Red Hat
Monthly
In the Linux kernel, the following vulnerability has been resolved: iommu/mediatek: fix use-after-free on probe deferral The driver is dropping the references taken to the larb devices during probe after successful lookup as well as on errors.
Local privilege-adjacent memory corruption in the Linux kernel ublk (userspace block driver) subsystem allows a ublk server process to trigger use-after-free-style access on completed I/O requests by exiting while requests are still in flight. The regression, introduced by commit e63d2228ef83 which stopped decrementing the request reference count in __ublk_fail_req(), leaves a leaked UBLK_REFCOUNT_INIT reference that permits illegitimate user-copy and zero-copy operations on already-completed requests and trips WARN_ON_ONCE reference-count assertions during queue teardown. It carries a CVSS 7.8 (high) with local vector and low privileges, EPSS is negligible at 0.03% (7th percentile), and there is no public exploit identified at time of analysis.
Local privilege escalation in the Linux kernel's ETS (net/sched) packet scheduler arises from a use-after-free on a struct Qdisc object, caused by a race between ets_qdisc_dequeue and ets_qdisc_change. When bands are reconfigured, a class can retain its entry on the active list after its child qdisc is freed and nulled, letting a concurrent dequeue operate on freed memory. Trend Micro ZDI reports a working reproducer and that an attacker able to create user and network namespaces can trigger the flaw and, per their analysis, control RIP; there is no CISA KEV listing and EPSS is very low (0.05%), so no public evidence of active exploitation.
Local privilege-escalation-capable memory corruption in the Linux kernel's ALPS PS/2 touchpad driver (drivers/input/mouse/alps.c) lets a use-after-free be triggered when an ALPS touchpad is detached while the dev3_register_work delayed work item is still pending. Because psmouse_disconnect() only flushes work already queued, alps_register_bare_ps2_mouse() can run and dereference the alps_data structure (priv->dev3) after alps_disconnect() has kfree()'d it, corrupting freed kernel heap memory. There is no public exploit identified at time of analysis, EPSS is negligible (0.02%), and the flaw was found by static analysis rather than in-the-wild activity.
Out-of-bounds memory access in the Linux kernel's dtv5100 DVB-USB media driver allows a local, low-privileged attacker to corrupt kernel memory via an unchecked, user-controlled rlen value in dtv5100_i2c_msg(). Because rlen is never validated against sizeof(st->data), a value larger than the buffer drives reads/writes past the allocated region, enabling denial of service and potential local privilege escalation. There is no public exploit identified at time of analysis, EPSS is very low (0.05%), and it is not listed in CISA KEV, but a vendor fix is available.
Kernel NULL pointer dereference in the Linux kernel's qla2xxx QLogic Fibre Channel driver crashes systems running in SCSI target mode during the command-abort path. The flawed code in __qla2x00_abort_all_cmds() assumed every command was TYPE_SRB and called sp->done() on TYPE_TGT_CMD entries, jumping to a NULL function pointer during ISP error recovery. The fix reverts the offending commit (0367076b0817) since the switch-case handling below it already covers the chip-down case correctly. EPSS is low (0.05%, 16th percentile) and there is no public exploit identified at time of analysis.
Local privilege escalation and kernel memory corruption in the Linux kernel's net/sched ETS (Enhanced Transmission Selection) qdisc allows a user with network-configuration capability to corrupt an internal linked list by toggling a scheduling class between DRR and strict modes. When an active DRR class is reconfigured to strict, ets_qdisc_change() fails to remove it from the active list, so switching it back to DRR performs a double list_add, corrupting kernel list pointers. No public exploit has been identified at time of analysis and the EPSS score is very low (0.05%), but the flaw is a classic kernel list-corruption primitive that can escalate to arbitrary write and full compromise.
Kernel memory corruption in the Linux svcrdma transport (server-side NFS-over-RDMA) arises because svc_rdma_copy_inline_range() added the page index (rc_curpage) instead of the byte offset (rc_pageoff) when computing the memcpy destination, so inline payload copies land at the wrong location within the receive pages. On systems exporting NFS over RDMA, a client on the RDMA fabric can drive copies that miscalculate their target offset, leading to out-of-bounds reads/writes in kernel page memory and potential information disclosure or corruption. There is no public exploit identified at time of analysis, EPSS is negligible (0.02%), and the issue is not in CISA KEV, but distro fixes (Ubuntu USN-8177-1) and upstream stable commits are available.
Local privilege escalation and memory corruption in the Linux kernel's KVM subsystem allows a user with VM control (e.g. a QEMU/hypervisor process) to trigger a use-after-free by clearing the KVM_MEM_GUEST_MEMFD flag on a memslot that was created with a guest_memfd binding. KVM blocks enabling the flag on an existing memslot but never blocked disabling it, leaving the guest_memfd instance bound to freed memory that is later written during kvm_gmem_release. The flaw carries a CVSS of 7.8 and has a vendor-released fix; it is not in CISA KEV and no public exploit has been identified at time of analysis, with an EPSS of 0.02%.
Concurrent SMB delete-on-close handling in the Linux kernel's in-kernel SMB3 server (ksmbd) mishandles locking on the ksmbd_inode->m_flags field, letting racing open/close/delete operations corrupt delete-on-close and pending-delete state. Affected builds ship the pre-fix ksmbd vfs_cache code where helpers such as ksmbd_inode_pending_delete() and ksmbd_fd_set_delete_on_close() read and modify m_flags without holding ci->m_lock. The result is inconsistent delete semantics - files that persist after a delete-on-close or vanish while still in use - with no public exploit identified at time of analysis and a very low EPSS score (0.03%).
Local privilege escalation risk in the Linux kernel's vidtv (Virtual Digital TV) media test driver stems from a double-free and use-after-free condition in vidtv_channel_si_init(), where local pointers are not reset to NULL after ownership of program/service/event list memory is transferred to the PAT/SDT/EIT tables. When vidtv_psi_pmt_create_sec_for_each_pat_entry() fails and control jumps to the free_eit error path, already-freed memory is destroyed a second time. Publicly this is patched upstream and shipped via Ubuntu USN-8096-1, but with EPSS at 0.05% and no public exploit identified at time of analysis, real-world exploitation is unlikely and confined to systems that actually load this virtual test driver.
Out-of-bounds buffer handling in the Linux kernel's ksmbd (in-kernel SMB3 server) Extended Attribute code allows a remote attacker to disclose adjacent kernel memory by sending a crafted SMB2 SET_INFO request. The smb2_set_ea() function validated the request buffer without accounting for the null-terminator byte of the EA name, so the size check under-counted the required length and permitted reads past the intended boundary. No public exploit is identified at time of analysis and EPSS probability is low (0.04%), but a vendor/upstream fix is available and it is not listed in CISA KEV.
Local privilege escalation and memory corruption in the Linux kernel FUSE subsystem's io-uring path allows an unprivileged local user to trigger list corruption and a resulting use-after-free. The flaw stems from a terminated-but-not-committed FUSE io-uring request never being unlinked from the queue's list, leaving a dangling entry that corrupts kernel memory. EPSS is very low (0.03%, 7th percentile) and there is no public exploit identified at time of analysis; a fix has been merged upstream and shipped by distributions such as Ubuntu.
Incorrect access-control enforcement in the Linux kernel's NFSv4 server (nfsd) causes POSIX ACLs supplied during file creation to be silently discarded, so a client that sets an ACL with a named principal ends up with only a default mode-derived ACL instead. This violates RFC 8881 6.4.1.3 and means files intended to carry restrictive or explicitly-granted permissions are instead governed by permissive default mode bits, potentially exposing their contents to unintended principals on the export. There is no public exploit identified at time of analysis, EPSS is very low (0.05%), and it is not on CISA KEV - this is a security-relevant correctness defect rather than an actively exploited threat.
Local privilege-holder-triggered use-after-free in the Linux kernel's mlxsw Spectrum switch driver (drivers/net/ethernet/mellanox/mlxsw) lets a network-admin operation on Nvidia/Mellanox Spectrum ASIC hardware dereference a freed neighbour object during RIF teardown. The driver cached a neighbour pointer without holding a reference, so freeing the neighbour (e.g. while removing a device from a VRF) leaves a dangling pointer that mlxsw_sp_neigh_entry_update() reads, corrupting kernel memory. CVSS is 7.8 (local, low privilege in the automated NVD scoring) but EPSS is only 0.05% (16th percentile) and there is no public exploit or CISA KEV listing - the flaw was found via an internal stress test, not observed attacks.
Local privilege escalation potential arising from a use-after-free in the Linux kernel's mlxsw Spectrum driver (spectrum_mr) allows an attacker with network-administration rights on a Mellanox Spectrum switch to trigger memory corruption during multicast route replacement. A prior fix introduced a dedicated mutex to protect the multicast route list, but one list-entry deletion path during route replace was missed, so the periodic mlxsw_sp_mr_stats_update worker can read a freed route object. EPSS is low (0.05%) and there is no public exploit identified at time of analysis; a vendor patch is available across multiple stable branches.
Denial of service and potential kernel memory disclosure in the Linux kernel's CAIF protocol stack arises because cffrml_receive() subtracts 2 from a packet-derived length field without checking that the value is at least 2 when FCS checking is disabled. An adjacent attacker sending a CAIF frame whose length field is 0 or 1 triggers an integer underflow, leading to memory exhaustion, kernel instability, and possible leakage of uninitialized kernel memory contained in packet padding. There is no public exploit identified at time of analysis, EPSS risk is very low (0.05%), and the flaw is not in CISA KEV.
