🚨 CVE-2026-53422
Observable Response Discrepancy vulnerability in Erlang OTP ssh (ssh_sftpd module) allows an authenticated SFTP user to enumerate the existence of files and directories outside the configured root directory.

The SSH_FXP_REALPATH handler in ssh_sftpd calls relate_file_name/3 with Canonicalize=false, unlike every other SFTP operation handler. This allows .. components in the requested path to bypass the is_within_root/2 check without being resolved. The un-canonicalized path then enters resolve_symlinks/2, which walks up the directory tree above the configured root and issues read_link() syscalls on arbitrary filesystem paths.

An authenticated SFTP client can exploit this by sending a REALPATH request with a crafted traversal path. The server response differs depending on whether the target path exists on the host filesystem (SSH_FXP_NAME when the path resolves successfully, SSH_FX_NO_SUCH_FILE when it does not). This creates a path-existence oracle that an attacker can use to enumerate the filesystem structure outside the configured root, including the existence of sensitive files, directories, and mount points.

The vulnerability leaks only the existence of paths. No file contents, credentials, or write access are obtainable through this issue alone. The information gained may assist further attacks when combined with other vulnerabilities.

This vulnerability is associated with program files lib/ssh/src/ssh_sftpd.erl and program routine ssh_sftpd:handle_op/4.

This issue affects OTP from OTP 17.0 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssh from 3.0.1 before 6.0.2, 5.5.2.2 and 5.2.11.9.

🎖@cveNotify

🚨 CVE-2026-54886
Loop with Unreachable Exit Condition ('Infinite Loop') vulnerability in Erlang OTP ssh (ssh_sftpd module) allows an authenticated SFTP user to render an SFTP channel permanently unresponsive.

The handle_data/4 function in ssh_sftpd contains a catch-all clause that accepts channel data of any type. When channel data with a non-zero type code (SSH_MSG_CHANNEL_EXTENDED_DATA) arrives with an empty pending buffer and a payload at or below the SFTP packet size limit, the clause tail-calls itself with identical arguments, creating an infinite loop.

The SFTP protocol operates exclusively on normal channel data (type 0). Extended data (non-zero type) is meaningless for SFTP and is never sent by conforming clients. However, the SSH protocol permits any channel participant to send extended data on an open channel, so an authenticated SFTP client can trigger the loop by sending SSH_MSG_CHANNEL_EXTENDED_DATA with any data_type_code and any non-empty payload at or below the size limit.

The targeted ssh_sftpd process enters an infinite tail-recursive loop. It never processes another message, its message queue grows without bound, and it can only be stopped by killing the process. BEAM's reduction-based scheduler preemption continues to function, so other processes on the node are not starved, but each stuck channel process consumes its full CPU time share continuously and accumulates unbounded message queue memory. Opening many channels amplifies the CPU and memory impact.

Erlang/OTP SSH configurations using the default max_channels setting (infinity) allow an authenticated user to open unlimited channels per connection, amplifying the attack without requiring multiple TCP connections or authentications.

No file contents, credentials, or write access are obtainable through this issue. The impact is limited to denial of service on targeted SFTP channels, with secondary CPU degradation and memory growth.

This vulnerability is associated with program file lib/ssh/src/ssh_sftpd.erl and program routine ssh_sftpd:handle_data/4.

This issue affects OTP from OTP 17.0 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssh from 3.0.1 before 6.0.2, 5.5.2.2 and 5.2.11.9.

🎖@cveNotify

🚨 CVE-2026-54887
Use of Default Cryptographic Key vulnerability in Erlang/OTP ssl (DTLS server) allows predictable DTLS cookie computation during the startup window, enabling source address verification bypass.

On DTLS server startup, dtls_server_connection:initial_hello/3 initializes previous_cookie_secret to the empty binary (<<>>) instead of a random value. Because HMAC with an empty key is deterministic, anyone who observes the plaintext ClientHello can compute dtls_handshake:cookie(<<>>, IP, Port, Hello) and forge a valid DTLS cookie before the first rotation of the cookie secret. The DTLS cookie (RFC 6347 §4.2.1) is a denial-of-service mitigation that prevents spoofed source IPs from forcing the server to allocate state and perform expensive cryptographic operations; it is not an authentication mechanism. During the window from server startup until the first secret rotation (0 to 15 seconds), an attacker who can observe the plaintext ClientHello can bypass the source address verification, enabling DTLS handshake amplification with spoofed source addresses.

