Base de CVEs

LightLLM through 1.2.0 contains an authentication bypass vulnerability in the /pd_register WebSocket endpoint that allows unauthenticated attackers to register arbitrary nodes by supplying crafted JSON without peer address validation. Attackers can disclose full user prompts routed to their socket, trigger denial of service by replacing legitimate nodes, or make the PD Master issue requests to internal network addresses.

O IBM Guardium Data Protection 12.2 pode permitir que um invasor remoto execute comandos SQL arbitrários devido à neutralização imprópria de elementos especiais usados em um comando SQL.

O IBM Guardium Data Protection 12.2 é vulnerável a uma falha de falta de autenticação no LoadBalancerServlet. Um usuário não autenticado pode acessar operações privilegiadas de balanceamento de carga, potencialmente resultando em ações não autorizadas e impacto na integridade e na disponibilidade do sistema afetado.

O IBM Guardium Data Protection 12.2 pode permitir que um invasor remoto contorne restrições de segurança devido à falta de autenticação para o ChangeTrackerServlet.

O IBM Guardium Data Protection 12.2 pode permitir que um atacante remoto autenticado execute comandos SQL arbitrários devido à neutralização inadequada de elementos especiais usados em um comando SQL.

O IBM Guardium Data Protection 12.2 pode permitir que um atacante remoto autenticado execute comandos SQL arbitrários devido à neutralização imprópria de elementos especiais usados em um comando SQL.

IBM Guardium Data Protection 12.2 could allow a remote authenticated attacker to execute arbitrary code due to improper neutralization of input during web page generation.

IBM Guardium Data Protection 12.2 is vulnerable to an authentication bypass that allows an unauthenticated remote attacker to bypass IP-based access controls and access the Guardium management interface.

IBM Guardium Data Protection 12.2 could allow a remote authenticated attacker to execute arbitrary code due to improper neutralization of input during web page generation.

IBM Guardium Data Protection 12.2 is vulnerable to unauthenticated insecure deserialization and attacker-controlled reflective method dispatch in the Change Audit System (CAS) listener. A network attacker able to reach TCP port 16017 may submit crafted serialized messages and potentially cause unintended code execution in the Guardium appliance.

O IBM Guardium Data Protection 12.2 pode permitir que um invasor remoto não autenticado execute código arbitrário no sistema devido à desserialização de dados não confiáveis.

IBM Guardium Data Protection 12.2 is vulnerable to an unauthenticated second-order SQL injection vulnerability in the generateInsertQuery functionality of change-tracker-data.sql. A remote attacker could inject malicious SQL that is subsequently processed by the application, potentially resulting in compromise of the confidentiality, integrity, and availability of the affected system.

IBM Sterling File Gateway could allow a remote attacker to bypass authentication and obtain a fully authenticated session due to improper authentication via an unvalidated SSO header.

CordysCRM is an open source AI-powered customer relationship management system that supports private deployment. Prior to 1.7.2, SseController exposes the anonymous /sse/subscribe, /sse/broadcast, and /sse/close endpoints because ShiroFilter.addPublicPathFilters permits the SSE paths, and the endpoints trust the caller-controlled userId instead of deriving an identity from an authenticated principal. An unauthenticated caller can use /sse/subscribe to read another user's workflow events, approval requests, mentions, and alerts, use /sse/broadcast to inject SYSTEM_HEARTBEAT messages into another user's stream, or use /sse/close to terminate another user's channel. This vulnerability is fixed in 1.7.2.

pg_partman is a PostgreSQL extension that manages partitioned tables by time or ID. Prior to 5.5.0, create_partition_time() reads the writable part_config.time_encoder text value and interpolates it without identifier quoting into a dynamically executed SELECT statement. A role with the documented partman_user INSERT and UPDATE privileges can store SQL rather than a function name. When pg_partman_bgw later creates a child partition for a text- or UUID-keyed set, the worker executes the stored SQL with pg_partman_bgw.role privileges, which default to PostgreSQL superuser. The persistent configuration row can repeatedly restore elevated access on later maintenance ticks, and successful exploitation can permit database-wide compromise and operating-system command execution as the PostgreSQL service account. This issue is fixed in version 5.5.0.

FluidSynth is a software synthesizer based on the SoundFont 2 specifications. From 1.1.2 until 2.5.6, the FluidSynth command handler accepts a pitch_bend_range command whose channel argument is not bounds checked before the supplied value is written through the selected synth channel. An out-of-range channel can therefore cause an out-of-bounds heap write, leading to denial of service or possible code execution. The issue is remotely reachable when the TCP server is enabled through new_fluid_server() or fluidsynth -s, and it is locally reachable through malicious commands delivered to the FluidSynth shell on standard input. Applications that do not use the shell, command handler, or TCP server are not affected. This issue is fixed in version 2.5.6.

