Dragos Tatulea [Thu, 5 Mar 2026 14:26:33 +0000 (16:26 +0200)]
net/mlx5e: RX, Fix XDP multi-buf frag counting for striding RQ
XDP multi-buf programs can modify the layout of the XDP buffer when the
program calls bpf_xdp_pull_data() or bpf_xdp_adjust_tail(). The
referenced commit in the fixes tag corrected the assumption in the mlx5
driver that the XDP buffer layout doesn't change during a program
execution. However, this fix introduced another issue: the dropped
fragments still need to be counted on the driver side to avoid page
fragment reference counting issues.
The issue was discovered by the drivers/net/xdp.py selftest,
more specifically the test_xdp_native_tx_mb:
- The mlx5 driver allocates a page_pool page and initializes it with
a frag counter of 64 (pp_ref_count=64) and the internal frag counter
to 0.
- The test sends one packet with no payload.
- On RX (mlx5e_skb_from_cqe_mpwrq_nonlinear()), mlx5 configures the XDP
buffer with the packet data starting in the first fragment which is the
page mentioned above.
- The XDP program runs and calls bpf_xdp_pull_data() which moves the
header into the linear part of the XDP buffer. As the packet doesn't
contain more data, the program drops the tail fragment since it no
longer contains any payload (pp_ref_count=63).
- mlx5 device skips counting this fragment. Internal frag counter
remains 0.
- mlx5 releases all 64 fragments of the page but page pp_ref_count is
63 => negative reference counting error.
The problem applies for XDP_PASS as well which is handled in a different
code path in the driver.
This patch fixes the issue by doing page frag counting on all the
original XDP buffer fragments for all relevant XDP actions (XDP_TX ,
XDP_REDIRECT and XDP_PASS). This is basically reverting to the original
counting before the commit in the fixes tag.
As frag_page is still pointing to the original tail, the nr_frags
parameter to xdp_update_skb_frags_info() needs to be calculated
in a different way to reflect the new nr_frags.
Gal Pressman [Thu, 5 Mar 2026 14:26:32 +0000 (16:26 +0200)]
net/mlx5e: Fix DMA FIFO desync on error CQE SQ recovery
In case of a TX error CQE, a recovery flow is triggered,
mlx5e_reset_txqsq_cc_pc() resets dma_fifo_cc to 0 but not dma_fifo_pc,
desyncing the DMA FIFO producer and consumer.
After recovery, the producer pushes new DMA entries at the old
dma_fifo_pc, while the consumer reads from position 0.
This causes us to unmap stale DMA addresses from before the recovery.
The DMA FIFO is a purely software construct with no HW counterpart.
At the point of reset, all WQEs have been flushed so dma_fifo_cc is
already equal to dma_fifo_pc. There is no need to reset either counter,
similar to how skb_fifo pc/cc are untouched.
Carolina Jubran [Thu, 5 Mar 2026 14:26:31 +0000 (16:26 +0200)]
net/mlx5: Fix peer miss rules host disabled checks
The check on mlx5_esw_host_functions_enabled(esw->dev) for adding VF
peer miss rules is incorrect. These rules match traffic from peer's VFs,
so the local device's host function status is irrelevant. Remove this
check to ensure peer VF traffic is properly handled regardless of local
host configuration.
Also fix the PF peer miss rule deletion to be symmetric with the add
path, so only attempt to delete the rule if it was actually created.
When moving to switchdev mode when the device doesn't support IPsec,
we try to clean up the IPsec resources anyway which causes the crash
below, fix that by correctly checking for IPsec support before trying
to clean up its resources.
Cosmin Ratiu [Thu, 5 Mar 2026 08:10:19 +0000 (10:10 +0200)]
net/mlx5: Fix deadlock between devlink lock and esw->wq
esw->work_queue executes esw_functions_changed_event_handler ->
esw_vfs_changed_event_handler and acquires the devlink lock.
.eswitch_mode_set (acquires devlink lock in devlink_nl_pre_doit) ->
mlx5_devlink_eswitch_mode_set -> mlx5_eswitch_disable_locked ->
mlx5_eswitch_event_handler_unregister -> flush_workqueue deadlocks
when esw_vfs_changed_event_handler executes.
Fix that by no longer flushing the work to avoid the deadlock, and using
a generation counter to keep track of work relevance. This avoids an old
handler manipulating an esw that has undergone one or more mode changes:
- the counter is incremented in mlx5_eswitch_event_handler_unregister.
- the counter is read and passed to the ephemeral mlx5_host_work struct.
- the work handler takes the devlink lock and bails out if the current
generation is different than the one it was scheduled to operate on.
- mlx5_eswitch_cleanup does the final draining before destroying the wq.
No longer flushing the workqueue has the side effect of maybe no longer
cancelling pending vport_change_handler work items, but that's ok since
those are disabled elsewhere:
- mlx5_eswitch_disable_locked disables the vport eq notifier.
- mlx5_esw_vport_disable disarms the HW EQ notification and marks
vport->enabled under state_lock to false to prevent pending vport
handler from doing anything.
- mlx5_eswitch_cleanup destroys the workqueue and makes sure all events
are disabled/finished.
Fixes: f1bc646c9a06 ("net/mlx5: Use devl_ API in mlx5_esw_offloads_devlink_port_register") Signed-off-by: Cosmin Ratiu <cratiu@nvidia.com> Reviewed-by: Moshe Shemesh <moshe@nvidia.com> Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com> Reviewed-by: Simon Horman <horms@kernel.org> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260305081019.1811100-1-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Laurent Vivier [Wed, 4 Mar 2026 13:43:38 +0000 (14:43 +0100)]
qmi_wwan: allow max_mtu above hard_mtu to control rx_urb_size
Commit c7159e960f14 ("usbnet: limit max_mtu based on device's hard_mtu")
capped net->max_mtu to the device's hard_mtu in usbnet_probe(). While
this correctly prevents oversized packets on standard USB network
devices, it breaks the qmi_wwan driver.
qmi_wwan relies on userspace (e.g. ModemManager) setting a large MTU on
the wwan0 interface to configure rx_urb_size via usbnet_change_mtu().
QMI modems negotiate USB transfer sizes of 16,383 or 32,767 bytes, and
the USB receive buffers must be sized accordingly. With max_mtu capped
to hard_mtu (~1500 bytes), userspace can no longer raise the MTU, the
receive buffers remain small, and download speeds drop from >300 Mbps
to ~0.8 Mbps.
Introduce a FLAG_NOMAXMTU driver flag that allows individual usbnet
drivers to opt out of the max_mtu cap. Set this flag in qmi_wwan's
driver_info structures to restore the previous behavior for QMI devices,
while keeping the safety fix in place for all other usbnet drivers.
Jakub Kicinski [Sat, 7 Mar 2026 00:25:19 +0000 (16:25 -0800)]
Merge branch 'bond-fix-2-link-state-issues'
Hangbin Liu says:
====================
bond: fix 2 link state issues
This patch set fixes two bonding link state issues:
1. Broadcast mode incorrectly sets usable_slaves, causing updelay to be ignored
2. BOND_LINK_FAIL and BOND_LINK_BACK are treated as invalid states, generating
confusing error messages
Here is the reproducer:
```
ip netns add ns
ip -n ns link add bond0 type bond mode 3 miimon 100 updelay 200 downdelay 200
ip -n ns link add type veth
ip -n ns link add type veth
ip -n ns link set veth1 up
ip -n ns link set veth3 up
ip -n ns link set veth0 master bond0
ip -n ns link set veth2 master bond0
ip -n ns link set bond0 up
sleep 1
ip -n ns link set veth3 down
sleep 1
ip -n ns link set veth3 up
sleep 1
dmesg | tail
```
====================
Hangbin Liu [Wed, 4 Mar 2026 07:13:54 +0000 (15:13 +0800)]
bonding: handle BOND_LINK_FAIL, BOND_LINK_BACK as valid link states
Before the fixed commit, we check slave->new_link during commit
state, which values are only BOND_LINK_{NOCHANGE, UP, DOWN}. After
the commit, we start using slave->link_new_state, which state also could
be BOND_LINK_{FAIL, BACK}.
For example, when we set updelay/downdelay, after a failover,
the slave->link_new_state could be set to BOND_LINK_{FAIL, BACK} in
bond_miimon_inspect(). And later in bond_miimon_commit(), it will treat
it as invalid and print an error, which would cause confusion for users.
[ 106.440254] bond0: (slave veth2): link status down for interface, disabling it in 200 ms
[ 106.440265] bond0: (slave veth2): invalid new link 1 on slave
[ 106.648276] bond0: (slave veth2): link status definitely down, disabling slave
[ 107.480271] bond0: (slave veth2): link status up, enabling it in 200 ms
[ 107.480288] bond0: (slave veth2): invalid new link 3 on slave
[ 107.688302] bond0: (slave veth2): link status definitely up, 10000 Mbps full duplex
Let's handle BOND_LINK_{FAIL, BACK} as valid link states.
Hangbin Liu [Wed, 4 Mar 2026 07:13:53 +0000 (15:13 +0800)]
bonding: do not set usable_slaves for broadcast mode
After commit e0caeb24f538 ("net: bonding: update the slave array for broadcast mode"),
broadcast mode will also set all_slaves and usable_slaves during
bond_enslave(). But if we also set updelay, during enslave, the
slave init state will be BOND_LINK_BACK. And later
bond_update_slave_arr() will alloc usable_slaves but add nothing.
This will cause bond_miimon_inspect() to have ignore_updelay
always true. So the updelay will be always ignored. e.g.
[ 6.498368] bond0: (slave veth2): link status definitely down, disabling slave
[ 7.536371] bond0: (slave veth2): link status up, enabling it in 0 ms
[ 7.536402] bond0: (slave veth2): link status definitely up, 10000 Mbps full duplex
To fix it, we can either always call bond_update_slave_arr() on every
place when link changes. Or, let's just not set usable_slaves for
broadcast mode.
Johan Hovold [Thu, 5 Mar 2026 10:45:49 +0000 (11:45 +0100)]
net: mctp: fix device leak on probe failure
Driver core holds a reference to the USB interface and its parent USB
device while the interface is bound to a driver and there is no need to
take additional references unless the structures are needed after
disconnect.
This driver takes a reference to the USB device during probe but does
not to release it on probe failures.
Drop the redundant device reference to fix the leak, reduce cargo
culting, make it easier to spot drivers where an extra reference is
needed, and reduce the risk of further memory leaks.
Fixes: 0791c0327a6e ("net: mctp: Add MCTP USB transport driver") Cc: stable@vger.kernel.org # 6.15 Signed-off-by: Johan Hovold <johan@kernel.org> Acked-by: Jeremy Kerr <jk@codeconstruct.com.au> Link: https://patch.msgid.link/20260305104549.16110-1-johan@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net: dsa: realtek: rtl8365mb: remove ifOutDiscards from rx_packets
rx_packets should report the number of frames successfully received:
unicast + multicast + broadcast. Subtracting ifOutDiscards (a TX
counter) is incorrect and can undercount RX packets. RX drops are
already reported via rx_dropped (e.g. etherStatsDropEvents), so
there is no need to adjust rx_packets.
