Red Hat
Monthly
Denial of service in Apache Qpid Proton-Dotnet 1.0.0 and earlier allows a pre-authentication remote attacker to crash the AMQP service by submitting messages with deeply nested type structures, triggering a StackOverflowError. The CVSS 7.5 (High) reflects a fully network-exposed, unauthenticated attack path with complete availability impact. No active exploitation has been confirmed (EPSS: 0.19%, 9th percentile; no CISA KEV listing), and the vendor has released a direct fix in version 1.1.0.
Excessive memory allocation in Apache Qpid Proton Dotnet through 1.0.0 enables pre-authentication denial of service via crafted AMQP messages. The .NET AMQP client library fails to properly validate type size or count fields during message parsing, allowing a remote unauthenticated attacker to trigger unbounded memory allocation and exhaust available resources. No public exploit code has been identified at time of analysis, but the pre-authentication vector and network reachability of AMQP services make this a realistic threat to messaging infrastructure.
Pre-authentication denial of service in Apache Qpid Proton-Dotnet through version 1.0.0 allows remote unauthenticated attackers to exhaust server resources by exploiting unbounded caching of AMQP symbol values. Because the vulnerability is reachable before any authentication handshake completes, no credentials or prior access are required to trigger resource exhaustion. No public exploit or KEV listing is identified at time of analysis, but the pre-auth nature and low attack complexity make this a meaningful availability risk for any internet- or network-exposed AMQP endpoint using the affected .NET library.
Excessive heap allocation in Apache Qpid Proton-J's AMQP type parser exposes any network-accessible AMQP endpoint to unauthenticated denial-of-service before an authentication handshake completes. Malformed AMQP frames containing attacker-controlled size or count field values cause the Java library to allocate unbounded memory, leading to JVM heap exhaustion and service crash. No public exploit code has been identified and the vulnerability is not listed in CISA KEV at time of analysis, but the pre-authentication attack vector and zero-privilege requirement make this a high-priority patch target for any internet-exposed Proton-J deployment.
Resource exhaustion leading to denial of service in Apache Qpid Proton-J through 0.34.1 can be triggered by a pre-authentication attacker via unbounded AMQP symbol value caching. Any unauthenticated network peer capable of establishing an AMQP connection can send crafted frames that cause the library to cache symbol values without bound, consuming memory until the hosting process becomes unavailable. No public exploit code has been identified at time of analysis, and this vulnerability is not listed in the CISA KEV catalog.
Use-after-free in the Linux kernel rhashtable walk subsystem allows a local attacker with low privileges to dereference a stale pointer into freed slab memory, enabling kernel memory disclosure, crash, or potential privilege escalation. The defect is in rhashtable_walk_start_check(): when a hash table is freed during resize (iter->walker.tbl == NULL), the restart path clears slot and skip but omits clearing iter->p, leaving a dangling pointer that rhashtable_walk_next() then dereferences. Confirmed callers affected include netlink_diag and TIPC socket diagnostics. No public exploit or CISA KEV listing exists at time of analysis; EPSS is 0.15% (5th percentile), reflecting low current exploitation activity despite real kernel memory-safety impact.
Race condition in the Linux kernel's KVM nVMX (nested VMX) subsystem allows a low-privileged local attacker to trigger a use-after-free via shadow VMCS teardown, with potential for full hypervisor compromise. The flaw arises because free_nested() releases the shadow VMCS page while vmcs01 still holds a dangling pointer; if the vCPU migrates to another physical CPU before the pointer is cleared, __loaded_vmcs_clear() executes VMCLEAR against the freed page, corrupting kernel memory. No public exploit code exists and no active exploitation is confirmed (EPSS 0.16%, 5th percentile), but the S:C scope change in the CVSS vector and the C:H/I:H/A:H impact triad indicate this could enable a guest-to-host escape under the right conditions.
Heap-buffer-overflow vulnerabilities in GIMP's APNG file loader and DDS plug-in allow code execution when a user opens a specially crafted image file. The APNG flaw is triggered when the fcTL chunk width exceeds the IHDR width, causing pixel data to overwrite memory beyond the heap allocation; the DDS flaw stems from a bits-per-pixel mismatch in load_layer(). Both affect GIMP as packaged on Red Hat Enterprise Linux 6 through 9. No public exploit identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Uncontrolled recursion in Bouncy Castle for Java's lazy ASN.1 parser enables remote unauthenticated denial of service across all three library release lines (bc-java, bc-lts-java, and bc-fja FIPS). When a LazyEncodedSequence is forced during tree traversal, it resets the nesting-depth guard to its configured maximum instead of continuing with the remaining depth budget, allowing crafted deeply-nested ASN.1 structures to bypass StreamUtil's recursion limit and trigger a StackOverflowError. No public exploit has been identified at time of analysis, but any Java application accepting externally-supplied ASN.1 data - TLS certificates, CMS signatures, PKCS structures - is potentially exposed without authentication.
Out-of-memory denial-of-service in Bouncy Castle for Java allows remote unauthenticated attackers to crash Java applications by submitting a crafted ASN.1 definite-length object with an inflated declared size. The ASN.1 parser in `DefiniteLengthInputStream.toByteArray()` allocated a full heap-sized byte array before reading any body bytes, meaning a few-byte header claiming a near-JVM-heap payload was sufficient to trigger OOM. All major Bouncy Castle release lines are affected - including the FIPS-certified BC-FJA variant - and a vendor patch is available; no public exploit code or CISA KEV listing exists at time of analysis.
Uncontrolled memory allocation in Bouncy Castle for Java's HSS public-key parser allows remote unauthenticated attackers to exhaust JVM heap memory by supplying a crafted public key with an arbitrarily large HSS level count (L value), causing denial of service during signature verification. All three product lines are affected: bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FIPS (bc-fja) before 2.0.2/2.1.3 depending on series. No public exploit has been identified at time of analysis and no CISA KEV listing exists, but the AV:N/AC:L/PR:N attack profile and Bouncy Castle's ubiquity as a Java crypto dependency make this a credible availability risk for any service accepting external HSS/LMS key material.
CCMBlockCipher in Bouncy Castle for Java decrypts ciphertext directly into the caller-supplied output buffer before verifying the authentication tag, leaving unverified CTR-decrypted plaintext in memory when tag verification fails and violating NIST SP 800-38C §6.2. Every edition - bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA across three active release lines (1.0.x, 2.0.x, 2.1.x) - is affected. Applications that reuse pooled byte buffers, log failure-path output, or otherwise expose the buffer on an authentication failure inadvertently become CTR decryption oracles, allowing attackers to submit forged ciphertexts and recover plaintext without possessing a valid authentication tag. No public exploit and no CISA KEV listing exist at time of analysis.
Signature bypass in Bouncy Castle for Java allows any attacker-supplied content wrapped in a zero-signer CMS envelope to pass `CMSSignedData.verifySignatures()` without any valid cryptographic signature. The method iterates an empty signerInfos SET, falls through to a vacuous `return true`, and any application using it as its top-level authenticity check silently accepts unsigned attacker-controlled content as authenticated - a classic fail-open flaw (CWE-347). No public exploit has been identified at time of analysis and the vulnerability is not listed in CISA KEV, but the attack primitive is trivially constructible using any standard CMS/PKCS#7 library.
Bouncy Castle for Java's OpenPGP session-key decryption path re-exposes the Mister-Zuccherato adaptive-chosen-ciphertext oracle by failing to suppress the legacy CFB 'quick check' on the PKESK/session-key route, enabling an attacker with decryption oracle access to recover plaintext from SEIPD v1 encrypted messages. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA (bcpg-fips) across the 1.0.X, 2.0.X, and 2.1.X series - three distinct product lines used broadly in enterprise and open-source Java applications. No public exploit or CISA KEV listing has been identified at time of analysis; however, the attack technique is well-established in cryptographic literature dating to the Mister-Zuccherato paper.
Signature forgery via back-dated signingTime in Bouncy Castle's S/MIME validator allows an attacker holding a revoked or expired signing certificate to have that certificate accepted as valid by unpatched validators. The SignedMailValidator in bc-java before 1.85, bc-lts-java before 2.73.12, and all bc-fja (FIPS) lines before their respective 1.0.7/2.0.7/2.1.7 fixes unconditionally used the signer-asserted CMS signingTime attribute as the PKIX path-validation date, making certificate expiry and revocation checks bypassable. No public exploit or CISA KEV listing exists at time of analysis, but the flaw is architecturally straightforward to exploit by any party that holds or has previously held a Bouncy Castle-validated signing certificate.
CMS AuthenticatedData integrity verification in Bouncy Castle for Java is bypassable when authenticated attributes (authAttrs) are present, allowing an attacker to substitute arbitrary content while the recipient MAC check passes without error. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA across the 1.0.x, 2.0.x, and 2.1.x series - all widely deployed in Java enterprise and FIPS environments. The CVSS 4.0 vector (AV:N/AC:L/PR:N/UI:N/VI:H) reflects unauthenticated remote exploitation with high integrity impact; no public exploit or CISA KEV listing is confirmed at time of analysis.
OpenPGP inline-signature policy enforcement in Bouncy Castle for Java silently discards rejection exceptions, causing signatures that use prohibited weak hash algorithms (MD5, SHA-1, RIPEMD-160) or otherwise violate the configured OpenPGPPolicy to be reported as cryptographically valid. Affected are bc-java before 1.85 and bcpg-fips (bc-fja) before 2.0.13; the detached-signature and prefixed-signature verification paths are not affected - only the inline one-pass path. A vendor patch is available; no public exploit code has been identified at time of analysis, though the CVSS 4.0 score of 8.7 reflects a high integrity impact driven by the unauthenticated network vector.
Unbounded recursion in Bouncy Castle's OER (Octet Encoding Rules) parser enables remote unauthenticated denial-of-service against any Java application processing IEEE 1609.2 V2X messages. Because the IEEE 1609.2 schema is cyclic - an Ieee1609Dot2Data structure can nest inside its own SignedData payload - a few kilobytes of crafted input drive the parser into unbounded recursive descent, terminating in a JVM StackOverflowError. Critically, parsing occurs before signature verification, so no authentication or trust relationship is needed to trigger the crash. No public exploit has been identified at time of analysis, but the patch commit exposes two additional heap-exhaustion vectors alongside the primary recursion flaw, all reachable from the same pre-verification code path.
Memory exhaustion denial-of-service in Bouncy Castle's DTLS implementation allows remote unauthenticated attackers to crash or degrade Java applications by sending crafted handshake messages with inflated 24-bit length fields. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA (bctls-fips) before 1.0.24, 2.0.24, or 2.1.24 depending on the deployed branch. The buffer allocation occurs pre-authentication during the DTLS handshake reassembly phase - before any signature or Finished verification - making this exploitable without credentials against any exposed DTLS endpoint. No public exploit code has been identified and the vulnerability is not listed in CISA KEV.
Memory exhaustion in Bouncy Castle for Java's OpenPGP parser enables unauthenticated remote denial of service via a crafted User Attribute subpacket with an oversized length header. All Java applications importing untrusted OpenPGP certificates using bc-java before 1.85, bc-lts-java before 2.73.12, or BC-FJA (FIPS) in affected series are vulnerable to pre-authentication JVM heap exhaustion. No KEV listing or public exploit has been confirmed at time of analysis, but the zero-complexity unauthenticated network attack vector and CVSS 4.0 score of 8.7 signal high operational priority for services that accept OpenPGP certificates from external sources.
Bouncy Castle for Java's BKS keystore implementation silently accepted legacy version 0/1 keystores that derive a critically weak 16-bit HMAC integrity key - brute-forceable offline in milliseconds against the 65,536-value keyspace - allowing an attacker who can supply or tamper with a keystore file to forge the integrity MAC, inject rogue trusted certificates, and have the manipulated keystore accepted without error. All bc-java releases before 1.85 and bc-lts-java releases before 2.73.12 are affected. No public exploit code has been identified and this vulnerability does not appear in the CISA KEV catalog at time of analysis.
