GHSA-GRP7-V8XH-RJ7H
Vulnerability from github – Published: 2026-08-25 18:09 – Updated: 2026-08-25 18:09Summary
GRPC.Codec.Erlpack.decode/2 calls :erlang.binary_to_term/1 directly on the raw gRPC message body without the :safe option. Any unauthenticated peer that can reach a gRPC endpoint with Content-Type: application/grpc+erlpack can crash the entire BEAM node via atom table exhaustion or, if a decoded fun term flows into a call site that invokes it, achieve remote code execution inside the server process.
Details
Root cause — lib/grpc/codec/erlpack.ex implements decode/2 as a bare :erlang.binary_to_term(binary) call with no :safe flag, no size limit, and no type validation. This has two independent exploitation paths:
1. DoS via atom exhaustion — BEAM atoms are never garbage-collected and the global atom table is bounded (~1,048,576 entries). A crafted payload encoding large numbers of fresh atoms saturates the table and crashes the entire VM, taking down all applications on the node.
2. RCE via fun materialization — Without :safe, binary_to_term/1 reconstructs fun and external-fun terms from wire data. If the decoded value reaches any call site that applies it (e.g. Enum.map, Task.async, direct invocation), attacker-controlled code executes inside the server process.
Configuration requirement: GRPC.Codec.Erlpack is not registered by default and must be explicitly added to the server's codecs option.
PoC
- Start a gRPC server with
codecs: [GRPC.Codec.Erlpack]. - Open an HTTP/2 connection to the server.
- Send a gRPC-framed POST to any RPC path with
Content-Type: application/grpc+erlpackand a body of:erlang.term_to_binary(fn -> <malicious_code> end). - The server's
decode/2materializes the fun; any downstream call site that invokes the decoded value executes the attacker's code. - For DoS only: send payloads encoding fresh atoms in a loop until the atom table is exhausted and the VM crashes.
Impact
Affects grpc ≥ 0.4.0. Any server that explicitly registers GRPC.Codec.Erlpack is vulnerable to unauthenticated node-level DoS and potentially RCE.
References
- Introduction commit: https://github.com/elixir-grpc/grpc/commit/25bcc569fe2cc4478531a6c546c923205fc751c9
- Patch commit: https://github.com/elixir-grpc/grpc/commit/272a97a5ea1b46af1819f14a831fcf35fc91f992
{
"affected": [
{
"package": {
"ecosystem": "Hex",
"name": "grpc"
},
"ranges": [
{
"events": [
{
"introduced": "0.4.0"
},
{
"fixed": "1.0.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-48853"
],
"database_specific": {
"cwe_ids": [
"CWE-502",
"CWE-770"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-25T18:09:53Z",
"nvd_published_at": "2026-06-15T23:16:45Z",
"severity": "CRITICAL"
},
"details": "### Summary\n\n`GRPC.Codec.Erlpack.decode/2` calls `:erlang.binary_to_term/1` directly on the raw gRPC message body without the `:safe` option. Any unauthenticated peer that can reach a gRPC endpoint with `Content-Type: application/grpc+erlpack` can crash the entire BEAM node via atom table exhaustion or, if a decoded fun term flows into a call site that invokes it, achieve remote code execution inside the server process.\n\n### Details\n\n**Root cause** \u2014 `lib/grpc/codec/erlpack.ex` implements `decode/2` as a bare `:erlang.binary_to_term(binary)` call with no `:safe` flag, no size limit, and no type validation. This has two independent exploitation paths:\n\n**1. DoS via atom exhaustion** \u2014 BEAM atoms are never garbage-collected and the global atom table is bounded (~1,048,576 entries). A crafted payload encoding large numbers of fresh atoms saturates the table and crashes the entire VM, taking down all applications on the node.\n\n**2. RCE via fun materialization** \u2014 Without `:safe`, `binary_to_term/1` reconstructs fun and external-fun terms from wire data. If the decoded value reaches any call site that applies it (e.g. `Enum.map`, `Task.async`, direct invocation), attacker-controlled code executes inside the server process.\n\n**Configuration requirement:** `GRPC.Codec.Erlpack` is not registered by default and must be explicitly added to the server\u0027s `codecs` option.\n\n### PoC\n\n1. Start a gRPC server with `codecs: [GRPC.Codec.Erlpack]`.\n2. Open an HTTP/2 connection to the server.\n3. Send a gRPC-framed POST to any RPC path with `Content-Type: application/grpc+erlpack` and a body of `:erlang.term_to_binary(fn -\u003e \u003cmalicious_code\u003e end)`.\n4. The server\u0027s `decode/2` materializes the fun; any downstream call site that invokes the decoded value executes the attacker\u0027s code.\n5. For DoS only: send payloads encoding fresh atoms in a loop until the atom table is exhausted and the VM crashes.\n\n### Impact\n\nAffects `grpc` \u2265 0.4.0. Any server that explicitly registers `GRPC.Codec.Erlpack` is vulnerable to unauthenticated node-level DoS and potentially RCE.\n\n### References\n\n* Introduction commit: https://github.com/elixir-grpc/grpc/commit/25bcc569fe2cc4478531a6c546c923205fc751c9\n* Patch commit: https://github.com/elixir-grpc/grpc/commit/272a97a5ea1b46af1819f14a831fcf35fc91f992",
"id": "GHSA-grp7-v8xh-rj7h",
"modified": "2026-08-25T18:09:53Z",
"published": "2026-08-25T18:09:53Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/elixir-grpc/grpc/security/advisories/GHSA-grp7-v8xh-rj7h"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-48853"
},
{
"type": "WEB",
"url": "https://github.com/elixir-grpc/grpc/pull/540"
},
{
"type": "WEB",
"url": "https://github.com/elixir-grpc/grpc/commit/272a97a5ea1b46af1819f14a831fcf35fc91f992"
},
{
"type": "WEB",
"url": "https://cna.erlef.org/cves/CVE-2026-48853.html"
},
{
"type": "PACKAGE",
"url": "https://github.com/elixir-grpc/grpc"
},
{
"type": "WEB",
"url": "https://github.com/elixir-grpc/grpc/releases/tag/v1.0.0"
},
{
"type": "WEB",
"url": "https://osv.dev/vulnerability/EEF-CVE-2026-48853"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N",
"type": "CVSS_V4"
}
],
"summary": "gRPC Erlang package vulnerable to Remote Code Execution with attacker-controlled gRPC payloads"
}
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.