This patch removes the subtraction of ifOutDiscards from rx_packets
in rtl8365mb_stats_update().
Linus Torvalds [Thu, 5 Mar 2026 19:00:46 +0000 (11:00 -0800)]
Merge tag 'net-7.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Pull networking fixes from Jakub Kicinski:
"Including fixes from CAN, netfilter and wireless.
Current release - new code bugs:
- sched: cake: fixup cake_mq rate adjustment for diffserv config
- wifi: fix missing ieee80211_eml_params member initialization
Previous releases - regressions:
- tcp: give up on stronger sk_rcvbuf checks (for now)
Previous releases - always broken:
- net: fix rcu_tasks stall in threaded busypoll
- sched:
- fq: clear q->band_pkt_count[] in fq_reset()
- only allow act_ct to bind to clsact/ingress qdiscs and shared
blocks
- bridge: check relevant per-VLAN options in VLAN range grouping
- xsk: fix fragment node deletion to prevent buffer leak
Misc:
- spring cleanup of inactive maintainers"
* tag 'net-7.0-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (138 commits)
xdp: produce a warning when calculated tailroom is negative
net: enetc: use truesize as XDP RxQ info frag_size
libeth, idpf: use truesize as XDP RxQ info frag_size
i40e: use xdp.frame_sz as XDP RxQ info frag_size
i40e: fix registering XDP RxQ info
ice: change XDP RxQ frag_size from DMA write length to xdp.frame_sz
ice: fix rxq info registering in mbuf packets
xsk: introduce helper to determine rxq->frag_size
xdp: use modulo operation to calculate XDP frag tailroom
selftests/tc-testing: Add tests exercising act_ife metalist replace behaviour
net/sched: act_ife: Fix metalist update behavior
selftests: net: add test for IPv4 route with loopback IPv6 nexthop
net: ipv6: fix panic when IPv4 route references loopback IPv6 nexthop
net: vxlan: fix nd_tbl NULL dereference when IPv6 is disabled
net: bridge: fix nd_tbl NULL dereference when IPv6 is disabled
MAINTAINERS: remove Thomas Falcon from IBM ibmvnic
MAINTAINERS: remove Claudiu Manoil and Alexandre Belloni from Ocelot switch
MAINTAINERS: replace Taras Chornyi with Elad Nachman for Marvell Prestera
MAINTAINERS: remove Jonathan Lemon from OpenCompute PTP
MAINTAINERS: replace Clark Wang with Frank Li for Freescale FEC
...
Linus Torvalds [Thu, 5 Mar 2026 16:05:05 +0000 (08:05 -0800)]
Merge tag 'trace-v7.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace
Pull tracing fixes from Steven Rostedt:
- Fix thresh_return of function graph tracer
The update to store data on the shadow stack removed the abuse of
using the task recursion word as a way to keep track of what
functions to ignore. The trace_graph_return() was updated to handle
this, but when function_graph tracer is using a threshold (only trace
functions that took longer than a specified time), it uses
trace_graph_thresh_return() instead.
This function was still incorrectly using the task struct recursion
word causing the function graph tracer to permanently set all
functions to "notrace"
- Fix thresh_return nosleep accounting
When the calltime was moved to the shadow stack storage instead of
being on the fgraph descriptor, the calculations for the amount of
sleep time was updated. The calculation was done in the
trace_graph_thresh_return() function, which also called the
trace_graph_return(), which did the calculation again, causing the
time to be doubled.
Remove the call to trace_graph_return() as what it needed to do
wasn't that much, and just do the work in
trace_graph_thresh_return().
- Fix syscall trace event activation on boot up
The syscall trace events are pseudo events attached to the
raw_syscall tracepoints. When the first syscall event is enabled, it
enables the raw_syscall tracepoint and doesn't need to do anything
when a second syscall event is also enabled.
When events are enabled via the kernel command line, syscall events
are partially enabled as the enabling is called before rcu_init. This
is due to allow early events to be enabled immediately. Because
kernel command line events do not distinguish between different types
of events, the syscall events are enabled here but are not fully
functioning. After rcu_init, they are disabled and re-enabled so that
they can be fully enabled.
The problem happened is that this "disable-enable" is done one at a
time. If more than one syscall event is specified on the command
line, by disabling them one at a time, the counter never gets to
zero, and the raw_syscall is not disabled and enabled, keeping the
syscall events in their non-fully functional state.
Instead, disable all events and re-enabled them all, as that will
ensure the raw_syscall event is also disabled and re-enabled.
- Disable preemption in ftrace pid filtering
The ftrace pid filtering attaches to the fork and exit tracepoints to
add or remove pids that should be traced. They access variables
protected by RCU (preemption disabled). Now that tracepoint callbacks
are called with preemption enabled, this protection needs to be added
explicitly, and not depend on the functions being called with
preemption disabled.
- Disable preemption in event pid filtering
The event pid filtering needs the same preemption disabling guards as
ftrace pid filtering.
- Fix accounting of the memory mapped ring buffer on fork
Memory mapping the ftrace ring buffer sets the vm_flags to DONTCOPY.
But this does not prevent the application from calling
madvise(MADVISE_DOFORK). This causes the mapping to be copied on
fork. After the first tasks exits, the mapping is considered unmapped
by everyone. But when he second task exits, the counter goes below
zero and triggers a WARN_ON.
Since nothing prevents two separate tasks from mmapping the ftrace
ring buffer (although two mappings may mess each other up), there's
no reason to stop the memory from being copied on fork.
Update the vm_operations to have an ".open" handler to update the
accounting and let the ring buffer know someone else has it mapped.
- Add all ftrace headers in MAINTAINERS file
The MAINTAINERS file only specifies include/linux/ftrace.h But misses
ftrace_irq.h and ftrace_regs.h. Make the file use wildcards to get
all *ftrace* files.
* tag 'trace-v7.0-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
ftrace: Add MAINTAINERS entries for all ftrace headers
tracing: Fix WARN_ON in tracing_buffers_mmap_close
tracing: Disable preemption in the tracepoint callbacks handling filtered pids
ftrace: Disable preemption in the tracepoint callbacks handling filtered pids
tracing: Fix syscall events activation by ensuring refcount hits zero
fgraph: Fix thresh_return nosleeptime double-adjust
fgraph: Fix thresh_return clear per-task notrace
====================
Address XDP frags having negative tailroom
Aside from the issue described below, tailroom calculation does not account
for pages being split between frags, e.g. in i40e, enetc and
AF_XDP ZC with smaller chunks. These series address the problem by
calculating modulo (skb_frag_off() % rxq->frag_size) in order to get
data offset within a smaller block of memory. Please note, xskxceiver
tail grow test passes without modulo e.g. in xdpdrv mode on i40e,
because there is not enough descriptors to get to flipped buffers.
Many ethernet drivers report xdp Rx queue frag size as being the same as
DMA write size. However, the only user of this field, namely
bpf_xdp_frags_increase_tail(), clearly expects a truesize.
Such difference leads to unspecific memory corruption issues under certain
circumstances, e.g. in ixgbevf maximum DMA write size is 3 KB, so when
running xskxceiver's XDP_ADJUST_TAIL_GROW_MULTI_BUFF, 6K packet fully uses
all DMA-writable space in 2 buffers. This would be fine, if only
rxq->frag_size was properly set to 4K, but value of 3K results in a
negative tailroom, because there is a non-zero page offset.
We are supposed to return -EINVAL and be done with it in such case,
but due to tailroom being stored as an unsigned int, it is reported to be
somewhere near UINT_MAX, resulting in a tail being grown, even if the
requested offset is too much(it is around 2K in the abovementioned test).
This later leads to all kinds of unspecific calltraces.
The issue can be fixed in all in-tree drivers, but we cannot just trust OOT
drivers to not do this. Therefore, make tailroom a signed int and produce a
warning when it is negative to prevent such mistakes in the future.
The issue can also be easily reproduced with ice driver, by applying
the following diff to xskxceiver and enjoying a kernel panic in xdpdrv mode:
diff --git a/tools/testing/selftests/bpf/prog_tests/test_xsk.c b/tools/testing/selftests/bpf/prog_tests/test_xsk.c
index 5af28f359cfd..042d587fa7ef 100644
--- a/tools/testing/selftests/bpf/prog_tests/test_xsk.c
+++ b/tools/testing/selftests/bpf/prog_tests/test_xsk.c
@@ -2541,8 +2541,8 @@ int testapp_adjust_tail_grow_mb(struct test_spec *test)
{
test->mtu = MAX_ETH_JUMBO_SIZE;
/* Grow by (frag_size - last_frag_Size) - 1 to stay inside the last fragment */
- return testapp_adjust_tail(test, (XSK_UMEM__MAX_FRAME_SIZE / 2) - 1,
- XSK_UMEM__LARGE_FRAME_SIZE * 2);
+ return testapp_adjust_tail(test, XSK_UMEM__MAX_FRAME_SIZE * 100,
+ 6912);
}
int testapp_tx_queue_consumer(struct test_spec *test)
If we print out the values involved in the tailroom calculation:
Larysa Zaremba [Thu, 5 Mar 2026 11:12:50 +0000 (12:12 +0100)]
xdp: produce a warning when calculated tailroom is negative
Many ethernet drivers report xdp Rx queue frag size as being the same as
DMA write size. However, the only user of this field, namely
bpf_xdp_frags_increase_tail(), clearly expects a truesize.
Such difference leads to unspecific memory corruption issues under certain
circumstances, e.g. in ixgbevf maximum DMA write size is 3 KB, so when
running xskxceiver's XDP_ADJUST_TAIL_GROW_MULTI_BUFF, 6K packet fully uses
all DMA-writable space in 2 buffers. This would be fine, if only
rxq->frag_size was properly set to 4K, but value of 3K results in a
negative tailroom, because there is a non-zero page offset.
We are supposed to return -EINVAL and be done with it in such case, but due
to tailroom being stored as an unsigned int, it is reported to be somewhere
near UINT_MAX, resulting in a tail being grown, even if the requested
offset is too much (it is around 2K in the abovementioned test). This later
leads to all kinds of unspecific calltraces.
The issue can be fixed in all in-tree drivers, but we cannot just trust OOT
drivers to not do this. Therefore, make tailroom a signed int and produce a
warning when it is negative to prevent such mistakes in the future.
Fixes: bf25146a5595 ("bpf: add frags support to the bpf_xdp_adjust_tail() API") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com> Acked-by: Martin KaFai Lau <martin.lau@kernel.org> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-10-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:49 +0000 (12:12 +0100)]
net: enetc: use truesize as XDP RxQ info frag_size
The only user of frag_size field in XDP RxQ info is
bpf_xdp_frags_increase_tail(). It clearly expects truesize instead of DMA
write size. Different assumptions in enetc driver configuration lead to
negative tailroom.
Set frag_size to the same value as frame_sz.
