Fix a typo and a redundant phrase in the block comment above
__sys_accept4(): "thats" -> "that's", and drop the trailing
"to recvmsg" that repeats the recvmsg() reference earlier in
the same sentence.
No functional change.
Signed-off-by: Avinash Duduskar <avinash.duduskar@gmail.com>
Link: https://patch.msgid.link/20260516101109.479042-1-avinash.duduskar@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
htb_dump_class_stats() and htb_offload_aggregate_stats()
call gnet_stats_basic_sync_init(&cl->bstats) which
is wrong on 32bit arches when syncp is cleared.
Make sure to acquire qdisc spinlock and use
_bstats_set() to ease future lockless dumps.
Fixes: 8327158624 ("sch_htb: Stats for offloaded HTB")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Maxim Mikityanskiy <maximmi@mellanox.com>
Cc: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260518090518.629245-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This patch continues the effort to refactor workqueue APIs, which has begun
with the changes introducing new workqueues and a new alloc_workqueue flag:
commit 128ea9f6cc ("workqueue: Add system_percpu_wq and system_dfl_wq")
commit 930c2ea566 ("workqueue: Add new WQ_PERCPU flag")
The point of the refactoring is to eventually alter the default behavior of
workqueues to become unbound by default so that their workload placement is
optimized by the scheduler.
Before that to happen, workqueue users must be converted to the better named
new workqueues with no intended behaviour changes:
system_wq -> system_percpu_wq
system_unbound_wq -> system_dfl_wq
This way the old obsolete workqueues (system_wq, system_unbound_wq) can be
removed in the future.
Cc: David Ahern <dsahern@kernel.org>
Cc: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/
Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260515135143.259669-2-marco.crivellari@suse.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jordan Rhee says:
====================
gve: add support for PTP gettimex64
This patch series adds support to obtain near-simultaneous NIC and
system timestamps with gettimex64. This enables daemons like
chrony and phc2sys to synchronize the system clock to the NIC clock.
GVE does not have direct register access to the NIC hardware clock, so
it must issue an AdminQ command to read the NIC clock. Two paths
for obtaining a cross-timestamp are implemented: a precise path using
system counter values sampled by the device, and a fallback path using
system counter values sampled in the driver using
ptp_read_system_prets()/postts().
To use the precise path, the current system clocksource must match the
units returned by the device, which on x86 is X86_TSC and on ARM64 is
ARM_ARCH_COUNTER. The clockid requested for the cross-timestamp must
be either CLOCK_REALTIME or CLOCK_MONOTONIC_RAW. These conditions hold
by default on GCP VMs using Chrony, so we expect the precise path to be
used the vast majority of the time. If the system clocksource is changed
to kvm-clock, it activates the fallback path. Ethtool counters have been
added to count how many times each path is used.
The uncertainty window in the precise path is typically around 1-2us,
while in the fallback path is around 60-80us. This table shows a
comparison in chrony tracking statistics between the precise path and
fallback path. The RMS offset is nearly 4 orders of magnitude smaller
in the precise path.
| | Fallback Path | Precise path |
| --------------- | --------------------- | ------------------------ |
| System time | 0.000000005 s slow | 0.000000001 s fast |
| Last offset | +0.000005606 seconds | +0.000000001 seconds |
| RMS offset | 0.000009020 seconds | 0.000000002 seconds |
| Frequency | 4.115 ppm fast | 0.362 ppm fast |
| Residual freq | +2.515 ppm | +0.000 ppm |
| Skew | 18.480 ppm | 0.001 ppm |
| Root delay | 0.000000001 seconds | 0.000000001 seconds |
| Root dispersion | 0.000081905 seconds | 0.000001169 seconds |
| Update interval | 0.5 seconds | 0.5 seconds |
| Leap status | Normal | Normal |
The first two patches pave the way for the PTP implementation by
quieting excessive logging and refactoring an existing routine for
thread safety.
====================
Link: https://patch.msgid.link/20260514225842.110706-1-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Enable chrony and phc2sys to synchronize system clock to NIC clock.