Local privilege-level buffer overflow in the Linux kernel's ethtool subsystem allows an attacker who can trigger a device stat-count change between the ETHTOOL_GSSET_INFO, ETHTOOL_GSTRINGS, and ETHTOOL_GSTATS ioctls to overflow a userspace stats buffer, corrupting memory and potentially escalating privileges. The flaw stems from a TOCTOU race where drivers using dynamic counters (mlx5, bnx2x, bna, ksz884x) can report a different stat count than the buffer sized by userspace. It is not in CISA KEV, EPSS exploitation probability is very low (0.06%), and no public exploit has been identified, though a vendor patch is available.
Local privilege escalation or kernel memory disclosure in the Linux kernel's amdgpu DRM driver stems from a use-after-free during GPU recovery, where a race between the scheduler timeout callback (drm_sched_job_timedout) and the TDR work queue can free a job structure before amdgpu_device_gpu_recover reads its pasid field. Affected systems are those running vulnerable kernels with AMD GPUs using the amdgpu driver; a local attacker able to submit GPU work and induce a GPU hang could trigger the freed-memory access, as demonstrated by the supplied KASAN slab-use-after-free trace. There is no public exploit identified at time of analysis, and the EPSS score is negligible (0.03%, 7th percentile), consistent with a difficult-to-win kernel race rather than a broadly exploited flaw.
Memory corruption in the Linux kernel's TPM 2.0 session-handling code (tpm2-sessions) arises because the 'name_size' value derived from a TPM_ALG_ID is used to index without range validation, so an unrecognized or malicious algorithm identifier can drive an out-of-bounds access. The affected TPM HMAC/authorization session path is reachable from local, low-privileged kernel-driver interactions; at worst this leads to kernel memory corruption affecting confidentiality, integrity, and availability. No public exploit is identified at time of analysis and the EPSS probability is negligible (0.03%), consistent with a locally-scoped kernel hardening fix rather than a mass-exploited flaw.
Out-of-bounds memory access in the Linux kernel's Open vSwitch (openvswitch) module allows a local user holding CAP_NET_ADMIN to leak kernel memory or crash the host by submitting a malformed OVS_ACTION_ATTR_PUSH_NSH netlink action with an incorrectly sized middle attribute. The push_nsh() action path validates the outer and inner NSH attributes but never checks the intermediate OVS_KEY_ATTR_NSH wrapper, so nsh_hdr_from_nlattr() performs a 184-byte read past a slab allocation (KASAN slab-out-of-bounds). A patch is available from the kernel stable tree and via Ubuntu USN-8096-1; there is no public exploit identified at time of analysis, and EPSS is very low (0.07%).
Remote denial of service in the Linux kernel SCSI target (LIO) subsystem allows an initiator to crash a storage server by triggering a NULL-pointer dereference on the command error path. When allocation of the per-command CDB buffer (cmd->t_task_cdb) fails, the pointer is left NULL yet subsequently dereferenced in the 'err' cleanup path, panicking the kernel. No public exploit identified at time of analysis and EPSS is very low (0.05%), consistent with a hard-to-trigger memory-pressure condition rather than a reliably weaponizable flaw.
Refcount underflow in the Linux kernel's net/handshake subsystem lets duplicate cancellation of the same TLS handshake request drop an extra reference on the underlying socket, corrupting its reference count and leaking/freeing the socket structure. The flaw affects kernels using the in-kernel TLS handshake upcall path (notably SUNRPC-over-TLS via tlshd), where a handshake timeout can trigger two cancellation requests - one from xs_reset_transport() and one from xs_tls_handshake_sync() - for a request still present in handshake_rhashtbl. Despite an assigned 9.8 CVSS, the practical impact is a kernel memory-safety fault (crash / potential use-after-free) rather than trivial remote code execution, and there is no public exploit identified at time of analysis; EPSS is very low (0.03%).
Denial of service in the Linux kernel's HFS+ (hfsplus) filesystem driver allows a local user with access to a mounted HFS+ volume to crash the kernel via a reference-counting race in __hfs_bnode_create(). When sync() and link() operations run concurrently, two threads can both attempt to create the same B-tree node; the thread that loses the race reuses the already-hashed node without incrementing its refcount, so a later put trips BUG_ON(!atomic_read(&node->refcnt)) and panics the system. No public exploit has been identified at time of analysis, EPSS risk is very low (0.05%), and the issue is not in CISA KEV.
Denial of service in the Linux kernel's bnxt_en driver (Broadcom NetXtreme-E) affects systems running an XDP program that uses the XDP_TX action. Incorrect clearing of NAPI event flags in the XDP_TX path causes the TX doorbell to be skipped when the TX ring is full, stalling the associated RX ring so it is never refilled and all subsequent packets are silently dropped. There is no public exploit identified at time of analysis, and EPSS exploitation probability is very low (0.02%, 6th percentile); an upstream fix is available in mainline and stable kernels.
Libpng versions 1.6.26 through 1.6.53 contain an integer truncation flaw in the simplified write API functions that triggers a heap buffer over-read when processing images with negative row strides or strides exceeding 65535 bytes. Local attackers can exploit this to read sensitive heap memory, potentially disclosing application data. No patch is currently available; users should avoid processing untrusted PNG images with these vulnerable libpng versions.
Libpng versions 1.6.51-1.6.53 contain a heap buffer over-read in the simplified API function png_image_finish_read when processing interlaced 16-bit PNG images with 8-bit output and non-minimal row stride, allowing local attackers to read out-of-bounds memory through a malicious image file. Public exploit code exists for this regression, which was introduced by a previous security fix. Upgrade to version 1.6.54 to remediate.
Fulcio versions prior to 1.8.5 allow unauthenticated attackers to bypass MetaIssuer URL validation through unanchored regex patterns, enabling blind SSRF attacks against internal services. Although the vulnerability is limited to read-only GET requests with no response exfiltration, attackers can probe internal networks to discover active services and infrastructure. Public exploit code exists for this medium-severity issue, and a patch is available in version 1.8.5.
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, avahi-daemon can be crashed by sending 2 unsolicited announcements with CNAME resource records 2 seconds apart. [CVSS 6.5 MEDIUM]
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, avahi-daemon can be crashed by sending unsolicited announcements containing CNAME resource records pointing it to resource records with short TTLs. [CVSS 6.5 MEDIUM]
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, an unprivileged local users can crash avahi-daemon (with wide-area disabled) by creating record browsers with the AVAHI_LOOKUP_USE_WIDE_AREA flag set via D-Bus. [CVSS 5.5 MEDIUM]
Vllm versions up to 0.12.0 is affected by allocation of resources without limits or throttling (CVSS 6.5).
Cosign's bundle verification mechanism fails to properly validate that embedded Rekor entries reference the correct artifact digest, signature, and public key, allowing an attacker with a compromised signing identity to forge valid bundles and bypass transparency log verification. A malicious actor could exploit this to create counterfeit signatures that pass validation checks, affecting users relying on Cosign for container and binary code signing verification. Public exploit code exists for this vulnerability; patches are available in versions 2.6.2 and 3.0.4.
Virtualenv versions up to 20.36.1 is affected by improper link resolution before file access (CVSS 4.5).
Python's filelock SoftFileLock implementation prior to version 3.20.3 contains a TOCTOU race condition that allows local attackers with symlink creation privileges to interfere with lock file operations between permission validation and file creation. An attacker can exploit this window to create a symlink at the target lock path, causing lock operations to fail or redirect to unintended files, resulting in denial of service or unexpected behavior. Upgrade to filelock version 3.20.3 or later to remediate this vulnerability.
HarfBuzz text shaping engine versions prior to 12.3.0 crash when the SubtableUnicodesCache::create function attempts to dereference a null pointer returned by failed memory allocation, enabling denial of service in applications processing untrusted font data. Public exploit code exists for this vulnerability. A patch is available in version 12.3.0 and later.
pypdf versions prior to 6.6.0 are vulnerable to denial of service through CPU exhaustion when processing malformed PDF files with crafted startxref entries in non-strict reading mode. An attacker can create a specially crafted PDF containing excessive whitespace that causes the library to consume significant processing resources during cross-reference table reconstruction. A patch is available in version 6.6.0 and later.
Denial of service via resource exhaustion in pypdf prior to version 6.6.0 allows remote attackers to trigger excessive processing times by submitting specially crafted PDF files with missing /Root objects and inflated /Size values. The vulnerability only affects non-strict parsing mode and causes the library to consume significant CPU resources when processing otherwise invalid documents. A patch is available in version 6.6.0 and later.
React Router is a router for React. In @remix-run/server-runtime version prior to 2.17.3. [CVSS 6.5 MEDIUM]
React-Router versions up to 6.30.1 is affected by url redirection to untrusted site (open redirect) (CVSS 6.5).
A memory initialization issue was addressed with improved memory handling. This issue is fixed in tvOS 26.2, Safari 26.2, watchOS 26.2, visionOS 26.2, iOS 26.2 and iPadOS 26.2, macOS Tahoe 26.2. [CVSS 4.3 MEDIUM]
A flaw was found in GNU Wget2. This vulnerability, a stack-based buffer overflow, occurs in the filename sanitization logic when processing attacker-controlled URL paths, particularly when filename restriction options are active. [CVSS 7.6 HIGH]
A security issue was discovered in GNU Wget2 when handling Metalink documents. The application fails to properly validate file paths provided in Metalink <file name> elements. [CVSS 8.8 HIGH]
The ECDSA implementation of the Elliptic package generates incorrect signatures if an interim value of 'k' (as computed based on step 3.2 of RFC 6979 https://datatracker.ietf.org/doc/html/rfc6979 ) has leading zeros and is susceptible to cryptanalysis, which can lead to secret key exposure. [CVSS 5.6 MEDIUM]
Authlib is a Python library which builds OAuth and OpenID Connect servers. [CVSS 5.7 MEDIUM]
CoreDNS is a DNS server that chains plugins. Prior to version 1.14.0, multiple CoreDNS server implementations (gRPC, HTTPS, and HTTP/3) lack critical resource-limiting controls. An unauthenticated remote attacker can exhaust memory and degrade or crash the server by opening many concurrent connections, streams, or sending oversized request bodies. The issue is similar in nature to CVE-2025-47950 (QUIC DoS) but affects additional server types that do not enforce connection limits, stream limit...
