When processing AUTH + COOKIE-ECHO packets, if skb_clone() fails
due to memory pressure, chunk->auth_chunk is NULL. The original
code still sets chunk->auth = 1 and continues, leaving the
COOKIE-ECHO to be processed without a valid auth_chunk for
deferred verification.
Discard the AUTH chunk early via pdiscard when skb_clone() fails,
so that the receive loop can continue processing remaining chunks
in the inqueue instead of stalling the entire packet.
Signed-off-by: Qing Luo <luoqing@kylinos.cn>
Acked-by: Xin Long <lucien.xin@gmail.com>
Link: https://patch.msgid.link/20260723061107.384106-1-l1138897701@163.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jakub Kicinski says:
====================
net_shaper: clarify kernel API docs
A handful of improvements to struct net_shaper_ops kdoc.
Today driver authors have to dig thru the shaper.c code
to understand the behavior. I'm covering things I wish
were explained already during recent reviews (within Meta).
There may be more things that need to be explained, incrementally.
====================
Link: https://patch.msgid.link/20260724210756.1553565-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Clarify the re-parenting expectations. Specifically that @delete
on a queue removes it from the hierarchy which is a bit unusual
in the overall API structure.
IIRC the implicit delete behavior was introduced because otherwise
it would not be possible to remove a queue from the hierarchy
without changing at least one handle of the shapers. Normally
"removal" is done by "adding" to the new parent, but "outside
the hierarchy" does not have a parent we can point at.
Link: https://patch.msgid.link/20260724210756.1553565-4-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It is true that the user-facing group() operation can only
be invoked with queues as leaves (see net_shaper_parse_leaf()),
but the driver facing op is also called when we delete a node.
When we delete a node we conceptually call
group(parent, node.list_of_leaves)
to add node's leaves to the parent. Node deletion "mid-hierarchy"
is supported so some of the leaves may themselves be nodes.
Therefore the driver facing group() may be called with nodes.
Remove the incorrect comment, and add a comment about differences
between the Netlink API and driver facing API.
Link: https://patch.msgid.link/20260724210756.1553565-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reuse the flowi6 template t->fl.u.ip6 built by ip6_tnl_link_config() in
ip6_tnl_fill_forward_path(), aligning the fast-path route lookup with
the slow path in ipxip6_tnl_xmit(). This automatically inherits the
correct conditional FLOWLABEL masking based on the
IP6_TNL_F_USE_ORIG_FLOWLABEL flag.
Return -EOPNOTSUPP when IP6_TNL_F_USE_ORIG_TCLASS,
IP6_TNL_F_USE_ORIG_FLOWLABEL or IP6_TNL_F_USE_ORIG_FWMARK is set,
or for collect_md tunnels, since fill_forward_path has no access to
the original skb and cannot recover the per-packet traffic class,
flowlabel, mark or tunnel destination needed for the route lookup.
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Link: https://patch.msgid.link/20260724-ip6ip6-route-lookup-fill_forward_path-v3-1-7b7991538614@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Commit 1b42e07af1 ("gve: Add Rx HWTS metadata to AF_XDP ZC mode")
exposes support for the XDP RX timestamping metadata operation in the DQ
RDA mode. While the operation works on its own, the intent was to enable
XDP metadata support for the queue format as a whole along with it.
Currently bpf_xdp_adjust_meta fails because meta_valid is set to false.
This change updates xdp_buff preparation to set meta_valid to true, so
metadata can be fully used by XDP programs.
Reviewed-by: Harshitha Ramamurthy <hramamurthy@google.com>
Reviewed-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Joshua Washington <joshwash@google.com>
Link: https://patch.msgid.link/20260722221634.186886-3-joshwash@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When scanning the MKEX profile to determine supported NPC features, warn
if the SPI extraction field overlaps with other key fields. AH and ESP
may legitimately use the same key offset for SPI, so continue to
advertise NPC_IPSEC_SPI via npc_is_field_present() instead of treating
the overlap as a hard failure.
Signed-off-by: Hariprasad Kelam <hkelam@marvell.com>
Signed-off-by: Ratheesh Kannoth <rkannoth@marvell.com>
Link: https://patch.msgid.link/20260721070303.986740-1-rkannoth@marvell.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Customer reported a problem with mounting CIFS shares where the server
hostname was 2 chars long. Turned out that the CIFS client wasn't
able to resolve NetBIOS names shorter than 3 chars.
