Commit graph

57,421 commits

Author SHA1 Message Date
Dmitriy Okunev
14cb1e7702 net: mvpp2: prevent buffer overflow in page_pool allocation
The per‑processor buffering scheme is supported only if the
number of pools (nrxqs * 2) does not exceed MVPP2_BM_MAX_POOLS (8).
This is already checked in mvpp2_probe() during the initial
activation of percpu_pools.

However, mvpp2_change_mtu() may later call
mvpp2_bm_switch_buffers(priv, true) without this check, which can
lead to an out-of-bounds access in the priv->page_pool array in
mvpp2_bm_init(). The array is sized to hold MVPP2_PORT_MAX_RXQ
entries, and mvpp2_get_nrxqs() may return exactly that value. The
per-CPU scheme then doubles it to nrxqs * 2, exceeding the array
bounds.

Check that the hardware version is MVPP22 or newer and that the
number of pools (nrxqs * 2) does not exceed MVPP2_BM_MAX_POOLS
before switching to per-CPU mode.

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

Fixes: 7d04b0b13b ("mvpp2: percpu buffers")
Signed-off-by: Dmitriy Okunev <dokunevdmitriy@gmail.com>
Link: https://patch.msgid.link/20260914091557.71769-1-dokunevdmitriy@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-17 14:43:29 +02:00
James Clark
9ca4ba2425 net: macb: fix ordering around PTP timestamp read
PTP_SYS_OFFSET_EXTENDED returns system timestamps that do not correctly
bracket the PHC register read on MACB/GEM. On a Raspberry Pi 5, the
returned interval can be as short as 37 ns, while an ordered register
read takes approximately 1 us. This biases the midpoint used by phc2sys,
causing CLOCK_REALTIME to run approximately 0.5 us ahead when synchronized
to the PHC.

gem_tsu_get_time() reads the nanoseconds register using the driver's
relaxed MMIO accessor. On weakly ordered systems, the subsequent system
timestamp can be taken before the register read completes. The internal
smp_rmb() in the pre-timestamp path also does not guarantee ordering
against the subsequent MMIO read.

Add rmb() before and after the bracketed nanoseconds read in both the
normal and seconds rollover paths so the system timestamps bracket the
PHC read. Adding the post-read barrier increases the minimum interval on
the same Raspberry Pi 5 to approximately 1 us.

Fixes: e51bb5c278 ("net: macb: ptp: Switch to gettimex64() interface")
Tested-by: Nicolai Buchwitz <nb@tipi-net.de> # Raspberry Pi CM5, min bracket 37 ns -> 981 ns
Reviewed-by: Nicolai Buchwitz <nb@tipi-net.de>
Reviewed-by: Théo Lebrun <theo.lebrun@bootlin.com>
Assisted-by: LLM
Signed-off-by: James Clark <jjc@jclark.com>
Link: https://patch.msgid.link/20260915045823.76100-1-jjc@jclark.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-17 14:25:52 +02:00
Lorenzo Bianconi
02fffd1939 net: stmmac: preserve real_num_tx_queues on mqprio setup failure
With the FPE preemption-class mapping error now propagated from
stmmac_fpe_map_preemption_class(), tc_setup_dwmac510_mqprio() can fail
on the mapping step. The error path used to call stmmac_reset_tc_mqprio(),
which resets the number of real TX queues to priv->plat->tx_queues_to_use
(the platform maximum), overwriting the value that was active before the
offload was attempted (for example a lower count left over from a previous
mqprio configuration).

The issue can be triggered using the following configuration:

  # First mqprio config lowers the hw queue count below the platform
  # default (e.g. 8 TX queues).
  $tc qdisc add dev eth0 root handle 1: mqprio queues 2@0 2@2

  # Replace mqprio configuration with a second one that fails FPE
  # preemption-class mapping. stmmac driver resets the real_num_tx_queues
  # to the platform maximum, losing the previous configuration.
  $tc qdisc replace dev eth0 root handle 2: mqprio queues 2@0 2@2 fp E P

Save ndev->real_num_tx_queues before lowering it and restore it,
together with the TC-to-queue and priority-to-TC mappings, when the FPE
preemption-class mapping fails, instead of resetting the queue count to
the platform maximum.

Note that a failed setup makes the qdisc layer run mqprio_destroy() on
the new qdisc. Because priv->hw_offload is only assigned after
ndo_setup_tc() succeeds, mqprio_destroy() calls netdev_set_num_tc(dev, 0),
so dev->num_tc ends up 0 regardless of the driver-side restore and the
previous qdisc is not reactivated. The restore is still needed to keep
real_num_tx_queues and to avoid leaving the failed configuration's
TC-to-queue and priority-to-TC mappings in place.

Fixes: 195e4f409a ("net: stmmac: support fp parameter of tc-mqprio")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Link: https://patch.msgid.link/20260911-stmmac-tc_setup_dwmac510_mqprio-error-path-v3-2-a76b1e2547c1@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 19:07:13 -07:00
Lorenzo Bianconi
90e4b849df net: stmmac: propagate FPE preemption-class mapping errors
stmmac_fpe_map_preemption_class() dispatches through the
stmmac_do_void_callback() helper, which forces the callback's return
value to 0 whenever the op pointer is populated. As a result the
-EINVAL returned by dwmac5_fpe_map_preemption_class() (e.g. when a
preemptible TC owns more than one TXQ under SP scheduling) is silently
swallowed by every caller.

Switch the dispatch macro to stmmac_do_callback() so the callback's real
result is propagated, and honour it in the taprio and mqprio qdisc
offload.

Note that the taprio "if (ret)" check in tc_taprio_configure() used to
be dead code and now becomes live: a preemptible TC spanning more than
one TXQ under SP scheduling cannot be programmed in hardware, so a
taprio or mqprio configuration that previously returned success while
leaving the preemption-class register unprogrammed now fails with
-EINVAL. For taprio, the failure also runs the disable path, tearing
down the schedule that was just installed; this is the intended
behaviour.

Fixes: 195e4f409a ("net: stmmac: support fp parameter of tc-mqprio")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Link: https://patch.msgid.link/20260911-stmmac-tc_setup_dwmac510_mqprio-error-path-v3-1-a76b1e2547c1@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 19:07:13 -07:00
Linus Walleij
1dd85662fe net: ethernet: cortina: Ack RX overrun interrupt correctly
The RX overrun interrupt is reported in interrupt status register 4, but
gmac_irq() acknowledges it using the RX descriptor error bit from status
register 0. For GMAC0 this writes the GMAC1 overrun bit, while for GMAC1
the shift leaves no bit in the 32-bit register.

Acknowledge the same per-port RX overrun bit that was detected.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260914-b4-gemini-ethernet-fixes-2-v2-1-5ab39a047b90@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 17:44:40 -07:00
Jakub Kicinski
490599ab23 eth: fbnic: ring the doorbell if a burst ends in a drop
fbnic_tx_map() skips the doorbell write, and the completion request,
for every packet handed to it with xmit_more set, counting on the
packet which ends the burst to publish them all. When that packet is
dropped instead - skb_put_padto(), skb_cow_head() or a DMA mapping
failure - nothing rings. The descriptors of the preceding packets stay
invisible to the HW until the next transmit on that queue, which for a
burst-then-idle workload may never come.

Remember the meta descriptor of the last packet left without a doorbell
and flush it from the error paths. The completion request has to be set
on that descriptor rather than simply writing the tail, otherwise the HW
would transmit the packets but never report a head, and the ring would
fill up and stall for good.

This is very similar to Joe's recent series of fixes for bnxt.
Not seen in real life, reproduced under QEMU with failure injection.

Fixes: 9a57bacd57 ("eth: fbnic: Add basic Tx handling")
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260915022327.913218-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 17:33:06 -07:00
Yige Jiang
5ae916fabc net: netsec: fix device_node reference leak on phy_np
netsec_of_probe() takes a reference on the PHY device_node with
of_parse_phandle() and stores it in priv->phy_np, but the driver never
drops it.  One device_node reference is leaked per probe, on the success
path as well as on every error path reached after netsec_of_probe().

