net: usb: Convert tasklet API to new bottom half workqueue mechanism

Migrate tasklet APIs to the new bottom half workqueue mechanism. It
replaces all occurrences of tasklet usage with the appropriate workqueue
APIs throughout the usbnet driver. This transition ensures compatibility
with the latest design and enhances performance.

Signed-off-by: Jun Miao <jun.miao@intel.com>
Link: https://patch.msgid.link/20250618173923.950510-1-jun.miao@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

igc: Make the const read-only array supported_sizes static

Don't populate the const read-only array supported_sizes on the
stack at run time, instead make it static.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Vitaly Lifshits <vitaly.lifshits@intel.com>>
Link: https://patch.msgid.link/20250618135408.1784120-1-colin.i.king@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'convert-lan78xx-driver-to-the-phylink'

Oleksij Rempel says:

====================
convert lan78xx driver to the PHYLINK

This series converts the lan78xx driver to use the PHYLINK framework,
which enhances PHY and MAC management. The changes also streamline the
driver by removing unused elements and improving link status reporting.

This is the final part of the previously split conversion series:
https://lore.kernel.org/all/20250428130542.3879769-1-o.rempel@pengutronix.de/

Tested on EVB_LAN7850.
====================

Link: https://patch.msgid.link/20250618122602.3156678-1-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: remove unused struct members

Remove unused members from struct lan78xx_net, including:

    driver_priv
    suspend_count
    mdix_ctrl

These fields are no longer used in the driver and can be safely removed
as part of a cleanup.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-7-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: Integrate EEE support with phylink LPI API

Refactor Energy-Efficient Ethernet (EEE) support in the LAN78xx driver to
fully integrate with the phylink Low Power Idle (LPI) API. This includes:

- Replacing direct calls to `phy_ethtool_get_eee` and `phy_ethtool_set_eee`
  with `phylink_ethtool_get_eee` and `phylink_ethtool_set_eee`.
- Implementing `.mac_enable_tx_lpi` and `.mac_disable_tx_lpi` to control
  LPI transitions via phylink.
- Configuring `lpi_timer_default` to align with recommended values from
  LAN7800 documentation.
- ensure EEE is disabled on controller reset

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-6-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: port link settings to phylink API

Refactor lan78xx_get_link_ksettings and lan78xx_set_link_ksettings to
use the phylink API (phylink_ethtool_ksettings_get and
phylink_ethtool_ksettings_set) instead of directly interfacing with the
PHY. This change simplifies the code and ensures better integration with
the phylink framework for link management.

Additionally, the explicit calls to usb_autopm_get_interface() and
usb_autopm_put_interface() have been removed. These were originally
needed to manage USB power management during register accesses. However,
lan78xx_mdiobus_read() and lan78xx_mdiobus_write() already handle USB
auto power management internally, ensuring that the interface remains
active when necessary. Since there are no other direct register accesses
in these functions that require explicit power management handling, the
extra calls have become redundant and are no longer needed.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-5-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: Use ethtool_op_get_link to reflect current link status

Replace the custom lan78xx_get_link implementation with the standard
ethtool_op_get_link helper, which uses netif_carrier_ok to reflect
the current link status accurately.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-4-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: Rename EVENT_LINK_RESET to EVENT_PHY_INT_ACK

The EVENT_LINK_RESET macro currently triggers deferred work after a PHY
interrupt. Prior to PHYLINK conversion, this work included reconfiguring
the MAC and PHY, effectively performing a 'link reset'.

However, after porting the driver to the PHYLINK framework, the logic
associated with this event now solely handles the acknowledgment of
the PHY interrupt. The MAC and PHY reconfiguration is now managed by
PHYLINK's dedicated callbacks.

To accurately reflect its current, narrowed functionality, rename
EVENT_LINK_RESET to EVENT_PHY_INT_ACK.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-3-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: lan78xx: Convert to PHYLINK for improved PHY and MAC management

Convert the LAN78xx USB Ethernet driver to use the PHYLINK framework for
managing PHY and MAC interactions. This improves consistency with other
network drivers, simplifies pause frame handling, and enables cleaner
suspend/resume support.

Key changes:
- Replace all PHYLIB-based logic with PHYLINK equivalents:
  - Replace phy_connect()/phy_disconnect() with phylink_connect_phy()
  - Replace phy_start()/phy_stop() with phylink_start()/phylink_stop()
  - Replace pauseparam handling with phylink_ethtool_get/set_pauseparam()
- Introduce lan78xx_phylink_setup() to configure PHYLINK
- Add phylink MAC operations:
  - lan78xx_mac_config()
  - lan78xx_mac_link_up()
  - lan78xx_mac_link_down()
- Remove legacy link state handling:
  - lan78xx_link_status_change()
  - lan78xx_link_reset()
- Handle fixed-link fallback for LAN7801 using phylink_set_fixed_link()
- Replace deprecated flow control handling with phylink-managed logic

Power management:
- Switch suspend/resume paths to use phylink_suspend()/phylink_resume()
- Ensure proper use of rtnl_lock() where required
- Note: full runtime testing of power management is currently limited
  due to hardware setup constraints

Note: Conversion of EEE (Energy Efficient Ethernet) handling to the
PHYLINK-managed API will be done in a follow-up patch. For now, the
legacy EEE enable logic is preserved in mac_link_up().

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250618122602.3156678-2-o.rempel@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: gianfar: Use device_get_named_child_node_count()

We can avoid open-coding the loop construct which counts firmware child
nodes with a specific name by using the newly added
device_get_named_child_node_count().

The gianfar driver has such open-coded loop. Replace it with the
device_get_child_node_count_named().

Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Matti Vaittinen <mazziesaccount@gmail.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/3a33988fc042588cb00a0bfc5ad64e749cb0eb1f.1750248902.git.mazziesaccount@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-fec-general-vlan-cleanups'

Marc Kleine-Budde says:

====================
net: fec: general + VLAN cleanups

This series first cleans up the fec driver a bit (typos, obsolete
comments, add missing header files, rename struct, replace magic
number by defines).

The last 5 patches clean up the fec_enet_rx_queue() function,
including VLAN handling.

v3: https://patch.msgid.link/20250617-fec-cleanups-v3-0-a57bfb38993f@pengutronix.de
v2: https://patch.msgid.link/20250612-fec-cleanups-v2-0-ae7c36df185e@pengutronix.de
v1: https://patch.msgid.link/20241016-fec-cleanups-v1-0-de783bd15e6a@pengutronix.de
====================

Link: https://patch.msgid.link/20250618-fec-cleanups-v4-0-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_enet_rx_queue(): factor out VLAN handling into separate function fec_enet_rx_vlan()

In order to clean up of the VLAN handling, factor out the VLAN
handling into separate function fec_enet_rx_vlan().

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-11-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_enet_rx_queue(): move_call to _vlan_hwaccel_put_tag()

Move __vlan_hwaccel_put_tag() into the if statement that sets
vlan_packet_rcvd = true. This change eliminates the unnecessary
vlan_packet_rcvd variable, simplifying the code and improving clarity.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-10-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_enet_rx_queue(): reduce scope of data

In order to clean up of the VLAN handling, reduce the scope of data.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-9-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_enet_rx_queue(): replace manual VLAN header calculation with skb_vlan_eth_hdr()

For better readability and maintainability, use the provided helper function
skb_vlan_eth_hdr() to replace manual the VLAN header calculation, and change
the type of vlan_header to struct vlan_ethhdr to take into account that the
Ethernet header plus VLAN header is returned.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-8-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_enet_rx_queue(): use same signature as fec_enet_tx_queue()

There are the functions fec_enet_rx_queue() and fec_enet_tx_queue(),
one for handling the RX queue the other one handles the TX queue.

However they don't have the same signature. Align fec_enet_rx_queue()
argument order with fec_enet_tx_queue() to make code more readable.

Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-7-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fec_restart(): introduce a define for FEC_ECR_SPEED

Replace "1 << 5" for configuring 1000 MBit/s with a defined constant to
improve code readability and maintainability.

Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-6-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: rename struct fec_devinfo fec_imx6x_info -> fec_imx6sx_info

In da722186f654 ("net: fec: set GPR bit on suspend by DT
configuration.") the platform_device_id fec_devtype::driver_data was
converted from holding the quirks to a pointing to struct fec_devinfo.

The struct fec_devinfo holding the information for the i.MX6SX was
named fec_imx6x_info.

Rename fec_imx6x_info to fec_imx6sx_info to align with the SoC's name.

Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-5-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: sort the includes by alphabetic order

This is a preparation patch to make addition of new includes easier
without breaking the alphabetic order.

Suggested-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-4-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: switch from asm/cacheflush.h to linux/cacheflush.h

To fix the checkpatch warning, use linux/cacheflush.h instead of
asm/cacheflush.h.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Reviewed-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-3-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: struct fec_enet_private: remove obsolete comment

In commit 4d494cdc92b3 ("net: fec: change data structure to support
multiqueue") the data structures were changed, so that the comment about
the sent-in-place skb doesn't apply any more. Remove it.

Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-2-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: fec: fix typos found by codespell

codespell has found some typos in the comments, fix them.

