DEV Community

Cover image for Apache HTTP/2 vulnerability CVE-2026-119618: find every h2 line, upgrade to 2.4.71, check the running process
System Design Simple
System Design Simple

Posted on Originally published at systemdesignsimple.com

Apache HTTP/2 vulnerability CVE-2026-119618: find every h2 line, upgrade to 2.4.71, check the running process

The Apache HTTP/2 vulnerability CVE-2026-119618 is a critical mod_http2 flaw in Apache HTTP Server 2.4.55 through 2.4.70 that, according to the Apache advisory of 2026-11-11, can lead to remote code execution on a build that offers HTTP/2 (h2 or h2c). Upgrade to 2.4.71 or your distro's fixed httpd and mod_http2 packages, do a full restart, and confirm the process serving traffic loaded the new build. If you cannot upgrade today, remove h2 and h2c from every effective Protocols line.

This guide is for Apache servers you own or maintain.

CVE-2026-119618 (Apache advisory, 2026-11-11)
Component mod_http2, assign_headers() in h2_request.c
Affected httpd 2.4.55 through 2.4.70; mod_h2 (GitHub) 2.0.8 through 2.0.42
Fixed httpd 2.4.71, commit 824beab
Workaround named none

Am I affected? How do I check for the Apache HTTP/2 vulnerability?

You are affected when three things hold for the running server: the httpd or mod_http2 build is in the range without a distro backport, http2_module is loaded, and an effective Protocols line offers h2 (HTTP/2 over TLS) or h2c (HTTP/2 over cleartext TCP). That line can sit in the main config, one <VirtualHost>, or a file pulled in by Include or IncludeOptional.

Run the dumps the way the service runs httpd: same executable, same -f, -d and -D arguments, same environment. Read those from the unit first. On Debian, apachectl loads /etc/apache2/envvars for you; a bare apache2 -M does not.

# how the service really starts httpd (RHEL: httpd instead of apache2)
systemctl cat apache2 | grep -E 'ExecStart|EnvironmentFile'

# version, loaded modules and include tree, with the service's arguments added
apachectl -v
apachectl -M 2>/dev/null | grep http2_module
apachectl -t -D DUMP_INCLUDES

# candidate Protocols lines in every included file
apachectl -t -D DUMP_INCLUDES 2>/dev/null | grep -oE '/[^ ]+$' | xargs grep -niE '^\s*protocols\s'
Enter fullscreen mode Exit fullscreen mode

The grep finds candidates. A hit inside <IfModule> or <IfDefine> only counts when that condition is true for this service, so read each hit in its scope.

On packaged builds the module may not come from the httpd package. On RHEL, Rocky, Alma and Fedora, mod_http2 ships as its own mod_http2 package, so find the owner of the loaded .so and compare that revision with the distro advisory:

rpm -qf /usr/lib64/httpd/modules/mod_http2.so   # RHEL family: usually mod_http2-<rev>
rpm -q httpd mod_http2
dpkg -S mod_http2.so && dpkg-query -W apache2-bin   # Debian, Ubuntu: compare with the DSA/USN
apk info --who-owns /usr/lib/apache2/mod_http2.so   # Alpine: apache2-http2
Enter fullscreen mode Exit fullscreen mode

Fix it now: how do I patch CVE-2026-119618?

# Debian, Ubuntu
sudo apt-get update && sudo apt-get install --only-upgrade apache2 apache2-bin

# RHEL, Rocky, Alma, Fedora: upgrade both packages
sudo dnf upgrade httpd mod_http2

# Alpine
sudo apk upgrade apache2 apache2-http2

# source build or mod_h2 from GitHub: build httpd 2.4.71, or mod_h2 newer than 2.0.42
Enter fullscreen mode Exit fullscreen mode

For the official container image, pulling changes nothing that is already running. Pull, recreate the container the way your deployment does it, then check inside the replacement:

docker pull httpd:2.4
# recreate: docker compose up -d, a rollout, or docker run with your usual flags
docker exec "$container_name" httpd -v     # want 2.4.71 or later
Enter fullscreen mode Exit fullscreen mode

On a host, test the config and do a full stop and start. A graceful restart replaces the children but keeps the parent, which can keep the old binary mapped.

sudo apachectl -t && sudo systemctl restart apache2     # httpd on RHEL

pid=$(systemctl show -p MainPID --value apache2)
ps -o lstart=,args= -p "$pid"                        # start time later than the package install
sudo ls -l /proc/"$pid"/exe                          # "(deleted)" = stale binary still running
sudo grep -E '\.so.*\(deleted\)' /proc/"$pid"/maps   # stale modules
Enter fullscreen mode Exit fullscreen mode

A (deleted) mapping means the process still runs files the package manager replaced. An empty result only means nothing stale is mapped. Whether the build is fixed comes from the package revision check above.

Why it is dangerous: how does the mod_http2 header write go wrong?

An HTTP/2 stream is one request and response carried on a shared connection. mod_http2 reads each stream on the main connection and collects its headers into an h2_request allocated from the stream's pool (h2_stream.c in 2.4.70). An APR pool is the arena allocator that owns an httpd object's memory. The stream is then processed on a secondary connection, where c2_process() builds an ordinary request_rec.

assign_headers() does that hand-off. It first clones the stream's header table into r->headers_in, which the request owns. One later branch, for a request with a Content-Type, no Content-Length and no body, sets Content-Length: 0. In 2.4.70 that write went to req->headers, the table still owned by the stream. The fix commit points it at r->headers_in.

So request-side code changes stream-side state. APR tables have no locking, and adding an entry can allocate from the owning pool. The advisory classifies this as concurrent use of a shared resource without proper synchronization and says it can lead to RCE, which here means a remote client's code running as the httpd worker user. It does not describe the corruption sequence, and we do not claim one.

