Merge patch series "Fix for unintended FUSE ACL cache"
Amir Goldstein <amir73il@gmail.com> says:
Fix for unintended FUSE ACL cache.
Christian, I bet you did not miss fuse acls...
Ghosts from the past have come back to haunt me now.
* patches from https://patch.msgid.link/20260713220932.413004-1-amir73il@gmail.com:
selftests/fuse: add ACL_DONT_CACHE regression test
fs: preserve ACL_DONT_CACHE state in forget_cached_acl()
Amir Goldstein [Mon, 13 Jul 2026 22:09:32 +0000 (00:09 +0200)]
selftests/fuse: add ACL_DONT_CACHE regression test
Add a test that reproduces the stale ACL bug fixed by:
"fs: preserve ACL_DONT_CACHE state in forget_cached_acl()"
A FUSE mount that does not negotiate FUSE_POSIX_ACL initialises inodes
with i_acl = ACL_DONT_CACHE. Before the fix, calling
forget_all_cached_acls() (e.g. from fuse_update_get_attr() on a
statx(AT_STATX_FORCE_SYNC)) would silently replace ACL_DONT_CACHE with
ACL_NOT_CACHED, enabling the kernel ACL cache. A subsequent getxattr
would populate the cache, and because fuse_set_acl() skips
forget_all_cached_acls() for !fc->posix_acl, later ACL changes were
not visible to callers — getxattr returned stale data.
The test mounts a minimal libfuse3 lowlevel filesystem (no
FUSE_POSIX_ACL negotiated) and:
1. Issues two getxattrs — both must reach the daemon, proving
ACL_DONT_CACHE suppresses caching before any trigger.
2. Calls statx(AT_STATX_FORCE_SYNC) to trigger forget_all_cached_acls().
3. Issues another getxattr (populates the cache on a buggy kernel).
4. Switches the daemon to a different-sized ACL (ACL_B).
5. Issues a final getxattr — expects ACL_B (44 bytes) and daemon
call count 4; a buggy kernel returns stale ACL_A (28 bytes).
fuse_acl_cache_test is only built when libfuse3 is detected via
pkg-config.
Christian Brauner <brauner@kernel.org> says:
Changed do_force_statx() to call the statx() libc wrapper instead of
syscall(SYS_statx, ...) as requested by Amir after review feedback from
Luis Henriques, and dropped the now unused <sys/syscall.h> include.
Amir Goldstein [Mon, 13 Jul 2026 22:09:31 +0000 (00:09 +0200)]
fs: preserve ACL_DONT_CACHE state in forget_cached_acl()
The ACL_DONT_CACHE state is meant to be a constant state for the inode
for filesystems that want to opt out of posix acl caching.
Commit facd61053cff1 ("fuse: fixes after adapting to new posix acl api")
used this facility to opt out of posix acl caching for fuse inodes with
fuse server that does not negotiate FUSE_POSIX_ACL (fc->posix_acl).
The commit also takes care to gate the forget_all_cached_acls() call in
fuse_set_acl() on fc->posix_acl because there is no need for it, but
there are other placed in fuse code which call forget_all_cached_acls()
unconditional to fc->posix_acl and those cause the loss of the
ACL_DONT_CACHE state.
This is not only a functional bug. Properly timed, a get_acl() from this
fuse filesystem can return a stale cached value, as was observed in tests,
because set_acl() does not invalidate the unintentional acl cache.
We could fix this in fuse, but it actually makes no sense for the vfs
helper forget_cached_acl() to invalidate the ACL_DONT_CACHE state, so
let it not do that to fix fuse and future users of ACL_DONT_CACHE.
Fixes: facd61053cff1 ("fuse: fixes after adapting to new posix acl api") Cc: stable@vger.kernel.org Signed-off-by: Amir Goldstein <amir73il@gmail.com> Link: https://patch.msgid.link/20260713220932.413004-2-amir73il@gmail.com Reviewed-by: Luis Henriques <luis@igalia.com> Signed-off-by: Christian Brauner (Amutable) <brauner@kernel.org>
Two paths add or delete the same slot (pf->vfs[vf_id].padev): a VF's
pdsc_reset_done() and the PF's devlink enable_vnet/disable_vnet handler.
They serialize on config_lock, but neither guards the slot under it
correctly.
add() registers and stores a new auxiliary device without first checking
the slot, so a second add of an already-populated slot leaks the first
device. del() makes that check outside config_lock, so two concurrent
dels can both pass it; the first clears the slot, and the second
dereferences a NULL pointer.
Check and update the slot under config_lock in both paths.
Fixes: b699bdc720c0 ("pds_core: specify auxiliary_device to be created") Reported-by: sashiko-bot@kernel.org # Running on a local machine Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Reviewed-by: Brett Creeley <brett.creeley@amd.com> Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Link: https://patch.msgid.link/20260714210745.1785625-1-nikhil.rao@amd.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
pds_core: order completion reads after the ownership check
pdsc_process_adminq() and pdsc_process_notifyq() decide a completion is
valid from its ownership field - the color bit for the adminq, the event
id for the notifyq - then read the rest of the descriptor, with no
barrier in between.
On a weakly ordered architecture the CPU may read the payload first. Add
dma_rmb() between the ownership read and the payload reads.
Fixes: 7e82a8745b95 ("pds_core: Prevent race issues involving the adminq") Reported-by: sashiko-bot <sashiko-bot@kernel.org> Closes: https://sashiko.dev/#/patchset/20260629200358.2626129-1-nikhil.rao%40amd.com?part=2 Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Reviewed-by: Eric Joyner <eric.joyner@amd.com> Link: https://patch.msgid.link/20260714204145.1782390-1-nikhil.rao@amd.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
pds_core: yield the CPU while waiting for the adminq to drain
pdsc_adminq_wait_and_dec_once_unused() busy-waits for adminq_refcnt to
drop to one:
while (!refcount_dec_if_one(&pdsc->adminq_refcnt))
cpu_relax();
The refcount is held by pdsc_adminq_post() for the duration of an
in-flight command, which can wait up to devcmd_timeout seconds
(PDS_CORE_DEVCMD_TIMEOUT is 5) for the hardware to complete. cpu_relax()
is not a reschedule point, so on a non-preemptible kernel this loop can
spin on the CPU for several seconds, starving other tasks on that core.
Add cond_resched() to the loop so the waiter yields to other runnable
tasks while it polls, keeping cpu_relax() as the busy-wait hint between
checks.
Fixes: 7e82a8745b95 ("pds_core: Prevent race issues involving the adminq") Reported-by: sashiko-bot <sashiko-bot@kernel.org> Closes: https://sashiko.dev/#/patchset/20260629200358.2626129-1-nikhil.rao%40amd.com?part=2 Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Reviewed-by: Eric Joyner <eric.joyner@amd.com> Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com> Link: https://patch.msgid.link/20260714201456.1776153-1-nikhil.rao@amd.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
pds_core: fix use-after-free on workqueue during remove
In pdsc_remove(), the workqueue is destroyed before pdsc_teardown()
is called. This ordering allows two paths to queue work on the
destroyed workqueue:
1. If pdsc_teardown() -> pdsc_devcmd_reset() times out, the error
path in pdsc_devcmd_locked() queues health_work.
2. A NotifyQ event can trigger the ISR and queue work before free_irq()
is called in pdsc_teardown().
Fix by moving destroy_workqueue() after pdsc_teardown() so the
workqueue outlives every queuer; destroy_workqueue() then flushes any
work still pending.
Draining the queued work also requires ordering the teardown so the
resources that work touches are freed last:
- In pdsc_qcq_free(), after freeing the interrupt, cancel_work_sync()
the queue's work and only then clear qcq->intx, so
pdsc_process_adminq()'s read of qcq->intx for interrupt-credit
return cannot race with the clear.
- Free adminqcq before notifyqcq: the shared adminq ISR is released
when adminqcq is freed, and the adminq work accesses notifyqcq, so
both must be stopped before notifyqcq is freed.
Fixes: 01ba61b55b20 ("pds_core: Add adminq processing and commands") Reported-by: sashiko-bot <sashiko-bot@kernel.org> Closes: https://patchwork.kernel.org/comment/27002369/ Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Link: https://patch.msgid.link/20260714180223.1642792-3-nikhil.rao@amd.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
pds_core: fix deadlock between reset thread and remove
pci_reset_function() acquires device_lock before performing the reset.
pdsc_remove() is called by the PCI core with device_lock already held.
If pdsc_pci_reset_thread() is running when pdsc_remove() is called,
destroy_workqueue() will block waiting for the work to complete, while
the work is blocked waiting for device_lock - deadlock.
Use pci_try_reset_function() which uses pci_dev_trylock() internally.
This acquires both the device lock and the PCI config access lock
without blocking - if either lock is contended, it returns -EAGAIN
immediately. This avoids the deadlock while also ensuring proper
config space access serialization during the reset.
The pci_dev_get/put calls are also removed as they were unnecessary -
the driver-owned workqueue is destroyed in pdsc_remove(), guaranteeing
the work completes before remove returns. The PCI core holds its
reference to pci_dev throughout the entire unbind sequence.
Fixes: 81665adf25d2 ("pds_core: Fix pdsc_check_pci_health function to use work thread") Reported-by: sashiko-bot <sashiko-bot@kernel.org> Closes: https://patchwork.kernel.org/comment/27002369/ Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Reviewed-by: Harshitha Ramamurthy <hramamurthy@google.com> Link: https://patch.msgid.link/20260714180223.1642792-2-nikhil.rao@amd.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
mailmap: fix wrong canonical name for mgr@kernel.org
After picking up some pending patches for the kernel to work on, I
realized my name in the mailmap file somehow got mixed up. When
switching to my kernel.org Address some time ago, I had never the
intention to use a scrambled variant of Polish and German used for my
first name to be found in this file. However, so here we are. Lets fix
it for good.
sctp: fix auth_chunk_list capacity check in sctp_auth_ep_add_chunkid
sctp_auth_ep_add_chunkid() uses SCTP_NUM_CHUNK_TYPES (20) as the
capacity limit for ep->auth_chunk_list, allowing it to hold up to
20 chunk entries (param_hdr.length up to 24). However, the copy
destination asoc->c.auth_chunks in struct sctp_cookie is only
SCTP_AUTH_MAX_CHUNKS (16) entries (20 bytes). When more than 16
chunks are added, sctp_association_init() memcpy overflows the
destination by up to 4 bytes.
Fix by using SCTP_AUTH_MAX_CHUNKS as the capacity limit, matching
the destination capacity.
Perfect FDIR filters can be added while the interface is down and are
kept on the software list for later restore. unregister_netdev() only
calls ndo_stop when the device is up, so txgbe_fdir_filter_exit() in
txgbe_close() is skipped in that case and the filters are leaked on
driver remove. Free the filter list from txgbe_remove() as well.
Fixes: 4bdb441105dc ("net: txgbe: support Flow Director perfect filters") Signed-off-by: Chenguang Zhao <zhaochenguang@kylinos.cn> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Link: https://patch.msgid.link/20260713091911.1614795-1-chenguang.zhao@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ilia Gavrilov [Thu, 9 Jul 2026 16:27:54 +0000 (16:27 +0000)]
rds: Fix inet6_addr_lst NULL dereference when IPv6 is disabled
When booting with the 'ipv6.disable=1' parameter, inet6_addr_lst
is never initialized because inet6_init() exits before addrconf_init()
is called to initialize it. An attempt to bind an RDS socket to
an ipv6 address results in a crash in __ipv6_chk_addr_and_flags()
net: txgbe: fix heap overflow when reading module EEPROM
txgbe_read_eeprom_hostif() always copies round_up(length, 4) bytes
into the caller buffer, which ethtool allocates with exactly 'length'
bytes. A non-4-aligned length therefore causes an out-of-bounds write.
Copy only the remaining bytes on the final dword instead.