Reviewed-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Frank Li <Frank.Li@nxp.com>
Reviewed-by: Csókás, Bence <csokas.bence@prolan.hu>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://patch.msgid.link/20250618-fec-cleanups-v4-1-c16f9a1af124@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-stmmac-loongson1-cleanups'

Russell King says:

====================
net: stmmac: loongson1: cleanups

A couple of patches to cleanup loongson1. First, introducing a match
data struct to allow the per-match data to be extended beyond the init
function pointer, and then adding a setup method to allow the resource
base address to be translated to the MAC index at probe time rather
than repeatedly in the setup function.
====================

Link: https://patch.msgid.link/aFKXzlno7HkG-cNh@shell.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: stmmac: loongson1: get ls1b resource only once

ls1b_dwmac_syscon_init() was getting the stmmac iomem resource to detect
which GMAC block is being used. Move this to a separate setup() function
that only runs at probe time, so it can sensibly behave with an
unrecognised resource adress. Use this to set a MAC index (id) which is
then used in place of testing the base address.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Keguang Zhang <keguang.zhang@gmail.com>
Tested-by: Keguang Zhang <keguang.zhang@gmail.com> # on LS1B & LS1C
Link: https://patch.msgid.link/E1uRqEE-004c7M-Go@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: stmmac: loongson1: provide match data struct

Provide a structure for match data rather than using the function
pointer as match data. This allows stronger type-checking for the
function itself, and allows extensions to the match data.

Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Keguang Zhang <keguang.zhang@gmail.com>
Tested-by: Keguang Zhang <keguang.zhang@gmail.com> # on LS1B & LS1C
Link: https://patch.msgid.link/E1uRqE9-004c7G-CB@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: stmmac: replace ioaddr with stmmac_priv for pcs_set_ane() method

Pass the stmmac_priv structure into the pcs_set_ane() MAC method rather
than having callers dereferencing this structure for the IO address.

Tested-by: Bartosz Golaszewski <bartosz.golaszewski@linaro.org> # sa8775p-ride-r3
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/E1uRqbQ-004djP-1l@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Cross-merge networking fixes after downstream PR (net-6.16-rc3).

No conflicts or adjacent changes.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag 'net-6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from wireless.

  The ath12k fix to avoid FW crashes requires adding support for a
  number of new FW commands so it's quite large in terms of LoC. The
  rest is relatively small.

  Current release - fix to a fix:

   - ptp: fix breakage after ptp_vclock_in_use() rework

  Current release - regressions:

   - openvswitch: allocate struct ovs_pcpu_storage dynamically, static
     allocation may exhaust module loader limit on smaller systems

  Previous releases - regressions:

   - tcp: fix tcp_packet_delayed() for peers with no selective ACK
     support

  Previous releases - always broken:

   - wifi: ath12k: don't activate more links than firmware supports

   - tcp: make sure sockets open via passive TFO have valid NAPI ID

   - eth: bnxt_en: update MRU and RSS table of RSS contexts on queue
     reset, prevent Rx queues from silently hanging after queue reset

   - NFC: uart: set tty->disc_data only in success path"

* tag 'net-6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (59 commits)
  net: airoha: Differentiate hwfd buffer size for QDMA0 and QDMA1
  net: airoha: Compute number of descriptors according to reserved memory size
  tools: ynl: fix mixing ops and notifications on one socket
  net: atm: fix /proc/net/atm/lec handling
  net: atm: add lec_mutex
  mlxbf_gige: return EPROBE_DEFER if PHY IRQ is not available
  net: airoha: Always check return value from airoha_ppe_foe_get_entry()
  NFC: nci: uart: Set tty->disc_data only in success path
  calipso: Fix null-ptr-deref in calipso_req_{set,del}attr().
  MAINTAINERS: Remove Shannon Nelson from MAINTAINERS file
  net: lan743x: fix potential out-of-bounds write in lan743x_ptp_io_event_clock_get()
  eth: fbnic: avoid double free when failing to DMA-map FW msg
  tcp: fix passive TFO socket having invalid NAPI ID
  selftests: net: add test for passive TFO socket NAPI ID
  selftests: net: add passive TFO test binary
  selftests: netdevsim: improve lib.sh include in peer.sh
  tipc: fix null-ptr-deref when acquiring remote ip of ethernet bearer
  Octeontx2-pf: Fix Backpresure configuration
  net: ftgmac100: select FIXED_PHY
  net: ethtool: remove duplicate defines for family info
  ...

Merge tag 'hwmon-for-v6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:

 - ltc4282: Avoid repeated register write operation

 - occ: Fix unaligned accesses, and rework attribute registration to
   reduce stack usage

 - ftsteutates: Fix TOCTOU race

* tag 'hwmon-for-v6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (ltc4282) avoid repeated register write
  hwmon: (occ) fix unaligned accesses
  hwmon: (occ) Rework attribute registration for stack usage
  hwmon: (ftsteutates) Fix TOCTOU race in fts_read()

Merge branch 'net-airoha-improve-hwfd-buffer-descriptor-queues-setup'

Lorenzo Bianconi says:

====================
net: airoha: Improve hwfd buffer/descriptor queues setup

Compute the number of hwfd buffers/descriptors according to the reserved
memory size if provided via DTS.

Reduce the required hwfd buffers queue size for QDMA1.

v3: https://lore.kernel.org/20250618-airoha-hw-num-desc-v3-0-18a6487cd75e@kernel.org
v1: https://lore.kernel.org/20250615-airoha-hw-num-desc-v1-0-8f88daa4abd7@kernel.org
====================

Link: https://patch.msgid.link/20250619-airoha-hw-num-desc-v4-0-49600a9b319a@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: airoha: Differentiate hwfd buffer size for QDMA0 and QDMA1

EN7581 SoC allows configuring the size and the number of buffers in
hwfd payload queue for both QDMA0 and QDMA1.
In order to reduce the required DRAM used for hwfd buffers queues and
decrease the memory footprint, differentiate hwfd buffer size for QDMA0
and QDMA1 and reduce hwfd buffer size to 1KB for QDMA1 (WAN) while
maintaining 2KB for QDMA0 (LAN).

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250619-airoha-hw-num-desc-v4-2-49600a9b319a@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: airoha: Compute number of descriptors according to reserved memory size

In order to not exceed the reserved memory size for hwfd buffers,
compute the number of hwfd buffers/descriptors according to the
reserved memory size and the size of each hwfd buffer (2KB).

Fixes: 3a1ce9e3d01b ("net: airoha: Add the capability to allocate hwfd buffers via reserved-memory")
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20250619-airoha-hw-num-desc-v4-1-49600a9b319a@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag 'wireless-2025-06-18' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless

Johannes Berg says:

====================
More fixes:
 - ath12k
   - avoid busy-waiting
   - activate correct number of links
 - iwlwifi
   - iwldvm regression (lots of warnings)
   - iwlmld merge damage regression (crash)
   - fix build with some old gcc versions
 - carl9170: don't talk to device w/o FW [syzbot]
 - ath6kl: remove bad FW WARN [syzbot]
 - ieee80211: use variable-length arrays [syzbot]
 - mac80211
   - remove WARN on delayed beacon update [syzbot]
   - drop OCB frames with invalid source [syzbot]

* tag 'wireless-2025-06-18' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
  wifi: iwlwifi: Fix incorrect logic on cmd_ver range checking
  wifi: iwlwifi: dvm: restore n_no_reclaim_cmds setting
  wifi: iwlwifi: cfg: Limit cb_size to valid range
  wifi: iwlwifi: restore missing initialization of async_handlers_list (again)
  wifi: ath6kl: remove WARN on bad firmware input
  wifi: carl9170: do not ping device which has failed to load firmware
  wifi: ath12k: don't wait when there is no vdev started
  wifi: ath12k: don't use static variables in ath12k_wmi_fw_stats_process()
  wifi: ath12k: avoid burning CPU while waiting for firmware stats
  wifi: ath12k: fix documentation on firmware stats
  wifi: ath12k: don't activate more links than firmware supports
  wifi: ath12k: update link active in case two links fall on the same MAC
  wifi: ath12k: support WMI_MLO_LINK_SET_ACTIVE_CMDID command
  wifi: ath12k: update freq range for each hardware mode
  wifi: ath12k: parse and save sbs_lower_band_end_freq from WMI_SERVICE_READY_EXT2_EVENTID event
  wifi: ath12k: parse and save hardware mode info from WMI_SERVICE_READY_EXT_EVENTID event for later use
  wifi: ath12k: Avoid CPU busy-wait by handling VDEV_STAT and BCN_STAT
  wifi: mac80211: don't WARN for late channel/color switch
  wifi: mac80211: drop invalid source address OCB frames
  wifi: remove zero-length arrays
====================

Link: https://patch.msgid.link/20250618210642.35805-6-johannes@sipsolutions.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

tools: ynl: fix mixing ops and notifications on one socket

The multi message support loosened the connection between the request
and response handling, as we can now submit multiple requests before
we start processing responses. Passing the attr set to NlMsgs decoding
no longer makes sense (if it ever did), attr set may differ message
by messsage. Isolate the part of decoding responsible for attr-set
specific interpretation and call it once we identified the correct op.

