GHSA-P9JM-Q85P-7MCP
Vulnerability from github – Published: 2026-08-07 17:16 – Updated: 2026-08-07 17:16Summary
RedisArrayAggregator clears retained partial aggregate state when the maxNestedArrayDepth limit is exceeded, but it does not clear the same state when the sibling maxElements limit is exceeded. A peer can start a valid RESP array, send a bulk-string child, then send a nested array header longer than the configured maxElements. Netty throws a decoder exception, but the existing partial aggregate remains retained in the handler.
If the application leaves the channel alive after the exception, later messages are still consumed into the pre-error aggregate. The supplied PoV proves both the retained ByteBuf reference and the stale parser state continuation.
Technical Details
RedisArrayAggregator.decode(...) retains non-array messages before adding them to depths.peek().children. In decodeRedisArrayHeader(...), the header.length() > maxElements branch throws immediately:
if (header.length() > maxElements) {
throw new CodecException("this codec doesn't support longer length than " + maxElements);
}
The immediately following nested-depth branch clears retained aggregate state before throwing:
if (depths.size() >= maxNestedArrayDepth) {
releaseAndClearDepths();
throw new CodecException("max nested array depth exceeded: " + maxNestedArrayDepth);
}
The missing cleanup in the first branch leaves retained children and aggregate state reachable after the exception.
PoC
Place the supplied RedisArrayAggregatorIncompleteCleanupPovTest.java under:
codec-redis/src/test/java/io/netty/handler/codec/redis/
Run:
./mvnw -pl codec-redis -am -Dtest=RedisArrayAggregatorIncompleteCleanupPovTest -Dsurefire.failIfNoSpecifiedTests=false -DskipNativeTests -DskipAutobahnTests test
The test suite includes:
- serialized RESP trigger through
RedisDecoder,RedisBulkStringAggregator, andRedisArrayAggregator; - direct refcount proof that max-elements overflow does not release the retained child immediately;
- post-exception continuation proof that the stale aggregate consumes a later message;
- nested-depth controls that clear the same partial aggregate state.
All five tests pass on current 4.2, 4.2.15.Final, and 4.1.135.Final.
Impact
For Redis codec pipelines that continue after codec exceptions, an unauthenticated peer can keep attacker-controlled aggregate state alive across a security-limit exception. This can pin retained pooled buffers until channel close/removal or until a later message completes the stale aggregate.
RedisBulkStringAggregator permits bulk strings up to RedisConstants.REDIS_MESSAGE_MAX_LENGTH (512MB), so the retained child can be large in deployments that aggregate untrusted Redis streams.
Applications that always close the channel or remove the handler on decoder exceptions will trigger existing cleanup; the issue is the missing immediate cleanup on the max-elements failure path while the handler remains installed.
Suggested Fix
Call releaseAndClearDepths() before throwing from the max-elements branch. Consider applying the same cleanup to all unrecoverable decodeRedisArrayHeader(...) error exits that can occur while depths is non-empty.
Affected Package/Versions
io.netty:netty-codec-redis
Confirmed on:
- current
4.2branch head7bae566a93e69409697fe57fa807910ba5c9720e 4.2.15.Finalata41f7b289ce14.1.135.Finalatf05f765d8146
Advisory History
This differs from the public Redis codec advisories because it reproduces on their patched tags:
GHSA-5w86-c3rq-vjj7GHSA-3244-j874-rhc2/CVE-2026-44250GHSA-6jv9-x5w9-2ccm/CVE-2026-48006GHSA-6ghj-frrj-jjj3/CVE-2026-44890
Why This Is Not Intended Behavior
The public API docs document RedisArrayAggregator as aggregating RedisMessage parts into ArrayRedisMessage and document a CodecException when an array header exceeds maxElements. They do not document preserving pre-exception partial aggregate state after that limit fires.
The adjacent nested-depth branch already calls releaseAndClearDepths() before throwing. The max-elements branch is the sibling aggregation-limit branch but throws without cleanup. Netty's later Redis lifecycle cleanup patch explicitly added release behavior for nested-array failure and handler removal, leaving the max-elements failure branch as a missed cleanup path.