Two paths are implemented: a precise path using system counter values
sampled by the device, and a fallback path using system counter values
sampled in the driver using ptp_read_system_prets()/postts().
To use the precise path, the current system clocksource must match the
units returned by the device, which on x86 is X86_TSC and on ARM64 is
ARM_ARCH_COUNTER. The clockid requested for the cross-timestamp must
be either CLOCK_REALTIME or CLOCK_MONOTONIC_RAW. These conditions hold
by default on GCP VMs using Chrony, so we expect the precise path to be
used the vast majority of the time. If the system clocksource is changed
to kvm-clock, it activates the fallback path. Ethtool counters have been
added to count how many times each path is used.
The uncertainty window in the precise path is typically around 1-2us,
while in the fallback path is around 60-80us.
Stub implementions of adjfine and adjtime are added to avoid NULL
dereference when phc2sys tries to adjust the clock.
Cc: John Stultz <jstultz@google.com>
Cc: Thomas Gleixner <tglx@kernel.org>
Cc: Stephen Boyd <sboyd@kernel.org>
Cc: David Woodhouse <dwmw2@infradead.org>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Kevin Yang <yyd@google.com>
Reviewed-by: Naman Gulati <namangulati@google.com>
Signed-off-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Harshitha Ramamurthy <hramamurthy@google.com>
Link: https://patch.msgid.link/20260514225842.110706-4-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a mutex to protect the shared DMA buffer that receives NIC
timestamp reports. The NIC timestamp will be read from two different
threads: the periodic worker and upcoming `gettimex64`.
Move clock registration to the last step of initialization to ensure
that all data needed by the clock module is initialized before
the clock is exposed to usermode.
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Joshua Washington <joshwash@google.com>
Signed-off-by: Ankit Garg <nktgrg@google.com>
Signed-off-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Harshitha Ramamurthy <hramamurthy@google.com>
Link: https://patch.msgid.link/20260514225842.110706-3-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
AdminQ commands may return -EAGAIN under certain transient conditions.
These commands are intended to be retried by the driver, so logging
a formal error to the system log is misleading and creates
unnecessary noise.
Modify the logging logic to skip the error message when the result
is -EAGAIN, and move logging to dev_err_ratelimited() to avoid
spamming the log.
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Joshua Washington <joshwash@google.com>
Signed-off-by: Jordan Rhee <jordanrhee@google.com>
Signed-off-by: Harshitha Ramamurthy <hramamurthy@google.com>
Link: https://patch.msgid.link/20260514225842.110706-2-hramamurthy@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When roaming, the FDB destination port can change without holding the
bridge's hash lock. Therefore, add missing READ_ONCE() annotations in
both RCU readers and readers that hold the lock. In the latter case, the
annotation is not needed in places where the FDB entry was already
validated to be a local entry since such entries cannot roam.
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260517115009.175163-1-idosch@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently, during probe defferal the driver will print multiple times:
mscc-miim e20101a8.mdio: Cannot register MDIO bus (-517)
So, lets silence that by using the dev_err_probe() for printing the probe
error as it handles probe defferal.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20260515120608.706361-1-robert.marko@sartura.hr
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Modern systems may have more than 16 PPS sources and current hard-coded
limit breaks registration of some devices. Let's bump the limit to 256
in hope it will be enough in foreseen future.
Signed-off-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Acked-by: Rodolfo Giometti <giometti@enneenne.com>
Link: https://patch.msgid.link/20260515135028.2021318-1-vadim.fedorenko@linux.dev
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
BUILD_BUG_ON() is evaluated at compile time. The same three checks
for __HWTSTAMP_TX_CNT, __HWTSTAMP_FILTER_CNT and __HWTSTAMP_FLAG_CNT
are already present earlier in this file (tsconfig_reply_size()).
Repeating them at the start of ethnl_set_tsconfig() does not
strengthen the guarantee; remove the redundant block there and
keep a single occurrence.