Signed-off-by: Chenguang Zhao <zhaochenguang@kylinos.cn> Reviewed-by: Jiawen Wu <jiawenwu@trustnetic.com> Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> Fixes: 9b97b6b5635b ("net: txgbe: support getting module EEPROM by page") Link: https://patch.msgid.link/20260713085111.1481884-1-chenguang.zhao@linux.dev Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Weiming Shi [Thu, 16 Jul 2026 02:52:04 +0000 (19:52 -0700)]
tipc: serialize udp bearer replicast list updates
tipc_udp_rcast_add() and cleanup_bearer() both update ub->rcast.list with
list_add_rcu() / list_del_rcu(), but nothing serializes them. The add runs
from the encap receive softirq (via tipc_udp_rcast_disc()) without
rtnl_lock(), so it can race the cleanup delete and corrupt the list:
list_del corruption. prev->next should be ffff8880298d7ab8,
but was ffff88802449ad38. (prev=ffff888027e3ec98)
kernel BUG at lib/list_debug.c:62!
RIP: __list_del_entry_valid_or_report+0x17a/0x200
Workqueue: events cleanup_bearer
Call Trace:
cleanup_bearer (net/tipc/udp_media.c:811)
process_one_work (kernel/workqueue.c:3302)
worker_thread (kernel/workqueue.c:3466)
The bearer can be enabled from an unprivileged user namespace, as the
TIPCv2 generic-netlink ops carry no GENL_ADMIN_PERM.
Add a spinlock to struct udp_bearer and take it around the list_add_rcu()
in tipc_udp_rcast_add() and the list_del_rcu() loop in cleanup_bearer() so
the two writers can no longer corrupt the list.
Reject a duplicate peer under the same lock before allocating, and remove
tipc_udp_is_known_peer(). The old lockless pre-check in
tipc_udp_rcast_disc() was racy: two softirqs discovering the same peer
could both find it absent and add it twice.
cleanup_bearer() runs from a workqueue after tipc_udp_disable() clears the
bearer's up bit, so an encap softirq can still reach tipc_udp_rcast_add()
and add a peer after cleanup_bearer() has already emptied the list, leaking
that entry when the bearer is freed. Mark the bearer disabled under
rcast_lock once the list is emptied and refuse further additions.
Fixes: ef20cd4dd163 ("tipc: introduce UDP replicast") Reported-by: Xiang Mei <xmei5@asu.edu> Suggested-by: Tung Nguyen <tung.quang.nguyen@est.tech> Signed-off-by: Weiming Shi <bestswngs@gmail.com> Reviewed-by: Tung Nguyen <tung.quang.nguyen@est.tech> Link: https://patch.msgid.link/20260716025203.9332-2-bestswngs@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Justin Lai [Thu, 9 Jul 2026 10:34:56 +0000 (18:34 +0800)]
rtase: Workaround for TX hang caused by hardware packet parsing
The hardware performs packet parsing before packet transmission.
Parsing incomplete IPv4, IPv6, TCP, or UDP headers may trigger a TX
hang because the hardware parser expects additional protocol header
data that is not present in the packet.
The hardware performs additional PTP parsing on UDP packets identified
by destination ports 319/320 at the expected UDP destination port
offset.
If such a packet has transport data smaller than RTASE_MIN_PAD_LEN,
the hardware parser expects additional packet data and may trigger a
TX hang.
To avoid these hardware issues, the driver applies the following
workarounds.
Drop malformed packets that may trigger this hardware issue before
transmission.
For IPv4 non-initial fragments, the hardware does not check the
fragment offset before parsing the expected transport header location.
As a result, these packets are still subject to transport header
parsing even though they do not contain a transport header. If the
transport data is shorter than the minimum transport header required
by the hardware parser, pad the transport data to the minimum
transport header length required by the hardware parser. Packets that
also match the hardware PTP parsing conditions continue to follow the
corresponding workaround.
For IPv6 fragmented packets, neither of the above hardware issues
occurs because the hardware only continues packet parsing when the
IPv6 Base Header Next Header field directly indicates UDP. Packets
carrying a Fragment Header do not continue through the subsequent
packet parsing stages.
For packets identified for hardware PTP parsing, pad the transport
data so it reaches RTASE_MIN_PAD_LEN before transmission.
Fixes: d6e882b89fdf ("rtase: Implement .ndo_start_xmit function") Cc: stable@vger.kernel.org Signed-off-by: Justin Lai <justinlai0215@realtek.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260709103456.83789-1-justinlai0215@realtek.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
mptcp: only set DATA_FIN when a mapping is present
mptcp_get_options() clears only the status group of struct
mptcp_options_received; data_seq, subflow_seq and data_len are filled in
by mptcp_parse_option() exclusively inside the DSS mapping block, which
runs only when the DSS M (mapping present) bit is set.
A peer can send a DSS option with the DATA_FIN flag set but the mapping
bit clear. The parser then records mp_opt->data_fin while leaving
data_len and data_seq uninitialized. For a zero-length segment
mptcp_incoming_options() evaluates
if (mp_opt.data_fin && mp_opt.data_len == 1 &&
mptcp_update_rcv_data_fin(msk, mp_opt.data_seq, mp_opt.dsn64))
which reads the uninitialized data_len and data_seq; KMSAN reports an
uninit-value in mptcp_incoming_options(). The stale data_seq can also be
fed into the receive-side DATA_FIN sequence tracking.
Record the DATA_FIN flag only when the DSS option carries a mapping, so
data_fin is never set without data_seq and data_len also being present.
data_fin is part of the status group that mptcp_get_options() clears up
front, so on the no-map path it stays zero and the zero-length DATA_FIN
branch is simply skipped. A DATA_FIN is always transmitted together with
a mapping (mptcp_write_data_fin() sets use_map along with data_seq and
data_len), so legitimate DATA_FIN handling is unaffected.
Move the pr_debug() that logs the parsed DSS flags below the mapping
block, so it reports the final data_fin value instead of the stale one
it would otherwise print before the assignment.
Fixes: 43b54c6ee382 ("mptcp: Use full MPTCP-level disconnect state machine") Suggested-by: Paolo Abeni <pabeni@redhat.com> Cc: stable@vger.kernel.org Signed-off-by: Michael Bommarito <michael.bommarito@gmail.com> Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org> Link: https://patch.msgid.link/20260709191925.2811195-1-michael.bommarito@gmail.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Antoine Tenart [Thu, 9 Jul 2026 12:50:00 +0000 (14:50 +0200)]
geneve: ensure the skb is writable before fixing its headers
Make sure the IPv4/6 and UDP headers are writable before fixing them up in
geneve_post_decap_hint. As skb_ensure_writable can reallocate the skb linear
area, reload the GRO hint header pointer and only set the IPv4/6 header ones
after the call.
Fixes: fd0dd796576e ("geneve: use GRO hint option in the RX path") Reported-by: Sashiko <sashiko-bot@kernel.org> Closes: https://sashiko.dev/#/patchset/20260529144713.780938-1-atenart%40kernel.org Signed-off-by: Antoine Tenart <atenart@kernel.org> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260709125000.141092-1-atenart@kernel.org Signed-off-by: Jakub Kicinski <kuba@kernel.org>
LiangCheng Wang [Wed, 15 Jul 2026 06:49:38 +0000 (14:49 +0800)]
wifi: brcmfmac: set F2 blocksize to 256 for BCM43752
The BCM43752 is not reliable with the default 512-byte SDIO function 2
block size: on an i.MX8MP board with an AMPAK AP6275S module at
SDR104 / 200 MHz, an iperf TX stress test kills WLAN within seconds:
Commit d2587c57ffd8 ("brcmfmac: add 43752 SDIO ids and initialization")
set up the 43752 like the 4373 for the F2 watermark but missed the F2
block size, which the 4373 limits to 256 bytes. The vendor driver
(bcmdhd) also programs a 256-byte F2 block size for this chip and runs
the same hardware without errors.
Group the 43752 with the 4373, matching the F2 watermark handling.
With this change a 10-minute bidirectional iperf3 soak completes with
zero SDIO errors at ~270 Mbit/s in each direction.
Backporting note: kernels before v6.18 name this id
SDIO_DEVICE_ID_BROADCOM_CYPRESS_43752, so on those trees the case
label added by this patch must be adjusted to that name. Cherry-picking
the rename commit 74e2ef72bd4b ("wifi: brcmfmac: fix 43752 SDIO FWVID
incorrectly labelled as Cypress (CYW)") first is not a clean
alternative: on trees before v6.17 its context collides with the 43751
additions, and trees before v6.2 lack the FWVID framework it touches.
Fixes: d2587c57ffd8 ("brcmfmac: add 43752 SDIO ids and initialization") Cc: stable@vger.kernel.org # see patch description, needs adjustments for <= 6.17 Signed-off-by: LiangCheng Wang <zaq14760@gmail.com> Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com> Link: https://patch.msgid.link/20260715-b43752-f2-blksz-v2-1-f9be49856050@gmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
net: Call net_enable_timestamp() before failure in sk_clone().
When sk_clone() fails, sk_destruct() is called for the new socket.
If the parent socket has SK_FLAGS_TIMESTAMP in sk->sk_flags,
net_disable_timestamp() is called for the child socket even though
net_enable_timestamp() is not called for it.
Let's call net_enable_timestamp() before any failure path in
sk_clone().
Fixes: 704da560c0a0 ("tcp: update the netstamp_needed counter when cloning sockets") Reported-by: Sashiko <sashiko-bot@kernel.org> Closes: https://lore.kernel.org/all/20260709032007.9E4D61F000E9@smtp.kernel.org/ Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Reviewed-by: Jason Xing <kerneljasonxing@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260709183315.965751-3-kuniyu@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
soreuseport: Clear sk_reuseport_cb before failure in sk_clone().
When sk_clone() fails, sk_destruct() is called for the new socket.
If the parent socket has sk->sk_reuseport_cb, the child will call
reuseport_detach_sock() for the reuseport group.
Let's clear sk->sk_reuseport_cb before any failure path in sk_clone().
Note that this was not a problem before the cited commit because
reuseport_detach_sock() did nothing if the socket was not found in
the reuseport array.
Fixes: 5dc4c4b7d4e8 ("bpf: Introduce BPF_MAP_TYPE_REUSEPORT_SOCKARRAY") Reported-by: Sashiko <sashiko-bot@kernel.org> Closes: https://lore.kernel.org/all/20260709032007.9E4D61F000E9@smtp.kernel.org/ Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Reviewed-by: Jason Xing <kerneljasonxing@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260709183315.965751-2-kuniyu@google.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
MAC_AUTO_SW (VR_MII_DIG_CTRL1 bit 9) enables automatic XPCS speed
mode switching after CL37 auto-negotiation and is only meaningful in
SGMII MAC mode. The original code unconditionally set this bit on
every call to xgbe_an37_set(), including when called from
xgbe_an37_disable() with enable=false. This left MAC_AUTO_SW=1 after
AN was disabled, causing the XPCS to autonomously switch speed from
stale AN state during subsequent mode changes, breaking SGMII speed
negotiation on 1G copper SFP modules.
Patrick: This was breaking negotiation for all 1G SFP modules,
not just copper modules.
net: gro: fix double aggregation of flush-marked skbs
Commit 0ab03f353d36 ("net-gro: Fix GRO flush when receiving a GSO
packet.") added a flush check to skb_gro_receive(), but
skb_gro_receive_list() lacks the same validation.
As a result, packets marked with NAPI_GRO_CB(skb)->flush may still be
re-aggregated.
This allows already-GRO'd packets with existing frag_list to be
re-aggregated into a new GRO session, corrupting the frag_list chain
structure. When skb_segment() attempts to unpack these malformed packets,
it encounters invalid state and triggers a kernel panic.
Scenario (Tethering/Device forwarding):
1. Driver: Generated aggregated packet P1 via LRO with frag_list
2. Dev A: Receives aggregated fraglist packet and flush flag set
3. Dev A: Re-enters GRO, skb_gro_receive_list() is called
4. Missing flush check allows re-aggregation despite flush flag
5. Frag_list chain becomes corrupted (loops or dangling refs)
6. Dev B: TX path calls skb_segment(), crashes on corrupted frag_list
Root cause in skb_segment():
The check at line ~4891:
if (hsize <= 0 && i >= nfrags && skb_headlen(list_skb) &&
(skb_headlen(list_skb) == len || sg)) {
When frag_list is corrupted by double aggregation, when list_skb is
a NULL pointer from skb->next, skb_headlen(list_skb) dereference
NULL/corrupted pointers occurs.
Fix: Add NAPI_GRO_CB(skb)->flush validation to the early-return check in
skb_gro_receive_list(), matching the defensive programming pattern of
skb_gro_receive().