HTML injection in Preact and React through unsafe JSON deserialization allows remote attackers to inject arbitrary scripts when applications pass unsanitized user-controlled data as component children, with public exploit code already available. The vulnerability stems from a regression in Preact 10.26.5 that weakened protections against constructing Virtual DOM elements from malicious JSON payloads. Affected applications are vulnerable if they consume unvalidated external data without sanitization and lack additional mitigations like Content Security Policy.
The RSA crate versions prior to 0.9.10 crash when constructing private keys with invalid prime components (such as 1), allowing an attacker to trigger a denial of service by providing malformed key material. This affects applications using the vulnerable RSA library for cryptographic operations. A patch is available in version 0.9.10 and later.
A flaw was found in Ansible Automation Platform (AAP). Read-only scoped OAuth2 API Tokens in AAP, are enforced at the Gateway level for Gateway-specific operations. [CVSS 8.5 HIGH]
Bokeh versions 3.8.1 and below allow attackers to bypass Origin validation in WebSocket connections by registering domains that suffix-match allowlisted domains (e.g., dashboard.corp.attacker.com for allowlist entry dashboard.corp), enabling unauthorized server interaction. Public exploit code exists for this vulnerability, which could allow attackers to access sensitive data or modify visualizations on behalf of victims. The issue is resolved in Bokeh 3.8.2.
Stack-based buffer overflow in libtasn1 version: v4.20.0. The function fails to validate the size of input data resulting in a buffer overflow in asn1_expend_octet_string. [CVSS 7.5 HIGH]
Visualdata versions up to - is affected by inefficient regular expression complexity (redos) (CVSS 5.3).
Quarkus is a Cloud Native, (Linux) Container First framework for writing Java applications. Prior to versions 3.31.0, 3.27.2, and 3.20.5, a vulnerability exists in the HTTP layer of Quarkus REST related to response handling. When a response is being written, the framework waits for previously written response chunks to be fully transmitted before proceeding. If the client connection is dropped during this waiting period, the associated worker thread is never released and becomes permanently b...
Undertow HTTP server (used in WildFly, JBoss EAP) fails to validate Host headers, enabling cache poisoning, internal network scanning, and session hijacking. Affects a widely-used Java application server component.
Crypt::Sodium::XS for Perl bundles a vulnerable version of libsodium (<= 1.0.20) that has a signature verification flaw. In atypical use cases with custom cryptography, this can compromise data authenticity guarantees. Patch available.
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. In versions 3.13.2 and below, handling of chunked messages can result in excessive blocking CPU usage when receiving a large number of chunks. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below allow a request to be crafted in such a way that an AIOHTTP server's memory fills up uncontrollably during processing. [CVSS 7.5 HIGH]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below allow for an infinite loop to occur when assert statements are bypassed, resulting in a DoS attack when processing a POST body. [CVSS 7.5 HIGH]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below contain parser logic which allows non-ASCII decimals to be present in the Range header. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below enable an attacker to ascertain the existence of absolute path components through the path normalization logic for static files meant to prevent path traversal. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below of the Python HTTP parser may allow a request smuggling attack with the presence of non-ASCII characters. [CVSS 6.5 MEDIUM]
vega-functions provides function implementations for the Vega expression language. Prior to version 6.1.1, for sites that allow users to supply untrusted user input, malicious use of an internal function (not part of the public API) could be used to run unintentional javascript (XSS). [CVSS 7.2 HIGH]
Vega is a visualization grammar, a declarative format for creating, saving, and sharing interactive visualization designs. Prior to versions 6.1.2 and 5.6.3, applications meeting two conditions are at risk of arbitrary JavaScript code execution, even if "safe mode" expressionInterpreter is used. [CVSS 8.1 HIGH]
Security-flag inheritance bypass in the Linux kernel NFS client allows a low-privileged local user to obtain access on automounted NFS submounts that ignores the restrictive superblock options (ro, noexec, nodev, sync) set on the parent mount. Because automounted filesystems fail to preserve these user-set mount flags, protections such as read-only or no-execute enforcement can be silently dropped on the child mount, undermining the security posture an administrator intended. No public exploit identified at time of analysis, and the EPSS score is negligible (0.05%, 16th percentile), consistent with a local, configuration-dependent defense-in-depth weakness rather than a mass-exploited flaw.
Memory corruption in the Linux kernel's HFS filesystem driver stems from a use-after-free in hfs_correct_next_unused_CNID(), where hfs_bnode_put() drops a b-tree node's refcount and the freed node is then dereferenced on the following line. The flaw affects systems that mount or process HFS (legacy Apple) volumes and can lead to a kernel crash or potential memory corruption. There is no public exploit identified at time of analysis, EPSS is negligible (0.03%), and it is not in CISA KEV; the tags characterize the practical impact as denial of service.
Out-of-bounds write in the Linux kernel's ALSA firewire-motu driver (snd-firewire-motu) allows a local process reading DSP events to trigger a kernel put_user() copy past the end of its supplied buffer when that buffer's size is not a multiple of 4 bytes. The flaw affects systems with a MOTU FireWire audio interface driver loaded and requires local low-privileged access; it is fixed in stable kernel releases and shipped in Ubuntu USN-8094-1. EPSS is very low (0.05%, 16th percentile), there is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Improper IV handling in libtpms 0.10.0 and 0.10.1 causes the library to return initial instead of final initialization vectors during symmetric cipher operations with OpenSSL 3.x, potentially weakening cryptographic security for local users who can interact with the TPM emulation. Public exploit code exists for this vulnerability affecting confidentiality of encrypted data. Update to libtpms 0.10.2 to remediate.
In Gitea before 1.25.2, /api/v1/user has different responses for failed authentication depending on whether a username exists. [CVSS 5.3 MEDIUM]
Availability impact in the Linux kernel's Intel VPU accelerator driver (accel/ivpu) stems from a race condition between Buffer Object (BO) freeing and context teardown. A local, low-privileged user on a system with Intel VPU/NPU hardware can trigger the ivpu_gem_bo_free() path to remove a BO from the list before it is fully unmapped, causing file_priv_unbind() to encounter an unclean memory manager state and trigger a drm_mm_takedown() warning, ultimately resulting in a kernel availability impact. No public exploit exists and EPSS sits at the 1st percentile, indicating negligible real-world exploitation interest at this time.
Local privilege escalation and memory corruption in the Linux kernel's drm/panthor GPU driver arises from a use-after-free race between device unplug (panthor_fw_unplug()) freeing firmware memory sections and the asynchronous process_fw_events_work() handler, which can still access those freed FW memory regions. An attacker with local access to a system using an ARM Mali CSF-based GPU can potentially corrupt kernel memory or leak information by racing device teardown against pending firmware event processing. CVSS is 7.8 (High) with full confidentiality/integrity/availability impact; no public exploit has been identified and it is not in CISA KEV.
Local privilege escalation and memory corruption in the Linux kernel's Panthor DRM driver (ARM Mali GPU) arises from a use-after-free on kernel buffer-object virtual-address nodes; when the GPU MMU is down, panthor_vm_unmap_range() returns early and leaves a va_node stuck in the drm_mm tree, so removing adjacent nodes later dereferences freed memory. A local user with GPU/render access can trigger the condition to corrupt kernel memory (CVSS 7.8, high confidentiality/integrity/availability impact). No public exploit identified at time of analysis; an upstream fix has been merged to stable trees.
Use-after-free in the Linux kernel's Tegra210 QSPI (tegra210-quad) SPI controller driver arises when a transfer times out before the threaded IRQ handler runs, leaving the curr_xfer field pointing at freed memory that the IRQ thread may later dereference. Local low-privileged users on NVIDIA Tegra-based systems could trigger memory corruption affecting confidentiality, integrity, and availability, though exploitation depends on winning a narrow timing race. There is no public exploit identified at time of analysis and it is not listed in CISA KEV; the fix is delivered as upstream stable-tree commits.
Kernel instability in the Linux qla2xxx SCSI driver (QLogic Fibre Channel target/host adapter) allows a chip-reset event to leave in-flight commands stuck and never freed, and can trigger a BUG_ON kernel panic (BUG_ON(cmd->sg_mapped) or BUG_ON(!valid_dma_direction(dir))) when two CPUs race to unmap the same scatter-gather list. This fix reverts commit 26f9ce53817a and partially reverts aefed3e5548f at __qla2x00_abort_all_cmds to restore correct command cleanup after a firmware/chip reset. No public exploit identified at time of analysis; the NVD-assigned CVSS of 9.8 (AV:N) is inconsistent with the local, hardware-dependent nature of the bug.
