Kohei Enju [Fri, 15 May 2026 18:24:15 +0000 (11:24 -0700)]
igc: set tx buffer type for SMD frames
Sashiko pointed out that igc_fpe_init_smd_frame() initializes
igc_tx_buffer fields for an SMD skb, but does not set the buffer type:
https://sashiko.dev/#/patchset/20260415025226.114115-1-kohei%40enjuk.jp
Since igc_tx_buffer entries are reused, a stale XDP or XSK type can
remain and make TX completion use the wrong cleanup path.
Set the buffer type to IGC_TX_BUFFER_TYPE_SKB.
Fixes: 5422570c0010 ("igc: add support for frame preemption verification") Signed-off-by: Kohei Enju <kohei@enjuk.jp> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Avigail Dahan <avigailx.dahan@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-9-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
ixgbevf: fix use-after-free in VEPA multicast source pruning
ixgbevf_clean_rx_irq() prunes frames whose source MAC matches the VF's
own address (VEPA multicast workaround) by freeing the skb and
continuing to the next descriptor:
dev_kfree_skb_irq(skb);
continue;
The skb pointer is declared outside the while loop and persists across
iterations. Because the continue skips the "skb = NULL" reset at the
bottom of the loop, the next iteration enters the "else if (skb)" path
and calls ixgbevf_add_rx_frag() on the freed skb, dereferencing
skb_shinfo(skb)->nr_frags - a use-after-free in NAPI softirq context.
The sibling driver iavf already handles this correctly by nulling the
pointer before continuing. Apply the same pattern here.
I do not have ixgbevf hardware; the bug was found by static analysis
(scan_drop_continue_loops.py + semgrep drop_continue_in_loop, multi-tool
corroboration with the highest score in the scan). The UAF was confirmed
under KASAN by loading a test module that reproduces the exact code
pattern (alloc skb, kfree_skb, then read skb_shinfo(skb)->nr_frags):
QEMU emulates igb (82576) but not ixgbe (82599), and the igbvf VF
driver does not include the VEPA source pruning path, so a full
end-to-end reproduction with emulated hardware was not possible.
Fixes: bad17234ba70 ("ixgbevf: Change receive model to use double buffered page based receives") Cc: stable@vger.kernel.org Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Rafal Romanowski <rafal.romanowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-8-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Grzegorz Nitka [Fri, 15 May 2026 18:24:13 +0000 (11:24 -0700)]
ice: restore PTP Rx timestamp config after ethtool set-channels
When ethtool -L changes queue counts, ice_vsi_recfg_qs() closes and
rebuilds the VSI, reallocating Rx rings. The newly allocated rings have
ptp_rx cleared, so RX hardware timestamps are no longer attached to skb
until hwtstamp configuration is applied again.
Restore timestamp mode after ice_vsi_open() in the queue reconfiguration
path, matching reset/rebuild behavior and ensuring newly rebuilt Rx rings
have PTP RX timestamping re-enabled.
Testing hints:
- run ptp4l application in client synchronization mode:
ptp4l -i ethX -m -s
- run PTP traffic
- change queue number on ethX netdev interface:
ethtool -L ethX combined new_queue_size
- observe ptp4l output
- expected result: no "received DELAY_REQ without timestamp" messages
Fixes: 77a781155a65 ("ice: enable receive hardware timestamping") Cc: stable@vger.kernel.org Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Alexander Nowlin <alexander.nowlin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-7-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Grzegorz Nitka [Fri, 15 May 2026 18:24:12 +0000 (11:24 -0700)]
ice: ptp: use primary NAC semaphore on E825
For E825 2xNAC configurations, PTP semaphore operations must hit the
primary NAC register block so both sides coordinate on the same lock.
Commit e2193f9f9ec9 ("ice: enable timesync operation on 2xNAC E825
devices") updated other primary-only PTP register accesses to
use the primary NAC on non-primary functions, but left ice_ptp_lock()
and ice_ptp_unlock() operating on the local NAC. As a result, secondary
NAC PTP paths can take a different semaphore than the primary side.
Select the primary hardware in ice_ptp_lock() and ice_ptp_unlock() when
the current function is not primary, keeping semaphore operations
symmetric and consistent with the rest of the 2xNAC PTP register access
path.
Fixes: e2193f9f9ec9 ("ice: enable timesync operation on 2xNAC E825 devices") Reviewed-by: Arkadiusz Kubalewski <Arkadiusz.kubalewski@intel.com> Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Alexander Nowlin <alexander.nowlin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-6-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Grzegorz Nitka [Fri, 15 May 2026 18:24:11 +0000 (11:24 -0700)]
ice: ptp: serialize E825 PHY timer start with PTP lock
ice_start_phy_timer_eth56g() programs TIMETUS registers and issues
INIT_INCVAL without holding the global PTP semaphore.
This allows concurrent PTP command paths to interleave with PHY timer
start, which can make the sequence fail and leave timer initialization
inconsistent.
Take the PTP lock around TIMETUS registers programming and INIT_INCVAL
command execution, and make sure the lock is released on all error paths.
Keep the subsequent sync step outside of this critical section, since
ice_sync_phy_timer_eth56g() takes the same semaphore internally.
Fixes: 7cab44f1c35f ("ice: Introduce ETH56G PHY model for E825C products") Reviewed-by: Arkadiusz Kubalewski <Arkadiusz.kubalewski@intel.com> Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com> Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Tested-by: Alexander Nowlin <alexander.nowlin@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-5-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Marcin Szycik [Fri, 15 May 2026 18:24:10 +0000 (11:24 -0700)]
ice: fix setting promisc mode while adding VID filter
There are at least two paths through which VSI promiscuous mode can be
independently configured via ice_fltr_set_vsi_promisc():
- ice_vlan_rx_add_vid() (netdev op)
- ice_service_task() -> ... -> ice_set_promisc()
Both paths may try to program promiscuous mode concurrently. One such
scenario is:
1. Add ice netdev to bond
2. Add the bond netdev to bridge
3. ice netdev enters allmulticast mode (IFF_ALLMULTI)
4. Service task programs promisc mode filter
5. Bridge -> bond calls ice_vlan_rx_add_vid()
Crucially, ice_vlan_rx_add_vid() fails if ice_fltr_set_vsi_promisc()
returns any error, including -EEXIST. This causes VLAN filtering setup
to fail on the bond interface. ice_set_promisc() already handles -EEXIST
correctly.
Fix by adding the same -EEXIST check to ice_vlan_rx_add_vid(): if the
promisc filter is already programmed, continue without returning error.
Fixes: 1273f89578f2 ("ice: Fix broken IFF_ALLMULTI handling") Cc: stable@vger.kernel.org Signed-off-by: Marcin Szycik <marcin.szycik@intel.com> Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-4-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
ice: fix VF queue configuration with low MTU values
The ice driver's VF queue configuration validation rejects
databuffer_size values below 1024 bytes, which prevents VFs from
using MTU values below 871 bytes.
The iavf driver calculates databuffer_size based on the MTU using:
databuffer_size = ALIGN(MTU + LIBETH_RX_LL_LEN, 128)
where LIBETH_RX_LL_LEN = 26 (ETH_HLEN + 2*VLAN_HLEN + ETH_FCS_LEN).
For MTU values below 871:
MTU 870: 870 + 26 = 896, aligned to 128 = 896 (< 1024, rejected)
MTU 871: 871 + 26 = 897, aligned to 128 = 1024 (>= 1024, accepted)
The 1024-byte minimum seems unnecessarily restrictive, because the hardware
supports databuffer_size as low as 128 bytes (the alignment boundary),
which should allow MTU values down to the standard minimum of 68 bytes.
I haven't found the reason why the limit was configured in the commit 9c7dd7566d18 ("ice: add validation in OP_CONFIG_VSI_QUEUES VF message"), so
with no more information and since it is working, change the minimum
databuffer_size validation from 1024 to 128 bytes to allow standard low
MTU values while still preventing invalid configurations.
Fixes: 9c7dd7566d18 ("ice: add validation in OP_CONFIG_VSI_QUEUES VF message")
cc: stable@vger.kernel.org Signed-off-by: Jose Ignacio Tornos Martinez <jtornosm@redhat.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com> Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de> Tested-by: Rafal Romanowski <rafal.romanowski@intel.com> Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-3-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jacob Keller [Fri, 15 May 2026 18:24:08 +0000 (11:24 -0700)]
ice: fix locking around wait_event_interruptible_locked_irq
Commit 50327223a8bb ("ice: add lock to protect low latency interface")
introduced a wait queue used to protect the low latency timer interface.
The queue is used with the wait_event_interruptible_locked_irq macro, which
unlocks the wait queue lock while sleeping. The irq variant uses
spin_lock_irq and spin_unlock_irq to manage this. The wait queue lock was
previously locked using spin_lock_irqsave. This difference in lock variants
could lead to issues, since wait_event would unlock the wait queue and
restore interrupts while sleeping.
The ice_read_phy_tstamp_ll_e810() function is ultimately called through
ice_read_phy_tstamp, which is called from ice_ptp_process_tx_tstamp or
ice_ptp_clear_unexpected_tx_ready. The former is called through the
miscellaneous IRQ thread function, while the latter is called from the
service task work queue thread. Neither of these functions has interrupts
disabled, so use spin_lock_irq instead of spin_lock_irqsave.
Fixes: 50327223a8bb ("ice: add lock to protect low latency interface") Cc: stable@vger.kernel.org Reported-by: Jakub Kicinski <kuba@kernel.org> Closes: https://lore.kernel.org/netdev/20250109181823.77f44c69@kernel.org/ Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com> Reviewed-by: Simon Horman <horms@kernel.org> Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel) Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com> Link: https://patch.msgid.link/20260515182419.1597859-2-anthony.l.nguyen@intel.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jakub Kicinski [Mon, 18 May 2026 23:59:29 +0000 (16:59 -0700)]
Merge tag 'nf-26-05-16' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
Pablo Neira Ayuso says:
====================
Netfilter/IPVS fixes for net
The following patchset contains Netfilter/IPVS fixes for net:
1) Fix small race windows in nf_ct_helper_log() when accessing helper,
from Florian Westphal.
2) Fix potential infinite loop and race conditions in IPVS caused by
frequent user-triggered service table changes, from Julia Anastasov.
3) Fix a race condition when dumping ipsets for restore,
from Jozsef Kadlecsik.
4) Fix inner transport offset in IPv6 in nft_inner when extension
headers come before the layer 4 transport header, from Yizhou Zhao.
5) Fix incorrect iteration over IPv4 ranges in several hash set types,
from Nan Li.
6) Fix incorrect order when restoring BH in nft_inner_restore_tun_ctx(),
from Florian Westphal.
7) Validate option array from ip6t_hbh checkpath() to fix an off-by-one
access, from Zhengchuan Liang.
8) Fix race condition between ipset list -terse and concurrent updates,
from Jozsef Kadlecisk.
9) Fix race condition when inserting elements into a hash bucket, also
from Jozsef.
10) Annotate access to first free slot in hashtable, from Jozsef Kadlecsik.
11) Ensure sufficient headroom in br_netfilter neigh transmission,
from Lorenzo Bianconi.
12) Hold reference on skb->dev in nfqueue exit path, bridge local input
is speciall since skb->dev != state->indev, allowing for net_device
to go away while packet is sitting in nfqueue. From Haoze Xie.
* tag 'nf-26-05-16' of git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
netfilter: nf_queue: hold bridge skb->dev while queued
netfilter: br_netfilter: Reallocate headroom if necessary in neigh_hh_bridge()
netfilter: ipset: annotate "pos" for concurrent readers/writers
netfilter: ipset: Fix data race between add and dump in all hash types
netfilter: ipset: Fix data race between add and list header in all hash types
netfilter: ip6t_hbh: reject oversized option lists
netfilter: nft_inner: release local_lock before re-enabling softirqs
netfilter: ipset: stop hash:* range iteration at end
netfilter: nft_inner: Fix IPv6 inner_thoff desync
netfilter: ipset: fix a potential dump-destroy race
ipvs: avoid possible loop in ip_vs_dst_event on resizing
netfilter: nf_conntrack_helper: fix possible null deref during error log
====================
Jakub Kicinski [Mon, 18 May 2026 23:40:04 +0000 (16:40 -0700)]
Merge tag 'for-net-2026-05-14' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth
Luiz Augusto von Dentz says:
====================
bluetooth pull request for net:
- af_bluetooth: serialize accept_q access
- L2CAP: ecred_reconfigure: send packed pdu, not stack pointer
- btmtk: accept too short WMT FUNC_CTRL events
- hci_qca: Convert timeout from jiffies to ms
* tag 'for-net-2026-05-14' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetooth/bluetooth:
Bluetooth: hci_qca: Convert timeout from jiffies to ms
Bluetooth: L2CAP: ecred_reconfigure: send packed pdu, not stack pointer
Bluetooth: btmtk: accept too short WMT FUNC_CTRL events
Bluetooth: serialize accept_q access
====================
Ilya Maximets [Thu, 14 May 2026 18:46:31 +0000 (20:46 +0200)]
openvswitch: vport: fix race between linking and the device notifier
Sashiko reports that it is technically possible that we got the device
reference, but by the time we're linking it to the OVS datapath, it
may be already in the process of being deleted. In this case if the
notifier wins the race for RTNL, it will see that the device is not
yet in the OVS datapath (ovs_netdev_get_vport() will fail in the
dp_device_event()) and will do nothing. Then the ovs_netdev_link()
will take the RTNL and link the unregistering device to OVS datapath.
