CVE-2026-43632
Full vulnerability record from the local CVENow database.
Description
llama.cpp builds b7492 through the latest b9060 contains a use-after-free vulnerability in llama-server affecting six tokenization endpoints (/tokenize, /detokenize, /infill, /apply-template, /rerank, and /anthropic/count_tokens) that bypass the task queue and access ctx_server.vocab directly on HTTP worker threads. Attackers can exploit a time-of-check-time-of-use race condition where the main thread destroys and frees vocab after the synchronization lock is released but before the handler finishes using it, causing a crash or potential code execution when --sleep-idle-seconds is configured.
Severity
CRITICAL
CVSS score
9.2
(v4.0)
Vulnerability status
Analyzed
Source identifier
Published
2026-08-06 22:17:06.343000
Last modified
2026-10-23 21:10:00.157000
View original CVE JSON
{
"id": "CVE-2026-43632",
"sourceIdentifier": "[email protected]",
"published": "2026-08-06T22:17:06.343",
"lastModified": "2026-10-23T21:10:00.157",
"vulnStatus": "Analyzed",
"cveTags": [],
"descriptions": [
{
"lang": "en",
"value": "llama.cpp builds b7492 through the latest b9060 contains a use-after-free vulnerability in llama-server affecting six tokenization endpoints (/tokenize, /detokenize, /infill, /apply-template, /rerank, and /anthropic/count_tokens) that bypass the task queue and access ctx_server.vocab directly on HTTP worker threads. Attackers can exploit a time-of-check-time-of-use race condition where the main thread destroys and frees vocab after the synchronization lock is released but before the handler finishes using it, causing a crash or potential code execution when --sleep-idle-seconds is configured."
},
{
"lang": "es",
"value": "Las compilaciones de llama.cpp desde la b7492 hasta la última b9060 contienen una vulnerabilidad de uso después de liberación en llama-server que afecta a seis puntos finales de tokenización (/tokenize, /detokenize, /infill, /apply-template, /rerank, y /anthropic/count_tokens) que omiten la cola de tareas y acceden directamente a ctx_server.vocab en los hilos de trabajo HTTP. Los atacantes pueden explotar una condición de carrera de tiempo de verificación-tiempo de uso donde el hilo principal destruye y libera vocab después de que se libera el bloqueo de sincronización pero antes de que el gestor termine de usarlo, causando un fallo o una posible ejecución de código cuando se configura --sleep-idle-seconds."
}
],
"affected": [
{
"source": "[email protected]",
"affectedData": [
{
"vendor": "ggml-org",
"product": "llama.cpp",
"defaultStatus": "affected",
"versions": [
{
"version": "b7492",
"lessThanOrEqual": "b9060",
"versionType": "custom",
"status": "affected"
}
]
}
]
}
],
"metrics": {
"cvssMetricV40": [
{
"source": "[email protected]",
"type": "Secondary",
"cvssData": {
"version": "4.0",
"vectorString": "CVSS:4.0/AV:N/AC:H/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
"baseScore": 9.2,
"baseSeverity": "CRITICAL",
"attackVector": "NETWORK",
"attackComplexity": "HIGH",
"attackRequirements": "PRESENT",
"privilegesRequired": "NONE",
"userInteraction": "NONE",
"vulnConfidentialityImpact": "HIGH",
"vulnIntegrityImpact": "HIGH",
"vulnAvailabilityImpact": "HIGH",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "NONE",
"subAvailabilityImpact": "NONE",
"exploitMaturity": "NOT_DEFINED",
"confidentialityRequirement": "NOT_DEFINED",
"integrityRequirement": "NOT_DEFINED",
"availabilityRequirement": "NOT_DEFINED",
"modifiedAttackVector": "NOT_DEFINED",
"modifiedAttackComplexity": "NOT_DEFINED",
"modifiedAttackRequirements": "NOT_DEFINED",
"modifiedPrivilegesRequired": "NOT_DEFINED",
"modifiedUserInteraction": "NOT_DEFINED",
"modifiedVulnConfidentialityImpact": "NOT_DEFINED",
"modifiedVulnIntegrityImpact": "NOT_DEFINED",
"modifiedVulnAvailabilityImpact": "NOT_DEFINED",
"modifiedSubConfidentialityImpact": "NOT_DEFINED",
"modifiedSubIntegrityImpact": "NOT_DEFINED",
"modifiedSubAvailabilityImpact": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"valueDensity": "NOT_DEFINED",
"vulnerabilityResponseEffort": "NOT_DEFINED",
"providerUrgency": "NOT_DEFINED"
}
}
],
"cvssMetricV31": [
{
"source": "[email protected]",
"type": "Secondary",
"cvssData": {
"version": "3.1",
"vectorString": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H",
"baseScore": 8.1,
"baseSeverity": "HIGH",
"attackVector": "NETWORK",
"attackComplexity": "HIGH",
"privilegesRequired": "NONE",
"userInteraction": "NONE",
"scope": "UNCHANGED",
"confidentialityImpact": "HIGH",
"integrityImpact": "HIGH",
"availabilityImpact": "HIGH"
},
"exploitabilityScore": 2.2,
"impactScore": 5.9
}
],
"ssvcV203": [
{
"source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
"ssvcData": {
"timestamp": "2026-08-08T02:09:08.150856Z",
"id": "CVE-2026-43632",
"options": [
{
"exploitation": "none"
},
{
"automatable": "no"
},
{
"technicalImpact": "total"
}
],
"role": "CISA Coordinator",
"version": "2.0.3"
}
}
]
},
"weaknesses": [
{
"source": "[email protected]",
"type": "Secondary",
"description": [
{
"lang": "en",
"value": "CWE-367"
},
{
"lang": "en",
"value": "CWE-416"
}
]
}
],
"configurations": [
{
"nodes": [
{
"operator": "OR",
"negate": false,
"cpeMatch": [
{
"vulnerable": true,
"criteria": "cpe:2.3:a:ggml:llama.cpp:*:*:*:*:*:*:*:*",
"versionStartIncluding": "b7492",
"versionEndIncluding": "b9060",
"matchCriteriaId": "0809463E-7195-4AD2-90E6-1CE6A05CC8BA"
}
]
}
]
}
],
"references": [
{
"url": "https://github.com/Vladimir-tokarev-cyera/llama-cpp-security-patches",
"source": "[email protected]",
"tags": [
"Patch",
"Third Party Advisory"
]
},
{
"url": "https://www.vulncheck.com/advisories/llama-cpp-b7492-b9060-use-after-free-in-tokenization-endpoints",
"source": "[email protected]",
"tags": [
"Patch",
"Third Party Advisory"
]
}
]
}