A malicious HTTP/2 client which rapidly creates requests and immediately resets them can cause excessive server resource consumption. While the total number of requests is bounded by the http2.Server.MaxConcurrentStreams setting, resetting an in-progress request allows the attacker to create a new request while the existing one is still executing.

With the fix applied, HTTP/2 servers now bound the number of simultaneously executing handler goroutines to the stream concurrency limit (MaxConcurrentStreams). New requests arriving when at the limit (which can only happen after the client has reset an existing, in-flight request) will be queued until a handler exits. If the request queue grows too large, the server will terminate the connection.

This issue is also fixed in golang.org/x/net/http2 for users manually configuring HTTP/2.

The default stream concurrency limit is 250 streams (requests) per HTTP/2 connection. This value may be adjusted using the golang.org/x/net/http2 package; see the Server.MaxConcurrentStreams setting and the ConfigureServer function.

HTTP/2 Rapid reset attack The HTTP/2 protocol allows clients to indicate to the server that a previous stream should be canceled by sending a RSTSTREAM frame. The protocol does not require the client and server to coordinate the cancellation in any way, the client may do it unilaterally. The client may also assume that the cancellation will take effect immediately when the server receives the RSTSTREAM frame, before any other data from that TCP connection is processed.

Abuse of this feature is called a Rapid Reset attack because it relies on the ability for an endpoint to send a RSTSTREAM frame immediately after sending a request frame, which makes the other endpoint start working and then rapidly resets the request. The request is canceled, but leaves the HTTP/2 connection open.

The HTTP/2 Rapid Reset attack built on this capability is simple: The client opens a large number of streams at once as in the standard HTTP/2 attack, but rather than waiting for a response to each request stream from the server or proxy, the client cancels each request immediately.

The ability to reset streams immediately allows each connection to have an indefinite number of requests in flight. By explicitly canceling the requests, the attacker never exceeds the limit on the number of concurrent open streams. The number of in-flight requests is no longer dependent on the round-trip time (RTT), but only on the available network bandwidth.

In a typical HTTP/2 server implementation, the server will still have to do significant amounts of work for canceled requests, such as allocating new stream data structures, parsing the query and doing header decompression, and mapping the URL to a resource. For reverse proxy implementations, the request may be proxied to the backend server before the RSTSTREAM frame is processed. The client on the other hand paid almost no costs for sending the requests. This creates an exploitable cost asymmetry between the server and the client.

Multiple software artifacts implementing HTTP/2 are affected. This advisory was originally ingested from the swift-nio-http2 repo advisory and their original conent follows.

swift-nio-http2 specific advisory swift-nio-http2 is vulnerable to a denial-of-service vulnerability in which a malicious client can create and then reset a large number of HTTP/2 streams in a short period of time. This causes swift-nio-http2 to commit to a large amount of expensive work which it then throws away, including creating entirely new Channels to serve the traffic. This can easily overwhelm an EventLoop and prevent it from making forward progress.

swift-nio-http2 1.28 contains a remediation for this issue that applies reset counter using a sliding window. This constrains the number of stream resets that may occur in a given window of time. Clients violating this limit will have their connections torn down. This allows clients to continue to cancel streams for legitimate reasons, while constraining malicious actors.

.NET and Visual Studio Denial of Service Vulnerability

Netlogon RPC Elevation of Privilege Vulnerability

Windows Kerberos RC4-HMAC Elevation of Privilege Vulnerability

Windows Kerberos Elevation of Privilege Vulnerability

An out-of-bounds read vulnerability exists due to a boundary error when reading SONMP packets. A remote user can send specially crafted packets to the application, trigger a heap-based buffer overflow read and leak memory values from lldpd application or crash it.

Reference:

https://www.cybersecurity-help.cz/vdb/SB2021111808 https://github.com/lldpd/lldpd/commit/73d42680fce8598324364dbb31b9bc3b8320adf7 https://lldpd.github.io/security.html

A flaw was found in the grafana package. Auth proxy allows authentication of a user by only providing the username (or email) in an X-WEBAUTH-USER HTTP header. The trust assumption is that a front proxy will take care of authentication and that the Grafana server is only publicly reachable with this front proxy.

CVE-2023-45145 - The wrong order of listen(2) and chmod(2) calls creates a race condition that can be used by another process to bypass desired Unix socket permissions on startup.