Signed-off-by: Chenguang Zhao <zhaochenguang@kylinos.cn>
Link: https://patch.msgid.link/20260515032550.138572-1-zhaochenguang@kylinos.cn
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Xuan Zhuo says:
====================
eea: Add basic driver framework for Alibaba Elastic Ethernet Adaptor
Add a driver framework for EEA that will be available in the future.
This driver is currently quite minimal, implementing only fundamental
core functionalities. Key features include: I/O queue management via
adminq, basic PCI-layer operations, and essential RX/TX data
communication capabilities. It also supports the creation,
initialization, and management of network devices (netdev). Furthermore,
the ring structures for both I/O queues and adminq have been abstracted
into a simple, unified, and reusable library implementation,
facilitating future extension and maintenance.
====================
Link: https://patch.msgid.link/20260514095138.80680-1-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This commit adds support for ndo_get_stats64 to provide accurate
interface statistics. With the TX and RX data paths now fully functional,
it is appropriate to register the netdevice and expose the interface to
userspace.
Registered the network device via register_netdev, and updated the
corresponding unregister_netdev and dev_close routines to ensure
synchronization.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-9-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Implement the core logic for transmitting packets in the EEA TX path,
including packet preparation and submission to the underlying transport.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-7-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Implement the core logic for receiving packets in the EEA RX path,
including packet buffering and basic validation.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-6-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add basic driver framework for the Alibaba Elastic Ethernet Adapter(EEA).
This commit introduces the implementation for the netdevice open and
stop.
This commit introduces HA to restore the device when error occurs,
but in HA scenarios the driver can't ensure to restore the status
correctly.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-5-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add basic driver framework for the Alibaba Elastic Ethernet Adapter(EEA).
This commit creates the netdevice after PCI probe,
and initializes the admin queue to send commands to the device.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-4-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add basic driver framework for the Alibaba Elastic Ethernet Adapter(EEA).
This commit introduces the ring and descriptor implementations.
These structures and ring APIs are used by the RX, TX, and admin queues.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-3-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add basic driver framework for the Alibaba Elastic Ethernet Adapter(EEA).
This commit implements the EEA PCI probe functionality.
Reviewed-by: Dust Li <dust.li@linux.alibaba.com>
Reviewed-by: Philo Lu <lulie@linux.alibaba.com>
Signed-off-by: Wen Gu <guwen@linux.alibaba.com>
Signed-off-by: Xuan Zhuo <xuanzhuo@linux.alibaba.com>
Link: https://patch.msgid.link/20260514095138.80680-2-xuanzhuo@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add PPPoE test-cases to the GRO selftest. Only run a subset of
common_tests to avoid changing the hardcoded L3 offsets everywhere.
Add a new "pppoe_sid" test case to verify that packets with different
PPPoE session IDs are correctly identified as separate flows and not
coalesced.
Signed-off-by: Qingfang Deng <qingfang.deng@linux.dev>
Link: https://patch.msgid.link/20260513013400.7467-2-qingfang.deng@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Only handles packets where the pppoe header length field matches the exact
packet length. Significantly improves rx throughput.
When running NAT traffic through a MediaTek MT7621 devices from a host
behind PPPoE to a host directly connected via ethernet, the TCP throughput
that the device is able to handle improves from ~130 Mbit/s to ~630 Mbit/s,
using fraglist GRO.
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Qingfang Deng <qingfang.deng@linux.dev>
Tested-by: Pablo Neira Ayuso <pablo@netfilter.org>
Link: https://patch.msgid.link/20260513013400.7467-1-qingfang.deng@linux.dev
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Bobby Eshleman says:
====================
net: devmem: support devmem with netkit devices
This series enables TCP devmem TX through netkit devices.
Netkit now supports queue leasing. A physical NIC's RX queue can be
leased to a netkit guest interface inside a container namespace. This
gives the container a devmem-capable data path on the RX side (bind-rx,
etc...). On the TX side, the container process binds to its netkit guest
interface and sends traffic that netkit redirects (via BPF or ip
forwarding) to the physical NIC for DMA.