Neither consumer takes ownership.  of_mdio_parse_addr() is a static
inline taking a const struct device_node * that only reads the "reg"
property.  of_phy_connect() borrows as well: of_phy_get_and_connect() in
drivers/net/mdio/of_mdio.c brackets its own call with of_node_get() at
:364 and of_node_put() at :373, which would be a double put if
of_phy_connect() consumed the reference.

The node is still in use at netsec_netdev_open() time, where it is
passed to of_phy_connect(), so it has device lifetime.  Release it at
the probe error label, which every failure path after the acquire
funnels through, and in netsec_remove().  Both releases precede
free_netdev(), since priv is netdev_priv(ndev).  The ACPI probe path
leaves priv->phy_np NULL and of_node_put(NULL) is a no-op.

There is no end-user visible symptom on currently supported platforms:
a device_node is only freed once OF_DYNAMIC is enabled and the node has
been detached, so on a static device tree the imbalance is inert.  It is
observable as a refcount that grows across bind/unbind cycles, and would
matter under device tree overlays.

Found by static analysis of reference acquire/release pairing rather
than from a runtime report.  No reproducer was produced and the change
has not been runtime tested; it is compile-tested only (arm64,
CONFIG_SNI_NETSEC=m via COMPILE_TEST).

Fixes: 533dd11a12 ("net: socionext: Add Synquacer NetSec driver")
Signed-off-by: Yige Jiang <yigejiang86@gmail.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260913064102.37452-1-yigejiang86@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 17:32:15 -07:00
Mark Amirkan
a9ce4053dc net: lan743x: fix RX checksum use-after-free
lan743x_rx_process_buffer() adds each non-first receive buffer to the
head skb's frag_list.  On the last descriptor, lan743x_rx_trim_skb()
linearizes the head and frees the fragment skb metadata.

The checksum-success path then writes ip_summed through the local skb
pointer, which still points to the final fragment.  This causes a
use-after-free write when a packet spans more than one receive buffer.

Set ip_summed on the surviving head skb instead.  Multi-buffer receive
can occur after a live MTU increase because existing ring entries keep
their old buffer size until they are replenished.

A KUnit test invoking lan743x_rx_process_buffer() with a two-buffer
packet produced a one-byte KASAN use-after-free write before this change.
The same test passed after the change.  The driver object also builds
with W=1.  This was not tested on physical LAN743x hardware.

Fixes: cd6910501c ("net: lan743x: Add support for Rx IP & TCP checksum offload")
Cc: stable@vger.kernel.org
Signed-off-by: Mark Amirkan <markdamirkan@gmail.com>
Reviewed-by: Chenguang Zhao <zhaochenguang@kylinos.cn>
Link: https://patch.msgid.link/20260913-b4-send-lan743x-uaf-v1-1-73d563d08ba9@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-16 17:26:40 -07:00
Lorenzo Bianconi
f0ef4b1eae net: stmmac: do not overwrite phc_index when no PTP clock is registered
stmmac_get_ts_info() reports phc_index as 0 when hardware timestamping
is supported but no PTP clock has been registered yet (e.g. while the
interface is down). Zero is a valid PHC index and would make userspace
resolve the wrong clock; the absence of a clock should be reported as
-1.

The ethtool core already initializes phc_index to -1 before invoking
the get_ts_info callback (ethtool_init_tsinfo()), so just drop the
erroneous assignment.

Fixes: 9364fa7fcf ("net: stmmac: Remove setting of RX software timestamp")
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20260914-stmmac-fix-phc_index-v2-1-bf3d90373fe4@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-15 18:25:06 -07:00
Lorenzo Bianconi
15989abd74 net: stmmac: fix TSO header length truncation
stmmac_tso_xmit() stores the protocol header length returned by
stmmac_tso_header_size() in a u8. stmmac_tso_valid_packet() admits
headers up to 1023 bytes, so a header longer than 255 bytes wraps modulo
256 (486 becomes 230, 256 becomes 0).

A TCP over IPv6 socket carrying a few hundred bytes of sticky
destination/hop-by-hop options makes skb_tcp_all_headers() exceed 255
while staying below the 1023-byte limit, so such an skb reaches
stmmac_tso_xmit().

Widen proto_hdr_len to unsigned int, which is sufficient since the value
is bounded by the hardware limit, and adjust the debug print specifier
accordingly.

Fixes: 9edfa7dab8 ("net: stmmac: enable TSO for IPv6")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20260911-stmmac-fix-header-length-v1-1-8fc103334327@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-15 16:55:52 -07:00
Nicolai Buchwitz
23ca4ddc4f net: bcmgenet: restore the hardware filters on open
bcmgenet_hfb_init() runs INIT_LIST_HEAD() on priv->rxnfc_list, which drops
every rule off the list, and bcmgenet_open() calls it on each ifup. Every
rule the user configured is silently lost:

  # ethtool -N eth0 flow-type ether dst $MAC action 0
  Added rule with ID 0
  # ethtool -n eth0 | grep -c Filter:
  1
  # ip link set eth0 down && ip link set eth0 up
  # ethtool -n eth0 | grep -c Filter:
  0

Initialise the lists once at probe and restore the rules on open, as
bcmgenet_resume() already does.

Fixes: 3e37095228 ("net: bcmgenet: add support for ethtool rxnfc flows")
Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>
Reviewed-by: Justin Chen <justin.chen@broadcom.com>
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Link: https://patch.msgid.link/20260913190052.939955-1-nb@tipi-net.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-14 19:21:58 -07:00
Linus Torvalds
7844502343 Nothing too exciting, usual stream of fixes.
Including fixes from Netfilter, Bluetooth and WPAN.
 
 Current release - new code bugs:
 
  - Bluetooth: hci_sync: fix not setting CE length properly
 
  - eth: enic: match mailbox replies to request numbers
 
 Previous releases - regressions:
 
  - tunnels: drop stale dst when building an ICMP error for PMTUD
 
  - ipv6: null-check fib6_node before accessing in __ip6_del_rt_siblings()
    (bug in the rtnl_lock -> RCU conversion)
 
  - eth: bnxt_en: fix crashes on Thor2 due to OOB coalescing buffer accesses
 
  - eth: bnxt_en: prevent queue stop with deferred completions
 
 Previous releases - always broken:
 
  - eth: ice: don't dereference pointers from TP_printk()
 
  - eth: fix OOB writes on ethtool flow rule dump in 3 drivers
 
  - eth: mlx5: fix FEC configuration with RS_544_514_INTERLEAVED_QUAD
 
  - dsa: tag_brcm: legacy FCS: request needed tailroom
 
 Misc:
 
  - net: cap tx_queue_len at S16_MAX to prevent oversized ring alloc
 
  - ipv6: flowlabel: cap duplicate leases per socket
 
 Signed-off-by: Jakub Kicinski <kuba@kernel.org>
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEE6jPA+I1ugmIBA4hXMUZtbf5SIrsFAmqi3pQACgkQMUZtbf5S
 IrvUIg//X9nIxY2F5PzJ5jD9p5ccXrQMLe7kT5AW2tP5PDC8d5PIv4Q5XzFKQPU7
 XElAUKvBxmofwU2lqILYGi8AeUpqHZtKPY7XzKeqd6i72KOD6mGzzYNijqttBXcM
 vFVtIeKExXjAwvNc2as1SeXVEAAAkBtrCFuMNHMq0C56yK4md/XVkCDHaJkomNit
 geke1U8gut3rZddWKxp4WDbL8Wmx9yM0uDMBznO/+cwITObA0Hme3IgRndglzz7n
 n4Ih+EG4tRrD3kUf6oePzKQ47cd+qnSVlVTCZUwB5E/HKqWJFXxSN4Sv+mez0sAS
 rrI5hl+luNKUYrZ8/jiNlvajgAL4+AYpCKPDJbXrOW+z+x4BC2VYZBAHLoUr5ZAq
 Z5OYU9SgD1oGntqkI8mAEiRTEu+4gjhIEhjENHEzqdjUogaBIp7MWwCrNBAnFWvs
 2McmNfZZMVhxKpyYnndUStsVQySVPASb0CXeqTIO6PJsAp/HBjoMYKYKPhhwk0Gp
 lE8zHjEnPVofRfXfT+oZnbS8is2nC9FjBy9ksIGcC7vyTOdPsoIBoB8JY0x/INRM
 SOJvyxrdnVkMjiBejkdOa5X9HbD1cA/NVyzT2WEaZBGPmIqfNBgUIOxnXD4CShuy
 9zX8qtHsUmmYxPteF30Uhfe0kyLQ9OnjUz2Bl++EruxpIE3+i48=
 =f9J9
 -----END PGP SIGNATURE-----

Merge tag 'net-7.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Nothing too exciting, usual stream of fixes. Including fixes from
  Netfilter, Bluetooth and WPAN.