Fix this by allowing a minimum of one character per hostname in
dns_query().
Reproducer with samba server:
# 'ab' and 'srv' hotnames resolve to same ip address
$ ssh srv ln -s 'msdfs:\\ab\\share' /home/shares/dfs/link1
$ mount.cifs //srv/dfs/link1 /mnt -o ... [EINVAL]
Reported-by: Pierguido Lambri <plambri@redhat.com>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: Frank Sorenson <sorenson@redhat.com>
Link: https://patch.msgid.link/20260722-net-dns_resolver-v1-1-c3385898ccf9@manguebit.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
On every VF/SF vport enable, esw_acl_egress_ofld_setup() allocates an
egress ACL flow table and a fwd_grp whenever the device supports
egress_acl_forward_to_vport. The only consumer of that group is the
active/passive fwd2vport rule installed when two representor netdevs
are bonded - a path that almost never fires. As a result, hosts with
many VFs/SFs pay a per-vport flow table and flow group cost for a
feature most ports never use.
Defer the flow table and fwd_grp creation to the moment they are
actually needed, when mlx5e_rep_esw_bond_netevent() drives
mlx5_esw_acl_egress_vport_bond() for the passive vport:
- esw_acl_egress_ofld_setup() now returns early unless
prio_tag_required is set. When prio_tag_required is set the
flow table is still allocated eagerly for the VLAN pop rule, and
its size is grown by one when fwd2vport is supported so the lazy
fwd_grp can later be added without re-creating the table. Only
the VLAN group is built up-front.
- A new helper, esw_acl_egress_ofld_fwd2vport_setup(), allocates
the egress ACL flow table (size 1) and the fwd_grp on demand,
and rolls back the flow table if group creation fails and the
helper had just allocated it. Existing cleanup paths
(esw_acl_egress_ofld_cleanup() -> *_groups_destroy() /
*_table_destroy()) already tolerate NULL fields, so vport
disable continues to free everything that was actually
allocated.
- mlx5_esw_acl_egress_vport_bond() calls the helper for the
passive vport before installing the fwd2vport rule. The active
vport does not need the flow table on its own: with a NULL
fwd_dest, esw_acl_egress_ofld_rules_create() is a no-op unless
prio_tag_required is set, in which case the eager path already
built the table.
mlx5_esw_acl_egress_vport_bond() and mlx5_esw_acl_egress_vport_unbond()
now take esw->state_lock for the duration of the operation, because
they may mutate vport->egress.acl, which is also written by the vport
enable/disable path under the same lock.
Signed-off-by: Yael Chemla <ychemla@nvidia.com>
Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260723070427.1861502-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The _once rate-limiting in slow_pci_heuristic() is unnecessary because
this function only runs during probe.
Worse, it interacts poorly with dynamic debug: if the first probe
happens before dynamic debug is enabled for this callsite, the _once
flag is permanently consumed and the message becomes unreachable without
reloading the module.
Additionally, only the first probed device values were printable in case
of multiple devices.
Replace with mlx5_core_dbg() which allows enabling the print via dynamic
debug at any time and observing it on the next probe.
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Alex Lazar <alazar@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260723081743.1868357-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Paolo Abeni says:
====================
docs: netdev: a couple of updates
Explicitly covers a couple of relevant spots currently left out in the
dark. Likely the first one was too obvious to ever spell out loud and
the latter too new.
Both cases are increasingly relevant, at least to me.
Keep the update in the netdev subsystem, as others may have slightly
different preferences.
====================
Link: https://patch.msgid.link/cover.1784884817.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Pablo Neira Ayuso says:
====================
Netfilter/IPVS updates for net-next
The following patchset contains Netfilter/IPVS updates for net-next,
just a small batch with accumulated pending updates:
1) In IPVS, use system_dfl_long_wq instead of system_long_wq, from
Ismael Luceno.
2) Add missing .checkentry in xt_tcpmss for IPv6, this is a follow up
to a recent harderning, from Florian Westphal.
3) Address a sashiko report in the NAT SIP helper, from Florian Westphal.
4) Tear down flow entries with stale routes using the GC, this is to
detect route updates when hardware offload is enabled.
5) Pass master conntrack as parameter to functions instead of using
exp->master as preparation work to turn exp->master into a cookie.