This vulnerability is associated with program file lib/ssl/src/dtls_server_connection.erl and program routine dtls_server_connection:initial_hello/3.

This issue affects OTP from OTP 20.0 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssl from 8.2 before 11.7.3, 11.6.0.3 and 11.2.12.10.

🎖@cveNotify

🚨 CVE-2026-54891
Improper Enforcement of Message Integrity During Transmission in a Communication Channel vulnerability in Erlang/OTP ssl (tls_gen_connection module) allows a network-positioned attacker to inject unauthenticated plaintext that the TLS client application later treats as authenticated server data.

The function tls_gen_connection:handle_protocol_record/3 rejects APPLICATION_DATA records that arrive in pre-handshake states when the TLS endpoint acts as a server, but does not apply the same check when the endpoint acts as a client. A network-positioned attacker can send plaintext APPLICATION_DATA records to the client during the handshake. The records are buffered and, once the handshake completes successfully, delivered to the application as if they were authenticated post-handshake data. The attacker cannot observe the client's response or steer the connection, so the impact is limited to blind injection of unauthenticated bytes. The injection window is wider for TLS versions prior to TLS 1.3 than for TLS 1.3.

This vulnerability is associated with program file lib/ssl/src/tls_gen_connection.erl.

This issue affects OTP from OTP 17.0 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssl from 5.3.4 before 11.7.3, 11.6.0.3 and 11.2.12.10. TLS 1.3 is affected starting with OTP 22.0, when TLS 1.3 support was added.

🎖@cveNotify

🚨 CVE-2026-55950
Time-of-check Time-of-use (TOCTOU) race condition vulnerability in Erlang/OTP ssl (dtls_packet_demux module) allows an unauthenticated remote attacker to crash all active DTLS sessions on a listener.

A DTLS server listener uses a single shared dtls_packet_demux gen_server process to route incoming UDP datagrams to the correct connection handler. When a DTLS client reconnects rapidly from the same source address and port (sending multiple ClientHello messages in quick succession), a race condition in the demux's internal gb_trees key-value store causes a {key_exists, {old, Client}} crash, terminating the demux process. Because the demux is shared across all DTLS associations on that listener, its crash immediately kills every active DTLS session, not just the attacker's.

The attack is pre-authentication: the attacker only needs to send UDP datagrams containing valid ClientHello messages from the same source IP and port before the intermediate DOWN monitor message is processed by the gen_server. No credentials, no completed handshake, and no special configuration are required, and the crash can be repeated indefinitely to create a persistent denial of service for all clients of that listener.

This vulnerability is associated with program file lib/ssl/src/dtls_packet_demux.erl.

This issue affects OTP from OTP 25.3 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssl from 10.9 before 11.7.3, 11.6.0.3 and 11.2.12.10.

🎖@cveNotify

🚨 CVE-2026-55952
The Erlang/OTP ssl application does not validate that the PSK identity list and binder list carried in a TLS 1.3 ClientHello pre-shared key extension have equal length before passing them to the session ticket handler. In tls_handshake_1_3:handle_pre_shared_key/3, an OfferedPreSharedKeys record with a mismatched number of identities and binders is forwarded directly to tls_server_session_ticket:use/4, which crashes the session ticket handler process.

An unauthenticated remote attacker can send a single crafted ClientHello to a TLS 1.3 server with session tickets enabled (stateful or stateless mode) and permanently disrupt session ticket handling on that listener. New TLS 1.3 handshakes complete but subsequently crash when the server attempts to issue a session ticket, effectively making TLS 1.3 unusable on the affected listener until the ssl application is restarted. TLS 1.2 connections are not affected.

This issue affects OTP from OTP 22.2 before OTP 29.0.3, OTP 28.5.0.3 and OTP 27.3.4.14, corresponding to ssl from 9.5 before 11.7.3, 11.6.0.3 and 11.2.12.10.

🎖@cveNotify

🚨 CVE-2026-53365
In the Linux kernel, the following vulnerability has been resolved:

vsock/virtio: fix zerocopy completion for multi-skb sends

When a large message is fragmented into multiple skbs, the zerocopy
uarg is only allocated and attached to the last skb in the loop.
Non-final skbs carry pinned user pages with no completion tracking,
so the kernel has no way to notify userspace when those pages are safe
to reuse. If the loop breaks early the uarg is never allocated at all,
leaking pinned pages with no completion notification.