  Current release - new code bugs:

   - Bluetooth: hci_sync: fix not setting CE length properly

   - eth: enic: match mailbox replies to request numbers

  Previous releases - regressions:

   - tunnels: drop stale dst when building an ICMP error for PMTUD

   - ipv6: null-check fib6_node before accessing in __ip6_del_rt_siblings()
     (bug in the rtnl_lock -> RCU conversion)

   - eth: bnxt_en:
       - fix crashes on Thor2 due to OOB coalescing buffer accesses
       - prevent queue stop with deferred completions

  Previous releases - always broken:

   - eth:
       - ice: don't dereference pointers from TP_printk()
       - fix OOB writes on ethtool flow rule dump in 3 drivers
       - mlx5: fix FEC configuration with RS_544_514_INTERLEAVED_QUAD

   - dsa: tag_brcm: legacy FCS: request needed tailroom

  Misc:

   - net: cap tx_queue_len at S16_MAX to prevent oversized ring alloc

   - ipv6: flowlabel: cap duplicate leases per socket"

* tag 'net-7.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (164 commits)
  selftests: tc-testing: test action batch failure cleanup
  net/sched: act_api: release all action references on NEWACTION failure
  openvswitch: fix wrong flag value in get_ipv6_ext_hdrs()
  ipmr: account multicast table and route memory
  net: phy: dp83td510: handle the active-high LED polarity mode
  net: macb: initialize PTP state before registering clock
  net: hsr: enable promiscuous mode on interlink port with fwd offload
  ipv6: fix fib6 walker UAF on seq stop
  net: stmmac: fix TX descriptor availability check for TSO traffic
  net/rds: fix tcp stream corruption with large pages
  net: mana: restore the XDP program pointer when pre-allocation fails
  net: phy: dp83867: handle the active-high LED polarity mode
  octeontx2-af: fix PF/CGX debugfs PCI bus lookup
  net: net_failover: Fix the deadlock in net_failover_slave_name_change()
  net: phy: mediatek-ge: disable EEE on the MT7530 PHY
  tcp: reject non zerocopy devmem tx
  net: ethernet: mtk_eth_soc: populate lpi_interfaces to fix EEE support
  net: dsa: mt7530: populate lpi_interfaces to fix EEE support
  net: hinic: fix mailbox segment buffer overflow
  net: sun4i-emac: fix missing of_node_put() for phy_node
  ...
2026-09-10 14:07:48 -07:00
Jakub Kicinski
fc266a5dd9 Merge branch '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue
Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2026-09-08 (idpf, ice)

For idpf:
Myeonghun Pak adds calls to disable DIM work and PTM to allow for proper
cleanup.

Josh adds check, and adjustment, for VLAN headers when processing RSC
packets.

For ice:
Jake adds call to xa_destroy for xarray sched_node_ids; also moving it
from port_info struct to ice_hw to simplify its lifecycle management.

Jakub Kicinski stores trace event data as scalars instead of
dereferencing pointers in TP_printk(), preventing use-after-free issues
during event printing and eliminating double-dereference warnings.

* '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  eth: ice: don't dereference pointers from TP_printk()
  ice: add missing xa_destroy for sched_node_ids
  idpf: account for VLAN header when parsing RSC packet header
  idpf: disable PTM on probe failure and on remove
  idpf: disable DIM work before freeing q_vectors
====================

Link: https://patch.msgid.link/20260908214502.528440-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-10 08:55:13 -07:00
Runyu Xiao
e1406330d7 net: macb: initialize PTP state before registering clock
gem_ptp_init() registers the PTP clock before initializing
bp->tsu_clk_lock and the TSU hardware. Since ptp_clock_register()
publishes the PTP character device, userspace may invoke PTP callbacks
before the lock and hardware are ready.

In addition, gem_ptp_init() is called from both the interface open and
resume paths. Reinitializing tsu_clk_lock there can reset the lock while
timestamp processing is using it.

This race is theoretical and has not been observed in practice.

Initialize tsu_clk_lock once during probe and initialize the TSU before
registering the PTP clock.

Fixes: ab91f0a9b5 ("net: macb: Add hardware PTP support")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/netdev/20260904030439.3994047-1-runyu.xiao@seu.edu.cn/
Reviewed-by: Théo Lebrun <theo.lebrun@bootlin.com>
Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Signed-off-by: Runyu Xiao <runyu.xiao@seu.edu.cn>
Link: https://patch.msgid.link/20260908103924.607033-1-runyu.xiao@seu.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-10 08:46:19 -07:00
Lorenzo Bianconi
5e38d732ec net: stmmac: fix TX descriptor availability check for TSO traffic
stmmac_tso_xmit() estimates the number of free TX descriptors required by
a TSO skb as:

	(skb->len - proto_hdr_len) / TSO_MAX_BUFF_SIZE + 1

which assumes the payload is split into TSO_MAX_BUFF_SIZE chunks. This
underestimates the descriptors actually consumed by stmmac_tso_allocator(),
since each fragment is mapped individually and so it needs at least one
descriptor regardless of its size. Moreover, one descriptor is used for
the L2/L3/L4 headers and, when the MSS changes, one more is consumed for
the MSS context descriptor.
For a highly fragmented TSO skb the check can therefore pass even when the
ring has too few free slots. stmmac_tso_allocator() then writes past the
available descriptors, overwriting descriptors still owned by the DMA
engine, corrupting the TX ring.
Add stmmac_tso_get_num_desc() to compute the exact number of descriptors
needed for the header, the linear payload and each fragment, plus the MSS
context descriptor when required, and use it in the availability check.

Fixes: f748be531d ("stmmac: support new GMAC4")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Link: https://patch.msgid.link/20260907-stmmac-fix-tso-nfrags-check-v1-1-328459906cdb@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-10 08:34:07 -07:00
Long Li
4c46beb807 net: mana: restore the XDP program pointer when pre-allocation fails
mana_xdp_set() publishes the new program into apc->bpf_prog before it
allocates anything, because mana_pre_alloc_rxbufs() sizes the buffers
from it via mana_get_rxbuf_cfg(). When that allocation fails the
function returns the error directly, skipping the err_dealloc_rxbuffs
label which is the only place that restores the previous pointer.

The attach is reported as failed, so the BPF core drops the reference it
held for the caller and the program can be freed, while apc->bpf_prog
still points at it. The next consumer of mana_xdp_get() - typically
mana_chn_setxdp() from mana_alloc_queues() on the following ifup, or
after a TX timeout reset - then calls bpf_prog_add() on freed memory.

This is reachable from an ordinary "ip link set dev ethX xdp obj ..."
whenever the per-queue RX buffer pre-allocation cannot be satisfied.

Restore the previous program on that error path.