Fixes: 3a1296a38d0c ("net: Support GRO/GSO fraglist chaining.") Cc: stable@vger.kernel.org Signed-off-by: Shiming Cheng <shiming.cheng@mediatek.com> Reviewed-by: Willem de Bruijn <willemb@google.com> Link: https://patch.msgid.link/20260709014704.3625-1-shiming.cheng@mediatek.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"RISC-V:
- Avoid redundant allocations when allocating IMSIC page tables
- Apply SBI FWFT LOCK flag only on successful set
- Bound SBI PMU counter mask scan to BITS_PER_LONG, since on RV32 the
PMU SBI start/stop helper can only access 32 PMU counters.
- Skip TLB flush when G-stage PTE becomes valid if the Svvptc
extension is available.
- Always show Zicbo[m|z|p] block sizes in ONE_REG
- Inject instruction access fault on unmapped guest fetch
- Use raw spinlock for irqs_pending and irqs_pending_mask
- Fix Spectre-v1 in vector register access via ONE_REG
x86:
- Fixes to SEV selftests
- Once free_nested() did a VMCLEAR of shadow VMCS, there's no need to
VMCLEAR it again if the kernel is preempted and thread migration
happens
- Preserve nested TDP shadow page tables if they are used as roots,
instead of clearing them unnecessarily
- Fix use of stale data if out-of-memory happens after vendor module
reload
- Check for invalid/obsolete root *after* making MMU pages available,
because the latter can make a page invalid
- Only reset TSC Deadline Timer in apic_timer_expired on KVM_RUN"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: x86: Only reset TSC Deadline Timer in apic_timer_expired on KVM_RUN
KVM: selftests: sev_init2_tests: Derive SEV availability from KVM
KVM: selftests: sev_smoke_test: Only run VM types the host offers
KVM: x86/mmu: Fix use-after-free on vendor module reload
KVM: x86/mmu: Preserve nested TDP shadow page tables if they are used as roots
KVM: x86: Check for invalid/obsolete root *after* making MMU pages available
KVM: nVMX: Hide shadow VMCS right after VMCLEAR
KVM: riscv: Fix Spectre-v1 in vector register access
RISC-V: KVM: Serialize virtual interrupt pending state updates
RISC-V: KVM: Inject instruction access fault on unmapped guest fetch
RISC-V: KVM: Zicbo[m|z|p] block sizes should be always present in ONE_REG
riscv: kvm: Skip TLB flush when G-stage PTE becomes valid with Svvptc
KVM: riscv: PMU: Bound counter mask scan to BITS_PER_LONG
KVM: riscv: SBI FWFT: Apply LOCK flag only on successful set
RISC-V: KVM: Avoid redundant page-table allocations in ioremap topup
Richard Cheng [Tue, 21 Jul 2026 10:00:26 +0000 (18:00 +0800)]
arm64/mm: Check the requested PFN range during memory removal
prevent_memory_remove_notifier() advances pfn while scanning the requested
range for early memory. When the loop completes, pfn is at or beyond
end_pfn. Passing it to can_unmap_without_split() therefore checks a range
after the one being offlined.
Consequently, a valid request can be rejected based on the following
range, while a request that would split a leaf mapping can be accepted if
the shifted range can be unmapped without a split. This was observed with
CXL DAX memory, where the final memory block was incorrectly allowed to
be offlined.
Pass arg->start_pfn into can_unmap_without_split() so it checks the
requested range.
Fixes: 95a58852b0e5 ("arm64/mm: Reject memory removal that splits a kernel leaf mapping") Signed-off-by: Richard Cheng <icheng@nvidia.com> Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com> Signed-off-by: Will Deacon <will@kernel.org>
Merge tag 'probes-fixes-v7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace
Pull probes fixes from Masami Hiramatsu:
- Avoid temporary buffer truncation in match_command_args()
Compare argument name, delimiter, and comm expression directly
instead of formatting into a stack buffer to prevent false
matching failures
- Prevent out-of-bounds write in __trace_probe_log_err()
Return early when trace_probe_log.argc is zero to prevent
out-of-bounds access when constructing the formatted error
command string
- Fix potential underflow in LEN_OR_ZERO macro
Ensure buffer length is greater than current position before
subtraction to prevent unsigned size underflow when formatting
print strings
- Fix exact system name matching in eprobe_dyn_event_match()
Check system name null-termination to avoid partial prefix
matching when comparing event probe target system names
* tag 'probes-fixes-v7.2-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
tracing/eprobe: Fix exact system name matching in eprobe_dyn_event_match()
tracing/probes: Fix potential underflow in LEN_OR_ZERO macro
tracing/probes: Prevent out-of-bounds write in __trace_probe_log_err()
tracing/probes: Avoid temporary buffer truncation in trace_probe_match_command_args()
Steven Price [Fri, 3 Jul 2026 13:48:35 +0000 (14:48 +0100)]
arm64: Correct value returned by ESR_ELx_FSC_ADDRSZ_nL()
Address size fault, level -1 is encoded as 0b101001 or 0x29 according to
the Arm ARM. Correct the value to match the spec. This also matches the
offset of "level -1 address size fault" in the fault_info array in
fault.c.
Fixes: fb8a3eba9c81 ("KVM: arm64: Only read HPFAR_EL2 when value is architecturally valid") Signed-off-by: Steven Price <steven.price@arm.com> Reviewed-by: Marc Zyngier <maz@kernel.org> Signed-off-by: Will Deacon <will@kernel.org>
Merge tag 'for-7.2-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"I'm catching up with the fix backlog in the development branch, so
here's a number of them and will probably send one more for this or
the next rc:
- relocation fixes:
- skip attempting compression on reloc inodes
- exclude inline extents from file extent offset checks
- fix minor memory leak after error when adding reloc root
- fix root cleanup after inserting and merging
- fix clearing folio tags after writeback
- clear logging flag of extent map before splitting
- fix unsigned 32/64 type conversions when accounting dirty metadata,
leading to continually exceeding threshold
- fix regression in 32bit compat ioctl for subvolume info
- fix type of SEARCH_TREE ioctl buffer in UAPI header
- fix expression in ASSERT expression which can be unconditionally
evaluated on some compilers
- only account delalloc bytes for regular inodes"
* tag 'for-7.2-rc4-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: fix GET_SUBVOL_INFO after compat refactor
btrfs: free mapping node on duplicate reloc root insert
btrfs: fix a regression where PAGECACHE_TAG_DIRTY is never cleared
btrfs: don't propagate EXTENT_FLAG_LOGGING to split extent maps
btrfs: fix u32 to s64 type conversion in dirty_metadata_bytes accounting
btrfs: fix NULL pointer deref during assertion in btrfs_backref_free_node()
btrfs: only account delalloc bytes for regular file inodes in btrfs_getattr()
btrfs: reject inline file extents item in get_new_location()
btrfs: do not try compression for data reloc inodes
btrfs: declare btrfs_ioctl_search_args_v2::buf as __u8
btrfs: fix reloc root cleanup in merge_reloc_roots()
btrfs: fix use-after-free on reloc root after error in insert_dirty_subvol()
pds_core: reject component parameter in legacy firmware update
The legacy firmware update path does not support per-component updates.
If a user specifies a component parameter with devlink flash, reject
the request with -EOPNOTSUPP rather than silently ignoring the component
parameter and flashing the entire firmware image.
Fixes: 49ce92fbee0b ("pds_core: add FW update feature to devlink") Signed-off-by: Nikhil P. Rao <nikhil.rao@amd.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260708163649.128620-1-nikhil.rao@amd.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Linmao Li [Tue, 21 Jul 2026 01:15:58 +0000 (09:15 +0800)]
drm/vc4: Prevent shader BO mappings from becoming writable
vc4_gem_object_mmap() rejects a writable mapping of a validated shader
BO, but leaves VM_MAYWRITE set. Userspace can map the BO read-only and
then turn it writable with mprotect().
Validated shader BOs must stay read-only: the validator checks the
instructions once and the GPU trusts them afterwards. A writable
mapping lets userspace rewrite the code after validation, bypassing the
validator.
Clear VM_MAYWRITE on the read-only path so the mapping cannot be
upgraded, as i915 already does for its read-only objects.
Fixes: 463873d57014 ("drm/vc4: Add an API for creating GPU shaders in GEM BOs.") Cc: stable@vger.kernel.org Reported-by: Sashiko <sashiko-bot@kernel.org> Closes: https://lore.kernel.org/dri-devel/20260720085554.B0AF01F000E9@smtp.kernel.org/ Signed-off-by: Linmao Li <lilinmao@kylinos.cn> Link: https://patch.msgid.link/20260721011558.1672477-1-lilinmao@kylinos.cn Reviewed-by: Maíra Canal <mcanal@igalia.com> Signed-off-by: Maíra Canal <mcanal@igalia.com>
Zhao Li [Wed, 8 Jul 2026 19:59:07 +0000 (03:59 +0800)]
wifi: cfg80211: guard optional PMSR nominal time
pmsr_parse_ftm() rejects a request that omits NOMINAL_TIME only for
non-trigger-based PD ranging. It then reads the attribute
unconditionally for every non-trigger-based request:
This PR contains quite a few fixes, mostly found by LLMs. Notably:
- missing validation of TLVs length
- missing validation on the revision of BIOS tables
- missing validation of fw responses length
- protect against double free of RX pointers
- avoid double deregistration of the tzone core
- validate fields in FW responses before using them
- fix off-by-one in WLAN_EID_EXT_CAPABILITY parsing
- Add missing support for UNII-9
====================
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
hwsim_tx_info_frame_received_nl() casts the HWSIM_ATTR_TX_INFO payload
to a struct hwsim_tx_rate * and unconditionally reads
IEEE80211_TX_MAX_RATES entries (8 bytes) from it. The policy only bounds
the attribute from above (NLA_BINARY .len is a maximum) and the op sets
GENL_DONT_VALIDATE_STRICT, so a short or zero-length attribute is
accepted and the loop reads past the payload.
Require the exact length in the policy, so a malformed attribute is
rejected before the handler runs.
Fan Wu [Sat, 18 Jul 2026 02:43:53 +0000 (02:43 +0000)]
wifi: brcmfmac: drain bus_reset work on device removal
brcmf_fw_crashed() and the debugfs "reset" entry both schedule
drvr->bus_reset, whose callback recovers drvr through container_of()
and dereferences it. The removal path frees drvr (brcmf_free ->
wiphy_free) without draining the work, so a bus_reset callback pending
or running during removal can outlive drvr.
Cancellation cannot live in brcmf_detach() or brcmf_free(): the work
callback reaches teardown through the bus .reset op (PCIe
brcmf_pcie_reset -> brcmf_detach; SDIO brcmf_sdio_bus_reset ->
brcmf_sdiod_remove -> brcmf_free), so cancelling there would wait for
the running work and deadlock.
Add a per-bus mutex (bus_reset_lock) and route all arming through
brcmf_bus_schedule_reset(), which under the lock skips when the bus is
marked removing. Each bus remove entry calls
brcmf_bus_cancel_reset_work(), which under the same lock sets removing
and cancels the work. Holding the mutex across cancel_work_sync() makes
the set-removing + drain step atomic. Every producer reaches the arming
path from process context -- the PCIe firmware-halt notification runs in
the threaded IRQ handler (brcmf_pcie_isr_thread) and the SDIO hostmail
path runs from the data workqueue -- so the mutex is taken only in
sleepable contexts. Where applicable the remove entry first stops the
firmware-crash producer: on PCIe mask the mailbox and synchronize_irq;
on SDIO unregister the bus interrupt and cancel the data worker, which
also reports firmware halts through brcmf_fw_crashed(). The mutex is
initialized at bus allocation. The SDIO suspend power-off path frees
drvr through the same brcmf_sdiod_remove() and takes the same lock;
resume re-allows the work only on a successful re-probe.
Also guard brcmf_fw_crashed() against a NULL bus_if/drvr: it can fire
before brcmf_attach() wires up drvr, and it dereferences drvr
(bphy_err/brcmf_dev_coredump) before reaching the arming gate.
