
again after ipheth_close() has already drained it, and stay armed
until the netdev whose private area embeds it is freed.

On unplug with a TX URB in flight, ipheth_disconnect() drains the work
through unregister_netdev() -> ipheth_close() ->
cancel_delayed_work_sync() and only then calls ipheth_kill_urbs().
usb_kill_urb() completes the in-flight TX URB with -ENOENT, so
ipheth_sndbulk_callback() runs after the drain and re-arms
carrier_work.

The same completion also re-arms the work if the interface is only
brought down while a TX URB is in flight, and
ipheth_carrier_check_work() then keeps re-queueing itself once a
second. unregister_netdev() does not call ipheth_close() for an
already-down interface, so nothing drains it on the later unplug
either.

In both cases free_netdev() frees the netdev while carrier_work is
still pending, and ipheth_carrier_check_work() dereferences freed
memory.

Tie the work to the interface state instead of chasing the completion:
disable it in ipheth_close() and enable it in ipheth_open(), so a
schedule_delayed_work() from the URB completion is a no-op whenever
the interface is not up. disable_delayed_work_sync() also waits for a
running instance, so it fully replaces the cancel_delayed_work_sync()
it takes the place of. The work starts out disabled in ipheth_probe()
so the enable/disable counts balance from the first open.

Reproduced under KASAN on linux-next (next-20260731) with dummy_hcd and
raw-gadget standing in for the device, driving the second path above (the
interface is already down, so unregister_netdev() does not call
ipheth_close()): 15 of 15 unpatched boots report a slab-use-after-free in
__run_timers(), freed by ipheth_disconnect() and re-armed from
ipheth_sndbulk_callback() via queue_delayed_work_on(). The
same trigger on a kernel differing only by this patch reports 0 of 15,
and the carrier check still functions across open/close cycles.

The reproducer needs an attached USB device that stops draining bulk OUT,
plus a link down and unplug, driven as root. It is not a privilege
boundary crossing and no exploit primitive was developed.

Found by 0sec (https://0sec.ai).

Fixes: bb1b40c7cb ("usbnet: ipheth: prevent TX queue timeouts when device not ready")
Cc: stable@vger.kernel.org
Signed-off-by: Doruk Tan Ozturk <doruk@0sec.ai>
Link: https://patch.msgid.link/20260802120602.42595-1-doruk@0sec.ai
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-06 08:21:57 -07:00
Fan XinRan
68bf02b6b4 net: thunderbolt: Tear down DMA paths before stopping the rings
tbnet_tear_down() stops both rings and frees their frame buffers before
calling tb_xdomain_disable_paths().  tb_ring_stop() zeroes the ring's
descriptor base and tbnet_free_buffers() unmaps and frees the pages the
frames sit in, so by the time __tb_path_deactivate_hop() polls the hop's
'pending' bit, anything still in flight has nowhere to drain to.

The teardown sequence has been in this order since the driver was added.
The setup path has not: commit ff7cd07f30 ("net: thunderbolt: Enable
DMA paths only after rings are enabled") moved the path enable to the end
of tbnet_connected_work() and documented why:

	/* Both logins successful so enable the rings, high-speed DMA
	 * paths and start the network device queue.
	 *
	 * Note we enable the DMA paths last to make sure we have primed
	 * the Rx ring before any incoming packets are allowed to
	 * arrive.
	 */

Teardown was never updated to match, so the rings and the paths now come
down in the same order they go up instead of in reverse.

On an ASMedia ASM4242 host router the 'pending' bit then never clears:
every teardown burns the full 500 ms timeout and
__tb_path_deactivate_hop() returns -ETIMEDOUT.  Raising the timeout to
5 s does not help, so the hop is not slow to drain, it never drains
at all.

The failure is invisible above the thunderbolt core.
__tb_path_deactivate_hops() is void and only calls tb_port_warn();
tb_path_deactivate(), tb_tunnel_deactivate() and
__tb_disconnect_xdomain_paths() are void as well, and
tb_disconnect_xdomain_paths() ends in an unconditional "return 0".  So
tb_xdomain_disable_paths() reports success and the netdev_warn() below
it never fires.  Repeated teardowns eventually take the XDomain control
channel down, after which the peer node is gone and only a power cycle
brings the controller back.

Deactivating the paths first fixes it.  Measured with kretprobes on a
stock v6.17 tree with no other patches applied, on a link that was up
and had just carried traffic:

  before: __tb_path_deactivate_hop() returns 0 for the first hop, then
          -ETIMEDOUT for the second 500335 us later
  after:  0 for both, 525 us apart

Alternating the two orderings ABBA over three load levels, four
teardowns per arm: every teardown failed before the change (21 of 21
that ran), none failed after (0 of 24).  The before arms ran short
because the link died partway through.  The same split shows up when
the interface is enslaved to a bond instead of just brought down, which
is how I ran into this in the first place.  Throughput and latency after
the change are unchanged.

Hosts whose routers drain the hop despite the stale descriptor base see
no functional difference, since the paths end up deactivated either way.

Fixes: e69b6c02b4 ("net: Add support for networking over Thunderbolt cable")
Signed-off-by: Fan XinRan <shinjiangjiang@gmail.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Link: https://patch.msgid.link/20260803-b4-tbnet-teardown-v2-1-27de6a13ca2d@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-08-06 08:19:53 -07:00
Sergey V. Frolov
7e2d693af0 net: octeontx2-pf: Fix UB in shift operation
In function otx2_get_egress_burst_cfg, when the parameter `burst` is
255 and the max mantissa is 255 (0xFFULL), `burst_exp` is set to
`ilog2(255) - 1`, which equals 6.

This results in an unsigned wrap-around when calculating
`(1ULL << (*burst_exp - 7))`, since `*burst_exp - 7` becomes -1,
which makes the shift operand 0xFFFFFFFF. This value is greater than
the width of the left operand.

According to standard 6.5.7 p.3:
"The type of the result is that of the promoted left operand.
If the value of the right operand is negative or is greater than
or equal to the width of the promoted left operand, the behavior
is undefined."

Fix the off-by-one boundary condition.

Add a WARN_ON(*burst_exp < 7) before the else branch as an
explicit safeguard. This ensures that if max_mantissa ever changes
in a way that reintroduces this condition, it will be immediately
caught at runtime rather than silently triggering UB.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: e638a83f16 ("octeontx2-pf: TC_MATCHALL egress ratelimiting offload")
Signed-off-by: Sergey V. Frolov <Sergey.V.Frolov@kaspersky.com>
Cc: stable@vger.kernel.org
Reviewed-by: Ratheesh Kannoth <rkannoth@marvell.com>
Reviewed-by: Sunil Goutham <sgoutham@marvell.com>
Link: https://patch.msgid.link/20260804120446.1955448-1-Sergey.V.Frolov@kaspersky.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-08-06 15:16:50 +02:00