Skip to main content
Mallory
UnratedPublic exploitVulnerability

CVE-2026-62356: Redis CMSketch RDB Loading Heap Out-of-Bounds Write

Updated 2 months ago
Trending this weekNot trendingOutside this week’s top 100
Mention activity · 30 daysNo mention activity in the last 30 days.
CVSS-Not scored
EPSS-30-day exploit probability
CISA KEVNoNot listed
In the wildNonePublic exploit, no exploitation reported
IdentifiersCVE-2026-62356CWE-787

CVE-2026-62356 is a buffer-size calculation error in Redis CMSketch handling during Redis Database (RDB) loading. A malformed or specially constructed RDB can cause Redis to perform a heap-based out-of-bounds write while loading CMSketch data.

Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
Analyst brief

Impact, mitigation & remediation

What it means and what to do now: patch path, interim mitigations, and the assume-compromise checklist.

Impact

What an attacker gets, and what they’ve been doing with it.

Successful exploitation can corrupt heap memory and may cause Redis process instability or denial of service. The available information does not establish code execution impact specifically for CVE-2026-62356.

Mitigation

If you can’t patch tonight, do this now.

Until updates can be deployed, prevent loading RDB files from untrusted or unverified sources and restrict access to Redis backup, restore, replication, and deployment workflows that could introduce attacker-controlled RDB content.

Remediation

Patch, then assume compromise.

Upgrade Redis to a fixed release: 8.2.9 or later, 8.4.6 or later, 8.6.6 or later, 8.8.2 or later, or 8.10.1 or later, according to the deployed release branch.

Public exploits

Exploits

1 exploit validated by Mallory's analysis.

View more in app →
CVE-2026-62356-POCThe repository contains 11 files: six code files, English and Chinese READMEs, a technical analysis document, a license, and .gitignore. It is a standalone exploit demonstration rather than a framework module. Repository URL, revision, and archive size were not provided; the empty metadata and zero size are placeholders, not measured values. The main Python entry point, poc/cms_overflow_crash.py, builds a malformed CMSketch RDB object, attempts to disable dump-payload sanitization, injects the object with RESTORE, and triggers CMS.QUERY. Declared width and depth of 2^32 cause the expected 4*width*depth array length to wrap to zero in 64-bit arithmetic. The retained dimensions then permit accesses beyond the supplied empty array. Cleanup deletes the selected key. Connection loss and failed PING are used as crash indicators, but do not independently establish SIGSEGV; server logs are needed. The script's predicted fault address is also not a reliable account of which row faults first. Supporting code goes beyond a simple crash reproducer. cmslib.py supplies RESP communication, serialization, a destructive FLUSHALL-based planting helper, and hash-directed out-of-bounds reads. leak_base.py queries a 1 MiB memory window through the malformed object, saves it locally, and uses pointer/SDS-header correlations to estimate the array base. This is active memory-disclosure tooling, not merely passive or read-only analysis of an existing dump. Its crash-free behavior depends on the memory layout and is not guaranteed. The two C programs brute-force MurmurHash2 preimages for selected hashes or slots. No executable write chain or RCE payload is implemented. The Bash setup script starts redis:8.8.1 and publishes host port 6381 to container port 6379. Analysis helpers instead default to 6380, while the crash script defaults to 6379. The setup script removes any existing same-named lab container and exposes the published port without a loopback-only bind. Persistent denial of service is documented for poisoned objects saved and reloaded through RDB, but persistence is not explicitly established by the main script. The claimed CVE and advisory attribution are supplied by the repository and were not independently verified. Its affected/fixed version tables conflict for the 8.2 and 8.4 branches, and broad RedisBloom affected ranges overlap listed fixed backports. Treat those ranges as claims requiring authoritative confirmation. No attacker-controlled callback, exfiltration server, reverse shell, or external HTTP request is present in the exploit code.zsy-arch · Disclosed Oct 22, 2026 · python · c · network · filePoC
FAQ

Frequently asked questions

Is CVE-2026-62356 exploited in the wild?
As of Sep 26, 2026, Mallory has no reports of CVE-2026-62356 being exploited in the wild. It is not listed in CISA KEV.
Is there a public exploit for CVE-2026-62356?
Yes. Mallory tracks 1 public exploit or proof of concept for CVE-2026-62356, the earliest disclosed Oct 22, 2026.
Is there a patch for CVE-2026-62356?
Upgrade Redis to a fixed release: 8.2.9 or later, 8.4.6 or later, 8.6.6 or later, 8.8.2 or later, or 8.10.1 or later, according to the deployed release branch.
Mallory platform

The version that knows your environment.

This page is what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which adversaries are exploiting it right now, and which detections to deploy tonight.
Exposure mappingQuery your assets running an affected version, and investigate the blast radius.
Threat actor evidenceEvery report linking this CVE to a named adversary.
Associated malwareMalware families linked to this CVE, with evidence and IOCs.
Detection signaturesYARA, Sigma, Snort, and vendor rules, auto-deployed to your SIEM.
Vendor-by-vendor mappingCross-references every affected SKU, including bundled OEM variants.
Social activity5Community discussion across Reddit, Mastodon, and other social sources.