Bouncy Castle for Java's BKS and UBER keystore implementation crashes with an OutOfMemoryError when loading a crafted keystore because heap allocation from stream-declared lengths occurs before the HMAC integrity check. Any application calling KeyStore.load() with attacker-controlled BKS or UBER keystore data is affected - bc-java before 1.85 and bc-lts-java before 2.73.12. A POC was submitted alongside the report, confirming that a file of just a few dozen bytes declaring Integer.MAX_VALUE-length fields is sufficient to exhaust JVM heap regardless of heap configuration or supplied password. No public exploit identified at time of analysis beyond the researcher's POC referenced in the patch commit.
Entropy source compromise in Data::Entropy (Perl) before version 0.010 allows any on-path network attacker to fully control the random bytes returned by rand_bits and rand_int for every application that selects the RandomOrg or RandomnumbersInfo remote sources via with_entropy_source. The library fetches entropy over plain HTTP with no transport security, and Data::Entropy::RawSource::RandomOrg's integrity check trivially accepts any non-empty byte string, so substituted attacker-controlled responses are never rejected. A vendor-released patch is confirmed at version 0.010; no active exploitation is confirmed (EPSS 0.10%, 1st percentile, not in CISA KEV), but the deterministic nature of the flaw means any secret derived from these sources is entirely known to the attacker.
Cross-tenant credential theft in pgAdmin 4 (9.0 through 9.16) lets an authenticated non-owner user hijack another user's shared server connection by abusing the Workspaces adhoc-connect endpoint (/misc/workspace/adhoc_connect_server), which clones a target server row verbatim - including its owner's stored database password and tunnel password - and reassigns ownership to the caller. The victim is typically an administrator whose shared server carries saved credentials, and the attacker gains use of those database credentials and any privileges they confer. There is no public exploit identified at time of analysis and it is not on CISA KEV, but the fix (pgAdmin 4 9.17) and the underlying commit are public, and EPSS was not provided.
SQL injection in pgAdmin 4 through an incomplete remediation of CVE-2026-12044 allows a low-privileged authenticated user to execute arbitrary SQL in a higher-privileged victim's database session. By creating a PostgreSQL object - table, index, publication, or subscription - with an apostrophe in its name (a character PostgreSQL legally permits in quoted identifiers), the attacker plants a payload that fires the moment any other pgAdmin user opens the object's Statistics or Dependencies tab, triggering raw Jinja2 template interpolation in coll_stats.sql, dependencies.sql, and get_position.sql without the qtLiteral escaping applied by the prior fix. The cross-user injection escalates impact well beyond the attacker's own database privileges, with the injected SQL executing under the victim's session credentials. No public exploit has been identified at time of analysis, and this CVE has not been added to the CISA KEV catalog.
Denial of service in gnome-remote-desktop as shipped in Red Hat Enterprise Linux 8, 9, and 10 lets an unauthenticated remote attacker exhaust server resources when the daemon runs in system mode with RDP enabled. Because the incoming connection handler bypasses the connection throttler, an attacker can open large numbers of parallel pre-authentication RDP connections, accumulating accepted sockets and pending routing-token operations until timeout and blocking legitimate RDP logins. There is no public exploit identified at time of analysis, and the flaw is specific to Red Hat's build - it does not affect the upstream project.
Deploy-time policy bypass in Red Hat Advanced Cluster Security for Kubernetes 4 lets a user with Kubernetes Deployment-create permission spoof workload identity by setting the openshift.io/encoded-deployment-config label to "null", forcing ACS to record the workload with empty UID, name and labels and a "default" namespace. This evades ACS deploy-time policy detection and enforcement visibility and corrupts Central's persistence, violation reporting and compliance correlation. No public exploit identified at time of analysis and the issue is not listed in CISA KEV.
Denial of service in the aws-smithy-json Rust runtime crate (versions before 0.62.7) lets remote unauthenticated attackers abort a process via stack exhaustion. Because the smithy-rs code generator wires this crate's deserializer into every generated struct, any smithy-rs generated server that parses JSON is exposed, and a single small HTTP request carrying deeply nested JSON triggers uncontrolled recursion in the unknown-key skip path. No public exploit was identified at time of analysis and the flaw is not in CISA KEV, but the trigger is trivial and a vendor patch is available.
Arbitrary code execution in the Banks Python LLM-templating library (versions prior to 2.4.3) allows an attacker who can influence Tool JSON definitions within a {% completion %} block to run any importable Python callable in the host process. Banks resolves the attacker-controlled import_path field via importlib.import_module() plus getattr() with no allowlist, so values like os.system or subprocess.getoutput become invokable when the LLM emits a matching tool_calls entry, passing decoded arguments as kwargs. No public exploit identified at time of analysis; this bug is explicitly distinct from and unfixed by the 2.4.2 SandboxedEnvironment patch (GHSA-gphh-9q3h-jgpp / CVE-2026-44209) because the malicious payload is plain text in the rendered body and never touches Jinja attribute access.
Arbitrary file disclosure in the Banks Python prompt-templating library (masci/banks) before 2.4.4 lets attackers who control template variables read any file the Python process can access. All four media filters (image, audio, video, document) wrap untrusted input in Path(value) and feed it straight to open(file_path, "rb") with no sanitization, so path traversal (../) reads files like .env, SSH keys, cloud credentials, source, /etc/passwd and /etc/shadow, returning them base64-encoded in the rendered prompt for trivial exfiltration. No public exploit identified at time of analysis and it is not in CISA KEV, but the fix in 2.4.4 is straightforward to reverse-engineer.
Confidential Active Directory attribute disclosure in Samba's Active Directory Domain Controller allows an authenticated low-privilege domain user to read protected directory attributes by abusing how LDAP Compare requests are handled. Because Samba fails to validate user-supplied attribute names (LDAP filter injection, CWE-90) and runs the resulting internal search in a trusted context, normal ACL enforcement is bypassed; in gMSA-enabled deployments this permits extraction of msKds-RootKeyData and offline derivation of Group Managed Service Account passwords, which can escalate to full domain compromise. No public exploit identified at time of analysis and the issue is not listed in CISA KEV, but a vendor patch is available.
Denial of service in IBM Enterprise Build of Quarkus (versions 3.27.1 through 3.27.4.SP2 and 3.33.1 through 3.33.2.SP2) allows a remote, unauthenticated attacker to exhaust memory by sending multipart requests whose MIME part-headers are accumulated without bound in the Quarkus REST layer. Because each part-header buffer grows without a size cap, an attacker can force excessive heap allocation and crash or hang the service, degrading availability for all users. There is no public exploit identified at time of analysis, but the flaw is trivially reachable (AV:N/AC:L/PR:N/UI:N) against any endpoint that accepts multipart input.
Service-account token disclosure in the Red Hat OpenShift koku-metrics-operator lets a user with edit rights on the CostManagementMetricsConfig custom resource redirect metrics uploads to an attacker-controlled URL, to which the operator blindly attaches its own Kubernetes bearer token. This CWE-918 server-side request forgery (tagged SSRF/Kubernetes/Red Hat) yields the operator's service-account credentials, enabling in-cluster privilege escalation. Reported by Red Hat with a CVSS 3.1 base of 7.6 (scope-changed); no public exploit identified at time of analysis and it is not in CISA KEV.
Out-of-bounds write in PHP's BCMath bccomp() function allows attacker-controlled inputs to corrupt stack and heap memory in PHP 8.4.x before 8.4.24 and 8.5.x before 8.5.9, with high confidentiality, integrity, and availability impact per the vendor CVSS 4.0 score of 8.1. Exploitation requires an application to pass attacker-influenced values into bccomp() with the BCMath extension enabled; there is no public exploit identified at time of analysis (CVSS E:U, no KEV listing). Memory corruption of this class can lead to crashes and, potentially, arbitrary code execution within the PHP process.
SQL injection in the PHP interpreter's string-escaping routine affects PHP 8.2.x before 8.2.33, 8.3.x before 8.3.33, 8.4.x before 8.4.24, and 8.5.x before 8.5.9, where backslashes in attacker-controlled parameters are improperly escaped, allowing malicious input to break out of quoted SQL string literals. Any web application relying on PHP's built-in escaping to sanitize user data before database queries can be subverted, enabling trivial injection. This is a CWE-89 flaw reported by the PHP project (advisory GHSA-7qpv-r5mr-78m4); no public exploit identified at time of analysis, and CVSS 4.0 rates it 8.1 with exploitation marked automatable.
Account takeover in osTicket (the open-source support ticketing system) before v1.17.8 and v1.18.4 stems from an inverted logic check (CWE-640) in the password-reset token validation routine, where the expiry check runs only when the token's timestamp lookup fails - so any token with a valid timestamp bypasses expiration entirely. An attacker who obtains a valid reset token can replay it after it should have expired to hijack the associated account. Rated CVSS 4.0 9.1 (Critical); no public exploit identified at time of analysis and not listed in CISA KEV.
Server-side request forgery in the Eclipse CSI PIA `POST /v1/upload/sbom` endpoint lets remote attackers coerce the server into connecting to an arbitrary host, port, and scheme of their choosing. By exploiting a parser-differential between Python's `urlparse` (used for the issuer allowlist check) and `requests`/`urllib3`/`urllib` (used for OIDC discovery and JWKS retrieval), an attacker crafts a JWT `iss` claim containing a backslash that passes validation but resolves to a different host during the actual fetch. The `iss` claim is only checked in its unverified form, so no legitimate token is needed; no public exploit has been identified at time of analysis and the issue is not listed in CISA KEV.
Improper input neutralization in Phoenix Contact CHARX SEC electric-vehicle charging controllers lets a remote, unauthenticated attacker inject malicious data through MQTT that the bundled ModbusServer application processes without validation, corrupting Modbus data and disrupting service. All CHARX SEC-3000/3050/3100/3150 units running firmware 1.0.0 through 1.9.0 are affected, with impact limited to integrity and availability (no data disclosure). No public exploit identified at time of analysis; the flaw was coordinated-disclosure reported through CERT@VDE and a fixed firmware (1.9.1) is available.
Memory-corruption remote code execution in Google Chrome's Dawn WebGPU graphics component affects all Desktop builds prior to 151.0.7922.72, where a crafted HTML page triggers a use-after-free (CWE-416) that lets a remote attacker run arbitrary code - though confined inside the renderer sandbox, which is why Chromium rated it Low despite the CVSS 8.8. Exploitation requires only that a victim load a malicious page (UI:R), with no authentication. There is no public exploit identified at time of analysis and the CVE is not on CISA KEV.
Remote code execution in Google Chrome's PDFium component (versions prior to 151.0.7922.72) allows a remote attacker to run arbitrary code - confined to the renderer sandbox - when a victim opens a maliciously crafted PDF. The flaw is a CWE-416 use-after-free reported by Google's own Chrome team and fixed in the July 2026 Stable channel update; Chromium rated its intrinsic severity Low because execution is contained within the sandbox, though NVD's CVSS scores it 8.8. There is no public exploit identified at time of analysis and it is not listed in CISA KEV.
Local privilege escalation in the Google Chrome Updater on Windows (versions prior to 151.0.7922.72) lets a low-privileged local attacker win a race condition and substitute a malicious file to elevate privileges. The flaw carries a CVSS 7.0 score, and while Chromium rates its security severity as Low, the SSVC technical impact is Total. There is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Arbitrary code execution in Google Chrome's V8 JavaScript engine allows a remote attacker to run code inside the renderer sandbox when a victim visits a crafted HTML page, affecting all desktop builds prior to 151.0.7922.72. The flaw is a CWE-843 type confusion in V8 and carries a CVSS 8.8 due to network reach and high CIA impact, though execution is confined to the sandbox and Chromium rates the security severity as Low. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but a vendor patch is available.