Without this fix performing SET operation on an ethtool socket, while
being subscribed to notifications causes:

 # File "tools/net/ynl/pyynl/lib/ynl.py", line 1096, in _op
 # Exception|     return self._ops(ops)[0]
 # Exception|            ~~~~~~~~~^^^^^
 # File "tools/net/ynl/pyynl/lib/ynl.py", line 1040, in _ops
 # Exception|     nms = NlMsgs(reply, attr_space=op.attr_set)
 # Exception|                                    ^^^^^^^^^^^

The value of op we use on line 1040 is stale, it comes form the previous
loop. If a notification comes before a response we will update op to None
and the next iteration thru the loop will break with the trace above.

Fixes: 6fda63c45fe8 ("tools/net/ynl: fix cli.py --subscribe feature")
Fixes: ba8be00f68f5 ("tools/net/ynl: Add multi message support to ynl")
Reviewed-by: Donald Hunter <donald.hunter@gmail.com>
Link: https://patch.msgid.link/20250618171746.1201403-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'with-a-mutex'

Eric Dumazet says:

====================
net: atm: protect dev_lec[] with a mutex

Based on an initial syzbot report.

First patch is adding lec_mutex to address the report.

Second patch protects /proc/net/atm/lec operations.

We probably should delete this driver, it seems quite broken.
====================

Link: https://patch.msgid.link/20250618140844.1686882-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: atm: fix /proc/net/atm/lec handling

/proc/net/atm/lec must ensure safety against dev_lec[] changes.

It appears it had dev_put() calls without prior dev_hold(),
leading to imbalance and UAF.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com> # Minor atm contributor
Link: https://patch.msgid.link/20250618140844.1686882-3-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: atm: add lec_mutex

syzbot found its way in net/atm/lec.c, and found an error path
in lecd_attach() could leave a dangling pointer in dev_lec[].

Add a mutex to protect dev_lecp[] uses from lecd_attach(),
lec_vcc_attach() and lec_mcast_attach().

Following patch will use this mutex for /proc/net/atm/lec.

BUG: KASAN: slab-use-after-free in lecd_attach net/atm/lec.c:751 [inline]
BUG: KASAN: slab-use-after-free in lane_ioctl+0x2224/0x23e0 net/atm/lec.c:1008
Read of size 8 at addr ffff88807c7b8e68 by task syz.1.17/6142

CPU: 1 UID: 0 PID: 6142 Comm: syz.1.17 Not tainted 6.16.0-rc1-syzkaller-00239-g08215f5486ec #0 PREEMPT(full)
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 05/07/2025
Call Trace:
 <TASK>
  __dump_stack lib/dump_stack.c:94 [inline]
  dump_stack_lvl+0x116/0x1f0 lib/dump_stack.c:120
  print_address_description mm/kasan/report.c:408 [inline]
  print_report+0xcd/0x680 mm/kasan/report.c:521
  kasan_report+0xe0/0x110 mm/kasan/report.c:634
  lecd_attach net/atm/lec.c:751 [inline]
  lane_ioctl+0x2224/0x23e0 net/atm/lec.c:1008
  do_vcc_ioctl+0x12c/0x930 net/atm/ioctl.c:159
  sock_do_ioctl+0x118/0x280 net/socket.c:1190
  sock_ioctl+0x227/0x6b0 net/socket.c:1311
  vfs_ioctl fs/ioctl.c:51 [inline]
  __do_sys_ioctl fs/ioctl.c:907 [inline]
  __se_sys_ioctl fs/ioctl.c:893 [inline]
  __x64_sys_ioctl+0x18e/0x210 fs/ioctl.c:893
  do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
  do_syscall_64+0xcd/0x4c0 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
 </TASK>

Allocated by task 6132:
  kasan_save_stack+0x33/0x60 mm/kasan/common.c:47
  kasan_save_track+0x14/0x30 mm/kasan/common.c:68
  poison_kmalloc_redzone mm/kasan/common.c:377 [inline]
  __kasan_kmalloc+0xaa/0xb0 mm/kasan/common.c:394
  kasan_kmalloc include/linux/kasan.h:260 [inline]
  __do_kmalloc_node mm/slub.c:4328 [inline]
  __kvmalloc_node_noprof+0x27b/0x620 mm/slub.c:5015
  alloc_netdev_mqs+0xd2/0x1570 net/core/dev.c:11711
  lecd_attach net/atm/lec.c:737 [inline]
  lane_ioctl+0x17db/0x23e0 net/atm/lec.c:1008
  do_vcc_ioctl+0x12c/0x930 net/atm/ioctl.c:159
  sock_do_ioctl+0x118/0x280 net/socket.c:1190
  sock_ioctl+0x227/0x6b0 net/socket.c:1311
  vfs_ioctl fs/ioctl.c:51 [inline]
  __do_sys_ioctl fs/ioctl.c:907 [inline]
  __se_sys_ioctl fs/ioctl.c:893 [inline]
  __x64_sys_ioctl+0x18e/0x210 fs/ioctl.c:893
  do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
  do_syscall_64+0xcd/0x4c0 arch/x86/entry/syscall_64.c:94
 entry_SYSCALL_64_after_hwframe+0x77/0x7f

Freed by task 6132:
  kasan_save_stack+0x33/0x60 mm/kasan/common.c:47
  kasan_save_track+0x14/0x30 mm/kasan/common.c:68
  kasan_save_free_info+0x3b/0x60 mm/kasan/generic.c:576
  poison_slab_object mm/kasan/common.c:247 [inline]
  __kasan_slab_free+0x51/0x70 mm/kasan/common.c:264
  kasan_slab_free include/linux/kasan.h:233 [inline]
  slab_free_hook mm/slub.c:2381 [inline]
  slab_free mm/slub.c:4643 [inline]
  kfree+0x2b4/0x4d0 mm/slub.c:4842
  free_netdev+0x6c5/0x910 net/core/dev.c:11892
  lecd_attach net/atm/lec.c:744 [inline]
  lane_ioctl+0x1ce8/0x23e0 net/atm/lec.c:1008
  do_vcc_ioctl+0x12c/0x930 net/atm/ioctl.c:159
  sock_do_ioctl+0x118/0x280 net/socket.c:1190
  sock_ioctl+0x227/0x6b0 net/socket.c:1311
  vfs_ioctl fs/ioctl.c:51 [inline]
  __do_sys_ioctl fs/ioctl.c:907 [inline]
  __se_sys_ioctl fs/ioctl.c:893 [inline]
  __x64_sys_ioctl+0x18e/0x210 fs/ioctl.c:893

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reported-by: syzbot+8b64dec3affaed7b3af5@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6852c6f6.050a0220.216029.0018.GAE@google.com/T/#u
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250618140844.1686882-2-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

mlxbf_gige: return EPROBE_DEFER if PHY IRQ is not available

The message "Error getting PHY irq. Use polling instead"
is emitted when the mlxbf_gige driver is loaded by the
kernel before the associated gpio-mlxbf driver, and thus
the call to get the PHY IRQ fails since it is not yet
available. The driver probe() must return -EPROBE_DEFER
if acpi_dev_gpio_irq_get_by() returns the same.

Fixes: 6c2a6ddca763 ("net: mellanox: mlxbf_gige: Replace non-standard interrupt handling")
Signed-off-by: David Thompson <davthompson@nvidia.com>
Reviewed-by: Asmaa Mnebhi <asmaa@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250618135902.346-1-davthompson@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: airoha: Always check return value from airoha_ppe_foe_get_entry()

airoha_ppe_foe_get_entry routine can return NULL, so check the returned
pointer is not NULL in airoha_ppe_foe_flow_l2_entry_update()

Fixes: b81e0f2b58be3 ("net: airoha: Add FLOW_CLS_STATS callback support")
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20250618-check-ret-from-airoha_ppe_foe_get_entry-v2-1-068dcea3cc66@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

NFC: nci: uart: Set tty->disc_data only in success path

Setting tty->disc_data before opening the NCI device means we need to
clean it up on error paths.  This also opens some short window if device
starts sending data, even before NCIUARTSETDRIVER IOCTL succeeded
(broken hardware?).  Close the window by exposing tty->disc_data only on
the success path, when opening of the NCI device and try_module_get()
succeeds.

The code differs in error path in one aspect: tty->disc_data won't be
ever assigned thus NULL-ified.  This however should not be relevant
difference, because of "tty->disc_data=NULL" in nci_uart_tty_open().

Cc: Linus Torvalds <torvalds@linuxfoundation.org>
Fixes: 9961127d4bce ("NFC: nci: add generic uart support")
Cc: <stable@vger.kernel.org>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Link: https://patch.msgid.link/20250618073649.25049-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

calipso: Fix null-ptr-deref in calipso_req_{set,del}attr().

syzkaller reported a null-ptr-deref in sock_omalloc() while allocating
a CALIPSO option.  [0]

The NULL is of struct sock, which was fetched by sk_to_full_sk() in
calipso_req_setattr().

Since commit a1a5344ddbe8 ("tcp: avoid two atomic ops for syncookies"),
reqsk->rsk_listener could be NULL when SYN Cookie is returned to its
client, as hinted by the leading SYN Cookie log.

