CVE & CISA-KEV Catalog

A flaw was found in Cockpit. An unauthenticated remote attacker can exploit this vulnerability by initiating and sustaining numerous simultaneous connections to the `cockpit-tls` service. This forces the service to create an unbounded number of detached threads, consuming system resources such as memory and file descriptors. The primary consequence is a denial of service (DoS), leading to degradation or complete unavailability of the Cockpit service for legitimate users.

Published 2026-09-18

SGLang through 0.5.19 in prefill/decode disaggregation mode with Mooncake KV transfer backend fails to validate bootstrap_room values, allowing unbounded transfer state allocation. Unauthenticated attackers can reach the decode engine's POST /generate endpoint and submit arbitrary bootstrap_room values to exhaust prefill process memory until out-of-memory termination.

Published 2026-09-18

libheif is a HEIF and AVIF file format decoder and encoder. In 1.23.1 and earlier, crafted grid, iovl, and iden reference graphs can repeatedly decode the same base image because processed_ids is copied per branch and ImageItem::decode_image() has no shared operation budget. This vulnerability is fixed in 1.23.2.

Published 2026-09-18

A flaw was found in the DERDecoder class within wildfly-elytron-asn1. A remote attacker can exploit this resource exhaustion vulnerability by sending a specially crafted DER (Distinguished Encoding Rules) payload. The decoder attempts to allocate excessive memory based on an inflated length value without proper validation, leading to Java Virtual Machine (JVM) memory exhaustion. This results in a remote Denial of Service (DoS) for services that process untrusted DER/ASN.1 input, including SASL (Simple Authentication and Security Layer) authentication mechanisms and X.500 certificate principal parsing paths.

Published 2026-09-18

A flaw was found in Netty's HttpServerCodec. A remote, unauthenticated attacker can exploit this vulnerability by pipelining HTTP/1.1 requests on a single connection and withholding reads. This action causes the methodOverflowQueue to grow without limit, leading to unbounded heap memory consumption and a denial of service due to memory exhaustion.

Published 2026-09-18

A flaw was found in Netty. SpdySessionHandler accepts an unlimited number of concurrent remote-initiated streams because localConcurrentStreams defaults to Integer.MAX_VALUE and the handler provides no API to change it. A remote peer can open a SPDY connection and send a large number of SYN_STREAM frames with FLAG_FIN=0, causing unbounded heap and direct memory allocation that can lead to JVM OutOfMemoryError and a denial of service.

Published 2026-09-18

## Summary `RedisArrayAggregator` recently added `maxElements` and `maxNestedArrayDepth` limits to fix public Redis resource-exhaustion advisories. The limits are independent, but the allocator remains eager: every positive nested RESP array header creates `new ArrayList<RedisMessage>(length)` before any child element exists. With the default constructor, an attacker can send nested array headers with length `1,000,000` until the default nesting limit of `1024` is reached. This can reserve up to `1,024,000,000` child slots from roughly 12 KB of RESP input. This is backing capacity, not logical list size: `ArrayList(int)` constructs an empty list with the specified initial capacity. ## Technical Details Current `decodeRedisArrayHeader(...)` checks the two limits independently:

Published 2026-09-18

A vulnerability was identified in O-RAN-SC SMO OAM 2025-06-10. This affects an unknown part of the component VES Collector. The manipulation leads to allocation of resources. Remote exploitation of the attack is possible. The exploit is publicly available and might be used. The project was informed of the problem early through a bug report but has not responded yet.

Published 2026-09-18

A vulnerability was determined in O-RAN-SC SMO OAM 2025-06-10. Affected by this issue is some unknown functionality of the component VES Collector. Executing a manipulation can lead to allocation of resources. The attack may be launched remotely. The exploit has been publicly disclosed and may be utilized. The project was informed of the problem early through a bug report but has not responded yet.

Published 2026-09-18

A vulnerability was found in O-RAN-SC SMO OAM 2025-06-10. Affected by this vulnerability is an unknown functionality of the component VES Collector. Performing a manipulation results in allocation of resources. The attack may be initiated remotely. The exploit has been made public and could be used. The project was informed of the problem early through a bug report but has not responded yet.