Fixes: 730ff06d3f ("net: mana: Use page pool fragments for RX buffers instead of full pages to improve memory efficiency.")
Signed-off-by: Long Li <longli@microsoft.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260904202640.3900685-1-longli@microsoft.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-10 08:06:07 -07:00
Ratheesh Kannoth
4f4b743c2d octeontx2-af: fix PF/CGX debugfs PCI bus lookup
rvu_dbg_rvu_pf_cgx_map_display() locates each RVU PF PCI device via
pci_get_domain_bus_and_slot() when printing the PF-to-CGX map. It
assumed PF0 always sits on PCI bus 1 and derived other PF bus numbers
as pf + 1, but the AF device can be enumerated on a different bus.

Use rvu->pdev->bus->number as the base bus instead, so each PF lookup
uses pf + start on systems where RVU functions are on contiguous buses
but do not start at bus 1.

Fixes: e2fb373038 ("octeontx2-af: Display CGX, NIX and PF map in debugfs.")
Signed-off-by: Subbaraya Sundeep <sbhatta@marvell.com>
Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260904085114.3385530-1-rkannoth@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-10 08:02:06 -07:00
Aleksei Sviridkin
d876c9cb2d net: ethernet: mtk_eth_soc: populate lpi_interfaces to fix EEE support
phylink_create() decides once and for all that a MAC supports managed
EEE, and it requires the tx_lpi ops plus non-empty lpi_capabilities and
lpi_interfaces. mtk_add_mac() leaves lpi_interfaces empty.

So ever since EEE support was added, ethtool has answered "Not
supported" on every MAC that uses mtk_phylink_ops, and
phy_disable_eee() has locked userspace out of turning EEE on. MT7628
is unaffected, as rt5350_phylink_ops has no tx_lpi methods.

Leave 2.5 Gbps out of both bitmaps, and the xGMII modes that
mtk_mac_enable_tx_lpi() already refuses. MAC_MCR folds SPEED_2500 onto
MAC_MCR_SPEED_1000, so MAC_MCR_EEE1G would govern LPI on such a link,
and that is unvalidated rather than known unsupported: MediaTek's SDK
driver sets the EEE force bits for 100 Mbps and 1 Gbps only, and the
unit of the wakeup timers is undocumented with the port clock at
2.5 times the rate.

mtk_mac_enable_tx_lpi() programs wake-up times taken from MT7531's
reset values, and the SoC's own field has no reset value to fall
back on. Only MT7981 has been seen to exit LPI cleanly with them, so
the LPI interfaces sit behind a new MTK_GMAC_EEE capability that only
MT7981 sets; every other SoC keeps the current behaviour until it has
been confirmed.

LPI stays off until userspace enables it, but the EEE advertisement of
a PHY that advertises it out of reset comes back, since phylink stops
force-clearing it.

Fixes: 952d732536 ("net: ethernet: mediatek: add EEE support")
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Link: https://patch.msgid.link/20260903123644.23800-3-f@lex.la
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-10 13:31:28 +02:00
Aamir Ahmed
5d4d985957 net: hinic: fix mailbox segment buffer overflow
check_mbox_seq_id_and_seg_len() validates that seq_id does not
exceed SEQ_ID_MAX_VAL (42) and seg_len does not exceed
MBOX_SEG_LEN (48).  However, this allows the last segment
(seq_id=42) to carry a full 48-byte payload, writing to offset
42*48=2016 for 48 bytes (ending at byte 2064).  The receive
buffer is only MBOX_MAX_BUF_SZ (2048) bytes, resulting in a
16-byte heap buffer overflow.

The hinic3 driver already handles this correctly by defining
MBOX_LAST_SEG_MAX_LEN and rejecting the last segment when it
exceeds the remaining buffer space.  Apply the same fix to the
hinic driver.

Fixes: a425b6e1c6 ("hinic: add mailbox function support")
Signed-off-by: Aamir Ahmed <elb12345@hotmail.co.uk>
Link: https://patch.msgid.link/AS8P251MB0001AE870B09020B46B5D7DBC8B22@AS8P251MB0001.EURP251.PROD.OUTLOOK.COM
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-10 12:37:09 +02:00
Li Youhong
af406abfec net: sun4i-emac: fix missing of_node_put() for phy_node
of_parse_phandle() returns a node pointer with an elevated refcount.
Add the missing of_node_put() on the probe error path after
register_netdev() fails and in emac_remove().

Fixes: 492205050d ("net: Add EMAC ethernet driver found on Allwinner A10 SoC's")
Signed-off-by: Li Youhong <liyouhong@kylinos.cn>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260904080758.2432748-1-dayou5941@163.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-10 12:22:01 +02:00
Carolina Jubran
7f26a5e804 net/mlx5e: Move representor vnic reporter to eswitch devlink port
The representor vnic devlink health reporter is created and destroyed
along the representor netdev (un)load path, which is not serialized by
the devlink instance lock. Destroying the reporter from there triggers
a devl_assert_locked() splat on driver unbind:
  WARNING: net/devlink/core.c:259 at devl_assert_locked+0x54/0x70, CPU#2: bash/3758
  Modules linked in: mlx5_vdpa vringh vdpa mlx5_ib mlx5_fwctl mlx5_core ...
  CPU: 2 UID: 0 PID: 3758 Comm: bash Tainted: G        W           6.19.0+ #1 PREEMPT
  Tainted: [W]=WARN
  Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), ...
  RIP: 0010:devl_assert_locked+0x54/0x70
  Call Trace:
   <TASK>
   devl_health_reporter_destroy+0x3a/0x1b0
   mlx5e_vport_rep_unload+0x12d/0x2b0 [mlx5_core]
   mlx5_eswitch_unregister_vport_reps+0x1b8/0x220 [mlx5_core]
   ? __esw_offloads_unload_rep+0x190/0x190 [mlx5_core]
   ? kernfs_remove_by_name_ns+0xc3/0xf0
   device_release_driver_internal+0x3b2/0x560
   unbind_store+0xce/0xf0

Move the reporter's lifecycle to the eswitch devlink port (un)register
paths, which are already serialized by the devlink instance lock, and
store the handle on mlx5_devlink_port. Use the port's mlx5_vport as the
reporter priv since the diagnose callback only needs a device handle and
a vport number, and mlx5_vport carries both and is initialized before
any representor driver probes.

Fixes: cf14af140a ("net/mlx5e: Add vnic devlink health reporter to representors")
Signed-off-by: Carolina Jubran <cjubran@nvidia.com>
Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260906090700.3761260-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-09 19:24:15 -07:00
Lorenzo Bianconi
0338c68e22 net: stmmac: initialize ptp_lock at probe time
priv->ptp_lock is only initialized in stmmac_ptp_register(), which runs
during __stmmac_open(). However, the lock is also used while the
interface is down and has never been opened: tc_taprio_configure()
invokes the PTP gettime64() callback to compute the EST base time when
offloading a TAPRIO schedule, and stmmac_get_time() takes
priv->ptp_lock. Using an uninitialized rwlock is undefined behaviour.
Move the rwlock_init() to __stmmac_dvr_probe(), together with the other
private locks, so that ptp_lock is always valid regardless of the
interface state.

Fixes: b60189e039 ("net: stmmac: Integrate EST with TAPRIO scheduler API")
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20260904-stmmac-fix-ptp-clock-init-v1-1-df70eb1eb04d@oss.qualcomm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-09 19:07:36 -07:00
Ratheesh Kannoth
ef39fca850 octeontx2-pf: reset HTB scheduler topology before freeing queues
HTB offload programs NIX_AF_TLxX_TOPOLOGY on QoS-allocated scheduler
queues via otx2_qos_txschq_set_parent_topology(), but teardown freed
those queues without clearing TOPOLOGY.  The AF only restores PARENT and
SCHEDULE on free, so PRIO_ANCHOR/RR_PRIO settings can survive in the
shared scheduler pool and affect later allocations.

Add otx2_qos_reset_schq_topology() and otx2_qos_free_hw_schq() to zero
TL4 through TL2 TOPOLOGY before each schq is returned to the AF during
hierarchy teardown and cfg rollback.  Skip the aggregation level (TL1):
it is a per-tx-link queue shared by the PF, default Tx hierarchy and VFs,
and is not freed back to the AF by nix_txschq_free_one().