Fix this by following the approach used by TCP: allocate the zerocopy
uarg (if not provided by the caller) before the send loop and attach
it to every skb via skb_zcopy_set(), which takes a reference per skb.
Each skb's completion properly decrements the refcount, and the
notification only fires after the last skb is freed.
On failure, if no data was sent, the uarg is cleanly aborted via
net_zcopy_put_abort().

This issue was initially discovered by sashiko while reviewing commit
1cb36e252211 ("vsock/virtio: fix MSG_ZEROCOPY pinned-pages accounting")
but was pre-existing.

🎖@cveNotify

🚨 CVE-2026-50652
Deserialization of untrusted data in Azure Active Directory allows an unauthorized attacker to deny service over a network.

🎖@cveNotify

🚨 CVE-2026-53366
In the Linux kernel, the following vulnerability has been resolved:

ipv4: account for fraggap on the paged allocation path

In __ip_append_data(), when the paged-allocation branch is taken,
alloclen and pagedlen are computed as

alloclen = fragheaderlen + transhdrlen;
pagedlen = datalen - transhdrlen;

datalen already includes fraggap, but the fraggap bytes carried over
from the previous skb are copied into the new skb's linear area at
offset transhdrlen by the subsequent skb_copy_and_csum_bits(). The
linear area is therefore undersized by fraggap bytes while pagedlen is
overstated by the same amount.

The non-paged branch sets alloclen to fraglen, which already accounts
for fraggap because datalen does. Bring the paged branch in line by
adding fraggap to alloclen and subtracting it from pagedlen.

After this adjustment, copy no longer collapses to -fraggap on the
paged path, so remove the stale comment describing that old arithmetic.

🎖@cveNotify

🚨 CVE-2026-53390
In the Linux kernel, the following vulnerability has been resolved:

ksmbd: fix out-of-bounds read in smb_check_perm_dacl()

The permission-check ACE walk in smb_check_perm_dacl() validates the ACE
header size and caps sid.num_subauth at SID_MAX_SUB_AUTHORITIES, but it
never checks that ace->size is actually large enough to contain
num_subauth sub-authorities before compare_sids() dereferences them.

CIFS_SID_BASE_SIZE covers the SID header up to but excluding the
sub_auth[] array, and offsetof(struct smb_ace, sid) is the ACE header,
so the existing guards only guarantee the 8-byte SID base, i.e. zero
sub-authorities. compare_sids() then reads ace->sid.sub_auth[i] for
i < min(local_sid->num_subauth, ace->sid.num_subauth). The local
comparison SIDs (sid_everyone, sid_unix_NFS_mode, and the id_to_sid()
result) always have at least one sub-authority, and an attacker controls
the ACE revision and authority bytes (which lie within the in-bounds SID
base), so they can match one of those SIDs and force the sub_auth read.

A crafted ACE with size == 16 and num_subauth >= 1 placed at the tail of
the security descriptor therefore causes a heap out-of-bounds read of up
to SID_MAX_SUB_AUTHORITIES * sizeof(__le32) bytes past the pntsd
allocation. The security descriptor is loaded by ksmbd_vfs_get_sd_xattr()
into a buffer sized exactly to the on-disk data (kzalloc(sd_size) in
ndr_decode_v4_ntacl()), so the read lands past the allocation. The
malformed descriptor can be stored verbatim via SMB2_SET_INFO (the DACL
is not normalised before being written to the security.NTACL xattr) and
the read fires on a subsequent SMB2_CREATE access check, making this
reachable by an authenticated client on a share that uses ACL xattrs.

Add the missing num_subauth-versus-ace_size check, mirroring the
identical guards already present in the sibling parsers parse_dacl() and
smb_inherit_dacl().

🎖@cveNotify

🚨 CVE-2026-53392
In the Linux kernel, the following vulnerability has been resolved:

NFSv4/flexfiles: reject zero filehandle version count

ff_layout_alloc_lseg() decodes the filehandle-version array count
from the flexfiles layout body. The value is used as the count for
kzalloc_objs(), and the current code only rejects NULL.