Published 2026-09-18

Caddy is an extensible server platform that uses TLS by default. In version 2.11.3 and earlier, three configuration-dependent weaknesses affect the handler and placeholder layer. In modules/caddyhttp/rewrite/rewrite.go, Rewrite.Rewrite() can pass attacker-controlled replacement bytes through buildQueryString for a second placeholder expansion when a rewrite URI ends with a literal question mark, allowing injected environment or request-variable placeholders to disclose data and, when the file provider is registered, allowing injected file placeholders to disclose readable files. The issue is fixed in version 2.11.4.

Published 2026-09-17

dd-trace-cpp is the Datadog distributed tracing library for C++. Prior to 2.1.0, dd-trace-cpp parses incoming W3C baggage headers without enforcing DD_TRACE_BAGGAGE_MAX_ITEMS or DD_TRACE_BAGGAGE_MAX_BYTES on the extraction path, even though those limits are enforced during injection. A remote unauthenticated attacker can send a header containing many comma-separated key-value pairs or one very large value, causing per-request hash-map allocation and unbounded CPU and memory consumption. Baggage extraction is enabled by default in most affected tracers unless baggage is removed from DD_TRACE_PROPAGATION_STYLE or DD_TRACE_PROPAGATION_STYLE_EXTRACT, so affected internet-facing services can be denied service. This issue is fixed in version 2.1.0.

Published 2026-09-17

The Datadog PHP Tracer provides application performance monitoring and distributed tracing for PHP. Prior to 1.19.2, ddtrace_deserialize_baggage in ext/distributed_tracing_headers.c parses incoming W3C baggage HTTP headers without enforcing DD_TRACE_BAGGAGE_MAX_ITEMS or DD_TRACE_BAGGAGE_MAX_BYTES. A remote unauthenticated client can send an arbitrarily large number of comma-separated key-value pairs or a single oversized value, causing the tracer to allocate hash-map entries and consume unbounded CPU and memory on each request. Baggage extraction is enabled by default in most affected deployments unless baggage is removed from DD_TRACE_PROPAGATION_STYLE or DD_TRACE_PROPAGATION_STYLE_EXTRACT. This issue is fixed in version 1.19.2.

Published 2026-09-17

Valhalla is an open source routing engine and accompanying libraries for use with OpenStreetMap data. In 3.7.0 and earlier, a POST request to /sources_to_targets containing an exclude_polygons ring formed by three collinear points can cause unbounded memory growth in the worker. The zero-area geometry, rather than the other request options, triggers processing in src/loki/polygon_search.cc until the process is terminated by the out-of-memory killer. A single unauthenticated request can therefore stop a public-facing worker. Other endpoints that accept exclude_polygons, including /route, were not verified as affected. No fixed version is available as of this review.

Published 2026-09-17

Pomerium is an identity and context-aware access proxy. Prior to 0.32.8, decodeQueryStringV2 in pkg/hpke/url.go performs zstd decompression of attacker-controlled data without an output-memory limit when DecryptURLValues processes HPKE V2 values for Stateless.Callback in internal/authenticateflow/stateless.go. In hosted or stateless authentication deployments, an unauthenticated attacker can obtain the receiver key from /.well-known/pomerium/hpke-public-key, provide a matching attacker-controlled sender key, and send a compressed payload to /.pomerium/callback that expands before validateSenderPublicKey rejects the sender. This can allocate hundreds of megabytes per request, exhaust proxy memory, crash or degrade the process, and block access to applications protected by the deployment. St

Published 2026-09-17

libp2p is a JavaScript implementation of the libp2p networking stack. Prior to 11.0.26, @libp2p/floodsub accepts unauthenticated RPC frames on /floodsub/1.0.0 through PeerStreams.attachInboundStream in packages/floodsub/src/peer-streams.ts without protobuf element limits, then processRpc and processRpcSubOpt in packages/floodsub/src/floodsub.ts synchronously process the subscriptions array without a per-frame cap. A single bounded-size frame can decode into millions of empty subscription entries that block the event loop, while hundreds of thousands of unique-topic SUBSCRIBE entries allocate PeerSet objects in this.topics that are not removed after peer removal or stop. Empty entries cause CPU exhaustion but do not grow this.topics; persistent memory growth requires unique topics. The subs

Published 2026-09-17

A flaw was found in quarkus-websockets-next. This vulnerability allows a remote attacker to cause a Denial of Service (DoS) by streaming messages over a single connection faster than the application can process them. Due to unbounded message buffering and a lack of read backpressure, this rapidly exhausts heap space, leading to a java.lang.OutOfMemoryError that crashes the Java Virtual Machine (JVM).