Remote code execution in Google Chrome's V8 JavaScript engine (all versions prior to 151.0.7922.72) allows a remote attacker who lures a victim to a crafted HTML page to execute arbitrary code, though confined inside the renderer sandbox rather than achieving a full sandbox escape. Google rated the Chromium security severity as Low, and there is no public exploit identified at time of analysis and no evidence of active exploitation (CISA SSVC exploitation status: none). Exploitation hinges on winning a timing race, so despite the 7.5 CVSS score it is regarded as difficult to trigger reliably.
Remote code execution within the browser's renderer sandbox affects Google Chrome desktop builds prior to 151.0.7922.72, stemming from an inappropriate implementation in the Passwords component (CWE-269). A remote attacker who lures a victim to a crafted HTML page can execute arbitrary code, but execution is confined to the sandbox - which is why Chromium rated the underlying issue Low severity despite the CVSS 8.8. No public exploit identified at time of analysis, and the fix is already shipping in the Stable channel.
Sandboxed remote code execution in Google Chrome desktop before 151.0.7922.72 lets a remote attacker run arbitrary code within the renderer sandbox when a victim opens a crafted HTML page. The flaw stems from an inappropriate implementation in Chrome's Scheduling component; Google rated it Chromium-Low severity and no public exploit has been identified, though the NVD CVSS of 8.8 reflects the high per-system impact once triggered. Exploitation requires user interaction (visiting a malicious page) and, because code executes inside the sandbox rather than escaping it, real-world impact is constrained absent a chained sandbox-escape bug.
Sandbox-constrained arbitrary code execution in Google Chrome's V8 JavaScript engine affects all desktop builds prior to 151.0.7922.72, where a crafted Chrome extension abuses an inappropriate V8 implementation to run attacker code inside the renderer sandbox. Exploitation requires convincing a user to install a malicious extension, and Google internally rated the Chromium security severity as Low. There is no public exploit identified at time of analysis and CISA SSVC records exploitation status as 'none'.
Remote code execution in Google Chrome on macOS prior to 151.0.7922.72 stems from an inappropriate implementation in the Safe Browsing component, where a remote attacker can trigger arbitrary code execution by luring a victim into opening or downloading a malicious file. The CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:R) requires no privileges but does require user interaction, and Google has shipped a fixed Stable channel build. Despite the 8.8 CVSS, Chromium rated the underlying issue Low severity, EPSS is only 0.20% (10th percentile), and there is no public exploit identified at time of analysis.
Cross-user information disclosure in undici's cache interceptor allows authenticated users' responses to be served from a shared cache to subsequent callers, including unauthenticated ones, when the upstream origin sends qualified `no-cache` or `private` Cache-Control directives with optional whitespace (OWS) around the `=` delimiter. Affected deployments are those explicitly running undici 7.0.0-7.28.x or 8.0.0-8.8.x with `interceptors.cache()` in shared mode while forwarding `Authorization` headers upstream. This is a bypass of the earlier CVE-2026-9678 fix; a public proof-of-concept exists and vendor-released patches are available in undici 7.29.0 and 8.9.0.
Server-side request forgery allowlist bypass in the V programming language standard library (net.urllib / net.http) through version 0.5.2 lets remote unauthenticated attackers reach internal network services that a host-based allowlist was meant to block. The flaw stems from a parser differential: net.urllib.parse() and net.http.get() disagree on how a backslash in a URL's authority is handled, so validation and connection resolve to different hosts. Publicly available exploit code exists and the issue was reported by VulnCheck, though it is not listed in CISA KEV (no public evidence of active exploitation at time of analysis).
Cross-user information disclosure and a parse-time denial of service in undici's HTTP cache interceptor affect versions 7.0.0-7.28.x and 8.0.0-8.8.x when the interceptor runs in shared mode (the default). A response bearing a degenerate qualified private directive (e.g. private="") slips past the shared-cache guard and is later served to a different caller sharing the same cache key, leaking private bodies and headers including Set-Cookie; separately, mixing an unqualified and a qualified private directive throws an uncaught TypeError that can crash the process. No public exploit identified at time of analysis, and EPSS is low (0.27%), but the fix is confirmed in undici 7.29.0 and 8.9.0.
Denial of service in open-iscsi's bundled open-isns library (iSNS implementation) lets an unauthenticated man-in-the-middle attacker crash the iSNS client/server by triggering a double free in the attribute-decoding routines. The flaw stems from freed pointers not being nulled in isns_attr_type_string_decode and isns_attr_type_opaque_decode, so a crafted iSNS message causes the same allocation to be freed twice. There is no public exploit identified at time of analysis and it is not listed in CISA KEV; impact is limited to availability, with no confidentiality or integrity effect.
Denial of service in the cJSON C library (versions through 1.7.19) arises from exponential algorithmic complexity in the cJSON_Compare() function, which recursively descends into each shared object subtree twice per level with no depth guard. Applications that call cJSON_Compare() on attacker-influenced JSON that is structurally equal to a reference document can be forced to burn hours of CPU on a document of only a few hundred bytes (nesting depth around 40), with cost roughly doubling per added level. No public exploit identified at time of analysis and the issue is not on CISA KEV; impact is compute exhaustion only, with no confidentiality or integrity effect.
Denial of service in cJSON (DaveGamble's C JSON parser) through version 1.7.19 lets a remote attacker crash any application that applies an untrusted RFC 6902 JSON Patch through cJSONUtils_ApplyPatches() or cJSONUtils_ApplyPatchesCaseSensitive(). By combining add and copy operations that graft duplicated subtrees, an attacker inflates document nesting far beyond the parser's 1000-level limit, and the unbounded recursion in cJSON_Delete() (plus a permissive CJSON_CIRCULAR_LIMIT of 10000 in cJSON_Duplicate()) exhausts the thread stack. Reported by VulnCheck with a public technical write-up; there is no public exploit identified at time of analysis and it is not on CISA KEV.
Local privilege escalation in open-iscsi's iscsiuio component lets any unprivileged local user connect to the iscsiuio control socket and issue commands normally restricted to root, because the daemon validated peer credentials against the wrong file descriptor. The flaw (CWE-863 Incorrect Authorization) was reported by SUSE and fixed upstream in commit 668ca1df; there is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Out-of-bounds write in GNU binutils' BFD library (bfd/elf32-dlx.c) lets a crafted ELF/DLX object file corrupt memory and achieve arbitrary code execution when parsed by any BFD-consuming tool (objdump, readelf, strip, ld, nm, objcopy). Reported by Red Hat and tracked against RHEL 6-10, Hardened Images and OpenShift 4, the flaw is only reachable when binutils is compiled with the DLX backend (typically --enable-targets=all), so exposure is gated by a non-default build option. There is no public exploit identified at time of analysis (not in CISA KEV), though the researcher demonstrated code execution via a File Stream Oriented Programming (FSOP) attack on glibc FILE structures.
SQL injection in ClickHouse Server versions 26.3.9.8 and earlier lets remote attackers inject arbitrary SQL through the CREATE DICTIONARY functionality, which the advisory claims can escalate to arbitrary code execution. Publicly available exploit code exists (GitHub repository CVE-2026-51992), but there is no CISA KEV listing and no EPSS score in the provided data, so no confirmed active exploitation. The provided CVSS is 9.1, though the description's reference to the CREATE DICTIONARY operation suggests an authenticated privilege may actually be required, which conflicts with the PR:N in the supplied vector.
Privilege escalation in CRIU allows a malicious container process to spoof credentials during checkpoint, resulting in elevated capabilities upon restore. On Red Hat platforms, effective exploitation is mitigated by mandatory access controls, user namespaces, and the requirement for privileged checkpoint initiation. No public exploit identified at time of analysis.
Remote code execution in Apache Axis2/Java through 2.0.0 arises from unsafe deserialization of Java objects in the optional Tribes clustering component, allowing unauthenticated remote attackers with network access to the clustering port to execute arbitrary code via crafted serialized payloads. The vulnerability is only exploitable when clustering is explicitly enabled, which is off by default. No active exploitation or public exploit code has been observed; the issue is resolved by upgrading to version 2.0.1, which removes the entire clustering feature.
Use-after-free vulnerability in Safari’s web content processing allows remote attackers to execute arbitrary code on affected Apple devices when a user visits a malicious website. Affected versions include Safari prior to 26.6, iOS and iPadOS prior to 26.6, macOS Tahoe prior to 26.6, visionOS prior to 26.6, and watchOS prior to 26.6. No public exploit or active exploitation has been identified, and EPSS indicates a low probability of exploitation (0.13%).
Memory corruption in Apple Safari 26.x allows remote code execution via processing maliciously crafted web content. This buffer overflow (CWE-119) affects all Apple platforms—iOS, iPadOS, macOS, visionOS, and watchOS—requiring user interaction to visit a malicious webpage. While the CVSS score is critical (8.8), EPSS (0.13%) indicates low current exploitation probability and no KEV listing or public exploit exists.
Information disclosure in Apple Safari and OS-integrated WebKit allows websites to detect if a user has visited specific URLs. Safari, iOS, iPadOS, macOS, tvOS, visionOS, and watchOS prior to version 26.6 are affected. No public exploit code or active attacks have been reported, and the probability of exploitation is very low (EPSS 0.17%).
Out-of-bounds read in Apple Safari's web content processing can crash the browser and potentially leak sensitive memory contents. This vulnerability affects all Apple platforms where Safari rendering is used and is fixed in Safari 26.6, iOS/iPadOS 26.6, macOS Tahoe 26.6, tvOS 26.6, visionOS 26.6, and watchOS 26.6. Exploitation requires user interaction to visit a malicious site; no known active exploits or public proof-of-concept exist.
Information disclosure and denial of service in GIMP's file-icns plugin allow local attackers to leak heap memory contents via a crafted ICNS image file. When processing a truncated mask resource, the plugin performs an out-of-bounds read, leaking heap data as alpha channel pixel values or crashing the application. User interaction is required to open the malicious file.
Arbitrary code execution in GIMP when opening a crafted FITS file due to an integer overflow in the file-fits plugin that leads to a heap buffer overflow. An attacker can achieve memory corruption and potentially execute arbitrary code with the user's privileges. No public exploit or active exploitation has been identified at time of analysis.
Reflected cross-site scripting in XMLRPC-C Library versions 1.07 through 1.67.01 allows unauthenticated attackers to execute arbitrary JavaScript when a victim clicks a crafted link, as the error page component reflects unsanitized user input. This can lead to theft of sensitive session data or other browser-accessible information from affected services. No public exploit code or active exploitation has been reported at this time.
Heap-based buffer overflow in LibRaw 0.21 allows remote code execution when a user opens a maliciously crafted RAW image file. The flaw resides in the stretch() and fuji_rotate() functions and requires user interaction, enabling a complete compromise of confidentiality, integrity, and availability. Proof-of-concept code exists, but no active exploitation has been confirmed.
Memory corruption in SQLite 3.41's ORDER BY parsing allows remote attackers to crash the database process or leak sensitive data via crafted SQL queries. A use-after-free in ExprList handling, triggered by an ORDER BY clause with many items, can also lead to arbitrary code execution under specific memory conditions. A public advisory exists, but no active exploitation (KEV) or EPSS data is available.
Use-after-free in SQLite 3.41 allows remote attackers supplying crafted SQL to trigger memory corruption during parsing, leading to denial of service, sensitive information leakage, and potential arbitrary code execution.
Use-after-free in SQLite 3.41's JSON parser allows remote code execution. A remote attacker can deliver a crafted JSON payload; if a user or application processes it with SQLite, the vulnerability triggers memory corruption, potentially enabling full system compromise. A proof-of-concept advisory is publicly available, though no active exploitation has been confirmed.