The bus_reset work is shared across buses, so the drain is applied to
every remove path: PCIe (the .reset op introduced by the Fixes commit),
SDIO (arms the same work through brcmf_fw_crashed()), and USB (via the
debugfs "reset" entry). cancel_work_sync() drains a running or pending
bus_reset work item before removal frees drvr, and patch 1/2 makes the
scratch-buffer release safe when reset teardown has already released
those DMA buffers.
This patch fixes the lifetime of the bus_reset work item itself. It does
not attempt to address the separate, pre-existing lifetime of the
asynchronous firmware completion started by the PCIe reset path. That
callback needs its own lifetime/ownership protocol and is being tracked
separately.
This issue was found by an in-house static analysis tool.
Fixes: 4684997d9eea ("brcmfmac: reset PCIe bus on a firmware crash") Cc: stable@vger.kernel.org Signed-off-by: Fan Wu <fanwu01@zju.edu.cn> Assisted-by: Codex:gpt-5.6 Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com> Link: https://patch.msgid.link/20260718024353.3147201-3-fanwu01@zju.edu.cn Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Fan Wu [Sat, 18 Jul 2026 02:43:52 +0000 (02:43 +0000)]
wifi: brcmfmac: make release_scratchbuffers idempotent
brcmf_pcie_release_scratchbuffers() frees the shared.scratch and
shared.ringupd DMA buffers with dma_free_coherent() but does not clear
the pointers afterwards, unlike the sibling release_ringbuffers() which
NULLs commonrings/flowrings/idxbuf on release.
Both the bus_reset .reset callback (brcmf_pcie_reset) and
brcmf_pcie_remove() call release_scratchbuffers. When reset teardown
has run before removal, remove's own teardown would call
dma_free_coherent() a second time on the already-freed DMA allocation.
NULL the pointers after free, matching release_ringbuffers(), so a later
release observes that the allocation has already been released. This
patch makes repeated sequential release safe; the reset-work lifetime is
handled separately by the following patch.
This issue was found by an in-house static analysis tool.
Fixes: 4684997d9eea ("brcmfmac: reset PCIe bus on a firmware crash") Cc: stable@vger.kernel.org Signed-off-by: Fan Wu <fanwu01@zju.edu.cn> Assisted-by: Codex:gpt-5.6 Acked-by: Arend van Spriel <arend.vanspriel@broadcom.com> Link: https://patch.msgid.link/20260718024353.3147201-2-fanwu01@zju.edu.cn Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Andrew Pope [Fri, 17 Jul 2026 01:17:51 +0000 (11:17 +1000)]
wifi: mac80211: recalculate TIM when a station enters power save
When an AP buffers frames for a station on its per-station TXQs and the
station subsequently enters power save, sta_ps_start() records the
buffered TIDs in txq_buffered_tids but does not update the TIM. The
station's TIM bit is only ever set when a further frame is buffered
while the station is already asleep
(ieee80211_tx_h_unicast_ps_buf() -> sta_info_recalc_tim()).
If no further downlink frame arrives for that station the beacon
TIM never advertises the buffered traffic. A station relying on the
TIM then remains in doze indefinitely on top of a non-empty queue. Its
TXQs were removed from the scheduler's active list at PS entry, nothing
pages it, and the flow deadlocks until an unrelated event wakes the
station.
Recalculate the TIM at the end of sta_ps_start(), so traffic
already buffered at PS entry is advertised immediately.
sta_info_recalc_tim() already consults txq_buffered_tids, which is
updated above, and is safe in this context (it is already called
from equivalent paths such as the tx handlers and
ieee80211_handle_filtered_frame()).
Fixes: ba8c3d6f16a1 ("mac80211: add an intermediate software queue implementation") Signed-off-by: Andrew Pope <andrew.pope@morsemicro.com> Link: https://patch.msgid.link/20260717011751.79524-1-andrew.pope@morsemicro.com
[add wifi: subject prefix] Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Li RongQing [Tue, 21 Jul 2026 09:34:10 +0000 (17:34 +0800)]
iommu/intel: Fix out-of-bounds memset in dmar_latency_disable()
dmar_latency_disable() intends to zero out only the single
latency_statistic entry for the given type, but the memset size was
computed as sizeof(*lstat) * DMAR_LATENCY_NUM, which clears the entire
array starting from &lstat[type].
When type > 0, this writes beyond the end of the allocated array,
corrupting adjacent memory.
Fix by using sizeof(*lstat) to clear only the target entry.
Fixes: 55ee5e67a59a ("iommu/vt-d: Add common code for dmar latency performance monitors") Signed-off-by: Li RongQing <lirongqing@baidu.com> Signed-off-by: Will Deacon <will@kernel.org>
The ivrs_acpihid command-line parser appends entries to a fixed
four-element early_acpihid_map array. Unlike the sibling IOAPIC and HPET
parsers, it does not reject a fifth entry before incrementing the map size.
Check the capacity at the common found label before parsing the HID and
UID or writing the entry.
Fixes: ca3bf5d47cec ("iommu/amd: Introduces ivrs_acpihid kernel parameter") Signed-off-by: Pengpeng Hou <pengpeng@iscas.ac.cn> Reviewed-by: Ankit Soni <Ankit.Soni@amd.com> Signed-off-by: Will Deacon <will@kernel.org>
wifi: mwifiex: fix NULL dereference when the AP has HT-cap but no HT-oper
mwifiex_tdls_add_ht_oper() gates its follow-the-AP-bandwidth path on
bss_desc->bcn_ht_cap being present, but then dereferences a different
pointer, bss_desc->bcn_ht_oper:
if (ISSUPP_CHANWIDTH40(priv->adapter->hw_dot_11n_dev_cap) &&
bss_desc->bcn_ht_cap &&
ISALLOWED_CHANWIDTH40(bss_desc->bcn_ht_oper->ht_param))
bcn_ht_cap and bcn_ht_oper are populated independently while parsing the
associated AP's beacon in mwifiex_update_bss_desc_with_ie(): an AP that
advertises an HT Capabilities element but no HT Operation element leaves
bcn_ht_cap non-NULL and bcn_ht_oper NULL. Setting up a TDLS link to a
peer while associated to such an AP then dereferences the NULL
bcn_ht_oper and crashes the kernel. Every other bcn_ht_oper user in the
driver NULL-checks it first.
Guard on the pointer that is actually dereferenced.
Found by 0sec automated security-research tooling (https://0sec.ai).
Fixes: 396939f94084 ("mwifiex: add HT operation IE in TDLS setup confirm") Cc: stable@vger.kernel.org Assisted-by: 0sec:multi-model Signed-off-by: Doruk Tan Ozturk <doruk@0sec.ai> Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com> Link: https://patch.msgid.link/20260716103042.88469-1-doruk@0sec.ai Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Georgi Valkov [Thu, 16 Jul 2026 00:17:28 +0000 (03:17 +0300)]
wifi: mwifiex: replace one-element arrays with flexible array members
Replace deprecated one-element arrays with flexible array members.
CONFIG_FORTIFY_SOURCE reports the following warning when
one-element arrays are used as variable-length buffers:
sta_cmd.c:1033 mwifiex_sta_prepare_cmd
memcpy: detected field-spanning write (size 84) of single field
"domain->triplet" at .../marvell/mwifiex/sta_cmd.c:1033 (size 3)
Convert affected structs to use flexible array members.
- Preserve existing wire layouts.
- Use DECLARE_FLEX_ARRAY() for structs inside affected unions.
Tested-on: WRT3200ACM, OpenWrt Signed-off-by: Georgi Valkov <gvalkov@gmail.com> Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com> Link: https://patch.msgid.link/20260716001728.57799-1-gvalkov@gmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
wifi: at76c50x-usb: avoid length underflow in at76_guess_freq()
at76_guess_freq() checks only that the received frame is at least a bare
802.11 header (24 bytes) before subtracting the fixed management-body
offset:
len -= el_off;
For both beacon and probe response frames, el_off is 36. If the frame is
shorter than el_off, subtracting it causes the calculated IE length to
wrap. The length is eventually passed to cfg80211_find_elem_match() as a
very large unsigned value, so the element walk runs beyond the RX skb.
This path is reached from at76_rx_tasklet() while scanning. If the device
delivers a truncated beacon or probe response, the oversized IE length
causes an out-of-bounds read during scanning.
Skip the IE lookup if the frame does not reach the variable elements,
before subtracting el_off.
Lu Baolu [Thu, 16 Jul 2026 05:35:53 +0000 (13:35 +0800)]
iommu/vt-d: Disallow SVA if page walk is not coherent
Hardware implementations report Scalable-Mode Page-walk Coherency Support
via the SMPWCS field in the extended capability register. If the hardware
does not support page-walk coherency, a clflush is required every time
the page table entries (which are walked by the IOMMU hardware) are
updated.
In the SVA case, page tables are managed by the CPU mm core, not by the
IOMMU driver. Because the IOMMU driver has no way of knowing whether the
CPU page table management code has ensured coherency via clflush, the
driver must deny SVA if the hardware does not support coherent paging.
Fixes: ff3dc6521f78 ("iommu/vt-d: Fix CPU and IOMMU SVM feature matching checks") Cc: stable@vger.kernel.org Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com> Reviewed-by: Kevin Tian <kevin.tian@intel.com> Reviewed-by: Samiullah Khawaja <skhawaja@google.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Will Deacon <will@kernel.org>
event->len is supplied by the device firmware and is never validated,
and the subtraction is unchecked. assoc_req_ies points into
adapter->event_body[MAX_EVENT_SIZE], a fixed-size array embedded in the
kmalloc()'d struct mwifiex_adapter.
On the ap_11n_enabled path mwifiex_set_sta_ht_cap() walks these IEs with
cfg80211_find_ie(), whose for_each_element() loop dereferences each
element header. A firmware-reported event->len larger than the bytes
actually received makes assoc_req_ies_len describe IEs that extend past
event_body, so the walk reads out of the adapter slab object, a
slab-out-of-bounds read (KASAN: slab-out-of-bounds in cfg80211_find_ie).
An event->len smaller than the header instead makes the int subtraction
negative, which wraps to a huge size_t when stored in assoc_req_ies_len.
The same length is handed to cfg80211_new_sta(), so a more modest
over-claim can also copy stale event_body bytes into the
NL80211_CMD_NEW_STATION notification.
A malicious or malfunctioning mwifiex device (USB/SDIO/PCIe) can deliver
such an event while the interface is in AP/uAP mode.
Validate event->len before use: reject a length that underflows the
header or that would place the IEs outside the event_body[] buffer the
event was copied into. event->len here is struct mwifiex_assoc_event.len,
a payload field internal to this event, not the transport frame length,
so it is validated in this handler rather than at the generic
MWIFIEX_TYPE_EVENT receive path, which only sees the event cause and the
transport frame length. The bound is against event_body[MAX_EVENT_SIZE]
rather than the actually-received length because the transports store the
event differently (USB and SDIO leave the 4-byte event header in
event_skb, PCIe strips it via skb_pull), whereas event_body is the single
fixed buffer all of them copy the event into. This is the event-path
analogue of the receive-path bounds checks added in commit 119585281617
("wifi: mwifiex: Fix OOB and integer underflow when rx packets").
Fixes: e568634ae7ac ("mwifiex: add AP event handling framework") Signed-off-by: HE WEI (ギカク) <skyexpoc@gmail.com> Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com> Link: https://patch.msgid.link/20260715135711.34688-1-skyexpoc@gmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
vhost-net: fix TX stall when vhost owns virtio-net header
When vhost owns the virtio-net header, i.e. when
VHOST_NET_F_VIRTIO_NET_HDR is negotiated, sock_hlen is 0,
meaning that no header will be forwarded to the TAP device.
In the current vhost_net_build_xdp() implementation,
when sock_hlen == 0, the gso pointer can point at the start of the
Ethernet frame instead of a virtio-net header.
This results in a wrong interpretation of the destination MAC address
bytes as struct virtio_net_hdr fields.
This can, for some MAC addresses, trigger -EINVAL and return early
before the TX descriptor is completed, which can stall vhost-net TX.
Before 97b2409f28e0, the gso pointer was set to the zeroed padding area,
using it as a synthetic virtio-net header. Restore that behavior.