Here are 3 options to fix the bug:

  1) Return 0 in calipso_req_setattr()
  2) Return an error in calipso_req_setattr()
  3) Alaways set rsk_listener

1) is no go as it bypasses LSM, but 2) effectively disables SYN Cookie
for CALIPSO.  3) is also no go as there have been many efforts to reduce
atomic ops and make TCP robust against DDoS.  See also commit 3b24d854cb35
("tcp/dccp: do not touch listener sk_refcnt under synflood").

As of the blamed commit, SYN Cookie already did not need refcounting,
and no one has stumbled on the bug for 9 years, so no CALIPSO user will
care about SYN Cookie.

Let's return an error in calipso_req_setattr() and calipso_req_delattr()
in the SYN Cookie case.

This can be reproduced by [1] on Fedora and now connect() of nc times out.

[0]:
TCP: request_sock_TCPv6: Possible SYN flooding on port [::]:20002. Sending cookies.
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000006: 0000 [#1] PREEMPT SMP KASAN NOPTI
KASAN: null-ptr-deref in range [0x0000000000000030-0x0000000000000037]
CPU: 3 UID: 0 PID: 12262 Comm: syz.1.2611 Not tainted 6.14.0 #2
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.3-0-ga6ed6b701f0a-prebuilt.qemu.org 04/01/2014
RIP: 0010:read_pnet include/net/net_namespace.h:406 [inline]
RIP: 0010:sock_net include/net/sock.h:655 [inline]
RIP: 0010:sock_kmalloc+0x35/0x170 net/core/sock.c:2806
Code: 89 d5 41 54 55 89 f5 53 48 89 fb e8 25 e3 c6 fd e8 f0 91 e3 00 48 8d 7b 30 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <80> 3c 02 00 0f 85 26 01 00 00 48 b8 00 00 00 00 00 fc ff df 4c 8b
RSP: 0018:ffff88811af89038 EFLAGS: 00010216
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: ffff888105266400
RDX: 0000000000000006 RSI: ffff88800c890000 RDI: 0000000000000030
RBP: 0000000000000050 R08: 0000000000000000 R09: ffff88810526640e
R10: ffffed1020a4cc81 R11: ffff88810526640f R12: 0000000000000000
R13: 0000000000000820 R14: ffff888105266400 R15: 0000000000000050
FS:  00007f0653a07640(0000) GS:ffff88811af80000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f863ba096f4 CR3: 00000000163c0005 CR4: 0000000000770ef0
PKRU: 80000000
Call Trace:
 <IRQ>
 ipv6_renew_options+0x279/0x950 net/ipv6/exthdrs.c:1288
 calipso_req_setattr+0x181/0x340 net/ipv6/calipso.c:1204
 calipso_req_setattr+0x56/0x80 net/netlabel/netlabel_calipso.c:597
 netlbl_req_setattr+0x18a/0x440 net/netlabel/netlabel_kapi.c:1249
 selinux_netlbl_inet_conn_request+0x1fb/0x320 security/selinux/netlabel.c:342
 selinux_inet_conn_request+0x1eb/0x2c0 security/selinux/hooks.c:5551
 security_inet_conn_request+0x50/0xa0 security/security.c:4945
 tcp_v6_route_req+0x22c/0x550 net/ipv6/tcp_ipv6.c:825
 tcp_conn_request+0xec8/0x2b70 net/ipv4/tcp_input.c:7275
 tcp_v6_conn_request+0x1e3/0x440 net/ipv6/tcp_ipv6.c:1328
 tcp_rcv_state_process+0xafa/0x52b0 net/ipv4/tcp_input.c:6781
 tcp_v6_do_rcv+0x8a6/0x1a40 net/ipv6/tcp_ipv6.c:1667
 tcp_v6_rcv+0x505e/0x5b50 net/ipv6/tcp_ipv6.c:1904
 ip6_protocol_deliver_rcu+0x17c/0x1da0 net/ipv6/ip6_input.c:436
 ip6_input_finish+0x103/0x180 net/ipv6/ip6_input.c:480
 NF_HOOK include/linux/netfilter.h:314 [inline]
 NF_HOOK include/linux/netfilter.h:308 [inline]
 ip6_input+0x13c/0x6b0 net/ipv6/ip6_input.c:491
 dst_input include/net/dst.h:469 [inline]
 ip6_rcv_finish net/ipv6/ip6_input.c:79 [inline]
 ip6_rcv_finish+0xb6/0x490 net/ipv6/ip6_input.c:69
 NF_HOOK include/linux/netfilter.h:314 [inline]
 NF_HOOK include/linux/netfilter.h:308 [inline]
 ipv6_rcv+0xf9/0x490 net/ipv6/ip6_input.c:309
 __netif_receive_skb_one_core+0x12e/0x1f0 net/core/dev.c:5896
 __netif_receive_skb+0x1d/0x170 net/core/dev.c:6009
 process_backlog+0x41e/0x13b0 net/core/dev.c:6357
 __napi_poll+0xbd/0x710 net/core/dev.c:7191
 napi_poll net/core/dev.c:7260 [inline]
 net_rx_action+0x9de/0xde0 net/core/dev.c:7382
 handle_softirqs+0x19a/0x770 kernel/softirq.c:561
 do_softirq.part.0+0x36/0x70 kernel/softirq.c:462
 </IRQ>
 <TASK>
 do_softirq arch/x86/include/asm/preempt.h:26 [inline]
 __local_bh_enable_ip+0xf1/0x110 kernel/softirq.c:389
 local_bh_enable include/linux/bottom_half.h:33 [inline]
 rcu_read_unlock_bh include/linux/rcupdate.h:919 [inline]
 __dev_queue_xmit+0xc2a/0x3c40 net/core/dev.c:4679
 dev_queue_xmit include/linux/netdevice.h:3313 [inline]
 neigh_hh_output include/net/neighbour.h:523 [inline]
 neigh_output include/net/neighbour.h:537 [inline]
 ip6_finish_output2+0xd69/0x1f80 net/ipv6/ip6_output.c:141
 __ip6_finish_output net/ipv6/ip6_output.c:215 [inline]
 ip6_finish_output+0x5dc/0xd60 net/ipv6/ip6_output.c:226
 NF_HOOK_COND include/linux/netfilter.h:303 [inline]
 ip6_output+0x24b/0x8d0 net/ipv6/ip6_output.c:247
 dst_output include/net/dst.h:459 [inline]
 NF_HOOK include/linux/netfilter.h:314 [inline]
 NF_HOOK include/linux/netfilter.h:308 [inline]
 ip6_xmit+0xbbc/0x20d0 net/ipv6/ip6_output.c:366
 inet6_csk_xmit+0x39a/0x720 net/ipv6/inet6_connection_sock.c:135
 __tcp_transmit_skb+0x1a7b/0x3b40 net/ipv4/tcp_output.c:1471
 tcp_transmit_skb net/ipv4/tcp_output.c:1489 [inline]
 tcp_send_syn_data net/ipv4/tcp_output.c:4059 [inline]
 tcp_connect+0x1c0c/0x4510 net/ipv4/tcp_output.c:4148
 tcp_v6_connect+0x156c/0x2080 net/ipv6/tcp_ipv6.c:333
 __inet_stream_connect+0x3a7/0xed0 net/ipv4/af_inet.c:677
 tcp_sendmsg_fastopen+0x3e2/0x710 net/ipv4/tcp.c:1039
 tcp_sendmsg_locked+0x1e82/0x3570 net/ipv4/tcp.c:1091
 tcp_sendmsg+0x2f/0x50 net/ipv4/tcp.c:1358
 inet6_sendmsg+0xb9/0x150 net/ipv6/af_inet6.c:659
 sock_sendmsg_nosec net/socket.c:718 [inline]
 __sock_sendmsg+0xf4/0x2a0 net/socket.c:733
 __sys_sendto+0x29a/0x390 net/socket.c:2187
 __do_sys_sendto net/socket.c:2194 [inline]
 __se_sys_sendto net/socket.c:2190 [inline]
 __x64_sys_sendto+0xe1/0x1c0 net/socket.c:2190
 do_syscall_x64 arch/x86/entry/common.c:52 [inline]
 do_syscall_64+0xc3/0x1d0 arch/x86/entry/common.c:83
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
RIP: 0033:0x7f06553c47ed
Code: 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 a8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007f0653a06fc8 EFLAGS: 00000246 ORIG_RAX: 000000000000002c
RAX: ffffffffffffffda RBX: 00007f0655605fa0 RCX: 00007f06553c47ed
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 000000000000000b
RBP: 00007f065545db38 R08: 0000200000000140 R09: 000000000000001c
R10: f7384d4ea84b01bd R11: 0000000000000246 R12: 0000000000000000
R13: 00007f0655605fac R14: 00007f0655606038 R15: 00007f06539e7000
 </TASK>
Modules linked in:

[1]:
dnf install -y selinux-policy-targeted policycoreutils netlabel_tools procps-ng nmap-ncat
mount -t selinuxfs none /sys/fs/selinux
load_policy
netlabelctl calipso add pass doi:1
netlabelctl map del default
netlabelctl map add default address:::1 protocol:calipso,1
sysctl net.ipv4.tcp_syncookies=2
nc -l ::1 80 &
nc ::1 80

Fixes: e1adea927080 ("calipso: Allow request sockets to be relabelled by the lsm.")
Reported-by: syzkaller <syzkaller@googlegroups.com>
Reported-by: John Cheung <john.cs.hey@gmail.com>
Closes: https://lore.kernel.org/netdev/CAP=Rh=MvfhrGADy+-WJiftV2_WzMH4VEhEFmeT28qY+4yxNu4w@mail.gmail.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Link: https://patch.msgid.link/20250617224125.17299-1-kuni1840@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

MAINTAINERS: Remove Shannon Nelson from MAINTAINERS file

Brett Creeley is taking ownership of AMD/Pensando drivers while I wander
off into the sunset with my retirement this month.  I'll still keep an
eye out on a few topics for awhile, and maybe do some free-lance work in
the future.