Eventually, netdev_wait_allrefs_any() will re-broadcast the event and
the device will be properly detached, but it will take at least a
second before that happens, so it's not something we should rely on.
Let's avoid linking the non-registered device in the first place.
Note: As per documentation, RTNL doesn't protect the reg_state, but
it actually does for all the state transitions we care about here,
so it should not be necessary to use READ_ONCE or taking the instance
lock. We can still do that, but we have a few more places even in
this file where the reg_state is accessed without those while under
RTNL, and many more places like this across the kernel code, so it
might make more sense to change all of them in a more centralized
fashion in the future, if necessary.
Weiming Shi [Thu, 14 May 2026 12:25:12 +0000 (05:25 -0700)]
net: qualcomm: rmnet: fix endpoint use-after-free in rmnet_dellink()
rmnet_dellink() removes the endpoint from the hash table with
hlist_del_init_rcu() and then immediately frees it with kfree(). However,
RCU readers on the receive path (rmnet_rx_handler ->
__rmnet_map_ingress_handler) may still hold a reference to the endpoint and
dereference ep->egress_dev after the memory has been freed. The endpoint is
a kmalloc-32 object, and the stale read at offset 8 corresponds to the
egress_dev pointer.
Add an rcu_head field to struct rmnet_endpoint and replace kfree() with
kfree_rcu() so the endpoint memory remains valid through the RCU grace
period. Also remove the rmnet_vnd_dellink() call and inline only the
nr_rmnet_devs decrement, since rmnet_vnd_dellink() would set
ep->egress_dev to NULL during the grace period, creating a data race
with lockless readers.
Fixes: ceed73a2cf4a ("drivers: net: ethernet: qualcomm: rmnet: Initial implementation") Reported-by: Xiang Mei <xmei5@asu.edu> Signed-off-by: Weiming Shi <bestswngs@gmail.com> Link: https://patch.msgid.link/20260514122511.3083479-2-bestswngs@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Weiming Shi [Thu, 14 May 2026 12:38:08 +0000 (05:38 -0700)]
net: appletalk: fix NULL pointer dereference in aarp_send_ddp()
aarp_send_ddp() calls atalk_find_dev_addr(dev) in the LocalTalk fast
path without checking for NULL. When the device has no AppleTalk
interface configured (dev->atalk_ptr == NULL), this leads to a NULL
pointer dereference at the at->s_net access.
Dragos Tatulea [Wed, 13 May 2026 06:46:13 +0000 (09:46 +0300)]
net/mlx5e: xsk: Fix unlocked writing to ICOSQ
During napi poll, when the affinity changes and there's still XSK work
to be done, we trigger an ICOSQ interrupt on the new CPU. However, this
triggering on the ICOSQ is done unprotected.
There are 2 such races:
A) mlx5e_trigger_irq() is called while mlx5e_xsk_alloc_rx_mpwqe() is
running from a different CPU due to affinity change. This can happen
because IRQ triggering is done after napi_complete_done(). At this point
the NAPI can be scheduled on a different CPU. Like this:
CPU A (old affinity, NAPI tail) CPU B (new affinity, fresh NAPI)
------------------------------- --------------------------------
napi_complete_done() clears SCHED
mlx5e_cq_arm(...)
napi_schedule_prep() sets SCHED
mlx5e_napi_poll()
mlx5e_xsk_alloc_rx_mpwqe()
mlx5e_icosq_sync_lock() // noop
memcpy 640 B UMR body
advance sq->pc by 10
mlx5e_trigger_irq(&c->icosq)
wqe_info[pi] = {NOP, 1}
mlx5e_post_nop() advances sq->pc
B) mlx5e_trigger_irq() is called on the ICOSQ when
mlx5e_trigger_napi_icosq() is running.
The obvious fix would be to lock the ICOSQ. But ICOSQ has an optimized
locking scheme that doesn't work for this scenario. Kick the async ICOSQ
instead which is always locked.
This issue was noticed in the wild with the following splat:
Haoze Xie [Fri, 15 May 2026 03:19:02 +0000 (11:19 +0800)]
netfilter: nf_queue: hold bridge skb->dev while queued
br_pass_frame_up() rewrites skb->dev from the ingress port to the bridge
master before queueing bridge LOCAL_IN packets. NFQUEUE only holds
references on state.in/out and bridge physdevs, so a queued bridge
packet can retain a freed bridge master in skb->dev until reinjection.
When the verdict is reinjected later, br_netif_receive_skb() re-enters
the receive path with skb->dev still pointing at the freed bridge master,
triggering a use-after-free.
Store skb->dev in the queue entry, hold a reference on it for the queue
lifetime, and use the saved device when dropping queued packets during
NETDEV_DOWN handling.
Fixes: ac2863445686 ("netfilter: bridge: add nf_afinfo to enable queuing to userspace") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Haoze Xie <royenheart@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Lorenzo Bianconi [Thu, 14 May 2026 14:46:38 +0000 (16:46 +0200)]
netfilter: br_netfilter: Reallocate headroom if necessary in neigh_hh_bridge()
neigh_hh_bridge() assumes the skb always has sufficient headroom to copy
the aligned L2 header. This assumption can trigger the crash reported
below using the following netfilter setup:
Fix the issue reallocating the skb headroom if necessary in neigh_hh_bridge routine.
Fixes: e179e6322ac33 ("netfilter: bridge-netfilter: Fix MAC header handling with IP DNAT") Reviewed-by: Ido Schimmel <idosch@nvidia.com> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Jozsef Kadlecsik [Thu, 14 May 2026 08:55:13 +0000 (10:55 +0200)]
netfilter: ipset: annotate "pos" for concurrent readers/writers
The "pos" structure member of struct hbucket stores the first
free slot in the hash bucket of a hash type of set and there
are concurrent readers/writers. Annotate accesses properly.
Fixes: 18f84d41d34f ("netfilter: ipset: Introduce RCU locking in hash:* types") Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
netfilter: ipset: Fix data race between add and dump in all hash types
When adding a new entry to the next position in the existing hash bucket,
the position index was incremented too early and parallel dump could
read it before the entry was populated with the value. Move the setting
of the position index after populating the entry.
v2: Position counting fixed, noticed by Florian Westphal.
Jozsef Kadlecsik [Thu, 14 May 2026 08:55:11 +0000 (10:55 +0200)]
netfilter: ipset: Fix data race between add and list header in all hash types
The "ipset list -terse" command is actually a dump operation which
may run parallel with "ipset add" commands, which can trigger an
internal resizing of the hash type of sets just being dumped. However,
dumping just the header part of the set was not protected against
underlying resizing. Fix it by protecting the header dumping part
as well.
Fixes: c4c997839cf9 ("netfilter: ipset: Fix parallel resizing and listing of the same set") Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
struct ip6t_opts stores at most IP6T_OPTS_OPTSNR option descriptors,
but hbh_mt6_check() does not reject larger optsnr values supplied from
userspace.
Validate optsnr in the rule setup path so only match data that fits the
fixed-size opts array can be installed. This follows the existing xtables
pattern of rejecting invalid user-provided counts in checkentry() and
keeps the packet matching path unchanged.
`struct ip6t_opts` has a fixed `opts[IP6T_OPTS_OPTSNR]` array,
where `IP6T_OPTS_OPTSNR` is 16, then off-by-one array access is possible:
[ 137.924693][ T8692] UBSAN: array-index-out-of-bounds in ../net/ipv6/netfilter/ip6t_hbh.c:110:29
[ 137.926167][ T8692] index 16 is out of range for type '__u16 [16]'
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Zhengchuan Liang <zcliangcn@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
The iterator must stop once the last address in the requested range has
been processed. Advancing it once more can move the traversal state past
the end of the request, so a later retry may continue from an unintended
position.
Handle the iterator increment explicitly at the end of the loop and stop
once the upper bound has been processed. This keeps the existing retry
behaviour intact for valid ranges while preventing traversal from
continuing past the original boundary.
Fixes: 48596a8ddc46 ("netfilter: ipset: Fix adding an IPv4 range containing more than 2^31 addresses") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Nan Li <tonanli66@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Yizhou Zhao [Mon, 11 May 2026 17:30:41 +0000 (01:30 +0800)]
netfilter: nft_inner: Fix IPv6 inner_thoff desync
In nft_inner_parse_l2l3(), when processing inner IPv6 packets,
ipv6_find_hdr() correctly computes the transport header offset
traversing all extension headers, but the result is immediately
overwritten with nhoff + sizeof(_ip6h) (40 bytes), which only
accounts for the IPv6 base header. This creates a desync between
inner_thoff (wrong — points to extension header start) and l4proto
(correct — e.g., IPPROTO_TCP), enabling transport header forgery
and potential firewall bypass. This issue affects stable versions
from Linux 6.2.
For comparison, the normal (non-inner) IPv6 path correctly
preserves ipv6_find_hdr()'s result. Removing the incorrect overwrite
ensures that ipv6_find_hdr()'s calculated transport header offset is
preserved, thereby fixing the desynchronization.
Fixes: 3a07327d10a0 ("netfilter: nft_inner: support for inner tunnel header matching") Cc: stable@vger.kernel.org Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn> Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn> Reported-by: Xuewei Feng <fengxw06@126.com> Reported-by: Qi Li <qli01@tsinghua.edu.cn> Reported-by: Ke Xu <xuke@tsinghua.edu.cn> Assisted-by: GLM:5.1 Z.ai Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn> Reviewed-by: Fernando Fernandez Mancera <fmancera@suse.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Jozsef Kadlecsik [Thu, 14 May 2026 08:55:10 +0000 (10:55 +0200)]
netfilter: ipset: fix a potential dump-destroy race
When dumping sets in order to create the proper order for restore,
the list type of sets dumped last. Therefore internally we run the
dumping loop twice: first with all non-list type of sets and skipping
the list type ones and then secondly for the list type of sets.
Sashiko noticed that there's a potential race between dump and destroy
if in the first loop the last set was a list type of set: its pointer
remains unreferenced and a concurrent destroy can free it.
Fix the issue by resetting the variable holding the pointer.
Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Julian Anastasov [Sun, 10 May 2026 10:46:05 +0000 (13:46 +0300)]
ipvs: avoid possible loop in ip_vs_dst_event on resizing
Sashiko points out that unprivileged user can frequently
call ip_vs_flush() or ip_vs_del_service() to trigger
svc_table_changes updates that can lead to infinite loop
in ip_vs_dst_event(). This can also happen if the user
triggers frequent table resizing without deleting all
services. We should also consider the possible effects
if the user triggers many NETDEV_DOWN events.
One way to solve it is to hold svc_resize_sem in
ip_vs_dst_event() but this can block the dev notifier
during the whole resizing process.
Instead, use new rw_semaphore svc_replace_sem to protect just
the svc_table replacement which is a short code section.
Then hold svc_replace_sem in ip_vs_dst_event() to serialize
with replacing the svc_table. As result, loop is avoided
as there is no need to repeat the table walking from the
start. By this way changes in svc_table_changes can happen
only when all services are removed and all dev references
dropped which allows us to abort the table walking.
As IP_VS_WORK_SVC_NORESIZE is the flag used to stop the
svc_resize_work under service_mutex, we should check only
this flag often but not while under service_mutex.
To remove the mutex_trylock() for service_mutex in the
second phase where the resizer installs the new table
after rehashing, we will avoid holding the service_mutex
there. As result, the code in configuration context which
is under service_mutex should access ipvs->svc_table under
RCU because it can be replaced at anytime and released
after a RCU grace period. As for ip_vs_zero_all(), it needs
different solution as a table walker which can escape
single RCU read-side critical section: to hold the
svc_replace_sem to prevent table to be replaced.
In ip_vs_status_show() prefer to hold svc_replace_sem
to avoid many loops, just detect if the svc_table is
removed.
Prefer the newly attached table for the u_thresh/l_thresh
checks to know when to grow/shrink while adding or deleting
services because the new table size is based on the latest
parameters.
net: hsr: defer node table free until after RCU readers
HSR node-list and node-status generic-netlink operations run under
rcu_read_lock(). They walk hsr->node_db through hsr_get_next_node() and
hsr_get_node_data(), but RTM_DELLINK teardown removes the same node table
with plain list_del() and frees each node immediately.
That lets a generic-netlink reader hold a struct hsr_node pointer across
hsr_dellink(). In a KASAN build, widening the reader window after
hsr_get_next_node() obtains the node reproduces a slab-use-after-free
when the reader copies node->macaddress_A; the freeing stack is
hsr_del_nodes() from hsr_dellink().
Use list_del_rcu() and defer the free through the existing
hsr_free_node_rcu() callback. This matches the lifetime rule used by the
HSR prune paths, which already delete nodes with list_del_rcu() and
call_rcu().
vsock/virtio: fix zerocopy completion for multi-skb sends
When a large message is fragmented into multiple skbs, the zerocopy
uarg is only allocated and attached to the last skb in the loop.
Non-final skbs carry pinned user pages with no completion tracking,
so the kernel has no way to notify userspace when those pages are safe
to reuse. If the loop breaks early the uarg is never allocated at all,
leaking pinned pages with no completion notification.
Fix this by following the approach used by TCP: allocate the zerocopy
uarg (if not provided by the caller) before the send loop and attach
it to every skb via skb_zcopy_set(), which takes a reference per skb.