Hongjing e-HR before 8.2 contains a SQL injection vulnerability in the /servlet/codesettree endpoint where the categories query parameter is passed to a database query without sanitization after HRMS-encoding is stripped. An unauthenticated remote attacker can supply a crafted UNION SELECT payload to read arbitrary database content, including credential tables such as operuser. Exploitation evidence was first observed by the Shadowserver Foundation on 2023-10-14.

Mongoid contains an unsafe reflection weakness in the query path used for embedded documents. An application that passes an externally supplied field name to certain in-memory query methods may allow an unauthenticated party to obtain unintended disclosure of stored document data and to permanently remove stored records.

Cocos AI is a confidential computing system for AI. In Cocos AI 0.8.2 and earlier, VerifyAttestationReportTLS in the AMD SEV-SNP verifier sets Policy.ReportData only when its caller supplies non-empty expected reportData. A relying-party integration that passes nil or empty expected reportData while relying on that value for current-session freshness can therefore accept SEV-SNP evidence unrelated to the current connection and make a trust, authorization, or application-data release decision for an unintended attestation context. This issue was fixed in Cocos AI 0.9.0 by removing the affected verifier path as part of the migration to post-handshake attestation.

Cocos AI is a confidential computing system for AI. In Cocos AI 0.8.2 and earlier, VerifTeeAttestation in the Intel TDX aTLS verifier receives an expected current-session value but does not apply it to TdQuoteBodyPolicy.ReportData before validating TDX QuoteV4 evidence. The verifier can consequently accept structurally valid evidence containing mismatched or reused REPORT_DATA for the current TLS session, causing the relying party to release application data or make a trust or authorization decision for an unrelated attestation context. This issue was fixed in Cocos AI 0.9.0 by removing the affected verifier path as part of the migration to post-handshake attestation.

Icinga 2 is an open source monitoring system. From 2.8 until 2.14.9, 2.15.4, and 2.16.2, certificate update JSON-RPC message handling does not validate that the sender is a trusted endpoint. An unauthenticated network attacker able to connect to TCP port 5665 can replace the node certificate and trusted CA certificate, impersonate a trusted node, and take control of the node. This issue is fixed in versions 2.14.9, 2.15.4, and 2.16.2.

Mnemosyne is a memory layer for artificial intelligence agents. Prior to v3.10.1, the auth check in mnemosyne/core/sync_server.py parsed the JWT's header and payload using base64 decoding, then passed the token to a jwt library call with options that effectively disabled signature verification. The server accepted any well-formed token regardless of the signature, including tokens with alg: none and tokens signed with the wrong key. The fix in v3.10.1 replaces the broken decode with a from-scratch HS256 verifier using only the Python standard library. For users who cannot upgrade immediately, restrict network access to the sync server endpoint to trusted clients only. Firewall, reverse proxy with mTLS, or localhost bind with SSH tunnel are all viable. The vulnerability is not exploitable against an unreachable endpoint.

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received through a ZeroMQ PULL socket. PyZMQ implements `recv_pyobj()` using Python pickle deserialization, which can execute arbitrary code while reconstructing an object. The peer address used by the receiver was supplied through the `POST /distserve/p2p_connect` HTTP endpoint. An attacker who could reach an affected DistServe API server could cause the server to connect to an attacker-controlled ZeroMQ endpoint and deserialize a crafted pickle payload. API-key authentication is not enabled unless the operator explicitly configures it. As a result, affected DistServe deployments without API keys allowed unauthenticated remote code execution with the privileges of the LMDeploy serving process. This issue affects the PyTorch backend when PD-disaggregation/DistServe is enabled. Ordinary deployments that do not use the affected disaggregated-serving path do not expose this data flow. The fix was released in LMDeploy 0.16.0. Users who cannot upgrade immediately should prevent untrusted clients from reaching `/distserve/*` endpoints, restrict the DistServe HTTP and ZeroMQ control planes to trusted cluster networks, configure API-key authentication, and block arbitrary outbound ZeroMQ connections from serving nodes. These measures reduce exposure but do not make pickle deserialization safe.

Mongoid contains an unsafe reflection weakness in the document persistence layer of its object-document mapping code. Input whose keys are passed through from an unauthenticated party by an embedding application can cause unintended internal method invocation instead of the intended array field update. This may result in unintended removal of stored records and in the embedding application becoming unresponsive.

CareCam CM2507 IP cameras store the device's root-account password using a fixed legacy password hash that provides insufficient resistance to offline cracking. An attacker who obtains the firmware image or password database could recover the associated credential, which may also be reusable across other devices running the same firmware.