Meanwhile, thank you all for the fun and support and the many learning
opportunities :-).

Special thanks go to DaveM for merging my first patch long ago, the big
ionic patchset a few years ago, and my last patchset last week.

Redirect things to a non-corporate account.

Signed-off-by: Shannon Nelson <shannon.nelson@amd.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Brett Creeley <brett.creeley@amd.com>
Link: https://patch.msgid.link/20250616224437.56581-1-shannon.nelson@amd.com
[Jakub: squash in the .mailmap update]
Signed-off-by: Shannon Nelson <sln@onemain.com>
Link: https://patch.msgid.link/20250619010603.1173141-1-sln@onemain.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag 'linux-can-next-for-6.17-20250618' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next

Marc Kleine-Budde says:

====================
pull-request: can-next 2025-06-18

All 10 patches are by Geert Uytterhoeven, target the rcar_canfd
driver, first cleanup/refactor the driver and then add support for
Transceiver Delay Compensation.

* tag 'linux-can-next-for-6.17-20250618' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next:
  can: rcar_canfd: Add support for Transceiver Delay Compensation
  can: rcar_canfd: Return early in rcar_canfd_set_bittiming() when not FD
  can: rcar_canfd: Share config code in rcar_canfd_set_bittiming()
  can: rcar_canfd: Rename rcar_canfd_setrnc() to rcar_canfd_set_rnc()
  can: rcar_canfd: Repurpose f_dcfg base for other registers
  can: rcar_canfd: Simplify data access in rcar_canfd_{ge,pu}t_data()
  can: rcar_canfd: Add helper variable dev to rcar_canfd_reset_controller()
  can: rcar_canfd: Add helper variable ndev to rcar_canfd_rx_pkt()
  can: rcar_canfd: Remove bittiming debug prints
  can: rcar_canfd: Consistently use ndev for net_device pointers
====================

Link: https://patch.msgid.link/20250618092336.2175168-1-mkl@pengutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: lan743x: fix potential out-of-bounds write in lan743x_ptp_io_event_clock_get()

Before calling lan743x_ptp_io_event_clock_get(), the 'channel' value
is checked against the maximum value of PCI11X1X_PTP_IO_MAX_CHANNELS(8).
This seems correct and aligns with the PTP interrupt status register
(PTP_INT_STS) specifications.

However, lan743x_ptp_io_event_clock_get() writes to ptp->extts[] with
only LAN743X_PTP_N_EXTTS(4) elements, using channel as an index:

    lan743x_ptp_io_event_clock_get(..., u8 channel,...)
    {
        ...
        /* Update Local timestamp */
        extts = &ptp->extts[channel];
        extts->ts.tv_sec = sec;
        ...
    }

To avoid an out-of-bounds write and utilize all the supported GPIO
inputs, set LAN743X_PTP_N_EXTTS to 8.

Detected using the static analysis tool - Svace.
Fixes: 60942c397af6 ("net: lan743x: Add support for PTP-IO Event Input External Timestamp (extts)")
Signed-off-by: Alexey Kodanev <aleksei.kodanev@bell-sw.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Rengarajan S <rengarajan.s@microchip.com>
Link: https://patch.msgid.link/20250616113743.36284-1-aleksei.kodanev@bell-sw.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Merge branch 'selftests-net-use-slowwait-to-make-sure-setup-finished'

Hangbin Liu says:

====================
selftests: net: use slowwait to make sure setup finished

The two updated tests sometimes failed because the network setup hadn't
completed. Used slowwait to ensure the setup finished and the tests
always passed. I ran both tests 50 times, and all of them passed.
====================

Link: https://patch.msgid.link/20250617105101.433718-1-liuhangbin@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

selftests: net: use slowwait to make sure IPv6 setup finished

Sometimes the vxlan vnifiltering test failed on slow machines due to
network setup not finished. e.g.

  TEST: VM connectivity over vnifiltering vxlan (ipv4 default rdst)   [ OK ]
  TEST: VM connectivity over vnifiltering vxlan (ipv6 default rdst)   [FAIL]

Let's use slowwait to make sure the connection is finished.

Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250617105101.433718-3-liuhangbin@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

selftests: net: use slowwait to stabilize vrf_route_leaking test

The vrf_route_leaking test occasionally fails due to connectivity issues
in our testing environment. A sample failure message shows that the ping
check fails intermittently

  PING 2001:db8:16:2::2 (2001:db8:16:2::2) 56 data bytes

  --- 2001:db8:16:2::2 ping statistics ---
  1 packets transmitted, 0 received, 100% packet loss, time 0ms

  TEST: Basic IPv6 connectivity                                       [FAIL]

This is likely due to insufficient wait time on slower machines. To address
this, switch to using slowwait, which provides a longer and more reliable
wait for setup completion.

Before this change, the test failed 3 out of 10 times. After applying this
fix, the test was run 30 times without any failure.

Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250617105101.433718-2-liuhangbin@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Merge branch 'support-bandwidth-clamping-in-mana-using-net-shapers'

Erni Sri Satya Vennela says:

====================
Support bandwidth clamping in mana using net shapers

This patchset introduces hardware-backed bandwidth rate limiting
for MANA NICs via the net_shaper_ops interface, enabling efficient and
fine-grained traffic shaping directly on the device.

Previously, MANA lacked a mechanism for user-configurable bandwidth
control. With this addition, users can now configure shaping parameters,
allowing better traffic management and performance isolation.

The implementation includes the net_shaper_ops callbacks in the MANA
driver and supports one shaper per vport. Add shaping support via
mana_set_bw_clamp(), allowing the configuration of bandwidth rates
in 100 Mbps increments (minimum 100 Mbps). The driver validates input
and rejects unsupported values. On failure, it restores the previous
configuration which is queried using mana_query_link_cfg() or
retains the current state.

To prevent potential deadlocks introduced by net_shaper_ops, switch to
_locked variants of NAPI APIs when netdevops_lock is held during
VF setup and teardown.

Also, Add support for ethtool get_link_ksettings to report the maximum
link speed supported by the SKU in mbps.

These APIs when invoked on hardware that are older or that do
not support these APIs, the speed would be reported as UNKNOWN and
the net-shaper calls to set speed would fail.
====================

Link: https://patch.msgid.link/1750144656-2021-1-git-send-email-ernis@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: mana: Handle unsupported HWC commands

If any of the HWC commands are not recognized by the
underlying hardware, the hardware returns the response
header status of -1. Log the information using
netdev_info_once to avoid multiple error logs in dmesg.

Signed-off-by: Erni Sri Satya Vennela <ernis@linux.microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Reviewed-by: Saurabh Singh Sengar <ssengar@linux.microsoft.com>
Reviewed-by: Dipayaan Roy <dipayanroy@linux.microsoft.com>
Link: https://patch.msgid.link/1750144656-2021-5-git-send-email-ernis@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: mana: Add speed support in mana_get_link_ksettings

Allow mana ethtool get_link_ksettings operation to report
the maximum speed supported by the SKU in mbps.

The driver retrieves this information by issuing a
HWC command to the hardware via mana_query_link_cfg(),
which retrieves the SKU's maximum supported speed.

These APIs when invoked on hardware that are older/do
not support these APIs, the speed would be reported as UNKNOWN.

Before:
$ethtool enP30832s1
> Settings for enP30832s1:
        Supported ports: [  ]
        Supported link modes:   Not reported
        Supported pause frame use: No
        Supports auto-negotiation: No
        Supported FEC modes: Not reported
        Advertised link modes:  Not reported
        Advertised pause frame use: No
        Advertised auto-negotiation: No
        Advertised FEC modes: Not reported
        Speed: Unknown!
        Duplex: Full
        Auto-negotiation: off
        Port: Other
        PHYAD: 0
        Transceiver: internal
        Link detected: yes

After:
$ethtool enP30832s1
> Settings for enP30832s1:
        Supported ports: [  ]
        Supported link modes:   Not reported
        Supported pause frame use: No
        Supports auto-negotiation: No
        Supported FEC modes: Not reported
        Advertised link modes:  Not reported
        Advertised pause frame use: No
        Advertised auto-negotiation: No
        Advertised FEC modes: Not reported
        Speed: 16000Mb/s
        Duplex: Full
        Auto-negotiation: off
        Port: Other
        PHYAD: 0
        Transceiver: internal
        Link detected: yes

Signed-off-by: Erni Sri Satya Vennela <ernis@linux.microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Reviewed-by: Saurabh Singh Sengar <ssengar@linux.microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
Link: https://patch.msgid.link/1750144656-2021-4-git-send-email-ernis@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: mana: Add support for net_shaper_ops

Introduce support for net_shaper_ops in the MANA driver,
enabling configuration of rate limiting on the MANA NIC.