Fixes: 2768b2e2f7d2 ("net: enetc: register XDP RX queues with frag_size") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-9-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:48 +0000 (12:12 +0100)]
libeth, idpf: use truesize as XDP RxQ info frag_size
The only user of frag_size field in XDP RxQ info is
bpf_xdp_frags_increase_tail(). It clearly expects whole buffer size instead
of DMA write size. Different assumptions in idpf driver configuration lead
to negative tailroom.
To make it worse, buffer sizes are not actually uniform in idpf when
splitq is enabled, as there are several buffer queues, so rxq->rx_buf_size
is meaningless in this case.
Use truesize of the first bufq in AF_XDP ZC, as there is only one. Disable
growing tail for regular splitq.
Fixes: ac8a861f632e ("idpf: prepare structures to support XDP") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-8-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:47 +0000 (12:12 +0100)]
i40e: use xdp.frame_sz as XDP RxQ info frag_size
The only user of frag_size field in XDP RxQ info is
bpf_xdp_frags_increase_tail(). It clearly expects whole buffer size instead
of DMA write size. Different assumptions in i40e driver configuration lead
to negative tailroom.
Set frag_size to the same value as frame_sz in shared pages mode, use new
helper to set frag_size when AF_XDP ZC is active.
Fixes: a045d2f2d03d ("i40e: set xdp_rxq_info::frag_size") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-7-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:46 +0000 (12:12 +0100)]
i40e: fix registering XDP RxQ info
Current way of handling XDP RxQ info in i40e has a problem, where frag_size
is not updated when xsk_buff_pool is detached or when MTU is changed, this
leads to growing tail always failing for multi-buffer packets.
Couple XDP RxQ info registering with buffer allocations and unregistering
with cleaning the ring.
Fixes: a045d2f2d03d ("i40e: set xdp_rxq_info::frag_size") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-6-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:45 +0000 (12:12 +0100)]
ice: change XDP RxQ frag_size from DMA write length to xdp.frame_sz
The only user of frag_size field in XDP RxQ info is
bpf_xdp_frags_increase_tail(). It clearly expects whole buff size instead
of DMA write size. Different assumptions in ice driver configuration lead
to negative tailroom.
This allows to trigger kernel panic, when using
XDP_ADJUST_TAIL_GROW_MULTI_BUFF xskxceiver test and changing packet size to
6912 and the requested offset to a huge value, e.g.
XSK_UMEM__MAX_FRAME_SIZE * 100.
Due to other quirks of the ZC configuration in ice, panic is not observed
in ZC mode, but tailroom growing still fails when it should not.
Use fill queue buffer truesize instead of DMA write size in XDP RxQ info.
Fix ZC mode too by using the new helper.
Fixes: 2fba7dc5157b ("ice: Add support for XDP multi-buffer on Rx side") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-5-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:44 +0000 (12:12 +0100)]
ice: fix rxq info registering in mbuf packets
XDP RxQ info contains frag_size, which depends on the MTU. This makes the
old way of registering RxQ info before calculating new buffer sizes
invalid. Currently, it leads to frag_size being outdated, making it
sometimes impossible to grow tailroom in a mbuf packet. E.g. fragments are
actually 3K+, but frag size is still as if MTU was 1500.
Always register new XDP RxQ info after reconfiguring memory pools.
Fixes: 2fba7dc5157b ("ice: Add support for XDP multi-buffer on Rx side") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-4-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:43 +0000 (12:12 +0100)]
xsk: introduce helper to determine rxq->frag_size
rxq->frag_size is basically a step between consecutive strictly aligned
frames. In ZC mode, chunk size fits exactly, but if chunks are unaligned,
there is no safe way to determine accessible space to grow tailroom.
Report frag_size to be zero, if chunks are unaligned, chunk_size otherwise.
Fixes: 24ea50127ecf ("xsk: support mbuf on ZC RX") Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-3-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Larysa Zaremba [Thu, 5 Mar 2026 11:12:42 +0000 (12:12 +0100)]
xdp: use modulo operation to calculate XDP frag tailroom
The current formula for calculating XDP tailroom in mbuf packets works only
if each frag has its own page (if rxq->frag_size is PAGE_SIZE), this
defeats the purpose of the parameter overall and without any indication
leads to negative calculated tailroom on at least half of frags, if shared
pages are used.
There are not many drivers that set rxq->frag_size. Among them:
* i40e and enetc always split page uniformly between frags, use shared
pages
* ice uses page_pool frags via libeth, those are power-of-2 and uniformly
distributed across page
* idpf has variable frag_size with XDP on, so current API is not applicable
* mlx5, mtk and mvneta use PAGE_SIZE or 0 as frag_size for page_pool
As for AF_XDP ZC, only ice, i40e and idpf declare frag_size for it. Modulo
operation yields good results for aligned chunks, they are all power-of-2,
between 2K and PAGE_SIZE. Formula without modulo fails when chunk_size is
2K. Buffers in unaligned mode are not distributed uniformly, so modulo
operation would not work.
To accommodate unaligned buffers, we could define frag_size as
data + tailroom, and hence do not subtract offset when calculating
tailroom, but this would necessitate more changes in the drivers.
Define rxq->frag_size as an even portion of a page that fully belongs to a
single frag. When calculating tailroom, locate the data start within such
portion by performing a modulo operation on page offset.
Fixes: bf25146a5595 ("bpf: add frags support to the bpf_xdp_adjust_tail() API") Acked-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Larysa Zaremba <larysa.zaremba@intel.com> Link: https://patch.msgid.link/20260305111253.2317394-2-larysa.zaremba@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Whenever an ife action replace changes the metalist, instead of
replacing the old data on the metalist, the current ife code is appending
the new metadata. Aside from being innapropriate behavior, this may lead
to an unbounded addition of metadata to the metalist which might cause an
out of bounds error when running the encode op:
ip -6 nexthop add id 100 dev lo
ip route add 172.20.20.0/24 nhid 100
ping -c1 172.20.20.1 # kernel crash
Problem Description
When a standalone IPv6 nexthop object is created with a loopback device,
fib6_nh_init() misclassifies it as a reject route. Nexthop objects have
no destination prefix (fc_dst=::), so fib6_is_reject() always matches
any loopback nexthop. The reject path skips fib_nh_common_init(), leaving
nhc_pcpu_rth_output unallocated. When an IPv4 route later references
this nexthop and triggers a route lookup, __mkroute_output() calls
raw_cpu_ptr(nhc->nhc_pcpu_rth_output) on a NULL pointer, causing a page
fault.
The reject classification was designed for regular IPv6 routes to prevent
kernel routing loops, but nexthop objects should not be subject to this
check since they carry no destination information. Loop prevention is
handled separately when the route itself is created.
[1] https://syzkaller.appspot.com/bug?extid=334190e097a98a1b81bb
====================
When a standalone IPv6 nexthop object is created with a loopback device
(e.g., "ip -6 nexthop add id 100 dev lo"), fib6_nh_init() misclassifies
it as a reject route. This is because nexthop objects have no destination
prefix (fc_dst=::), causing fib6_is_reject() to match any loopback
nexthop. The reject path skips fib_nh_common_init(), leaving
nhc_pcpu_rth_output unallocated. If an IPv4 route later references this
nexthop, __mkroute_output() dereferences NULL nhc_pcpu_rth_output and
panics.
Simplify the check in fib6_nh_init() to only match explicit reject
routes (RTF_REJECT) instead of using fib6_is_reject(). The loopback
promotion heuristic in fib6_is_reject() is handled separately by
ip6_route_info_create_nh(). After this change, the three cases behave
as follows:
2. Implicit loopback reject route ("ip -6 route add 2001:db8::/32 dev lo"):
RTF_REJECT is not set, takes normal path, fib_nh_common_init() is
called. ip6_route_info_create_nh() still promotes it to reject
afterward. nhc_pcpu_rth_output is allocated but unused, which is
harmless.
3. Standalone nexthop object ("ip -6 nexthop add id 100 dev lo"):
RTF_REJECT is not set, takes normal path, fib_nh_common_init() is
called. nhc_pcpu_rth_output is properly allocated, fixing the crash
when IPv4 routes reference this nexthop.
Suggested-by: Ido Schimmel <idosch@nvidia.com> Fixes: 493ced1ac47c ("ipv4: Allow routes to use nexthop objects") Reported-by: syzbot+334190e097a98a1b81bb@syzkaller.appspotmail.com Closes: https://lore.kernel.org/all/698f8482.a70a0220.2c38d7.00ca.GAE@google.com/T/ Signed-off-by: Jiayuan Chen <jiayuan.chen@shopee.com> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Reviewed-by: David Ahern <dsahern@kernel.org> Link: https://patch.msgid.link/20260304113817.294966-2-jiayuan.chen@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net: vxlan: fix nd_tbl NULL dereference when IPv6 is disabled
When booting with the 'ipv6.disable=1' parameter, the nd_tbl is never
initialized because inet6_init() exits before ndisc_init() is called
which initializes it. If an IPv6 packet is injected into the interface,
route_shortcircuit() is called and a NULL pointer dereference happens on
neigh_lookup().
Fix this by adding an early check on route_shortcircuit() when protocol
is ETH_P_IPV6. Note that ipv6_mod_enabled() cannot be used here because
VXLAN can be built-in even when IPv6 is built as a module.
net: bridge: fix nd_tbl NULL dereference when IPv6 is disabled
When booting with the 'ipv6.disable=1' parameter, the nd_tbl is never
initialized because inet6_init() exits before ndisc_init() is called
which initializes it. Then, if neigh_suppress is enabled and an ICMPv6
Neighbor Discovery packet reaches the bridge, br_do_suppress_nd() will
dereference ipv6_stub->nd_tbl which is NULL, passing it to
neigh_lookup(). This causes a kernel NULL pointer dereference.
Fix this by replacing IS_ENABLED(IPV6) call with ipv6_mod_enabled() in
the callers. This is in essence disabling NS/NA suppression when IPv6 is
disabled.
Fixes: ed842faeb2bd ("bridge: suppress nd pkts on BR_NEIGH_SUPPRESS ports") Reported-by: Guruprasad C P <gurucp2005@gmail.com> Closes: https://lore.kernel.org/netdev/CAHXs0ORzd62QOG-Fttqa2Cx_A_VFp=utE2H2VTX5nqfgs7LDxQ@mail.gmail.com/ Signed-off-by: Fernando Fernandez Mancera <fmancera@suse.de> Reviewed-by: Ido Schimmel <idosch@nvidia.com> Acked-by: Nikolay Aleksandrov <razor@blackwall.org> Link: https://patch.msgid.link/20260304120357.9778-1-fmancera@suse.de Signed-off-by: Jakub Kicinski <kuba@kernel.org>
====================
MAINTAINERS: annual cleanup of inactive maintainers
Annual cleanup of inactive maintainers under networking.
The goal is to make sure MAINTAINERS reflect reality for
code which is relatively actively changed (at least 70 commits
in the last 2 years or at least 120 commits in the last 5 years).
Those who either:
- were the initial author / "upstreamer" of the driver; or
- authored at least 1/3rd of the exiting code base (per git blame); or
- authored at least 25% of commits before becoming inactive
are moved to CREDITS.