Memory corruption in the Linux kernel's qla2xxx QLogic Fibre Channel HBA driver arises from an error-handling path in qla2xxx_process_purls_iocb() that calls kfree() on a purex item which may have originated from a per-adapter pre-allocated pool rather than kzalloc(). Freeing a pool-backed object with kfree() can corrupt kernel heap state. The fix routes deallocation through qla24xx_free_purex_item(), which correctly distinguishes dynamically allocated from pooled items; no public exploit identified at time of analysis, and the issue is not in CISA KEV.
Access-control inconsistency in the Linux kernel's Landlock LSM allows a sandboxed local task to widen its effective filesystem access rights by operating on "disconnected" directories - files opened through a bind mount and then renamed or moved out of scope of the mount point. Because the mount point could not be located during the hierarchy walk, Landlock collected access rights straight down to the filesystem root, ignoring the more restrictive mount-point hierarchy and permitting renames (notably under LANDLOCK_ACCESS_FS_REFER) that should have been denied. Only Linux systems actively enforcing Landlock sandboxing are affected; there is no public exploit identified at time of analysis and EPSS is negligible (0.02%, 6th percentile).
Local privilege escalation via a use-after-free in the Linux kernel's drm/panthor driver (the DRM driver for newer Arm Mali CSF-based GPUs) allows a local user with GPU render-node access to free and reuse a group object during creation. The GROUP_CREATE ioctl stored a group pointer in an Xarray and then continued to dereference it, letting a second thread call GROUP_DESTROY on a guessed handle to trigger the UAF. No public exploit or CISA KEV listing has been identified; the fix adds an Xarray entry mark so a group cannot be destroyed until GROUP_CREATE has fully returned.
Kernel panic in the Linux octeontx2-af driver exposes Marvell CN103XX OcteonTX2 systems to a denial-of-service crash when firmware reports a physical link change event with an invalid lmac_type_id value. The driver's cgx_link_change_handler() uses the unvalidated firmware-supplied ID to index into an array and calls strncpy with the resulting out-of-bounds pointer, producing a fatal kernel Oops in a workqueue context. No public exploit has been identified and EPSS sits at 0.17% (6th percentile), but the crash is deterministic once the invalid firmware value is processed, making Availability impact high on affected hardware.
Use-after-free in the Linux kernel NTFS3 filesystem driver allows a local attacker with low privileges to corrupt kernel heap memory via a crafted NTFS volume with malformed extended attributes. The `ntfs_read_ea` function fails to propagate error codes on encountering inconsistent EA structures, causing `ntfs_set_ea` to perform a 4-byte write to already-freed memory - confirmed by an in-description KASAN trace captured at 6.2.0-rc1+. No active exploitation is confirmed (not in CISA KEV), and an EPSS of 0.16% (6th percentile) signals very low observed exploitation activity, though a researcher-developed proof-of-concept is implied by the kernel trace.
Local privilege-tier memory corruption in the Linux kernel's Intel VPU/NPU accelerator driver (accel/ivpu) arises from a dangling buffer object left on a per-device list after a failed allocation, later dereferenced in ivpu_bo_unbind_all_bos_from_context() to cause a page fault. A local user with access to the ivpu device can trigger this use-after-free via the DRM/accel interface, with CVSS 7.8 (High) and CISA-style impact of confidentiality, integrity, and availability all High. There is no public exploit identified at time of analysis and the flaw is not in CISA KEV; the primary observed effect is a kernel page fault (crash), though the dangling-pointer condition carries corruption potential.
Kernel-side memory corruption in the Linux kernel's crypto AEAD framework arises because AEAD algorithms began populating the generic cra_reqsize field (added by commit afddce13ce81d) without the framework being updated to translate it into a proper per-request size via crypto_aead_set_reqsize(). As a result, request buffers for affected AEAD transforms are undersized, producing out-of-bounds memory writes and kernel crashes when the algorithm is used. This is tracked as a Denial-of-Service class issue; no public exploit identified at time of analysis, and it is not listed in CISA KEV.
BPF test infrastructure in the Linux kernel allows a local user with BPF program execution privileges to emit packets with malformed GSO (Generic Segmentation Offload) metadata to the network stack via bpf_clone_redirect(), triggering skb_warn_bad_offload() and disabling GSO-related offload features - a denial-of-service condition. The root cause is convert___skb_to_skb() populating gso_segs and gso_size without a corresponding gso_type, producing an internally inconsistent socket buffer that fails GSO validation in netif_skb_features(). No active exploitation has been confirmed (not in CISA KEV) and no public exploit code has been identified; the issue was originally surfaced through fuzzing by Yinhao et al. Patches across six stable kernel branches have been released.
Memory corruption in the Linux kernel's asymmetric_keys crypto subsystem allows a local low-privileged user to trigger a heap buffer overflow by supplying a crafted X.509 certificate. The flaw lives in asymmetric_key_generate_id(), where the sizes of attacker-controlled binary blobs (such as arbitrarily large ASN.1 INTEGER serial numbers and issuer names) are added to the size of an asymmetric_key_id structure without overflow checking, so an integer overflow yields an undersized allocation and a subsequent out-of-bounds copy. No public exploit identified at time of analysis, and EPSS is low (0.15%, 5th percentile).
Local denial of service in the Linux kernel's software RDMA driver (RDMA/rxe, rdma_rxe module) allows a user with access to an rxe device to crash the host via a NULL pointer dereference. When ibv_modify_srq() is called to resize a shared receive queue and rxe_queue_resize() fails, rxe_srq_from_attr() leaves srq->rq.queue set to NULL; a subsequent modify call dereferences srq->rq.queue->buf->index_mask in rxe_srq_chk_attr() and panics. No public exploit identified at time of analysis, and the EPSS score is low (0.16%, 6th percentile).
Local privilege-limited denial of service in the Linux kernel's CoreSight TMC (Trace Memory Controller) driver allows a user with performance-monitoring access to trigger a kernel oops via an invalid pointer dereference in tmc_etr_get_buffer(). The flaw occurs because the perf event handle was not attached to the coresight_path, so dependent devices such as the CATU could not retrieve the per-CPU AUX_EVENT during path enablement. It is reproducible with a standard perf record command against CoreSight ETM tracing; there is no public exploit identified at time of analysis and EPSS is low (0.16%).
Denial of service in the Linux kernel's ntfs3 filesystem driver stems from a run_lock rwsem that is never initialized when the special $Extend metadata inode is loaded. When $Extend's inode mode is set to a regular file, a truncate() call reaches ntfs_set_size() and acquires the uninitialized lock, tripping lockdep and risking kernel instability. The regression was introduced by commit 4e8011ffec79 and was found via syzbot fuzzing; no public exploit identified at time of analysis, and EPSS is low at 0.17%.
Uninitialized memory use in the Linux kernel NTFS3 filesystem driver (fs/ntfs3) enables a local low-privileged user to trigger a kernel crash, causing a denial of service. The `__getname()` function allocates slab memory via `kmem_cache_alloc()` without zeroing it; `ntfs_read_hdr` and `bcmp` then operate on this indeterminate data, as detected by KMSAN. No active exploitation is confirmed (not in CISA KEV), and the EPSS score of 0.18% at the 7th percentile reflects very low real-world exploitation probability. Upstream patches are available across multiple stable kernel branches.
Denial of service in the Linux kernel's mt76 wireless driver (MT7996 WiFi 7 support) allows the kernel to crash (NULL/invalid pointer dereference) when hardware-accelerated flow offloading runs on a secondary Wireless Ethernet Dispatch instance (wed_hif2, e.g. a 6GHz link). The mt76 WED callbacks unconditionally referenced the primary wed device, so tearing down or disabling offload on wed_hif2 dereferences an invalid address and triggers a kernel Oops. There is no public exploit identified at time of analysis, EPSS risk is low (0.18%, 8th percentile), and it is not in CISA KEV; impact is availability loss (system crash), not code execution.
Memory corruption (double free) in the Linux kernel btrfs filesystem's quota group (qgroup) accounting affects the add_delayed_ref_head() code path when a qgroup record is reported as already existing but a subsequent error occurs, causing both caller and callee to free the same 'qrecord' object. The flaw is a local btrfs-triggered condition on systems using btrfs with quota groups; there is no public exploit identified at time of analysis and it is not listed in CISA KEV, with a very low EPSS of 0.17%. Despite an automated CVSS of 9.8, the realistic attack surface is local to the machine and gated by a hard-to-reach error path.
Out-of-bounds memory access in the Linux kernel's ch341 SPI driver (drivers/spi/spi-ch341) allows a local low-privileged actor to trigger heap buffer overreads and overwrites during SPI transfers on systems using a WCH CH341 USB-to-SPI bridge. The flaw stems from ch341_transfer_one copying 'len' bytes (which includes a 1-byte command header) instead of 'len - 1' from the user-supplied transfer buffer, causing both an out-of-bounds read from trans->tx_buf and, when len equals the 32-byte packet length, a 1-byte overflow of ch341->tx_buf. EPSS is low at 0.18% and there is no public exploit identified at time of analysis, consistent with a niche hardware-dependent kernel driver rather than a widely exploited target.
Denial of service in the Linux kernel's NFSv4/pNFS client allows a null-pointer dereference and system crash during inode writeback, affecting hosts mounting parallel-NFS (pNFS) exports. The flaw occurs because NFS_INO_LAYOUTCOMMIT was not cleared in pnfs_mark_layout_stateid_invalid, so the write_inode → nfs4_write_inode → pnfs_layoutcommit_inode path may dereference a layout that has already been freed. There is no public exploit identified at time of analysis, and the low EPSS score (0.18%) indicates minimal observed exploitation interest despite the CVSS 7.5 rating.