Fixes: 5e6808b4c6 ("octeontx2-pf: Add support for HTB offload")
Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260903020533.3068041-1-rkannoth@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-09 18:17:19 -07:00
Marek Vasut
66ef5adb75 net: ks8851: Fix receiver error in 100BASE-TX mode following software power-down
KSZ8851 errata sheet DS80000716D-page 4 Module 3 [1] states that,
when issuing a software power-down (PMECR[1:0] = 10) followed by a
power-on (PMECR[1:0] = 00), the receiver circuit can fail to start
properly preventing communication. The Transmitter will still send
data, but no data will be received.

The errata sheet also includes a workaround, which states that,
it is recommended that the software power-down feature not be used.

Implement that workaround and drop the entry into software power-down
mode. The ks8851_write_mac_addr() calls entry into normal power-on
mode at the very beginning of the function, therefore dropping the
second call to enter software power-down mode is sufficient here.
The ks8851_net_stop() can only be called after ks8851_net_start()
was already called, and ks8851_net_start() also makes the MAC enter
normal power-on mode, therefore it is also fine to drop the call to
enter software power-down mode from ks8851_net_stop().

This will lead to a slight increase in power consumption, but it also
fixes a sporadic reliability problem on at least KSZ8851-16MLL, which
is where the problem was reported and this fix was tested.

[1] https://ww1.microchip.com/downloads/en/DeviceDoc/80000716D.pdf

Fixes: 3ba81f3ece ("net: Micrel KS8851 SPI network driver")
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Marek Vasut <marex@nabladev.com>
Link: https://patch.msgid.link/20260905130327.203851-1-marex@nabladev.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-09 18:04:05 -07:00
Nicolai Buchwitz
382a373d9e net: macb: put the "mdio" child node reference on success
macb_mii_init() holds the reference returned by of_get_child_by_name()
for macb_mdiobus_register() and drops it only on the error paths, so
every successful probe leaks a node reference. On a CM5, overlay
removal after four bind cycles reports

  OF: ERROR: memory leak, expected refcount 1 instead of 5

Drop the reference after registration, where __mdiobus_register() has
already taken its own for the lifetime of the bus.

Fixes: 8a6631f1ce ("net: macb: avoid redundant lookup for "mdio" child node in MDIO setup")
Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>
Link: https://patch.msgid.link/20260907210856.1673589-3-nb@tipi-net.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-08 17:52:05 -07:00
Nicolai Buchwitz
7d059f3907 net: macb: destroy the phylink instance on the probe error path
macb_mii_init() creates a phylink instance on both of its success paths,
but the probe unwind frees the netdev without destroying it, so a failing
macb_alloc_tieoff() or register_netdev() leaks the instance.

Destroy it at err_out_unregister_mdio, which is only reachable once
macb_mii_init() has succeeded, so bp->phylink is valid there.

Fixes: 7897b071ac ("net: macb: convert to phylink")
Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>
Link: https://patch.msgid.link/20260907210856.1673589-2-nb@tipi-net.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-08 17:52:05 -07:00
Long Li
f6d61fe4c1 net: mana: Clear RDMA teardown and suspend state in mana_rdma_probe()
mana_rdma_remove() sets gd->rdma_teardown to stop
mana_rdma_service_handle() from acting on servicing events, but nothing
ever clears it. A hardware service reset (GDMA_EQE_HWC_RESET_REQUEST)
goes through mana_gd_suspend() -> mana_rdma_remove() and mana_gd_resume()
-> mana_rdma_probe(), so from the first reset onwards every
GDMA_EQE_HWC_SOC_SERVICE event returns early and RDMA suspend/resume
servicing is silently dropped for the life of the device.

gd->is_suspended has the same problem: it is set when servicing removes
the adev and is cleared only by a matching resume. A reset while RDMA is
suspended re-adds the adev but leaves is_suspended set, so a later resume
event calls add_adev() on top of a live gd->adev and leaks it. This is
currently masked by the rdma_teardown bug.

Clear both in mana_rdma_probe(). On the reset path mana_rdma_remove()
has closed the gate and drained the service workqueue, so clear
is_suspended first and re-open the gate with smp_store_release(), paired
with smp_load_acquire() in the handler, so the handler cannot observe an
open gate with a stale is_suspended. On the initial probe path the gate
was never closed and both flags are already clear.

This does not order gd->adev, which add_adev() publishes afterwards. A
servicing event arriving in that window is still dropped, as it is in
mainline today on the initial probe path; closing it needs probe and the
handler to be serialized and is left to a separate change.

Fixes: 505cc26bca ("net: mana: Add support for auxiliary device servicing events")
Signed-off-by: Long Li <longli@microsoft.com>
Link: https://patch.msgid.link/20260902175153.3410560-1-longli@microsoft.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-08 15:53:29 -07:00
Jakub Kicinski
b8bf9bfda5 eth: ice: don't dereference pointers from TP_printk()
After forwarding net-next during the v7.3 merge window we started
seeing:

  TRACE EVENT ERROR: Event ice_tx_dim_work has double dereference in TP_printk: REC->q_vector->tx.tx_ring->q_index
  WARNING: kernel/trace/trace_events.c:420 at test_double_dereference.cold+0x39/0x4b

this is due to extra checks added in tracing subsystem in
commit b5cc230af5 ("tracing: Warn when an event dereferences a pointer in TP_printk()").

Printing happens long after the event was recorded, by which point
the pointers may be invalid (the ring or the dim instance).
Copy the eight scalars into the event instead.

Fixes: 3089cf6d3c ("ice: add tracepoints")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tested-by: Alexander Nowlin <alexander.nowlin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-09-08 13:10:35 -07:00
Jacob Keller
53432c4c3e ice: add missing xa_destroy for sched_node_ids
Commit 16dfa49406 ("ice: Introduce new parameters in ice_sched_node")
added a sched_node_ids xarray to the port info structure, but never called
xa_destroy on it.

Since xarrays can allocate internal memory, this can result in a memory
leak even if every element in the xarray has been removed.

The xarray is currently embedded in the port_info structure. This appears
to have been done because its use is within functions that take the
port_info as a primary argument.

However, this complicates managing the lifecycle of the field. The
port_info structure is allocated in ice_init_hw() using devm, and it is
not released until the devm cleanup when the driver is unloaded.

The ice_init_hw() function is called in many places, including devlink
reload, and possibly during DDP load after updating the Tx scheduler
layout.

Adding a call of xa_destroy to the ice_deinit_hw() causes Sashiko to raise
multiple concerns due to potential ordering issues and possible ways that
port_info could be a dangling reference.

To handle this, move the sched_node_ids out of port_info and into the hw
structure. All users of the array already have a pointer to hw anyways, and
there is only one sched_node_ids per adapter. While here, remove the overly
verbose comment explaining the nature of the sched_node_ids xarray.

Add the missing xa_destroy to the cleanup path and to ice_deinit_hw(),
ensuring that we properly release the xarray memory.

This was caught by Sashiko during development of unrelated code.

Fixes: 16dfa49406 ("ice: Introduce new parameters in ice_sched_node")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@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>
2026-09-08 13:10:35 -07:00
Joshua Hay
cc6d60ef92 idpf: account for VLAN header when parsing RSC packet header
While parsing the header of a Receive Side Coalesced (RSC) packet, check
if a VLAN tag is present and adjust the header parsing accordingly.
Otherwise, Rx TCP traffic is completely broken for any VLAN interface
whose underlying interface has RSC (rx-gro-hw) enabled.

We only need to worry about one VLAN header since Rx packets with
multiple VLAN headers are not candidates for RSC.

Fixes: 3a8845af66 ("idpf: add RX splitq napi poll support")
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Emil Tantilov <emil.s.tantilov@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-09-08 13:10:35 -07:00
Myeonghun Pak
650f197d8e idpf: disable PTM on probe failure and on remove
idpf_probe() enables PCIe Precision Time Measurement with
pci_enable_ptm(), which takes a reference on the device and on every
PTM-capable device up the path to the PTM Root.