The discovery of inactive maintainers was done using gitdm tools,
with a bunch of ad-hoc scripts on top to do the rest. I tried to
double check the results but this is mostly a scripted cleanup
so please report inaccuracies if any.
====================
Jakub Kicinski [Tue, 3 Mar 2026 21:53:11 +0000 (13:53 -0800)]
MAINTAINERS: remove Thomas Falcon from IBM ibmvnic
We have not seen emails or tags from Thomas's IBM address
(tlfalcon@linux.ibm.com) in over 5 years. Looks like Thomas
is active in perf tooling at Intel (thomas.falcon@intel.com).
Subsystem IBM Power SRIOV Virtual NIC Device Driver
Changes 49 / 134 (36%)
Last activity: 2025-08-26
Haren Myneni <haren@linux.ibm.com>:
Tags 3c14917953a5 2025-08-26 00:00:00 2
Rick Lindsley <ricklind@linux.ibm.com>:
Nick Child <nnac123@linux.ibm.com>:
Author d93a6caab5d7 2025-03-25 00:00:00 14
Tags d93a6caab5d7 2025-03-25 00:00:00 16
Thomas Falcon <tlfalcon@linux.ibm.com>:
Top reviewers:
[22]: drt@linux.ibm.com
[13]: horms@kernel.org
[9]: ricklind@linux.vnet.ibm.com
[3]: davemarq@linux.ibm.com
INACTIVE MAINTAINER Thomas Falcon <tlfalcon@linux.ibm.com>
Move Thomas to CREDITS as the initial author of ibmvnic.
Jakub Kicinski [Tue, 3 Mar 2026 21:53:10 +0000 (13:53 -0800)]
MAINTAINERS: remove Claudiu Manoil and Alexandre Belloni from Ocelot switch
We have not seen tags from Claudiu for the Ocelot switch driver
in over 5 years. He is active upstream in other NXP subsystems
(ENETC, gianfar), with 46 emails on lore since 2024.
We have not seen tags from Alexandre for the Ocelot switch driver
in over 5 years. He is very active upstream in other subsystems
(RTC, I3C, Atmel/Microchip SoC), with over 1,200 emails on lore
since 2024.
Vladimir Oltean is active.
Jakub Kicinski [Tue, 3 Mar 2026 21:53:09 +0000 (13:53 -0800)]
MAINTAINERS: replace Taras Chornyi with Elad Nachman for Marvell Prestera
We have not seen emails or tags from Taras in over 5 years,
and there is no recent mailing list activity.
Elad Nachman has been providing reviews in the last couple
of years and is the top reviewer for this subsystem.
Jakub Kicinski [Tue, 3 Mar 2026 21:53:07 +0000 (13:53 -0800)]
MAINTAINERS: replace Clark Wang with Frank Li for Freescale FEC
We have not seen tags from Clark for FEC in over 5 years.
He has some limited recent activity on the mailing list in other
NXP subsystems (stmmac, phy). Wei Fang and Shenwei Wang are active,
with decent review coverage (61%).
Frank Li has been reviewing code actively more recenty, let's
make it official.
Subsystem FREESCALE IMX / MXC FEC DRIVER
Changes 57 / 92 (61%)
Last activity: 2026-02-10
Wei Fang <wei.fang@nxp.com>:
Author 25eb3058eb70 2026-02-10 00:00:00 33
Tags 25eb3058eb70 2026-02-10 00:00:00 61
Shenwei Wang <shenwei.wang@nxp.com>:
Author d466c16026e9 2025-09-14 00:00:00 6
Tags d466c16026e9 2025-09-14 00:00:00 6
Clark Wang <xiaoning.wang@nxp.com>:
Top reviewers:
[23]: Frank.Li@nxp.com
[17]: andrew@lunn.ch
[4]: csokas.bence@prolan.hu
[3]: horms@kernel.org
[2]: maxime.chevallier@bootlin.com
INACTIVE MAINTAINER Clark Wang <xiaoning.wang@nxp.com>
Jakub Kicinski [Tue, 3 Mar 2026 21:53:06 +0000 (13:53 -0800)]
MAINTAINERS: remove DENG Qingfang from MediaTek switch
We have not seen tags from DENG Qingfang for the MediaTek
switch driver in over 5 years. He is active upstream with
PPP/PPPoE patches in net-next. Chester and Daniel are active.
Subsystem MEDIATEK SWITCH DRIVER
Changes 26 / 70 (37%)
Last activity: 2025-12-01
Chester A. Unal <chester.a.unal@arinc9.com>:
Tags 585943b7ad30 2025-12-01 00:00:00 7
Daniel Golle <daniel@makrotopia.org>:
Author 497041d76301 2025-04-23 00:00:00 2
Tags 3b87e60d2131 2025-12-01 00:00:00 14
DENG Qingfang <dqfext@gmail.com>:
Sean Wang <sean.wang@mediatek.com>:
Top reviewers:
[4]: andrew@lunn.ch
[4]: florian.fainelli@broadcom.com
[4]: arinc.unal@arinc9.com
[2]: olteanv@gmail.com
INACTIVE MAINTAINER DENG Qingfang <dqfext@gmail.com>
Jakub Kicinski [Tue, 3 Mar 2026 21:53:05 +0000 (13:53 -0800)]
MAINTAINERS: remove Sean Wang from MediaTek Ethernet and switch
We have not seen tags from Sean in over 5 years,
with only one mailing list post since 2024.
Felix and Lorenzo are active for the Ethernet driver,
and Chester, Daniel and DENG Qingfang are active for
the switch driver.
Subsystem MEDIATEK ETHERNET DRIVER
Changes 55 / 113 (48%)
Last activity: 2025-10-12
Felix Fietkau <nbd@nbd.name>:
Author d4736737110f 2025-09-02 00:00:00 3
Tags d4736737110f 2025-09-02 00:00:00 4
Sean Wang <sean.wang@mediatek.com>:
Lorenzo Bianconi <lorenzo@kernel.org>:
Author 96326447d466 2025-08-13 00:00:00 35
Tags 3abc0e55ea1f 2025-10-12 00:00:00 40
Top reviewers:
[26]: horms@kernel.org
[5]: andrew@lunn.ch
[4]: jacob.e.keller@intel.com
[3]: shannon.nelson@amd.com
[3]: michal.swiatkowski@linux.intel.com
INACTIVE MAINTAINER Sean Wang <sean.wang@mediatek.com>
Subsystem MEDIATEK SWITCH DRIVER
Changes 26 / 70 (37%)
Last activity: 2025-12-01
Chester A. Unal <chester.a.unal@arinc9.com>:
Tags 585943b7ad30 2025-12-01 00:00:00 7
Daniel Golle <daniel@makrotopia.org>:
Author 497041d76301 2025-04-23 00:00:00 2
Tags 3b87e60d2131 2025-12-01 00:00:00 14
DENG Qingfang <dqfext@gmail.com>:
Sean Wang <sean.wang@mediatek.com>:
Top reviewers:
[4]: andrew@lunn.ch
[4]: florian.fainelli@broadcom.com
[4]: arinc.unal@arinc9.com
[2]: olteanv@gmail.com
INACTIVE MAINTAINER Sean Wang <sean.wang@mediatek.com>
Jakub Kicinski [Tue, 3 Mar 2026 21:53:03 +0000 (13:53 -0800)]
MAINTAINERS: remove Jerin Jacob from Marvell OcteonTX2
We have not seen tags from Jerin for OcteonTX2 in over 5 years.
Recent lore activity is in DPDK (non-kernel), not Linux.
Sunil, Linu, Geetha, hariprasad, and Subbaraya are active,
though the review coverage isn't great (38%).
Jakub Kicinski [Tue, 3 Mar 2026 21:53:02 +0000 (13:53 -0800)]
MAINTAINERS: remove Manish Chopra from QLogic QL4xxx (now orphan)
We have not seen tags from Manish for the QL4xxx driver in over 5 years,
and there is no mailing list activity since Oct 2023. There has been
no maintainer activity in this subsystem at all.
Since there is no other maintainer for this driver it becomes an Orphan.
Jakub Kicinski [Tue, 3 Mar 2026 21:53:01 +0000 (13:53 -0800)]
MAINTAINERS: remove Johan Hedberg from Bluetooth subsystem
We have not seen emails or tags from Johan in over 5 years,
and there is no recent mailing list activity.
Marcel Holtmann hasn't provided any tags in the Bluetooth
subsystem in over 5 years, but he is active on the Bluetooth
mailing list, providing informal review.
Luiz Augusto von Dentz is very active, handling essentially
all commits and reviews (12% coverage, but Luiz is the sole
active committer).
Subsystem BLUETOOTH SUBSYSTEM
Changes 50 / 411 (12%)
Last activity: 2026-02-23
Marcel Holtmann <marcel@holtmann.org>:
Johan Hedberg <johan.hedberg@gmail.com>:
Luiz Augusto von Dentz <luiz.dentz@gmail.com>:
Author 138d7eca445e 2026-02-23 00:00:00 164
Committer 138d7eca445e 2026-02-23 00:00:00 361
Tags 138d7eca445e 2026-02-23 00:00:00 362
Top reviewers:
[15]: pmenzel@molgen.mpg.de
[8]: keescook@chromium.org
[5]: willemb@google.com
[4]: horms@kernel.org
[3]: kuniyu@amazon.com
[3]: luiz.von.dentz@intel.com
INACTIVE MAINTAINER Johan Hedberg <johan.hedberg@gmail.com>
Sun Jian [Wed, 25 Feb 2026 11:14:51 +0000 (19:14 +0800)]
selftests: net: tun: don't abort XFAIL cases
The tun UDP tunnel GSO fixture contains XFAIL-marked variants intended to
exercise failure paths (e.g. EMSGSIZE / "Message too long").
Using ASSERT_EQ() in these tests aborts the subtest, which prevents the
harness from classifying them as XFAIL and can make the overall net: tun
test fail.
Switch the relevant ASSERT_EQ() checks to EXPECT_EQ() so the subtests
continue running and the failures are correctly reported and accounted
as XFAIL where applicable.
Sun Jian [Wed, 25 Feb 2026 11:14:50 +0000 (19:14 +0800)]
selftests/harness: order TEST_F and XFAIL_ADD constructors
TEST_F() allocates and registers its struct __test_metadata via mmap()
inside its constructor, and only then assigns the
_##fixture_##test##_object pointer.
XFAIL_ADD() runs in a constructor too and reads
_##fixture_##test##_object to initialize xfail->test. If XFAIL_ADD runs
first, xfail->test can be NULL and the expected failure will be reported
as FAIL.
Use constructor priorities to ensure TEST_F registration runs before
XFAIL_ADD, without adding extra state or runtime lookups.
Jakub Kicinski [Thu, 5 Mar 2026 15:33:25 +0000 (07:33 -0800)]
Merge tag 'nf-26-03-05' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Florian Westphal says:
====================
netfilter: updates for net
1) Inseo An reported a bug with the set element handling in nf_tables:
When set cannot accept more elements, we unlink and immediately free
an element that was inserted into a public data structure, freeing it
without waiting for RCU grace period. Fix this by doing the
increment earlier and by deferring possible unlink-and-free to the
existing abort path, which performs the needed synchronize_rcu before
free. From Pablo Neira Ayuso. This is an ancient bug, dating back to
kernel 4.10.