6) Move expectation event_mask to the nf_conntrack_expect object,
again as preparation work to turn exp->master into a cookie.
7) In IPVS, use kzalloc_obj{s}() typesafe allocations, from Subasri S.
* tag 'nf-next-26-07-24' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf-next:
ipvs: use type-safe allocation helpers in ip_vs_rht_alloc
netfilter: nf_conntrack_expect: store event cache in expectation
netfilter: conntrack_helper: pass master conntrack to helper functions
netfilter: flowtable: tear down flow entries with stale dst from GC
netfilter: nf_nat_sip: rewind offset when NAT shrinks the packet
netfilter: xt_tcpmss: extend checkentry to ipv6
ipvs: Move defense_work and est_reload_work to system_dfl_long_wq
====================
Link: https://patch.msgid.link/20260724104932.437729-1-pablo@netfilter.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Xin Xie says:
====================
net: hsr: PRP RedBox (PRP-SAN) support
This series adds PRP RedBox support to the hsr driver: a PRP node that
proxies one or more SANs sitting behind an interlink port (IEC 62439-3,
PRP-SAN). HSR-SAN has been supported since commit 5055cccfc2 ("net: hsr:
Provide RedBox support (HSR-SAN)"); this extends the equivalent capability
to PRP, reusing the existing protocol-neutral proxy machinery
(proxy_node_db, hsr_proxy_announce(), hsr_prune_proxy_nodes()).
A SAN behind the interlink does bidirectional unicast with peers on the PRP
network, its source MAC is preserved on the wire, the PRP RCT is correct,
and the RedBox announces each proxied SAN with the RedBox-MAC TLV (Type 30)
in its supervision frames.
The series is bisect-safe: the datapath, duplicate discard and supervision
support are added first; the rtnetlink rejection of "type hsr ... interlink
<dev> proto 1" is removed only in patch 3, once the feature is complete.
Design notes:
- prp_drop_frame() does not walk the node tables. The destination
classification (PRP-network node vs proxied SAN) is resolved once per
frame in fill_frame_info() and cached in struct hsr_frame_info, so the
per egress-port drop decision is O(1) in the softIRQ path. The
classification is gated on PRP RedBox devices (prot_version == PRP_V1 &&
hsr->redbox), so HSR RedBox traffic is not affected.
- The LAN A/B duplicate test is factored into prp_is_lan_dup() so the new
PRP interlink rules in prp_drop_frame() do not change hsr_drop_frame()
behaviour, including the NETIF_F_HW_HSR_FWD path. This is software PRP
RedBox only; it adds no new hardware-offload contract.
- The supervision emitter uses pre-reserved tailroom (hsr_init_skb() +
skb_put()) on the existing GFP_ATOMIC path; no skb_linearize() or
pskb_expand_head(). The RedBox-MAC TLV is followed by an explicit EOT
(Type 0, Length 0); padding via skb_put_padto(ETH_ZLEN) and the 6-byte
PRP RCT remain at the absolute tail of the egress frame.
- The hsr_get_node() hsr_ethhdr length guard is relaxed only for PRP
supervision frames (prot_version == PRP_V1 && ETH_P_PRP && is_sup), which
are untagged with mac_len == ETH_HLEN. HSR (ETH_P_HSR) supervision is
front-tagged and keeps the original length requirement, so HSR
malformed-frame filtering is unchanged.
====================
Link: https://patch.msgid.link/20260717201457.54-1-xiexinet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a kselftest that builds a PRP RedBox (interlink) with a SAN behind the
interlink and a peer DANP, and checks bidirectional unicast across the
interlink, preservation of the SAN source MAC on the PRP network, and that
the proxy-announce supervision frame carries the RedBox-MAC TLV (Type 30)
terminated by an EOT marker. It reuses the hsr_common.sh / lib.sh helpers
and skips cleanly on a kernel or iproute2 without PRP interlink support.
The background ping is killed by its exact PID: ip netns exec does not
isolate the PID namespace, so a pattern-based pkill could hit unrelated
processes on the host.
Signed-off-by: Xin Xie <xiexinet@gmail.com>
Link: https://patch.msgid.link/20260717201457.54-5-xiexinet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
With the PRP interlink datapath, duplicate discard and supervision support
in place, a PRP device can act as a RedBox. Remove the rtnetlink rejection
of "type hsr ... interlink <dev> proto 1"; the feature is implemented
unconditionally by the preceding patches.