Each skb's completion properly decrements the refcount, and the
notification only fires after the last skb is freed.
On failure, if no data was sent, the uarg is cleanly aborted via
net_zcopy_put_abort().
This issue was initially discovered by sashiko while reviewing commit 1cb36e252211 ("vsock/virtio: fix MSG_ZEROCOPY pinned-pages accounting")
but was pre-existing.
Fixes: 581512a6dc93 ("vsock/virtio: MSG_ZEROCOPY flag support") Closes: https://sashiko.dev/#/patchset/20260420132051.217589-1-sgarzare%40redhat.com Reported-by: Maher Azzouzi <maherazz04@gmail.com> Signed-off-by: Stefano Garzarella <sgarzare@redhat.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Acked-by: Arseniy Krasnov <avkrasnov@salutedevices.com> Link: https://patch.msgid.link/20260514092948.268720-1-sgarzare@redhat.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Sam Daly [Wed, 13 May 2026 16:42:53 +0000 (18:42 +0200)]
octeontx2-af: CGX: add bounds check to cgx_speed_mbps index
cgx_speed_mbps has 13 elements but RESP_LINKSTAT_SPEED can yield values
0-15. If it returns a value >= 13, this causes an out-of-bounds array
access. Add a bounds check and default to speed 0 if the index is out of
range.
Fixes: 61071a871ea6 ("octeontx2-af: Forward CGX link notifications to PFs") Cc: Sunil Goutham <sgoutham@marvell.com> Cc: Linu Cherian <lcherian@marvell.com> Cc: Geetha sowjanya <gakula@marvell.com> Cc: hariprasad <hkelam@marvell.com> Cc: Subbaraya Sundeep <sbhatta@marvell.com> Cc: Andrew Lunn <andrew+netdev@lunn.ch> Cc: stable <stable@kernel.org> Signed-off-by: Sam Daly <sam@samdaly.ie> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://patch.msgid.link/2026051352-refined-demise-e88d@gregkh Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Dragos Tatulea [Wed, 13 May 2026 12:45:18 +0000 (15:45 +0300)]
IB/IPoIB: ndo_set_rx_mode_async conversion
The commit in the fixes tag added a warning for devices
that are netdev ops locked that they should be converted
to .ndo_set_rx_mode_async. IPoIB for mlx5 is such a
driver which was missed during the conversion because the
flow is more complex:
- mlx5 part of IPoIB device was converted to ops-lock in commit [1].
- ipoib_intf_init() then overrides netdev_ops with
ipoib_netdev_ops_{pf,vf}, which still wired ndo_set_rx_mode to the
legacy sync path -- tripping the new warning on every probe.
So now we have the following splat:
netdevice: ib0 (uninitialized): ops-locked drivers should use ndo_set_rx_mode_async
WARNING: net/core/dev.c:11366 at register_netdevice+0x83c/0x21d0
...
register_netdev+0x1f/0x40
ipoib_add_one+0x35c/0x880 [ib_ipoib]
This patch implements .ndo_set_rx_mode_async but it simply schedules the
multicast restart task like before. This is done to maintain the
assumption that this task and others [2] must run on the same order
workqueue to avoid racing with themselves. The race between
ipoib_mcast_join_task() and ipoib_mcast_restart_task() would be the most
obvious example.
if (iphlen >= sizeof(*iph)) {
/* fix up saddr, tot_len, id, csum, transport_header */
}
It does not, however, reject ihl < 5. For such a packet the
"if (iphlen >= sizeof(*iph))" branch is skipped, leaving the
crafted iphdr untouched, but the packet is still handed to
__ip_local_out() and onward. Downstream consumers that read
iph->ihl assume a sane value: net/ipv4/ah4.c:ah_output() in
particular subtracts sizeof(struct iphdr) from top_iph->ihl * 4
and passes the (signed-int-negative, then cast to size_t)
result to memcpy(), producing an OOB access of length close to
SIZE_MAX and a host kernel panic.
An IPv4 header with ihl < 5 is malformed by definition (RFC 791:
"Internet Header Length is the length of the internet header in
32 bit words ... Note that the minimum value for a correct header
is 5."). The kernel should not be willing to inject such a
packet into its own output path.
Reject "iphlen < sizeof(*iph)" alongside the existing
"iphlen > length" check. This matches the principle that locally
constructed packets that re-enter the IP stack must pass the same
basic sanity tests that a foreign packet would be subjected to.
Once this lands, the "if (iphlen >= sizeof(*iph))" wrapper around
the fixup branch becomes redundant; left in place to keep the
patch minimal and backport-friendly. A follow-up can unwrap it.
Note that commit 86f4c90a1c5c ("ipv4, ipv6: ensure raw socket
message is big enough to hold an IP header") ensures the message
buffer is large enough to hold an iphdr, but does not constrain
the self-reported iph->ihl.
Reachability: the malformed packet source is any caller with
CAP_NET_RAW, including an unprivileged process in a user+net
namespace on a kernel with CONFIG_USER_NS=y. The reproduced AH
crash also requires a matching xfrm AH policy on the outgoing
route; a container granted CAP_NET_ADMIN can install that state
and policy in its netns. Loopback bypasses xfrm_output, so the
trigger uses a real netdev.
Reproduced on UML + KASAN: kernel-mode fault at addr 0x0 with
memcpy_orig at the crash site. Same shape reproduces inside a
rootless Docker container with --cap-add NET_ADMIN on a stock
distro kernel.
Sven Eckelmann [Wed, 13 May 2026 08:43:54 +0000 (10:43 +0200)]
batman-adv: tp_meter: directly shut down timer on cleanup
batadv_tp_sender_cleanup() was calling timer_delete_sync() followed by
timer_delete() to guard against the timer handler re-arming itself between
the two calls. This double-deletion hack relied on the sending status being
set to 0 to suppress re-arming.
Replace both calls with a single timer_shutdown_sync(). This function both
waits for any running timer callback to complete (like timer_delete_sync())
and permanently disarms the timer so it cannot be re-armed afterwards,
making re-arming prevention unconditional and self-documenting.
The re-arming property is also required because otherwise:
1. context 0 (batadv_tp_recv_ack()) checks in
batadv_tp_reset_sender_timer() if sending is still 1 -> it is
2. context 1 changes in batadv_tp_sender_shutdown() sending to 0 and in
this process forces the kthread to stop timer in
batadv_tp_sender_cleanup()
3. context 0 continues in batadv_tp_reset_sender_timer() and rearms the
timer -> but the reference for it is already gone
Cc: stable@kernel.org Fixes: 33a3bb4a3345 ("batman-adv: throughput meter implementation") Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Wed, 13 May 2026 07:01:36 +0000 (09:01 +0200)]
batman-adv: frag: disallow unicast fragment in fragment
batadv_frag_skb_buffer() is called by batadv_batman_skb_recv() when a
BATADV_UNICAST_FRAG packet is received. Once all fragments are collected
and the packet is reassembled, batadv_recv_frag_packet() calls
batadv_batman_skb_recv() again to process the defragmented payload.
A malicious sender can craft a BATADV_UNICAST_FRAG packet whose reassembled
payload is itself a BATADV_UNICAST_FRAG packet (matryoshka-style nesting).
Each nesting level recurses through batadv_batman_skb_recv() without bound,
growing the kernel stack until it is exhausted.
Since refragmentation or fragments in fragments are not actually allowed,
discard all packets which are still BATADV_UNICAST_FRAG packets after the
defragmentation process.
Cc: stable@kernel.org Fixes: 610bfc6bc99b ("batman-adv: Receive fragmented packets and merge") Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Reviewed-by: Yuan Tan <yuantan098@gmail.com> Signed-off-by: Sven Eckelmann <sven@narfation.org>
ifb_dev_init() allocates dp->tx_private to dev->num_tx_queues
entries via kzalloc_objs(*txp, dev->num_tx_queues). Both IFB
per-queue RX and TX stats live in those entries: ifb_xmit() updates
txp->rx_stats using the skb queue mapping, ifb_ri_tasklet() updates
txp->tx_stats, and ifb_stats64() aggregates both over
dev->num_tx_queues.
The ethtool stats callbacks instead size and walk the per-queue
stats with dev->real_num_rx_queues and dev->real_num_tx_queues. With
an asymmetric device where the RX queue count exceeds the TX queue
count, for example:
ip link add name ifb10 numtxqueues 1 numrxqueues 8 type ifb
ethtool -S ifb10
ifb_get_ethtool_stats() indexes past the tx_private allocation and
copies adjacent slab data through ETHTOOL_GSTATS.
Use dev->num_tx_queues consistently for the stats strings, the
stats count, and the stats data walks. This reports one RX stats
group and one TX stats group for each backing ifb_q_private entry,
which is the queue set IFB can actually populate.
Reproduced under UML+KASAN at v7.1-rc2:
BUG: KASAN: slab-out-of-bounds in ifb_fill_stats_data+0x3c/0xae
Read of size 8 at addr 0000000062dbd228 by task ethtool/36
ifb_fill_stats_data+0x3c/0xae
ifb_get_ethtool_stats+0xc0/0x129
__dev_ethtool+0x1ca5/0x363c
dev_ethtool+0x123/0x1b3
dev_ioctl+0x56c/0x744
sock_do_ioctl+0x15f/0x1b2
sock_ioctl+0x4d5/0x50a
sys_ioctl+0xd8b/0xde9
With the patch applied, the same UML+KASAN repro is silent and
ethtool -S ifb10 reports only the stats backed by the single
allocated tx_private entry.
Or Har-Toov [Wed, 13 May 2026 06:36:40 +0000 (09:36 +0300)]
net/mlx5: Skip disabled vports when setting max TX speed
When setting vports max TX speed during LAG activation or bond state
changes, the code iterates over all eswitch vports. However, some
vports may not be enabled yet.
Skip vports that are not enabled to avoid sending FW commands for
uninitialized vports. Save the LAG aggregated speed in the vport
struct so it can be applied when the vport is enabled later.
Fixes: 50f1d188c580 ("net/mlx5: Propagate LAG effective max_tx_speed to vports") Signed-off-by: Or Har-Toov <ohartoov@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260513063640.334132-1-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jeroen Massar [Wed, 13 May 2026 06:33:02 +0000 (09:33 +0300)]
net/mlx5: Do not restore destination-less TC rules
After IPsec policy/state TX rules are added, any TC flow rule, which
forwards packets to uplink, is modified to forward to IPsec TX tables.
As these tables are destroyed dynamically, whenever there is no
reference to them, the destinations of this kind of rules must be
restored to uplink, unless there is no destination for that rule.
The flow rules FLOW_ACTION_ACCEPT, DROP, TRAP, GOTO and SAMPLE do not
have a destination port, and thus out_count = 0.
At cleanup time of the rules in mlx5_esw_ipsec_modify_flow_dests
we call mlx5_eswitch_restore_ipsec_rule but as the above types
do not have a destination we get an underflow of out_count, as
the port is passed, which is esw_attr->out_count - 1.
This change avoids calling mlx5_eswitch_restore_ipsec_rule when
there are no output destinations and thus avoids the underflow.
Fixes: d1569537a837 ("net/mlx5e: Modify and restore TC rules for IPSec TX rules") Signed-off-by: Jeroen Massar <jmassar@nvidia.com> Reviewed-by: Jianbo Liu <jianbol@nvidia.com> Reviewed-by: Cosmin Ratiu <cratiu@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260513063302.333761-1-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Gal Pressman [Wed, 13 May 2026 06:27:37 +0000 (09:27 +0300)]
net/mlx5e: Don't leak RSS context in case of error
If mlx5e_rx_res_rss_set_rxfh() fails during mlx5e_create_rxfh_context(),
the RSS context is not cleaned up.
This leaves a stale entry in 'res->rss[rss_idx]' that occupies a context
slot.
Destroy the RSS context before returning the error.
Fixes: 6c2509d44636 ("net/mlx5e: Add error flow for ethtool -X command") Signed-off-by: Gal Pressman <gal@nvidia.com> Reviewed-by: Nimrod Oren <noren@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20260513062737.333259-1-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Chuck Lever [Wed, 13 May 2026 12:58:25 +0000 (08:58 -0400)]
tls: Preserve sk_err across recvmsg() when data has been copied
The sk_err check in tls_rx_rec_wait() consumes the error via
sock_error(), which clears sk_err atomically. When the caller
(tls_sw_recvmsg, tls_sw_splice_read, or tls_sw_read_sock) already
has bytes copied to userspace, it returns those bytes and discards
the error from this call. sk_err is now zero on the socket, so the
next read syscall observes only RCV_SHUTDOWN and reports a clean
EOF instead of the actual error (typically -ECONNRESET).
The race is reachable when tls_read_flush_backlog()'s periodic
sk_flush_backlog() triggers tcp_reset() in the middle of a
multi-record read.
Pass a has_copied flag to tls_rx_rec_wait(). When has_copied is
false, consume sk_err via sock_error() as before. When has_copied
is true, report the error from READ_ONCE() but leave sk_err set:
the caller returns the byte count and discards the err from this
call, and the next read syscall surfaces the preserved sk_err. This
mirrors the tcp_recvmsg() preserve-and-surface pattern.
