Skip to main content
Mallory
HighPublic exploitVulnerability

CVE-2026-119618: Unauthenticated RCE in Apache HTTP Server mod_http2 assign_headers()

Published Nov 11, 2026 · Updated today
Trending this week#1of 427,843 vulnerabilities
Mention activity · 30 daysAt least 200 mentions · peak 89 in a dayToday
CVSS8.1High · v3.1
EPSS1.6%76th percentile
CISA KEVNoNot listed
In the wildNonePublic exploit, no exploitation reported
Vendor
Icing
IdentifiersCVE-2026-119618CWE-362 · Concurrent Execution using Shared…

CVE-2026-119618 is a race condition in the assign_headers() function of Apache HTTP Server's mod_http2 module. Improper synchronization between connection and stream worker threads allows concurrent access to shared header-processing memory, potentially causing heap corruption. An unauthenticated remote attacker can trigger the vulnerability through specially crafted concurrent HTTP/2 requests and potentially execute arbitrary code with the web-server worker process's privileges. Affected versions are Apache HTTP Server 2.4.55 through 2.4.70 and separately distributed mod_h2 2.0.8 through 2.0.42. Exploitation requires mod_http2 to be loaded and HTTP/2 to be enabled; HTTP/1.1-only deployments are not affected.

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 compromise the confidentiality, integrity, and availability of resources accessible to the Apache worker process. Consequences include unauthorized access to application data and credentials, modification of accessible files, and execution of attacker-controlled commands. Heap corruption can also crash worker processes and disrupt service. Further compromise of backend systems or lateral movement depends on the server's permissions, credentials, and network access; elevated operating-system privileges are not inherently granted.

Mitigation

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

Until patching is possible, disable HTTP/2 by removing h2 and h2c from every applicable server-wide and virtual-host Protocols directive, leaving Protocols http/1.1, or unload mod_http2. Restart Apache to apply the configuration and verify that HTTP/2 is no longer available. Check dependencies such as gRPC backends before disabling HTTP/2. Monitor for unexpected worker crashes, child-process execution, and outbound connections while remediation proceeds.

Remediation

Patch, then assume compromise.

Upgrade Apache HTTP Server to version 2.4.71 or later, or install a distribution or product-vendor security update that backports the fix. Restart affected Apache processes after updating. Rebuild and redeploy container images containing vulnerable components rather than merely restarting existing containers. Standalone mod_h2 users can migrate to the fixed module bundled with Apache HTTP Server; the availability of a standalone 2.0.43 fix is inconsistently documented.

Public exploits

Exploits

4 exploits validated by Mallory's analysis. Hidden: 11 rejected by Mallory's analysis as not a working exploit; 1 not yet analyzed, such as Exploit-DB entries.