A zero count yields ZERO_SIZE_PTR, which can be stored in
dss_info->fh_versions even though later flexfiles paths assume that at
least one filehandle version exists.

Reject fh_count == 0 before the allocation, matching the existing zero
version_count validation in the flexfiles GETDEVICEINFO parser.

A QEMU/KASAN run with a malformed flexfiles layout hit:

KASAN: null-ptr-deref in range [0x0000000000000010-0x0000000000000017]
RIP: 0010:ff_layout_encode_ff_layoutupdate.isra.0+0x15f/0x750
ff_layout_encode_layoutreturn+0x683/0x970
nfs4_xdr_enc_layoutreturn+0x278/0x3a0
Kernel panic - not syncing: Fatal exception

The patched kernel rejects the malformed layout without KASAN/oops/panic,
and a valid fh_count=1 regression still opens, reads, and unmounts cleanly.

🎖@cveNotify

🚨 CVE-2026-53393
In the Linux kernel, the following vulnerability has been resolved:

nfsd: reset write verifier on deferred writeback errors

nfsd_vfs_write() and nfsd_commit() both call filemap_check_wb_err() to
detect deferred writeback errors, but neither rotates the server's write
verifier (nn->writeverf) when this check fails. Every other
durable-storage-failure path in these functions calls
commit_reset_write_verifier() before returning an error.

The missing rotation means clients holding UNSTABLE write data under the
current verifier will COMMIT, receive the unchanged verifier back, and
conclude their data is durable — silently dropping data that failed
writeback. This violates the UNSTABLE+COMMIT durability contract
(RFC 1813 §3.3.7, RFC 8881 §18.32).

Add commit_reset_write_verifier() calls at both filemap_check_wb_err()
error sites, matching the pattern used by adjacent error paths in the
same functions. The helper already filters -EAGAIN and -ESTALE
internally, so the calls are unconditionally safe.

🎖@cveNotify

🚨 CVE-2026-53399
In the Linux kernel, the following vulnerability has been resolved:

nfsd: release layout stid on setlease failure

nfs4_alloc_stid() publishes the new stid into cl->cl_stateids via
idr_alloc_cyclic() under cl_lock before returning to
nfsd4_alloc_layout_stateid(). When nfsd4_layout_setlease() then
fails, the error path frees the layout stateid directly with
kmem_cache_free() without ever calling idr_remove(), leaving the
IDR slot pointing at freed slab memory. Any subsequent IDR walker
(states_show, client teardown) dereferences the dangling pointer.

The correct teardown for an IDR-published stid is nfs4_put_stid(),
which removes the IDR slot under cl_lock, dispatches sc_free
(nfsd4_free_layout_stateid) to release ls->ls_file via
nfsd4_close_layout(), and drops the nfs4_file reference in its
tail.

A second issue blocks that switch: nfsd4_free_layout_stateid()
unconditionally inspects ls->ls_fence_work via
delayed_work_pending() under ls_lock, but
INIT_DELAYED_WORK(&ls->ls_fence_work, ...) currently runs only
after the setlease call. On the setlease-failure path the
destructor would touch an uninitialized delayed_work.

nfsd4_alloc_layout_stateid()
nfs4_alloc_stid() /* idr_alloc_cyclic under cl_lock */
nfsd4_layout_setlease() /* fails */
nfs4_put_stid()
nfsd4_free_layout_stateid()
delayed_work_pending(&ls->ls_fence_work) /* needs INIT */
nfsd4_close_layout() /* nfsd_file_put(ls->ls_file) */
put_nfs4_file()

Fix by hoisting the ls_fenced / ls_fence_delay / INIT_DELAYED_WORK
initialization above the nfsd4_layout_setlease() call, and replace
the manual nfsd_file_put + put_nfs4_file + kmem_cache_free cleanup
with a single nfs4_put_stid(stp).

🎖@cveNotify

🚨 CVE-2026-53400
In the Linux kernel, the following vulnerability has been resolved:

i2c: core: fix adapter registration race

Adapters can be looked up based on their id using i2c_get_adapter()
which takes a reference to the embedded struct device.

Make sure that the adapter (including its struct device) has been
initialised before adding it to the IDR to avoid accessing uninitialised
data which could, for example, lead to NULL-pointer dereferences or
use-after-free.