The decrypt-abort path is unaffected: tls_err_abort() raises
sk_err to EBADMSG after tls_rx_rec_wait() returns, and nothing
on the caller's return path consumes it, so the EBADMSG surfaces
on the next read.
tls_sw_splice_read() passes has_copied=false: it processes
one record per call, so no bytes have been copied within the
function when tls_rx_rec_wait() runs. A reset that arrives
between iterations of splice_direct_to_actor() (the sendfile()
path) is still consumed by sock_error() in the later call, and the
outer loop returns the prior iterations' byte count and drops the
error. tcp_splice_read() exhibits the same pattern at the iteration
boundary; addressing it belongs at the splice_direct_to_actor()
layer and is out of scope here.
Fixes: c46b01839f7a ("tls: rx: periodically flush socket backlog") Suggested-by: Jakub Kicinski <kuba@kernel.org> Signed-off-by: Chuck Lever <chuck.lever@oracle.com> Link: https://patch.msgid.link/20260513125825.205189-1-cel@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Dawei Feng [Wed, 13 May 2026 15:13:20 +0000 (23:13 +0800)]
octeontx2-pf: fix double free in rvu_rep_rsrc_init()
rvu_rep_rsrc_init() allocates queue memory before calling
otx2_init_hw_resources(). When hardware resource setup fails,
otx2_init_hw_resources() already unwinds the partially initialized
SQ, CQ, and aura state before returning an error. The representor
error path then calls otx2_free_hw_resources() again and can free
the same resources a second time.
Fix this by splitting the cleanup labels so that a failure from
otx2_init_hw_resources() only releases queue memory. Keep the
otx2_free_hw_resources() call for failures that happen after
hardware resource initialization completed successfully.
The bug was first flagged by an experimental analysis tool we are
developing for kernel memory-management bugs while analyzing
v6.13-rc1. The tool is still under development and is not yet publicly
available. Manual inspection confirms that the bug is still
present in v7.1-rc3.
Runtime validation was not performed because reproducing this path
requires OcteonTX2 representor hardware.
William Bowling [Wed, 13 May 2026 04:16:35 +0000 (04:16 +0000)]
net: skbuff: preserve shared-frag marker during coalescing
skb_try_coalesce() can attach paged frags from @from to @to. If @from
has SKBFL_SHARED_FRAG set, the resulting @to skb can contain the same
externally-owned or page-cache-backed frags, but the shared-frag marker
is currently lost.
That breaks the invariant relied on by later in-place writers. In
particular, ESP input checks skb_has_shared_frag() before deciding
whether an uncloned nonlinear skb can skip skb_cow_data(). If TCP
receive coalescing has moved shared frags into an unmarked skb, ESP can
see skb_has_shared_frag() as false and decrypt in place over page-cache
backed frags.
Propagate SKBFL_SHARED_FRAG when skb_try_coalesce() transfers paged
frags. The tailroom copy path does not need the marker because it copies
bytes into @to's linear data rather than transferring frag descriptors.
Fixes: cef401de7be8 ("net: fix possible wrong checksum generation") Fixes: f4c50a4034e6 ("xfrm: esp: avoid in-place decrypt on shared skb frags") Signed-off-by: William Bowling <vakzz@zellic.io> Reviewed-by: Eric Dumazet <edumazet@google.com> Tested-by: Jiayuan Chen <jiayuan.chen@linux.dev> Link: https://patch.msgid.link/20260513041635.1289541-1-vakzz@zellic.io Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Matt Fleming [Wed, 13 May 2026 11:22:26 +0000 (12:22 +0100)]
net/mlx5e: Fix use-after-free in mlx5e_tx_reporter_timeout_recover
mlx5e_tx_reporter_timeout_recover() accesses sq->netdev after
mlx5e_safe_reopen_channels() has torn down and freed the channel (and
its embedded SQs). Replace the three sq->netdev references with
priv->netdev which is safe because priv outlives channel teardown.
The netdev_err() call already used priv->netdev for this reason; make
the trylock/unlock and health_channel_eq_recover calls consistent.
This fixes the following KASAN splat:
BUG: KASAN: use-after-free in mlx5e_tx_reporter_timeout_recover+0x1dd/0x360 [mlx5_core]
Read of size 8 at addr ffff889860ed0b28 by task kworker/u113:2/5277
Maoyi Xie [Tue, 12 May 2026 14:28:07 +0000 (22:28 +0800)]
rds_tcp: close NULL deref window in rds_tcp_set_callbacks
rds_tcp_set_callbacks() links a new rds_tcp_connection onto
rds_tcp_tc_list under rds_tcp_tc_list_lock. It releases the
lock, then assigns tc->t_sock = sock outside the lock.
rds_tcp_tc_info() and rds6_tcp_tc_info() walk rds_tcp_tc_list
under the same lock. Both dereference tc->t_sock->sk without
a NULL check.
A reader can acquire rds_tcp_tc_list_lock between the writer's
spin_unlock and the t_sock store. It then sees a list entry
whose t_sock is NULL. The dereference of tc->t_sock->sk is a
NULL access.
Move tc->t_sock = sock inside rds_tcp_tc_list_lock, before
list_add_tail. A reader holding the lock then observes the
linkage and the t_sock store together.
The restore path is safe. rds_tcp_restore_callbacks() does
list_del_init inside the lock. The matching tc->t_sock = NULL
after unlink is harmless to readers holding the lock.
Fixes: 70041088e3b9 ("RDS: Add TCP transport to RDS") Suggested-by: Simon Horman <horms@kernel.org> Signed-off-by: Maoyi Xie <maoyi.xie@ntu.edu.sg> Reviewed-by: Allison Henderson <achender@kernel.org> Link: https://patch.msgid.link/20260512142807.1855619-1-maoyi.xie@ntu.edu.sg Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Sven Eckelmann [Wed, 13 May 2026 07:01:35 +0000 (09:01 +0200)]
batman-adv: tp_meter: avoid use of uninit sender vars
batadv_tp_recv_ack() and batadv_tp_stop() are only valid for tp_vars in the
BATADV_TP_SENDER role. When called with a BATADV_TP_RECEIVER role, it
proceeds to read sender-only members that were never initialized, leading
to undefined behavior.
This can be triggered when a node that is currently acting as a receiver in
an ongoing tp_meter session receives a malicious ACK packet.
Guard against this by checking tp_vars->role immediately after the
lookup and bailing out if it is not BATADV_TP_SENDER, before any of
those members are accessed.
Cc: stable@kernel.org Fixes: 33a3bb4a3345 ("batman-adv: throughput meter implementation") Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Reviewed-by: Yuan Tan <yuantan098@gmail.com> Signed-off-by: Sven Eckelmann <sven@narfation.org>
Sven Eckelmann [Wed, 13 May 2026 07:01:34 +0000 (09:01 +0200)]
batman-adv: dat: handle forward allocation error
batadv_dat_forward_data() calls pskb_copy_for_clone() to duplicate an skb
for each DHT candidate, but does not check the return value before passing
it to batadv_send_skb_prepare_unicast_4addr(). That function dereferences
the skb unconditionally, so a failed allocation triggers a NULL pointer
dereference.
Skip forwarding to the current DHT candidate on allocation failure.
Cc: stable@kernel.org Fixes: 785ea1144182 ("batman-adv: Distributed ARP Table - create DHT helper functions") Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Reviewed-by: Yuan Tan <yuantan098@gmail.com> Signed-off-by: Sven Eckelmann <sven@narfation.org>
Ruijie Li [Thu, 14 May 2026 08:13:25 +0000 (16:13 +0800)]
batman-adv: clear current gateway during teardown
batadv_gw_node_free() removes the gateway list entries during mesh teardown,
but it does not clear the currently selected gateway. This leaves stale
gateway state behind across cleanup and can break a later mesh recreation.
Clear bat_priv->gw.curr_gw before walking the gateway list so the selected
gateway reference is dropped as part of teardown.
Fixes: 2265c1410864 ("batman-adv: gateway election code refactoring") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Ruijie Li <ruijieli51@gmail.com> Signed-off-by: Zhanpeng Li <lzhanpeng2025@lzu.edu.cn> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Sven Eckelmann <sven@narfation.org>
Ruide Cao [Wed, 13 May 2026 03:58:15 +0000 (11:58 +0800)]
batman-adv: fix fragment reassembly length accounting
batman-adv keeps a running payload length for queued fragments and uses it
to validate a fragment chain before reassembly.
That accounting currently allows the accumulated fragment length to be
truncated during updates. As a result, malformed fragment chains can
bypass the intended validation and drive reassembly with inconsistent
length state, leading to a local denial of service.
Fix the accounting by storing the accumulated length in a length-typed
field and rejecting update overflows before the existing validation logic
runs.
The fix was verified against the original reproducer and against valid
fragment reassembly paths.
Fixes: 610bfc6bc99b ("batman-adv: Receive fragmented packets and merge") Cc: stable@kernel.org Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Ruide Cao <caoruide123@gmail.com> Tested-by: Ren Wei <enjou1224z@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Sven Eckelmann <sven@narfation.org>
- ipv6: flowlabel: enforce per-netns limit for unprivileged callers
- tls: fix off-by-one in sg_chain entry count for wrapped sk_msg ring
- smc: avoid NULL deref of conn->lnk in smc_msg_event tracepoint
- sctp: revalidate list cursor after sctp_sendmsg_to_asoc() in SCTP_SENDALL
- batman-adv:
- reject new tp_meter sessions during teardown
- purge non-released claims
- eth:
- i40e: cleanup PTP registration on probe failure
- idpf: fix double free and use-after-free in aux device error paths
- ena: fix potential use-after-free in get_timestamp"
* tag 'net-7.1-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (88 commits)
net: phy: DP83TC811: add reading of abilities
net: tls: prevent chain-after-chain in plain text SG
net: tls: fix off-by-one in sg_chain entry count for wrapped sk_msg ring
net/smc: reject CHID-0 ACCEPT that matches an empty ism_dev slot
macsec: use rcu_work to defer TX SA crypto cleanup out of softirq
macsec: use rcu_work to defer RX SA crypto cleanup out of softirq
macsec: introduce dedicated workqueue for SA crypto cleanup
net: net_failover: Fix the deadlock in slave register
MAINTAINERS: update atlantic driver maintainer
selftests/tc-testing: Add QFQ/CBS qlen underflow test
net/sched: sch_cbs: Call qdisc_reset for child qdisc
FDDI: defza: Sanitise the reset safety timer
net: ethernet: ravb: Do not check URAM suspension when WoL is active
ethtool: fix ethnl_bitmap32_not_zero() bit interval semantics
net/smc: avoid NULL deref of conn->lnk in smc_msg_event tracepoint
net/smc: fix sleep-inside-lock in __smc_setsockopt() causing local DoS
net: atm: fix skb leak in sigd_send() default branch
net: ethtool: phy: avoid NULL deref when PHY driver is unbound
net: atlantic: preserve PCI wake-from-D3 on shutdown when WOL enabled
net: shaper: reject QUEUE scope handle with missing id
...
Linus Torvalds [Thu, 14 May 2026 15:53:24 +0000 (08:53 -0700)]
Merge tag 'audit-pr-20260513' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit
Pull audit fixes from Paul Moore:
- Correctly log the inheritable capabilities
- Honor AUDIT_LOCKED in the AUDIT_TRIM and AUDIT_MAKE_EQUIV commands
* tag 'audit-pr-20260513' of git://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/audit:
audit: enforce AUDIT_LOCKED for AUDIT_TRIM and AUDIT_MAKE_EQUIV
audit: fix incorrect inheritable capability in CAPSET records
Linus Torvalds [Wed, 13 May 2026 18:37:18 +0000 (11:37 -0700)]
ptrace: slightly saner 'get_dumpable()' logic
The 'dumpability' of a task is fundamentally about the memory image of
the task - the concept comes from whether it can core dump or not - and
makes no sense when you don't have an associated mm.
And almost all users do in fact use it only for the case where the task
has a mm pointer.
But we have one odd special case: ptrace_may_access() uses 'dumpable' to
check various other things entirely independently of the MM (typically
explicitly using flags like PTRACE_MODE_READ_FSCREDS). Including for
threads that no longer have a VM (and maybe never did, like most kernel
threads).
It's not what this flag was designed for, but it is what it is.
The ptrace code does check that the uid/gid matches, so you do have to
be uid-0 to see kernel thread details, but this means that the
traditional "drop capabilities" model doesn't make any difference for
this all.
Make it all make a *bit* more sense by saying that if you don't have a
MM pointer, we'll use a cached "last dumpability" flag if the thread
ever had a MM (it will be zero for kernel threads since it is never
set), and require a proper CAP_SYS_PTRACE capability to override.
Shuai Zhang [Mon, 11 May 2026 13:58:37 +0000 (21:58 +0800)]
Bluetooth: hci_qca: Convert timeout from jiffies to ms
Since the timer uses jiffies as its unit rather than ms, the timeout value
must be converted from ms to jiffies when configuring the timer. Otherwise,
the intended 8s timeout is incorrectly set to approximately 33s.
To improve readability, embed msecs_to_jiffies() directly in the macro
definitions and drop the _MS suffix from macros that now yield jiffies
values: MEMDUMP_TIMEOUT, FW_DOWNLOAD_TIMEOUT, IBS_DISABLE_SSR_TIMEOUT,
CMD_TRANS_TIMEOUT, and IBS_BTSOC_TX_IDLE_TIMEOUT.
IBS_WAKE_RETRANS_TIMEOUT_MS and IBS_HOST_TX_IDLE_TIMEOUT_MS are
intentionally left unchanged. Their values are stored in the struct fields
wake_retrans and tx_idle_delay, which hold ms values at runtime and can be
modified via debugfs. The msecs_to_jiffies() conversion happens at each
call site against the field value, so it cannot be embedded in the macro.