Fixes: 97b2409f28e0 ("vhost-net: reduce one userspace copy when building XDP buff") Signed-off-by: Enrico Zanda <enrico.zanda@arm.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Link: https://patch.msgid.link/20260708152242.2268848-1-enrico.zanda@arm.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
wifi: wilc1000: validate assoc response length before subtracting header
wilc_parse_assoc_resp_info() computes the trailing IE length as
ies_len = buffer_len - sizeof(*res);
without first checking that buffer_len is at least sizeof(struct
wilc_assoc_resp) (6 bytes). buffer_len is the length reported for a
received association response (host_int_parse_assoc_resp_info() passes
hif_drv->assoc_resp / assoc_resp_info_len straight in) and must be
validated before the driver accesses the fixed header.
For a frame shorter than the 6-byte fixed header, the subtraction wraps.
For a four-byte response the result is truncated to a u16 ies_len of
65534, so kmemdup() then attempts to copy 65534 bytes starting at
buffer + sizeof(*res), beyond the valid association-response data
(CWE-125). A response shorter than four bytes can also cause an
out-of-bounds read of res->status_code at offsets 2 and 3.
Reject frames too short to hold the fixed header before touching the
header or computing ies_len. Also set the connection status to a failure
on this path: the caller falls through to a
"conn_info->status == WLAN_STATUS_SUCCESS" check after the parser
returns, so leaving the status untouched could let a malformed short
response be treated as a successful association.
Georgi Valkov [Sun, 12 Jul 2026 22:17:09 +0000 (01:17 +0300)]
wifi: mwifiex: fix freeze for 60 seconds caused by request_firmware
Fix regression in rgpower table loading, caused by using
request_firmware(): when the requested firmware does not exist, e.g.
nxp/rgpower_WW.bin does not exist on OpenWRT builds for WRT3200ACM,
request_firmware() falls back to firmware_fallback_sysfs(), which expects
the firmware to be provided by user space using SYSFS. No such utility is
provided in this configuration, so the entire system locks up for 60
seconds, until the request times out. During this time, no other log
messages are observed, and the device does not respond to commands over
UART.
The request_firmware() call is performed in the following context:
current->comm kworker/1:2 in_task 1 irqs_disabled 0 in_atomic 0
Fixed by using request_firmware_direct(). This prevents fallback to SYSFS,
and avoids delay. The rgpower table is optional. The driver falls back
to the device tree power table if the firmware is not present.
The error code is printed for debugging and returned to the caller,
which only cares for success or failure, so there are no side effects.
Fixes: 7b6f16a25806 ("wifi: mwifiex: add rgpower table loading support") Signed-off-by: Georgi Valkov <gvalkov@gmail.com> Reviewed-by: Francesco Dolcini <francesco.dolcini@toradex.com> Link: https://patch.msgid.link/20260712221709.7099-1-gvalkov@gmail.com Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The program header scan handles PT_INTERP from a switch nested in the
scan loop, so its break leaves the switch and not the loop. A binary
carrying more than one PT_INTERP runs the case again and overwrites both
interpreter_name and interpreter. The previous name allocation leaks and
so does the previous interpreter reference, along with the write denial
open_exec() took on it. The denial is never released, so the file stays
unwritable for as long as the system runs.
An unprivileged caller reaches this with a crafted binary and repeats it
at will. binfmt_elf stops at the first PT_INTERP. Do the same here.
The flaw dates back to the driver's introduction in the pre-git history
tree introduced in v2.6.11 by 91808d6ebe39 ("[PATCH] FRV: Add FDPIC ELF
binary format driver").
Ruoyu Wang [Wed, 8 Jul 2026 14:34:15 +0000 (22:34 +0800)]
wan: wanxl: Only reset hardware after BAR mapping
wanxl_pci_init_one() stores the freshly allocated card in driver data
before the PLX BAR is mapped. Several early probe failures then unwind
through wanxl_pci_remove_one(), including failure to allocate the coherent
status area or to restore the DMA mask.
wanxl_pci_remove_one() unconditionally calls wanxl_reset(), and
wanxl_reset() dereferences card->plx. On those early failures card->plx
is still NULL, so the error path can dereference a NULL MMIO pointer.
Only issue the hardware reset once the BAR mapping exists. The remaining
cleanup in wanxl_pci_remove_one() already checks whether later resources
were allocated.
This issue was found by a static analysis checker and confirmed by
manual source review.
Ruoyu Wang [Wed, 8 Jul 2026 14:34:08 +0000 (22:34 +0800)]
nfp: Check resource mutex allocation
nfp_cpp_resource_find() allocates a CPP mutex handle for the matching
resource-table entry and then reports success. nfp_resource_try_acquire()
immediately passes that handle to nfp_cpp_mutex_trylock().
However, nfp_cpp_mutex_alloc() returns NULL on failure. If that happens
for a matching table entry, the resource lookup still returns success and
the following trylock dereferences a NULL mutex pointer while opening the
resource.
nfp_resource_acquire() already treats failure to allocate the table mutex
as -ENOMEM. Do the same for the resource mutex and fail the lookup before
publishing the rest of the resource handle.
This issue was found by a static analysis checker and confirmed by
manual source review.
Xiang Mei [Sat, 11 Jul 2026 21:03:02 +0000 (14:03 -0700)]
wifi: mac80211: tear down new links on vif update error path
When ieee80211_vif_update_links() adds new links it allocates a link
container for each and calls ieee80211_link_init() (which registers the
per-link debugfs files with file->private_data pointing into the container)
and ieee80211_link_setup(). If the subsequent drv_change_vif_links() fails,
the error path restores the old pointers and jumps to 'free', which frees
the new containers but never removes their debugfs entries or stops the
links. The debugfs files survive with file->private_data dangling at the
freed container, so a later open()+read() (e.g. link-1/txpower)
dereferences freed memory in ieee80211_if_read_link(), a use-after-free.
The removal path already dismantles links correctly via
ieee80211_tear_down_links(), which removes each link's keys and debugfs
entries and calls ieee80211_link_stop(); the add path on the error branch
does not. Commit be1ba9ed221f ("wifi: mac80211: avoid weird state in error
path") hardened this same error path for the link-removal case
(new_links == 0) but left the newly-added links' teardown unaddressed.
drv_change_vif_links() can fail at runtime on MLO drivers (internal
allocation / queue / firmware command failures).
Remove the new links' debugfs entries and stop them before freeing.
BUG: KASAN: slab-use-after-free in ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
Read of size 8 at addr ffff888011290000 by task exploit/145
Call Trace:
...
ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
short_proxy_read (fs/debugfs/file.c:373)
vfs_read (fs/read_write.c:572)
ksys_read (fs/read_write.c:716)
do_syscall_64 (arch/x86/entry/syscall_64.c:94)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
...
Oops: general protection fault, probably for non-canonical address 0xdffffc000000000a
RIP: 0010:ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
Kernel panic - not syncing: Fatal exception
Fixes: 170cd6a66d9a ("wifi: mac80211: add netdev per-link debugfs data and driver hook") Reported-by: Weiming Shi <bestswngs@gmail.com> Assisted-by: Claude:claude-opus-4-8 Signed-off-by: Xiang Mei <xmei5@asu.edu> Link: https://patch.msgid.link/20260711210302.2098404-1-xmei5@asu.edu Signed-off-by: Johannes Berg <johannes.berg@intel.com>
exec: fix unsigned loop counter wrap in transfer_args_to_stack()
The stop value is derived from bprm->p >> PAGE_SHIFT. The index variable
is an unsigned long. If bprm->p drops below PAGE_SIZE and stop becomes
zero the loop condition index >= stop is always true.
After the index == 0 iteration the decrement wraps to ULONG_MAX and
bprm->page[ULONG_MAX] reads sizeof(void *) bytes in front of the array.
The pointer has wrapped to -1. That garbage pointer is then passed to
kmap_local_page() and PAGE_SIZE bytes are copied from wherever that
lands into the stack of the process being created. And the loop doesn't
terminate either...
Getting there only requires bprm->p < PAGE_SIZE. On !MMU
bprm_set_stack_limit() and bprm_hit_stack_limit() are empty. So the only
constraint on how far bprm->p is pushed down is valid_arg_len(), i.e.
that each individual string still fits in what is left.
bprm->p starts at PAGE_SIZE * MAX_ARG_PAGES - sizeof(void *) so a
single argument or environment string of a little over 31 pages leaves
it in the first page:
Count down from MAX_ARG_PAGES so the loop ends when index reaches stop,
stop == 0 included. The iterations performed are unchanged for every
other value of stop.
Only CONFIG_MMU=n builds are affected, transfer_args_to_stack() is used
by binfmt_flat and binfmt_elf_fdpic on nommu only.
The loop predates git history. commit 7e7ec6a93434
("elf_fdpic_transfer_args_to_stack(): make it generic") only moved it
from binfmt_elf_fdpic.c into fs/exec.c and narrowed the copy to the used
part of the first page. The condition and the decrement are unchanged
from 2.6.12-rc2.
iommu/amd: Wait for completion instead of returning early in iommu_completion_wait()
need_sync is a per-IOMMU flag shared by all domains and devices behind
that IOMMU. It is set whenever a command is queued with sync == true and
cleared when a completion-wait (CWAIT) command is queued. However, a
cleared need_sync only means that a covering CWAIT has been queued, not
that all previously queued commands have actually completed in hardware.
iommu_completion_wait() read need_sync locklessly and returned early
when it was false. This breaks the "block until all previously queued
commands have completed" contract in a multi-CPU scenario:
CPU2 returns without waiting for any sequence number even though its
inv-B may not have completed yet (CWAIT(N), queued after inv-B, has not
been signaled). CPU2 then proceeds to, for example, free page-table
pages while the IOMMU can still walk stale translations, opening a
use-after-free window. This is a logical race in the meaning of the
flag, not a memory-visibility issue, so barriers alone do not help.
Fix it without losing the optimization of avoiding redundant CWAIT
commands: take iommu->lock before testing need_sync, and when it is
false do not return early but wait for the last allocated sequence
number (cmd_sem_val). Since need_sync == false implies no sync command
was queued after the last CWAIT, that CWAIT is FIFO-ordered after every
not-yet-completed command, so waiting for its sequence number guarantees
all prior commands (possibly queued by another CPU) have completed. The
common path with pending work is unchanged and no extra hardware command
is issued.
Paolo Bonzini [Tue, 21 Jul 2026 10:25:45 +0000 (12:25 +0200)]
Merge tag 'kvm-riscv-fixes-7.2-1' of https://github.com/kvm-riscv/linux into HEAD
KVM/riscv fixes for 7.2, take #1
- Avoid redundant page-table allocations in ioremap pcache topup
- Apply SBI FWFT LOCK flag only on successful set
- Bound SBI PMU counter mask scan to BITS_PER_LONG
- Skip TLB flush when G-stage PTE becomes valid with Svvptc
- Zicbo[m|z|p] block sizes should be always present in ONE_REG
- Inject instruction access fault on unmapped guest fetch
- Serialize virtual interrupt pending state updates using raw spinlock
- Fix Spectre-v1 in vector register access via ONE_REG
KVM: x86: Only reset TSC Deadline Timer in apic_timer_expired on KVM_RUN
On Intel platforms with a VMX preemption timer and APICv, if a VMM
calls KVM_GET_LAPIC before KVM_GET_MSRS to save the vCPU state, it is
possible to lose a pending timer interrupt.
If the thread running these ioctls is migrated to another core after
calling KVM_GET_LAPIC but before KVM_GET_MSRS and the guest is using
their LAPIC timer in TSC-deadline mode, not only does the save LAPIC
state not carry the pending interrupt, the TSCDEADLINE MSR will be
zeroed.
After migration across CPUs, KVM_GET_MSRS calls vcpu_load, posting the
interrupt and clearing the MSR:
vcpu_load() ->
kvm_arch_vcpu_load() ->
kvm_lapic_restart_hv_timer() ->
start_hv_timer() ->
apic_timer_expired() ->
kvm_apic_inject_pending_timer_irqs()
. post interrupt into the LAPIC state
. clear IA32_TSCDEADLINE
The saved LAPIC state will be missing the pending interrupt and the saved
MSR will be zero. Oops.
Fix by only posting an interrupt when we're attempting to enter the guest
(vcpu->wants_to_run == true), not for vcpu_load from other paths.