Note that the i2c-dev chardev, which is registered from a bus notifier,
currently uses i2c_get_adapter() so the adapter needs to be added to the
IDR before registration.

🎖@cveNotify

🚨 CVE-2026-63797
In the Linux kernel, the following vulnerability has been resolved:

rpmsg: char: Fix use-after-free on probe error path

rpmsg_chrdev_probe() stores the newly allocated eptdev in the default
endpoint's priv pointer before calling rpmsg_chrdev_eptdev_add(). If
rpmsg_chrdev_eptdev_add() then fails, its error path frees eptdev while
the default endpoint may still dispatch callbacks with the stale priv
pointer.

Avoid publishing eptdev through the default endpoint until
rpmsg_chrdev_eptdev_add() succeeds. Messages received before the priv
pointer is published should be ignored by rpmsg_ept_cb(). Flow-control
updates can hit rpmsg_ept_flow_cb() in the same window, so make both
callbacks return success when priv is NULL.

🎖@cveNotify

🚨 CVE-2026-63803
In the Linux kernel, the following vulnerability has been resolved:

hdlc_ppp: sync per-proto timers before freeing hdlc state

Each PPP control protocol (LCP/IPCP/IPV6CP) embedded in struct ppp
registers a timer via timer_setup(). That struct ppp is the
hdlc->state allocation, which detach_hdlc_protocol() frees with kfree()
in both teardown paths: unregister_hdlc_device() and the re-attach inside
attach_hdlc_protocol().

The ppp proto never registered a .detach callback, so
detach_hdlc_protocol() performs no timer synchronization before the
kfree(). The only cancel, timer_delete(&proto->timer) in ppp_cp_event(),
is partial (it does not wait for a running callback) and only runs on the
->CLOSED transition; ppp_stop()/ppp_close() do not sync either. A
ppp_timer callback already executing (blocked on ppp->lock) survives the
kfree and then dereferences proto->state / ppp->lock in freed memory,
leading to a use-after-free.

Fix this by adding a .detach helper that calls timer_shutdown_sync() on
every per-proto timer. detach_hdlc_protocol() invokes proto->detach(dev)
before kfree(hdlc->state), so timer_shutdown_sync()
now runs on both free paths.
timer_shutdown_sync() is used instead of timer_delete_sync() because the
keepalive path re-arms the timer through add_timer()/mod_timer() and
shutdown blocks any re-activation during teardown.

Initialize the per-protocol timers in ppp_ioctl() when the protocol is
attached, and remove the now-redundant timer_setup() from ppp_start(), so
that the timers are initialized exactly once at attach time and
ppp_timer_release() never operates on uninitialized timer_list
structures. attach_hdlc_protocol() uses kmalloc() (not kzalloc), so
struct ppp's protos[i].timer is uninitialized garbage until the first
timer_setup(); without this init-at-attach, attaching the PPP protocol
without ever bringing the device up would leave timer_shutdown_sync()
operating on uninitialized memory in .detach. Moving the init out of
ppp_start() (which only runs on NETDEV_UP) into the attach path makes the
initialization unconditional and avoids initializing the same timer_list
twice.

This bug was found by static analysis.

🎖@cveNotify

🚨 CVE-2026-63806
In the Linux kernel, the following vulnerability has been resolved:

KVM: Replace guest-triggerable BUG_ON() in ioeventfd datamatch with get_unaligned()

Drop a BUG_ON() that has been reachable since it was first added, way back
in 2009, and instead use get_unaligned() to perform potentially-unaligned
accesses.

For a given store, KVM x86's emulator tracks the entire value in the
destination operand, x86_emulate_ctxt.dst. If the destination is memory,
and the target splits multiple pages and/or is emulated MMIO, then KVM
handles each fragment independently. E.g. on a page split starting at page
offset 0xffc, KVM writes 4 bytes to the first page, then the remaining
bytes to the second page, using ctxt->dst as the source for both (with
appropriate offsets).

If the destination splits a page *and* hits emulated MMIO on the second
page, then KVM will complete the write to the first page, then emulate the
MMIO access to the second page. If there is a datamatch-enabled ioeventfd
at offset 0 of the second page, then KVM will process the remainder of the
store as a potential ioeventfd signal.

