Apache HTTP Server versions 2.4.6 to 2.4.46 mod_proxy_wstunnel configured on an URL that is not necessarily Upgraded by the origin server was tunneling the whole connection regardless, thus allowing for subsequent requests on the same connection to pass through with no HTTP validation, authentication or authorization possibly configured.
Apache HTTP Server versions 2.4.41 to 2.4.46 mod_proxy_http can be made to crash (NULL pointer dereference) with specially crafted requests using both Content-Length and Transfer-Encoding headers, leading to a Denial of Service
Apache HTTP Server versions 2.4.0 to 2.4.46 A specially crafted Digest nonce can cause a stack overflow in mod_auth_digest. There is no report of this overflow being exploitable, nor the Apache HTTP Server team could create one, though some particular compiler and/or compilation option might make it possible, with limited consequences anyway due to the size (a single byte) and the value (zero byte) of the overflow
Apache HTTP Server versions 2.4.0 to 2.4.46 A specially crafted Cookie header handled by mod_session can cause a NULL pointer dereference and crash, leading to a possible Denial Of Service
Apache HTTP Server versions 2.4.0 to 2.4.46 A specially crafted SessionHeader sent by an origin server could cause a heap overflow
Apache HTTP Server versions 2.4.39 to 2.4.46 Unexpected matching behavior with 'MergeSlashes OFF'
Apache HTTP Server protocol handler for the HTTP/2 protocol checks received request headers against the size limitations as configured for the server and used for the HTTP/1 protocol as well. On violation of these restrictions and HTTP response is sent to the client with a status code indicating why the request was rejected.
This rejection response was not fully initialised in the HTTP/2 protocol handler if the offending header was the very first one received or appeared in a a footer. This led to a NULL pointer dereference on initialised memory, crashing reliably the child process. Since such a triggering HTTP/2 request is easy to craft and submit, this can be exploited to DoS the server.
The notice on the website and the mail announcement for CVE-2021-31618 say different versions, the website saying <=2.4.48 is affected and the mail saying <2.4.47 is affected. Unsure which are actually affected.
2.4.47 is the one with the fixes:
Please stabilize >=2.4.47.
Sanity check failed:
> pdepend amd64 dev profile default/linux/amd64/17.0/x32 (40 total)
> pdepend amd64 stable profile default/linux/amd64/17.1 (63 total)
All sanity-check issues have been resolved
Changes with Apache 2.4.48
*) SECURITY: CVE-2021-31618 (cve.mitre.org)
mod_http2: Fix a potential NULL pointer dereference [Ivan Zhakov]
Unable to check for sanity:
> no match for package: app-admin/apache-tools-2.4.48
Unable to check for sanity:
> no match for package: www-servers/apache-2.4.48-r1
all arches done
GLSA request filed.
This issue was resolved and addressed in
GLSA 202107-38 at https://security.gentoo.org/glsa/202107-38
by GLSA coordinator John Helmert III (ajak).
The bug has been referenced in the following commit(s):
Author: Hans de Graaff <firstname.lastname@example.org>
AuthorDate: 2021-07-18 09:33:51 +0000
Commit: Hans de Graaff <email@example.com>
CommitDate: 2021-07-18 09:33:51 +0000
www-servers/apache: cleanup vulnerable versions
Package-Manager: Portage-3.0.20, Repoman-3.0.2
Signed-off-by: Hans de Graaff <firstname.lastname@example.org>
www-servers/apache/Manifest | 3 -
www-servers/apache/apache-2.4.46-r7.ebuild | 246 ----------------------------
www-servers/apache/apache-2.4.46-r8.ebuild | 253 -----------------------------
www-servers/apache/apache-2.4.47-r1.ebuild | 253 -----------------------------
4 files changed, 755 deletions(-)
Thanks graaff, not sure how I missed this