View more in app →
CVE-2026-119618The supplied repository contains five files: a partially displayed README, three Python modules, and one shell script. It is a standalone HTTP/2 race-triggering and allocator-grooming tool, not a framework module. The README attributes the issue to CVE-2026-119618 and claims that assign_headers() modifies a shared header table rather than the request-private table. Those vulnerability identifiers, version ranges, severity claims, and upstream fixes are repository assertions, not independently verified findings. exploit.py is the main driver. It constructs GET requests with Content-Type but no Content-Length, fills the header table near modeled allocation boundaries, primes streams, and repeatedly sends concurrent carrier POSTs and trigger GETs followed by stream resets. Its default configuration creates connections × procs worker processes. It accepts custom target paths, header padding, payload files, duration, and concurrency settings. The --check mode sends one request and asks the operator to inspect access logs; it does not automatically determine vulnerability. groom.py models APR table growth and allocator size classes, supplies the repeated-byte payload, and includes an unused forged-entry serializer. h2.py implements minimal HTTP/2 framing and literal HPACK encoding using only Python's standard library, supporting cleartext prior knowledge and TLS with ALPN. post.sh is an independent local reconnaissance script that inventories identity, sudo permissions, SUID files, capabilities, cron paths, web roots, configuration filenames, and Docker access. It neither escalates privileges nor runs automatically after the Python driver. Despite the driver's 'weaponised' description, the supplied code does not implement a verified memory overlap, address disclosure, arbitrary memory primitive, control-flow transfer, or remote command execution. Its basic supplied payload makes it operational as a traffic-generation and corruption-attempt tool, not as a demonstrated RCE chain. Connection drops and GOAWAY frames are weak failure indicators rather than proof of successful exploitation. Additional limitations include simplified allocator assumptions, incomplete HTTP/2 flow-control handling, no DATA-frame splitting despite accepting oversized payloads, and request counters that overcount when carrier bodies are disabled. No fixed external callback, exfiltration destination, or command-and-control endpoint appears in the code; README badge and advisory URLs are documentation only. Repository URL, revision, and archive size were not supplied; the five listed file sizes total 30,174 bytes, which is not an archive-size measurement.LLCForge · Disclosed Nov 12, 2026 · python · sh · networkOperationalCVE-2026-119618The supplied repository contains a standalone Python exploit and a differential container lab for the claimed CVE-2026-119618. Its implemented impact is unauthenticated denial of service through heap corruption; although the documentation references an upstream RCE advisory, no RCE chain is included. The described root cause is assign_headers() inserting a synthesized Content-Length: 0 into req->headers, owned by the HTTP/2 stream pool, instead of request-private r->headers_in. When this insertion grows a full table and exhausts the pool's allocation fast path, a worker can enter the same mutex-free APR allocator being used by the connection thread. Concurrent allocation and freeing can corrupt allocator state. The default request shape fills an 80-entry table, and padding changes whether growth reaches the shared allocator. A missed race is explicitly distinguished from an unaffected target. poc/h2.py implements a standard-library-only cleartext HTTP/2 client, including stateless HPACK literal encoding, SETTINGS negotiation, HEADERS/CONTINUATION fragmentation, DATA, resets, and response monitoring. poc/poc.py provides check, exploit, probe, and mitigation modes, configurable traffic generation, header-table arithmetic, and machine-readable summaries. Functional classification uses the lab access log: the synthesized Content-Length appears as absent on the vulnerable build and as zero on the fixed build. This is not shown to be a universal response-only remote detector. run-lab.sh and the Makefile orchestrate vulnerable 2.4.70, fixed 2.4.71, and HTTP/2-disabled containers, parameter sweeps, and matched repeated experiments. lab/ holds the server image, pinned dependency installer, source-build script, runtime entrypoint, and diagnostic logging configuration. The lab is documented as publishing no host ports. patches/ contains C diagnostic changes for allocator-concurrency assertions and an optional race-window delay; these are laboratory instruments, not fixes. tests/ contains two Python unittest modules covering framing, HPACK, request construction, table arithmetic, reporting, and exit codes. The 18-file inventory includes four Python files, four Bash scripts, two container recipes, one Makefile, two C patch files, one Apache configuration template, and four metadata/documentation files. The code-file count excludes patches and configuration. No malicious callback, exfiltration destination, or operator-destructive command is evident; cleanup commands shown relate to container builds and lab lifecycle. Public advisory and attribution URLs are references rather than exploit callbacks. Several substantial files are truncated, so full control-flow