CVE Watch30 jul 2026

CVE-2026-65624

Allocation of Resources Without Limits or Throttling vulnerability in ninenines cowboy allows an unauthenticated remote attacker to exhaust

CVSS

—

Sin CVSS

EPSS

0.4%

p32

KEV

—

Exploit Today

0-100

Publicado: 28 jul 2026 · Última mod.: 30 jul 2026 · CWE-770

EPSS · 30d

0.4%EPSS · 30 días0.4%

2026-08-242026-09-21

Descripción técnica

Allocation of Resources Without Limits or Throttling vulnerability in ninenines cowboy allows an unauthenticated remote attacker to exhaust connection process memory over HTTP/1.1. The HTTP/1.1 handler in cowboy_http enforces the max_headers limit by counting the number of distinct header names in a map (maps:size(Headers)). When a request contains multiple header lines with the same name, the values are concatenated into a single ever-growing binary stored under that one map key (", " for regular headers, "; " for cookies), so the map size stays at one and the max_headers cap (default 100) is never reached. Because no accumulator bounds the total number of header lines or the total byte size of the header block (only per-line max_header_name_length and max_header_value_length apply), an unauthenticated client can send an arbitrary number of header lines with the same name and grow the connection process's binary memory to arbitrary size within the request window. The impact per connection is bounded by request_timeout (default 5 seconds, not reset by header data), and by max_heap_size when set (the offending connection process is killed once its heap grows past the limit). When max_heap_size is left at the default (unset), sustained abuse can drive the Erlang VM into out-of-memory conditions. This issue affects cowboy from 2.0.0-pre.4 before 2.18.0.

Referencias oficiales

CVEs relacionados

CVECVSSEPSSKEVRExplotTítuloVis.

CVE-2026-68956—

—

———Allocation of Resources Without Limits or Throttling vulnerability in Erlang/OTP ssh allows an authenticated remote attacker to exhaust node memory by repeatedly opening session channels that are never assigned a handler. The "session" clause of ssh_connection:handle_msg/4 checks only minimal_remote_max_packet_size before calling setup_session/5, which unconditionally builds a #channel{} record and stores it in the ETS channel cache. The max_channels daemon option is consulted only by ssh_channel_sup:max_num_channels_not_exceeded/2, which counts supervisor children, so a channel that never gets a shell, exec, or subsystem handler is invisible to the limit and setting the option to a finite value does not mitigate the attack. RFC 4254 section 5.1 permits many session channels per connection, and each record costs only a few hundred bytes, so a single authenticated connection can accumulate channels until the node runs out of memory and the emulator terminates, affecting every application on it. No file contents, credentials, or write access are obtainable. This issue affects OTP from OTP 18.1.2 before OTP 27.3.4.18, OTP 28.5.0.7, and OTP 29.1.1, corresponding to ssh from 4.1.1 before 5.2.11.13, 5.5.2.6, and 6.0.6. Whether OTP before OTP 18.1.2, corresponding to ssh before 4.1.1, is affected is unknown.3h CVE-2026-946247.5 ALT

—

———vLLM through 0.29.0 contains a denial of service vulnerability in P2P KV offloading when OffloadingConnector is configured with TieringOffloadingSpec and a peer-to-peer secondary tier. Attackers can supply arbitrary remote host and port values in kv_transfer_params to create unreachable peer sessions that retain ZeroMQ sockets until the context quota is exhausted, causing an uncaught ZMQError that crashes EngineCore and stops all inference.16h CVE-2026-61652—

—

———Zapros, a Python HTTP client, prior to version 0.14.0 is vulnerable to denial of service via memory exhaustion. The issue affects all callers who streamed compressed responses relying on the chunk size — explicit (`iter_bytes(chunk_size=...)`) or the default — to bound memory. The decoder ignored that bound, so a chunk could be far larger than requested and a single compressed response could overflow memory. Version 0.14.0 contains a patch. Some workarounds are available. Read the still-compressed body with `Response.iter_raw()` / `Response.async_iter_raw()`, which bypass the built-in decoders, and decompress it yourself with an explicit output-size bound (e.g. `zlib`'s `max_length`), aborting once a configured limit is exceeded. Where feasible, send `Accept-Encoding: identity` to disable response compression so bodies are not decompressed client-side. Avoid decoding response bodies from untrusted servers.16h CVE-2026-61541—

—

———Zapros, a Python HTTP client, prior to version 0.14.0 is vulnerable to denial of service when an application requests content from an untrusted server, or follows a redirect to one, because a malicious response containing an excessive number of chained `Content-Encoding` values causes Zapros to construct a deeply nested decompression chain that consumes excessive resources. Version 0.14.0 patches the vulnerability by limiting responses to five content-encoding layers and raising `DecodingError` when that limit is exceeded. As a workaround, applications can add response middleware that inspects the `Content-Encoding` header and rejects responses containing more than a safe number of encoding layers.16h CVE-2026-852193.7 BAJ

—

———Denial-of-Service in Redis module in Thinkst Canary's OpenCanary 0.9.9 allows an unauthenticated remote attacker cause unconstrained memory usage.18h CVE-2026-616297.5 ALT

—

——0nginx ignition is a user interface for the nginx web server. In versions 2.29.0 through 2.40.0, the gin i18n middleware in nginx-ignition's API server runs in front of every HTTP request and calls `golang.org/x/text/language.ParseAcceptLanguage` on the raw `Accept-Language` header without imposing any size or shape filter. The underlying parser has quadratic-time behaviour on long lists of malformed language tags. The CVE-2022-32149 guard that golang.org/x/text added in v0.3.8 caps the number of `-` characters in the input at 1000, but it does not cap `_` characters even though the parser's internal scanner aliases `_` to `-` before parsing. A single unauthenticated GET request with an `Accept-Language` header built out of `_` separators burns about 2.4 seconds of server CPU on the host running nginx-ignition; ten concurrent attackers saturate a ten-core box for the duration of the attack while consuming ~10 MiB/s of upstream bandwidth. Version 2.40.1 fixes this issue.23h