Two things in the existing devmem TX path prevent this from working:
1. validate_xmit_unreadable_skb() requires dev->netmem_tx before it will
forward a dmabuf-backed (unreadable) skb. This protects skbs from
landing on devices that don't have the IOMMU mappings for the backing
dmabuf or that don't speak netmem. Netkit, however, does not support
DMA, doesn't attempt to read unreadable skb pages and so doesn't
break netmem (it is pure skb routing and redirection). It is
functionally capable of routing unreadable skbs, but there is no way
for the TX validation pathway to distinguish between a device that
will actually attempt DMA-ing the skb and another device
(like netkit) that does not DMA but also does not break
netmem.
2. bind_tx_doit uses the bound device as the DMA device. When the user
binds devmem TX to the netkit guest, the bind handler attempts to
create DMA mappings against netkit, which has no DMA capability and
no IOMMU mappings.
This series solves these problems as follows:
1. Extend netmem_tx to two bits, assigned to one of three values:
NETMEM_TX_NONE - netmem not supported
NETMEM_TX_DMA - netmem supported and performs DMA
NETMEM_TX_NO_DMA - netmem supported, but does not DMA
With these bits, phys devices can set NETMEM_TX_DMA and devices like
netkit set NETMEM_TX_NO_DMA. The validation TX path ensures that any
DMA-capable netdev exactly matches the bound device, guaranteeing the
correct mapping of the bound dmabuf. The validation TX path also
allows devices with NETMEM_TX_NO_DMA to pass, knowing these devices
will not misuse netmem or run into IOMMU faults. After redirection or
routing and the skb finally makes its way through the stack to a
physical device's TX path, the above NETMEM_TX_DMA check is performed
again to guarantee the device has the appropriate binding/mappings.
2. On TX bind, the bind handler recognizes NETMEM_TX_NO_DMA devices and
finds the phys TX device and binds to that instead. For the netkit
case, if it has been leased a queue from a DMA-capable device
already, then the bind action is performed on the DMA-capable device
instead and the dmabuf is mapped correctly.
====================
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-0-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add nk_devmem.py with four tests for TCP devmem through a netkit device.
These tests are just duplicates of the original devmem tests, with some
adjusted parameters such as telling ncdevmem to avoid device setup
(since it only has access to netkit, not a phys device).
Each test uses NetDrvContEnv with primary_rx_redirect=True to set up the
BPF redirect program on the primary netkit interface, then calls a
shared run_*() helper which probes for devmem support and configures
the NIC (HDS, RSS, queue lease) before driving the test. NIC state is
restored per-test via defer() callbacks registered inside the helper.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-8-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When sending from a namespace that has access to a netkit device with a
leased queue, the nk primary in the host namespace needs to redirect its
RX to the physical device. This patch adds that redirection bpf program
and teaches the harness to install it.
Add primary_rx_redirect=False parameter to NetDrvContEnv.__init__().
When enabled, _attach_primary_rx_redirect_bpf() attaches a new BPF TC
program (nk_primary_rx_redirect.bpf.c) to the primary (host-side) netkit
interface. The program redirects non-ICMPv6 IPv6 packets to the physical
NIC via bpf_redirect_neigh(), with the physical ifindex configured via
the .bss map. ICMPv6 is left on the host's netkit primary so IPv6
neighbor discovery still work locally.
Extract _find_bss_map_id() from _attach_bpf() into a reusable helper so
other BPF attachment methods can use it.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-7-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Adding netkit-based devmem tests is a straight-forward copy of devmem
test commands plus some args for the nk cases, so this patch breaks out
these command builders into helpers used by both.
Though we tried to avoid libraries to avoid increasing the barrier of
entry/complexity (see selftests/drivers/net/README.md, section "Avoid
libraries and frameworks"), factoring out these functions seemed like
the lesser of two evils in this case of using the same commands, just
with slightly different args per environment.