2) syzbot reported WARN_ON() splat in nf_tables that occurs on memory
allocation failure. Fix this by a new iterator annotation:
The affected walker does not need to clone the data structure and
can just use the live version if no clone exists yet.
Also from Pablo. This bug existed since 6.10 days.
3) Ancient forever bug in nft_pipapo data structure:
The garbage collection logic to remove expired elements is broken.
We must unlink from data structure and can only hand the freeing
to call_rcu after the clone/live pointers of the data structures
have been swapped. Else, readers can observe the free'd element.
Reported by Yiming Qian.
* tag 'nf-26-03-05' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
netfilter: nft_set_pipapo: split gc into unlink and reclaim phase
netfilter: nf_tables: clone set on flush only
netfilter: nf_tables: unconditionally bump set->nelems before insertion
====================
Jerome Marchand [Thu, 5 Mar 2026 09:31:17 +0000 (10:31 +0100)]
ftrace: Add MAINTAINERS entries for all ftrace headers
There is currently no entry for ftrace_irq.h and ftrace_regs.h. Add a
generic entry for all *ftrace* headers to include them and prevent
overlooking future ftrace headers.
netfilter: nft_set_pipapo: split gc into unlink and reclaim phase
Yiming Qian reports Use-after-free in the pipapo set type:
Under a large number of expired elements, commit-time GC can run for a very
long time in a non-preemptible context, triggering soft lockup warnings and
RCU stall reports (local denial of service).
We must split GC in an unlink and a reclaim phase.
We cannot queue elements for freeing until pointers have been swapped.
Expired elements are still exposed to both the packet path and userspace
dumpers via the live copy of the data structure.
call_rcu() does not protect us: dump operations or element lookups starting
after call_rcu has fired can still observe the free'd element, unless the
commit phase has made enough progress to swap the clone and live pointers
before any new reader has picked up the old version.
This a similar approach as done recently for the rbtree backend in commit 35f83a75529a ("netfilter: nft_set_rbtree: don't gc elements on insert").
Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Reported-by: Yiming Qian <yimingqian591@gmail.com> Signed-off-by: Florian Westphal <fw@strlen.de>
Restrict set clone to the flush set command in the preparation phase.
Add NFT_ITER_UPDATE_CLONE and use it for this purpose, update the rbtree
and pipapo backends to only clone the set when this iteration type is
used.
As for the existing NFT_ITER_UPDATE type, update the pipapo backend to
use the existing set clone if available, otherwise use the existing set
representation. After this update, there is no need to clone a set that
is being deleted, this includes bound anonymous set.
An alternative approach to NFT_ITER_UPDATE_CLONE is to add a .clone
interface and call it from the flush set path.
Reported-by: syzbot+4924a0edc148e8b4b342@syzkaller.appspotmail.com Fixes: 3f1d886cc7c3 ("netfilter: nft_set_pipapo: move cloning of match info to insert/removal path") Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Florian Westphal <fw@strlen.de>
netfilter: nf_tables: unconditionally bump set->nelems before insertion
In case that the set is full, a new element gets published then removed
without waiting for the RCU grace period, while RCU reader can be
walking over it already.
To address this issue, add the element transaction even if set is full,
but toggle the set_full flag to report -ENFILE so the abort path safely
unwinds the set to its previous state.
As for element updates, decrement set->nelems to restore it.
A simpler fix is to call synchronize_rcu() in the error path.
However, with a large batch adding elements to already maxed-out set,
this could cause noticeable slowdown of such batches.
Fixes: 35d0ac9070ef ("netfilter: nf_tables: fix set->nelems counting with no NLM_F_EXCL") Reported-by: Inseo An <y0un9sa@gmail.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org> Signed-off-by: Florian Westphal <fw@strlen.de>
net: Provide a PREEMPT_RT specific check for netdev_queue::_xmit_lock
After acquiring netdev_queue::_xmit_lock the number of the CPU owning
the lock is recorded in netdev_queue::xmit_lock_owner. This works as
long as the BH context is not preemptible.
On PREEMPT_RT the softirq context is preemptible and without the
softirq-lock it is possible to have multiple user in __dev_queue_xmit()
submitting a skb on the same CPU. This is fine in general but this means
also that the current CPU is recorded as netdev_queue::xmit_lock_owner.
This in turn leads to the recursion alert and the skb is dropped.
Instead checking the for CPU number, that owns the lock, PREEMPT_RT can
check if the lockowner matches the current task.
Add netif_tx_owned() which returns true if the current context owns the
lock by comparing the provided CPU number with the recorded number. This
resembles the current check by negating the condition (the current check
returns true if the lock is not owned).
On PREEMPT_RT use rt_mutex_owner() to return the lock owner and compare
the current task against it.
Use the new helper in __dev_queue_xmit() and netif_local_xmit_active()
which provides a similar check.
Update comments regarding pairing READ_ONCE().
Reported-by: Bert Karwatzki <spasswolf@web.de> Closes: https://lore.kernel.org/all/20260216134333.412332-1-spasswolf@web.de Fixes: 3253cb49cbad4 ("softirq: Allow to drop the softirq-BKL lock on PREEMPT_RT") Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Reported-by: Bert Karwatzki <spasswolf@web.de> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Link: https://patch.msgid.link/20260302162631.uGUyIqDT@linutronix.de Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
net: stmmac: Fix VLAN handling when interface is down
VLAN register accesses on the MAC side require the PHY RX clock to be
active. When the network interface is down, the PHY is suspended and
the RX clock is unavailable, causing VLAN operations to fail with
timeouts.
The VLAN core automatically removes VID 0 after the interface goes down
and re-adds it when it comes back up, so these timeouts happen during
normal interface down/up:
# ip link set end1 down
renesas-gbeth 15c40000.ethernet end1: Timeout accessing MAC_VLAN_Tag_Filter
renesas-gbeth 15c40000.ethernet end1: failed to kill vid 0081/0
Adding VLANs while the interface is down also fails:
# ip link add link end1 name end1.10 type vlan id 10
renesas-gbeth 15c40000.ethernet end1: Timeout accessing MAC_VLAN_Tag_Filter
RTNETLINK answers: Device or resource busy
Patch 4 fixes this by adding checks in the VLAN paths for netif_running(),
and skipping register accesses if the interface is down. Only the software
state is updated in this case. When the interface is brought up, the VLAN
state is restored to hardware.
Patches 1-3 fix some issues in the existing VLAN implementation.
====================
Ovidiu Panait [Tue, 3 Mar 2026 14:58:28 +0000 (14:58 +0000)]
net: stmmac: Defer VLAN HW configuration when interface is down
VLAN register accesses on the MAC side require the PHY RX clock to be
active. When the network interface is down, the PHY is suspended and
the RX clock is unavailable, causing VLAN operations to fail with
timeouts.
The VLAN core automatically removes VID 0 after the interface goes down
and re-adds it when it comes back up, so these timeouts happen during
normal interface down/up:
# ip link set end1 down
renesas-gbeth 15c40000.ethernet end1: Timeout accessing MAC_VLAN_Tag_Filter
renesas-gbeth 15c40000.ethernet end1: failed to kill vid 0081/0
Adding VLANs while the interface is down also fails:
# ip link add link end1 name end1.10 type vlan id 10
renesas-gbeth 15c40000.ethernet end1: Timeout accessing MAC_VLAN_Tag_Filter
RTNETLINK answers: Device or resource busy
To fix this, check if the interface is up before accessing VLAN registers.
The software state is always kept up to date regardless of interface state.
When the interface is brought up, stmmac_vlan_restore() is called
to write the VLAN state to hardware.
Ovidiu Panait [Tue, 3 Mar 2026 14:58:27 +0000 (14:58 +0000)]
net: stmmac: Fix VLAN HW state restore
When the network interface is opened or resumed, a DMA reset is performed,
which resets all hardware state, including VLAN state. Currently, only
the resume path is restoring the VLAN state via
stmmac_restore_hw_vlan_rx_fltr(), but that is incomplete: the VLAN hash
table and the VLAN_TAG control bits are not restored.
Therefore, add stmmac_vlan_restore(), which restores the full VLAN
state by updating both the HW filter entries and the hash table, and
call it from both the open and resume paths.
The VLAN restore is moved outside of phylink_rx_clk_stop_block/unblock
in the resume path because receive clock stop is already disabled when
stmmac supports VLAN.
Also, remove the hash readback code in vlan_restore_hw_rx_fltr() that
attempts to restore VTHM by reading VLAN_HASH_TABLE, as it always reads
zero after DMA reset, making it dead code.
Fixes: 3cd1cfcba26e ("net: stmmac: Implement VLAN Hash Filtering in XGMAC") Fixes: ed64639bc1e0 ("net: stmmac: Add support for VLAN Rx filtering") Signed-off-by: Ovidiu Panait <ovidiu.panait.rb@renesas.com> Link: https://patch.msgid.link/20260303145828.7845-4-ovidiu.panait.rb@renesas.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ovidiu Panait [Tue, 3 Mar 2026 14:58:26 +0000 (14:58 +0000)]
net: stmmac: Improve double VLAN handling
The double VLAN bits (EDVLP, ESVL, DOVLTC) are handled inconsistently
between the two vlan_update_hash() implementations:
- dwxgmac2_update_vlan_hash() explicitly clears the double VLAN bits when
is_double is false, meaning that adding a 802.1Q VLAN will disable
double VLAN mode:
$ ip link add link eth0 name eth0.200 type vlan id 200 protocol 802.1ad
$ ip link add link eth0 name eth0.100 type vlan id 100
# Double VLAN bits no longer set
- vlan_update_hash() sets these bits and only clears them when the last
VLAN has been removed, so double VLAN mode remains enabled even after all
802.1AD VLANs are removed.
Address both issues by tracking the number of active 802.1AD VLANs in
priv->num_double_vlans. Pass this count to stmmac_vlan_update() so both
implementations correctly set the double VLAN bits when any 802.1AD
VLAN is active, and clear them only when none remain.
Also update vlan_update_hash() to explicitly clear the double VLAN bits
when is_double is false, matching the dwxgmac2 behavior.
Ovidiu Panait [Tue, 3 Mar 2026 14:58:25 +0000 (14:58 +0000)]
net: stmmac: Fix error handling in VLAN add and delete paths
stmmac_vlan_rx_add_vid() updates active_vlans and the VLAN hash
register before writing the HW filter entry. If the filter write
fails, it leaves a stale VID in active_vlans and the hash register.
stmmac_vlan_rx_kill_vid() has the reverse problem: it clears
active_vlans before removing the HW filter. On failure, the VID is
gone from active_vlans but still present in the HW filter table.
To fix this, reorder the operations to update the hash table first,
then attempt the HW filter operation. If the HW filter fails, roll
back both the active_vlans bitmap and the hash table by calling
stmmac_vlan_update() again.
Larysa removes VF restriction for LLDP filters on ice to allow for LLDP
traffic to reach the correct destination.