Signed-off-by: Xin Xie <xiexinet@gmail.com>
Link: https://patch.msgid.link/20260717201457.54-4-xiexinet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
A PRP RedBox must announce the SANs it proxies so peers populate their
proxy node tables. The proxy-announce machinery (hsr_proxy_announce(),
armed via hsr->redbox) already iterates proxy_node_db under RCU and calls
send_sv_frame() once per SAN, but the PRP sender emitted neither the
announced SAN MAC nor the RedBox-MAC TLV that IEC 62439-3 requires.
Extend send_prp_supervision_frame() so that, for a proxy-announce
(identified by the interlink port, an O(1) test), the frame carries the
proxied SAN MAC as MacAddressA followed by the RedBox-MAC TLV (Type 30)
and an explicit End-of-TLV marker before padding.
hsr_get_node() must also accept the reinjected proxy-announce: a PRP
supervision frame is an untagged ETH_P_PRP frame (mac_len == ETH_HLEN, the
RCT is appended only on egress) sourced from macaddress_redbox, which is
never learned from data. Exempt only PRP supervision frames from the
hsr_ethhdr length guard; HSR (ETH_P_HSR) supervision is front-tagged and
keeps the original guard, so HSR malformed-frame filtering is unchanged.
Also align macaddress_redbox so that ether_addr_copy() and
ether_addr_equal() on it are safe on architectures without efficient
unaligned access.
Signed-off-by: Xin Xie <xiexinet@gmail.com>
Link: https://patch.msgid.link/20260717201457.54-3-xiexinet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
A PRP RedBox proxies SANs that sit behind an interlink port: their frames
must reach the PRP network with the SAN source MAC preserved, and PRP
unicast must be steered between the LAN and the SAN segment correctly.
Add the PRP interlink forwarding rules to prp_drop_frame() and give RedBox
nodes a second duplicate-discard slot so the two LAN copies of a frame
destined to a SAN collapse to a single delivery out the interlink.
The destination classification (is the unicast DA a PRP-network node or a
proxied SAN) is resolved once per frame in fill_frame_info(), gated to PRP
RedBox devices, and cached in struct hsr_frame_info, so prp_drop_frame()
stays O(1) and does not walk the node tables for every candidate egress
port in the softIRQ path. HSR RedBox frame classification is untouched.
Factor the LAN A/B duplicate test into prp_is_lan_dup() so the new PRP
interlink rules do not change hsr_drop_frame() behaviour, including the
NETIF_F_HW_HSR_FWD path which keeps using the LAN-duplicate test only.
Publish the RedBox state before the first hsr_add_port(): the slave and
interlink rx handlers are live from hsr_add_port() on and rtnl does not
stop softirq processing, so a frame could otherwise be handled while
hsr->redbox is still false. hsr_add_node() sizes each node's per-port
sequence state from hsr->redbox; a node learned in that window would get
a single-port sequence block, breaking the interlink duplicate discard
(WARN_ON_ONCE plus duplicate delivery to the SAN) and letting the
supervision sequence-block merge read beyond the source node's allocated
sequence bitmap. Publishing the flag before any port exists makes the
per-node sizing uniform by construction. This is safe: the proxy
announce timer is only armed from hsr_check_announce() once the master
is running, the packet-path readers of hsr->redbox tolerate an empty
proxy node database and an absent interlink port, and the
prune_proxy_timer is still armed only after the interlink port has been
attached successfully.
Additionally bound the supervision sequence-block merge by the smaller
of the two nodes' seq_port_cnt as defense in depth against mismatched
node sizes.
Signed-off-by: Xin Xie <xiexinet@gmail.com>
Link: https://patch.msgid.link/20260717201457.54-2-xiexinet@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Wei Fang says:
====================
net: enetc: cleanups and improvements
The first group of patches (1, 2, 5-7) eliminates code duplication
between the ENETC v1 and v4 drivers. Since both hardware generations
share identical register layouts for MAC promiscuous mode, MAC hash
filters, and VLAN promiscuous mode - differing only in register address
offsets - common helper functions are extracted into enetc_pf_common.c
and shared by both drivers.
Patch 3 converts ndo_set_rx_mode() to ndo_set_rx_mode_async(), removing
the dedicated workqueue that was previously needed to defer MAC address
list updates to a sleepable context.