Putting it all together, if the guest emits a store that splits a page
starting at page offset N, and the second page has a datamatch-enabled
ioeventfd at offset 0, then KVM will check for datamatch using
&dst.valptr[N] as the source. Due to dst (and thus dst.valptr) being
32-byte aligned, if N is not aligned to @len, the BUG_ON() fires.

E.g. with a 16-byte store at page offset 0xffc, to an ioeventfd of len 8,
all initial checks in ioeventfd_in_range() will succeed, and the BUG_ON()
fires due to @val being 4-byte aligned, but not 8-byte aligned.

------------[ cut here ]------------
kernel BUG at arch/x86/kvm/../../../virt/kvm/eventfd.c:783!
Oops: invalid opcode: 0000 [#1] SMP
CPU: 0 UID: 1000 PID: 615 Comm: repro Not tainted 7.1.0-rc2-ff238429d1ea #365 PREEMPT
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
RIP: 0010:ioeventfd_write+0x6c/0x70 [kvm]
Call Trace:
<TASK>
__kvm_io_bus_write+0x85/0xb0 [kvm]
kvm_io_bus_write+0x53/0x80 [kvm]
vcpu_mmio_write+0x66/0xf0 [kvm]
emulator_read_write_onepage+0x12a/0x540 [kvm]
emulator_read_write+0x109/0x2b0 [kvm]
x86_emulate_insn+0x4f8/0xfb0 [kvm]
x86_emulate_instruction+0x181/0x790 [kvm]
kvm_mmu_page_fault+0x313/0x630 [kvm]
vmx_handle_exit+0x18a/0x590 [kvm_intel]
kvm_arch_vcpu_ioctl_run+0xc81/0x1c90 [kvm]
kvm_vcpu_ioctl+0x2d5/0x970 [kvm]
__x64_sys_ioctl+0x8a/0xd0
do_syscall_64+0xb7/0x890
entry_SYSCALL_64_after_hwframe+0x4b/0x53
RIP: 0033:0x7f19c931a9bf
</TASK>
Modules linked in: kvm_intel kvm irqbypass
---[ end trace 0000000000000000 ]---

In a perfect world, the fix would be to simply delete the BUG_ON(), as KVM
x86 doesn't perform alignment checks on "normal" memory accesses at CPL0.
Sadly, C99 ruins all the fun; while the x86 architecture plays nice,
dereferencing an unaligned pointer directly is undefined behavior in C,
e.g. triggers splats when running with CONFIG_UBSAN_ALIGNMENT=y.

🎖@cveNotify

🚨 CVE-2026-63810
In the Linux kernel, the following vulnerability has been resolved:

block: Avoid mounting the bdev pseudo-filesystem in userspace

The bdev pseudo-filesystem is an internal kernel filesystem with which
userspace should not interfere. Unregister it so that userspace cannot
even attempt to mount it.

This fixes a bug [1] that occurs when attempting to access files,
because the system call move_mount() uses pointers declared in the
inode_operations structure, which for the bdev pseudo-filesystem
are always equal to 0. `inode->i_op = &empty_iops;`

[1]

BUG: kernel NULL pointer dereference, address: 0000000000000000
#PF: supervisor instruction fetch in kernel mode
#PF: error_code(0x0010) - not-present page
PGD 23380067 P4D 23380067 PUD 23381067 PMD 0
Oops: 0010 [#1] PREEMPT SMP KASAN NOPTI
CPU: 2 PID: 17125 Comm: syz-executor.0 Not tainted 6.1.155-syzkaller-00350-g84221fde2681 #0
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014
RIP: 0010:0x0

Call Trace:
<TASK>
lookup_open.isra.0+0x700/0x1180 fs/namei.c:3460
open_last_lookups fs/namei.c:3550 [inline]
path_openat+0x953/0x2700 fs/namei.c:3780
do_filp_open+0x1c5/0x410 fs/namei.c:3810
do_sys_openat2+0x171/0x4d0 fs/open.c:1318
do_sys_open fs/open.c:1334 [inline]
__do_sys_openat fs/open.c:1350 [inline]
__se_sys_openat fs/open.c:1345 [inline]
__x64_sys_openat+0x13c/0x1f0 fs/open.c:1345
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x35/0x80 arch/x86/entry/common.c:81
entry_SYSCALL_64_after_hwframe+0x6e/0xd8

Found by Linux Verification Center (linuxtesting.org) with Syzkaller.