Bluetooth: L2CAP: ecred_reconfigure: send packed pdu, not stack pointer
Commit 1c08108f3014 ("Bluetooth: L2CAP: Avoid -Wflex-array-member-not-at-end
warnings") converted the on-stack request PDU in l2cap_ecred_reconfigure()
from an explicit packed struct to DEFINE_RAW_FLEX(), but did not adjust the
size and source-pointer arguments to l2cap_send_cmd():
After the conversion, DEFINE_RAW_FLEX() expands to declare an anonymous
union pdu_u plus a local pointer "pdu" pointing at it. Therefore:
- sizeof(pdu) is now sizeof(struct l2cap_ecred_reconf_req *) = 8 on
64-bit (4 on 32-bit), not the 6 bytes of (mtu, mps, scid[1]).
- &pdu is the address of the local pointer's stack storage, not the
address of the request payload.
l2cap_send_cmd() forwards (data, count) to l2cap_build_cmd(), which calls
skb_put_data(skb, data, count). The L2CAP_ECRED_RECONFIGURE_REQ packet
body therefore contains 8 bytes copied from the kernel stack starting at
&pdu -- the 8 bytes overlap the pdu pointer's value, leaking a kernel
stack address to the paired Bluetooth peer. The intended (mtu, mps, scid)
fields are not transmitted at all, so the peer rejects the request as
malformed and the L2CAP_ECRED_RECONFIGURE feature itself has been broken
for the local-side initiator since the introducing commit landed.
The sibling site l2cap_ecred_conn_req() in the same commit was converted
correctly (sizeof(*pdu) + len, pdu); only this site was missed.
Restore the original semantics: pass the full flex-struct size via
struct_size(pdu, scid, 1) and the pdu pointer (the struct address) as
the source.
Validated on a stock 7.0-based host kernel via the real call path:
setsockopt(SOL_BLUETOOTH, BT_RCVMTU, ...) on a BT_CONNECTED
L2CAP_MODE_EXT_FLOWCTL socket emits an L2CAP_ECRED_RECONFIGURE_REQ
whose body is 8 bytes (the on-stack pdu local's value) rather than
the expected 6. Three captures from fresh socket / fresh hciemu peer
on the same host -- low bytes vary per call, high 0xffff confirms a
kernel virtual address (KASLR-randomised stack slot, not a fixed
string):
RECONF_REQ body (ident=0x02 len=8): 42 fb 54 af 0e ca ff ff
RECONF_REQ body (ident=0x02 len=8): 52 3d 2e af 0e ca ff ff
RECONF_REQ body (ident=0x02 len=8): b2 fc 5b af 0e ca ff ff
After this patch the body is 6 bytes carrying the expected
little-endian (mtu, mps, scid).
Cc: stable@vger.kernel.org Fixes: 1c08108f3014 ("Bluetooth: L2CAP: Avoid -Wflex-array-member-not-at-end warnings") Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Pauli Virtanen [Fri, 24 Apr 2026 19:24:29 +0000 (22:24 +0300)]
Bluetooth: btmtk: accept too short WMT FUNC_CTRL events
MT7925 (USB ID 0e8d:e025) on fw version 20260106153314 sends WMT
FUNC_CTRL events that are missing the status field.
Prior to commit 006b9943b982 ("Bluetooth: btmtk: validate WMT event SKB
length before struct access") the status was read from out-of-bounds of
SKB data, which usually would result to success with
BTMTK_WMT_ON_UNDONE, although I don't know the intent here. The bounds
check added in that commit returns with error instead, producing
"Bluetooth: hci0: Failed to send wmt func ctrl (-22)" and makes the
device unusable.
Fix the regression by interpreting too short packet as status
BTMTK_WMT_ON_UNDONE, which makes the device work normally again.
Fixes: 634a4408c061 ("Bluetooth: btmtk: validate WMT event SKB length before struct access") Signed-off-by: Pauli Virtanen <pav@iki.fi> Tested-by: Mikhail Gavrilov <mikhail.v.gavrilov@gmail.com> # MT7922 (0489:e0e2) Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Jiexun Wang [Wed, 6 May 2026 11:43:30 +0000 (19:43 +0800)]
Bluetooth: serialize accept_q access
bt_sock_poll() walks the accept queue without synchronization, while
child teardown can unlink the same socket and drop its last reference.
The unsynchronized accept queue walk has existed since the initial
Bluetooth import.
Protect accept_q with a dedicated lock for queue updates and polling.
Also rework bt_accept_dequeue() to take temporary child references under
the queue lock before dropping it and locking the child socket.
Fixes: 1da177e4c3f41524e886b7f1b8a0c1fc7321cac2 ("Linux-2.6.12-rc2") Cc: stable@vger.kernel.org Reported-by: Jann Horn <jannh@google.com> Reported-by: Yuan Tan <yuantan098@gmail.com> Reported-by: Yifan Wu <yifanwucs@gmail.com> Reported-by: Juefei Pu <tomapufckgml@gmail.com> Reported-by: Xin Liu <bird@lzu.edu.cn> Signed-off-by: Jiexun Wang <wangjiexun2025@gmail.com> Signed-off-by: Ren Wei <n05ec@lzu.edu.cn> Signed-off-by: Jiexun Wang <wangjiexun2025@gmail.com> Reviewed-by: Jann Horn <jannh@google.com> Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Sven Schuchmann [Tue, 12 May 2026 07:19:47 +0000 (09:19 +0200)]
net: phy: DP83TC811: add reading of abilities
At this time the driver is not listing any speeds
it supports. This should be ETHTOOL_LINK_MODE_100baseT1_Full_BIT
for DP83TC811. Add the missing call for phylib to read the abilities.
Fixes: b753a9faaf9a ("net: phy: DP83TC811: Introduce support for the DP83TC811 phy") Suggested-by: Andrew Lunn <andrew@lunn.ch> Signed-off-by: Sven Schuchmann <schuchmann@schleissheimer.de> Reviewed-by: Andrew Lunn <andrew@lunn.ch> Link: https://patch.msgid.link/20260512071949.6218-1-schuchmann@schleissheimer.de
[pabeni@redhat.com: dropped revision history] Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Jakub Kicinski [Mon, 11 May 2026 17:49:18 +0000 (10:49 -0700)]
net: tls: prevent chain-after-chain in plain text SG
Sashiko points out that if end = 0 (start != 0) the current
code will create a chain link to content type right after
the wrap link:
This would create a chain where the wrap link points directly
to another chain link. The scatterlist API sg_next iterator
does not recursively resolve consecutive chain links.
meaning this is illegal input to crypto.
The wrapping link is unnecessary if end = 0. end is the entry after
the last one used so end = 0 means there's nothing pushed after
the wrap:
end start i
v v v
[ ]...[ ][ d ][ d ][ d ][ d ][rsv for wrap]
Skip the wrapping in this case.
TLS 1.3 can use the "wrapping slot" for it's chaining if end = 0.
This avoids the chain-after-chain.
Move the wrap chaining before marking END and chaining off content
type, that feels like more logical ordering to me, but should not
matter from functional perspective.
Reported-by: Sashiko <sashiko-bot@kernel.org> Fixes: 9aaaa56845a0 ("bpf: Sockmap/tls, skmsg can have wrapped skmsg that needs extra chaining") Signed-off-by: Jakub Kicinski <kuba@kernel.org> Link: https://patch.msgid.link/20260511174920.433155-3-kuba@kernel.org Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Jakub Kicinski [Mon, 11 May 2026 17:49:17 +0000 (10:49 -0700)]
net: tls: fix off-by-one in sg_chain entry count for wrapped sk_msg ring
When an sk_msg scatterlist ring wraps (sg.end < sg.start),
tls_push_record() chains the tail portion of the ring to the head
using sg_chain(). An extra entry in the sg array is reserved for
this:
struct sk_msg_sg {
[...]
/* The extra two elements:
* 1) used for chaining the front and sections when the list becomes
* partitioned (e.g. end < start). The crypto APIs require the
* chaining;
* 2) to chain tailer SG entries after the message.
*/
struct scatterlist data[MAX_MSG_FRAGS + 2];
The current code uses MAX_SKB_FRAGS + 1 as the ring size:
instead of the true last entry. This is likely due to a "race" of
the commit under Fixes landing close to
commit 031097d9e079 ("bpf: sk_msg, zap ingress queue on psock down")
Convert to ARRAY_SIZE and drop the data[start] / - start (as suggested
by Sabrina).
Reported-by: 钱一铭 <yimingqian591@gmail.com> Fixes: 9aaaa56845a0 ("bpf: Sockmap/tls, skmsg can have wrapped skmsg that needs extra chaining") Signed-off-by: Jakub Kicinski <kuba@kernel.org> Reviewed-by: Sabrina Dubroca <sd@queasysnail.net> Link: https://patch.msgid.link/20260511174920.433155-2-kuba@kernel.org Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Xiang Mei [Mon, 11 May 2026 06:21:38 +0000 (23:21 -0700)]
net/smc: reject CHID-0 ACCEPT that matches an empty ism_dev slot
On the SMC-D client, slot 0 of ini->ism_dev[]/ini->ism_chid[] is
reserved for an SMC-Dv1 device. smc_find_ism_v2_device_clnt()
populates V2 entries starting at index 1, so when no V1 device is
selected slot 0 is left in its kzalloc()'ed state with ism_dev[0] ==
NULL and ism_chid[0] == 0.
smc_v2_determine_accepted_chid() then matches the peer's CHID against
the array starting from index 0 using the CHID alone. A malicious
peer replying to a SMC-Dv2-only proposal with d1.chid == 0 matches
the empty slot, ini->ism_selected becomes 0, and the subsequent
ism_dev[0]->lgr_lock dereference in smc_conn_create() faults at
offsetof(struct smcd_dev, lgr_lock) == 0x68:
BUG: KASAN: null-ptr-deref in _raw_spin_lock_bh+0x79/0xe0
Write of size 4 at addr 0000000000000068 by task exploit/144
Call Trace:
_raw_spin_lock_bh
smc_conn_create (net/smc/smc_core.c:1997)
__smc_connect (net/smc/af_smc.c:1447)
smc_connect (net/smc/af_smc.c:1720)
__sys_connect
__x64_sys_connect
do_syscall_64
Require ism_dev[i] to be non-NULL before accepting a CHID match.
Fixes: a7c9c5f4af7f ("net/smc: CLC accept / confirm V2") Reported-by: Weiming Shi <bestswngs@gmail.com> Assisted-by: Claude:claude-opus-4-7 Signed-off-by: Xiang Mei <xmei5@asu.edu> Link: https://patch.msgid.link/20260511062138.2839584-1-xmei5@asu.edu Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
macsec: use rcu_work to fix crypto cleanup in softirq context
From: Jinliang Zheng <alexjlzheng@tencent.com>
crypto_free_aead() can internally call vunmap() (e.g. via dma_free_attrs()
in hardware crypto drivers like hisi_sec2), which must not be invoked from
softirq context. Both free_rxsa() and free_txsa() are RCU callbacks that
run in softirq, causing a kernel crash on affected hardware.
This series fixes the issue by deferring the actual cleanup to a workqueue
using rcu_work, which combines the RCU grace period and workqueue dispatch
into a single primitive.
Two design decisions worth noting:
1. rcu_work instead of schedule_work() + synchronize_rcu()
An alternative would be to call schedule_work() directly from
macsec_rxsa_put()/macsec_txsa_put(), then call synchronize_rcu() at
the start of the work handler to replace the grace period previously
provided by call_rcu(). However, synchronize_rcu() blocks the worker
thread for the duration of a full RCU grace period. Under high SA
churn (e.g. tearing down an interface with many SAs), each SA would
occupy a worker thread while waiting, and multiple concurrent calls
cannot share the same grace period — leading to unnecessary latency
and resource waste.
rcu_work uses call_rcu_hurry() internally, which is fully asynchronous:
the worker thread is only dispatched after the grace period has elapsed,
and multiple concurrent queue_rcu_work() calls naturally batch under the
same grace period via the RCU subsystem's existing coalescing mechanism.
2. Dedicated workqueue instead of system_wq
Using a dedicated workqueue (macsec_wq) allows macsec_exit() to drain
exactly the work items belonging to this module — by calling
destroy_workqueue() after rcu_barrier(). If system_wq were used,
flush_scheduled_work() would drain all pending work items across the
entire system, creating unnecessary coupling with unrelated subsystems
and potentially causing unexpected delays. The dedicated workqueue
provides a clean, contained teardown path.
====================
Jinliang Zheng [Mon, 11 May 2026 15:31:00 +0000 (23:31 +0800)]
macsec: use rcu_work to defer TX SA crypto cleanup out of softirq
free_txsa() is an RCU callback running in softirq context, but calls
crypto_free_aead() which can invoke vunmap() internally on hardware
crypto drivers (e.g. hisi_sec2), triggering a kernel crash.
Use rcu_work to defer the cleanup to a workqueue, for the same reasons
as the analogous fix to free_rxsa() in the previous patch.