Assisted-by: gemini:gemini-3.1-pro-preview Debugged-by: David Matlack <dmatlack@google.com> Debugged-by: Sean Christopherson <seanjc@google.com> Debugged-by: Jim Mattson <jmattson@google.com> Debugged-by: James Houghton <jthoughton@google.com> Signed-off-by: Venkatesh Srinivas <venkateshs@chromium.org>
Message-ID: <20260715234234.15382-2-venkateshs@chromium.org> Reviewed-by: James Houghton <jthoughton@google.com> Reviewed-by: Chao Gao <chao.gao@intel.com> Cc: stable@vger.kernel.org Fixes: ae95f566b3d2 ("KVM: X86: TSCDEADLINE MSR emulation fastpath", 2020-05-15) Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
David Woodhouse [Mon, 20 Jul 2026 11:03:03 +0000 (12:03 +0100)]
KVM: selftests: sev_init2_tests: Derive SEV availability from KVM
The test asserted that the X86_FEATURE_SEV CPUID bit exactly matches
whether KVM offers KVM_X86_SEV_VM. That is not an invariant: when all
SEV ASIDs are assigned to SEV-SNP, KVM does not offer the SEV VM type
even though CPUID reports SEV, so the test aborts on an SNP-only host.
Derive SEV availability from KVM_CAP_VM_TYPES (as already done for SEV-ES
and SNP), assert only the one-way implication that a type offered by KVM
is also reported in CPUID, and TEST_REQUIRE() the SEV VM type so the test
skips cleanly when it is unavailable.
Reviewed-by: Tycho Andersen (AMD) <tycho@kernel.org> Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Message-ID: <5d3c345113748f39b7982e365d241abaf3e11086.1784545391.git.dwmw@amazon.co.uk> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
David Woodhouse [Mon, 20 Jul 2026 11:03:02 +0000 (12:03 +0100)]
KVM: selftests: sev_smoke_test: Only run VM types the host offers
sev_smoke_test ran the plain SEV subtest unconditionally, gated only on
the X86_FEATURE_SEV CPUID bit, while gating SEV-ES and SNP on the
KVM_CAP_VM_TYPES bits. CPUID reporting SEV does not mean KVM offers the
SEV VM type: when all SEV ASIDs are assigned to SEV-SNP, KVM_X86_SEV_VM
is unavailable even though X86_FEATURE_SEV is set. On such a host the
test aborts in KVM_CREATE_VM instead of exercising the available modes.
Gate the SEV subtest on KVM_CAP_VM_TYPES like the others, so the test
runs the VM types the host actually offers.
Reviewed-by: Tycho Andersen (AMD) <tycho@kernel.org> Signed-off-by: David Woodhouse <dwmw@amazon.co.uk>
Message-ID: <2b5e7a83d277134294199a455469bb436196b902.1784545391.git.dwmw@amazon.co.uk> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Phil Rosenthal [Sat, 18 Jul 2026 16:50:23 +0000 (12:50 -0400)]
KVM: x86/mmu: Fix use-after-free on vendor module reload
mmu_destroy_caches() destroys pte_list_desc_cache and
mmu_page_header_cache, but leaves both pointers unchanged. The pointers
live in kvm.ko, and therefore survive when a vendor module is unloaded
while kvm.ko remains loaded.
If creation of pte_list_desc_cache fails during a subsequent vendor
module load, its assignment sets pte_list_desc_cache to NULL and the
error path calls mmu_destroy_caches(). mmu_page_header_cache still
points to the cache destroyed during the preceding vendor module
unload. Passing that stale pointer to kmem_cache_destroy() causes a
slab use-after-free.
Reproduce the issue on a v7.1.3 kernel with CONFIG_KASAN=y,
CONFIG_KASAN_GENERIC=y, CONFIG_KVM=m, and CONFIG_KVM_INTEL=m. A
one-shot test hook forces pte_list_desc_cache to NULL on the second
invocation of kvm_mmu_vendor_module_init():
1. Load kvm.ko and kvm-intel.ko, creating both caches.
2. Unload only kvm_intel, leaving kvm.ko loaded.
3. Reload kvm_intel and force initialization through the -ENOMEM path.
KASAN reports:
BUG: KASAN: slab-use-after-free in
kvm_mmu_vendor_module_init+0x5b/0x170 [kvm]
...
kmem_cache_destroy+0x21/0x1d0
kvm_mmu_vendor_module_init+0x5b/0x170 [kvm]
...
Allocated by task 16817:
__kmem_cache_create_args+0x12c/0x3b0
__kmem_cache_create.constprop.0+0xb6/0xf0 [kvm]
kvm_mmu_vendor_module_init+0x13b/0x170 [kvm]
...
Freed by task 16820:
kmem_cache_destroy+0x117/0x1d0
kvm_mmu_vendor_module_exit+0x21/0x30 [kvm]
Clear both pointers immediately after destroying their caches so that
the stored state reflects the caches' lifetime and repeated cleanup is
safe.
With the fix applied, the same injected vendor module reload fails with
-ENOMEM as expected and produces no KASAN report.
Fixes: cb498ea2ce1d ("KVM: Portability: Combine kvm_init and kvm_init_x86") Cc: stable@vger.kernel.org Signed-off-by: Phil Rosenthal <phil@phil.gs>
Message-ID: <20260718-kvm-mmu-cache-uaf-v3-1-e103b93c74e1@phil.gs> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyunwoo Kim [Sun, 12 Jul 2026 01:14:50 +0000 (10:14 +0900)]
KVM: x86/mmu: Preserve nested TDP shadow page tables if they are used as roots
kvm_mmu_zap_oldest_mmu_pages() excludes a shadow page whose root_count
is non-zero from top-level reclaim, because such a page cannot be
freed. The path in mmu_page_zap_pte() that recursively zaps a parentless
nested TDP child has no such check. As a result, a shadow page can
be zapped even if the page itself can't be freed; as the comment in
kvm_mmu_zap_oldest_mmu_pages() notes, zapping it will just force vCPUs
to rebuild the page.
As in top-level reclaim, do not recursively prepare zapping of a
nested TDP child whose root_count is non-zero.
Fixes: 2de4085cccea ("KVM: x86/MMU: Recursively zap nested TDP SPs when zapping last/only parent") Signed-off-by: Hyunwoo Kim <imv4bel@gmail.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
KVM: x86: Check for invalid/obsolete root *after* making MMU pages available
Check for a "stale" page fault, i.e. for an invalid and/or obsolete root,
after making MMU pages available for the shadow MMU. If reclaiming shadow
pages zaps an in-use root, i.e. marks it invalid, then KVM will attempt to
map memory into an invalid root. On its own, populating an invalid root is
"fine", but because child shadow pages inherit their parent's role, any
children created during the map/fetch will be created as invalid pages,
thus violating KVM's invariant that invalid pages are never on the list of
active MMU pages.
Note, the underlying flaw has existed since KVM first started tracking
invalid roots in 2008 (commit 2e53d63acba7, "KVM: MMU: ignore zapped root
pagetables"), but the true badness only came along in 2020 (Linux 5.9)
with the invariant that invalid shadow pages can't be on the list of
active pages.
Note #2, inheriting role.invalid when creating child shadow pages is also
far from ideal; that flaw will be addressed separately.
Reported-by: Hyunwoo Kim <imv4bel@gmail.com> Fixes: f95eec9bed76 ("KVM: x86/mmu: Don't put invalid SPs back on the list of active pages") Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Hyunwoo Kim [Fri, 17 Jul 2026 10:30:11 +0000 (12:30 +0200)]
KVM: nVMX: Hide shadow VMCS right after VMCLEAR
free_nested() frees the shadow VMCS while vmcs01 still points to it. But
because it is asynchronous with respect to loaded_vmcs_clear(), the vCPU
might migrate before the pointer is cleared and __loaded_vmcs_clear()
may then execute VMCLEAR.
The VMCS needs to stay attached until its explicit VMCLEAR completes, but
then it can be hidden and the page safely freed.
Fixes: 355f4fb1405e ("kvm: nVMX: VMCLEAR an active shadow VMCS after last use") Cc: stable@vger.kernel.org Signed-off-by: Hyunwoo Kim <imv4bel@gmail.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Wayen Yan [Wed, 8 Jul 2026 11:35:29 +0000 (19:35 +0800)]
net: airoha: Fix DMA direction for NPU mailbox buffer
airoha_npu_send_msg() always maps the mailbox buffer with DMA_TO_DEVICE,
but some callers expect the NPU to write response data back into the
same buffer:
- airoha_npu_wlan_msg_get() (NPU_OP_GET): NPU writes response into
the buffer, then the caller reads it via memcpy()
- airoha_npu_ppe_stats_setup() (NPU_OP_SET): NPU writes back
npu_stats_addr field in the response
On non-cache-coherent architectures like EN7581 (Cortex-A53 without
hardware cache coherency for NPU DMA), DMA_TO_DEVICE unmap is a no-op
— it does not invalidate the CPU cache. If the NPU-written cache line
is still present in the CPU cache when the caller reads the buffer,
the CPU observes stale data instead of the NPU response.
This is a timing-sensitive bug: small mailbox buffers (~24 bytes)
typically fit in a single cache line and may survive in the cache
until the caller reads them, producing silent data corruption rather
than a crash. The bug is more likely to trigger when the caller reads
the response immediately after dma_unmap_single() without intervening
cache-evicting operations.
Fix by using DMA_BIDIRECTIONAL for both map and unmap, which ensures
dma_unmap_single() invalidates the CPU cache on non-coherent systems.
The mailbox buffers are small so there is no performance concern.
Fixes: c52918744ee1e49cea86622a2633b9782446428f ("net: airoha: npu: Move memory allocation in airoha_npu_send_msg() caller") Signed-off-by: Wayen Yan <win847@gmail.com> Acked-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://patch.msgid.link/178351055214.98729.11403147818632027428@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Guangshuo Li [Wed, 8 Jul 2026 11:17:37 +0000 (19:17 +0800)]
dpaa2-eth: put MAC endpoint device on disconnect
fsl_mc_get_endpoint() returns the MAC endpoint device with a reference
taken through device_find_child(). The Ethernet connect path stores that
device in mac->mc_dev and keeps it for the lifetime of the connected MAC
object.
However, the disconnect path only disconnects and closes the MAC before
freeing the dpaa2_mac object. It does not drop the endpoint device
reference stored in mac->mc_dev, so every successful connect leaks that
device reference when the MAC is later disconnected.
Drop the endpoint device reference after closing the MAC and before
freeing the dpaa2_mac object.
Fixes: 719479230893 ("dpaa2-eth: add MAC/PHY support through phylink") Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com> Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com> Reviewed-by: Ioana Ciornei <ioana.ciornei@nxp.com> Link: https://patch.msgid.link/20260708111738.750391-1-lgs201920130244@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Wayen Yan [Wed, 8 Jul 2026 11:16:16 +0000 (19:16 +0800)]
net: airoha: Fix potential use-after-free in airoha_ppe_deinit()
airoha_ppe_deinit() replaces the NPU pointer with NULL via
rcu_replace_pointer() but does not wait for existing RCU readers
to exit before calling ppe_deinit() and airoha_npu_put(). This can
cause a use-after-free if a reader in an RCU read-side critical
section still holds a reference to the NPU when it is freed.
The init path (airoha_ppe_init) already calls synchronize_rcu()
after rcu_assign_pointer(), but the deinit path introduced in
commit 6abcf751bc08 ("net: airoha: Fix schedule while atomic in
airoha_ppe_deinit()") omitted the matching barrier when switching
from rcu_read_lock()/rcu_dereference() to rcu_replace_pointer().
Add synchronize_rcu() before ppe_deinit() to ensure all existing
RCU readers have completed before the NPU resources are released.
Fixes: 6abcf751bc084804a9e5b3051442e8a2ce67f48a ("net: airoha: Fix schedule while atomic in airoha_ppe_deinit()") Signed-off-by: Wayen Yan <win847@gmail.com> Acked-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://patch.msgid.link/178351022574.97989.6880403520276841703@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Guangshuo Li [Wed, 8 Jul 2026 11:10:25 +0000 (19:10 +0800)]
dpaa2-switch: put MAC endpoint device on disconnect
fsl_mc_get_endpoint() returns the MAC endpoint device with a reference
taken through device_find_child(). The switch port connect path stores
that device in mac->mc_dev and keeps it for the lifetime of the connected
MAC object.