Remote denial of service and heap memory leak in SQLite 3.41 via the json_remove() function allows attackers with the ability to execute SQL queries to crash the database process or leak sensitive heap data. A use-after-free in jsonRemoveFunc is triggered by malformed JSON and a deep path parameter; publicly available proof-of-concept code exists, and low EPSS risk indicates limited exploitation likelihood.
Use-after-free in SQLite 3.41's expression-evaluation path lets an attacker who can submit a crafted SQL statement crash the process, leak adjacent memory, or potentially achieve code execution. The flaw stems from sqlite3ReleaseTempReg() prematurely freeing a temporary register that exprComputeOperands() then dereferences. Rated CVSS 7.5 (AV:N/AC:H/PR:L); publicly available exploit code exists per the CISA SSVC 'poc' determination, but it is not on the CISA KEV list and EPSS is low (0.24%, 16th percentile), indicating no evidence of widespread active exploitation.
Use-after-free in SQLite 3.41 expression parsing leads to memory corruption, allowing remote attackers to crash applications and leak sensitive memory contents via crafted SQL queries. Public proof-of-concept exists but there is no evidence of active exploitation. The vulnerability is triggered by sending malicious SQL statements that cause the parser to access a freed expression object.
State-counter underflow in the Linux kernel's mlx5e (NVIDIA/Mellanox ConnectX) IPsec offload driver lets the eswitch mode-block reference counter (esw->offloads.num_block_mode) be decremented without a matching increment when a temporary acquire-flow security association is installed. The affected code is the mlx5e_xfrm_add_state() acquire-SA path, which skips hardware offload but still falls through to the common success label that unconditionally calls mlx5_eswitch_unblock_mode(). Fixed upstream in the stable kernel series; no public exploit identified at time of analysis and EPSS probability is low (0.21%).
Local privilege escalation in the Linux kernel's ARM FFA firmware interface allows an unprivileged local attacker to achieve arbitrary code execution by triggering an out-of-bounds write when processing PARTITION_INFO_GET descriptors. The kernel fails to validate firmware-supplied indices, enabling a heap buffer overflow that can corrupt memory. Patches are available, no public exploit exists, and EPSS indicates low exploitation probability (0.21%).
Heap-based information disclosure and denial of service in the Linux kernel's mac80211 WiFi subsystem can be triggered by adjacent network attackers. The vulnerability arises from incorrect pointer usage during MLE defragmentation, potentially leading to kernel heap memory over-read and exposure of sensitive data, as well as system crashes. Affected kernels from version 6.9 are vulnerable until patched; no public exploit is known and EPSS indicates a low exploitation probability (0.21%).
Use-after-free in the Linux kernel's usbio (USB I/O bridge) driver allows a local, low-privileged attacker to corrupt kernel slab memory during USB device disconnect, potentially achieving privilege escalation or triggering a kernel panic. The bug lies in usbio_disconnect(), which iterates the client list in reverse using list_for_each_entry_reverse() - a macro that reads client->link.prev after the loop body, but auxiliary_device_uninit() can free the current client before that read, dereferencing freed slab memory. Patches are confirmed in Linux stable branches 6.18.39, 7.1.4, and 7.2-rc3; no public exploit or active exploitation is identified at time of analysis.
Heap buffer overflow in the Linux kernel's rtl8723bs staging WiFi driver (rtw_cfg80211_set_wpa_ie()) allows a local low-privileged user to corrupt kernel memory by one byte via a crafted nl80211 connect request, enabling potential privilege escalation to kernel context. The 256-byte supplicant_ie array in struct security_priv overflows when a WPA IE with length 255 is supplied, writing 257 bytes and clobbering the adjacent last_mic_err_time field; a flawed length check in rtw_parse_wpa_ie() using u8 narrowing silently permits the overflow. Patches are available across all active stable kernels; EPSS is 0.23% and no public exploit or CISA KEV listing exists at time of analysis.
Use-after-free in the Linux kernel Bluetooth MGMT subsystem allows a local attacker with low privileges to trigger kernel memory corruption via a race condition between add_device and remove_device operations. The flaw exists in add_device_complete(), which dereferences a hci_conn_params pointer without holding hci_dev_lock, while a concurrent MGMT_OP_REMOVE_DEVICE path frees the same object under that lock. KASAN confirms exploitation as slab-use-after-free; no public exploit or active exploitation (CISA KEV) has been identified, and patches have been released across multiple stable branches.
Use-after-free memory corruption in the Linux kernel's netpoll subsystem allows a local attacker or concurrent kernel path to access freed memory during the netpoll shutdown sequence. The flaw exists because cancel_delayed_work() - called from softirq context during __netpoll_cleanup() - does not block until the TX worker finishes, permitting queue_process() to dereference a kfree'd npinfo struct via container_of(). KASAN confirms the UAF; patches are available across multiple stable kernel branches, and no public exploit or KEV listing exists at time of analysis.
Use-after-free memory corruption in the Linux kernel's NXP i.MX8 ISI (Image Sensing Interface) media driver is triggered during module removal (rmmod of imx8_isi), confirmed by KASAN slab-use-after-free reports. The root cause is an incorrect teardown order in mxc_isi_remove(): the crossbar cleanup frees media entity pads before the V4L2 cleanup has finished removing media links that reference those same pads, resulting in a dangling pointer dereference in __media_entity_remove_link(). No public exploit has been identified at time of analysis, and with EPSS at 0.21% (11th percentile), real-world weaponization appears unlikely despite the CVSS 7.8 score.
Out-of-bounds memory read in the Linux kernel's btnxpuart Bluetooth driver exposes kernel memory to a malicious or compromised NXP Bluetooth UART controller. The vulnerability exists in nxp_recv_fw_req_v3(), which handles firmware download requests during v3 controller initialization: when the controller supplies a crafted 32-bit offset or 16-bit length value that pushes past the end of the firmware image, the driver reads arbitrary kernel memory and transmits it back to the controller over UART. The parallel v1 path (nxp_recv_fw_req_v1()) already contained the equivalent bounds check; the v3 path lacked it. No active exploitation is confirmed (not in CISA KEV) and EPSS is 0.21% (11th percentile), consistent with the narrow local/hardware attack surface.
Memory leak in the Linux kernel's SMB client encrypted receive path (receive_encrypted_standard()) allows a local low-privileged user or a cooperative malicious SMB server to exhaust kernel heap memory, with potential for heap grooming-based privilege escalation. The bug was introduced with commit b24df3e30cbf and affects multiple stable kernel series from 4.19 onward. No public exploit code exists and EPSS is 0.22%, but the kernel team's classification of CWE-787 and a C:H/I:H/A:H CVSS score reflect the Linux security team's assessment that the leaked kernel buffer is exploitable beyond simple resource exhaustion.
Double-free memory corruption in the Linux kernel's Qualcomm CPU frequency hardware driver (qcom-cpufreq-hw) allows a local low-privileged attacker on Qualcomm SoC-based systems to trigger kernel heap corruption, potentially escalating privileges to root. The flaw originates in qcom_cpufreq_hw_cpu_exit(), which incorrectly calls kfree() on memory allocated via devm_kzalloc() - memory already tracked by the kernel's device resource management (devres) system. This results in a double-free when the platform device is subsequently detached and devres performs its own cleanup. No public exploit code exists and EPSS is low (0.21%), consistent with a static-analysis discovery rather than active exploitation.
Use-After-Free in the Linux kernel HID letsketch driver allows local exploitation during device unbind, potentially enabling kernel memory corruption and privilege escalation on affected systems. The letsketch_probe() function arms a per-device kernel timer (inrange_timer) with a 100 ms timeout, but no .remove callback exists to drain the timer before devm cleanup frees letsketch_data and input_dev structures on USB unplug or rmmod, producing a UAF read/write into freed kernel memory. Patches are backported across six stable kernel branches; no public exploit code exists and EPSS is 0.21% (12th percentile), reflecting the niche device requirement and race-condition complexity.
Use-after-free in the Linux kernel HID appleir driver allows a local attacker to corrupt kernel memory during device removal, potentially achieving privilege escalation. The race condition exists between the key_up_timer callback (armed by appleir_raw_event() on each keydown/key-repeat) and hid_hw_stop(), which frees the underlying input_dev before the timer is drained. A second race window, where a USB URB completion during hid_hw_stop() can re-arm the timer via mod_timer(), means a simple teardown reorder is insufficient — the fix introduces a spinlock-gated 'removing' flag combined with timer_shutdown_sync() to permanently disable further callbacks. No public exploit has been identified at time of analysis, and EPSS is 0.22% (13th percentile).
Use-after-free in the Linux kernel HID lg-g15 driver allows a local attacker to trigger kernel memory corruption by racing device input events against USB keyboard removal. The driver allocates lg_g15_data via devm and schedules work items from HID report handlers (backlight cycle key on G15, G15 v2, G510), but lacked a remove callback to cancel pending work before devres freed the allocation. On kernel versions from approximately 5.5 through the unfixed stable branches, unplugging the keyboard while a work item is pending causes the worker to dereference freed memory via container_of, potentially enabling local privilege escalation. No public exploit has been identified and EPSS is 0.22% (13th percentile), indicating no observed widespread exploitation.
Incorrect RCU lock annotations in the Linux kernel's BPF LPM trie map implementation (`kernel/bpf/lpm_trie.c`) cause lockdep false positives on debug kernels when sleepable BPF programs access `BPF_MAP_TYPE_LPM_TRIE` maps via lookup, update, or delete operations. The mismatch arises because sleepable BPF programs hold `rcu_read_lock_trace()` while the trie code asserts `rcu_read_lock_bh_held()` or `rcu_read_lock_held()`, producing repeated 'suspicious RCU usage' kernel warnings per distinct callsite. The Linux kernel developers explicitly state no actual use-after-free occurs, since Tasks Trace RCU already serializes against trie node reclaim; however, the incorrect annotation undermines lockdep's ability to detect future regressions and is increasingly impactful as sleepable BPF LSM programs become common in production security tooling. No public exploit exists and no CISA KEV listing has been issued.
Use-after-free and memory leak in the Linux kernel's EHCI and FOTG210 USB isochronous submit paths expose systems running kernel versions from 2.6.15 through 7.1.3 to local kernel memory corruption by a low-privileged user. When an isochronous URB submission fails before the staged schedule handoff - either because the host controller becomes inaccessible or usb_hcd_link_urb_to_ep() returns an error - the ehci_iso_sched allocation is not freed and urb->hcpriv retains a dangling pointer, creating a UAF condition. No public exploit exists and the EPSS score is 0.21% (12th percentile), but vendor-released patches across multiple stable branches are available and should be applied to eliminate the risk.
Use-after-free in the Linux kernel USB gadget UDC subsystem exposes systems with DWC3-based USB controllers to a race condition between usb_del_gadget() and gadget_match_driver(), enabling a NULL pointer dereference that can crash the kernel or enable privilege escalation on unpatched systems. The flaw affects all major Linux stable branches from 5.15.x through 7.1.x prior to the patched releases, and is particularly relevant for embedded Linux, Android-based devices, and IoT hardware that use USB gadget/device mode. No public exploit has been identified at time of analysis, and EPSS remains at 0.21% (12th percentile), reflecting current low exploitation interest.
Use-after-free in the Linux kernel's idmouse USB fingerprint driver allows a low-privileged local attacker to corrupt kernel memory by racing a device disconnect against a file-descriptor release, potentially enabling local privilege escalation to root. The bug stems from `mutex_unlock()` accessing the freed driver-data structure after the lock is released, a documented non-atomic behavior; the fix introduces a `kref` reference counter to correctly manage object lifetime. Patches are available across all active stable branches; no active exploitation or public exploit has been reported.