Patch 4 replaces counter-based MAFT entry tracking with a bitmap, which
keeps hardware and software state in sync and avoids partial failures
during entry allocation.
Patches 8 and 9 fix phylink-related issues: removing invalid code from
enetc4_pl_mac_link_up() and properly differentiating phylink capabilities
between pseudo-MAC and standalone MAC.
The remaining patches (10-14) are minor cleanups: removing a redundant
VLAN promiscuous mode initialization in probe, using the PCI device name
for the debugfs directory, simplifying port speed configuration, removing
a redundant num_vsi field, using alloc_etherdev_mqs() for the VF driver,
and using kzalloc_flex() for a flexible array allocation.
====================
Link: https://patch.msgid.link/20260720014317.1059359-1-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The VF driver uses alloc_etherdev_mq() with ENETC_MAX_NUM_TXQS as the
queue count, which forces the TX and RX queue counts to be equal and
uses a compile-time constant rather than the actual hardware capability.
After enetc_get_si_caps() is called, si->num_tx_rings and
si->num_rx_rings reflect the actual number of rings assigned to the VF
by the PF. For the ENETC VF on LS1028A and the upcoming i.MX95/94, their
SoCs have no more than 6 CPUs, and the number of TX/RX rings allocated
to the VF is less than 8.
Therefore, switch to alloc_etherdev_mqs() so that the TX and RX queue
counts are set independently, each capped at ENETC_MAX_NUM_TXQS, based
on the actual number of rings assigned to the VF by the PF.
Note that if future SoCs have more than 6 CPUs and more than 6 RX rings
allocated to VFs, the size of the int_vector array in struct
enetc_ndev_priv will need to be modified. Similarly, if more than 8 TX
rings are allocated to each int_vector, ENETC_MAX_NUM_TXQS will also
need to be modified.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260720014317.1059359-14-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The num_vsi field in struct enetc_port_caps is populated by reading the
NUM_VSI field of the ECAPR1 register, which reports the number of VSIs
supported by the ENETC4 port. When CONFIG_PCI_IOV is enabled, this value
always matches pf->total_vfs, which is obtained from the read-only
PCI_SRIOV_TOTAL_VF register via pci_sriov_get_totalvfs() during probe.
Both ECAPR1[NUM_VSI] and PCI_SRIOV_TOTAL_VF are derived from the same
IERB register EaVFRIDAR[NUM_VF] (a 4-bit field), so they are guaranteed
to be equal. When CONFIG_PCI_IOV is disabled, pci_sriov_get_totalvfs()
returns 0, but this is benign since pci_enable_sriov() is also stubbed
to return -ENODEV, so no VF can be created, and enetc4_enable_all_si()
only enables the PF SI (PSI).
Since pf->total_vfs already reflects the number of VFs that can actually
be used, and is the established convention in the sibling FSL_ENETC PF
driver, there is no need to read and cache num_vsi separately in the port
capabilities structure. Remove the num_vsi field from enetc_port_caps,
and replace all uses of pf->caps.num_vsi with pf->total_vfs in the ring
allocation, SI enable, and debugfs code paths. Note that in the MSI-X
configuration, it is still necessary to obtain the actual number of VSIs
from ECAPR1.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20260720014317.1059359-13-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The PSIPVMR in ENETC v4 has the same bit layout and functionality as the
PSIPVMR register in ENETC v1: bit n (n <= 15) controls VLAN promiscuous
mode for SI n. The only difference between the two hardware generations
is the register address offset.
Since the register functionality is identical, the VLAN promiscuous mode
setting code can be shared between ENETC v1 and v4 drivers.
Move enetc_set_si_vlan_promisc() from enetc_pf.c to enetc_pf_common.c
and export it so that it can be shared between the two drivers. Add a
revision check using is_enetc_rev1() to select the correct register
offset (ENETC_PSIPVMR for v1 and ENETC4_PSIPVMR for v4) while keeping
the same logic.
Remove the v4-specific enetc4_pf_set_si_vlan_promisc() from enetc4_pf.c
and replace its call site with the new common enetc_set_si_vlan_promisc()
to eliminate code duplication.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260720014317.1059359-12-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Remove the enetc_set_vlan_promisc(), enetc_enable_si_vlan_promisc() and
enetc_disable_si_vlan_promisc() functions, and introduce a new unified
function enetc_set_si_vlan_promisc() to enable or disable VLAN
promiscuous mode for a specific SI. This simplifies the logic and makes
the interface more straightforward. The vlan_promisc_simap field in
struct enetc_pf is no longer needed to track the current state.