Neither the probe error path nor idpf_remove() drops that reference, so
the PTM enable counts of this device and of its upstream path stay
elevated with no bound driver, and the device's PTM control bits remain
set. pcim_enable_device() only arranges for pci_disable_device() and
does not undo the PTM enable.

Add the matching pci_disable_ptm() to the common unwind path.
pci_enable_ptm() failure is not fatal here, so guard the call with
pcie_ptm_enabled(): pci_disable_ptm() decrements dev->ptm_enable_cnt
unconditionally and then recurses upstream, so calling it after a failed
enable would drive this device's count negative and wrongly decrement
parents shared with other endpoints.

This issue was identified during our ongoing static-analysis research
while reviewing kernel code.

Fixes: 8d5e12c592 ("idpf: add initial PTP support")
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
[TN moved call due to commit 6b284aa2dd ("idpf: refactor idpf to use libie_pci APIs")]
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-09-08 13:10:35 -07:00
Myeonghun Pak
7dd4c829ba idpf: disable DIM work before freeing q_vectors
idpf never drains the Tx/Rx DIM works before freeing the memory they
live in.  tx_dim and rx_dim are embedded in struct idpf_q_vector, they
are queued from the NAPI poll via net_dim(), and idpf_vport_intr_rel()
ends with kfree(rsrc->q_vectors).  Nothing in the driver cancels them.

idpf_tx_dim_work() and idpf_rx_dim_work() then run on freed memory:
idpf_vport_intr_write_itr() writes the ITR register through
q_vector->intr_reg.tx_itr / rx_itr, void __iomem pointers loaded out of
the freed q_vector.  No configuration is needed to get there --
IDPF_ITR_IS_DYNAMIC() is defined as (itr_mode) and idpf_vport_alloc()
initialises both modes to IDPF_ITR_DYNAMIC.

Draining after idpf_vport_intr_napi_dis_all() is not enough on its own.
idpf_net_dim() is called from inside the
"if (napi_complete_done(napi, work_done))" branch of the poll, and
napi_complete_done() has already cleared NAPIF_STATE_SCHED by then.
napi_disable_locked() waits only while (val & (NAPIF_STATE_SCHED |
NAPIF_STATE_NPSVC)), so napi_disable() can return while the poll tail is
still queueing the work, and a plain cancel_work_sync() would be
re-armed behind the drain.

Use disable_work_sync(): schedule_work() on a work with a non-zero
disable count is dropped by clear_pending_if_disabled() before
__queue_work() is reached.

Move idpf_init_dim() to idpf_vport_intr_alloc() so the works are
initialised on every path that can reach the drain -- the three
"goto intr_deinit" sites between idpf_vport_intr_init() and
idpf_vport_intr_ena() get there without the enable side having run.
Nothing re-enables them: rsrc->q_vectors is freed on every exit from
idpf_vport_open() and on every idpf_vport_stop(), so the count dies with
the object.

It is a race, not a deterministic failure -- net_dim() only schedules
once DIM_NEVENTS events have accumulated and the profile index changes.
A KASAN ifup/ifdown loop under load is the way to see it.

Fixes: c2d548cad1 ("idpf: add TX splitq napi poll support")
Fixes: 3a8845af66 ("idpf: add RX splitq napi poll support")
Cc: <stable@vger.kernel.org> # see patch description, needs adjustments for <= 6.9
Co-developed-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Ijae Kim <ae878000@gmail.com>
Signed-off-by: Myeonghun Pak <mhun512@gmail.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2026-09-08 13:10:35 -07:00
Linus Walleij
e89e88ad41 net: ethernet: cortina: Count RX descriptors for freeq refill
The software free queue provides one buffer fragment for every descriptor
moved to an RX queue. The refill heuristic instead advances by NAPI work,
which counts frames. A fragmented or discarded frame can consume several
queue entries while adding only one to the refill count.

Count the RX descriptors as they are consumed and report that separately
from NAPI work. Use the descriptor count to drive free queue refills.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Assisted-by: LLM
Reviewed-by: Joe Damato <joe@dama.to>
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260903-gemini-ethernet-fixes-v2-5-2bbbd598ca6e@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 12:35:22 +02:00
Linus Walleij
6520198c43 net: ethernet: cortina: Count RX drops once per frame
The absence of a partial skb means either that the driver is not
assembling a frame or that the current frame was already dropped.
Consequently, repeated descriptor errors can increment rx_dropped more
than once, while an orphaned descriptor chain can reach EOF without being
counted at all.

Track the dropping state across NAPI polls. Clear it at frame boundaries
and route mapping failures and orphaned continuations through the common
drop path so each discarded frame is counted exactly once.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Reported-by: Joe Damato <joe@dama.to>
Closes: https://lore.kernel.org/netdev/apdK5aMmvYssz35F@devvm20253.cco0.facebook.com/
Assisted-by: LLM
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260903-gemini-ethernet-fixes-v2-4-2bbbd598ca6e@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 12:35:22 +02:00
Linus Walleij
b856c552f5 net: ethernet: cortina: Count dropped frames as NAPI work
The RX loop only consumes budget when it successfully delivers a frame.
Error paths keep consuming descriptors without reducing the budget, so a
stream of bad frames can process the entire receive ring in one poll.

Move the budget accounting to a common end-of-frame path. This counts
each completed frame as NAPI work whether it was delivered or dropped,
matching the behavior of the vendor driver.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Assisted-by: LLM
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260903-gemini-ethernet-fixes-v2-3-2bbbd598ca6e@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 12:35:21 +02:00
Linus Walleij
baa26841cb net: ethernet: cortina: Finish RX updates before NAPI completion
napi_complete_done() releases ownership of the NAPI instance, but the
Gemini poll keeps the RX statistics writer section open and updates the
free queue after calling it. A new poll can therefore start while the old
writer is still active.

Finish the statistics and free queue updates before releasing ownership.
Only re-enable RX interrupts when napi_complete_done() reports successful
completion.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Suggested-by: Joe Damato <joe@dama.to>
Assisted-by: LLM
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260903-gemini-ethernet-fixes-v2-2-2bbbd598ca6e@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 12:35:21 +02:00
Linus Walleij
a0de06d0da net: ethernet: cortina: Fix budget accounting
The gmac_rx() function returns the remaining NAPI budget, but its
caller treats the return value as the number of packets received. An
idle poll therefore reports a full budget and remains scheduled.

Return the number of received packets instead. Preserve the existing
free queue refill accounting by adding that count directly; continuing
to subtract it from the budget would invert the refill behavior.

Fixes: 4d5ae32f5e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Link: https://lore.kernel.org/r/20260509-gemini-ethernet-fixes-v1-4-6c5d20ddc35b@kernel.org
Link: https://lore.kernel.org/r/20260512131456.189452-1-pabeni@redhat.com
Assisted-by: LLM
Reviewed-by: Joe Damato <joe@dama.to>
Signed-off-by: Linus Walleij <linusw@kernel.org>
Link: https://patch.msgid.link/20260903-gemini-ethernet-fixes-v2-1-2bbbd598ca6e@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 12:35:21 +02:00
Aleksei Sviridkin
2b6c0e25a3 net: macb: reject an unknown link speed in the taprio setup
speed is a u32, so SPEED_UNKNOWN arrives as 0xffffffff and passes the
"speed <= 0" check, which only ever catches zero. That is what an
autonegotiating link reports while it is down: the limit derived from
the speed collapses to a nanosecond at most and the first entry fails
with a misleading "exceeds hardware limit". Zero stays covered, it is
what an interface that was never opened reports, and
enst_max_hw_interval() divides by it. Say which case it was in the
error.