Jakub adds retry mechanism for AdminQ Read/Write SFF EEPROM call to
follow hardware specification on ice.
Zilin Guan adds cleanup path to free XDP rings on failure in
ice_set_ringparam().
Michal bypasses firmware logging unroll in libie when it isn't supported.
Kohei Enju fixes iavf to take into account hardware MTU support when
setting max MTU values.
Vivek Behera fixes issues on igb and igc using incorrect IRQs when Tx/Rx
queues do not share the same IRQ.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
igc: Fix trigger of incorrect irq in igc_xsk_wakeup function
igb: Fix trigger of incorrect irq in igb_xsk_wakeup
iavf: fix netdev->max_mtu to respect actual hardware limit
libie: don't unroll if fwlog isn't supported
ice: Fix memory leak in ice_set_ringparam()
ice: fix retry for AQ command 0x06EE
ice: reintroduce retry mechanism for indirect AQ
ice: fix adding AQ LLDP filter for VF
====================
Jakub Kicinski [Thu, 5 Mar 2026 02:21:15 +0000 (18:21 -0800)]
Merge branch 'mptcp-misc-fixes-for-v7-0-rc2'
Matthieu Baerts says:
====================
mptcp: misc fixes for v7.0-rc2
Here are various unrelated fixes:
- Patch 1: avoid bufferbloat in simult_flows selftest which can cause
instabilities. A fix for v5.10.
- Patches 2-3: reduce RM_ADDR lost by not sending it over the same
subflow as the one being removed, if possible. A fix for v5.13.
- Patches 4-5: avoid a WARN when using signal + subflow endpoints with a
subflow limit of 0, and removing such endpoints during an active
connection. A fix for v5.17.
====================
This validates the previous commit: endpoints with both the signal and
subflow flags should always be marked as used even if it was not
possible to create new subflows due to the MPTCP PM limits.
For this test, an extra endpoint is created with both the signal and the
subflow flags, and limits are set not to create extra subflows. In this
case, an ADD_ADDR is sent, but no subflows are created. Still, the local
endpoint is marked as used, and no warning is fired when removing the
endpoint, after having sent a RM_ADDR.
The 'Fixes' tag here below is the same as the one from the previous
commit: this patch here is not fixing anything wrong in the selftests,
but it validates the previous fix for an issue introduced by this commit
ID.
Fixes: 85df533a787b ("mptcp: pm: do not ignore 'subflow' if 'signal' flag is also set") Cc: stable@vger.kernel.org Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20260303-net-mptcp-misc-fixes-7-0-rc2-v1-5-4b5462b6f016@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
- Set the MPTCP subflows limit to 0
- Create an MPTCP endpoint with both the 'signal' and 'subflow' flags
- Create a new MPTCP connection from a different address: an ADD_ADDR
linked to the MPTCP endpoint will be sent ('signal' flag), but no
subflows is initiated ('subflow' flag)
- Remove the MPTCP endpoint
In this case, msk->pm.local_addr_used has been kept to 0 -- because no
subflows have been created -- but the corresponding bit in
msk->pm.id_avail_bitmap has been cleared when the ADD_ADDR has been
sent. This later causes a splat when removing the MPTCP endpoint because
msk->pm.local_addr_used has been kept to 0.
Now, if an endpoint has both the signal and subflow flags, but it is not
possible to create subflows because of the limits or the c-flag case,
then the local endpoint counter is still incremented: the endpoint is
used at the end. This avoids issues later when removing the endpoint and
calling __mark_subflow_endp_available(), which expects
msk->pm.local_addr_used to have been previously incremented if the
endpoint was marked as used according to msk->pm.id_avail_bitmap.
Note that signal_and_subflow variable is reset to false when the limits
and the c-flag case allows subflows creation. Also, local_addr_used is
only incremented for non ID0 subflows.
Fixes: 85df533a787b ("mptcp: pm: do not ignore 'subflow' if 'signal' flag is also set") Cc: stable@vger.kernel.org Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/613 Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20260303-net-mptcp-misc-fixes-7-0-rc2-v1-4-4b5462b6f016@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
selftests: mptcp: join: check RM_ADDR not sent over same subflow
This validates the previous commit: RM_ADDR were sent over the first
found active subflow which could be the same as the one being removed.
It is more likely to loose this notification.
For this check, RM_ADDR are explicitly dropped when trying to send them
over the initial subflow, when removing the endpoint attached to it. If
it is dropped, the test will complain because some RM_ADDR have not been
received.
Note that only the RM_ADDR are dropped, to allow the linked subflow to
be quickly and cleanly closed. To only drop those RM_ADDR, a cBPF byte
code is used. If the IPTables commands fail, that's OK, the tests will
continue to pass, but not validate this part. This can be ignored:
another subtest fully depends on such command, and will be marked as
skipped.
The 'Fixes' tag here below is the same as the one from the previous
commit: this patch here is not fixing anything wrong in the selftests,
but it validates the previous fix for an issue introduced by this commit
ID.
mptcp: pm: avoid sending RM_ADDR over same subflow
RM_ADDR are sent over an active subflow, the first one in the subflows
list. There is then a high chance the initial subflow is picked. With
the in-kernel PM, when an endpoint is removed, a RM_ADDR is sent, then
linked subflows are closed. This is done for each active MPTCP
connection.
MPTCP endpoints are likely removed because the attached network is no
longer available or usable. In this case, it is better to avoid sending
this RM_ADDR over the subflow that is going to be removed, but prefer
sending it over another active and non stale subflow, if any.
This modification avoids situations where the other end is not notified
when a subflow is no longer usable: typically when the endpoint linked
to the initial subflow is removed, especially on the server side.
Fixes: 8dd5efb1f91b ("mptcp: send ack for rm_addr") Cc: stable@vger.kernel.org Reported-by: Frank Lorenz <lorenz-frank@web.de> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/612 Reviewed-by: Mat Martineau <martineau@kernel.org> Signed-off-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20260303-net-mptcp-misc-fixes-7-0-rc2-v1-2-4b5462b6f016@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Paolo Abeni [Tue, 3 Mar 2026 10:56:02 +0000 (11:56 +0100)]
selftests: mptcp: more stable simult_flows tests
By default, the netem qdisc can keep up to 1000 packets under its belly
to deal with the configured rate and delay. The simult flows test-case
simulates very low speed links, to avoid problems due to slow CPUs and
the TCP stack tend to transmit at a slightly higher rate than the
(virtual) link constraints.
All the above causes a relatively large amount of packets being enqueued
in the netem qdiscs - the longer the transfer, the longer the queue -
producing increasingly high TCP RTT samples and consequently increasingly
larger receive buffer size due to DRS.
When the receive buffer size becomes considerably larger than the needed
size, the tests results can flake, i.e. because minimal inaccuracy in the
pacing rate can lead to a single subflow usage towards the end of the
connection for a considerable amount of data.
Address the issue explicitly setting netem limits suitable for the
configured link speeds and unflake all the affected tests.
====================
nfc: fix leaks and races surfaced by NIPA
I recently added the nci test to NIPA. Somewhat surprisingly it runs
without much settup but hits kmemleaks fairly often. Fix a handful of
issues to make the test pass in a stable way.
====================
Jakub Kicinski [Tue, 3 Mar 2026 16:23:45 +0000 (08:23 -0800)]
nfc: rawsock: cancel tx_work before socket teardown
In rawsock_release(), cancel any pending tx_work and purge the write
queue before orphaning the socket. rawsock_tx_work runs on the system
workqueue and calls nfc_data_exchange which dereferences the NCI
device. Without synchronization, tx_work can race with socket and
device teardown when a process is killed (e.g. by SIGKILL), leading
to use-after-free or leaked references.
Set SEND_SHUTDOWN first so that if tx_work is already running it will
see the flag and skip transmitting, then use cancel_work_sync to wait
for any in-progress execution to finish, and finally purge any
remaining queued skbs.
Jakub Kicinski [Tue, 3 Mar 2026 16:23:44 +0000 (08:23 -0800)]
nfc: nci: clear NCI_DATA_EXCHANGE before calling completion callback
Move clear_bit(NCI_DATA_EXCHANGE) before invoking the data exchange
callback in nci_data_exchange_complete().
The callback (e.g. rawsock_data_exchange_complete) may immediately
schedule another data exchange via schedule_work(tx_work). On a
multi-CPU system, tx_work can run and reach nci_transceive() before
the current nci_data_exchange_complete() clears the flag, causing
test_and_set_bit(NCI_DATA_EXCHANGE) to return -EBUSY and the new
transfer to fail.
This causes intermittent flakes in nci/nci_dev in NIPA:
# # RUN NCI.NCI1_0.t4t_tag_read ...
# # t4t_tag_read: Test terminated by timeout
# # FAIL NCI.NCI1_0.t4t_tag_read
# not ok 3 NCI.NCI1_0.t4t_tag_read
Jakub Kicinski [Tue, 3 Mar 2026 16:23:43 +0000 (08:23 -0800)]
nfc: nci: complete pending data exchange on device close
In nci_close_device(), complete any pending data exchange before
closing. The data exchange callback (e.g.
rawsock_data_exchange_complete) holds a socket reference.
Jakub Kicinski [Tue, 3 Mar 2026 16:23:41 +0000 (08:23 -0800)]
nfc: nci: free skb on nci_transceive early error paths
nci_transceive() takes ownership of the skb passed by the caller,
but the -EPROTO, -EINVAL, and -EBUSY error paths return without
freeing it.
Due to issues clearing NCI_DATA_EXCHANGE fixed by subsequent changes
the nci/nci_dev selftest hits the error path occasionally in NIPA,
and kmemleak detects leaks:
Bobby Eshleman [Tue, 3 Mar 2026 00:32:56 +0000 (16:32 -0800)]
net: devmem: use READ_ONCE/WRITE_ONCE on binding->dev
binding->dev is protected on the write-side in
mp_dmabuf_devmem_uninstall() against concurrent writes, but due to the
concurrent bare reads in net_devmem_get_binding() and
validate_xmit_unreadable_skb() it should be wrapped in a
READ_ONCE/WRITE_ONCE pair to make sure no compiler optimizations play
with the underlying register in unforeseen ways.
Doesn't present a critical bug because the known compiler optimizations
don't result in bad behavior. There is no tearing on u64, and load
omissions/invented loads would only break if additional binding->dev
references were inlined together (they aren't right now).
This just more strictly follows the linux memory model (i.e.,
"Lock-Protected Writes With Lockless Reads" in
tools/memory-model/Documentation/access-marking.txt).
Koichiro Den [Sat, 28 Feb 2026 14:53:07 +0000 (23:53 +0900)]
net: sched: avoid qdisc_reset_all_tx_gt() vs dequeue race for lockless qdiscs
When shrinking the number of real tx queues,
netif_set_real_num_tx_queues() calls qdisc_reset_all_tx_gt() to flush
qdiscs for queues which will no longer be used.
qdisc_reset_all_tx_gt() currently serializes qdisc_reset() with
qdisc_lock(). However, for lockless qdiscs, the dequeue path is
serialized by qdisc_run_begin/end() using qdisc->seqlock instead, so
qdisc_reset() can run concurrently with __qdisc_run() and free skbs
while they are still being dequeued, leading to UAF.