In the Linux kernel, the following vulnerability has been resolved: iommu/mediatek: fix use-after-free on probe deferral The driver is dropping the references taken to the larb devices during probe after successful lookup as well as on errors.
Local privilege-adjacent memory corruption in the Linux kernel ublk (userspace block driver) subsystem allows a ublk server process to trigger use-after-free-style access on completed I/O requests by exiting while requests are still in flight. The regression, introduced by commit e63d2228ef83 which stopped decrementing the request reference count in __ublk_fail_req(), leaves a leaked UBLK_REFCOUNT_INIT reference that permits illegitimate user-copy and zero-copy operations on already-completed requests and trips WARN_ON_ONCE reference-count assertions during queue teardown. It carries a CVSS 7.8 (high) with local vector and low privileges, EPSS is negligible at 0.03% (7th percentile), and there is no public exploit identified at time of analysis.
Local privilege escalation in the Linux kernel's ETS (net/sched) packet scheduler arises from a use-after-free on a struct Qdisc object, caused by a race between ets_qdisc_dequeue and ets_qdisc_change. When bands are reconfigured, a class can retain its entry on the active list after its child qdisc is freed and nulled, letting a concurrent dequeue operate on freed memory. Trend Micro ZDI reports a working reproducer and that an attacker able to create user and network namespaces can trigger the flaw and, per their analysis, control RIP; there is no CISA KEV listing and EPSS is very low (0.05%), so no public evidence of active exploitation.
Local privilege-escalation-capable memory corruption in the Linux kernel's ALPS PS/2 touchpad driver (drivers/input/mouse/alps.c) lets a use-after-free be triggered when an ALPS touchpad is detached while the dev3_register_work delayed work item is still pending. Because psmouse_disconnect() only flushes work already queued, alps_register_bare_ps2_mouse() can run and dereference the alps_data structure (priv->dev3) after alps_disconnect() has kfree()'d it, corrupting freed kernel heap memory. There is no public exploit identified at time of analysis, EPSS is negligible (0.02%), and the flaw was found by static analysis rather than in-the-wild activity.
Out-of-bounds memory access in the Linux kernel's dtv5100 DVB-USB media driver allows a local, low-privileged attacker to corrupt kernel memory via an unchecked, user-controlled rlen value in dtv5100_i2c_msg(). Because rlen is never validated against sizeof(st->data), a value larger than the buffer drives reads/writes past the allocated region, enabling denial of service and potential local privilege escalation. There is no public exploit identified at time of analysis, EPSS is very low (0.05%), and it is not listed in CISA KEV, but a vendor fix is available.
Kernel NULL pointer dereference in the Linux kernel's qla2xxx QLogic Fibre Channel driver crashes systems running in SCSI target mode during the command-abort path. The flawed code in __qla2x00_abort_all_cmds() assumed every command was TYPE_SRB and called sp->done() on TYPE_TGT_CMD entries, jumping to a NULL function pointer during ISP error recovery. The fix reverts the offending commit (0367076b0817) since the switch-case handling below it already covers the chip-down case correctly. EPSS is low (0.05%, 16th percentile) and there is no public exploit identified at time of analysis.
Local privilege escalation and kernel memory corruption in the Linux kernel's net/sched ETS (Enhanced Transmission Selection) qdisc allows a user with network-configuration capability to corrupt an internal linked list by toggling a scheduling class between DRR and strict modes. When an active DRR class is reconfigured to strict, ets_qdisc_change() fails to remove it from the active list, so switching it back to DRR performs a double list_add, corrupting kernel list pointers. No public exploit has been identified at time of analysis and the EPSS score is very low (0.05%), but the flaw is a classic kernel list-corruption primitive that can escalate to arbitrary write and full compromise.
Kernel memory corruption in the Linux svcrdma transport (server-side NFS-over-RDMA) arises because svc_rdma_copy_inline_range() added the page index (rc_curpage) instead of the byte offset (rc_pageoff) when computing the memcpy destination, so inline payload copies land at the wrong location within the receive pages. On systems exporting NFS over RDMA, a client on the RDMA fabric can drive copies that miscalculate their target offset, leading to out-of-bounds reads/writes in kernel page memory and potential information disclosure or corruption. There is no public exploit identified at time of analysis, EPSS is negligible (0.02%), and the issue is not in CISA KEV, but distro fixes (Ubuntu USN-8177-1) and upstream stable commits are available.
Local privilege escalation and memory corruption in the Linux kernel's KVM subsystem allows a user with VM control (e.g. a QEMU/hypervisor process) to trigger a use-after-free by clearing the KVM_MEM_GUEST_MEMFD flag on a memslot that was created with a guest_memfd binding. KVM blocks enabling the flag on an existing memslot but never blocked disabling it, leaving the guest_memfd instance bound to freed memory that is later written during kvm_gmem_release. The flaw carries a CVSS of 7.8 and has a vendor-released fix; it is not in CISA KEV and no public exploit has been identified at time of analysis, with an EPSS of 0.02%.
Concurrent SMB delete-on-close handling in the Linux kernel's in-kernel SMB3 server (ksmbd) mishandles locking on the ksmbd_inode->m_flags field, letting racing open/close/delete operations corrupt delete-on-close and pending-delete state. Affected builds ship the pre-fix ksmbd vfs_cache code where helpers such as ksmbd_inode_pending_delete() and ksmbd_fd_set_delete_on_close() read and modify m_flags without holding ci->m_lock. The result is inconsistent delete semantics - files that persist after a delete-on-close or vanish while still in use - with no public exploit identified at time of analysis and a very low EPSS score (0.03%).
Local privilege escalation risk in the Linux kernel's vidtv (Virtual Digital TV) media test driver stems from a double-free and use-after-free condition in vidtv_channel_si_init(), where local pointers are not reset to NULL after ownership of program/service/event list memory is transferred to the PAT/SDT/EIT tables. When vidtv_psi_pmt_create_sec_for_each_pat_entry() fails and control jumps to the free_eit error path, already-freed memory is destroyed a second time. Publicly this is patched upstream and shipped via Ubuntu USN-8096-1, but with EPSS at 0.05% and no public exploit identified at time of analysis, real-world exploitation is unlikely and confined to systems that actually load this virtual test driver.
Out-of-bounds buffer handling in the Linux kernel's ksmbd (in-kernel SMB3 server) Extended Attribute code allows a remote attacker to disclose adjacent kernel memory by sending a crafted SMB2 SET_INFO request. The smb2_set_ea() function validated the request buffer without accounting for the null-terminator byte of the EA name, so the size check under-counted the required length and permitted reads past the intended boundary. No public exploit is identified at time of analysis and EPSS probability is low (0.04%), but a vendor/upstream fix is available and it is not listed in CISA KEV.
Local privilege escalation and memory corruption in the Linux kernel FUSE subsystem's io-uring path allows an unprivileged local user to trigger list corruption and a resulting use-after-free. The flaw stems from a terminated-but-not-committed FUSE io-uring request never being unlinked from the queue's list, leaving a dangling entry that corrupts kernel memory. EPSS is very low (0.03%, 7th percentile) and there is no public exploit identified at time of analysis; a fix has been merged upstream and shipped by distributions such as Ubuntu.
Incorrect access-control enforcement in the Linux kernel's NFSv4 server (nfsd) causes POSIX ACLs supplied during file creation to be silently discarded, so a client that sets an ACL with a named principal ends up with only a default mode-derived ACL instead. This violates RFC 8881 6.4.1.3 and means files intended to carry restrictive or explicitly-granted permissions are instead governed by permissive default mode bits, potentially exposing their contents to unintended principals on the export. There is no public exploit identified at time of analysis, EPSS is very low (0.05%), and it is not on CISA KEV - this is a security-relevant correctness defect rather than an actively exploited threat.
Local privilege-holder-triggered use-after-free in the Linux kernel's mlxsw Spectrum switch driver (drivers/net/ethernet/mellanox/mlxsw) lets a network-admin operation on Nvidia/Mellanox Spectrum ASIC hardware dereference a freed neighbour object during RIF teardown. The driver cached a neighbour pointer without holding a reference, so freeing the neighbour (e.g. while removing a device from a VRF) leaves a dangling pointer that mlxsw_sp_neigh_entry_update() reads, corrupting kernel memory. CVSS is 7.8 (local, low privilege in the automated NVD scoring) but EPSS is only 0.05% (16th percentile) and there is no public exploit or CISA KEV listing - the flaw was found via an internal stress test, not observed attacks.
Local privilege escalation potential arising from a use-after-free in the Linux kernel's mlxsw Spectrum driver (spectrum_mr) allows an attacker with network-administration rights on a Mellanox Spectrum switch to trigger memory corruption during multicast route replacement. A prior fix introduced a dedicated mutex to protect the multicast route list, but one list-entry deletion path during route replace was missed, so the periodic mlxsw_sp_mr_stats_update worker can read a freed route object. EPSS is low (0.05%) and there is no public exploit identified at time of analysis; a vendor patch is available across multiple stable branches.
Denial of service and potential kernel memory disclosure in the Linux kernel's CAIF protocol stack arises because cffrml_receive() subtracts 2 from a packet-derived length field without checking that the value is at least 2 when FCS checking is disabled. An adjacent attacker sending a CAIF frame whose length field is 0 or 1 triggers an integer underflow, leading to memory exhaustion, kernel instability, and possible leakage of uninitialized kernel memory contained in packet padding. There is no public exploit identified at time of analysis, EPSS risk is very low (0.05%), and the flaw is not in CISA KEV.