{
"affected": [
{
"package": {
"ecosystem": "Maven",
"name": "io.netty:netty-codec-redis"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "4.1.136.Final"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "Maven",
"name": "io.netty:netty-codec-redis"
},
"ranges": [
{
"events": [
{
"introduced": "4.2.0-Final"
},
{
"fixed": "4.2.16.Final"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-56818"
],
"database_specific": {
"cwe_ids": [
"CWE-401",
"CWE-703"
],
"github_reviewed": true,
"github_reviewed_at": "2026-08-07T17:16:13Z",
"nvd_published_at": null,
"severity": "MODERATE"
},
"details": "## Summary\n\n`RedisArrayAggregator` clears retained partial aggregate state when the `maxNestedArrayDepth` limit is exceeded, but it does not clear the same state when the sibling `maxElements` limit is exceeded. A peer can start a valid RESP array, send a bulk-string child, then send a nested array header longer than the configured `maxElements`. Netty throws a decoder exception, but the existing partial aggregate remains retained in the handler.\n\nIf the application leaves the channel alive after the exception, later messages are still consumed into the pre-error aggregate. The supplied PoV proves both the retained `ByteBuf` reference and the stale parser state continuation.\n\n## Technical Details\n\n`RedisArrayAggregator.decode(...)` retains non-array messages before adding them to `depths.peek().children`. In `decodeRedisArrayHeader(...)`, the `header.length() \u003e maxElements` branch throws immediately:\n\n```java\nif (header.length() \u003e maxElements) {\n throw new CodecException(\"this codec doesn\u0027t support longer length than \" + maxElements);\n}\n```\n\nThe immediately following nested-depth branch clears retained aggregate state before throwing:\n\n```java\nif (depths.size() \u003e= maxNestedArrayDepth) {\n releaseAndClearDepths();\n throw new CodecException(\"max nested array depth exceeded: \" + maxNestedArrayDepth);\n}\n```\n\nThe missing cleanup in the first branch leaves retained children and aggregate state reachable after the exception.\n\n## PoC\n\nPlace the supplied `RedisArrayAggregatorIncompleteCleanupPovTest.java` under:\n\n`codec-redis/src/test/java/io/netty/handler/codec/redis/`\n\nRun:\n\n```fish\n./mvnw -pl codec-redis -am -Dtest=RedisArrayAggregatorIncompleteCleanupPovTest -Dsurefire.failIfNoSpecifiedTests=false -DskipNativeTests -DskipAutobahnTests test\n```\n\nThe test suite includes:\n\n- serialized RESP trigger through `RedisDecoder`, `RedisBulkStringAggregator`, and `RedisArrayAggregator`;\n- direct refcount proof that max-elements overflow does not release the retained child immediately;\n- post-exception continuation proof that the stale aggregate consumes a later message;\n- nested-depth controls that clear the same partial aggregate state.\n\nAll five tests pass on current `4.2`, `4.2.15.Final`, and `4.1.135.Final`.\n\n## Impact\n\nFor Redis codec pipelines that continue after codec exceptions, an unauthenticated peer can keep attacker-controlled aggregate state alive across a security-limit exception. This can pin retained pooled buffers until channel close/removal or until a later message completes the stale aggregate.\n\n`RedisBulkStringAggregator` permits bulk strings up to `RedisConstants.REDIS_MESSAGE_MAX_LENGTH` (`512MB`), so the retained child can be large in deployments that aggregate untrusted Redis streams.\n\nApplications that always close the channel or remove the handler on decoder exceptions will trigger existing cleanup; the issue is the missing immediate cleanup on the max-elements failure path while the handler remains installed.\n\n## Suggested Fix\n\nCall `releaseAndClearDepths()` before throwing from the max-elements branch. Consider applying the same cleanup to all unrecoverable `decodeRedisArrayHeader(...)` error exits that can occur while `depths` is non-empty.\n\n## Affected Package/Versions\n\n`io.netty:netty-codec-redis`\n\nConfirmed on:\n\n- current `4.2` branch head `7bae566a93e69409697fe57fa807910ba5c9720e`\n- `4.2.15.Final` at `a41f7b289ce1`\n- `4.1.135.Final` at `f05f765d8146`\n\n## Advisory History\n\nThis differs from the public Redis codec advisories because it reproduces on their patched tags:\n\n- `GHSA-5w86-c3rq-vjj7`\n- `GHSA-3244-j874-rhc2` / `CVE-2026-44250`\n- `GHSA-6jv9-x5w9-2ccm` / `CVE-2026-48006`\n- `GHSA-6ghj-frrj-jjj3` / `CVE-2026-44890`\n\n## Why This Is Not Intended Behavior\n\nThe public API docs document `RedisArrayAggregator` as aggregating `RedisMessage` parts into `ArrayRedisMessage` and document a `CodecException` when an array header exceeds `maxElements`. They do not document preserving pre-exception partial aggregate state after that limit fires.\n\nThe adjacent nested-depth branch already calls `releaseAndClearDepths()` before throwing. The max-elements branch is the sibling aggregation-limit branch but throws without cleanup. Netty\u0027s later Redis lifecycle cleanup patch explicitly added release behavior for nested-array failure and handler removal, leaving the max-elements failure branch as a missed cleanup path.",
"id": "GHSA-p9jm-q85p-7mcp",
"modified": "2026-08-07T17:16:13Z",
"published": "2026-08-07T17:16:13Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/netty/netty/security/advisories/GHSA-p9jm-q85p-7mcp"
},
{
"type": "WEB",
"url": "https://github.com/netty/netty/pull/17065"
},
{
"type": "WEB",
"url": "https://github.com/netty/netty/commit/5b68c61f37aa4a3045cba624cbea239655c9003b"
},
{
"type": "WEB",
"url": "https://github.com/netty/netty/commit/bb2ff68a1fb71cb4b0eb9a9e17b66c52aff680c6"
},
{
"type": "PACKAGE",
"url": "https://github.com/netty/netty"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L",
"type": "CVSS_V3"
}
],
"summary": "Netty: RedisArrayAggregator max-elements failure leaves retained partial aggregate state"
}
Drag nodes to arrange them. Double-click a related vulnerability to open it, or a source to visit its link. The Physics button on the left tunes the layout and the simulation; the legend at the bottom left hides or shows nodes by kind and by sighting type.
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.
Vulnerability timeline
Sightings and recorded dates, grouped by day · UTC
How durations are calculated
Following the VTEM framework (Black Hat USA 2015), time to patch starts when the vendor or a coordinator was informed, total time to patch at discovery, and disclosure to patch at public disclosure; vendor response time runs from the report to its acknowledgement. Time to exploit and time unpatched start at the earliest discovery or report, falling back to public disclosure. Unpatched time ends at the first recorded patch, or the calculation time when no patch is recorded. Exploitation to patch shows the interval between those two events; a published proof of concept is a public exploit, not exploitation.
Each result shows its source dates. A negative interval means the end event came first; when date-only inputs leave the order open, the result says so. Date-only inputs are approximate. The earliest evidence wins: KEV report dates and sightings are labelled as such. Records describing another vulnerability are listed but never calculated from, and a PoC alone does not establish exploitation. Missing patch evidence does not prove that no patch exists. Filters below change the event list only.
Loading timeline…
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.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.