Reproduce

Our lab, apache-http2-check-lab, classifies configurations; it never starts httpd or sends a request. It builds 15 throwaway hosts, each with a sample httpd -v, apachectl -V and httpd.conf, and follows Include into directories and wildcards. In our test on 2026-11-13 (Node 22.13.1, 305 ms), it classified all 8 synthetic in-range fixtures with h2 or h2c as affected, including h2 only inside vhost *:8443 and h2c reached through Include conf/extra, a directory. No vulnerability reproduction was performed.

Experiment

fixture (version, setting) before after
2.4.62, Protocols h2 h2c http/1.1 affected patched (2.4.71)
2.4.55 and 2.4.70 (Debian), h2 on affected patched
2.4.66, h2 only in vhost *:8443 affected mitigated
2.4.64, h2c only affected mitigated
2.4.58 / 2.4.60, h2 or h2c via IncludeOptional / Include affected mitigated
2.4.65, loadmodule HTTP2_module, H2 (parser case) affected mitigated
2.4.71 and 2.4.72, h2 on patched patched
2.4.62, module commented out, no or commented Protocols, or h2 without the module mitigated mitigated
2.4.54, h2 on outside outside

Fix

Each "after" applies one change: the upgrade, or removing h2 and h2c from every Protocols line. A half fix that dropped h2 but kept h2c stayed affected, since h2c still reaches mod_http2.

Report

in our test on 2026-11-13 result
synthetic fixtures classified affected 8/8 (one parser case)
classified patched or mitigated after changes 8/8
control fixtures unchanged 7/7
vulnerability reproduction none performed
this Mac Apache 2.4.66, in range; mod_http2 not loaded, 0 vhosts with h2 or h2c: mitigated

The lab never runs httpd, so it shows neither the race nor that 2.4.71 removes it; that rests on the advisory and the commit. Its parser lowercases module and protocol names, so the HTTP2_module fixture describes a config a real httpd would refuse to load: count 7 real configurations plus one parser case. The 2026-11-13 run also let the last Protocols line in a scope replace earlier ones, which httpd does not do; the lab code now adds them up. It reads files on disk, not the running process or distro changelogs.

If you cannot update today: how do I verify HTTP/2 is disabled?

The advisory names no workaround. Turning HTTP/2 off reduces exposure, because no connection reaches mod_http2:

# main config AND every <VirtualHost>, including Include files
Protocols http/1.1
Enter fullscreen mode Exit fullscreen mode

Repeated Protocols directives in one scope accumulate protocols. Adding Protocols http/1.1 after Protocols h2 leaves h2 enabled; edit or remove every earlier h2/h2c entry (set_protocols() in core.c).

Or unload the module (sudo a2dismod http2 on Debian, or comment out LoadModule http2_module), then apachectl -t, a full restart, and apachectl -M showing no http2_module. Then test each origin listener directly, not through a CDN or proxy. --resolve keeps Host and SNI (the hostname a TLS client sends) correct; --noproxy '*' skips proxy variables.

# TLS: prints 2 before the change, 1.1 after
curl --noproxy '*' -sS -o /dev/null -w '%{http_version}
' --http2 \
  --resolve app.example.com:443:10.0.0.5 https://app.example.com/; echo "exit $?"

# cleartext Upgrade: 2 before, 1.1 after
curl --noproxy '*' -sS -o /dev/null -w '%{http_version}
' --http2 \
  --resolve app.example.com:80:10.0.0.5 http://app.example.com/; echo "exit $?"

# direct h2c (prior knowledge): 2 before; after, expect a failed transfer and a nonzero exit
curl --noproxy '*' -sS -o /dev/null -w '%{http_version}
' --http2-prior-knowledge \
  --resolve app.example.com:80:10.0.0.5 http://app.example.com/; echo "exit $?"
Enter fullscreen mode Exit fullscreen mode

The direct test sends HTTP/2 with no HTTP/1.1 step, so a server without h2c answers it with an error rather than 1.1 (curl docs). H2Upgrade and H2Direct control the two cleartext paths separately, so run both. A connection refused or certificate error is inconclusive. If the before-run already prints 1.1, check curl -V | grep HTTP2.

Timeline and sources

Date Event
2026-10-26 reported to the Apache security team (advisory credit: Anthropic's OSS Scanner)
2026-11-03 reported again (advisory credit: Google Cloud Vulnerability Research)
2026-11-11 fixed by commit 824beab; 2.4.71 released
2026-11-12 news coverage (The Hacker News)
2026-11-13 our lab run

For holding a fixed release you cannot deploy yet, the OWASP Vulnerable Dependency Management cheat sheet covers the cases. Directive details are in the Protocols docs.

Our take

assign_headers() clones the stream's headers so the request owns private state, then one write goes to the original. The const on req did not catch it, because the table pointer inside the struct is not const. If you maintain C modules that hand data between threads, search for writes through the source struct after a clone.

On RHEL-family hosts, httpd -v and rpm -q httpd describe one package while the vulnerable code can live in another. A ticket that checks only the server version can close with the module unchanged. We would close it only when the package owning the loaded mod_http2.so matches the distro advisory and the restarted PID shows no stale mappings.

Our own lab got httpd's parsing wrong in two places. Any config scanner carries that risk, since its verdicts depend on how closely it copies the server's parser. We now start from DUMP_INCLUDES, the include tree httpd resolves itself, and end with a direct curl to each origin listener. The OpenSSH bypass check used the same /proc step, the Apache OFBiz check found a version file that could not tell patched from unpatched, and the Node.js Brotli check follows the same order.

Top comments (0)