Local privilege-level buffer overflow in the Linux kernel's ethtool subsystem allows an attacker who can trigger a device stat-count change between the ETHTOOL_GSSET_INFO, ETHTOOL_GSTRINGS, and ETHTOOL_GSTATS ioctls to overflow a userspace stats buffer, corrupting memory and potentially escalating privileges. The flaw stems from a TOCTOU race where drivers using dynamic counters (mlx5, bnx2x, bna, ksz884x) can report a different stat count than the buffer sized by userspace. It is not in CISA KEV, EPSS exploitation probability is very low (0.06%), and no public exploit has been identified, though a vendor patch is available.
Local privilege escalation or kernel memory disclosure in the Linux kernel's amdgpu DRM driver stems from a use-after-free during GPU recovery, where a race between the scheduler timeout callback (drm_sched_job_timedout) and the TDR work queue can free a job structure before amdgpu_device_gpu_recover reads its pasid field. Affected systems are those running vulnerable kernels with AMD GPUs using the amdgpu driver; a local attacker able to submit GPU work and induce a GPU hang could trigger the freed-memory access, as demonstrated by the supplied KASAN slab-use-after-free trace. There is no public exploit identified at time of analysis, and the EPSS score is negligible (0.03%, 7th percentile), consistent with a difficult-to-win kernel race rather than a broadly exploited flaw.
Memory corruption in the Linux kernel's TPM 2.0 session-handling code (tpm2-sessions) arises because the 'name_size' value derived from a TPM_ALG_ID is used to index without range validation, so an unrecognized or malicious algorithm identifier can drive an out-of-bounds access. The affected TPM HMAC/authorization session path is reachable from local, low-privileged kernel-driver interactions; at worst this leads to kernel memory corruption affecting confidentiality, integrity, and availability. No public exploit is identified at time of analysis and the EPSS probability is negligible (0.03%), consistent with a locally-scoped kernel hardening fix rather than a mass-exploited flaw.
Out-of-bounds memory access in the Linux kernel's Open vSwitch (openvswitch) module allows a local user holding CAP_NET_ADMIN to leak kernel memory or crash the host by submitting a malformed OVS_ACTION_ATTR_PUSH_NSH netlink action with an incorrectly sized middle attribute. The push_nsh() action path validates the outer and inner NSH attributes but never checks the intermediate OVS_KEY_ATTR_NSH wrapper, so nsh_hdr_from_nlattr() performs a 184-byte read past a slab allocation (KASAN slab-out-of-bounds). A patch is available from the kernel stable tree and via Ubuntu USN-8096-1; there is no public exploit identified at time of analysis, and EPSS is very low (0.07%).
Remote denial of service in the Linux kernel SCSI target (LIO) subsystem allows an initiator to crash a storage server by triggering a NULL-pointer dereference on the command error path. When allocation of the per-command CDB buffer (cmd->t_task_cdb) fails, the pointer is left NULL yet subsequently dereferenced in the 'err' cleanup path, panicking the kernel. No public exploit identified at time of analysis and EPSS is very low (0.05%), consistent with a hard-to-trigger memory-pressure condition rather than a reliably weaponizable flaw.
Refcount underflow in the Linux kernel's net/handshake subsystem lets duplicate cancellation of the same TLS handshake request drop an extra reference on the underlying socket, corrupting its reference count and leaking/freeing the socket structure. The flaw affects kernels using the in-kernel TLS handshake upcall path (notably SUNRPC-over-TLS via tlshd), where a handshake timeout can trigger two cancellation requests - one from xs_reset_transport() and one from xs_tls_handshake_sync() - for a request still present in handshake_rhashtbl. Despite an assigned 9.8 CVSS, the practical impact is a kernel memory-safety fault (crash / potential use-after-free) rather than trivial remote code execution, and there is no public exploit identified at time of analysis; EPSS is very low (0.03%).
Denial of service in the Linux kernel's HFS+ (hfsplus) filesystem driver allows a local user with access to a mounted HFS+ volume to crash the kernel via a reference-counting race in __hfs_bnode_create(). When sync() and link() operations run concurrently, two threads can both attempt to create the same B-tree node; the thread that loses the race reuses the already-hashed node without incrementing its refcount, so a later put trips BUG_ON(!atomic_read(&node->refcnt)) and panics the system. No public exploit has been identified at time of analysis, EPSS risk is very low (0.05%), and the issue is not in CISA KEV.
Denial of service in the Linux kernel's bnxt_en driver (Broadcom NetXtreme-E) affects systems running an XDP program that uses the XDP_TX action. Incorrect clearing of NAPI event flags in the XDP_TX path causes the TX doorbell to be skipped when the TX ring is full, stalling the associated RX ring so it is never refilled and all subsequent packets are silently dropped. There is no public exploit identified at time of analysis, and EPSS exploitation probability is very low (0.02%, 6th percentile); an upstream fix is available in mainline and stable kernels.
Libpng versions 1.6.26 through 1.6.53 contain an integer truncation flaw in the simplified write API functions that triggers a heap buffer over-read when processing images with negative row strides or strides exceeding 65535 bytes. Local attackers can exploit this to read sensitive heap memory, potentially disclosing application data. No patch is currently available; users should avoid processing untrusted PNG images with these vulnerable libpng versions.
Libpng versions 1.6.51-1.6.53 contain a heap buffer over-read in the simplified API function png_image_finish_read when processing interlaced 16-bit PNG images with 8-bit output and non-minimal row stride, allowing local attackers to read out-of-bounds memory through a malicious image file. Public exploit code exists for this regression, which was introduced by a previous security fix. Upgrade to version 1.6.54 to remediate.
Fulcio versions prior to 1.8.5 allow unauthenticated attackers to bypass MetaIssuer URL validation through unanchored regex patterns, enabling blind SSRF attacks against internal services. Although the vulnerability is limited to read-only GET requests with no response exfiltration, attackers can probe internal networks to discover active services and infrastructure. Public exploit code exists for this medium-severity issue, and a patch is available in version 1.8.5.
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, avahi-daemon can be crashed by sending 2 unsolicited announcements with CNAME resource records 2 seconds apart. [CVSS 6.5 MEDIUM]
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, avahi-daemon can be crashed by sending unsolicited announcements containing CNAME resource records pointing it to resource records with short TTLs. [CVSS 6.5 MEDIUM]
Avahi is a system which facilitates service discovery on a local network via the mDNS/DNS-SD protocol suite. In 0.9-rc2 and earlier, an unprivileged local users can crash avahi-daemon (with wide-area disabled) by creating record browsers with the AVAHI_LOOKUP_USE_WIDE_AREA flag set via D-Bus. [CVSS 5.5 MEDIUM]
Vllm versions up to 0.12.0 is affected by allocation of resources without limits or throttling (CVSS 6.5).
Cosign's bundle verification mechanism fails to properly validate that embedded Rekor entries reference the correct artifact digest, signature, and public key, allowing an attacker with a compromised signing identity to forge valid bundles and bypass transparency log verification. A malicious actor could exploit this to create counterfeit signatures that pass validation checks, affecting users relying on Cosign for container and binary code signing verification. Public exploit code exists for this vulnerability; patches are available in versions 2.6.2 and 3.0.4.
Virtualenv versions up to 20.36.1 is affected by improper link resolution before file access (CVSS 4.5).
Python's filelock SoftFileLock implementation prior to version 3.20.3 contains a TOCTOU race condition that allows local attackers with symlink creation privileges to interfere with lock file operations between permission validation and file creation. An attacker can exploit this window to create a symlink at the target lock path, causing lock operations to fail or redirect to unintended files, resulting in denial of service or unexpected behavior. Upgrade to filelock version 3.20.3 or later to remediate this vulnerability.
HarfBuzz text shaping engine versions prior to 12.3.0 crash when the SubtableUnicodesCache::create function attempts to dereference a null pointer returned by failed memory allocation, enabling denial of service in applications processing untrusted font data. Public exploit code exists for this vulnerability. A patch is available in version 12.3.0 and later.
pypdf versions prior to 6.6.0 are vulnerable to denial of service through CPU exhaustion when processing malformed PDF files with crafted startxref entries in non-strict reading mode. An attacker can create a specially crafted PDF containing excessive whitespace that causes the library to consume significant processing resources during cross-reference table reconstruction. A patch is available in version 6.6.0 and later.
Denial of service via resource exhaustion in pypdf prior to version 6.6.0 allows remote attackers to trigger excessive processing times by submitting specially crafted PDF files with missing /Root objects and inflated /Size values. The vulnerability only affects non-strict parsing mode and causes the library to consume significant CPU resources when processing otherwise invalid documents. A patch is available in version 6.6.0 and later.
React Router is a router for React. In @remix-run/server-runtime version prior to 2.17.3. [CVSS 6.5 MEDIUM]
React-Router versions up to 6.30.1 is affected by url redirection to untrusted site (open redirect) (CVSS 6.5).
A memory initialization issue was addressed with improved memory handling. This issue is fixed in tvOS 26.2, Safari 26.2, watchOS 26.2, visionOS 26.2, iOS 26.2 and iPadOS 26.2, macOS Tahoe 26.2. [CVSS 4.3 MEDIUM]
A flaw was found in GNU Wget2. This vulnerability, a stack-based buffer overflow, occurs in the filename sanitization logic when processing attacker-controlled URL paths, particularly when filename restriction options are active. [CVSS 7.6 HIGH]
A security issue was discovered in GNU Wget2 when handling Metalink documents. The application fails to properly validate file paths provided in Metalink <file name> elements. [CVSS 8.8 HIGH]
The ECDSA implementation of the Elliptic package generates incorrect signatures if an interim value of 'k' (as computed based on step 3.2 of RFC 6979 https://datatracker.ietf.org/doc/html/rfc6979 ) has leading zeros and is susceptible to cryptanalysis, which can lead to secret key exposure. [CVSS 5.6 MEDIUM]
Authlib is a Python library which builds OAuth and OpenID Connect servers. [CVSS 5.7 MEDIUM]
CoreDNS is a DNS server that chains plugins. Prior to version 1.14.0, multiple CoreDNS server implementations (gRPC, HTTPS, and HTTP/3) lack critical resource-limiting controls. An unauthenticated remote attacker can exhaust memory and degrade or crash the server by opening many concurrent connections, streams, or sending oversized request bodies. The issue is similar in nature to CVE-2025-47950 (QUIC DoS) but affects additional server types that do not enforce connection limits, stream limit...