Denial of service in Apache Qpid Proton-Dotnet 1.0.0 and earlier allows a pre-authentication remote attacker to crash the AMQP service by submitting messages with deeply nested type structures, triggering a StackOverflowError. The CVSS 7.5 (High) reflects a fully network-exposed, unauthenticated attack path with complete availability impact. No active exploitation has been confirmed (EPSS: 0.19%, 9th percentile; no CISA KEV listing), and the vendor has released a direct fix in version 1.1.0.
Excessive memory allocation in Apache Qpid Proton Dotnet through 1.0.0 enables pre-authentication denial of service via crafted AMQP messages. The .NET AMQP client library fails to properly validate type size or count fields during message parsing, allowing a remote unauthenticated attacker to trigger unbounded memory allocation and exhaust available resources. No public exploit code has been identified at time of analysis, but the pre-authentication vector and network reachability of AMQP services make this a realistic threat to messaging infrastructure.
Pre-authentication denial of service in Apache Qpid Proton-Dotnet through version 1.0.0 allows remote unauthenticated attackers to exhaust server resources by exploiting unbounded caching of AMQP symbol values. Because the vulnerability is reachable before any authentication handshake completes, no credentials or prior access are required to trigger resource exhaustion. No public exploit or KEV listing is identified at time of analysis, but the pre-auth nature and low attack complexity make this a meaningful availability risk for any internet- or network-exposed AMQP endpoint using the affected .NET library.
Excessive heap allocation in Apache Qpid Proton-J's AMQP type parser exposes any network-accessible AMQP endpoint to unauthenticated denial-of-service before an authentication handshake completes. Malformed AMQP frames containing attacker-controlled size or count field values cause the Java library to allocate unbounded memory, leading to JVM heap exhaustion and service crash. No public exploit code has been identified and the vulnerability is not listed in CISA KEV at time of analysis, but the pre-authentication attack vector and zero-privilege requirement make this a high-priority patch target for any internet-exposed Proton-J deployment.
Resource exhaustion leading to denial of service in Apache Qpid Proton-J through 0.34.1 can be triggered by a pre-authentication attacker via unbounded AMQP symbol value caching. Any unauthenticated network peer capable of establishing an AMQP connection can send crafted frames that cause the library to cache symbol values without bound, consuming memory until the hosting process becomes unavailable. No public exploit code has been identified at time of analysis, and this vulnerability is not listed in the CISA KEV catalog.
Use-after-free in the Linux kernel rhashtable walk subsystem allows a local attacker with low privileges to dereference a stale pointer into freed slab memory, enabling kernel memory disclosure, crash, or potential privilege escalation. The defect is in rhashtable_walk_start_check(): when a hash table is freed during resize (iter->walker.tbl == NULL), the restart path clears slot and skip but omits clearing iter->p, leaving a dangling pointer that rhashtable_walk_next() then dereferences. Confirmed callers affected include netlink_diag and TIPC socket diagnostics. No public exploit or CISA KEV listing exists at time of analysis; EPSS is 0.15% (5th percentile), reflecting low current exploitation activity despite real kernel memory-safety impact.
Race condition in the Linux kernel's KVM nVMX (nested VMX) subsystem allows a low-privileged local attacker to trigger a use-after-free via shadow VMCS teardown, with potential for full hypervisor compromise. The flaw arises because free_nested() releases the shadow VMCS page while vmcs01 still holds a dangling pointer; if the vCPU migrates to another physical CPU before the pointer is cleared, __loaded_vmcs_clear() executes VMCLEAR against the freed page, corrupting kernel memory. No public exploit code exists and no active exploitation is confirmed (EPSS 0.16%, 5th percentile), but the S:C scope change in the CVSS vector and the C:H/I:H/A:H impact triad indicate this could enable a guest-to-host escape under the right conditions.
Heap-buffer-overflow vulnerabilities in GIMP's APNG file loader and DDS plug-in allow code execution when a user opens a specially crafted image file. The APNG flaw is triggered when the fcTL chunk width exceeds the IHDR width, causing pixel data to overwrite memory beyond the heap allocation; the DDS flaw stems from a bits-per-pixel mismatch in load_layer(). Both affect GIMP as packaged on Red Hat Enterprise Linux 6 through 9. No public exploit identified at time of analysis, and the vulnerability is not listed in CISA KEV.
Uncontrolled recursion in Bouncy Castle for Java's lazy ASN.1 parser enables remote unauthenticated denial of service across all three library release lines (bc-java, bc-lts-java, and bc-fja FIPS). When a LazyEncodedSequence is forced during tree traversal, it resets the nesting-depth guard to its configured maximum instead of continuing with the remaining depth budget, allowing crafted deeply-nested ASN.1 structures to bypass StreamUtil's recursion limit and trigger a StackOverflowError. No public exploit has been identified at time of analysis, but any Java application accepting externally-supplied ASN.1 data - TLS certificates, CMS signatures, PKCS structures - is potentially exposed without authentication.
Out-of-memory denial-of-service in Bouncy Castle for Java allows remote unauthenticated attackers to crash Java applications by submitting a crafted ASN.1 definite-length object with an inflated declared size. The ASN.1 parser in `DefiniteLengthInputStream.toByteArray()` allocated a full heap-sized byte array before reading any body bytes, meaning a few-byte header claiming a near-JVM-heap payload was sufficient to trigger OOM. All major Bouncy Castle release lines are affected - including the FIPS-certified BC-FJA variant - and a vendor patch is available; no public exploit code or CISA KEV listing exists at time of analysis.
Uncontrolled memory allocation in Bouncy Castle for Java's HSS public-key parser allows remote unauthenticated attackers to exhaust JVM heap memory by supplying a crafted public key with an arbitrarily large HSS level count (L value), causing denial of service during signature verification. All three product lines are affected: bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FIPS (bc-fja) before 2.0.2/2.1.3 depending on series. No public exploit has been identified at time of analysis and no CISA KEV listing exists, but the AV:N/AC:L/PR:N attack profile and Bouncy Castle's ubiquity as a Java crypto dependency make this a credible availability risk for any service accepting external HSS/LMS key material.
CCMBlockCipher in Bouncy Castle for Java decrypts ciphertext directly into the caller-supplied output buffer before verifying the authentication tag, leaving unverified CTR-decrypted plaintext in memory when tag verification fails and violating NIST SP 800-38C §6.2. Every edition - bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA across three active release lines (1.0.x, 2.0.x, 2.1.x) - is affected. Applications that reuse pooled byte buffers, log failure-path output, or otherwise expose the buffer on an authentication failure inadvertently become CTR decryption oracles, allowing attackers to submit forged ciphertexts and recover plaintext without possessing a valid authentication tag. No public exploit and no CISA KEV listing exist at time of analysis.
Signature bypass in Bouncy Castle for Java allows any attacker-supplied content wrapped in a zero-signer CMS envelope to pass `CMSSignedData.verifySignatures()` without any valid cryptographic signature. The method iterates an empty signerInfos SET, falls through to a vacuous `return true`, and any application using it as its top-level authenticity check silently accepts unsigned attacker-controlled content as authenticated - a classic fail-open flaw (CWE-347). No public exploit has been identified at time of analysis and the vulnerability is not listed in CISA KEV, but the attack primitive is trivially constructible using any standard CMS/PKCS#7 library.
Bouncy Castle for Java's OpenPGP session-key decryption path re-exposes the Mister-Zuccherato adaptive-chosen-ciphertext oracle by failing to suppress the legacy CFB 'quick check' on the PKESK/session-key route, enabling an attacker with decryption oracle access to recover plaintext from SEIPD v1 encrypted messages. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA (bcpg-fips) across the 1.0.X, 2.0.X, and 2.1.X series - three distinct product lines used broadly in enterprise and open-source Java applications. No public exploit or CISA KEV listing has been identified at time of analysis; however, the attack technique is well-established in cryptographic literature dating to the Mister-Zuccherato paper.
Signature forgery via back-dated signingTime in Bouncy Castle's S/MIME validator allows an attacker holding a revoked or expired signing certificate to have that certificate accepted as valid by unpatched validators. The SignedMailValidator in bc-java before 1.85, bc-lts-java before 2.73.12, and all bc-fja (FIPS) lines before their respective 1.0.7/2.0.7/2.1.7 fixes unconditionally used the signer-asserted CMS signingTime attribute as the PKIX path-validation date, making certificate expiry and revocation checks bypassable. No public exploit or CISA KEV listing exists at time of analysis, but the flaw is architecturally straightforward to exploit by any party that holds or has previously held a Bouncy Castle-validated signing certificate.
CMS AuthenticatedData integrity verification in Bouncy Castle for Java is bypassable when authenticated attributes (authAttrs) are present, allowing an attacker to substitute arbitrary content while the recipient MAC check passes without error. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA across the 1.0.x, 2.0.x, and 2.1.x series - all widely deployed in Java enterprise and FIPS environments. The CVSS 4.0 vector (AV:N/AC:L/PR:N/UI:N/VI:H) reflects unauthenticated remote exploitation with high integrity impact; no public exploit or CISA KEV listing is confirmed at time of analysis.
OpenPGP inline-signature policy enforcement in Bouncy Castle for Java silently discards rejection exceptions, causing signatures that use prohibited weak hash algorithms (MD5, SHA-1, RIPEMD-160) or otherwise violate the configured OpenPGPPolicy to be reported as cryptographically valid. Affected are bc-java before 1.85 and bcpg-fips (bc-fja) before 2.0.13; the detached-signature and prefixed-signature verification paths are not affected - only the inline one-pass path. A vendor patch is available; no public exploit code has been identified at time of analysis, though the CVSS 4.0 score of 8.7 reflects a high integrity impact driven by the unauthenticated network vector.
Unbounded recursion in Bouncy Castle's OER (Octet Encoding Rules) parser enables remote unauthenticated denial-of-service against any Java application processing IEEE 1609.2 V2X messages. Because the IEEE 1609.2 schema is cyclic - an Ieee1609Dot2Data structure can nest inside its own SignedData payload - a few kilobytes of crafted input drive the parser into unbounded recursive descent, terminating in a JVM StackOverflowError. Critically, parsing occurs before signature verification, so no authentication or trust relationship is needed to trigger the crash. No public exploit has been identified at time of analysis, but the patch commit exposes two additional heap-exhaustion vectors alongside the primary recursion flaw, all reachable from the same pre-verification code path.
Memory exhaustion denial-of-service in Bouncy Castle's DTLS implementation allows remote unauthenticated attackers to crash or degrade Java applications by sending crafted handshake messages with inflated 24-bit length fields. Affected are bc-java before 1.85, bc-lts-java before 2.73.12, and BC-FJA (bctls-fips) before 1.0.24, 2.0.24, or 2.1.24 depending on the deployed branch. The buffer allocation occurs pre-authentication during the DTLS handshake reassembly phase - before any signature or Finished verification - making this exploitable without credentials against any exposed DTLS endpoint. No public exploit code has been identified and the vulnerability is not listed in CISA KEV.
Memory exhaustion in Bouncy Castle for Java's OpenPGP parser enables unauthenticated remote denial of service via a crafted User Attribute subpacket with an oversized length header. All Java applications importing untrusted OpenPGP certificates using bc-java before 1.85, bc-lts-java before 2.73.12, or BC-FJA (FIPS) in affected series are vulnerable to pre-authentication JVM heap exhaustion. No KEV listing or public exploit has been confirmed at time of analysis, but the zero-complexity unauthenticated network attack vector and CVSS 4.0 score of 8.7 signal high operational priority for services that accept OpenPGP certificates from external sources.
Bouncy Castle for Java's BKS keystore implementation silently accepted legacy version 0/1 keystores that derive a critically weak 16-bit HMAC integrity key - brute-forceable offline in milliseconds against the 65,536-value keyspace - allowing an attacker who can supply or tamper with a keystore file to forge the integrity MAC, inject rogue trusted certificates, and have the manipulated keystore accepted without error. All bc-java releases before 1.85 and bc-lts-java releases before 2.73.12 are affected. No public exploit code has been identified and this vulnerability does not appear in the CISA KEV catalog at time of analysis.