This can easily be reproduced on e.g. virtio-net by imposing heavy
traffic while frequently changing the number of queue pairs:
Allocated by task 1270 on cpu 5 at 44.558414s:
...
alloc_skb_with_frags+0x84/0x7c0
sock_alloc_send_pskb+0x69a/0x830
__ip_append_data+0x1b86/0x48c0
ip_make_skb+0x1e8/0x2b0
udp_sendmsg+0x13a6/0x1fc0
...
Freed by task 1306 on cpu 3 at 44.558445s:
...
kmem_cache_free+0x117/0x5e0
pfifo_fast_reset+0x14d/0x580
qdisc_reset+0x9e/0x5f0
netif_set_real_num_tx_queues+0x303/0x840
virtnet_set_channels+0x1bf/0x260 [virtio_net]
ethnl_set_channels+0x684/0xae0
ethnl_default_set_doit+0x31a/0x890
...
Serialize qdisc_reset_all_tx_gt() against the lockless dequeue path by
taking qdisc->seqlock for TCQ_F_NOLOCK qdiscs, matching the
serialization model already used by dev_reset_queue().
Additionally clear QDISC_STATE_NON_EMPTY after reset so the qdisc state
reflects an empty queue, avoiding needless re-scheduling.
Jakub Kicinski [Thu, 5 Mar 2026 00:47:45 +0000 (16:47 -0800)]
Merge tag 'linux-can-fixes-for-7.0-20260302' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can
Marc Kleine-Budde says:
====================
pull-request: can 2026-03-02
The first 2 patches are by Oliver Hartkopp. The first fixes the
locking for CAN Broadcast Manager op runtime updates, the second fixes
the packet statisctics for the CAN dummy driver.
Alban Bedel's patch fixes a potential problem in the error path of the
mcp251x's ndo_open callback.
A patch by Ziyi Guo add USB endpoint type validation to the esd_usb
driver.
The next 6 patches are by Greg Kroah-Hartman and fix URB data parsing
for the ems_usb and ucan driver, fix URB anchoring in the etas_es58x,
and in the f81604 driver fix URB data parsing, add URB error handling
and fix URB anchoring.
A patch by me targets the gs_usb driver and fixes interoperability
with the CANable-2.5 firmware by always configuring the bit rate
before starting the device.
The last patch is by Frank Li and fixes a CHECK_DTBS warning for the
nxp,sja1000 dt-binding.
* tag 'linux-can-fixes-for-7.0-20260302' of git://git.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can:
dt-bindings: net: can: nxp,sja1000: add reference to mc-peripheral-props.yaml
can: gs_usb: gs_can_open(): always configure bitrates before starting device
can: usb: f81604: correctly anchor the urb in the read bulk callback
can: usb: f81604: handle bulk write errors properly
can: usb: f81604: handle short interrupt urb messages properly
can: usb: etas_es58x: correctly anchor the urb in the read bulk callback
can: ucan: Fix infinite loop from zero-length messages
can: ems_usb: ems_usb_read_bulk_callback(): check the proper length of a message
can: esd_usb: add endpoint type validation
can: mcp251x: fix deadlock in error path of mcp251x_open
can: dummy_can: dummy_can_init(): fix packet statistics
can: bcm: fix locking for bcm_op runtime updates
====================
Linus Torvalds [Wed, 4 Mar 2026 23:42:24 +0000 (15:42 -0800)]
Merge tag 'modules-7.0-rc3.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/modules/linux
Pull module fixes from Sami Tolvanen:
- Fix a potential kernel panic in the module loader by adding a bounds
check for the ELF section index. This prevents crashes if attempting
to load a module that uses SHN_XINDEX or is corrupted.
- Fix the Kconfig menu layout for module versioning, signing, and
compression options so they correctly appear as submenus in
menuconfig.
- Remove a redundant lockdep_free_key_range() call in the load_module()
error path. This is already handled by module_deallocate() calling
free_mod_mem() since the module_memory rework.
* tag 'modules-7.0-rc3.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/modules/linux:
module: Fix kernel panic when a symbol st_shndx is out of bounds
module: Fix the modversions and signing submenus
module: Remove duplicate freeing of lockdep classes
Jakub Kicinski [Wed, 4 Mar 2026 23:29:56 +0000 (15:29 -0800)]
Merge tag 'wireless-2026-03-04' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless
Johannes Berg says:
====================
Some more fixes:
- mt76 gets three almost identical new length checks
- cw1200 & ti: locking fixes
- mac80211 has a fix for the recent EML frame handling
- rsi driver no longer oddly responds to config, which
had triggered a warning in mac80211
- ath12k has two fixes for station statistics handling
* tag 'wireless-2026-03-04' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
wifi: mt76: Fix possible oob access in mt76_connac2_mac_write_txwi_80211()
wifi: mt76: mt7925: Fix possible oob access in mt7925_mac_write_txwi_80211()
wifi: mt76: mt7996: Fix possible oob access in mt7996_mac_write_txwi_80211()
wifi: wlcore: Fix a locking bug
wifi: cw1200: Fix locking in error paths
wifi: mac80211: fix missing ieee80211_eml_params member initialization
wifi: rsi: Don't default to -EOPNOTSUPP in rsi_mac80211_config
wifi: ath12k: fix station lookup failure when disconnecting from AP
wifi: ath12k: use correct pdev id when requesting firmware stats
====================
Linus Torvalds [Wed, 4 Mar 2026 16:21:11 +0000 (08:21 -0800)]
Merge tag 'sysctl-7.00-fixes-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/sysctl/sysctl
Pull sysctl fix from Joel Granados:
- Fix error when reporting jiffies converted values back to user space
Return the converted value instead of "Invalid argument" error
* tag 'sysctl-7.00-fixes-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/sysctl/sysctl:
time/jiffies: Fix sysctl file error on configurations where USER_HZ < HZ
Linus Torvalds [Wed, 4 Mar 2026 16:12:06 +0000 (08:12 -0800)]
Merge tag 'media/v7.0-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media
Pull media fix from Mauro Carvalho Chehab:
"Fix for MPEG-TS decoder in dvb-net"
* tag 'media/v7.0-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
media: dvb-net: fix OOB access in ULE extension header tables
Linus Torvalds [Wed, 4 Mar 2026 16:03:43 +0000 (08:03 -0800)]
Merge tag 'pinctrl-v7.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl
Pull pin control fixes from Linus Walleij:
"All of these are driver fixes except a memory leak in the
pinconf_generic_parse_dt_config() helper which is the most
important fix.
- Rename and fix up the Intel Equilibrium immutable interrupt chip
- Handle the Qualcomm QCS615 dual edge GPIO IRQ by adding the right
flag
- Fix a memory leak in the widely used pinconf_generic_parse_dt_config()
and a more local leak in aml_dt_node_to_map_pinmux()
- Fix double put in the Cirrus cs42l43_pin_probe()
- Staticize amdisp_pinctrl_ops, Qualcomm SDM660 groups and functions
- Unexport CIX sky1_pinctrl_pm_ops
- Fix configuration of deferred pin in the Rockchip driver
- Implement .get_direction() in the Sunxi driver squelching a dmesg
warning message
- Fix a readout of the last bank of registers in the Cypress CY8C95x0
driver"
* tag 'pinctrl-v7.0-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
pinctrl: cy8c95x0: Don't miss reading the last bank registers
pinctrl: sunxi: Implement gpiochip::get_direction()
pinctrl: rockchip: Fix configuring a deferred pin
pinctrl: cirrus: cs42l43: Fix double-put in cs42l43_pin_probe()
pinctrl: meson: amlogic-a4: Fix device node reference leak in aml_dt_node_to_map_pinmux()
pinctrl: qcom: sdm660-lpass-lpi: Make groups and functions variables static
pinctrl: cix: sky1: Unexport sky1_pinctrl_pm_ops
pinctrl: amdisp: Make amdisp_pinctrl_ops variable static
pinctrl: pinconf-generic: Fix memory leak in pinconf_generic_parse_dt_config()
pinctrl: qcom: qcs615: Add missing dual edge GPIO IRQ errata flag
pinctrl: equilibrium: fix warning trace on load
pinctrl: equilibrium: rename irq_chip function callbacks
Darrick J. Wong [Mon, 2 Mar 2026 17:30:02 +0000 (09:30 -0800)]
iomap: reject delalloc mappings during writeback
Filesystems should never provide a delayed allocation mapping to
writeback; they're supposed to allocate the space before replying.
This can lead to weird IO errors and crashes in the block layer if the
filesystem is being malicious, or if it hadn't set iomap->dev because
it's a delalloc mapping.
Fix this by failing writeback on delalloc mappings. Currently no
filesystems actually misbehave in this manner, but we ought to be
stricter about things like that.
Cc: stable@vger.kernel.org # v5.5 Fixes: 598ecfbaa742ac ("iomap: lift the xfs writeback code to iomap") Signed-off-by: Darrick J. Wong <djwong@kernel.org> Link: https://patch.msgid.link/20260302173002.GL13829@frogsfrogsfrogs Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com> Signed-off-by: Christian Brauner <brauner@kernel.org>
Merge patch "iomap: don't mark folio uptodate if read IO has bytes pending"
Joanne Koong <joannelkoong@gmail.com> says:
This is a fix for this scenario:
->read_folio() gets called on a folio size that is 16k while the file is 4k:
a) ifs->read_bytes_pending gets initialized to 16k
b) ->read_folio_range() is called for the 4k read
c) the 4k read succeeds, ifs->read_bytes_pending is now 12k and the
0 to 4k range is marked uptodate
d) the post-eof blocks are zeroed and marked uptodate in the call to
iomap_set_range_uptodate()
e) iomap_set_range_uptodate() sees all the ranges are marked
uptodate and it marks the folio uptodate
f) iomap_read_end() gets called to subtract the 12k from
ifs->read_bytes_pending. it too sees all the ranges are marked
uptodate and marks the folio uptodate using XOR
g) the XOR call clears the uptodate flag on the folio
The same situation can occur if the last range read for the folio is done as
an inline read and all the previous ranges have already completed by the time
the inline read completes.
For more context, the full discussion can be found in [1]. There was a
discussion about alternative approaches in that thread, but they had more
complications.
There is another discussion in v1 [2] about consolidating the read paths.
Until that is resolved, this patch fixes the issue.
Joanne Koong [Tue, 3 Mar 2026 23:34:20 +0000 (15:34 -0800)]
iomap: don't mark folio uptodate if read IO has bytes pending
If a folio has ifs metadata attached to it and the folio is partially
read in through an async IO helper with the rest of it then being read
in through post-EOF zeroing or as inline data, and the helper
successfully finishes the read first, then post-EOF zeroing / reading
inline will mark the folio as uptodate in iomap_set_range_uptodate().
This is a problem because when the read completion path later calls
iomap_read_end(), it will call folio_end_read(), which sets the uptodate
bit using XOR semantics. Calling folio_end_read() on a folio that was
already marked uptodate clears the uptodate bit.