Upstream have released version 7.0.14/7.2.2 to fix CVE-2023-45145.

Reference: - https://github.com/redis/redis/releases/tag/7.0.14 - https://bugs.mageia.org/32406

CVE-2022-24834 - A specially crafted Lua script executing in Redis can trigger a heap overflow in the cjson and cmsgpack libraries, and result in heap corruption and potentially remote code execution. The problem exists in all versions of Redis with Lua scripting support, starting from 2.6, and affects only authenticated and authorized users.

https://github.com/redis/redis/security/advisories/GHSA-p8x2-9v9q-c838

A vulnerability was found in Redis. This flaw allows authenticated users to use the HINCRBYFLOAT command to create an invalid hash field that may crash Redis on access.

A security vulnerability was found in zlib. The flaw triggered a heap-based buffer in inflate in the inflate.c function via a large gzip header extra field. This flaw is only applicable in the call inflateGetHeader.

Impact

using string-to-date parsing in moment (more specifically rfc2822 parsing, which is tried by default) has quadratic (N^2) complexity on specific inputs noticeable slowdown is observed with inputs above 10k characters users who pass user-provided strings without sanity length checks to moment constructor are vulnerable to (Re)DoS attacks

Patches The problem is patched in 2.29.4, the patch can be applied to all affected versions with minimal tweaking.

Workarounds In general, given the proliferation of ReDoS attacks, it makes sense to limit the length of the user input to something sane, like 200 characters or less. I haven't seen legitimate cases of date-time strings longer than that, so all moment users who do pass a user-originating string to constructor are encouraged to apply such a rudimentary filter, that would help with this but also most future ReDoS vulnerabilities.

References There is an excellent writeup of the issue here: https://github.com/moment/moment/pull/6015#issuecomment-1152961973=

Details The issue is rooted in the code that removes legacy comments (stuff inside parenthesis) from strings during rfc2822 parsing. moment("(".repeat(500000)) will take a few minutes to process, which is unacceptable.

A use-after-free flaw was found in the Expat package, caused by destruction of a shared DTD in XMLExternalEntityParserCreate in out-of-memory situations. This may lead to availability disruptions.

A flaw in XML parsing could have led to a use-after-free causing a potentially exploitable crash.In official releases of Firefox this vulnerability is mitigated by wasm sandboxing; versions managed by Linux distributions may have other settings.

A vulnerability exists in Async through 3.2.1 for 3.x and through 2.6.3 for 2.x (fixed in 3.2.2 and 2.6.4), which could let a malicious user obtain privileges via the mapValues() method.

A specially crafted web page can abuse this vulnerability to cause memory corruption and potentially arbitrary code execution. A user would need to visit a malicious webpage to trigger this vulnerability.

Reference: https://webkitgtk.org/security/WSA-2023-0009.html#CVE-2023-39928

NGINX Open Source before versions 1.23.2 and 1.22.1, NGINX Open Source Subscription before versions R2 P1 and R1 P1, and NGINX Plus before versions R27 P1 and R26 P1 have a vulnerability in the module ngxhttpmp4module that might allow a local attacker to cause a worker process crash, or might result in worker process memory disclosure by using a specially crafted audio or video file. The issue affects only NGINX products that are built with the module ngxhttpmp4module, when the mp4 directive is used in the configuration file. Further, the attack is possible only if an attacker can trigger processing of a specially crafted audio or video file with the module ngxhttpmp4module.

NGINX Open Source before versions 1.23.2 and 1.22.1, NGINX Open Source Subscription before versions R2 P1 and R1 P1, and NGINX Plus before versions R27 P1 and R26 P1 have a vulnerability in the module ngxhttpmp4module that might allow a local attacker to corrupt NGINX worker memory, resulting in its termination or potential other impact using a specially crafted audio or video file. The issue affects only NGINX products that are built with the ngxhttpmp4module, when the mp4 directive is used in the configuration file. Further, the attack is possible only if an attacker can trigger processing of a specially crafted audio or video file with the module ngxhttpmp4module.

A flaw has been identified in glibc. In an extremely rare situation, the getaddrinfo function may access memory that has been freed, resulting in an application crash. This issue is only exploitable when a NSS module implements only the nssgethostbyname2r and nssgetcanonnamer hooks without implementing the nssgethostbyname3r hook. The resolved name should return a large number of IPv6 and IPv4, and the call to the getaddrinfo function should have the AFINET6 address family with AICANONNAME, AIALL and AIV4MAPPED as flags.