Bouncy Castle for Java's BKS and UBER keystore implementation crashes with an OutOfMemoryError when loading a crafted keystore because heap allocation from stream-declared lengths occurs before the HMAC integrity check. Any application calling KeyStore.load() with attacker-controlled BKS or UBER keystore data is affected - bc-java before 1.85 and bc-lts-java before 2.73.12. A POC was submitted alongside the report, confirming that a file of just a few dozen bytes declaring Integer.MAX_VALUE-length fields is sufficient to exhaust JVM heap regardless of heap configuration or supplied password. No public exploit identified at time of analysis beyond the researcher's POC referenced in the patch commit.
Entropy source compromise in Data::Entropy (Perl) before version 0.010 allows any on-path network attacker to fully control the random bytes returned by rand_bits and rand_int for every application that selects the RandomOrg or RandomnumbersInfo remote sources via with_entropy_source. The library fetches entropy over plain HTTP with no transport security, and Data::Entropy::RawSource::RandomOrg's integrity check trivially accepts any non-empty byte string, so substituted attacker-controlled responses are never rejected. A vendor-released patch is confirmed at version 0.010; no active exploitation is confirmed (EPSS 0.10%, 1st percentile, not in CISA KEV), but the deterministic nature of the flaw means any secret derived from these sources is entirely known to the attacker.
Cross-tenant credential theft in pgAdmin 4 (9.0 through 9.16) lets an authenticated non-owner user hijack another user's shared server connection by abusing the Workspaces adhoc-connect endpoint (/misc/workspace/adhoc_connect_server), which clones a target server row verbatim - including its owner's stored database password and tunnel password - and reassigns ownership to the caller. The victim is typically an administrator whose shared server carries saved credentials, and the attacker gains use of those database credentials and any privileges they confer. There is no public exploit identified at time of analysis and it is not on CISA KEV, but the fix (pgAdmin 4 9.17) and the underlying commit are public, and EPSS was not provided.
SQL injection in pgAdmin 4 through an incomplete remediation of CVE-2026-12044 allows a low-privileged authenticated user to execute arbitrary SQL in a higher-privileged victim's database session. By creating a PostgreSQL object - table, index, publication, or subscription - with an apostrophe in its name (a character PostgreSQL legally permits in quoted identifiers), the attacker plants a payload that fires the moment any other pgAdmin user opens the object's Statistics or Dependencies tab, triggering raw Jinja2 template interpolation in coll_stats.sql, dependencies.sql, and get_position.sql without the qtLiteral escaping applied by the prior fix. The cross-user injection escalates impact well beyond the attacker's own database privileges, with the injected SQL executing under the victim's session credentials. No public exploit has been identified at time of analysis, and this CVE has not been added to the CISA KEV catalog.
Denial of service in gnome-remote-desktop as shipped in Red Hat Enterprise Linux 8, 9, and 10 lets an unauthenticated remote attacker exhaust server resources when the daemon runs in system mode with RDP enabled. Because the incoming connection handler bypasses the connection throttler, an attacker can open large numbers of parallel pre-authentication RDP connections, accumulating accepted sockets and pending routing-token operations until timeout and blocking legitimate RDP logins. There is no public exploit identified at time of analysis, and the flaw is specific to Red Hat's build - it does not affect the upstream project.
Deploy-time policy bypass in Red Hat Advanced Cluster Security for Kubernetes 4 lets a user with Kubernetes Deployment-create permission spoof workload identity by setting the openshift.io/encoded-deployment-config label to "null", forcing ACS to record the workload with empty UID, name and labels and a "default" namespace. This evades ACS deploy-time policy detection and enforcement visibility and corrupts Central's persistence, violation reporting and compliance correlation. No public exploit identified at time of analysis and the issue is not listed in CISA KEV.
Denial of service in the aws-smithy-json Rust runtime crate (versions before 0.62.7) lets remote unauthenticated attackers abort a process via stack exhaustion. Because the smithy-rs code generator wires this crate's deserializer into every generated struct, any smithy-rs generated server that parses JSON is exposed, and a single small HTTP request carrying deeply nested JSON triggers uncontrolled recursion in the unknown-key skip path. No public exploit was identified at time of analysis and the flaw is not in CISA KEV, but the trigger is trivial and a vendor patch is available.
Arbitrary code execution in the Banks Python LLM-templating library (versions prior to 2.4.3) allows an attacker who can influence Tool JSON definitions within a {% completion %} block to run any importable Python callable in the host process. Banks resolves the attacker-controlled import_path field via importlib.import_module() plus getattr() with no allowlist, so values like os.system or subprocess.getoutput become invokable when the LLM emits a matching tool_calls entry, passing decoded arguments as kwargs. No public exploit identified at time of analysis; this bug is explicitly distinct from and unfixed by the 2.4.2 SandboxedEnvironment patch (GHSA-gphh-9q3h-jgpp / CVE-2026-44209) because the malicious payload is plain text in the rendered body and never touches Jinja attribute access.
Arbitrary file disclosure in the Banks Python prompt-templating library (masci/banks) before 2.4.4 lets attackers who control template variables read any file the Python process can access. All four media filters (image, audio, video, document) wrap untrusted input in Path(value) and feed it straight to open(file_path, "rb") with no sanitization, so path traversal (../) reads files like .env, SSH keys, cloud credentials, source, /etc/passwd and /etc/shadow, returning them base64-encoded in the rendered prompt for trivial exfiltration. No public exploit identified at time of analysis and it is not in CISA KEV, but the fix in 2.4.4 is straightforward to reverse-engineer.
Confidential Active Directory attribute disclosure in Samba's Active Directory Domain Controller allows an authenticated low-privilege domain user to read protected directory attributes by abusing how LDAP Compare requests are handled. Because Samba fails to validate user-supplied attribute names (LDAP filter injection, CWE-90) and runs the resulting internal search in a trusted context, normal ACL enforcement is bypassed; in gMSA-enabled deployments this permits extraction of msKds-RootKeyData and offline derivation of Group Managed Service Account passwords, which can escalate to full domain compromise. No public exploit identified at time of analysis and the issue is not listed in CISA KEV, but a vendor patch is available.
Denial of service in IBM Enterprise Build of Quarkus (versions 3.27.1 through 3.27.4.SP2 and 3.33.1 through 3.33.2.SP2) allows a remote, unauthenticated attacker to exhaust memory by sending multipart requests whose MIME part-headers are accumulated without bound in the Quarkus REST layer. Because each part-header buffer grows without a size cap, an attacker can force excessive heap allocation and crash or hang the service, degrading availability for all users. There is no public exploit identified at time of analysis, but the flaw is trivially reachable (AV:N/AC:L/PR:N/UI:N) against any endpoint that accepts multipart input.
Service-account token disclosure in the Red Hat OpenShift koku-metrics-operator lets a user with edit rights on the CostManagementMetricsConfig custom resource redirect metrics uploads to an attacker-controlled URL, to which the operator blindly attaches its own Kubernetes bearer token. This CWE-918 server-side request forgery (tagged SSRF/Kubernetes/Red Hat) yields the operator's service-account credentials, enabling in-cluster privilege escalation. Reported by Red Hat with a CVSS 3.1 base of 7.6 (scope-changed); no public exploit identified at time of analysis and it is not in CISA KEV.
Out-of-bounds write in PHP's BCMath bccomp() function allows attacker-controlled inputs to corrupt stack and heap memory in PHP 8.4.x before 8.4.24 and 8.5.x before 8.5.9, with high confidentiality, integrity, and availability impact per the vendor CVSS 4.0 score of 8.1. Exploitation requires an application to pass attacker-influenced values into bccomp() with the BCMath extension enabled; there is no public exploit identified at time of analysis (CVSS E:U, no KEV listing). Memory corruption of this class can lead to crashes and, potentially, arbitrary code execution within the PHP process.
SQL injection in the PHP interpreter's string-escaping routine affects PHP 8.2.x before 8.2.33, 8.3.x before 8.3.33, 8.4.x before 8.4.24, and 8.5.x before 8.5.9, where backslashes in attacker-controlled parameters are improperly escaped, allowing malicious input to break out of quoted SQL string literals. Any web application relying on PHP's built-in escaping to sanitize user data before database queries can be subverted, enabling trivial injection. This is a CWE-89 flaw reported by the PHP project (advisory GHSA-7qpv-r5mr-78m4); no public exploit identified at time of analysis, and CVSS 4.0 rates it 8.1 with exploitation marked automatable.
Account takeover in osTicket (the open-source support ticketing system) before v1.17.8 and v1.18.4 stems from an inverted logic check (CWE-640) in the password-reset token validation routine, where the expiry check runs only when the token's timestamp lookup fails - so any token with a valid timestamp bypasses expiration entirely. An attacker who obtains a valid reset token can replay it after it should have expired to hijack the associated account. Rated CVSS 4.0 9.1 (Critical); no public exploit identified at time of analysis and not listed in CISA KEV.
Server-side request forgery in the Eclipse CSI PIA `POST /v1/upload/sbom` endpoint lets remote attackers coerce the server into connecting to an arbitrary host, port, and scheme of their choosing. By exploiting a parser-differential between Python's `urlparse` (used for the issuer allowlist check) and `requests`/`urllib3`/`urllib` (used for OIDC discovery and JWKS retrieval), an attacker crafts a JWT `iss` claim containing a backslash that passes validation but resolves to a different host during the actual fetch. The `iss` claim is only checked in its unverified form, so no legitimate token is needed; no public exploit has been identified at time of analysis and the issue is not listed in CISA KEV.
Improper input neutralization in Phoenix Contact CHARX SEC electric-vehicle charging controllers lets a remote, unauthenticated attacker inject malicious data through MQTT that the bundled ModbusServer application processes without validation, corrupting Modbus data and disrupting service. All CHARX SEC-3000/3050/3100/3150 units running firmware 1.0.0 through 1.9.0 are affected, with impact limited to integrity and availability (no data disclosure). No public exploit identified at time of analysis; the flaw was coordinated-disclosure reported through CERT@VDE and a fixed firmware (1.9.1) is available.
Memory-corruption remote code execution in Google Chrome's Dawn WebGPU graphics component affects all Desktop builds prior to 151.0.7922.72, where a crafted HTML page triggers a use-after-free (CWE-416) that lets a remote attacker run arbitrary code - though confined inside the renderer sandbox, which is why Chromium rated it Low despite the CVSS 8.8. Exploitation requires only that a victim load a malicious page (UI:R), with no authentication. There is no public exploit identified at time of analysis and the CVE is not on CISA KEV.
Remote code execution in Google Chrome's PDFium component (versions prior to 151.0.7922.72) allows a remote attacker to run arbitrary code - confined to the renderer sandbox - when a victim opens a maliciously crafted PDF. The flaw is a CWE-416 use-after-free reported by Google's own Chrome team and fixed in the July 2026 Stable channel update; Chromium rated its intrinsic severity Low because execution is contained within the sandbox, though NVD's CVSS scores it 8.8. There is no public exploit identified at time of analysis and it is not listed in CISA KEV.
Local privilege escalation in the Google Chrome Updater on Windows (versions prior to 151.0.7922.72) lets a low-privileged local attacker win a race condition and substitute a malicious file to elevate privileges. The flaw carries a CVSS 7.0 score, and while Chromium rates its security severity as Low, the SSVC technical impact is Total. There is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Arbitrary code execution in Google Chrome's V8 JavaScript engine allows a remote attacker to run code inside the renderer sandbox when a victim visits a crafted HTML page, affecting all desktop builds prior to 151.0.7922.72. The flaw is a CWE-843 type confusion in V8 and carries a CVSS 8.8 due to network reach and high CIA impact, though execution is confined to the sandbox and Chromium rates the security severity as Low. There is no public exploit identified at time of analysis and it is not listed in CISA KEV, but a vendor patch is available.