I experimented with just having all of the tests in the same file to
avoid having helpers in a library file, but because ksft_run() is
limited to a single call per file, and the new tests will require
different environments (NetDrvContEnv/NetDrvEpEnv), it would have been
necessary to have each test set up its own environment instead of
sharing one for the entire ksft_run() run. This came at the cost of
ballooning the test time (from under 5s to 30s on my test system), so to
strike a balance these tests were placed in separate files so they could
keep a shared environment across a single ksft_run() run shared across
all tests using the same env type (introduced in subsequent patches).
The helpers work transparently with both plain and netkit environments
by inspecting cfg for netkit-specific attributes (netns, nk_queue,
etc...).
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-6-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Subsequent patches will use the _nk_guest_ifname as a public attr for
setting up devmem. Rename to nk_guest_ifname to avoid angering the
linter about the '_' prefix being used for a non-private attr.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-5-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a -n (skip_config) flag that causes ncdevmem to skip NIC
configuration when operating as an RX server. When -n is passed,
ncdevmem skips configuring header split, RSS, and flow steering, as well
as their teardown on exit.
This allows ksft tests to pre-configure the NIC in the host namespace
before launching ncdevmem in the guest namespace. This is needed for
netkit devmem tests where the test harness namespace has direct access
to the NIC and the ncdevmem namespace does not.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-4-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When a netkit virtual device leases queues from a physical NIC, devmem
TX bindings created on the netkit device must still result in the dmabuf
being mapped for dma by the physical device. This patch accomplishes
this by teaching the bind handler to search for the underlying
DMA-capable device by looking it up via leased rx queues. The function
netdev_find_netmem_tx_dev(), used for finding the underlying DMA-capable
device, can be extended to support other non-netkit NETMEM_TX_NO_DMA
devices in the future if needed.
Additionally, this patch extends validate_xmit_unreadable_skb() to
support the netkit case, where the skb is validated twice: once on the
netkit guest device and again on the physical NIC after BPF redirect or
ip forwarding.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-3-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Some virtual devices like netkit (or ifb) never DMA and never touch frag
contents, they just forward the skb to another device. They are unable
to forward unreadable skbs, however, because they fail to pass TX
validation checks on dev->netmem_tx. The existing two-state
NETMEM_TX_NONE / NETMEM_TX_DMA doesn't give the TX validator enough
information to differentiate devices that will attempt DMA on the
unreadable skb from those that will simply route it untouched.
Add a third mode to the enum so drivers can indicate 1) if they have
netmem TX support, and 2) if they do, whether they are DMA-capable:
NETMEM_TX_NO_DMA - pass-through, device never DMAs
Widen dev->netmem_tx from a 1-bit field to 2 bits to fit the new value,
and declare netkit as NETMEM_TX_NO_DMA. Devmem TX support over these
devices comes in a follow-up patch.
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-2-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Devices that support netmem TX previously set dev->netmem_tx = true.
This was checked in validate_xmit_unreadable_skb() to drop unreadable
skbs (skbs with dmabuf-backed frags) before they reach drivers that
would mishandle them or devices that would not have the iommu mappings
for them.
A subsequent patch will introduce a third state for virtual devices
that forward unreadable skbs without ever performing DMA on them. To
prepare for that, convert the boolean dev->netmem_tx into an enum:
NETMEM_TX_NONE - no netmem TX support (drop unreadable skbs)
NETMEM_TX_DMA - full support, device does DMA
Update the existing NIC drivers (bnxt, gve, mlx5, fbnic) and the
validators in net/core to use the new enum. No functional change.
Acked-by: Harshitha Ramamurthy <hramamurthy@google.com>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Signed-off-by: Bobby Eshleman <bobbyeshleman@meta.com>
Link: https://patch.msgid.link/20260514-tcp-dm-netkit-v5-1-408c59b91e66@meta.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tariq Toukan says:
====================
net/mlx5e: simplify and optimize napi poll flow
This series simplifies the mlx5e napi poll flow and reduces branching in
hot paths by leveraging existing dependencies between channel features.