Published 2026-09-17

vm2 before 3.11.6 fails to enforce bufferAllocLimit on ArrayBuffer, SharedArrayBuffer, and TypedArray constructors, allowing attackers to allocate arbitrary host memory. Attackers can bypass the buffer allocation cap by using these V8 intrinsics to exhaust host process memory and trigger out-of-memory conditions.

Published 2026-09-17

WWBN AVideo through commit e01e41ecc (no patched version available) contains a broken access control flaw in objects/userVerifyEmail.php. The script disables the login requirement ($global['ignoreUserMustBeLoggedIn'] = 1), takes users_id directly from the query string, and calls User::sendVerificationLink() with no session requirement, no CSRF/global token, no relationship check between caller and target, and no enforceRateLimit() call. The only intended throttle is keyed to the caller's own session, so cookie-less requests are never limited. An unauthenticated remote attacker can therefore cause an arbitrary number of verification emails to be sent to any account ID, and can enumerate accounts and their verification status from the three distinct JSON responses ("Verification Sent", "Alre

Published 2026-09-17

Transient DOS when processing large or numerous request buffers without sufficient memory allocation validation.

Published 2026-09-17

Novel vulnerabilities to launch algorithmic complexity attacks on DNSSEC have been researched under the term 'ReTrap'. These result in degradation of service when malicious zones are used to serve the algorithmic complexity vulnerabilities. NLnet Labs Unbound up to and including 1.26.0 is vulnerable to some of them. TagTrap, where the triple(Zone, Algo, KeyTag) matching mechanism introduces a significant attack vector when resolvers handle malicious responses containing numerous mismatched DNSKEY, RRSIG, and DS record. DelegationTrap, where constructing the chain-of-trust requires iterative validation of DNSKEY and DS records from the root zone downward. For deeply nested domains, this results in significant computational overhead. NsecTrap, where responses with excessive invalid NSEC re

Published 2026-09-17

In NLnetLabs Unbound up to and including 1.26.0, a degradation of service vulnerability is present in the TCP/DoT reading procedure where there is no limit on consecutive reads. A malicious actor that can stream and sustain a rate of distinct uncached names over the TCP/DoT connection, monopolizes a single worker's entire event loop for as long as its writes stay ahead of the drain.

Published 2026-09-17

CoreDNS is a DNS server written in Go. Prior to 1.14.7, the DNS-over-HTTPS, DNS-over-HTTP/3, DNS-over-QUIC, and DNS-over-gRPC request paths in plugin/pkg/doh/doh.go, core/dnsserver/server_quic.go, and core/dnsserver/server_grpc.go call dns.Msg.Unpack on attacker-controlled DNS section counts before dns.DefaultMsgAcceptFunc validates the fixed header. An unauthenticated client can use DNS name compression and excessive section counts to amplify allocation before the plugin chain, so plugin-level rate limiting cannot prevent concurrent requests from exhausting memory and terminating CoreDNS. The ordinary UDP and TCP listeners are not affected because they validate the header first. This issue is fixed in version 1.14.7.

Published 2026-09-16

Svelte devalue is a JavaScript library that serializes values into strings when JSON.stringify isn't sufficient for the job. Prior to 5.9.2, devalue.parse does not reject out-of-bounds indices that are greater than or equal to values.length in src/parse.js. A specially crafted untrusted payload can make the parser alternate between array representations, producing quadratic work as the payload grows and causing denial of service in applications that parse untrusted devalue data. This issue is fixed in version 5.9.2.

Published 2026-09-16

The RabbitMQ Java client library allows Java and JVM-based applications to connect to and interact with RabbitMQ nodes. Prior to 5.34.0, AMQConnection.start() applies Math.min(maxInboundMessageBodySize, frameMax) after Connection.Tune negotiation even though AMQP defines frameMax value zero as unlimited and ConnectionFactory.DEFAULT_FRAME_MAX is zero. When the client default and server-negotiated value are both zero, the result is passed to Utils.framePayloadLimit(int), which interprets zero as Integer.MAX_VALUE and disables the configured maxInboundMessageBodySize cap. A malicious AMQP server, or a man-in-the-middle attacker able to modify Connection.Tune and inject frames into the connection, can then send an oversized frame of any frame type, causing Frame.readFrom() to allocate a large