Fixes: 89934dbf16 ("net: macb: Add TAPRIO traffic scheduling support")
Assisted-by: LLM
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Link: https://patch.msgid.link/20260903123652.23900-3-f@lex.la
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 11:46:34 +02:00
Aleksei Sviridkin
0523d5c52a net: macb: zero the link settings taprio reads back
macb_taprio_setup_replace() calls phylink_ethtool_ksettings_get() with
an uninitialised kset, and kset is not only an out-parameter. On a
fixed link, or an in-band link with no PHY, phylink writes speed and
duplex only if kset->base.rate_matching already reads RATE_MATCH_NONE,
a field it never writes itself; in PHY mode before the PHY is attached
it writes port and supported and nothing more. Either way the speed
read back afterwards can be stack garbage. The ethtool core zeroes the
structure on every path into the op, which is why its callers never
see this; taprio is the only in-kernel caller passing its own variable.

Fixes: 89934dbf16 ("net: macb: Add TAPRIO traffic scheduling support")
Assisted-by: LLM
Signed-off-by: Aleksei Sviridkin <f@lex.la>
Link: https://patch.msgid.link/20260903123652.23900-2-f@lex.la
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 11:46:34 +02:00
Joe Damato
c0aceaf65b bnxt_en: Bound SW TPA IDs to prevent crashes
FW supports up to 1024 concurrent TPAs, so the FW TPA ID is in the range
0..1023 (see commit ec4d8e7cf0 ("bnxt_en: Add TPA ID mapping logic for
57500 chips.")). bnxt_alloc_agg_idx is intended to wrap the FW ID down to a
software ID which is used to index rxr->rx_tpa, and to generate a mapping
between FW IDs and the wrapped software ID.

On a 57608 with firmware version 233, the firmware advertises 32
concurrent TPAs. As of the commit under fixes, bp->max_tpa on this NIC
is set to 32.

If the software ID from bnxt_alloc_agg_idx is above 31, this results in
an invalid address being loaded on this line:

  tpa_info = &rxr->rx_tpa[agg_id];

because rx_tpa is allocated with only bp->max_tpa (32) entries. Writes
to tpa_info later in the code are out of bounds.

This bug results in a crash at boot:

Oops: general protection fault, kernel NULL pointer dereference 0x8: 0000 [#1] SMP NOPTI
RIP: 0010:bnxt_rx_pkt+0xc0/0x1560
RSP: 0018:ffffc900009b8c78 EFLAGS: 00010246
RAX: 0000000000000000 RBX: 0000000000000048 RCX: 0000000206682516
RDX: ffffc900009b8db4 RSI: 0000000000000000 RDI: 01ffffff038fe1c0
RBP: ffffc9006e687480 R08: ffffc9006e687000 R09: 0000000000003048
R10: 0000000000000480 R11: ffff8881c6083900 R12: 0000000006682516
R13: ffff8881c6095400 R14: 0000000000000016 R15: ffff8881c6b66680
FS:  0000000000000000(0000) GS:ffff88fef3c77000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fc8bda40584 CR3: 000000807c812001 CR4: 0000000008772ef0
PKRU: 55555554
Call Trace:
 <IRQ>
 ? __netif_receive_skb_list_core+0x1ca/0x250
 __bnxt_poll_work+0x152/0x280
 bnxt_poll_p5+0x1cd/0x480
 __napi_poll+0x30/0x180
 net_rx_action+0x20b/0x3b0
 ? note_gp_changes+0x53/0xe0
 ? tick_setup_sched_timer+0x180/0x180
 ? __napi_schedule+0x9a/0xb0
 ? bnxt_msix+0x24/0x30
 handle_softirqs+0xdd/0x2c0
 __irq_exit_rcu.llvm.3171231171502365008+0x47/0xf0
 common_interrupt+0x85/0x90
 </IRQ>
 <TASK>
 asm_common_interrupt+0x22/0x40

This stack trace is from a crash triggered when an out of bounds rx_tpa
is dereferenced. The invalid write mentioned above is silent in this
particular crash.

Fix this by allocating rx_tpa with bp->max_tpa rounded up to the next
power of 2 (bp->max_tpa_roundup_size) entries and masking the FW TPA ID
with that size, so the wrapped ID can never index past the end of the
array.

Fixes: 54c28fab2f ("bnxt_en: Set bp->max_tpa according to what the FW supports")
Reported-by: Raphael Cardoso Fernandes <raphaelcf@meta.com>
Suggested-by: Michael Chan <michael.chan@broadcom.com>
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-7-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:03 +02:00
Joe Damato
8e6a850c07 bnxt_en: Propagate RX ring init failures in bnxt_init_nic()
bnxt_init_rx_rings() returns an error when bnxt_alloc_one_rx_ring()
fails, but bnxt_init_nic() discards that return value and calls
bnxt_init_chip(), which enables TPA.

If an allocation fails, this could leave rxr->rx_tpa[] partially zeroed
and TPA would be enabled over an array with zeroed entries. This would
lead to a zeroed DMA address being handed out if the agg_idx is
translated to a SW index at a zeroed entry.

Fix this by propagating the error out of bnxt_init_nic(). Both callers
already check its return value and unwind with bnxt_free_skbs() and
bnxt_free_mem(), which tolerate a partially initialized RX ring.

Fixes: c0c050c58d ("bnxt_en: New Broadcom ethernet driver.")
Reported-by: Sashiko <sashiko-bot+sashiko@kernel.org>
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260828190900.1767611-1-joe%40dama.to
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-6-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:03 +02:00
Joe Damato
961e2a17c5 bnxt_en: Handle buffer allocation failure in bnxt_rx_ring_reset()
bnxt_rx_ring_reset() frees the ring buffers and then reallocates them,
ignoring the result.

bnxt_alloc_one_rx_ring() can fail in bnxt_alloc_one_tpa_info_data(), which
returns -ENOMEM on the first failed allocation and leaves the remaining
rxr->rx_tpa[] entries zeroed.

The error isn't propagated up, so the loop in bnxt_rx_ring_reset
continues and at the end the code re-enables TPA with partially
unallocated rx_tpa array.

This means that when the agg_id from hardware is mapped to a SW index in
rxr->rx_tpa[], an uninitialized slot can be chosen which would hand a
zero DMA address to the device.

Fix this by falling back to a global reset, which is what the existing
code already does when other functions fail, but unlike the other
failure cases this particular failure has to return because TPA can't
be re-enabled since the allocation failed.

Fixes: 8fbf58e17d ("bnxt_en: Implement RX ring reset in response to buffer errors.")
Reported-by: Sashiko <sashiko-bot+sashiko@kernel.org>
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260828190900.1767611-1-joe%40dama.to
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-5-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:03 +02:00
Joe Damato
b814dfbfeb bnxt_en: Propagate TPA buffer allocation failures in bnxt_queue_mem_alloc()
bnxt_alloc_one_tpa_info_data() returns -ENOMEM as soon as one allocation
fails. This leaves the remaining rxr->rx_tpa[] entries zeroed.

bnxt_queue_mem_alloc() discards that return value, so the partially
initialized ring is installed by bnxt_queue_start().

Since the agg_id is picked by the hardware and bnxt_alloc_agg_idx maps
it to a SW index in rxr->rx_tpa[], it is possible that an uninitialized
slot can be chosen which would hand a zero DMA address to the device.

Fix this by checking the return value of bnxt_alloc_one_tpa_info_data
and unwinding, freeing the ring buffers.

Fixes: bd649c5cc9 ("bnxt_en: handle tpa_info in queue API implementation")
Reported-by: Sashiko <sashiko-bot+sashiko@kernel.org>
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260828190900.1767611-1-joe%40dama.to
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-4-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:02 +02:00
Joe Damato
5ce7f36c33 bnxt_en: Don't free the live ring's TPA state on queue restart failure
bnxt_queue_mem_alloc() shallow copies the live RX ring into the clone:

  memcpy(clone, rxr, sizeof(*rxr));

the code currently clears pointers that the clone owns (such as
rx_agg_bmap), but rx_tpa and rx_tpa_idx_map are left pointing at memory
of the live ring that was cloned.

If an allocation failure happens later and the err_free_tpa_info label
is taken, the live ring's memory can be freed while still in use.

