GHSA-F9QC-QG88-7PQ5
Vulnerability from github – Published: 2026-08-28 18:34 – Updated: 2026-08-28 18:34The decode_unknown_field function in buffa's protobuf decoder allocated heap memory in proportion to untrusted input (unknown fields in the serialized protobuf) without enforcing an allocation budget. Any message decoded from untrusted input using code generated with preserve_unknown_fields=true (the default) was affected. A small, well-formed payload of nested unknown fields inside a StartGroup could trigger roughly 22× memory amplification (e.g., a 64 MiB input forcing ~1.4 GB of heap allocation), and length-delimited unknown fields could be sized arbitrarily, enabling an unauthenticated attacker to crash a process via memory exhaustion. This was reachable from the default decode APIs, since the top-level message size cap did not account for in-decode amplification.
For users of connectrpc - the DEFAULT_MAX_MESSAGE_SIZE for connectrpc is 4MiB, which limits amplification in the worst case to ~88 MiB of memory. A flood of concurrent requests with this pattern could still be used to exhaust available memory, however.
Users are advised to either set preserve_unknown_fields=false on their current generated code, or upgrade to 0.8.0, which enforces per-message unknown field count limits - this is configurable, with a default of 1 million unknown fields, or ~40MiB of allocation overhead per message.
Users are advised to update to the latest version, which enforces per-message unknown field count limits.
Thank you to @p80n-sec for reporting this issue.
{
"affected": [
{
"package": {
"ecosystem": "crates.io",
"name": "buffa"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.8.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-55407"
],
"database_specific": {
"cwe_ids": [
"CWE-400",
"CWE-770",
"CWE-789"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-28T18:34:22Z",
"nvd_published_at": "2026-07-16T17:16:57Z",
"severity": "MODERATE"
},
"details": "The `decode_unknown_field` function in buffa\u0027s protobuf decoder allocated heap memory in proportion to untrusted input (unknown fields in the serialized protobuf) without enforcing an allocation budget. Any message decoded from untrusted input using code generated with `preserve_unknown_fields=true` (the default) was affected. A small, well-formed payload of nested unknown fields inside a StartGroup could trigger roughly 22\u00d7 memory amplification (e.g., a 64 MiB input forcing ~1.4 GB of heap allocation), and length-delimited unknown fields could be sized arbitrarily, enabling an unauthenticated attacker to crash a process via memory exhaustion. This was reachable from the default decode APIs, since the top-level message size cap did not account for in-decode amplification. \n\nFor users of connectrpc - the DEFAULT_MAX_MESSAGE_SIZE for connectrpc is 4MiB, which limits amplification in the worst case to ~88 MiB of memory. A flood of concurrent requests with this pattern could still be used to exhaust available memory, however.\n\nUsers are advised to either set `preserve_unknown_fields=false` on their current generated code, or upgrade to 0.8.0, which enforces per-message unknown field count limits - this is configurable, with a default of 1 million unknown fields, or ~40MiB of allocation overhead per message.\nUsers are advised to update to the latest version, which enforces per-message unknown field count limits.\n\nThank you to @p80n-sec for reporting this issue.",
"id": "GHSA-f9qc-qg88-7pq5",
"modified": "2026-08-28T18:34:22Z",
"published": "2026-08-28T18:34:22Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/anthropics/buffa/security/advisories/GHSA-f9qc-qg88-7pq5"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-55407"
},
{
"type": "WEB",
"url": "https://github.com/anthropics/buffa/pull/184"
},
{
"type": "WEB",
"url": "https://github.com/anthropics/buffa/commit/278fa43fcff661d4ee6bd83b75955a153d4281fc"
},
{
"type": "PACKAGE",
"url": "https://github.com/anthropics/buffa"
},
{
"type": "WEB",
"url": "https://github.com/anthropics/buffa/releases/tag/v0.8.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "Buffa Vulnerable to Memory Exhaustion Denial of Service in decode_unknown_field via Unbounded Allocation"
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.