To apply rate limiting, the driver issues a HWC command via
mana_set_bw_clamp() and updates the corresponding shaper object
in the net_shaper cache. If an error occurs during this process,
the driver restores the previous speed by querying the current link
configuration using mana_query_link_cfg().

The minimum supported bandwidth is 100 Mbps, and only values that are
exact multiples of 100 Mbps are allowed. Any other values are rejected.

To remove a shaper, the driver resets the bandwidth to the maximum
supported by the SKU using mana_set_bw_clamp() and clears the
associated cache entry. If an error occurs during this process,
the shaper details are retained.

On the hardware that does not support these APIs, the net-shaper
calls to set speed would fail.

Set the speed:
./tools/net/ynl/pyynl/cli.py \
 --spec Documentation/netlink/specs/net_shaper.yaml \
 --do set --json '{"ifindex":'$IFINDEX',
   "handle":{"scope": "netdev", "id":'$ID' },
   "bw-max": 200000000 }'

Get the shaper details:
./tools/net/ynl/pyynl/cli.py \
 --spec Documentation/netlink/specs/net_shaper.yaml \
 --do get --json '{"ifindex":'$IFINDEX',
      "handle":{"scope": "netdev", "id":'$ID' }}'

> {'bw-max': 200000000,
> 'handle': {'scope': 'netdev'},
> 'ifindex': $IFINDEX,
> 'metric': 'bps'}

Delete the shaper object:
./tools/net/ynl/pyynl/cli.py \
 --spec Documentation/netlink/specs/net_shaper.yaml \
 --do delete --json '{"ifindex":'$IFINDEX',
      "handle":{"scope": "netdev","id":'$ID' }}'

Signed-off-by: Erni Sri Satya Vennela <ernis@linux.microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Reviewed-by: Saurabh Singh Sengar <ssengar@linux.microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
Link: https://patch.msgid.link/1750144656-2021-3-git-send-email-ernis@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net: mana: Fix potential deadlocks in mana napi ops

When net_shaper_ops are enabled for MANA, netdev_ops_lock
becomes active.

MANA VF setup/teardown by netvsc follows this call chain:

netvsc_vf_setup()
        dev_change_flags()
...
         __dev_open() OR __dev_close()

dev_change_flags() holds the netdev mutex via netdev_lock_ops.

Meanwhile, mana_create_txq() and mana_create_rxq() in mana_open()
path call NAPI APIs (netif_napi_add_tx(), netif_napi_add_weight(),
napi_enable()), which also try to acquire the same lock, risking
deadlock.

Similarly in the teardown path (mana_close()), netif_napi_disable()
and netif_napi_del(), contend for the same lock.

Switch to the _locked variants of these APIs to avoid deadlocks
when the netdev_ops_lock is held.

Fixes: d4c22ec680c8 ("net: hold netdev instance lock during ndo_open/ndo_stop")
Signed-off-by: Erni Sri Satya Vennela <ernis@linux.microsoft.com>
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Shradha Gupta <shradhagupta@linux.microsoft.com>
Reviewed-by: Saurabh Singh Sengar <ssengar@linux.microsoft.com>
Reviewed-by: Long Li <longli@microsoft.com>
Link: https://patch.msgid.link/1750144656-2021-2-git-send-email-ernis@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

eth: fbnic: avoid double free when failing to DMA-map FW msg

The semantics are that caller of fbnic_mbx_map_msg() retains
the ownership of the message on error. All existing callers
dutifully free the page.

Fixes: da3cde08209e ("eth: fbnic: Add FW communication mechanism")
Reviewed-by: Alexander Duyck <alexanderduyck@fb.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/20250616195510.225819-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

ipv6: Simplify link-local address generation for IPv6 GRE.

Since commit 3e6a0243ff00 ("gre: Fix again IPv6 link-local address
generation."), addrconf_gre_config() has stopped handling IP6GRE
devices specially and just calls the regular addrconf_addr_gen()
function to create their link-local IPv6 addresses.

We can thus avoid using addrconf_gre_config() for IP6GRE devices and
use the normal IPv6 initialisation path instead (that is, jump directly
to addrconf_dev_config() in addrconf_init_auto_addrs()).

See commit 3e6a0243ff00 ("gre: Fix again IPv6 link-local address
generation.") for a deeper explanation on how and why GRE devices
started handling their IPv6 link-local address generation specially,
why it was a problem, and why this is not even necessary in most cases
(especially for GRE over IPv6).

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/a9144be9c7ec3cf09f25becae5e8fdf141fde9f6.1750075076.git.gnault@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

net/mlx4_en: Remove the redundant NULL check for the 'my_ets' object

Static analysis shows that pointer "my_ets" cannot be NULL because it
points to the object "struct ieee_ets".

Remove the extra NULL check. It is meaningless and harms the readability
of the code.

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

Signed-off-by: Andrey Vatoropin <a.vatoropin@crpt.ru>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/20250616045034.26000-1-a.vatoropin@crpt.ru
Signed-off-by: Paolo Abeni <pabeni@redhat.com>

Merge branch 'add-support-for-pse-budget-evaluation-strategy'

Kory Maincent says:

====================
Add support for PSE budget evaluation strategy

This series brings support for budget evaluation strategy in the PSE
subsystem. PSE controllers can set priorities to decide which ports should
be turned off in case of special events like over-current.

This patch series adds support for two budget evaluation strategy.
1. Static Method:

   This method involves distributing power based on PD classification.
   It’s straightforward and stable, the PSE core keeping track of the
   budget and subtracting the power requested by each PD’s class.

   Advantages: Every PD gets its promised power at any time, which
   guarantees reliability.

   Disadvantages: PD classification steps are large, meaning devices
   request much more power than they actually need. As a result, the power
   supply may only operate at, say, 50% capacity, which is inefficient and
   wastes money.

2. Dynamic Method:

   To address the inefficiencies of the static method, vendors like
   Microchip have introduced dynamic power budgeting, as seen in the
   PD692x0 firmware. This method monitors the current consumption per port
   and subtracts it from the available power budget. When the budget is
   exceeded, lower-priority ports are shut down.

   Advantages: This method optimizes resource utilization, saving costs.

   Disadvantages: Low-priority devices may experience instability.

The UAPI allows adding support for software port priority mode managed from
userspace later if needed.

Patches 1-2: Add support for interrupt event report in PSE core, ethtool
     and ethtool specs.
Patch 3: Adds support for interrupt and event report in TPS23881 driver.
Patches 4,5: Add support for PSE power domain in PSE core and ethtool.
Patches 6-8: Add support for budget evaluation strategy in PSE core,
     ethtool and ethtool specs.
Patches 9-11: Add support for port priority and power supplies in PD692x0
      drivers.
Patches 12,13: Add support for port priority in TPS23881 drivers.
====================

Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-0-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

dt-bindings: net: pse-pd: ti,tps23881: Add interrupt description

Add an interrupt property to the device tree bindings for the TI TPS23881
PSE controller. The interrupt is primarily used to detect classification
and disconnection events, which are essential for managing the PSE
controller in compliance with the PoE standard.

Interrupt support is essential for the proper functioning of the TPS23881
controller. Without it, after a power-on (PWON), the controller will
no longer perform detection and classification. This could lead to
potential hazards, such as connecting a non-PoE device after a PoE device,
which might result in magic smoke.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-13-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: tps23881: Add support for static port priority feature

This patch enhances PSE callbacks by introducing support for the static
port priority feature. It extends interrupt management to handle and report
detection, classification, and disconnection events. Additionally, it
introduces the pi_get_pw_req() callback, which provides information about
the power requested by the Powered Devices.

Interrupt support is essential for the proper functioning of the TPS23881
controller. Without it, after a power-on (PWON), the controller will
no longer perform detection and classification. This could lead to
potential hazards, such as connecting a non-PoE device after a PoE device,
which might result in magic smoke.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-12-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

dt-bindings: net: pse-pd: microchip,pd692x0: Add manager regulator supply

Adds the regulator supply parameter of the managers.
Update also the example as the regulator supply of the PSE PIs
should be the managers itself and not an external regulator.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-11-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: pd692x0: Add support for controller and manager power supplies

Add support for managing the VDD and VDDA power supplies for the PD692x0
PSE controller, as well as the VAUX5 and VAUX3P3 power supplies for the
PD6920x PSE managers.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-10-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: pd692x0: Add support for PSE PI priority feature

This patch extends the PSE callbacks by adding support for the newly
introduced pi_set_prio() callback, enabling the configuration of PSE PI
priorities. The current port priority is now also included in the status
information returned to users.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-9-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ethtool: Add PSE port priority support feature

This patch expands the status information provided by ethtool for PSE c33
with current port priority and max port priority. It also adds a call to
pse_ethtool_set_prio() to configure the PSE port priority.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-8-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: Add support for budget evaluation strategies

This patch introduces the ability to configure the PSE PI budget evaluation
strategies. Budget evaluation strategies is utilized by PSE controllers to
determine which ports to turn off first in scenarios such as power budget
exceedance.