Published 2026-09-16

vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set media_io_kwargs.video.video_backend to pynvvideocodec, and MediaConnector.fetch_video forwards that choice to VideoMediaIO even when startup configuration selected a software decoder. The engine's _reserve_mm_ipc_gpu_memory logic budgets decoder memory only from static configuration, so the request-selected VIDEO_LOADER_REGISTRY backend can create a CUDA context, decoder surfaces, and decoded-frame allocations that were not removed from the engine's KV-cache budget. An attacker able to submit video requests to a video-capable GPU deployment with PyNvVideoCodec installed can exhaust shared GPU memory, causing request failures, worker crashes, or denia

Published 2026-09-16

vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the input_audio handling path for /v1/chat/completions calls AudioMediaIO.load_bytes or AudioMediaIO.load_file without passing VLLM_MAX_AUDIO_DECODE_DURATION_S to the shared audio decoder. An unauthenticated client can therefore submit a small compressed audio input that expands into a very large float32 PCM allocation, bypassing the duration guard already used by /v1/audio/transcriptions and causing an out-of-memory worker crash. Inline data URLs reach this path without being bounded by VLLM_AUDIO_FETCH_TIMEOUT. The issue affects deployments serving an audio-capable model, and authentication changes only the deployment-specific reachability. This issue is fixed in version 0.24.0.

Published 2026-09-16

RabbitMQ amqp091-go is a Go AMQP 0.9.1 client. Prior to 1.13.0, Channel.dispatch in channel.go, confirms.confirm in confirms.go, and Connection.dispatch0 in connection.go synchronously send publisher confirmations, flow-control events, consumer cancellations, returned messages, including NotifyConfirm events and connection block notifications, to application-provided channels. If a listener channel is unbuffered, full, or not drained promptly, the sole reader goroutine blocks and stops processing frames, acknowledgments, deliveries, and heartbeats. Broker-driven event bursts can therefore cause connection stalls, missed heartbeats, deadlocks, and disconnection. This issue is fixed in version 1.13.0.

Published 2026-09-16

RabbitMQ amqp091-go is a Go AMQP 0.9.1 client. Prior to 1.13.0, Connection.openTune in connection.go accepts a server-advertised FrameMax below the AMQP frameMinSize value of 4096 bytes because the connection negotiation loop does not enforce the protocol minimum. A malicious or compromised AMQP broker can therefore advertise an extremely small FrameMax, causing later client publications to be fragmented into excessive numbers of frames and write operations. This can consume CPU and stall the client or its host. This issue is fixed in version 1.13.0.

Published 2026-09-16

GitLab has remediated an issue in GitLab CE/EE affecting all versions from 18.4.6 before 19.1.8, 19.2 before 19.2.6, and 19.3 before 19.3.2 that under certain conditions could have allowed an unauthenticated user to cause denial of service due to improper resource allocation limits in the GraphQL complexity calculation logic.

Published 2026-09-16

GitLab has remediated an issue in GitLab CE/EE affecting all versions from 18.4.6 before 19.1.8, 19.2 before 19.2.6, and 19.3 before 19.3.2 that under certain conditions could have allowed an unauthenticated user to cause denial of service due to improper resource allocation limits in the GraphQL complexity calculation logic.

Published 2026-09-16

Http4s is a Scala interface for HTTP services. Prior to 0.23.35 and 1.0.0-M47, When Ember receives an HTTP/2 HEADERS or PUSH_PROMISE frame without END_HEADERS, H2Connection buffers the header block and subsequent CONTINUATION fragments without a size bound. A remote peer can keep an incomplete block open and exhaust heap memory before request decoding, affecting an ember-server or ember-client configured with withHttp2. The remediation tracks accumulated size against SETTINGS_MAX_HEADER_LIST_SIZE derived from EmberServerBuilder.maxHeaderSize or EmberClientBuilder.maxResponseHeaderSize, sends GOAWAY when the limit is exceeded, and applies receiveHeadersTimeout to incomplete blocks. This issue is fixed in versions 0.23.35 and 1.0.0-M47.

Published 2026-09-15

Http4s is a Scala interface for HTTP services. Prior to 0.23.35 and 1.0.0-M47, An Ember server with HTTP/2 enabled through withHttp2 does not enforce SETTINGS_MAX_CONCURRENT_STREAMS for peer-created streams. One unauthenticated connection can open an unbounded number of streams, each retaining per-stream state until heap exhaustion. The same unchecked allocation is reachable in an ember-client through server-initiated PUSH_PROMISE frames because enablePush is not enforced. This issue is fixed in versions 0.23.35 and 1.0.0-M47.