verification was not possible. No execution or independent validation of the vulnerability, published results, or version claims was performed. The original repository URL, git reference, and archive size were not supplied; empty strings and zero represent those unavailable metadata values.d4ytox · Disclosed Nov 14, 2026 · python · bash · networkOperationalCVE-2026-119618This is a standalone differential research laboratory and crash-oriented proof of concept, not a weaponized RCE exploit. Its claimed CVE is CVE-2026-119618. Documentation lists affected httpd 2.4.55–2.4.70 and standalone mod_h2 2.0.8–2.0.42, with fixes in httpd 2.4.71 and mod_h2 2.0.43. The supplied laboratory specifically compares httpd 2.4.70 with 2.4.71; these CVE and release claims were not independently verified. The described defect is a worker-thread write in assign_headers(): a bodyless request with Content-Type but no Content-Length causes synthesized Content-Length: 0 to be inserted into req->headers rather than request-private r->headers_in. When the shared table is full, growth can enter the connection's unsynchronized APR allocator concurrently with connection-thread allocation or freeing. The documented fix changes the destination table to r->headers_in. The 44-file inventory contains six Python files and three shell scripts, plus a Makefile, two container build recipes, configuration, documentation, and captured evidence. poc/poc.py supplies check, exploit, probe, and mitigation modes. poc/h2.py implements a minimal standard-library cleartext HTTP/2 client and stateless HPACK literal encoding. run-lab.sh and Makefile orchestrate Podman targets and evidence capture. lab/ builds upstream sources, configures servers, and optionally instruments APR/mod_http2. exp/ contains diagnostic-log analysis, TSan classification, and a functional differential helper. evidence/ contains access/error logs, sanitizer traces, test results, version records, and image metadata. The code_files count excludes declarative build recipes and configuration. Capabilities include configurable connection/stream concurrency, duration, header fill and padding, multiple HTTP/2 request shapes, shared-log inspection, fixed-build sweeps, and workaround testing. Supplied evidence reports multiple vulnerable child crashes and 31 allocator-race TSan reports on 2.4.70 versus zero matching allocator-race reports on 2.4.71. The fixed TSan run still contains eight reports classified as other, so it is not globally sanitizer-clean. Negative traffic results do not prove absence of a timing-dependent flaw. Remote HTTP/2 refusal likewise does not alone prove mitigation rather than an unrelated connection failure. The supplied client does not implement TLS, and reliable classification depends on laboratory log access. No controlled instruction-pointer takeover, shellcode, reverse shell, persistence, exfiltration, or command-execution payload is shown. Network destinations are configurable targets or legitimate build dependencies; no command-and-control destination is evident. Cleanup commands concern container/build state and are not evidence of a fake exploit. Several source files are truncated in the provided material, limiting complete implementation validation. Repository archive size was not supplied; size_bytes=0 is an unknown-value placeholder, not a measured archive size.googlesky · Disclosed Nov 12, 2026 · python · bash · networkPoCCVE-2026-119618---RCE-in-APACHEStandalone Python proof-of-concept repository containing README.md, poc.py, and h2.py. README.md describes the claimed CVE, laboratory requirements, four operating modes, exit codes, and remediation. poc.py is the CLI entry point, advertising check, exploit, probe, and mitigation modes. Its documented traffic defaults are 16 concurrent connections, 32 streams per connection, and a 20-second run. Visible code handles target parsing, configurable log locations, sanitizer-report discovery, and CLI dispatch. h2.py implements a minimal standard-library HTTP/2 client over raw TCP, including the connection preface, SETTINGS exchange, stateless HPACK literal encoding, HEADERS/CONTINUATION framing, DATA, resets, flow-control updates, and basic response handling. It does not implement TLS. The claimed issue is a race in mod_http2 assign_headers(): synthesizing Content-Length: 0 in a shared stream-header table can cause concurrent APR allocation and heap corruption. The repository identifies commit 824beab9b9424d6cfb29595248fdbf6c3dddeab1 and Apache HTTP Server 2.4.71 as the fix; these claims and the CVE identifier are not independently verified by the supplied material. No command-execution payload, persistence, or exfiltration destination is visible. Important limitations: the supplied poc.py content is explicitly truncated, omitting the main request-shape and traffic-generation implementations, so their operation cannot be fully validated. The report_kind documentation also treats any AddressSanitizer memory error during testing as attributable to this CVE, which is not sufficient causal evidence by itself. Likewise, HTTP/2 refusal or a run without corruption does not conclusively establish remediation. Repository URL, revision, and archive size were not supplied; empty metadata strings and size 0 represent unavailable values. The three listed file sizes total 39,299 bytes, which is not a measured archive size.Doseuser · Disclosed Nov 16, 2026 · python · networkPoC
Exposure surface