The pis_prio_max value is used to define the maximum priority level
supported by the controller. Both the current priority and the maximum
priority are exposed to the user through the pse_ethtool_get_status call.

This patch add support for two mode of budget evaluation strategies.
1. Static Method:

   This method involves distributing power based on PD classification.
   It’s straightforward and stable, the PSE core keeping track of the
   budget and subtracting the power requested by each PD’s class.

   Advantages: Every PD gets its promised power at any time, which
   guarantees reliability.

   Disadvantages: PD classification steps are large, meaning devices
   request much more power than they actually need. As a result, the power
   supply may only operate at, say, 50% capacity, which is inefficient and
   wastes money.

   Priority max value is matching the number of PSE PIs within the PSE.

2. Dynamic Method:

   To address the inefficiencies of the static method, vendors like
   Microchip have introduced dynamic power budgeting, as seen in the
   PD692x0 firmware. This method monitors the current consumption per port
   and subtracts it from the available power budget. When the budget is
   exceeded, lower-priority ports are shut down.

   Advantages: This method optimizes resource utilization, saving costs.

   Disadvantages: Low-priority devices may experience instability.

   Priority max value is set by the PSE controller driver.

For now, budget evaluation methods are not configurable and cannot be
mixed. They are hardcoded in the PSE driver itself, as no current PSE
controller supports both methods.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Acked-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-7-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: Add helper to report hardware enable status of the PI

Refactor code by introducing a helper function to retrieve the hardware
enabled state of the PI, avoiding redundant implementations in the
future.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-6-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ethtool: Add support for new power domains index description

Report the index of the newly introduced PSE power domain to the user,
enabling improved management of the power budget for PSE devices.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-5-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: Add support for PSE power domains

Introduce PSE power domain support as groundwork for upcoming port
priority features. Multiple PSE PIs can now be grouped under a single
PSE power domain, enabling future enhancements like defining available
power budgets, port priority modes, and disconnection policies. This
setup will allow the system to assess whether activating a port would
exceed the available power budget, preventing over-budget states
proactively.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-4-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: tps23881: Add support for PSE events and interrupts

Add support for PSE event reporting through interrupts. Set up the newly
introduced devm_pse_irq_helper helper to register the interrupt. Events are
reported for over-current and over-temperature conditions.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-3-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: Add support for reporting events

Add support for devm_pse_irq_helper() to register PSE interrupts and report
events such as over-current or over-temperature conditions. This follows a
similar approach to the regulator API but also sends notifications using a
dedicated PSE ethtool netlink socket.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-2-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: pse-pd: Introduce attached_phydev to pse control

In preparation for reporting PSE events via ethtool notifications,
introduce an attached_phydev field in the pse_control structure.
This field stores the phy_device associated with the PSE PI,
ensuring that notifications are sent to the correct network
interface.

The attached_phydev pointer is directly tied to the PHY lifecycle. It
is set when the PHY is registered and cleared when the PHY is removed.
There is no need to use a refcount, as doing so could interfere with
the PHY removal process.

Signed-off-by: Kory Maincent (Dent Project) <kory.maincent@bootlin.com>
Reviewed-by: Oleksij Rempel <o.rempel@pengutronix.de>
Link: https://patch.msgid.link/20250617-feature_poe_port_prio-v14-1-78a1a645e2ee@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'phc-support-in-ena-driver'

David Arinzon says:

====================
PHC support in ENA driver

This patchset adds the support for PHC (PTP Hardware Clock)
in the ENA driver. The documentation part of the patchset
includes additional information, including statistics,
utilization and invocation examples through the testptp
utility.
====================

Link: https://patch.msgid.link/20250617110545.5659-1-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Add PHC documentation

Provide the relevant information and guidelines
about the feature support in the ENA driver.

Signed-off-by: Amit Bernstein <amitbern@amazon.com>
Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-10-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: View PHC stats using debugfs

Add an entry named `phc_stats` to view the PHC statistics.
If PHC is enabled, the stats are printed, as below:

phc_cnt: 0
phc_exp: 0
phc_skp: 0
phc_err_dv: 0
phc_err_ts: 0

If PHC is disabled, no statistics will be displayed.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-9-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Add debugfs support to the ENA driver

Adding the base directory of debugfs to the driver.
In order for the folder to be unique per driver instantiation,
the chosen name is the device name.

This commit contains the initialization and the
base folder.

The creation of the base folder may fail, but is considered
non-fatal.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-8-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Control PHC enable through devlink

Add the capability to set parameters through the devlink framework.

The parameter used for controlling PHC (enable/disable) details
are as follows:
- Name: enable_phc
- Type: Boolean (true - enable/false - disable)
- Mode: DEVLINK_PARAM_CMODE_DRIVERINIT
- Effect: Changes take place during driver initialization,
          any changes require a devlink reload to take effect.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-7-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

devlink: Add new "enable_phc" generic device param

Add a new device generic parameter to enable/disable the
PHC (PTP Hardware Clock) functionality in the device associated
with the devlink instance.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://patch.msgid.link/20250617110545.5659-6-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Add devlink port support

Add the basic functionality to support devlink port
for devlink model completeness purposes.
Current support is for registration/un-registration.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-5-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Add device reload capability through devlink

Adding basic devlink capability support of reloading the driver.
This capability is required to support driver init type
devlink params (DEVLINK_PARAM_CMODE_DRIVERINIT). Such params
require reloading of the driver (destroy/restore sequence).
The reloading is done by the devlink framework using the
hooks provided by the driver.

Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-4-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: PHC silent reset

Each PHC device kernel registration receives a unique kernel index,
which is associated with a new PHC device file located at
"/dev/ptp<index>".
This device file serves as an interface for obtaining PHC timestamps.
Examples of tools that use "/dev/ptp" include testptp [1]
and chrony [2].

A reset flow may occur in the ENA driver while PHC is active.
During a reset, the driver will unregister and then re-register the
PHC device with the kernel.
Under race conditions, particularly during heavy PHC loads,
the driver’s reset flow might complete faster than the kernel’s PHC
unregister/register process.
This can result in the PHC index being different from what it was prior
to the reset, as the PHC index is selected using kernel ID
allocation [3].

While driver rmmod/insmod are done by the user, a reset may occur
at anytime, without the user awareness, consequently, the driver
might receive a new PHC index after the reset, potentially compromising
the user experience.

To prevent this issue, the PHC flow will detect the reset during PHC
destruction and will skip the PHC unregister/register calls to preserve
the kernel PHC index.
During the reset flow, any attempt to get a PHC timestamp will fail as
expected, but the kernel PHC index will remain unchanged.

[1]: https://github.com/torvalds/linux/blob/v6.1/tools/testing/selftests/ptp/testptp.c
[2]: https://github.com/mlichvar/chrony
[3]: https://www.kernel.org/doc/html/latest/core-api/idr.html

Signed-off-by: Amit Bernstein <amitbern@amazon.com>
Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-3-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: ena: Add PHC support in the ENA driver

The ENA driver will be extended to support the new PHC feature using
ptp_clock interface [1]. this will provide timestamp reference for user
space to allow measuring time offset between the PHC and the system
clock in order to achieve nanosecond accuracy.

[1] - https://www.kernel.org/doc/html/latest/driver-api/ptp.html

Signed-off-by: Amit Bernstein <amitbern@amazon.com>
Signed-off-by: David Arinzon <darinzon@amazon.com>
Link: https://patch.msgid.link/20250617110545.5659-2-darinzon@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'udp_tunnel-remove-rtnl_lock-dependency'

Stanislav Fomichev says:

====================
udp_tunnel: remove rtnl_lock dependency

Recently bnxt had to grow back a bunch of rtnl dependencies because
of udp_tunnel's infra. Add separate (global) mutext to protect
udp_tunnel state.
====================

Link: https://patch.msgid.link/20250616162117.287806-1-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Revert "bnxt_en: bring back rtnl_lock() in the bnxt_open() path"

This reverts commit 325eb217e41fa14f307c7cc702bd18d0bb38fe84.

udp_tunnel infra doesn't need RTNL, should be safe to get back
to only netdev instance lock.

Cc: Michael Chan <michael.chan@broadcom.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Link: https://patch.msgid.link/20250616162117.287806-7-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

netdevsim: remove udp_ports_sleep

Now that there is only one path in udp_tunnel, there is no need to
have udp_ports_sleep knob. Remove it and adjust the test.

Cc: Michael Chan <michael.chan@broadcom.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Link: https://patch.msgid.link/20250616162117.287806-6-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: remove redundant ASSERT_RTNL() in queue setup functions

The existing netdev_ops_assert_locked() already asserts that either
the RTNL lock or the per-device lock is held, making the explicit
ASSERT_RTNL() redundant.

Cc: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Link: https://patch.msgid.link/20250616162117.287806-5-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

udp_tunnel: remove rtnl_lock dependency

Drivers that are using ops lock and don't depend on RTNL lock
still need to manage it because udp_tunnel's RTNL dependency.
Introduce new udp_tunnel_nic_lock and use it instead of
rtnl_lock. Drop non-UDP_TUNNEL_NIC_INFO_MAY_SLEEP mode from
udp_tunnel infra (udp_tunnel_nic_device_sync_work needs to
grab udp_tunnel_nic_lock mutex and might sleep).