Fix this by not marking the folio as uptodate if the read IO has bytes
pending. The folio uptodate state will be set in the read completion
path through iomap_end_read() -> folio_end_read().
Gerd Rausch [Wed, 25 Feb 2026 23:37:49 +0000 (15:37 -0800)]
time/jiffies: Fix sysctl file error on configurations where USER_HZ < HZ
Commit 2dc164a48e6fd ("sysctl: Create converter functions with two new
macros") incorrectly returns error to user space when jiffies sysctl
converter is used. The old overflow check got replaced with an
unconditional one:
+ if (USER_HZ < HZ)
+ return -EINVAL;
which will always be true on configurations with "USER_HZ < HZ".
Remove the check; it is no longer needed as clock_t_to_jiffies() returns
ULONG_MAX for the overflow case and proc_int_u2k_conv_uop() checks for
"> INT_MAX" after conversion
Fixes: 2dc164a48e6fd ("sysctl: Create converter functions with two new macros") Reported-by: Colm Harrington <colm.harrington@oracle.com> Signed-off-by: Gerd Rausch <gerd.rausch@oracle.com> Signed-off-by: Joel Granados <joel.granados@kernel.org>
Naveen Anandhan [Sat, 28 Feb 2026 07:47:35 +0000 (13:17 +0530)]
selftests: tc-testing: fix list_categories() crash on list type
list_categories() builds a set directly from the 'category'
field of each test case. Since 'category' is a list,
set(map(...)) attempts to insert lists into a set, which
raises:
TypeError: unhashable type: 'list'
Flatten category lists and collect unique category names
using set.update() instead.
Signed-off-by: Naveen Anandhan <mr.navi8680@gmail.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Qing Wang [Fri, 27 Feb 2026 02:58:42 +0000 (10:58 +0800)]
tracing: Fix WARN_ON in tracing_buffers_mmap_close
When a process forks, the child process copies the parent's VMAs but the
user_mapped reference count is not incremented. As a result, when both the
parent and child processes exit, tracing_buffers_mmap_close() is called
twice. On the second call, user_mapped is already 0, causing the function to
return -ENODEV and triggering a WARN_ON.
Normally, this isn't an issue as the memory is mapped with VM_DONTCOPY set.
But this is only a hint, and the application can call
madvise(MADVISE_DOFORK) which resets the VM_DONTCOPY flag. When the
application does that, it can trigger this issue on fork.
Fix it by incrementing the user_mapped reference count without re-mapping
the pages in the VMA's open callback.
Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Cc: Vincent Donnefort <vdonnefort@google.com> Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com> Link: https://patch.msgid.link/20260227025842.1085206-1-wangqing7171@gmail.com Fixes: cf9f0f7c4c5bb ("tracing: Allow user-space mapping of the ring-buffer") Reported-by: syzbot+3b5dd2030fe08afdf65d@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=3b5dd2030fe08afdf65d Tested-by: syzbot+3b5dd2030fe08afdf65d@syzkaller.appspotmail.com Signed-off-by: Qing Wang <wangqing7171@gmail.com> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The tracepoint callbacks recently were changed to allow preemption. The
event PID filtering callbacks that were attached to the fork and exit
tracepoints expected preemption disabled in order to access the RCU
protected PID lists.
Add a guard(preempt)() to protect the references to the PID list.
Steven Rostedt [Tue, 3 Mar 2026 02:35:46 +0000 (21:35 -0500)]
ftrace: Disable preemption in the tracepoint callbacks handling filtered pids
When function trace PID filtering is enabled, the function tracer will
attach a callback to the fork tracepoint as well as the exit tracepoint
that will add the forked child PID to the PID filtering list as well as
remove the PID that is exiting.
Commit a46023d5616e ("tracing: Guard __DECLARE_TRACE() use of
__DO_TRACE_CALL() with SRCU-fast") removed the disabling of preemption
when calling tracepoint callbacks.
The callbacks used for the PID filtering accounting depended on preemption
being disabled, and now the trigger a "suspicious RCU usage" warning message.
Make them explicitly disable preemption.
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Link: https://patch.msgid.link/20260302213546.156e3e4f@gandalf.local.home Fixes: a46023d5616e ("tracing: Guard __DECLARE_TRACE() use of __DO_TRACE_CALL() with SRCU-fast") Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org> Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Huiwen He [Tue, 24 Feb 2026 02:35:44 +0000 (10:35 +0800)]
tracing: Fix syscall events activation by ensuring refcount hits zero
When multiple syscall events are specified in the kernel command line
(e.g., trace_event=syscalls:sys_enter_openat,syscalls:sys_enter_close),
they are often not captured after boot, even though they appear enabled
in the tracing/set_event file.
The issue stems from how syscall events are initialized. Syscall
tracepoints require the global reference count (sys_tracepoint_refcount)
to transition from 0 to 1 to trigger the registration of the syscall
work (TIF_SYSCALL_TRACEPOINT) for tasks, including the init process (pid 1).
The current implementation of early_enable_events() with disable_first=true
used an interleaved sequence of "Disable A -> Enable A -> Disable B -> Enable B".
If multiple syscalls are enabled, the refcount never drops to zero,
preventing the 0->1 transition that triggers actual registration.
Fix this by splitting early_enable_events() into two distinct phases:
1. Disable all events specified in the buffer.
2. Enable all events specified in the buffer.
This ensures the refcount hits zero before re-enabling, allowing syscall
events to be properly activated during early boot.
The code is also refactored to use a helper function to avoid logic
duplication between the disable and enable phases.
Cc: stable@vger.kernel.org Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> Link: https://patch.msgid.link/20260224023544.1250787-1-hehuiwen@kylinos.cn Fixes: ce1039bd3a89 ("tracing: Fix enabling of syscall events on the command line") Signed-off-by: Huiwen He <hehuiwen@kylinos.cn> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
trace_graph_thresh_return() called handle_nosleeptime() and then delegated
to trace_graph_return(), which calls handle_nosleeptime() again. When
sleep-time accounting is disabled this double-adjusts calltime and can
produce bogus durations (including underflow).
Fix this by computing rettime once, applying handle_nosleeptime() only
once, using the adjusted calltime for threshold comparison, and writing
the return event directly via __trace_graph_return() when the threshold is
met.
Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260221113314048jE4VRwIyZEALiYByGK0My@zte.com.cn Fixes: 3c9880f3ab52b ("ftrace: Use a running sleeptime instead of saving on shadow stack") Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Shengming Hu [Sat, 21 Feb 2026 03:30:07 +0000 (11:30 +0800)]
fgraph: Fix thresh_return clear per-task notrace
When tracing_thresh is enabled, function graph tracing uses
trace_graph_thresh_return() as the return handler. Unlike
trace_graph_return(), it did not clear the per-task TRACE_GRAPH_NOTRACE
flag set by the entry handler for set_graph_notrace addresses. This could
leave the task permanently in "notrace" state and effectively disable
function graph tracing for that task.
Mirror trace_graph_return()'s per-task notrace handling by clearing
TRACE_GRAPH_NOTRACE and returning early when set.
Cc: stable@vger.kernel.org Link: https://patch.msgid.link/20260221113007819YgrZsMGABff4Rc-O_fZxL@zte.com.cn Fixes: b84214890a9bc ("function_graph: Move graph notrace bit to shadow stack global var") Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org> Signed-off-by: Shengming Hu <hu.shengming@zte.com.cn> Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Mark the skge and sky2 drivers as orphan.
I no longer have any Marvell/SysKonnect boards to test with and
mail to Mirko Lindner bounced because Marvell sold off that divsion.
Raju Rangoju [Mon, 2 Mar 2026 04:21:24 +0000 (09:51 +0530)]
amd-xgbe: fix sleep while atomic on suspend/resume
The xgbe_powerdown() and xgbe_powerup() functions use spinlocks
(spin_lock_irqsave) while calling functions that may sleep:
- napi_disable() can sleep waiting for NAPI polling to complete
- flush_workqueue() can sleep waiting for pending work items
This causes a "BUG: scheduling while atomic" error during suspend/resume
cycles on systems using the AMD XGBE Ethernet controller.
The spinlock protection in these functions is unnecessary as these
functions are called from suspend/resume paths which are already serialized
by the PM core
Fix this by removing the spinlock. Since only code that takes this lock
is xgbe_powerdown() and xgbe_powerup(), remove it completely.
Breno Leitao [Mon, 2 Mar 2026 11:40:46 +0000 (03:40 -0800)]
netconsole: fix sysdata_release_enabled_show checking wrong flag
sysdata_release_enabled_show() checks SYSDATA_TASKNAME instead of
SYSDATA_RELEASE, causing the configfs release_enabled attribute to
reflect the taskname feature state rather than the release feature
state. This is a copy-paste error from the adjacent
sysdata_taskname_enabled_show() function.
The corresponding _store function already uses the correct
SYSDATA_RELEASE flag.
Yung Chih Su [Mon, 2 Mar 2026 06:02:47 +0000 (14:02 +0800)]
net: ipv4: fix ARM64 alignment fault in multipath hash seed
`struct sysctl_fib_multipath_hash_seed` contains two u32 fields
(user_seed and mp_seed), making it an 8-byte structure with a 4-byte
alignment requirement.
In `fib_multipath_hash_from_keys()`, the code evaluates the entire
struct atomically via `READ_ONCE()`:
While this silently works on GCC by falling back to unaligned regular
loads which the ARM64 kernel tolerates, it causes a fatal kernel panic
when compiled with Clang and LTO enabled.
Commit e35123d83ee3 ("arm64: lto: Strengthen READ_ONCE() to acquire
when CONFIG_LTO=y") strengthens `READ_ONCE()` to use Load-Acquire
instructions (`ldar` / `ldapr`) to prevent compiler reordering bugs
under Clang LTO. Since the macro evaluates the full 8-byte struct,
Clang emits a 64-bit `ldar` instruction. ARM64 architecture strictly
requires `ldar` to be naturally aligned, thus executing it on a 4-byte
aligned address triggers a strict Alignment Fault (FSC = 0x21).
Fix the read side by moving the `READ_ONCE()` directly to the `u32`
member, which emits a safe 32-bit `ldar Wn`.
Furthermore, Eric Dumazet pointed out that `WRITE_ONCE()` on the entire
struct in `proc_fib_multipath_hash_set_seed()` is also flawed. Analysis
shows that Clang splits this 8-byte write into two separate 32-bit
`str` instructions. While this avoids an alignment fault, it destroys
atomicity and exposes a tear-write vulnerability. Fix this by
explicitly splitting the write into two 32-bit `WRITE_ONCE()`
operations.
Finally, add the missing `READ_ONCE()` when reading `user_seed` in
`proc_fib_multipath_hash_seed()` to ensure proper pairing and
concurrency safety.
Fixes: 4ee2a8cace3f ("net: ipv4: Add a sysctl to set multipath hash seed") Signed-off-by: Yung Chih Su <yuuchihsu@gmail.com> Reviewed-by: Eric Dumazet <edumazet@google.com> Link: https://patch.msgid.link/20260302060247.7066-1-yuuchihsu@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
projects / thirdparty / linux.git / log