Affected products & vendors

Products and vendors Mallory has correlated with this vulnerability. Open in Mallory to drill down to specific CPE configurations and version ranges.

VendorProductType
IcingMod Http2application
FAQ

Frequently asked questions

Is CVE-2026-119618 exploited in the wild?
As of Nov 17, 2026, Mallory has no reports of CVE-2026-119618 being exploited in the wild. It is not listed in CISA KEV.
Is there a public exploit for CVE-2026-119618?
Yes. Mallory tracks 4 public exploits or proofs of concept for CVE-2026-119618, the earliest disclosed Nov 12, 2026.
Is there a patch for CVE-2026-119618?
Upgrade Apache HTTP Server to version 2.4.71 or later, or install a distribution or product-vendor security update that backports the fix. Restart affected Apache processes after updating.
What products are affected by CVE-2026-119618?
Icing Mod Http2.
How severe is CVE-2026-119618?
CVE-2026-119618 has a CVSS v3.1 score of 8.1 (High) and an EPSS score of 1.6% (76th percentile) as of Nov 17, 2026.
Activity feed

Recent activity

334 sources tracked across advisories, community write-ups, and news. New activity surfaces here as Mallory finds it.

View more in app →
xakepNews · Nov 17, 2026Критический баг в Apache HTTP Server позволяет выполнять код без аутентификации - ХакерA critical race condition in mod_http2's assign_headers() function can cause heap corruption and unauthenticated remote code execution with web-server process privileges, without user interaction. It affects Apache HTTP Server 2.4.55–2.4.70 and standalone mod_h2 2.0.8–2.0.42, but only when HTTP/2 is enabled. Red Hat assigned a CVSS score of 9.8. Researchers demonstrated exploitation, although no public exploit is available and Wiz reported no evidence of exploitation in real attacks. Apache HTTP Server 2.4.71 fixes the issue. As a temporary mitigation, remove h2 and h2c from the Protocols directive, leaving only http/1.1.Read more ↗orca security blogNews · Nov 16, 2026CVE-2026-119618: Critical Apache HTTP/2 Race Condition Enables Unauthenticated Remote Code Execution| Orca SecurityA critical race condition in mod_http2's assign_headers() function causes heap memory corruption, allowing unauthenticated attackers to execute arbitrary code with web server process privileges through crafted HTTP/2 requests. It affects Apache HTTP Server 2.4.55–2.4.70 and standalone mod_h2 2.0.8–2.0.42 when HTTP/2 is enabled server-wide or on a virtual host. The article reports a reliable public exploit but no confirmed in-the-wild exploitation. Apache HTTP Server 2.4.71 fixes the vulnerability; standalone mod_h2 users must migrate to the bundled module. Disabling HTTP/2 by removing h2 and h2c from all applicable Protocols directives is a temporary mitigation.Read more ↗cyberveilleNews · Nov 16, 2026🪲 Semaine 46 - CVE les plus discutées | CyberVeilleImproper synchronization in mod_http2's assign_headers() function can lead to remote code execution. Affected versions are Apache HTTP Server 2.4.55–2.4.70 and mod_h2 2.0.8–2.0.42. The page reports published proof-of-concept material and recommends upgrading to corrected versions, but does not explicitly report exploitation in the wild.Read more ↗malware newsNews · Nov 16, 202616th November - Threat Intelligence Report - Malware News - Malware Analysis, News and IndicatorsA critical vulnerability with a CVSS score of 9.8 affecting Apache HTTP Server 2.4.55 through 2.4.70 when HTTP/2 is enabled. It allows unauthenticated remote arbitrary code execution and is fixed in version 2.4.71.Read more ↗
+40 more in the timeline. Sign up to see them
Do you run this?See affected versions in your environment and whether adversaries are exploiting this vulnerability against assets like yours.
Check your exposure
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 activity151Community discussion across Reddit, Mastodon, and other social sources.