However, the disconnect path only closes the MAC and frees the dpaa2_mac
object. It does not drop the endpoint device reference stored in
mac->mc_dev, so every successful connect leaks that device reference when
the MAC is later disconnected.
Drop the endpoint device reference before freeing the dpaa2_mac object.
Fixes: 84cba72956fd ("dpaa2-switch: integrate the MAC endpoint support") Signed-off-by: Guangshuo Li <lgs201920130244@gmail.com> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260708111025.749311-1-lgs201920130244@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
====================
vsock/virtio: collapse receive queue under memory pressure
This series contains a patch (the first one) that is part of work I'm
doing to improve the tracking of memory used by AF_VSOCK sockets.
The second patch is a test for our suite that highlights the issue.
Since Brien reported an issue with his environment (based on Linux 6.12.y)
related to the work I’m doing, I extracted this patch and tried to make it
as easy as possible to backport. Brien tested it by backporting it to
6.12.y, which now contains the backport of the 059b7dbd20a6
("vsock/virtio: fix potential unbounded skb queue").
This patch primarily fixes STREAM sockets, but also partially fixes
SEQPACKET (with the exception of EOMs, which are kept in separate skbs to
avoid overcomplicating the code).
The rest of the work, I feel, is more net-next material and still needs
some work to be completed.
vsock/test: add test for small packets under pressure
Add a test that sends 2 MB of data using randomly sized small packets
(129-512 bytes) over a SOCK_STREAM connection. Packets above
GOOD_COPY_LEN (128) bypass the in-place coalescing in recv_enqueue(),
forcing each one into its own skb.
Without receive queue collapsing, the per-skb overhead eventually
exceeds buf_alloc and the connection is reset. The test verifies
that all data arrives and that content integrity is preserved.
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com> Acked-by: Michael S. Tsirkin <mst@redhat.com> Reviewed-by: Bobby Eshleman <bobbyeshleman@meta.com> Link: https://patch.msgid.link/20260708102904.50732-3-sgarzare@redhat.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
vsock/virtio: collapse receive queue under memory pressure
When many small packets accumulate in the receive queue, the skb overhead
can exceed buf_alloc even while the payload is within bounds. This causes
virtio_transport_inc_rx_pkt() to reject packets, leading to connection
resets during large transfers under backpressure.
The issue was reported by Brien, who has a reproducer, but it is also
easily reproducible with iperf-vsock [1] using a small packet size:
iperf3 --vsock -c $CID -l 129
which fails immediately without this patch but with commit 059b7dbd20a6
("vsock/virtio: fix potential unbounded skb queue").
Inspired by TCP's tcp_collapse() which solves a similar problem, add
virtio_transport_collapse_rx_queue() that walks the receive queue and
re-copies data into compact linear skbs to reduce the overhead.
The collapse is triggered proactively from when the number of skb queued
is close to exceeding the overhead budget.
A pre-scan counts the eligible bytes to size each allocation precisely,
avoiding waste for isolated small packets. Partially consumed skbs are
kept as-is to preserve buf_used/fwd_cnt accounting, EOM-marked skbs to
maintain SEQPACKET message boundaries, and skbs already larger than the
collapse target because they already have a good data-to-overhead ratio.
Walking a large queue may take a significant amount of time and cache
misses, causing traffic burstiness. To limit this, the collapse stops
once enough room is freed for this packet and the next one, but may
opportunistically free more to fill each collapsed skb to capacity.
Xuanqiang Luo [Wed, 8 Jul 2026 09:35:34 +0000 (17:35 +0800)]
rxrpc: fix io_thread race in rxrpc_wake_up_io_thread()
rxrpc_wake_up_io_thread() checks local->io_thread before waking it, but
then reloads the pointer for wake_up_process().
local->io_thread is cleared with WRITE_ONCE() when the I/O thread exits, so
the second load can see NULL even if the first load did not.
Take a READ_ONCE() snapshot and use it for both the NULL check and the
wake_up_process() call, as rxrpc_encap_rcv() already does.
Fixes: 5800b1cf3fd8 ("rxrpc: Allow CHALLENGEs to the passed to the app for a RESPONSE") Signed-off-by: Xuanqiang Luo <luoxuanqiang@kylinos.cn> Reviewed-by: Simon Horman <horms@kernel.org> Link: https://patch.msgid.link/20260708093534.53486-1-xuanqiang.luo@linux.dev Signed-off-by: Paolo Abeni <pabeni@redhat.com>
gtp: parse extension headers before reading inner protocol
GTPv1-U packets may carry a chain of extension headers before the inner
IP packet. The receive path already parses and skips these extension
headers, but it currently reads the inner protocol before doing so.
As a result, the first extension header byte is interpreted as the inner
IP version. Packets with extension headers are then dropped before PDP
lookup.
Parse the extension header chain before calling gtp_inner_proto(), so the
inner protocol is read from the actual inner IP header.
rds: drop incoming messages that cross network namespace boundaries
rds_find_bound() looks up the destination socket using a global
rhashtable keyed solely on (addr, port, scope_id). Network namespaces
are not part of the key, so a sender in netns A can deliver an incoming
message (inc) to a socket that lives in a different netns B.
When this happens, inc->i_conn points to an rds_connection whose c_net
is netns A, but the receiving rs lives in netns B. Once the child
process that created netns A exits, cleanup_net() calls
rds_loop_exit_net() -> rds_loop_kill_conns() -> rds_conn_destroy(),
freeing that connection. If the survivor socket in netns B still holds
the inc, any subsequent dereference of inc->i_conn is a use-after-free.
There are two dangerous sites in rds_clear_recv_queue():
1. inc->i_conn->c_lcong (offset 88 of freed rds_connection, size 200)
read via rds_recv_rcvbuf_delta() -- confirmed by KASAN.
2. inc->i_conn->c_trans->inc_free(inc) (function pointer at offset 80)
called via rds_inc_put() when the inc refcount reaches zero -- same
race window, potential call-through-freed-object primitive.
The bug is reachable from unprivileged user namespaces
(CLONE_NEWUSER + CLONE_NEWNET), available since Linux 3.8.
Fix this by rejecting the delivery in rds_recv_incoming() when the
socket returned by rds_find_bound() belongs to a different network
namespace than the connection that carried the message. Use the
existing rds_conn_net() / sock_net() helpers and net_eq() for the
comparison.
After PCIe DPC recovery, mlx5 reloads the affected functions and
replays multiport affiliation events. In the reported failure, the
first relevant device error was:
mlx5 recovered the PCI functions and resumed 0000:11:00.1. During
that resume, RDMA multiport binding replayed
MLX5_DRIVER_EVENT_AFFILIATION_DONE and mlx5e sent
MPV_DEVCOM_MASTER_UP. The host then panicked with:
MPV devcom registration publishes mlx5e private data to the component
peer list before mlx5e_devcom_init_mpv() stores the returned component
device in priv->devcom. A concurrent master-up event can therefore
reach a peer whose private data is visible but whose priv->devcom
backpointer is still NULL.
MPV_DEVCOM_MASTER_UP already carries the sender/master mlx5e private
data as event_data. The ready bit is stored on the shared devcom
component, not on an individual peer. Use the sender devcom when
marking the MPV component ready.
This preserves the readiness transition while avoiding a NULL
dereference of the peer devcom pointer during affiliation replay after
PCI error recovery.
OVS_ACTION_ATTR_TRUNC currently stores a delta from the original skb
length in OVS_CB(skb)->cutlen. When a later userspace action segments a
GSO skb, queue_gso_packets() reuses that delta for each smaller segment.
A segment can then reach queue_userspace_packet() with cutlen greater
than skb->len, underflowing the length passed to skb_zerocopy().
Store the maximum preserved length instead and bound each consumer
against the current skb length. Use U32_MAX as the no-truncation
sentinel so the value remains valid if skb geometry changes before a
consumer handles it.
net: airoha: fix MIB stats collection to be lossless
REG_FE_GDM_MIB_CLEAR after every read creates a race window where
packets arriving between read and clear are lost from statistics.
Switch to a delta-based approach instead:
- 64-bit H+L registers (ok pkts/bytes, E64..L1023): read absolute
hardware total directly into a local variable; clamp with max(new, old)
to prevent torn-read regression when the counter carries between the
two reads.
- tx/rx_len[0] ([0,64] bucket): combines RUNT_CNT (32-bit, delta via
tx_runt/rx_runt) and E64_CNT (64-bit, absolute) into a single local
accumulator; max(new, old) applied here too to guard against a torn
read of E64 when the RUNT accumulator is unchanged between polls.
MIB counters are zeroed by the SCU FE reset (EN7581_FE_RST) asserted
in airoha_hw_init() at module load, so no explicit MIB clear is needed
in airoha_fe_init().
Merge airoha_dev_get_hw_stats() into airoha_update_hw_stats() and
move stats_lock inside. Plain spin_lock() is correct: the function
is only called from ndo_get_stats64() in process context. Each dev
refreshes only its own MIB counters; sibling devs on a shared GDM3/4
port are polled when their own netdev is queried.
Fixes: 8f4695fb67b2 ("net: airoha: better handle MIBs for GDM ports with multiple devs attached") Signed-off-by: Aniket Negi <aniket.negi03@gmail.com> Acked-by: Lorenzo Bianconi <lorenzo@kernel.org> Link: https://patch.msgid.link/20260707152639.105628-1-aniket.negi03@gmail.com Signed-off-by: Paolo Abeni <pabeni@redhat.com>
drm_gpusvm_scan_mm() asks HMM to report the current CPU page-table
state without faulting missing entries by leaving default_flags set to
zero. The HMM PFN array is still caller-owned input/output state, and
the framework may preserve input bits while filling entries. It is not
safe for the caller to hand HMM an uninitialized array and then treat
entries without HMM_PFN_VALID as an authoritative unpopulated result.
Use kvcalloc() for the temporary PFN array so entries that are not
reported as valid start from the documented zero state. This prevents
random stack or heap contents from being interpreted as HMM PFN flags or
PFN values during the scan.
Fixes: f1d08a586482 ("drm/gpusvm: Introduce a function to scan the current migration state") Cc: stable@vger.kernel.org Signed-off-by: Stanislav Kinsburskii <skinsburskii@gmail.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Signed-off-by: Matthew Brost <matthew.brost@intel.com> Link: https://patch.msgid.link/178406967042.1113483.2116704310277917086.stgit@skinsburskii
net/iucv: fix use-after-free of a severed iucv_path
af_iucv queues not-yet-received message notifications on iucv->message_q,
each holding a raw pointer to the connection's iucv_path. When the peer
severs the connection, iucv_sever_path() frees that path with
iucv_path_free() but leaves the notifications queued. A later recvmsg()
drains message_q via iucv_process_message_q() and hands the stale path to
message_receive() -- a use-after-free of the freed iucv_path.
Drop the queued notifications when the path is severed; once the path is
gone they can no longer be received. This also frees the notifications
leaked when a socket is closed with messages still queued.
btrfs: raid56: fix scrub read assembly submitting no reads
Commit 5387bd958180 ("btrfs: raid56: remove sector_ptr structure")
converted the bio-list membership checks from sector pointers to
physical addresses. The two conversions in rmw_assemble_write_bios()
kept their polarity (skip the sector when it is NOT in the bio list,
i.e. when there is nothing to write), but scrub_assemble_read_bios()
has the opposite polarity -- skip the sector when it IS in the bio
list, because then there is nothing to read -- and the conversion
flipped it:
Since a parity-scrub rbio's bio list only holds the empty completion
bio, the result is that scrub_assemble_read_bios() submits no reads at
all. finish_parity_scrub() then compares the parity it computes from
the (cached, correct) data stripes against whatever happens to be in
the freshly allocated, uninitialized stripe pages:
- if the garbage differs from the computed parity, the sector is
"repaired" and written back -- accidentally producing the correct
on-disk result;
- if a recycled page happens to still hold the old (correct) parity
content, the sector is deemed clean, dropped from dbitmap, and the
actually-corrupt on-disk parity is left in place. (Scrub reports
no errors either way: there is no counter for P/Q corruption by
design, so the bug here is purely the failure to read and repair.)
The second case is intermittent because it depends on page-allocator
recycling. Observed with fstests btrfs/297 (raid5, 2 devices): the
corrupted P stripe intermittently stays corrupt after a scrub --
roughly 1/10 runs on x86-64 KVM and up to 7/8 on a UML build whose
timing favors page reuse.