HTML injection in Preact and React through unsafe JSON deserialization allows remote attackers to inject arbitrary scripts when applications pass unsanitized user-controlled data as component children, with public exploit code already available. The vulnerability stems from a regression in Preact 10.26.5 that weakened protections against constructing Virtual DOM elements from malicious JSON payloads. Affected applications are vulnerable if they consume unvalidated external data without sanitization and lack additional mitigations like Content Security Policy.
The RSA crate versions prior to 0.9.10 crash when constructing private keys with invalid prime components (such as 1), allowing an attacker to trigger a denial of service by providing malformed key material. This affects applications using the vulnerable RSA library for cryptographic operations. A patch is available in version 0.9.10 and later.
A flaw was found in Ansible Automation Platform (AAP). Read-only scoped OAuth2 API Tokens in AAP, are enforced at the Gateway level for Gateway-specific operations. [CVSS 8.5 HIGH]
Bokeh versions 3.8.1 and below allow attackers to bypass Origin validation in WebSocket connections by registering domains that suffix-match allowlisted domains (e.g., dashboard.corp.attacker.com for allowlist entry dashboard.corp), enabling unauthorized server interaction. Public exploit code exists for this vulnerability, which could allow attackers to access sensitive data or modify visualizations on behalf of victims. The issue is resolved in Bokeh 3.8.2.
Stack-based buffer overflow in libtasn1 version: v4.20.0. The function fails to validate the size of input data resulting in a buffer overflow in asn1_expend_octet_string. [CVSS 7.5 HIGH]
Visualdata versions up to - is affected by inefficient regular expression complexity (redos) (CVSS 5.3).
Quarkus is a Cloud Native, (Linux) Container First framework for writing Java applications. Prior to versions 3.31.0, 3.27.2, and 3.20.5, a vulnerability exists in the HTTP layer of Quarkus REST related to response handling. When a response is being written, the framework waits for previously written response chunks to be fully transmitted before proceeding. If the client connection is dropped during this waiting period, the associated worker thread is never released and becomes permanently b...
Undertow HTTP server (used in WildFly, JBoss EAP) fails to validate Host headers, enabling cache poisoning, internal network scanning, and session hijacking. Affects a widely-used Java application server component.
Crypt::Sodium::XS for Perl bundles a vulnerable version of libsodium (<= 1.0.20) that has a signature verification flaw. In atypical use cases with custom cryptography, this can compromise data authenticity guarantees. Patch available.
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. In versions 3.13.2 and below, handling of chunked messages can result in excessive blocking CPU usage when receiving a large number of chunks. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below allow a request to be crafted in such a way that an AIOHTTP server's memory fills up uncontrollably during processing. [CVSS 7.5 HIGH]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below allow for an infinite loop to occur when assert statements are bypassed, resulting in a DoS attack when processing a POST body. [CVSS 7.5 HIGH]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below contain parser logic which allows non-ASCII decimals to be present in the Range header. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below enable an attacker to ascertain the existence of absolute path components through the path normalization logic for static files meant to prevent path traversal. [CVSS 5.3 MEDIUM]
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Versions 3.13.2 and below of the Python HTTP parser may allow a request smuggling attack with the presence of non-ASCII characters. [CVSS 6.5 MEDIUM]
vega-functions provides function implementations for the Vega expression language. Prior to version 6.1.1, for sites that allow users to supply untrusted user input, malicious use of an internal function (not part of the public API) could be used to run unintentional javascript (XSS). [CVSS 7.2 HIGH]
Vega is a visualization grammar, a declarative format for creating, saving, and sharing interactive visualization designs. Prior to versions 6.1.2 and 5.6.3, applications meeting two conditions are at risk of arbitrary JavaScript code execution, even if "safe mode" expressionInterpreter is used. [CVSS 8.1 HIGH]
Security-flag inheritance bypass in the Linux kernel NFS client allows a low-privileged local user to obtain access on automounted NFS submounts that ignores the restrictive superblock options (ro, noexec, nodev, sync) set on the parent mount. Because automounted filesystems fail to preserve these user-set mount flags, protections such as read-only or no-execute enforcement can be silently dropped on the child mount, undermining the security posture an administrator intended. No public exploit identified at time of analysis, and the EPSS score is negligible (0.05%, 16th percentile), consistent with a local, configuration-dependent defense-in-depth weakness rather than a mass-exploited flaw.
Memory corruption in the Linux kernel's HFS filesystem driver stems from a use-after-free in hfs_correct_next_unused_CNID(), where hfs_bnode_put() drops a b-tree node's refcount and the freed node is then dereferenced on the following line. The flaw affects systems that mount or process HFS (legacy Apple) volumes and can lead to a kernel crash or potential memory corruption. There is no public exploit identified at time of analysis, EPSS is negligible (0.03%), and it is not in CISA KEV; the tags characterize the practical impact as denial of service.
Out-of-bounds write in the Linux kernel's ALSA firewire-motu driver (snd-firewire-motu) allows a local process reading DSP events to trigger a kernel put_user() copy past the end of its supplied buffer when that buffer's size is not a multiple of 4 bytes. The flaw affects systems with a MOTU FireWire audio interface driver loaded and requires local low-privileged access; it is fixed in stable kernel releases and shipped in Ubuntu USN-8094-1. EPSS is very low (0.05%, 16th percentile), there is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Improper IV handling in libtpms 0.10.0 and 0.10.1 causes the library to return initial instead of final initialization vectors during symmetric cipher operations with OpenSSL 3.x, potentially weakening cryptographic security for local users who can interact with the TPM emulation. Public exploit code exists for this vulnerability affecting confidentiality of encrypted data. Update to libtpms 0.10.2 to remediate.
In Gitea before 1.25.2, /api/v1/user has different responses for failed authentication depending on whether a username exists. [CVSS 5.3 MEDIUM]
Availability impact in the Linux kernel's Intel VPU accelerator driver (accel/ivpu) stems from a race condition between Buffer Object (BO) freeing and context teardown. A local, low-privileged user on a system with Intel VPU/NPU hardware can trigger the ivpu_gem_bo_free() path to remove a BO from the list before it is fully unmapped, causing file_priv_unbind() to encounter an unclean memory manager state and trigger a drm_mm_takedown() warning, ultimately resulting in a kernel availability impact. No public exploit exists and EPSS sits at the 1st percentile, indicating negligible real-world exploitation interest at this time.
Local privilege escalation and memory corruption in the Linux kernel's drm/panthor GPU driver arises from a use-after-free race between device unplug (panthor_fw_unplug()) freeing firmware memory sections and the asynchronous process_fw_events_work() handler, which can still access those freed FW memory regions. An attacker with local access to a system using an ARM Mali CSF-based GPU can potentially corrupt kernel memory or leak information by racing device teardown against pending firmware event processing. CVSS is 7.8 (High) with full confidentiality/integrity/availability impact; no public exploit has been identified and it is not in CISA KEV.
Local privilege escalation and memory corruption in the Linux kernel's Panthor DRM driver (ARM Mali GPU) arises from a use-after-free on kernel buffer-object virtual-address nodes; when the GPU MMU is down, panthor_vm_unmap_range() returns early and leaves a va_node stuck in the drm_mm tree, so removing adjacent nodes later dereferences freed memory. A local user with GPU/render access can trigger the condition to corrupt kernel memory (CVSS 7.8, high confidentiality/integrity/availability impact). No public exploit identified at time of analysis; an upstream fix has been merged to stable trees.
Use-after-free in the Linux kernel's Tegra210 QSPI (tegra210-quad) SPI controller driver arises when a transfer times out before the threaded IRQ handler runs, leaving the curr_xfer field pointing at freed memory that the IRQ thread may later dereference. Local low-privileged users on NVIDIA Tegra-based systems could trigger memory corruption affecting confidentiality, integrity, and availability, though exploitation depends on winning a narrow timing race. There is no public exploit identified at time of analysis and it is not listed in CISA KEV; the fix is delivered as upstream stable-tree commits.
Kernel instability in the Linux qla2xxx SCSI driver (QLogic Fibre Channel target/host adapter) allows a chip-reset event to leave in-flight commands stuck and never freed, and can trigger a BUG_ON kernel panic (BUG_ON(cmd->sg_mapped) or BUG_ON(!valid_dma_direction(dir))) when two CPUs race to unmap the same scatter-gather list. This fix reverts commit 26f9ce53817a and partially reverts aefed3e5548f at __qla2x00_abort_all_cmds to restore correct command cleanup after a firmware/chip reset. No public exploit identified at time of analysis; the NVD-assigned CVSS of 9.8 (AV:N) is inconsistent with the local, hardware-dependent nature of the bug.