Patch 1 avoids passing the full channel object to kTLS RX code paths
when only the async ICOSQ is needed, and slightly optimizes the common
flow in the pending resync handling logic.
Patch 2 further reduces branches in napi poll by relying on established
feature dependencies between XDP, async ICOSQ, XSK, and kTLS RX state.
====================
Link: https://patch.msgid.link/20260514111038.338251-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Do not pass channel just to extract the async ICOSQ.
It's already extracted, pass it.
Re-order the checks in mlx5e_ktls_rx_pending_resync_list to optimize the
common flow.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Link: https://patch.msgid.link/20260514111038.338251-3-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reduce the number of branches in napi poll, based on the following list
of dependencies:
1. xsk_open=t only if c->xdp and c->async_icosq.
2. c->xdpsq only if c->xdp.
3. c->xdp implies c->async_icosq.
4. ktls_rx_was_enabled implies c->async_icosq.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Link: https://patch.msgid.link/20260514111038.338251-2-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The last user was removed in commit 72628da6d6 ("net: ks8851:
Remove ks8851_mll.c") which consolidated the driver into a
common implementation. No file includes this header.
Signed-off-by: Costa Shulyupin <costa.shul@redhat.com>
Link: https://patch.msgid.link/20260515184531.1515418-1-costa.shul@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The Kconfig prompts for the DIBS options read:
DIBS support (DIBS) [N/m/y/?]
intra-OS shortcut with dibs loopback (DIBS_LO) [N/y/?]
Clarify the DIBS prompt by expanding the acronym.
Capitalize the first character of the DIBS_LO prompt.
While at it, join the first two lines of the DIBS help text into a real
sentence.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Alexandra Winter <wintera@linux.ibm.com>
Link: https://patch.msgid.link/9093259c43e5d1996965d1522562444419196a19.1778838921.git.geert+renesas@glider.be
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The PSP spec states that the lower 31b of the SPI need to be
non-zero. Though not in the spec, I think it is reasonable to reset
the lower 31b of the spi space after a key rotation, and to also
decline to generate session keys when the lower 31b saturate.
Signed-off-by: Daniel Zahka <daniel.zahka@gmail.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260515-spi-handle-v1-1-debf8cb467cb@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tariq Toukan says:
====================
net/mlx5: frag buffer improvements
This series adds observability for mlx5 fragment buffer DMA pools and
improves the default NUMA placement policy for fragment buffer
allocations.
Patch 1 adds a debugfs interface exposing per-node DMA pool usage
statistics for mlx5_frag_buf allocations, helping with debugging and
visibility into pool utilization.
Patch 2 improves locality of default fragment buffer allocations by
using numa_mem_id() when no explicit NUMA node is requested, allowing
allocations to prefer the current CPU's local memory node.
Together, these changes improve both introspection and memory locality
behavior of mlx5 fragment buffer allocations.
====================
Link: https://patch.msgid.link/20260514104925.337570-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Use the current CPU's local memory node when callers do not request a
specific NUMA node for mlx5_frag_buf allocations.
Signed-off-by: Nimrod Oren <noren@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260514104925.337570-2-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
addrconf_dad_failure() transitions ifp->state from DAD to POSTDAD
via addrconf_dad_end(), which drops ifp->lock on return. The lock
is re-acquired after net_info_ratelimited(). A concurrent
ipv6_del_addr() can take the lock in that window, set ifp->state
to DEAD and run list_del_rcu(&ifp->if_list).
addrconf_dad_failure() then overwrites DEAD with ERRDAD at errdad:
and schedules a new dad_work. The work calls ipv6_del_addr()
again, hitting the already-poisoned list entry:
general protection fault: 0000 [#1] SMP NOPTI
CPU: 4 PID: 217 Comm: kworker/4:1
Workqueue: ipv6_addrconf addrconf_dad_work
RIP: 0010:ipv6_del_addr+0xe9/0x280
RAX: dead000000000122
Call Trace:
addrconf_dad_stop+0x113/0x140
addrconf_dad_work+0x28c/0x430
process_one_work+0x1eb/0x3b0
worker_thread+0x4d/0x400
kthread+0x104/0x140
ret_from_fork+0x35/0x40
Fold the addrconf_dad_end() logic into addrconf_dad_failure() under
a single ifp->lock critical section. The STABLE_PRIVACY branch
temporarily drops ifp->lock around address regeneration, so at
lock_errdad: verify the state is still POSTDAD before transitioning
to ERRDAD; bail out otherwise to avoid overwriting a state set by
another path while the lock was released.