Cover more places in v4:

- netlink
  - udp_tunnel_notify_add_rx_port (ndo_open)
    - triggers udp_tunnel_nic_device_sync_work
  - udp_tunnel_notify_del_rx_port (ndo_stop)
    - triggers udp_tunnel_nic_device_sync_work
  - udp_tunnel_get_rx_info (__netdev_update_features)
    - triggers NETDEV_UDP_TUNNEL_PUSH_INFO
  - udp_tunnel_drop_rx_info (__netdev_update_features)
    - triggers NETDEV_UDP_TUNNEL_DROP_INFO
  - udp_tunnel_nic_reset_ntf (ndo_open)

- notifiers
  - udp_tunnel_nic_netdevice_event, depending on the event:
    - triggers NETDEV_UDP_TUNNEL_PUSH_INFO
    - triggers NETDEV_UDP_TUNNEL_DROP_INFO

- ethnl_tunnel_info_reply_size
- udp_tunnel_nic_set_port_priv (two intel drivers)

Cc: Michael Chan <michael.chan@broadcom.com>
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Link: https://patch.msgid.link/20250616162117.287806-4-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

vxlan: drop sock_lock

We won't be able to sleep soon in vxlan_offload_rx_ports and won't be
able to grab sock_lock. Instead of having separate spinlock to
manage sockets, rely on rtnl lock. This is similar to how geneve
manages its sockets.

Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/20250616162117.287806-3-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

geneve: rely on rtnl lock in geneve_offload_rx_ports

udp_tunnel_push_rx_port will grab mutex in the next patch so
we can't use rcu. geneve_offload_rx_ports is called
from geneve_netdevice_event for NETDEV_UDP_TUNNEL_PUSH_INFO and
NETDEV_UDP_TUNNEL_DROP_INFO which both have ASSERT_RTNL.
Entries are added to and removed from the sock_list under rtnl
lock as well (when adding or removing a tunneling device).

Signed-off-by: Stanislav Fomichev <stfomichev@gmail.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/20250616162117.287806-2-stfomichev@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'net-fix-passive-tfo-socket-having-invalid-napi-id'

David Wei says:

====================
net: fix passive TFO socket having invalid NAPI ID

Found a bug where an accepted passive TFO socket returns an invalid NAPI
ID (i.e. 0) from SO_INCOMING_NAPI_ID. Add a selftest for this using
netdevsim and fix the bug.

Patch 1 is a drive-by fix for the lib.sh include in an existing
drivers/net/netdevsim/peer.sh selftest.

Patch 2 adds a test binary for a simple TFO server/client.

Patch 3 adds a selftest for checking that the NAPI ID of a passive TFO
socket is valid. This will currently fail.

Patch 4 adds a fix for the bug.
====================

Link: https://patch.msgid.link/20250617212102.175711-1-dw@davidwei.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

tcp: fix passive TFO socket having invalid NAPI ID

There is a bug with passive TFO sockets returning an invalid NAPI ID 0
from SO_INCOMING_NAPI_ID. Normally this is not an issue, but zero copy
receive relies on a correct NAPI ID to process sockets on the right
queue.

Fix by adding a sk_mark_napi_id_set().

Fixes: e5907459ce7e ("tcp: Record Rx hash and NAPI ID in tcp_child_process")
Signed-off-by: David Wei <dw@davidwei.uk>
Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250617212102.175711-5-dw@davidwei.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests: net: add test for passive TFO socket NAPI ID

Add a test that checks that the NAPI ID of a passive TFO socket is valid
i.e. not zero.

Signed-off-by: David Wei <dw@davidwei.uk>
Link: https://patch.msgid.link/20250617212102.175711-4-dw@davidwei.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests: net: add passive TFO test binary

Add a simple passive TFO server and client test binary. This will be
used to test the SO_INCOMING_NAPI_ID of passive TFO accepted sockets.

Signed-off-by: David Wei <dw@davidwei.uk>
Link: https://patch.msgid.link/20250617212102.175711-3-dw@davidwei.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

selftests: netdevsim: improve lib.sh include in peer.sh

Fix the peer.sh test to run from INSTALL_PATH.

Signed-off-by: David Wei <dw@davidwei.uk>
Link: https://patch.msgid.link/20250617212102.175711-2-dw@davidwei.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

tcp: Remove inet_hashinfo2_free_mod()

DCCP was removed, inet_hashinfo2_free_mod() is unused now.

Signed-off-by: Yue Haibing <yuehaibing@huawei.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20250617130613.498659-1-yuehaibing@huawei.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

dpaa_eth: don't use fixed_phy_change_carrier

This effectively reverts 6e8b0ff1ba4c ("dpaa_eth: Add change_carrier()
for Fixed PHYs"). Usage of fixed_phy_change_carrier() requires that
fixed_phy_register() has been called before, directly or indirectly.
And that's not the case in this driver.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Link: https://patch.msgid.link/7eb189b3-d5fd-4be6-8517-a66671a4e4e3@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge tag '6.16-rc2-ksmbd-server-fixes' of git://git.samba.org/ksmbd

Pull smb server fixes from Steve French:

 - Fix alternate data streams bug

 - Important fix for null pointer deref with Kerberos authentication

 - Fix oops in smbdirect (RDMA) in free_transport

* tag '6.16-rc2-ksmbd-server-fixes' of git://git.samba.org/ksmbd:
  ksmbd: handle set/get info file for streamed file
  ksmbd: fix null pointer dereference in destroy_previous_session
  ksmbd: add free_transport ops in ksmbd connection

Merge tag 'driver-core-6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core

Pull driver core fixes from Danilo Krummrich:

 - Fix a race condition in Devres::drop(). This depends on two other
   patches:
     - (Minimal) Rust abstractions for struct completion
     - Let Revocable indicate whether its data is already being revoked

 - Fix Devres to avoid exposing the internal Revocable

 - Add .mailmap entry for Danilo Krummrich

 - Add Madhavan Srinivasan to embargoed-hardware-issues.rst

* tag 'driver-core-6.16-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core:
  Documentation: embargoed-hardware-issues.rst: Add myself for Power
  mailmap: add entry for Danilo Krummrich
  rust: devres: do not dereference to the internal Revocable
  rust: devres: fix race in Devres::drop()
  rust: revocable: indicate whether `data` has been revoked already
  rust: completion: implement initial abstraction

Merge tag 'cgroup-for-6.16-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup

Pull cgroup fix from Tejun Heo:

 - In cgroup1 freezer, a task migrating into a frozen cgroup might not
   get frozen immediately due to the wrong operation order. Fix it.

* tag 'cgroup-for-6.16-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
  cgroup,freezer: fix incomplete freezing when attaching tasks

Merge tag 'wq-for-6.16-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq

Pull workqueue fix from Tejun Heo:

 - Fix missed early init of wq_isolated_cpumask

* tag 'wq-for-6.16-rc2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: Initialize wq_isolated_cpumask in workqueue_init_early()

tipc: fix null-ptr-deref when acquiring remote ip of ethernet bearer

The reproduction steps:
1. create a tun interface
2. enable l2 bearer
3. TIPC_NL_UDP_GET_REMOTEIP with media name set to tun

tipc: Started in network mode
tipc: Node identity 8af312d38a21, cluster identity 4711
tipc: Enabled bearer <eth:syz_tun>, priority 1
Oops: general protection fault
KASAN: null-ptr-deref in range
CPU: 1 UID: 1000 PID: 559 Comm: poc Not tainted 6.16.0-rc1+ #117 PREEMPT
Hardware name: QEMU Ubuntu 24.04 PC
RIP: 0010:tipc_udp_nl_dump_remoteip+0x4a4/0x8f0

the ub was in fact a struct dev.

when bid != 0 && skip_cnt != 0, bearer_list[bid] may be NULL or
other media when other thread changes it.

fix this by checking media_id.

Fixes: 832629ca5c313 ("tipc: add UDP remoteip dump to netlink API")
Signed-off-by: Haixia Qu <hxqu@hillstonenet.com>
Reviewed-by: Tung Nguyen <tung.quang.nguyen@est.tech>
Link: https://patch.msgid.link/20250617055624.2680-1-hxqu@hillstonenet.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Octeontx2-pf: Fix Backpresure configuration

NIX block can receive packets from multiple links such as
MAC (RPM), LBK and CPT.

       -----------------
 RPM --|     NIX       |
       -----------------
             |
             |
            LBK

Each link supports multiple channels for example RPM link supports
16 channels. In case of link oversubsribe, NIX will assert backpressure
on receive channels.

The previous patch considered a single channel per link, resulting in
backpressure not being enabled on the remaining channels

Fixes: a7ef63dbd588 ("octeontx2-af: Disable backpressure between CPT and NIX")
Signed-off-by: Hariprasad Kelam <hkelam@marvell.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250617063403.3582210-1-hkelam@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/mlx4e: Don't redefine IB_MTU_XXX enum

Rely on existing IB_MTU_XXX definitions which exist in ib_verbs.h.

Reviewed-by: Patrisious Haddad <phaddad@nvidia.com>
Signed-off-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/382c91ee506e7f1f3c1801957df6b28963484b7d.1750147222.git.leon@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>