Memory corruption in the Linux kernel's qla2xxx QLogic Fibre Channel HBA driver arises from an error-handling path in qla2xxx_process_purls_iocb() that calls kfree() on a purex item which may have originated from a per-adapter pre-allocated pool rather than kzalloc(). Freeing a pool-backed object with kfree() can corrupt kernel heap state. The fix routes deallocation through qla24xx_free_purex_item(), which correctly distinguishes dynamically allocated from pooled items; no public exploit identified at time of analysis, and the issue is not in CISA KEV.
Access-control inconsistency in the Linux kernel's Landlock LSM allows a sandboxed local task to widen its effective filesystem access rights by operating on "disconnected" directories - files opened through a bind mount and then renamed or moved out of scope of the mount point. Because the mount point could not be located during the hierarchy walk, Landlock collected access rights straight down to the filesystem root, ignoring the more restrictive mount-point hierarchy and permitting renames (notably under LANDLOCK_ACCESS_FS_REFER) that should have been denied. Only Linux systems actively enforcing Landlock sandboxing are affected; there is no public exploit identified at time of analysis and EPSS is negligible (0.02%, 6th percentile).
Local privilege escalation via a use-after-free in the Linux kernel's drm/panthor driver (the DRM driver for newer Arm Mali CSF-based GPUs) allows a local user with GPU render-node access to free and reuse a group object during creation. The GROUP_CREATE ioctl stored a group pointer in an Xarray and then continued to dereference it, letting a second thread call GROUP_DESTROY on a guessed handle to trigger the UAF. No public exploit or CISA KEV listing has been identified; the fix adds an Xarray entry mark so a group cannot be destroyed until GROUP_CREATE has fully returned.
Kernel panic in the Linux octeontx2-af driver exposes Marvell CN103XX OcteonTX2 systems to a denial-of-service crash when firmware reports a physical link change event with an invalid lmac_type_id value. The driver's cgx_link_change_handler() uses the unvalidated firmware-supplied ID to index into an array and calls strncpy with the resulting out-of-bounds pointer, producing a fatal kernel Oops in a workqueue context. No public exploit has been identified and EPSS sits at 0.17% (6th percentile), but the crash is deterministic once the invalid firmware value is processed, making Availability impact high on affected hardware.
Use-after-free in the Linux kernel NTFS3 filesystem driver allows a local attacker with low privileges to corrupt kernel heap memory via a crafted NTFS volume with malformed extended attributes. The `ntfs_read_ea` function fails to propagate error codes on encountering inconsistent EA structures, causing `ntfs_set_ea` to perform a 4-byte write to already-freed memory - confirmed by an in-description KASAN trace captured at 6.2.0-rc1+. No active exploitation is confirmed (not in CISA KEV), and an EPSS of 0.16% (6th percentile) signals very low observed exploitation activity, though a researcher-developed proof-of-concept is implied by the kernel trace.
Local privilege-tier memory corruption in the Linux kernel's Intel VPU/NPU accelerator driver (accel/ivpu) arises from a dangling buffer object left on a per-device list after a failed allocation, later dereferenced in ivpu_bo_unbind_all_bos_from_context() to cause a page fault. A local user with access to the ivpu device can trigger this use-after-free via the DRM/accel interface, with CVSS 7.8 (High) and CISA-style impact of confidentiality, integrity, and availability all High. There is no public exploit identified at time of analysis and the flaw is not in CISA KEV; the primary observed effect is a kernel page fault (crash), though the dangling-pointer condition carries corruption potential.
Kernel-side memory corruption in the Linux kernel's crypto AEAD framework arises because AEAD algorithms began populating the generic cra_reqsize field (added by commit afddce13ce81d) without the framework being updated to translate it into a proper per-request size via crypto_aead_set_reqsize(). As a result, request buffers for affected AEAD transforms are undersized, producing out-of-bounds memory writes and kernel crashes when the algorithm is used. This is tracked as a Denial-of-Service class issue; no public exploit identified at time of analysis, and it is not listed in CISA KEV.
BPF test infrastructure in the Linux kernel allows a local user with BPF program execution privileges to emit packets with malformed GSO (Generic Segmentation Offload) metadata to the network stack via bpf_clone_redirect(), triggering skb_warn_bad_offload() and disabling GSO-related offload features - a denial-of-service condition. The root cause is convert___skb_to_skb() populating gso_segs and gso_size without a corresponding gso_type, producing an internally inconsistent socket buffer that fails GSO validation in netif_skb_features(). No active exploitation has been confirmed (not in CISA KEV) and no public exploit code has been identified; the issue was originally surfaced through fuzzing by Yinhao et al. Patches across six stable kernel branches have been released.
Memory corruption in the Linux kernel's asymmetric_keys crypto subsystem allows a local low-privileged user to trigger a heap buffer overflow by supplying a crafted X.509 certificate. The flaw lives in asymmetric_key_generate_id(), where the sizes of attacker-controlled binary blobs (such as arbitrarily large ASN.1 INTEGER serial numbers and issuer names) are added to the size of an asymmetric_key_id structure without overflow checking, so an integer overflow yields an undersized allocation and a subsequent out-of-bounds copy. No public exploit identified at time of analysis, and EPSS is low (0.15%, 5th percentile).
Local denial of service in the Linux kernel's software RDMA driver (RDMA/rxe, rdma_rxe module) allows a user with access to an rxe device to crash the host via a NULL pointer dereference. When ibv_modify_srq() is called to resize a shared receive queue and rxe_queue_resize() fails, rxe_srq_from_attr() leaves srq->rq.queue set to NULL; a subsequent modify call dereferences srq->rq.queue->buf->index_mask in rxe_srq_chk_attr() and panics. No public exploit identified at time of analysis, and the EPSS score is low (0.16%, 6th percentile).
Local privilege-limited denial of service in the Linux kernel's CoreSight TMC (Trace Memory Controller) driver allows a user with performance-monitoring access to trigger a kernel oops via an invalid pointer dereference in tmc_etr_get_buffer(). The flaw occurs because the perf event handle was not attached to the coresight_path, so dependent devices such as the CATU could not retrieve the per-CPU AUX_EVENT during path enablement. It is reproducible with a standard perf record command against CoreSight ETM tracing; there is no public exploit identified at time of analysis and EPSS is low (0.16%).
Denial of service in the Linux kernel's ntfs3 filesystem driver stems from a run_lock rwsem that is never initialized when the special $Extend metadata inode is loaded. When $Extend's inode mode is set to a regular file, a truncate() call reaches ntfs_set_size() and acquires the uninitialized lock, tripping lockdep and risking kernel instability. The regression was introduced by commit 4e8011ffec79 and was found via syzbot fuzzing; no public exploit identified at time of analysis, and EPSS is low at 0.17%.
Uninitialized memory use in the Linux kernel NTFS3 filesystem driver (fs/ntfs3) enables a local low-privileged user to trigger a kernel crash, causing a denial of service. The `__getname()` function allocates slab memory via `kmem_cache_alloc()` without zeroing it; `ntfs_read_hdr` and `bcmp` then operate on this indeterminate data, as detected by KMSAN. No active exploitation is confirmed (not in CISA KEV), and the EPSS score of 0.18% at the 7th percentile reflects very low real-world exploitation probability. Upstream patches are available across multiple stable kernel branches.
Denial of service in the Linux kernel's mt76 wireless driver (MT7996 WiFi 7 support) allows the kernel to crash (NULL/invalid pointer dereference) when hardware-accelerated flow offloading runs on a secondary Wireless Ethernet Dispatch instance (wed_hif2, e.g. a 6GHz link). The mt76 WED callbacks unconditionally referenced the primary wed device, so tearing down or disabling offload on wed_hif2 dereferences an invalid address and triggers a kernel Oops. There is no public exploit identified at time of analysis, EPSS risk is low (0.18%, 8th percentile), and it is not in CISA KEV; impact is availability loss (system crash), not code execution.
Memory corruption (double free) in the Linux kernel btrfs filesystem's quota group (qgroup) accounting affects the add_delayed_ref_head() code path when a qgroup record is reported as already existing but a subsequent error occurs, causing both caller and callee to free the same 'qrecord' object. The flaw is a local btrfs-triggered condition on systems using btrfs with quota groups; there is no public exploit identified at time of analysis and it is not listed in CISA KEV, with a very low EPSS of 0.17%. Despite an automated CVSS of 9.8, the realistic attack surface is local to the machine and gated by a hard-to-reach error path.
Out-of-bounds memory access in the Linux kernel's ch341 SPI driver (drivers/spi/spi-ch341) allows a local low-privileged actor to trigger heap buffer overreads and overwrites during SPI transfers on systems using a WCH CH341 USB-to-SPI bridge. The flaw stems from ch341_transfer_one copying 'len' bytes (which includes a 1-byte command header) instead of 'len - 1' from the user-supplied transfer buffer, causing both an out-of-bounds read from trans->tx_buf and, when len equals the 32-byte packet length, a 1-byte overflow of ch341->tx_buf. EPSS is low at 0.18% and there is no public exploit identified at time of analysis, consistent with a niche hardware-dependent kernel driver rather than a widely exploited target.
Denial of service in the Linux kernel's NFSv4/pNFS client allows a null-pointer dereference and system crash during inode writeback, affecting hosts mounting parallel-NFS (pNFS) exports. The flaw occurs because NFS_INO_LAYOUTCOMMIT was not cleared in pnfs_mark_layout_stateid_invalid, so the write_inode → nfs4_write_inode → pnfs_layoutcommit_inode path may dereference a layout that has already been freed. There is no public exploit identified at time of analysis, and the low EPSS score (0.18%) indicates minimal observed exploitation interest despite the CVSS 7.5 rating.