Jinliang Zheng [Mon, 11 May 2026 15:30:59 +0000 (23:30 +0800)]
macsec: use rcu_work to defer RX SA crypto cleanup out of softirq
crypto_free_aead() can internally invoke vunmap() (e.g. via
dma_free_attrs() in hardware crypto drivers such as hisi_sec2).
vunmap() must not be called from softirq context, but free_rxsa()
is an RCU callback that runs in softirq, leading to a kernel crash:
Use rcu_work to defer the cleanup to a workqueue. rcu_work dispatches
the worker asynchronously after the RCU grace period, so no thread
blocks waiting, and concurrent releases of multiple SAs naturally
share the same grace period.
Jinliang Zheng [Mon, 11 May 2026 15:30:58 +0000 (23:30 +0800)]
macsec: introduce dedicated workqueue for SA crypto cleanup
Introduce a dedicated ordered workqueue, macsec_wq, which will be used
by subsequent patches to defer SA crypto cleanup (crypto_free_aead and
related teardown) out of softirq context.
Using a dedicated workqueue instead of system_wq allows macsec_exit()
to drain exactly the work items belonging to this module via
destroy_workqueue(), without interfering with unrelated work items on
system_wq or causing unexpected delays elsewhere.
rcu_barrier() in macsec_exit() ensures all in-flight rcu_work callbacks
have enqueued their work items before destroy_workqueue() drains and
destroys the queue, making the two-step teardown correct and complete.
The same sequence is kept in the error path of macsec_init() as a
precaution, to mirror macsec_exit() and stay safe if work ever becomes
queueable before this point in the future.
While at it, rename the error labels in macsec_init() from the
resource-named style (rtnl:, notifier:, wq:) to the err_xxx: style
(err_rtnl:, err_notifier:, err_destroy_wq:) to align with the broader
kernel convention.
Faicker Mo [Mon, 11 May 2026 14:05:51 +0000 (22:05 +0800)]
net: net_failover: Fix the deadlock in slave register
There is netdev_lock_ops() before the NETDEV_REGISTER notifier
in register_netdevice(), so use the non-locking functions
in net_failover_slave_register().
failover_slave_register() in failover_existing_slave_register() adds lock
and unlock ops too.
Fixes: 4c975fd70002 ("net: hold instance lock during NETDEV_REGISTER/UP") Signed-off-by: Faicker Mo <faicker.mo@gmail.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Victor Nogueira [Mon, 11 May 2026 18:30:58 +0000 (14:30 -0400)]
selftests/tc-testing: Add QFQ/CBS qlen underflow test
Since CBS was not calling reset for its child qdisc, there are scenarios
where it could cause an underflow on its parent's qlen/backlog. When the
parent is QFQ, a null-ptr deref could occur.
Add a test case that reproduces the underflow followed by a null-ptr
deref scenario.
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: Victor Nogueira <victor@mojatatu.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Jamal Hadi Salim [Mon, 11 May 2026 18:30:57 +0000 (14:30 -0400)]
net/sched: sch_cbs: Call qdisc_reset for child qdisc
During a reset, CBS is not calling reset on its child qdisc, which
might cause qlen/backlog accounting issues. For example, if we have CBS
with a QFQ parent and a netem child with delay, we can create a scenario
where the parent's qlen underflows. QFQ, specifically, uses qlen to
check whether it should deference a pointer, so this scenario may cause
a null-ptr deref in QFQ:
Fix this by calling qdisc_reset for CBS' child qdisc.
Sashiko caught an issue which could result in a null ptr deref if
qdisc_create_dflt() is invoked on an unitialised cbs qdisc which is exposed
by this patch. We add an early return if the qdisc is null to address this.
This is a similar approach used by two other fixes[1][2].
The proper fix for this specific issue elucidated by sashiko is to remove
the call to qdisc_reset when qdisc_create_dflt fails. Since the dflt qdisc
isn't attached anywhere yet at that point, calling the reset callback doesn't
make much sense (and as stated has been a source of two other bugs).
We plan on submitting this fix in a later patch.
[1] https://lore.kernel.org/netdev/20221018063201.306474-2-shaozhengchao@huawei.com/
[2] https://lore.kernel.org/netdev/20221018063201.306474-4-shaozhengchao@huawei.com/
Fixes: 585d763af09c ("net/sched: Introduce Credit Based Shaper (CBS) qdisc") Reported-by: Junyoung Jang <graypanda.inzag@gmail.com> Tested-by: Junyoung Jang <graypanda.inzag@gmail.com> Tested-by: Victor Nogueira <victor@mojatatu.com> Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com> Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The reset actions of the DEFZA adapters are exceedingly slow, taking up
to 30 seconds to complete by the device spec and typically in the range
of 10 seconds in reality, as required for the device RTOS to boot, still
quite a lot. Therefore a state machine is used that's interrupt driven,
however a safety mechanism is required in case of adapter malfunction,
so that if no state change interrupt has arrived in time, then the
situation is taken care of.
The safety mechanism depends on the origin of the reset. For regular
adapter initialisation at the device probe time a sleep is requested.
However a reset is also required by the device spec when the adapter has
transitioned into the halted state, such as in response to a PC Trace
event in the course of ring fault recovery, possibly a common network
event. In that case no sleep is possible as a device halt is reported
at the hardirq level.
A timer is therefore set up to ensure progress in case no adapter state
change interrupt has arrived in time, but as from commit 168f6b6ffbee
("timers: Use del_timer_sync() even on UP") a warning is issued as the
timer is deleted in the hardirq handler upon an expected state change:
The immediate origin of the new warning is the switch away from aliasing
del_timer_sync() to del_timer() (timer_delete_sync() to timer_delete()
in terms of current function names) for UP configurations, which however
is the only choice for this driver anyway as no SMP hardware supports
the TURBOchannel bus this device interfaces to. Therefore there is a
very remote issue only this is a sign of.
Specifically if an adapter reset issued upon a transition to the halted
state times out and first triggers fza_reset_timer() for another reset
assertion, which then schedules fza_reset_timer() for reset deassertion
and then that second call is pre-empted after poking at the hardware,
but before the timer has been rearmed and owing to high system load
causing exceedingly high scheduling latency control is not handed back
before a transition to the uninitialised state has caused the timer to
be deleted even before it has been started, then fza_reset_timer() will
be called yet again and issue another reset even though by then the
adapter has already recovered.
Prevent this situation from happening by switching to timer_delete() for
the transition to the halted state and protect the code region affected
with a spinlock, also to make sure add_timer() has not been called twice
in a row due to an execution race between the interrupt handler and the
timer handler (though it could only happen on SMP, but let's keep the
driver clean). It's a very unlikely sequence of events to happen and
therefore there's no point in trying to be overly clever about it, such
as by placing printk() calls outside the protection. For the transition
to the uninitialised state switch to timer_delete_sync_try() instead, so
that a timer isn't deleted that's just been rearmed by the timer handler
and needs to watch for the device to come out of reset again (again, an
SMP scenario only).
Retain timer_delete_sync() invocations outside the hardirq context for a
stray timer not to fire once device structures have been released.
Fixes: 61414f5ec9834 ("FDDI: defza: Add support for DEC FDDIcontroller 700 TURBOchannel adapter") Signed-off-by: Maciej W. Rozycki <macro@orcam.me.uk> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Linus Torvalds [Wed, 13 May 2026 22:00:40 +0000 (15:00 -0700)]
Merge tag 'sched_ext-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext
Pull sched_ext fixes from Tejun Heo:
"The bulk of this is hardening of the new sub-scheduler infrastructure.
- UAFs and lifecycle bugs on the sub-sched attach/detach paths:
parent sub_kset freed under a racing child, list_del_rcu on an
uninitialized list head, ops->priv stomped by concurrent
attach/detach, and a UAF in the init-failure error path
- Task state-machine reorg closing concurrent enable-vs-dead races: a
task exiting during the unlocked init window could trip NULL ops
derefs or skip exit_task() cleanup
- A scx_link_sched() self-deadlock on scx_sched_lock
- isolcpus: stop dereferencing the now-RCU-protected HK_TYPE_DOMAIN
cpumask without RCU, and stop rejecting BPF schedulers when only
cpuset isolated partitions are active
- PREEMPT_RT: disable irq_work runs in hardirq context so dumps show
the failing task rather than the irq_work kthread
- Assorted !CONFIG_EXT_SUB_SCHED, randconfig, and selftest build
fixes"
* tag 'sched_ext-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext:
sched_ext: Use HK_TYPE_DOMAIN_BOOT to detect isolcpus= domain isolation
sched_ext: Defer sub_kset base put to scx_sched_free_rcu_work
sched_ext: INIT_LIST_HEAD() &sch->all in scx_alloc_and_add_sched()
sched_ext: Drop NONE early return in scx_disable_and_exit_task()
sched_ext: Avoid UAF in scx_root_enable_workfn() init failure path
sched_ext: Clear ops->priv on scx_alloc_and_add_sched() error paths
sched_ext: Fix ops->priv clobber on concurrent attach/detach
selftests/sched_ext: Fix build error in dequeue selftest
sched_ext: Handle SCX_TASK_NONE in disable/switched_from paths
sched_ext: Close sub-sched init race with post-init DEAD recheck
sched_ext: Close root-enable vs sched_ext_dead() race with SCX_TASK_INIT_BEGIN
sched_ext: Replace SCX_TASK_OFF_TASKS flag with SCX_TASK_DEAD state
sched_ext: Inline scx_init_task() and move RESET_RUNNABLE_AT into scx_set_task_state()
sched_ext: Cleanups in preparation for the SCX_TASK_INIT_BEGIN/DEAD work
sched_ext: Use IRQ_WORK_INIT_HARD() to initialize sch->disable_irq_work
sched_ext: Fix !CONFIG_EXT_SUB_SCHED build warnings
sched_ext: Drop unused scx_find_sub_sched() stub
sched_ext: Move scx_error() out of scx_link_sched()'s lock region
Linus Torvalds [Wed, 13 May 2026 21:56:31 +0000 (14:56 -0700)]
Merge tag 'cgroup-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup
Pull cgroup fixes from Tejun Heo:
- cpuset fixes:
- Partition invalidation could return CPUs still in use by sibling
partitions, producing overlapping effective_cpus
- cpuset_can_attach() over-reserved DL bandwidth on moves that
stayed within the same root domain
- Pending DL migration state leaked into later attaches when a
later can_attach() check failed
- Reorder PF_EXITING and __GFP_HARDWALL checks so dying tasks can
allocate from any node and exit quickly
- dmem: propagate -ENOMEM instead of spinning forever when the fallback
pool allocation also fails
- selftests/cgroup: percpu test error-path leak, bogus numeric
comparison of cpuset strings, and a zero-length read() that silently
passed OOM-kill tests
* tag 'cgroup-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup/cpuset: Return only actually allocated CPUs during partition invalidation
selftests/cgroup: Fix error path leaks in test_percpu_basic
cgroup/cpuset: Reserve DL bandwidth only for root-domain moves
cgroup/cpuset: Reset DL migration state on can_attach() failure
selftests/cgroup: Fix string comparison in write_test
selftests/cgroup: Fix cg_read_strcmp() empty string comparison
cgroup/dmem: Return -ENOMEM on failed pool preallocation
cgroup/cpuset: move PF_EXITING check before __GFP_HARDWALL in cpuset_current_node_allowed()
Linus Torvalds [Wed, 13 May 2026 21:49:13 +0000 (14:49 -0700)]
Merge tag 'wq-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq
Pull workqueue fixes from Tejun Heo:
- Plug a wq->cpu_pwq leak on the WQ_UNBOUND allocation failure path
- Fix a cancel_delayed_work_sync() livelock against drain_workqueue()
caused by the drain/destroy reject path leaving WORK_STRUCT_PENDING
set with no owner
* tag 'wq-for-7.1-rc3-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
workqueue: Fix wq->cpu_pwq leak in alloc_and_link_pwqs() WQ_UNBOUND path
workqueue: Release PENDING in __queue_work() drain/destroy reject path
Andrea Righi [Wed, 13 May 2026 11:24:38 +0000 (13:24 +0200)]
sched_ext: Use HK_TYPE_DOMAIN_BOOT to detect isolcpus= domain isolation
scx_enable() refuses to attach a BPF scheduler when isolcpus=domain is
in effect by comparing housekeeping_cpumask(HK_TYPE_DOMAIN) against
cpu_possible_mask.
Since commit 27c3a5967f05 ("sched/isolation: Convert housekeeping
cpumasks to rcu pointers"), HK_TYPE_DOMAIN's cpumask is RCU protected
and dereferencing it requires either RCU read lock, the cpu_hotplug
write lock, or the cpuset lock; scx_enable() holds none of these, so
booting with isolcpus=domain and attaching any BPF scheduler triggers
the following lockdep splat:
In addition, commit 03ff73510169 ("cpuset: Update HK_TYPE_DOMAIN cpumask
from cpuset") made HK_TYPE_DOMAIN include cpuset isolated partitions as
well, which means the current check also rejects BPF schedulers when a
cpuset partition is active. That contradicts the original intent of
commit 9f391f94a173 ("sched_ext: Disallow loading BPF scheduler if
isolcpus= domain isolation is in effect"), which explicitly noted that
cpuset partitions are honored through per-task cpumasks and should not
be rejected.
Switch to housekeeping_enabled(HK_TYPE_DOMAIN_BOOT), which reads only
the housekeeping flag bit (no RCU dereference) and reflects exactly the
boot-time isolcpus= configuration that the error message refers to.
where xcpus = user_xcpus(cs) which returns cs->exclusive_cpus (if set)
or cs->cpus_allowed. When exclusive_cpus is not set, user_xcpus(cs) can
contain CPUs that were never actually granted to the partition due to
sibling exclusion in compute_excpus(). Consequently, the invalidation
may return CPUs to the parent that remain in use by sibling partitions,
causing overlapping effective_cpus and triggering the
WARN_ON_ONCE(1) in generate_sched_domains().