Remote code execution in Google Chrome's V8 JavaScript engine (all versions prior to 151.0.7922.72) allows a remote attacker who lures a victim to a crafted HTML page to execute arbitrary code, though confined inside the renderer sandbox rather than achieving a full sandbox escape. Google rated the Chromium security severity as Low, and there is no public exploit identified at time of analysis and no evidence of active exploitation (CISA SSVC exploitation status: none). Exploitation hinges on winning a timing race, so despite the 7.5 CVSS score it is regarded as difficult to trigger reliably.
Remote code execution within the browser's renderer sandbox affects Google Chrome desktop builds prior to 151.0.7922.72, stemming from an inappropriate implementation in the Passwords component (CWE-269). A remote attacker who lures a victim to a crafted HTML page can execute arbitrary code, but execution is confined to the sandbox - which is why Chromium rated the underlying issue Low severity despite the CVSS 8.8. No public exploit identified at time of analysis, and the fix is already shipping in the Stable channel.
Sandboxed remote code execution in Google Chrome desktop before 151.0.7922.72 lets a remote attacker run arbitrary code within the renderer sandbox when a victim opens a crafted HTML page. The flaw stems from an inappropriate implementation in Chrome's Scheduling component; Google rated it Chromium-Low severity and no public exploit has been identified, though the NVD CVSS of 8.8 reflects the high per-system impact once triggered. Exploitation requires user interaction (visiting a malicious page) and, because code executes inside the sandbox rather than escaping it, real-world impact is constrained absent a chained sandbox-escape bug.
Sandbox-constrained arbitrary code execution in Google Chrome's V8 JavaScript engine affects all desktop builds prior to 151.0.7922.72, where a crafted Chrome extension abuses an inappropriate V8 implementation to run attacker code inside the renderer sandbox. Exploitation requires convincing a user to install a malicious extension, and Google internally rated the Chromium security severity as Low. There is no public exploit identified at time of analysis and CISA SSVC records exploitation status as 'none'.
Remote code execution in Google Chrome on macOS prior to 151.0.7922.72 stems from an inappropriate implementation in the Safe Browsing component, where a remote attacker can trigger arbitrary code execution by luring a victim into opening or downloading a malicious file. The CVSS 3.1 vector (AV:N/AC:L/PR:N/UI:R) requires no privileges but does require user interaction, and Google has shipped a fixed Stable channel build. Despite the 8.8 CVSS, Chromium rated the underlying issue Low severity, EPSS is only 0.20% (10th percentile), and there is no public exploit identified at time of analysis.
Cross-user information disclosure in undici's cache interceptor allows authenticated users' responses to be served from a shared cache to subsequent callers, including unauthenticated ones, when the upstream origin sends qualified `no-cache` or `private` Cache-Control directives with optional whitespace (OWS) around the `=` delimiter. Affected deployments are those explicitly running undici 7.0.0-7.28.x or 8.0.0-8.8.x with `interceptors.cache()` in shared mode while forwarding `Authorization` headers upstream. This is a bypass of the earlier CVE-2026-9678 fix; a public proof-of-concept exists and vendor-released patches are available in undici 7.29.0 and 8.9.0.
Server-side request forgery allowlist bypass in the V programming language standard library (net.urllib / net.http) through version 0.5.2 lets remote unauthenticated attackers reach internal network services that a host-based allowlist was meant to block. The flaw stems from a parser differential: net.urllib.parse() and net.http.get() disagree on how a backslash in a URL's authority is handled, so validation and connection resolve to different hosts. Publicly available exploit code exists and the issue was reported by VulnCheck, though it is not listed in CISA KEV (no public evidence of active exploitation at time of analysis).
Cross-user information disclosure and a parse-time denial of service in undici's HTTP cache interceptor affect versions 7.0.0-7.28.x and 8.0.0-8.8.x when the interceptor runs in shared mode (the default). A response bearing a degenerate qualified private directive (e.g. private="") slips past the shared-cache guard and is later served to a different caller sharing the same cache key, leaking private bodies and headers including Set-Cookie; separately, mixing an unqualified and a qualified private directive throws an uncaught TypeError that can crash the process. No public exploit identified at time of analysis, and EPSS is low (0.27%), but the fix is confirmed in undici 7.29.0 and 8.9.0.
Denial of service in open-iscsi's bundled open-isns library (iSNS implementation) lets an unauthenticated man-in-the-middle attacker crash the iSNS client/server by triggering a double free in the attribute-decoding routines. The flaw stems from freed pointers not being nulled in isns_attr_type_string_decode and isns_attr_type_opaque_decode, so a crafted iSNS message causes the same allocation to be freed twice. There is no public exploit identified at time of analysis and it is not listed in CISA KEV; impact is limited to availability, with no confidentiality or integrity effect.
Denial of service in the cJSON C library (versions through 1.7.19) arises from exponential algorithmic complexity in the cJSON_Compare() function, which recursively descends into each shared object subtree twice per level with no depth guard. Applications that call cJSON_Compare() on attacker-influenced JSON that is structurally equal to a reference document can be forced to burn hours of CPU on a document of only a few hundred bytes (nesting depth around 40), with cost roughly doubling per added level. No public exploit identified at time of analysis and the issue is not on CISA KEV; impact is compute exhaustion only, with no confidentiality or integrity effect.
Denial of service in cJSON (DaveGamble's C JSON parser) through version 1.7.19 lets a remote attacker crash any application that applies an untrusted RFC 6902 JSON Patch through cJSONUtils_ApplyPatches() or cJSONUtils_ApplyPatchesCaseSensitive(). By combining add and copy operations that graft duplicated subtrees, an attacker inflates document nesting far beyond the parser's 1000-level limit, and the unbounded recursion in cJSON_Delete() (plus a permissive CJSON_CIRCULAR_LIMIT of 10000 in cJSON_Duplicate()) exhausts the thread stack. Reported by VulnCheck with a public technical write-up; there is no public exploit identified at time of analysis and it is not on CISA KEV.
Local privilege escalation in open-iscsi's iscsiuio component lets any unprivileged local user connect to the iscsiuio control socket and issue commands normally restricted to root, because the daemon validated peer credentials against the wrong file descriptor. The flaw (CWE-863 Incorrect Authorization) was reported by SUSE and fixed upstream in commit 668ca1df; there is no public exploit identified at time of analysis, and it is not listed in CISA KEV.
Out-of-bounds write in GNU binutils' BFD library (bfd/elf32-dlx.c) lets a crafted ELF/DLX object file corrupt memory and achieve arbitrary code execution when parsed by any BFD-consuming tool (objdump, readelf, strip, ld, nm, objcopy). Reported by Red Hat and tracked against RHEL 6-10, Hardened Images and OpenShift 4, the flaw is only reachable when binutils is compiled with the DLX backend (typically --enable-targets=all), so exposure is gated by a non-default build option. There is no public exploit identified at time of analysis (not in CISA KEV), though the researcher demonstrated code execution via a File Stream Oriented Programming (FSOP) attack on glibc FILE structures.
SQL injection in ClickHouse Server versions 26.3.9.8 and earlier lets remote attackers inject arbitrary SQL through the CREATE DICTIONARY functionality, which the advisory claims can escalate to arbitrary code execution. Publicly available exploit code exists (GitHub repository CVE-2026-51992), but there is no CISA KEV listing and no EPSS score in the provided data, so no confirmed active exploitation. The provided CVSS is 9.1, though the description's reference to the CREATE DICTIONARY operation suggests an authenticated privilege may actually be required, which conflicts with the PR:N in the supplied vector.
Privilege escalation in CRIU allows a malicious container process to spoof credentials during checkpoint, resulting in elevated capabilities upon restore. On Red Hat platforms, effective exploitation is mitigated by mandatory access controls, user namespaces, and the requirement for privileged checkpoint initiation. No public exploit identified at time of analysis.
Remote code execution in Apache Axis2/Java through 2.0.0 arises from unsafe deserialization of Java objects in the optional Tribes clustering component, allowing unauthenticated remote attackers with network access to the clustering port to execute arbitrary code via crafted serialized payloads. The vulnerability is only exploitable when clustering is explicitly enabled, which is off by default. No active exploitation or public exploit code has been observed; the issue is resolved by upgrading to version 2.0.1, which removes the entire clustering feature.
Use-after-free vulnerability in Safari’s web content processing allows remote attackers to execute arbitrary code on affected Apple devices when a user visits a malicious website. Affected versions include Safari prior to 26.6, iOS and iPadOS prior to 26.6, macOS Tahoe prior to 26.6, visionOS prior to 26.6, and watchOS prior to 26.6. No public exploit or active exploitation has been identified, and EPSS indicates a low probability of exploitation (0.13%).
Memory corruption in Apple Safari 26.x allows remote code execution via processing maliciously crafted web content. This buffer overflow (CWE-119) affects all Apple platforms—iOS, iPadOS, macOS, visionOS, and watchOS—requiring user interaction to visit a malicious webpage. While the CVSS score is critical (8.8), EPSS (0.13%) indicates low current exploitation probability and no KEV listing or public exploit exists.
Information disclosure in Apple Safari and OS-integrated WebKit allows websites to detect if a user has visited specific URLs. Safari, iOS, iPadOS, macOS, tvOS, visionOS, and watchOS prior to version 26.6 are affected. No public exploit code or active attacks have been reported, and the probability of exploitation is very low (EPSS 0.17%).
Out-of-bounds read in Apple Safari's web content processing can crash the browser and potentially leak sensitive memory contents. This vulnerability affects all Apple platforms where Safari rendering is used and is fixed in Safari 26.6, iOS/iPadOS 26.6, macOS Tahoe 26.6, tvOS 26.6, visionOS 26.6, and watchOS 26.6. Exploitation requires user interaction to visit a malicious site; no known active exploits or public proof-of-concept exist.
Information disclosure and denial of service in GIMP's file-icns plugin allow local attackers to leak heap memory contents via a crafted ICNS image file. When processing a truncated mask resource, the plugin performs an out-of-bounds read, leaking heap data as alpha channel pixel values or crashing the application. User interaction is required to open the malicious file.
Arbitrary code execution in GIMP when opening a crafted FITS file due to an integer overflow in the file-fits plugin that leads to a heap buffer overflow. An attacker can achieve memory corruption and potentially execute arbitrary code with the user's privileges. No public exploit or active exploitation has been identified at time of analysis.
Reflected cross-site scripting in XMLRPC-C Library versions 1.07 through 1.67.01 allows unauthenticated attackers to execute arbitrary JavaScript when a victim clicks a crafted link, as the error page component reflects unsanitized user input. This can lead to theft of sensitive session data or other browser-accessible information from affected services. No public exploit code or active exploitation has been reported at this time.
Heap-based buffer overflow in LibRaw 0.21 allows remote code execution when a user opens a maliciously crafted RAW image file. The flaw resides in the stretch() and fuji_rotate() functions and requires user interaction, enabling a complete compromise of confidentiality, integrity, and availability. Proof-of-concept code exists, but no active exploitation has been confirmed.
Memory corruption in SQLite 3.41's ORDER BY parsing allows remote attackers to crash the database process or leak sensitive data via crafted SQL queries. A use-after-free in ExprList handling, triggered by an ORDER BY clause with many items, can also lead to arbitrary code execution under specific memory conditions. A public advisory exists, but no active exploitation (KEV) or EPSS data is available.
Use-after-free in SQLite 3.41 allows remote attackers supplying crafted SQL to trigger memory corruption during parsing, leading to denial of service, sensitive information leakage, and potential arbitrary code execution.
Use-after-free in SQLite 3.41's JSON parser allows remote code execution. A remote attacker can deliver a crafted JSON payload; if a user or application processes it with SQLite, the vulnerability triggers memory corruption, potentially enabling full system compromise. A proof-of-concept advisory is publicly available, though no active exploitation has been confirmed.