As ENETC V4 only changes the address offset of PSIPVMR register compared
to V1 without any functional difference, enetc_set_si_vlan_promisc() can
be moved to enetc_pf_common.c in the future with minor adjustments to be
reused by the ENETC V4 driver
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260720014317.1059359-11-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When adding phylink MAC operations support to the NETC switch driver,
Russell King pointed out several pieces of invalid logic in the
.mac_link_up() implementation (see [1] and [2]):
1) Half-duplex backpressure is not supported by the kernel, Ethernet
relies on packet dropping for congestion management.
2) phylink_autoneg_inband() is unnecessary, as RGMII in-band status is
not supported.
3) TX and RX pause are disabled in half-duplex mode, so there is no
need to override them in .mac_link_up().
The same invalid logic is also present in enetc4_pl_mac_link_up(), so
remove the invalid code from it.
Given enetc4_set_hd_flow_control() is removed, pf->caps.half_duplex has
also become useless and should therefore be removed as well.
Link: https://lore.kernel.org/imx/acEIQqI-_oyCym8O@shell.armlinux.org.uk/ # 1
Link: https://lore.kernel.org/imx/acEFwqmAvWls_9Ef@shell.armlinux.org.uk/ # 2
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20260720014317.1059359-9-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The ENETC pseudo-MACs are proprietary internal links that do not
implement any standard MII interface, so restrict their supported PHY
interface modes to PHY_INTERFACE_MODE_INTERNAL only.
Since pseudo-MACs can operate at any speed between 10Mbps and 25Gbps
in multiples of 10Mbps, set their MAC capabilities to cover the full
range of standard full-duplex speeds: 10/100/1000/2500/5000/10000/
20000/25000 Mbps.
For standalone ENETC (v4), expand the supported interface modes to
include 10GBASER in addition to the existing RGMII, SGMII, 1000BASEX,
2500BASEX and USXGMII modes, with MAC capabilities up to 10G. MAC_1000
is replaced with MAC_1000FD to explicitly exclude 1000M half-duplex,
which is not supported.
Note that 10GBASE-R mode of ENETC v4 has not supported yet, the current
patch adds PHY_INTERFACE_MODE_10GBASER simply as preparation for the
upcoming support of the 10GBASE-R mode.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Link: https://patch.msgid.link/20260720014317.1059359-8-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
enetc_create_debugfs() is called right after register_netdev(), at which
point ndev->name still holds the format "eth%d" (e.g., eth0) rather than
the final assigned name (e.g., via udev rules).
Use pci_name() instead of netdev_name() to name the debugfs directory.
The PCI device name is unique, stable, and available from the start,
making it a more reliable identifier for the debugfs entry. Therefore,
the observable debugfs path from something like
/sys/kernel/debug/eth0/mac_filter to a PCI BDF-style path such as
/sys/kernel/debug/0002:00:00.0/mac_filter.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260720014317.1059359-6-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Replace the counter-based MAFT entry tracking (num_mfe/mac_filter_num)
with a bitmap (maft_eid_bitmap) stored in struct ntmp_user, which is a
more appropriate place for NTMP resource management.
The bitmap approach brings two improvements. First, the entry deletion
in enetc4_pf_clear_maft_entries() now checks the return value of
ntmp_maft_delete_entry() and only clears the corresponding bit on
success, keeping hardware and software state in sync. Previously, the
counter was reset unconditionally regardless of whether the hardware
deletion actually succeeded.
Second, entry allocation in enetc4_pf_add_maft_entries() uses
ntmp_lookup_free_eid() to find available IDs dynamically, with an
upfront capacity check via bitmap_weight() to avoid partial failures.
The MAFT entry count is moved into ntmp_user.maft_num_entries and
initialized once during enetc4_init_ntmp_user(). Helper functions
enetc4_ntmp_bitmap_init() and enetc4_ntmp_bitmap_free() manage the
bitmap lifetime. The debugfs show function is updated accordingly to
iterate over set bits under rtnl_lock().
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20260720014317.1059359-5-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The current ndo_set_rx_mode() is called under netif_addr_lock spinlock
with BHs disabled, which prevents drivers from sleeping. To work around
this limitation, the enetc driver uses a dedicated workqueue to defer
MAC address list updates to a sleepable context.