Since the bio-list check can never be true for a parity-scrub rbio --
raid56_parity_alloc_scrub_rbio() adds a single empty completion bio
(asserting bi_size == 0), bio_paddrs[] is only populated by
index_rbio_pages() which is never called for BTRFS_RBIO_PARITY_SCRUB,
and rbio_can_merge() refuses to merge rbios of different operations --
remove the dead check entirely and assert the invariant instead, as
suggested by Qu Wenruo.
After this fix the injected corruption is read, detected and repaired
in every run (8/8 UML, 10/10 KVM), and the new assertion never fires
across the full fstests raid group.
btrfs: zoned: skip fully truncated ordered extents at zone finish
A fully truncated ordered extent (truncated_len == 0) wrote no data, so its
->csum_list is empty and btrfs_finish_ordered_zoned() trips:
assertion failed: !list_empty(&ordered->csum_list), in fs/btrfs/zoned.c:2141
Since commit 66ff4d366e7e a short or cancelled direct IO write finishes the
unsubmitted ordered extent as truncated with uptodate = true instead of
setting BTRFS_ORDERED_IOERR, so it now reaches btrfs_finish_ordered_zoned()
rather than being skipped by the IOERR check in btrfs_finish_ordered_io().
generic/208 hits this on a zoned filesystem.
Return early for these, like the BTRFS_ORDERED_PREALLOC case; there is no
zone append result to record and btrfs_finish_one_ordered() skips them too.
Fixes: 66ff4d366e7e ("btrfs: fix false IO failure after falling back to buffered write") Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: David Sterba <dsterba@suse.com>
Paul E. McKenney [Tue, 14 Jul 2026 20:32:24 +0000 (13:32 -0700)]
btrfs: initialize 'args' to avoid compiler warning in btrfs_ioctl_get_csums()
[COMPILER WARNING]
With GCC 11.5.0 and KASAN enabled on ARM, the following warning is
triggered during compiling:
In file included from ./include/asm-generic/rwonce.h:26,
from ./arch/arm64/include/asm/rwonce.h:81,
from ./include/linux/compiler.h:369,
from ./include/linux/array_size.h:5,
from ./include/linux/kernel.h:16,
from fs/btrfs/ioctl.c:6:
In function ‘instrument_copy_from_user_before’,
inlined from ‘_inline_copy_from_user’ at ./include/linux/uaccess.h:184:2,
inlined from ‘copy_from_user’ at ./include/linux/uaccess.h:222:9,
inlined from ‘btrfs_ioctl_get_csums.isra’ at fs/btrfs/ioctl.c:5220:6:
./include/linux/kasan-checks.h:38:27: warning: ‘args’ may be used uninitialized [-Wmaybe-uninitialized]
38 | #define kasan_check_write __kasan_check_write
./include/linux/instrumented.h:146:9: note: in expansion of macro ‘kasan_check_write’
146 | kasan_check_write(to, n);
| ^~~~~~~~~~~~~~~~~
fs/btrfs/ioctl.c: In function ‘btrfs_ioctl_get_csums.isra’:
./include/linux/kasan-checks.h:20:6: note: by argument 1 of type ‘const volatile void *’ to ‘__kasan_check_write’ declared here
20 | bool __kasan_check_write(const volatile void *p, unsigned int size);
| ^~~~~~~~~~~~~~~~~~~
fs/btrfs/ioctl.c:5201:43: note: ‘args’ declared here
5201 | struct btrfs_ioctl_get_csums_args args;
| ^~~~
[POSSIBLE FALSE ALERTS]
This seems to be a false alert from certain GCC versions.
The @args is immediately over-written by copy_from_user(), and there is
no code touching that @args until copy_from_user() finished correctly.
[WORKAROUND]
Initialize 'args' to zero, which suppresses the warning.
Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Paul E. McKenney <paulmck@kernel.org> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
reserve_chunk_space() stores the return value of
btrfs_zoned_activate_one_bg() in ret. The helper can return 1 after
successfully activating a block group, but ret is later used to decide
whether to reserve metadata for chunk tree updates.
As a result, successful activation skips btrfs_block_rsv_add() and leaves
trans->chunk_bytes_reserved unchanged. Use a separate variable for the
activation result so positive success does not affect the later
reservation. Keep activation failures in ret instead of returning early so
the function uses the common tail path.
Fixes: b6a98021e401 ("btrfs: zoned: activate necessary block group") CC: stable@vger.kernel.org Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Guanghui Yang <3497809730@qq.com> Signed-off-by: David Sterba <dsterba@suse.com>
btrfs: raid56: fix an incorrect csum skip during scrub
Commit 7425a2894019 ("btrfs: introduce btrfs_bio_for_each_block_all()
helper") uses the new helper to replace the nested loop inside
verify_bio_data_sectors(), which simplifies the code.
However that also changed the behavior of "continue" when a block has no
data checksum.
Previously the "continue" would skip the old for() loop, which would also
increase @total_sector_nr.
Now the "continue" will skip the new btrfs_bio_for_each_block_all()
loop, which doesn't update @total_sector_nr.
This means if we hit a block that has no data checksum, we will skip all
the remaining blocks no matter if they have data checksum.
As @total_sector_nr will never be updated, and that test_bit() will
always return false.
Fix it by increasing @total_sector_nr before calling "continue".
Fixes: 7425a2894019 ("btrfs: introduce btrfs_bio_for_each_block_all() helper") Reviewed-by: Daniel Vacek <neelx@suse.com> Signed-off-by: Qu Wenruo <wqu@suse.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
btrfs: report missing raid stripe tree root during lookup
When rescue=ibadroots ignores a failure to load the raid stripe tree root,
fs_info->stripe_root remains NULL. After the rescue mount proceeds, reading
file data that requires the raid stripe tree reaches
btrfs_get_raid_extent_offset().
Currently btrfs_search_slot() handles the NULL root and returns -EINVAL.
This avoids a NULL pointer dereference, but provides no diagnostic and
incorrectly describes missing filesystem metadata as an invalid argument.
Check stripe_root before allocating a path, emit a rate-limited error with
the logical address, and return -EUCLEAN.
Lookups with a valid stripe root are unchanged.
Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Dongjiang Zhu <zhudongjiang@fnnas.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
The same crash occurs with a corrupted raid stripe tree root, via
btrfs_read_block_groups() instead of fill_dummy_bgs().
[CAUSE]
With rescue=ibadroots, btrfs_read_roots() allows the mount to continue
when either root cannot be read, leaving the corresponding root pointer
NULL while its on-disk feature bit remains set.
btrfs_update_global_block_rsv() then dereferences the missing root based
on the feature bit alone.
[FIX]
Rescue mounts are fully read-only and cannot start transactions, so the
global reserve is never consumed. Under btrfs_is_full_ro(), mark the
reserve as full and return before performing the accounting.
And since we need to check if the fs is mount fully RO, export
fs_is_full_ro() as btrfs_is_full_ro(), and move it to fs.h.
Fixes: 8dbfc14fc736 ("btrfs: account block group tree when calculating global reserve size") Fixes: 515020900d44 ("btrfs: read raid stripe tree from disk") Suggested-by: Qu Wenruo <wqu@suse.com> Signed-off-by: Dongjiang Zhu <zhudongjiang@fnnas.com>
[ Squash the fs_is_full_ro() export commit into this one. ] Reviewed-by: Qu Wenruo <wqu@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
btrfs: zoned: reset meta_write_pointer on zone reset
btrfs_reset_unused_block_groups() resets a block group's zone and sets
alloc_offset back to 0 so the space can be reused, but it leaves
meta_write_pointer pointing at the previous end of the zone.
Once the block group is reactivated and reused for metadata, newly
allocated tree blocks live before that stale write pointer.
btrfs_check_meta_write_pointer() then sees them behind the write pointer,
so they can never be written out in sequential order: the dirty extent
buffers are stranded and pin their btree_inode folios until unmount.
Reset meta_write_pointer back to the start of the block group for
metadata and system block groups.
Fixes: 453a73c3069a ("btrfs: zoned: reclaim unused zone by zone resetting") Reviewed-by: Naohiro Aota <naohiro.aota@wdc.com> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: David Sterba <dsterba@suse.com>
btrfs: zoned: fix deadlock between metadata writeback and transaction commit
When writing out metadata extent buffers in a zoned filesystem,
btree_writepages() holds fs_info->zoned_meta_io_lock across the whole
writeback loop, including the call to btrfs_check_meta_write_pointer() ->
check_bg_is_active().
For the tree-log block group, check_bg_is_active() may fail to activate
the zone and fall back to btrfs_zone_finish_one_bg() to free an active
zone. That path waits for the running transaction to commit while still
holding zoned_meta_io_lock, but the committer needs that same lock to
write out the tree extents, so the two tasks deadlock:
Task A (kworker, metadata writeback) Task B (fsstress, transaction commit)
------------------------------------ -------------------------------------
wb_workfn() btrfs_commit_transaction(T)
btree_writepages() btrfs_write_and_wait_transaction()
btrfs_zoned_meta_io_lock() btrfs_write_marked_extents()
btrfs_check_meta_write_pointer() btree_writepages()
check_bg_is_active() [treelog_bg] btrfs_zoned_meta_io_lock()
btrfs_zone_finish_one_bg() <blocks on zoned_meta_io_lock,
btrfs_zone_finish() held by Task A>
do_zone_finish()
btrfs_inc_block_group_ro()
btrfs_wait_for_commit()
<blocks waiting for commit
of transaction T, done by
Task B>
The sibling branch in check_bg_is_active() already drops zoned_meta_io_lock
around do_zone_finish() for this exact reason. Do the same in the tree-log
branch: release the lock around btrfs_zone_finish_one_bg() and re-acquire
it afterwards. The lock only protects fs_info->active_{meta,system}_bg,
which this branch does not touch, and ctx->zoned_bg keeps a reference to
the block group across the unlock, so nothing is lost while the lock
is dropped.
This hang occasionally reproduces with fstests generic/475 on a zoned
btrfs filesystem.
Fixes: 13bb483d32ab ("btrfs: zoned: activate metadata block group on write time") Reviewed-by: Naohiro Aota <naohiro.aota@wdc.com> Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: David Sterba <dsterba@suse.com>
The above short script will be converted to a proper fstests case.
[CAUSE]
Inside btrfs_reconfigure(), if either btrfs_check_options() or
btrfs_check_features() failed, we will always have
BTRFS_FS_STATE_REMOUNTING set for the fs until the next successful
remount.
That BTRFS_FS_STATE_REMOUNTING flag will interrupt several operations,
including:
- Qgroup rescan
- Auto defrag
- Space reclaim
[FIX]
Change the error handling of btrfs_check_options() and
btrfs_check_features() to goto restore label.
Fixes: eddb1a433f26 ("btrfs: add reconfigure callback for fs_context") Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com> Signed-off-by: Qu Wenruo <wqu@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
cdrom: fix stack out-of-bounds read in CDROMVOLCTRL
mmc_ioctl_cdrom_volume() first reads the audio control mode page into a
32-byte stack buffer with cgc->buflen set to 24. If the device reports a
block descriptor, the function increases cgc->buflen to include that
descriptor and reads the page again.
For CDROMVOLCTRL, the function then builds a MODE SELECT parameter list
by moving cgc->buffer forward by offset - 8 bytes. This drops the block
descriptor from the outgoing payload and leaves a new 8-byte mode
parameter header in front of the audio control page. However, cgc->buflen
is left unchanged.
With a standard 8-byte block descriptor, cgc->buffer points at buffer + 8
but cgc->buflen remains 32. cdrom_mode_select() therefore asks the low
level packet path to write 32 bytes from that adjusted pointer, reading 8
bytes past the end of the 32-byte stack buffer.
This is not hit by CDROMVOLREAD, and CDROMVOLCTRL only triggers it on
drives that return a non-zero block descriptor length, which helps explain
why it has gone unnoticed. The overread is also sent to the device as
extra MODE SELECT payload, so it may not produce an obvious local failure.
Reduce cgc->buflen by the same amount as the buffer pointer adjustment so
the MODE SELECT transfer covers only the intended parameter list.
projects / thirdparty / linux.git / log