Fix this by initializing the clone's pointers to NULL to prevent live
ring state from being freed inadvertently.

Fixes: bd649c5cc9 ("bnxt_en: handle tpa_info in queue API implementation")
Reported-by: Sashiko <sashiko-bot+sashiko@kernel.org>
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260828190900.1767611-1-joe%40dama.to
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-3-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:02 +02:00
Joe Damato
4e17b5007b bnxt_en: Only restore LRO if the device supports TPA
With a P5+ device with firmware that reports max_aggs_supported == 0, it is
possible to make LRO settable by attaching and detaching an XDP program
even though the device does not support TPA.

Fix this by testing BNXT_SUPPORTS_TPA before restoring the feature bit.

Fixes: f0aa6a37a3 ("eth: bnxt: always recalculate features after XDP clearing, fix null-deref")
Reported-by: Sashiko <sashiko-bot+sashiko@kernel.org>
Link: https://netdev-ai.bots.linux.dev/sashiko/#/patchset/20260828190900.1767611-1-joe%40dama.to
Cc: stable@vger.kernel.org
Signed-off-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260902015652.2421609-2-joe@dama.to
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2026-09-08 10:47:02 +02:00
Vineeth Karumanchi
38b6be1010 net: macb: fix NULL pointer dereference on unbind with fixed-link
When the device tree describes a fixed-link and has no "mdio" child
node, macb_mii_init() returns early without allocating the MDIO bus,
leaving bp->mii_bus as NULL.

Two cleanup paths then dereference this NULL bus:

1. On driver unbind, macb_remove() unconditionally calls
   mdiobus_unregister(bp->mii_bus), which oopses:

  Unable to handle kernel NULL pointer dereference at virtual address 00000000000004a8
  pc : mdiobus_unregister+0x14/0xa4
  lr : macb_remove+0x38/0xa4
  Call trace:
   mdiobus_unregister+0x14/0xa4 (P)
   macb_remove+0x38/0xa4
   platform_remove+0x20/0x30
   device_release_driver_internal+0x1c8/0x224
   unbind_store+0xb4/0xbc

2. On the probe error path in macb_probe(), reached when
   macb_mii_init() has succeeded but a subsequent step fails, the
   err_out_unregister_mdio label runs the same unconditional cleanup.

mdiobus_unregister() and mdiobus_free() do not guard against a NULL
bus, so guard the calls in both macb_remove() and the probe error
path.

Fixes: d0c3601f2c ("net: macb: Avoid 20s boot delay by skipping MDIO bus registration for fixed-link PHY")
Signed-off-by: Vineeth Karumanchi <vineeth.karumanchi@amd.com>
Reviewed-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn>
Reviewed-by: Nicolai Buchwitz <nb@tipi-net.de>
Link: https://patch.msgid.link/20260902102836.2019355-1-vineeth.karumanchi@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-05 13:58:52 -07:00
Carolina Jubran
df99553f84 net/mlx5e: Keep HW timestamp stats monotonic across reconfiguration
`mlx5e_stats_ts_get()` currently selects either DMA or port timestamp
counters based on `tx_ptp_opened`. This flag is intentionally kept set
once the PTP TX queues have been opened so their statistics remain
available after queue teardown. As a result, DMA timestamps are no
longer reported after switching from port timestamping back to DMA
timestamping.

The function also reads statistics only from the currently active
channels and TCs. Reducing the number of channels or TCs can therefore
drop previously accumulated timestamp counters from the reported value.

Read the persistent channel statistics instead and always include DMA
timestamp counters. Once the PTP TX queues have been opened, also
include the port timestamp counters.

This also drops state_lock. It previously protected live channel/PTP
pointers, the new code only reads persistent channel_stats and
ptp_stats via mlx5e_stats_nch_read(), which is already safe for
lockless stats access.

Fixes: 3579032c08 ("net/mlx5e: Implement ethtool hardware timestamping statistics")
Signed-off-by: Carolina Jubran <cjubran@nvidia.com>
Reviewed-by: Shahar Shitrit <shshitrit@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260902193731.3668958-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-05 13:23:37 -07:00
Lama Kayal
c0c6f4ba8a net/mlx5: E-Switch, prevent mc_list repopulation during vport disable
In mlx5_esw_vport_disable(), move esw_apply_vport_rx_mode() ahead
of esw_vport_change_handle_locked() so vport->allmulti_rule is
NULL before the change handler observes it.

During FW-fatal recovery the disable runs while dev->state ==
INTERNAL_ERROR. The promisc query inside esw_update_vport_rx_mode()
fails and returns early, leaving vport->allmulti_rule intact, so
esw_update_vport_mc_promisc() runs and adds MLX5_ACTION_ADD entries
to vport->mc_list whose flow rules are then installed in the FDB
by esw_add_mc_addr(). esw_destroy_legacy_table() tears down the
FDB with those refs still held, corrupting the sub-tree and
leaving dangling flow_rule pointers in vport->mc_list.

Two-stage failure on `echo 1 > /sys/bus/pci/devices/<bdf>/reset`:

  refcount_t: underflow; use-after-free.
   tree_put_node+0xef/0x110 [mlx5_core]
   clean_tree+0x44/0xd0 [mlx5_core] (x5)
   mlx5_fs_core_cleanup+0x57/0x1c0 [mlx5_core]
   mlx5_unload+0x65/0xd0 [mlx5_core]
   ... mlx5_health_try_recover

  BUG: unable to handle page fault for address: 0000000003000055
   down_write+0x1c/0x60
   mlx5_del_flow_rules+0x33/0x1f0 [mlx5_core]
   esw_del_mc_addr+0x7b/0x170 [mlx5_core]
   esw_apply_vport_addr_list+0x56/0xf0 [mlx5_core]
   esw_vport_change_handle_locked+0x28b/0x310 [mlx5_core]
   mlx5_esw_vport_enable+0x270/0x4a0 [mlx5_core]
   ... mlx5_load ... mlx5_health_try_recover

esw_apply_vport_rx_mode(false, false) clears vport->allmulti_rule
via its local state machine even when the FW del fails. With the
rule NULL the !IS_ERR_OR_NULL(allmulti_rule) gate in the change
handler closes, no rules are installed during disable, and the
reload starts with a clean mc_list.

Fixes: 922f56e9a7 ("net/mlx5: Fix steering rules cleanup")
Signed-off-by: Lama Kayal <lkayal@nvidia.com>
Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260902193854.3669035-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-05 13:21:43 -07:00
Yael Chemla
7ee07f601f net/mlx5: E-Switch: fix use-after-free in mlx5_eswitch_termtbl_put
In mlx5_eswitch_termtbl_put(), the zero-ref cleanup check reads
tt->ref_count after termtbl_mutex has been released.  Two concurrent
callers on the same mlx5_termtbl_handle race: one decrements ref_count
to zero, removes the hash entry, and calls kfree(tt) while the other
has already dropped the mutex and is about to evaluate
if (!tt->ref_count), producing a use-after-free.

Fix this by capturing the result of the decrement into a stack-local
last variable before dropping the mutex.  The cleanup decision is now
made entirely under termtbl_mutex, and tt is not touched after
kfree.

Fixes: 10caabdaad ("net/mlx5e: Use termination table for VLAN push actions")
Signed-off-by: Yael Chemla <ychemla@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260902193514.3668880-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-05 13:20:26 -07:00
Carolina Jubran
af3aef0245 net/mlx5e: Fix use-after-free race in sample_restore_put()
Concurrent teardown of TC sample rules sharing the same restore
context may re-read restore->count after dropping restore_lock.
At that point another thread may already have completed cleanup and
freed the restore object.

Use the result of the refcount decrement while holding restore_lock to
determine whether cleanup is needed.

Fixes: 36a3196256 ("net/mlx5e: TC, Add sampler restore handle API")
Signed-off-by: Carolina Jubran <cjubran@nvidia.com>
Reviewed-by: Shahar Shitrit <shshitrit@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260902193341.3668809-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2026-09-05 13:19:22 -07:00