Since commit 3554b4345d ("net: introduce ndo_set_rx_mode_async and
netdev_rx_mode_work") introduced the ndo_set_rx_mode_async() callback,
drivers can now handle address list updates directly in a sleepable
context.
Therefore, convert the enetc driver to use ndo_set_rx_mode_async() and
remove the dedicated workqueue and the deferred work item accordingly.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20260720014317.1059359-4-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The PSIUMHFR and PSIMMHFR registers in ENETC v4 have the same bit layout
as in ENETC v1. The only difference between the two hardware generations
is the register address offsets.
Since the register functionality is identical, the MAC hash filter
configuration code can be shared between the ENETC v1 and v4 drivers.
Extract two new common helper functions, enetc_set_si_uc_hash_filter()
and enetc_set_si_mc_hash_filter(), into enetc_pf_common.c. These helpers
select the correct register offset based on the hardware revision via
is_enetc_rev1().
Remove v1-specific enetc_clear_mac_ht_flt() and enetc_set_mac_ht_flt()
from enetc_pf.c, and v4-specific enetc4_pf_set_si_uc_hash_filter() and
enetc4_pf_set_si_mc_hash_filter() from enetc4_pf.c, as they are now
superseded by the shared implementations.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Link: https://patch.msgid.link/20260720014317.1059359-3-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The PSIPMMR (Port Station Interface Promiscuous MAC Mode Register) in
ENETC v4 has the same bit layout as the PSIPMR register in ENETC v1: bit
n controls unicast promiscuous mode for SI n, and bit (n + 16) controls
multicast promiscuous mode for SI n. The only difference between the two
hardware generations is the register address offset.
Since the register functionality is identical, the MAC promiscuous mode
setting code can be shared between ENETC v1 and v4 drivers.
Rename ENETC_PSIPMR to ENETC_PSIPMMR in enetc_hw.h to match the actual
register name used in the reference manual, and extract two new common
helper functions, enetc_set_si_uc_promisc() and
enetc_set_si_mc_promisc(), into enetc_pf_common.c. These helpers select
the correct register offset based on the hardware revision via
is_enetc_rev1().
Remove the v4-specific enetc4_pf_set_si_mac_promisc() function from
enetc4_pf.c and the duplicate PSIPMMR_SI_MAC_UP/MP macro definitions
from enetc4_hw.h, as they are now superseded by the shared code.
Signed-off-by: Wei Fang <wei.fang@nxp.com>
Reviewed-by: Joe Damato <joe@dama.to>
Link: https://patch.msgid.link/20260720014317.1059359-2-wei.fang@oss.nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Johannes Berg says:
====================
wireless-next-2026-07-26
Mostly driver changes this time:
- new driver mm81x for an S1G device
- new driver nxpwifi for NXP devices
(mostly forked off from mwifiex)
- ath12k: much kernel infrastructure integration work
- brcmfmac: DPP support, some Cypress part update
- nl80211: per-link statistics support
====================
Link: https://patch.msgid.link/20260726105205.942922-60-johannes@sipsolutions.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add packet parsing support for custom L2 headers. Also add support
to include a field from the custom header for flow tag generation.
Introduce a new flow key type NIX_FLOW_KEY_TYPE_CH_LEN_90B which
maps to the NPC_LT_LA_CUSTOM_L2_90B_ETHER layer type. This extracts
a 2-byte field at a 24-byte offset in layer A to be used in flow
tag generation.
Signed-off-by: Satheesh Paul A <psatheesh@marvell.com>
Signed-off-by: Nitin Shetty J <nshettyj@marvell.com>
Link: https://patch.msgid.link/20260715072035.617544-1-nshettyj@marvell.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In both hns_rcb_get_tx_coalesced_frames() and
hns_rcb_set_tx_coalesced_frames(), the local variable reg holds a
register offset passed to dsaf_read_dev() or dsaf_write_dev().
Register offsets on this hardware are 32-bit values. Use u32 for reg
to match the register access interfaces and avoid implying that
64-bit offsets are supported.
Signed-off-by: Daniil Agalakov <ade@amicon.ru>
Signed-off-by: Daniil Iskhakov <dish@amicon.ru>
Link: https://patch.msgid.link/20260715125856.19346-1-dish@amicon.ru
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