Published 2026-09-15

Http4s is a Scala interface for HTTP services. Prior to 0.23.35 and 1.0.0-M47, Ember’s HTTP/2 flow-control window is replenished according to bytes received from the network rather than bytes consumed by the application, while each stream stores DATA in an unbounded channel. A hostile peer can therefore send a body faster than a slow or non-draining application consumes it, retaining payloads in heap on an ember-server or ember-client configured with withHttp2. The patch bounds the per-stream H2Connection body channel so application consumption applies backpressure. This issue is fixed in versions 0.23.35 and 1.0.0-M47.

Published 2026-09-15

Http4s is a Scala interface for HTTP services. Prior to 0.23.35 and 1.0.0-M47, Ember HTTP/2 serializes outbound frames through one unbounded queue consumed by writeLoop. When the peer stops reading, an unauthenticated HTTP/2 client can continue sending PING, SETTINGS, or DATA frames that cause Ember to enqueue acknowledgments or WINDOW_UPDATE frames faster than the writer drains them, exhausting heap memory on a server built with withHttp2. The shared behavior also affects an ember-client connected to a hostile HTTP/2 server, and the patch replaces the unbounded path with bounded, backpressured outbound queues. This issue is fixed in versions 0.23.35 and 1.0.0-M47.

Published 2026-09-15

Http4s is a Scala interface for HTTP services. Prior to 0.23.35 and 1.0.0-M47, The shared WebSocket decoder permits unbounded message buffering because defragmentation accumulates fragments without a limit and FrameTranscoder accepts declared lengths up to Int.MaxValue. A remote client that completes a WebSocket handshake against an http4s-blaze-server or http4s-ember-server endpoint can exhaust server memory with oversized frames or fragmented messages. The patched decoder applies a configurable 64 MiB default limit to individual frames and defragmented messages through EmberServerBuilder.withMaxWebSocketMessageSize. This issue is fixed in versions 0.23.35 and 1.0.0-M47.

Published 2026-09-15

Inspektor Gadget is a set of tools and framework for data collection and system inspection on Kubernetes clusters and Linux hosts using eBPF. From 0.27.0 until 0.53.1, the uprobe library resolver can allow an unprivileged container to consume excessive CPU and block other containers from starting by supplying a crafted /etc/ld.so.cache file while an uprobe-based gadget is active. The parser in pkg/uprobetracer/ldcache_parser.go trusts EntryCount enough to perform excessive iteration, computes cache1Len with overflowing uint32 arithmetic, and repeatedly calls readStringFromBytes in pkg/uprobetracer/bytes.go, where byte-by-byte immutable string concatenation produces quadratic work. While this processing occurs, pkg/container-hook holds the fanotify container-start pause, allowing a crafted

Published 2026-09-15

Tornado before 6.5.8 contains a memory amplification vulnerability in parse_multipart_form_data that splits multipart data before validating the max_parts limit. Attackers can send crafted multipart requests with many parts to create large transient lists, exhausting server memory and causing denial of service.

Published 2026-09-15

atomic-agents-stack before 1.1.0 contains a cost-guardrail bypass in the _estimate_batch_cost function that returns zero cost for unknown models not in the pricing table. Attackers can configure deployments with unknown model identifiers to bypass daily cost caps and exceed budget limits in parallel batch operations.

Published 2026-09-15

Vikunja versions before 2.6.0 contain a resource exhaustion vulnerability in the Planka migrator that fails to enforce aggregate memory budgets during migration jobs. Authenticated attackers can submit migration requests pointing to attacker-controlled servers advertising numerous size-compliant attachments, exhausting worker memory and causing denial of service for all users.

Published 2026-09-15

pyLoad is a free and open-source download manager written in Python. Prior to 0.5.0b3.dev101, EventManager in src/pyload/core/managers/event_manager.py appends a Client object to the clients list for each unique uuid submitted to the authenticated getEvents API endpoint, but get_events does not invoke the available clean method to remove inactive clients. An authenticated user can repeatedly submit unique UUID values, causing retained client objects and process memory to grow without bound even after requests stop. The resulting memory exhaustion can trigger an operating-system out-of-memory termination of pyLoad or host-wide instability and denial of service. This issue is fixed in version 0.5.0b3.dev101.