Use cs->effective_xcpus instead, which reflects the CPUs actually
granted to this partition.
# b1 becomes partition root with CPUs 1-2, but sibling exclusion
# reduces its effective_xcpus to CPU 2 only
echo "1-2" > b1/cpuset.cpus
echo "root" > b1/cpuset.cpus.partition
Linus Torvalds [Wed, 13 May 2026 18:53:51 +0000 (11:53 -0700)]
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"arm64:
- Add the pKVM side of the workaround for ARM's erratum 4193714,
provided that the EL3 firmware does its part of the job. KVM will
refuse to initialise otherwise
- Correctly handle 52bit VAs for guest EL2 stage-1 translations when
running under NV with E2H==0
- Correctly deal with permission faults in guest_memfd memslots
- Fix the steal-time selftest after the infrastructure was reworked
- Make sure the host cannot pass a non-sensical clock update to the
EL2 tracing infrastructure
- Appoint Steffen Eiden as a reviewer in anticipation of the KVM/s390
ability to run arm64 guests, which will inevitably lead to arm64
code being directly used on s390
- Make sure that EL2 is configured with both exception entry and exit
being Context Synchronization Events
- Handle the current vcpu being NULL on EL2 panic
- Fix the selftest_vcpu memcache being empty at the point of donation
or sharing
- Check that the memcache has enough capacity before engaging on the
share/donate path
- Fix __deactivate_fgt() to use its parameter rather than a variable
in the macro context
s390:
- Fix array overrun with large amounts of PCI devices
x86:
- Never use L0's PAUSE loop exiting while L2 is running, since it's
unlikely that a nested guest will help solving the hypervisor's
spinlock contention
- Fix emulation of MOVNTDQA
- Fix typo in Xen hypercall tracepoint
- Add back an optimization that was left behind when recently fixing
a bug
- Add module parameter to disable CET, whose implementation seems to
have issues. For now it remains enabled by default
Generic:
- Reject offset causing an unsigned overflow in kvm_reset_dirty_gfn()
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (22 commits)
KVM: VMX: introduce module parameter to disable CET
KVM: x86: Swap the dst and src operand for MOVNTDQA
KVM: x86: use again the flush argument of __link_shadow_page()
KVM: selftests: Ensure gmem file sizes are multiple of host page size
Documentation: kvm: update links in the references section of AMD Memory Encryption
KVM: nSVM: Never use L0's PAUSE loop exiting while L2 is running
KVM: x86: Fix Xen hypercall tracepoint argument assignment
KVM: Reject wrapped offset in kvm_reset_dirty_gfn()
KVM: arm64: Pre-check vcpu memcache for host->guest donate
KVM: arm64: Pre-check vcpu memcache for host->guest share
KVM: arm64: Seed pkvm_ownership_selftest vcpu memcache
KVM: arm64: Fix __deactivate_fgt macro parameter typo
KVM: arm64: Guard against NULL vcpu on VHE hyp panic path
KVM: arm64: Make EL2 exception entry and exit context-synchronization events
MAINTAINERS: Add Steffen as reviewer for KVM/arm64
KVM: arm64: Remove potential UB on nvhe tracing clock update
KVM: selftests: arm64: Fix steal_time test after UAPI refactoring
KVM: arm64: Handle permission faults with guest_memfd
KVM: arm64: nv: Consider the DS bit when translating TCR_EL2
KVM: arm64: Work around C1-Pro erratum 4193714 for protected guests
...
Yu Miao [Wed, 13 May 2026 02:39:07 +0000 (10:39 +0800)]
selftests/cgroup: Fix error path leaks in test_percpu_basic
When cg_name_indexed() returns NULL partway through the child creation
loop, the code returned -1 without running cleanup_children and cleanup.
That left the `parent` pathname allocation unreleased and did not remove
child cgroup directories already created under the parent. Fix by jumping
to cleanup_children instead of returning.
When cg_create() fails, `child` (the pathname from cg_name_indexed())
was not freed before cleanup_children. Fix by freeing `child` before
branching to cleanup_children.
Paolo Bonzini [Tue, 12 May 2026 14:58:48 +0000 (16:58 +0200)]
KVM: VMX: introduce module parameter to disable CET
There have been reports of host hangs caused by CET virtualization.
Until these are analyzed further, introduce a module parameter that
makes it possible to easily disable it.
Niklas Söderlund [Sun, 10 May 2026 10:30:17 +0000 (12:30 +0200)]
net: ethernet: ravb: Do not check URAM suspension when WoL is active
When updating the driver to match latest datasheet to suspend access to
URAM when suspending DMA transfers a corner-case was missed, URAM access
will not be suspended if WoL is enabled. This lead to the error message
(correctly) being triggered as URAM access is not suspended even tho
it's requested as part of stopping DMA.
Avoid checking if URAM access is suspended and printing the error
message if WoL is enabled when we suspend the system, as we know it will
not be.
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org> Closes: https://lore.kernel.org/all/CAMuHMdWnjV%3DHGE1o08zLhUfTgOSene5fYx1J5GG10mB%2BToq8qg@mail.gmail.com/ Fixes: 353d8e7989b6 ("net: ethernet: ravb: Suspend and resume the transmission flow") Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se> Reviewed-by: Sai Krishna <saikrishnag@marvell.com> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Chenguang Zhao [Mon, 11 May 2026 01:43:43 +0000 (09:43 +0800)]
ethtool: fix ethnl_bitmap32_not_zero() bit interval semantics
ethnl_bitmap32_not_zero() should return true if some bit in [start, end)
is set:
- Fix inverted memchr_inv() sense: return true when the scan finds a
non-zero byte, not when the middle words are all zero.
- Return false for an empty interval (end <= start).
- When end is 32-bit aligned, indices in [start, end) do not include any
bits from map[end_word]; return false after earlier checks found no
non-zero data.
Xiang Mei [Sun, 10 May 2026 22:26:40 +0000 (15:26 -0700)]
net/smc: avoid NULL deref of conn->lnk in smc_msg_event tracepoint
The smc_msg_event tracepoint class, shared by smc_tx_sendmsg and
smc_rx_recvmsg, unconditionally dereferences smc->conn.lnk:
__string(name, smc->conn.lnk->ibname)
conn->lnk is only set for SMC-R; for SMC-D it is NULL. Other code on
these paths already handles this (e.g. !conn->lnk in
SMC_STAT_RMB_TX_SIZE_SMALL()). With the tracepoint enabled, the first
sendmsg()/recvmsg() on an SMC-D socket crashes:
Oops: general protection fault, probably for non-canonical address
KASAN: null-ptr-deref in range [...]
RIP: 0010:strlen+0x1e/0xa0
Call Trace:
trace_event_raw_event_smc_msg_event (net/smc/smc_tracepoint.h:44)
smc_rx_recvmsg (net/smc/smc_rx.c:515)
smc_recvmsg (net/smc/af_smc.c:2859)
__sys_recvfrom (net/socket.c:2315)
__x64_sys_recvfrom (net/socket.c:2326)
do_syscall_64
The faulting address 0x3e0 is offsetof(struct smc_link, ibname),
confirming the NULL ->lnk deref. Enabling the tracepoint requires
root, but the trigger itself is unprivileged: socket(AF_SMC, ...) has
no capability check, and SMC-D negotiation needs no admin step on
s390 or on x86 with the loopback ISM device loaded.
Log an empty device name for SMC-D instead of dereferencing NULL.
Fixes: aff3083f10bf ("net/smc: Introduce tracepoints for tx and rx msg") Reported-by: Weiming Shi <bestswngs@gmail.com> Signed-off-by: Xiang Mei <xmei5@asu.edu> Reviewed-by: Dust Li <dust.li@linux.alibaba.com> Reviewed-by: Sidraya Jayagond <sidraya@linux.ibm.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Nicolò Coccia [Sun, 10 May 2026 16:34:13 +0000 (12:34 -0400)]
net/smc: fix sleep-inside-lock in __smc_setsockopt() causing local DoS
A logic flaw in __smc_setsockopt() allows a local unprivileged user to
cause a Denial of Service (DoS) by holding the socket lock indefinitely.
The function __smc_setsockopt() calls copy_from_sockptr() while holding
lock_sock(sk). By passing a userfaultfd-monitored memory page (or
FUSE-backed memory on systems where unprivileged userfaultfd is disabled)
as the optval, an attacker can halt execution during the copy operation,
keeping the lock held.
Combined with asynchronous tear-down operations like shutdown(), this
exhausts the kernel wq (kworkers) and triggers the hung task watchdog.
[ 240.123456] INFO: task kworker/u8:2 blocked for more than 120 seconds.
[ 240.123489] Call Trace:
[ 240.123501] smc_shutdown+...
[ 240.123512] lock_sock_nested+...
This patch moves the user-space copy outside the lock_sock() critical
section to prevent the issue.
Fixes: a6a6fe27bab4 ("net/smc: Dynamic control handshake limitation by socket options") Signed-off-by: Nicolò Coccia <n.coccia96@gmail.com> Reviewed-by: Dust Li <dust.li@linux.alibaba.com> Tested-by: Dust Li <dust.li@linux.alibaba.com> Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Wei Yang [Sat, 9 May 2026 12:23:58 +0000 (20:23 +0800)]
net: atm: fix skb leak in sigd_send() default branch
The default branch in sigd_send() calls sock_put() and returns -EINVAL
without freeing the skb, while all other exit paths do so. Add the
missing dev_kfree_skb() before sock_put() to fix the leak.
phy_remove() clears phydev->drv but doesn't call phy_detach(), so the
phy_device stays in the link topology xarray and ethnl_req_get_phydev()
still hands it back. ETHTOOL_MSG_PHY_GET then oopses on:
drvname is already treated as optional by phy_reply_size(),
phy_fill_reply() and phy_cleanup_data(), so just skip the allocation
when there is no driver bound.
Fixes: 9dd2ad5e92b9 ("net: ethtool: phy: Convert the PHY_GET command to generic phy dump") Cc: stable@vger.kernel.org # 6.13.x Signed-off-by: David Carlier <devnexen@gmail.com> Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com> Tested-by: Maxime Chevallier <maxime.chevallier@bootlin.com> Link: https://patch.msgid.link/20260509215046.107157-1-devnexen@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Zoran Ilievski [Mon, 11 May 2026 06:40:02 +0000 (08:40 +0200)]
net: atlantic: preserve PCI wake-from-D3 on shutdown when WOL enabled
The shutdown handler aq_pci_shutdown() unconditionally calls
pci_wake_from_d3(pdev, false), clearing the PCI PME_En bit even when
wake-on-LAN has been configured. While aq_nic_shutdown() correctly
programs the NIC firmware via aq_nic_set_power() to listen for magic
packets, the PCI subsystem will not propagate the resulting PME wake
event from D3, so the system never wakes after poweroff.
WOL from suspend (S3) is unaffected because aq_suspend_common() does
not touch pci_wake_from_d3() and relies on the PM core's wake
configuration via device_may_wakeup().
This affects all atlantic-supported NICs (AQC107/108/111/112/113);
users have reported that WOL works if the atlantic driver is never
loaded, but breaks once it has run its shutdown path.
Pass the configured WOL state to pci_wake_from_d3() instead of a
literal false, so the PCI PME_En bit is preserved when the user has
armed WOL via ethtool.
Tejun Heo [Mon, 11 May 2026 23:18:23 +0000 (13:18 -1000)]
sched_ext: Defer sub_kset base put to scx_sched_free_rcu_work
scx_sub_enable_workfn() pins parent->kobj before dropping scx_sched_lock,
but that does not pin parent->sub_kset. Concurrent disable can
kset_unregister and free sub_kset before scx_alloc_and_add_sched()
dereferences it.
Split sub_kset teardown: kobject_del() at disable keeps sysfs removal; defer
kobject_put() to scx_sched_free_rcu_work so the memory survives. A racing
child sees state_in_sysfs=0 with valid memory, sysfs_create_dir() fails, and
the existing exit_kind gate in scx_link_sched() turns it away with -ENOENT.
Tejun Heo [Mon, 11 May 2026 23:18:19 +0000 (13:18 -1000)]
sched_ext: INIT_LIST_HEAD() &sch->all in scx_alloc_and_add_sched()
On scx_link_sched() error paths (parent disabled, hash insert failure),
&sch->all is never added to scx_sched_all. The cleanup path runs
scx_unlink_sched() unconditionally, which calls list_del_rcu(&sch->all) on a
list_head that was never initialized triggering a corruption warning.
Initialize &sch->all.
Fixes: 54be8de4236a ("sched_ext: Factor out scx_link_sched() and scx_unlink_sched()") Signed-off-by: Tejun Heo <tj@kernel.org>
Tejun Heo [Tue, 12 May 2026 20:30:00 +0000 (10:30 -1000)]
sched_ext: Drop NONE early return in scx_disable_and_exit_task()
d3e73a0808dd ("sched_ext: Handle SCX_TASK_NONE in disable/switched_from
paths") skipped the trailing scx_set_task_sched(p, NULL) on NONE tasks.