Remote denial of service and heap memory leak in SQLite 3.41 via the json_remove() function allows attackers with the ability to execute SQL queries to crash the database process or leak sensitive heap data. A use-after-free in jsonRemoveFunc is triggered by malformed JSON and a deep path parameter; publicly available proof-of-concept code exists, and low EPSS risk indicates limited exploitation likelihood.
Use-after-free in SQLite 3.41's expression-evaluation path lets an attacker who can submit a crafted SQL statement crash the process, leak adjacent memory, or potentially achieve code execution. The flaw stems from sqlite3ReleaseTempReg() prematurely freeing a temporary register that exprComputeOperands() then dereferences. Rated CVSS 7.5 (AV:N/AC:H/PR:L); publicly available exploit code exists per the CISA SSVC 'poc' determination, but it is not on the CISA KEV list and EPSS is low (0.24%, 16th percentile), indicating no evidence of widespread active exploitation.
Use-after-free in SQLite 3.41 expression parsing leads to memory corruption, allowing remote attackers to crash applications and leak sensitive memory contents via crafted SQL queries. Public proof-of-concept exists but there is no evidence of active exploitation. The vulnerability is triggered by sending malicious SQL statements that cause the parser to access a freed expression object.
State-counter underflow in the Linux kernel's mlx5e (NVIDIA/Mellanox ConnectX) IPsec offload driver lets the eswitch mode-block reference counter (esw->offloads.num_block_mode) be decremented without a matching increment when a temporary acquire-flow security association is installed. The affected code is the mlx5e_xfrm_add_state() acquire-SA path, which skips hardware offload but still falls through to the common success label that unconditionally calls mlx5_eswitch_unblock_mode(). Fixed upstream in the stable kernel series; no public exploit identified at time of analysis and EPSS probability is low (0.21%).
Local privilege escalation in the Linux kernel's ARM FFA firmware interface allows an unprivileged local attacker to achieve arbitrary code execution by triggering an out-of-bounds write when processing PARTITION_INFO_GET descriptors. The kernel fails to validate firmware-supplied indices, enabling a heap buffer overflow that can corrupt memory. Patches are available, no public exploit exists, and EPSS indicates low exploitation probability (0.21%).
Heap-based information disclosure and denial of service in the Linux kernel's mac80211 WiFi subsystem can be triggered by adjacent network attackers. The vulnerability arises from incorrect pointer usage during MLE defragmentation, potentially leading to kernel heap memory over-read and exposure of sensitive data, as well as system crashes. Affected kernels from version 6.9 are vulnerable until patched; no public exploit is known and EPSS indicates a low exploitation probability (0.21%).
Use-after-free in the Linux kernel's usbio (USB I/O bridge) driver allows a local, low-privileged attacker to corrupt kernel slab memory during USB device disconnect, potentially achieving privilege escalation or triggering a kernel panic. The bug lies in usbio_disconnect(), which iterates the client list in reverse using list_for_each_entry_reverse() - a macro that reads client->link.prev after the loop body, but auxiliary_device_uninit() can free the current client before that read, dereferencing freed slab memory. Patches are confirmed in Linux stable branches 6.18.39, 7.1.4, and 7.2-rc3; no public exploit or active exploitation is identified at time of analysis.
Heap buffer overflow in the Linux kernel's rtl8723bs staging WiFi driver (rtw_cfg80211_set_wpa_ie()) allows a local low-privileged user to corrupt kernel memory by one byte via a crafted nl80211 connect request, enabling potential privilege escalation to kernel context. The 256-byte supplicant_ie array in struct security_priv overflows when a WPA IE with length 255 is supplied, writing 257 bytes and clobbering the adjacent last_mic_err_time field; a flawed length check in rtw_parse_wpa_ie() using u8 narrowing silently permits the overflow. Patches are available across all active stable kernels; EPSS is 0.23% and no public exploit or CISA KEV listing exists at time of analysis.
Use-after-free in the Linux kernel Bluetooth MGMT subsystem allows a local attacker with low privileges to trigger kernel memory corruption via a race condition between add_device and remove_device operations. The flaw exists in add_device_complete(), which dereferences a hci_conn_params pointer without holding hci_dev_lock, while a concurrent MGMT_OP_REMOVE_DEVICE path frees the same object under that lock. KASAN confirms exploitation as slab-use-after-free; no public exploit or active exploitation (CISA KEV) has been identified, and patches have been released across multiple stable branches.
Use-after-free memory corruption in the Linux kernel's netpoll subsystem allows a local attacker or concurrent kernel path to access freed memory during the netpoll shutdown sequence. The flaw exists because cancel_delayed_work() - called from softirq context during __netpoll_cleanup() - does not block until the TX worker finishes, permitting queue_process() to dereference a kfree'd npinfo struct via container_of(). KASAN confirms the UAF; patches are available across multiple stable kernel branches, and no public exploit or KEV listing exists at time of analysis.
Use-after-free memory corruption in the Linux kernel's NXP i.MX8 ISI (Image Sensing Interface) media driver is triggered during module removal (rmmod of imx8_isi), confirmed by KASAN slab-use-after-free reports. The root cause is an incorrect teardown order in mxc_isi_remove(): the crossbar cleanup frees media entity pads before the V4L2 cleanup has finished removing media links that reference those same pads, resulting in a dangling pointer dereference in __media_entity_remove_link(). No public exploit has been identified at time of analysis, and with EPSS at 0.21% (11th percentile), real-world weaponization appears unlikely despite the CVSS 7.8 score.
Out-of-bounds memory read in the Linux kernel's btnxpuart Bluetooth driver exposes kernel memory to a malicious or compromised NXP Bluetooth UART controller. The vulnerability exists in nxp_recv_fw_req_v3(), which handles firmware download requests during v3 controller initialization: when the controller supplies a crafted 32-bit offset or 16-bit length value that pushes past the end of the firmware image, the driver reads arbitrary kernel memory and transmits it back to the controller over UART. The parallel v1 path (nxp_recv_fw_req_v1()) already contained the equivalent bounds check; the v3 path lacked it. No active exploitation is confirmed (not in CISA KEV) and EPSS is 0.21% (11th percentile), consistent with the narrow local/hardware attack surface.
Memory leak in the Linux kernel's SMB client encrypted receive path (receive_encrypted_standard()) allows a local low-privileged user or a cooperative malicious SMB server to exhaust kernel heap memory, with potential for heap grooming-based privilege escalation. The bug was introduced with commit b24df3e30cbf and affects multiple stable kernel series from 4.19 onward. No public exploit code exists and EPSS is 0.22%, but the kernel team's classification of CWE-787 and a C:H/I:H/A:H CVSS score reflect the Linux security team's assessment that the leaked kernel buffer is exploitable beyond simple resource exhaustion.
Double-free memory corruption in the Linux kernel's Qualcomm CPU frequency hardware driver (qcom-cpufreq-hw) allows a local low-privileged attacker on Qualcomm SoC-based systems to trigger kernel heap corruption, potentially escalating privileges to root. The flaw originates in qcom_cpufreq_hw_cpu_exit(), which incorrectly calls kfree() on memory allocated via devm_kzalloc() - memory already tracked by the kernel's device resource management (devres) system. This results in a double-free when the platform device is subsequently detached and devres performs its own cleanup. No public exploit code exists and EPSS is low (0.21%), consistent with a static-analysis discovery rather than active exploitation.
Use-After-Free in the Linux kernel HID letsketch driver allows local exploitation during device unbind, potentially enabling kernel memory corruption and privilege escalation on affected systems. The letsketch_probe() function arms a per-device kernel timer (inrange_timer) with a 100 ms timeout, but no .remove callback exists to drain the timer before devm cleanup frees letsketch_data and input_dev structures on USB unplug or rmmod, producing a UAF read/write into freed kernel memory. Patches are backported across six stable kernel branches; no public exploit code exists and EPSS is 0.21% (12th percentile), reflecting the niche device requirement and race-condition complexity.
Use-after-free in the Linux kernel HID appleir driver allows a local attacker to corrupt kernel memory during device removal, potentially achieving privilege escalation. The race condition exists between the key_up_timer callback (armed by appleir_raw_event() on each keydown/key-repeat) and hid_hw_stop(), which frees the underlying input_dev before the timer is drained. A second race window, where a USB URB completion during hid_hw_stop() can re-arm the timer via mod_timer(), means a simple teardown reorder is insufficient — the fix introduces a spinlock-gated 'removing' flag combined with timer_shutdown_sync() to permanently disable further callbacks. No public exploit has been identified at time of analysis, and EPSS is 0.22% (13th percentile).
Use-after-free in the Linux kernel HID lg-g15 driver allows a local attacker to trigger kernel memory corruption by racing device input events against USB keyboard removal. The driver allocates lg_g15_data via devm and schedules work items from HID report handlers (backlight cycle key on G15, G15 v2, G510), but lacked a remove callback to cancel pending work before devres freed the allocation. On kernel versions from approximately 5.5 through the unfixed stable branches, unplugging the keyboard while a work item is pending causes the worker to dereference freed memory via container_of, potentially enabling local privilege escalation. No public exploit has been identified and EPSS is 0.22% (13th percentile), indicating no observed widespread exploitation.
Incorrect RCU lock annotations in the Linux kernel's BPF LPM trie map implementation (`kernel/bpf/lpm_trie.c`) cause lockdep false positives on debug kernels when sleepable BPF programs access `BPF_MAP_TYPE_LPM_TRIE` maps via lookup, update, or delete operations. The mismatch arises because sleepable BPF programs hold `rcu_read_lock_trace()` while the trie code asserts `rcu_read_lock_bh_held()` or `rcu_read_lock_held()`, producing repeated 'suspicious RCU usage' kernel warnings per distinct callsite. The Linux kernel developers explicitly state no actual use-after-free occurs, since Tasks Trace RCU already serializes against trie node reclaim; however, the incorrect annotation undermines lockdep's ability to detect future regressions and is increasingly impactful as sleepable BPF LSM programs become common in production security tooling. No public exploit exists and no CISA KEV listing has been issued.
Use-after-free and memory leak in the Linux kernel's EHCI and FOTG210 USB isochronous submit paths expose systems running kernel versions from 2.6.15 through 7.1.3 to local kernel memory corruption by a low-privileged user. When an isochronous URB submission fails before the staged schedule handoff - either because the host controller becomes inaccessible or usb_hcd_link_urb_to_ep() returns an error - the ehci_iso_sched allocation is not freed and urb->hcpriv retains a dangling pointer, creating a UAF condition. No public exploit exists and the EPSS score is 0.21% (12th percentile), but vendor-released patches across multiple stable branches are available and should be applied to eliminate the risk.
Use-after-free in the Linux kernel USB gadget UDC subsystem exposes systems with DWC3-based USB controllers to a race condition between usb_del_gadget() and gadget_match_driver(), enabling a NULL pointer dereference that can crash the kernel or enable privilege escalation on unpatched systems. The flaw affects all major Linux stable branches from 5.15.x through 7.1.x prior to the patched releases, and is particularly relevant for embedded Linux, Android-based devices, and IoT hardware that use USB gadget/device mode. No public exploit has been identified at time of analysis, and EPSS remains at 0.21% (12th percentile), reflecting current low exploitation interest.
Use-after-free in the Linux kernel's idmouse USB fingerprint driver allows a low-privileged local attacker to corrupt kernel memory by racing a device disconnect against a file-descriptor release, potentially enabling local privilege escalation to root. The bug stems from `mutex_unlock()` accessing the freed driver-data structure after the lock is released, a documented non-atomic behavior; the fix introduces a `kref` reference counter to correctly manage object lifetime. Patches are available across all active stable branches; no active exploitation or public exploit has been reported.