Published 2026-09-15

Affected versions of MISP do not consistently apply the existing authentication-failure logging throttle. Two API authentication failure branches wrote directly to the Log model: - API requests with no authentication key; - requests supplying an API key with an incorrect length Unlike other authentication failures, these paths bypassed _shouldLog(), so every request could create another durable auth_fail entry. Version affected: ≤2.5.45

Published 2026-09-15

This issue was addressed with additional entitlement checks. This issue is fixed in iOS 26.7 and iPadOS 26.7, iOS 27 and iPadOS 27, visionOS 27. An app may be able to cause a denial-of-service.

Published 2026-09-14

IBM Verify Identity Access could allow a remote attacker to cause a denial of service due to insufficient validation of incoming request resources.

Published 2026-09-14

Skipper is an HTTP router and reverse proxy for service composition. Prior to 0.26.22, Handler in dataclients/kubernetes/admission/admission.go passes the body of requests to the Kubernetes admission endpoint at :9443/admission directly to io.ReadAll(r.Body) without a size limit. An attacker with in-cluster network access and a valid Kubernetes client certificate can send a very large body that causes unbounded memory allocation and an out-of-memory termination of the Skipper process. The disruption is limited to Ingress and RouteGroup admission rather than pod creation or unrelated admission controllers, and Kubernetes normally restarts the process. This issue is fixed in version 0.26.22.

Published 2026-09-14

webhook through 2.8.3 reads the entire request body into memory before evaluating trigger rules, allowing unauthenticated attackers to exhaust memory by sending oversized bodies. Attackers can send multi-gigabyte request bodies with invalid signatures to trigger out-of-memory conditions and crash the service.

Published 2026-09-14

docx4j is an open source Java library for creating, editing, and saving OpenXML packages, including DOCX, PPTX, and XLSX files. Prior to 11.5.14, PropertyResolver and adjacent helpers recursively follow the WordprocessingML w:basedOn style inheritance chain without cycle detection. A well-formed DOCX containing mutually based styles causes unbounded recursion in PropertyResolver.fillPPrStack and related effective-style resolution paths, resulting in StackOverflowError. Server-side conversion and table-of-contents processing of an untrusted document can terminate a worker thread, degrade a thread pool, or deny service, although isolation in disposable workers or safe containment of StackOverflowError can reduce the practical effect. The fix adds cyclic-style tracking and CyclicStylesExcepti

Published 2026-09-14

dd-trace-rb is Datadog's client library for Ruby. Prior to 2.32.0, W3C baggage extraction does not enforce DD_TRACE_BAGGAGE_MAX_ITEMS, which defaults to 64, or DD_TRACE_BAGGAGE_MAX_BYTES, which defaults to 8192, although those limits apply during baggage injection. A remote unauthenticated attacker can send a baggage HTTP header containing many comma-separated key-value pairs or a single very large value. The extraction path allocates entries while parsing the attacker-controlled header on every request, causing unbounded CPU and memory consumption in an HTTP service where the baggage propagation style is enabled, which is the default for most affected tracers. This can cause denial of service. This issue is fixed in version 2.32.0.

Published 2026-09-14

dd-trace-java is a Datadog APM client for Java. Prior to 1.62.0, W3C baggage extraction does not enforce DD_TRACE_BAGGAGE_MAX_ITEMS, which defaults to 64, or DD_TRACE_BAGGAGE_MAX_BYTES, which defaults to 8192, although those limits apply during baggage injection. A remote unauthenticated attacker can send a baggage HTTP header containing many comma-separated key-value pairs or a single very large value. The extraction path allocates map entries while parsing the attacker-controlled header on every request, causing unbounded CPU and memory consumption in an HTTP service where the baggage propagation style is enabled, which is the default for most affected tracers. This can cause denial of service. This issue is fixed in version 1.62.0.

Published 2026-09-14

webtransport-go is an implementation of the WebTransport protocol. Prior to 0.11.1, Session.parseNextCapsule() in session.go skips an unknown WebTransport capsule on the HTTP/3 request stream by calling io.ReadAll on the capsule reader, retaining the complete declared capsule body in memory. A malicious peer can send a large unknown capsule and cause a client or server to allocate memory for the full payload; QUIC flow control does not bound the total retained allocation because reading advances the flow-control window while the received bytes remain in memory. The resulting memory and resource exhaustion can disrupt or crash the affected process. This issue is fixed in version 0.11.1.

Published 2026-09-14