After scx_fail_parent() parks a task at NONE/sched=parent and the parent
is later freed via queue_rcu_work() during root_disable, the preserved
p->scx.sched dangles - print_scx_info() from sched_show_task() reads
sch->ops.name from freed memory.
Drop the early return. __scx_disable_and_exit_task() already short-
circuits on NONE and the SUB_INIT block was cleared by
scx_fail_parent()'s earlier call, so clearing p->scx.sched is the only
work left - and the one thing the path actually needs.
v2: Extend the SUB_INIT block comment to note that the flag is only
set on the sub-enable path, so it's always clear on the NONE
re-entry (Andrea).
Fixes: d3e73a0808dd ("sched_ext: Handle SCX_TASK_NONE in disable/switched_from paths") Signed-off-by: Tejun Heo <tj@kernel.org> Reviewed-by: Andrea Righi <arighi@nvidia.com>
KVM: x86: Swap the dst and src operand for MOVNTDQA
Swap the MOVNTDQA operands, as MOVNTDQA does NOT in fact have "the same
characteristics as 0F E7 (MOVNTDQ)"; MOVNTDQA loads from memory and stores
to registers, while MOVNTDQ loads from registers and stores to memory.
Per the SDM:
MOVNTDQ - Move packed integer values in xmm1 to m128 using non-temporal
hint.
MOVNTDQA - Move double quadword from m128 to xmm1 using non-temporal hint
if WC memory type.
Reported-by: Josh Eads <josheads@google.com> Fixes: c57d9bafbd0b ("KVM: x86: Add support for emulating MOVNTDQA") Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260506213514.2781948-1-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Sun, 3 May 2026 21:09:17 +0000 (23:09 +0200)]
KVM: x86: use again the flush argument of __link_shadow_page()
Except in the case of parentless nested-TDP pages, mmu_page_zap_pte()
clears the SPTE but leaves the invalid_list empty. In this case, using
kvm_flush_remote_tlbs() as kvm_mmu_remote_flush_or_zap() does is overkill.
Avoid flushing the entirety of the remote TLBs unless the invalid_list
was populated: instead, use a more efficient gfn-targeting flush (if
available) and skip it altogether if the caller guarantees that a TLB
flush is not necessary.
Based-on: <20260503201029.106481-1-pbonzini@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <20260503210917.121840-1-pbonzini@redhat.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: selftests: Ensure gmem file sizes are multiple of host page size
When creating a guest_memfd file and associated memslot to validate shared
guest memory, size the file+memslot to the maximum of the host or guest
page size. Attempting to allocate a single guest page will fail if the
host page size is greater than the guest page size, as KVM requires that
the size of memslots and guest_memfd files are a multiple of the host page
size.
For simplicity, verify the entire file can be shared between guest and host,
e.g. instead of trying to validate "partial" mappings.
Fixes: 42188667be38 ("KVM: selftests: Add guest_memfd testcase to fault-in on !mmap()'d memory") Reported-by: Zenghui Yu <zenghui.yu@linux.dev> Closes: https://lore.kernel.org/all/0064952b-048c-455d-ad89-e27e5cb82591@linux.dev Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-ID: <20260512155634.772602-1-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Paolo Bonzini [Tue, 12 May 2026 20:19:20 +0000 (22:19 +0200)]
Merge tag 'kvmarm-fixes-7.1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm into HEAD
KVM/arm64 fixes for 7.1, take #2
- Add the pKVM side of the workaround for ARM's erratum 4193714, provided
that the EL3 firmware does its part of the job. KVM will refuse to
initialise otherwise.
- Correctly handle 52bit VAs for guest EL2 stage-1 translations when
running under NV with E2H==0.
- Correctly deal with permission faults in guest_memfd memslots.
- Fix the steal-time selftest after the infrastructure was reworked.
- Make sure the host cannot pass a non-sensical clock update to the
EL2 tracing infrastructure.
- Appoint Steffen Eiden as a reviewer in anticipation of the KVM/s390
ability to run arm64 guests, which will inevitably lead to arm64
code being directly used on s390.
- Make sure that EL2 is configured with both exception entry and exit
being Context Synchronization Events.
- Handle the current vcpu being NULL on EL2 panic.
- Fix the selftest_vcpu memcache being empty at the point of donation or
sharing.
- Check that the memcache has enough capacity before engaging on the
share/donate path.
- Fix __deactivate_fgt() to use its parameter rather than a variable
in the macro context.
KVM: nSVM: Never use L0's PAUSE loop exiting while L2 is running
Never use L0's (KVM's) PAUSE loop exiting controls while L2 is running,
and instead always configure vmcb02 according to L1's exact capabilities
and desires.
The purpose of intercepting PAUSE after N attempts is to detect when the
vCPU may be stuck waiting on a lock, so that KVM can schedule in a
different vCPU that may be holding said lock. Barring a very interesting
setup, L1 and L2 do not share locks, and it's extremely unlikely that an
L1 vCPU would hold a spinlock while running L2. I.e. having a vCPU
executing in L1 yield to a vCPU running in L2 will not allow the L1 vCPU
to make forward progress, and vice versa.
While teaching KVM's "on spin" logic to only yield to other vCPUs in L2 is
doable, in all likelihood it would do more harm than good for most setups.
KVM has limited visibility into which L2 "vCPUs" belong to the same VM,
and thus share a locking domain. And even if L2 vCPUs are in the same
VM, KVM has no visilibity into L2 vCPU's that are scheduled out by the
L1 hypervisor.
Furthermore, KVM doesn't actually steal PAUSE exits from L1. If L1 is
intercepting PAUSE, KVM will route PAUSE exits to L1, not L0, as
nested_svm_intercept() gives priority to the vmcb12 intercept. As such,
overriding the count/threshold fields in vmcb02 with vmcb01's values is
nonsensical, as doing so clobbers all the training/learning that has been
done in L1.
Even worse, if L1 is not intercepting PAUSE, i.e. KVM is handling PAUSE
exits, then KVM will adjust the PLE knobs based on L2 behavior, which could
very well be detrimental to L1, e.g. due to essentially poisoning L1 PLE
training with bad data.
And copying the count from vmcb02 to vmcb01 on a nested VM-Exit makes even
less sense, because again, the purpose of PLE is to detect spinning vCPUs.
Whether or not a vCPU is spinning in L2 at the time of a nested VM-Exit
has no relevance as to the behavior of the vCPU when it executes in L1.
The only scenarios where any of this actually works is if at least one
of KVM or L1 is NOT intercepting PAUSE for the guest. Per the original
changelog, those were the only scenarios considered to be supported.
Disabling KVM's use of PLE makes it so the VM is always in a "supported"
mode.
Last, but certainly not least, using KVM's count/threshold instead of the
values provided by L1 is a blatant violation of the SVM architecture.
Fixes: 74fd41ed16fd ("KVM: x86: nSVM: support PAUSE filtering when L0 doesn't intercept PAUSE") Cc: Maxim Levitsky <mlevitsk@redhat.com> Tested-by: David Kaplan <david.kaplan@amd.com> Signed-off-by: Sean Christopherson <seanjc@google.com> Link: https://patch.msgid.link/20260508213321.373309-1-seanjc@google.com/ Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Aaron Sacks [Tue, 12 May 2026 06:07:42 +0000 (02:07 -0400)]
KVM: Reject wrapped offset in kvm_reset_dirty_gfn()
kvm_reset_dirty_gfn() guards the gfn range with
if (!memslot || (offset + __fls(mask)) >= memslot->npages)
return;
but offset is u64 and the addition is unchecked. The check can be
silently bypassed by a u64 wrap.
The dirty ring backing those entries is MAP_SHARED at
KVM_DIRTY_LOG_PAGE_OFFSET of the vcpu fd, so the VMM can rewrite the
slot and offset fields of any entry between when the kernel pushes
them and when KVM_RESET_DIRTY_RINGS consumes them. On reset,
kvm_dirty_ring_reset() re-reads the values via READ_ONCE() and feeds
them straight back into this check; only the flags handshake is
treated as the handover, the slot/offset payload is taken on trust.
makes the coalescing loop in kvm_dirty_ring_reset() compute
delta = (s64)(0 - 0xffffffffffffffc1) = 63
which falls in [0, BITS_PER_LONG), so it folds entry[i+1] into the
existing mask by setting bit 63. The trailing kvm_reset_dirty_gfn()
call then sees offset = 0xffffffffffffffc1 and __fls(mask) = 63;
the sum is 0 in u64 and the bounds check passes.
That offset propagates into kvm_arch_mmu_enable_log_dirty_pt_masked()
unchanged. On the legacy MMU path -- kvm_memslots_have_rmaps() ==
true, i.e. shadow paging, any VM that has allocated shadow roots, or
a write-tracked slot -- it reaches gfn_to_rmap(), which indexes
slot->arch.rmap[0][] with a near-U64_MAX gfn. That is an
out-of-bounds load of a kvm_rmap_head, followed by a conditional
clear of PT_WRITABLE_MASK in whatever the loaded pointer points at.
The path is reachable from any process holding /dev/kvm.
Range-check offset on its own first, so the addition cannot wrap.
memslot->npages is bounded well below U64_MAX, so once offset <
npages holds, offset + __fls(mask) (with __fls(mask) < BITS_PER_LONG)
stays in range.
Sergio Correia [Tue, 12 May 2026 13:28:59 +0000 (14:28 +0100)]
audit: enforce AUDIT_LOCKED for AUDIT_TRIM and AUDIT_MAKE_EQUIV
AUDIT_ADD_RULE and AUDIT_DEL_RULE correctly check for AUDIT_LOCKED
and return -EPERM, but AUDIT_TRIM and AUDIT_MAKE_EQUIV do not. This
allows a process with CAP_AUDIT_CONTROL to modify directory tree
watches and equivalence mappings even when the audit configuration
has been locked, undermining the purpose of the lock.
Add AUDIT_LOCKED checks to both commands.
Cc: stable@vger.kernel.org Reviewed-by: Ricardo Robaina <rrobaina@redhat.com> Assisted-by: Claude:claude-opus-4-6 Signed-off-by: Sergio Correia <scorreia@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
Sergio Correia [Tue, 12 May 2026 13:28:33 +0000 (14:28 +0100)]
audit: fix incorrect inheritable capability in CAPSET records
__audit_log_capset() records the effective capability set into the
inheritable field due to a copy-paste error. Every CAPSET audit
record therefore reports cap_pi (process inheritable) with the value
of cap_effective instead of cap_inheritable.
This silently corrupts audit data used for compliance and forensic
analysis: an attacker who modifies inheritable capabilities to
prepare for a privilege-escalating exec would have the change masked
in the audit trail.
The bug has been present since the original introduction of CAPSET
audit records in 2008.
Cc: stable@vger.kernel.org Fixes: e68b75a027bb ("When the capset syscall is used it is not possible for audit to record the actual capbilities being added/removed. This patch adds a new record type which emits the target pid and the eff, inh, and perm cap sets.") Reviewed-by: Ricardo Robaina <rrobaina@redhat.com> Assisted-by: Claude:claude-opus-4-6 Signed-off-by: Sergio Correia <scorreia@redhat.com> Signed-off-by: Paul Moore <paul@paul-moore.com>
Linus Torvalds [Tue, 12 May 2026 17:18:02 +0000 (10:18 -0700)]
Merge tag 'probes-fixes-v7.1-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace
Pull probes fixes from Masami Hiramatsu:
- kprobes: skip non-symbol addresses in kprobe_add_ksym_blacklist()
Since the ftrace adds its NOPs at .kprobes.text section (which stores
an array), a wrong entry is added when loading a module which uses
"__kprobes" attribute.
To solve this, add "notrace" to __kprobes functions
- test_kprobes: clear kprobes between test runs
Clear all kprobes in the test program after running a test set,
because Kunit test can run several times
- fprobe: Fix unregister_fprobe() to wait for RCU grace period
Since the fprobe data structure is removed with hlist_del_rcu(), it
should wait for the RCU grace period. If the caller waits for RCU, we
can use the async variant (e.g. eBPF)
* tag 'probes-fixes-v7.1-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
fprobe: Fix unregister_fprobe() to wait for RCU grace period
test_kprobes: clear kprobes between test runs
kprobes: skip non-symbol addresses in kprobe_add_ksym_blacklist()
Paolo Abeni [Tue, 12 May 2026 14:15:03 +0000 (16:15 +0200)]
Merge branch 'net-shaper-fix-various-minor-bugs'
Jakub Kicinski says:
====================
net: shaper: fix various minor bugs
Fix various minor bugs in the net shaper API.
First 2 patches deal with ordering issues around inserting
and publishing new shapers. Shapers are inserted "tentatively"
and marked valid only after HW op succeeded, this used to
be slightly racy.
Only other patch of note is patch 8. We want to add a Netlink
policy check on the handle ID. This necessitates patch 7.
Jakub Kicinski [Sun, 10 May 2026 19:29:04 +0000 (12:29 -0700)]
net: shaper: reject QUEUE scope handle with missing id
net_shaper_parse_handle() does not enforce that the user provides
the handle ID. For NODE the ID defaults to UNSPEC for all other
cases it defaults to 0.
For NETDEV 0 is the only option. For QUEUE defaulting to 0 makes
less intuitive sense. Specifically because the behavior should
(IMHO) be the same for all cases where there may be more than
one ID (QUEUE and NODE).
We should either document this as intentional or reject.
I picked the latter with no strong conviction.
projects / thirdparty / kernel / linux.git / log