Accessibility. A privacy issue was addressed with improved private data redaction for log entries.

A buffer overflow was discovered in the GNU C Library's dynamic loader ld.so while processing the GLIBCTUNABLES environment variable. This issue could allow a local attacker to use maliciously crafted GLIBCTUNABLES environment variables when launching binaries with SUID permission to execute code with elevated privileges.

A flaw was found in glibc. When the getaddrinfo function is called with the AFUNSPEC address family and the system is configured with no-aaaa mode via /etc/resolv.conf, a DNS response via TCP larger than 2048 bytes can potentially disclose stack contents through the function returned address data, and may cause a crash.

Impact

Since Requests v2.3.0, Requests has been vulnerable to potentially leaking Proxy-Authorization headers to destination servers, specifically during redirects to an HTTPS origin. This is a product of how rebuildproxies is used to recompute and reattach the Proxy-Authorization header to requests when redirected. Note this behavior has only been observed to affect proxied requests when credentials are supplied in the URL user information component (e.g. https://username:password@proxy:8080).

Current vulnerable behavior(s):

1. HTTP → HTTPS: leak 2. HTTPS → HTTP: no leak 3. HTTPS → HTTPS: leak 4. HTTP → HTTP: no leak

For HTTP connections sent through the proxy, the proxy will identify the header in the request itself and remove it prior to forwarding to the destination server. However when sent over HTTPS, the Proxy-Authorization header must be sent in the CONNECT request as the proxy has no visibility into further tunneled requests. This results in Requests forwarding the header to the destination server unintentionally, allowing a malicious actor to potentially exfiltrate those credentials.

The reason this currently works for HTTPS connections in Requests is the Proxy-Authorization header is also handled by urllib3 with our usage of the ProxyManager in adapters.py with proxymanagerfor. This will compute the required proxy headers in proxyheaders and pass them to the Proxy Manager, avoiding attaching them directly to the Request object. This will be our preferred option going forward for default usage.

Patches Starting in Requests v2.31.0, Requests will no longer attach this header to redirects with an HTTPS destination. This should have no negative impacts on the default behavior of the library as the proxy credentials are already properly being handled by urllib3's ProxyManager.

For users with custom adapters, this may be potentially breaking if you were already working around this behavior. The previous functionality of rebuildproxies doesn't make sense in any case, so we would encourage any users impacted to migrate any handling of Proxy-Authorization directly into their custom adapter.

Workarounds For users who are not able to update Requests immediately, there is one potential workaround.

You may disable redirects by setting allowredirects to False on all calls through Requests top-level APIs. Note that if you're currently relying on redirect behaviors, you will need to capture the 3xx response codes and ensure a new request is made to the redirect destination. import requests r = requests.get('http://github.com/', allowredirects=False)

Credits

This vulnerability was discovered and disclosed by the following individuals.

Dennis Brinkrolf, Haxolot (https://haxolot.com/) Tobias Funke, (tobiasfunke93@gmail.com)

Cookie HTTP header isn't stripped on cross-origin redirects

A command injection vulnerability was found in Python 2.x and 3.x, specifically within the mailcap module. Mailcap core-module is based on the format documented in RFC 1524. The “findmatch()” function does not sanitise the second argument (filename). As a result, the legitimate command (that is used for opening the specified mime type) is concatenated with an arbitrary command, injected by an attacker.

A vulnerability was found in libssh, where the authentication check of the connecting client can be bypassed in thepkiverifydatasignature function in memory allocation problems. This issue may happen if there is insufficient memory or the memory usage is limited. The problem is caused by the return value rc, which is initialized to SSHERROR and later rewritten to save the return value of the function call pkikeycheckhashcompatible. The value of the variable is not changed between this point and the cryptographic verification. Therefore any error between them calls goto error returning SSHOK.

A flaw was found in the json5 package. The affected version of the json5 package could allow an attacker to set arbitrary and unexpected keys on the object returned from JSON5.parse.

A use-after-free flaw was found in the xorg-x11-server. An X server crash may occur in a very specific and legacy configuration (a multi-screen setup with multiple protocol screens, also known as Zaphod mode) if the pointer is warped from within a window on one screen to the root window of the other screen and if the original window is destroyed followed by another window being destroyed.

Last updated 24 July 2024