Fixes: c15b1ccadb ("ipv6: move DAD and addrconf_verify processing to workqueue")
Signed-off-by: Linmao Li <lilinmao@kylinos.cn>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20260513025509.3776405-1-lilinmao@kylinos.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Eric Dumazet says:
====================
net/sched: sch_htb: first round of fixes
First round of fixes in sch_htb.
I chose to send a small series, to reduce chances of multiple versions.
====================
Link: https://patch.msgid.link/20260514095935.3926276-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
htb_dump_class_stats() will soon run locklessly.
(no RTNL, not qdisc spinlock).
Add READ_ONCE()/WRITE_ONCE() annotations on these fields:
- cl->overlimits
- cl->drops
- cl->tokens
- cl->ctokens
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260514095935.3926276-5-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
htb_dump_class_stats() will soon run locklessly.
(no RTNL, not qdisc spinlock).
Remove cl->xstats and replace it with two fields:
- xstats_lends
- xstats_borrows
Then use READ_ONCE()/WRITE_ONCE() annotations on them, and change
htb_dump_class_stats to use a private struct tc_htb_xstats.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260514095935.3926276-4-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
htb_dump() runs without holding qdisc spinlock.
Add missing READ_ONCE()/WRITE_ONCE() annotations around
q->overlimits and q->direct_pkts.
Fixes: edb09eb17e ("net: sched: do not acquire qdisc spinlock in qdisc/class stats dump")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260514095935.3926276-3-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
htb_dump() runs without holding qdisc spinlock.
It is illegal to touch sch->flags with non locked RMW,
as concurrent readers might see intermediate wrong values.
Set TCQ_F_OFFLOADED in control path (htb_init()) instead.
Fixes: d03b195b5a ("sch_htb: Hierarchical QoS hardware offload")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260514095935.3926276-2-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
llc_conn_ac_inc_npta_value() and llc_conn_ac_inc_tx_win_size()
clamp their counters to the maximum valid 7-bit value via
(u8) ~LLC_2_SEQ_NBR_MODULO. LLC_2_SEQ_NBR_MODULO is defined as
((u8) 128) in include/net/llc_pdu.h, but the (u8) cast does not
prevent integer promotion of the operand of ~: ~128 is computed
as int (0xffffff7f), and the surrounding (u8) cast truncates
back to 0x7f. The result is correct (127), but the implicit
truncation is flagged by sparse:
net/llc/llc_c_ac.c:1008:38: warning: cast truncates bits from
constant value (ffffff7f becomes 7f)
(and three more at lines 1009, 1099, 1100)
Replace the (u8) ~LLC_2_SEQ_NBR_MODULO expression with
LLC_2_SEQ_NBR_MODULO - 1, which evaluates to 127 directly and
silences sparse.
The same ~LLC_2_SEQ_NBR_MODULO pattern also appears in
include/net/llc_pdu.h:148 as part of PDU_GET_NEXT_Vr, but there
the result is immediately &-masked, so the int promotion is
harmless and sparse does not flag it; it is left alone.
This patch is the minimum diff to silence the warning. The
counter-clamp idiom itself could be modernized to
min_t(u8, ..., LLC_2_SEQ_NBR_MODULO - 1), but that is a
separate cleanup left for another patch.
No functional change.
Signed-off-by: Avinash Duduskar <avinash.duduskar@gmail.com>
Link: https://patch.msgid.link/20260513092253.3035961-1-avinash.duduskar@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