🎖@cveNotify

🚨 CVE-2026-63815
In the Linux kernel, the following vulnerability has been resolved:

f2fs: bound i_inline_xattr_size for non-inline-xattr inodes

When the flexible_inline_xattr feature is enabled, do_read_inode() loads
the on-disk i_inline_xattr_size unconditionally:

if (f2fs_sb_has_flexible_inline_xattr(sbi))
fi->i_inline_xattr_size = le16_to_cpu(ri->i_inline_xattr_size);

but sanity_check_inode() only range-checks it when the inode also has the
FI_INLINE_XATTR flag set. An inode that carries an inline dentry or inline
data but not FI_INLINE_XATTR -- the normal layout for an inline
directory -- therefore keeps a fully attacker-controlled
i_inline_xattr_size from a crafted image.

get_inline_xattr_addrs() returns that value with no flag gating, so it
feeds the inode geometry:

MAX_INLINE_DATA() = 4 * (CUR_ADDRS_PER_INODE - i_inline_xattr_size - 1)
NR_INLINE_DENTRY() = MAX_INLINE_DATA() * BITS_PER_BYTE / (...)
addrs_per_page() = CUR_ADDRS_PER_INODE - i_inline_xattr_size

A large i_inline_xattr_size drives MAX_INLINE_DATA() and NR_INLINE_DENTRY()
negative, so make_dentry_ptr_inline() sets d->max (int) to a negative
value. The inline directory walk then compares an unsigned long bit_pos
against that negative d->max, which is promoted to a huge unsigned bound,
and reads far past the inline area:

while (bit_pos < d->max) /* fs/f2fs/dir.c */
... test_bit_le(bit_pos, d->bitmap) / d->dentry[bit_pos] ...

Mounting a crafted image and reading such a directory triggers an
out-of-bounds read in f2fs_fill_dentries(); the same underflow also
corrupts ADDRS_PER_INODE for regular files.

Validate i_inline_xattr_size against MAX_INLINE_XATTR_SIZE whenever the
flexible_inline_xattr feature is enabled -- i.e. whenever the value is
loaded from disk and consumed -- and keep the lower MIN_INLINE_XATTR_SIZE
bound gated on inodes that actually carry an inline xattr, so legitimate
inodes with i_inline_xattr_size == 0 are still accepted.

🎖@cveNotify

🚨 CVE-2026-63816
In the Linux kernel, the following vulnerability has been resolved:

f2fs: atomic: fix UAF issue on f2fs_inode_info.atomic_inode

- ioctl(F2FS_IOC_GARBAGE_COLLECT_RANGE) - shrink
- f2fs_gc
- gc_data_segment
- ra_data_block(cow_inode)
- mapping = F2FS_I(inode)->atomic_inode->i_mapping
: f2fs_is_cow_file(cow_inode) is true
- f2fs_evict_inode(atomic_inode)
- clear_inode_flag(fi->cow_inode, FI_COW_FILE)
- F2FS_I(fi->cow_inode)->atomic_inode = NULL
...
- truncate_inode_pages_final(atomic_inode)
- f2fs_grab_cache_folio(mapping)
: create folio in atomic_inode->mapping
- clear_inode(atomic_inode)
- BUG_ON(atomic_inode->i_data.nrpages)

We need to add a reference on fi->atomic_inode before using its mapping
field during garbage collection, otherwise, it will cause UAF issue.

🎖@cveNotify

🚨 CVE-2026-63817
In the Linux kernel, the following vulnerability has been resolved:

f2fs: validate compress cache inode only when enabled

F2FS_COMPRESS_INO() uses NM_I(sbi)->max_nid as the synthetic inode
number for the compressed page cache inode. That inode only exists when
the compress_cache mount option is enabled.

When compress_cache is disabled, max_nid is outside the valid inode
range. A corrupted directory entry that points to ino == max_nid should
therefore be rejected by f2fs_check_nid_range(). However, is_meta_ino()
currently treats F2FS_COMPRESS_INO() as a meta inode unconditionally,
so f2fs_iget() bypasses do_read_inode() and its nid range check, and
instantiates a fake internal inode instead.

Gate the compressed cache inode case on COMPRESS_CACHE, matching
f2fs_init_compress_inode(). With compress_cache disabled, ino ==
max_nid now follows the normal inode path and is rejected as an
out-of-range nid.

🎖@cveNotify