CVE

14 Articles

This Week In Security: Curl Reveal, Rapid Reset DDoS, And Libcue

October 13, 2023 by 17 Comments

Curl gave us all a big warning that a severe security problem had been found in that code-base. Given the staggering number of Curl installs around the world, we held our collective breath and waited for the bombshell to drop this Wednesday. It turns out, it’s not quite as bad as feared — so long as you don’t have a SOCKS proxy.

In hindsight, shipping a heap overflow in code installed in over twenty billion instances is not an experience I would recommend. — Daniel Stenberg

The trouble started when the SOCKS5 proxy support was converted to a non-blocking implementation. It’s a win for libcurl to work on requests asynchronously, but refactoring code and new features always runs a bit of risk. SOCKS5 proxying has some quirks, like allowing DNS resolution to happen locally or at the proxy. The new async code starts out with:

bool socks5_resolve_local =
(proxytype == CURLPROXY_SOCKS5) ? TRUE : FALSE;

First off, unnecessary ternary is unnecessary. But note that this local variable gets set by the proxytype. If that’s CURLPROXY_SOCKS5_HOSTNAME, then it uses remote resolution. But inherited from old code is a check for a hostname that is too long for a SOCKS request (255 bytes). This code converts back to local resolution in this case.

The important detail here is that this function is now a state machine, that potentially runs multiple times for a single request, to achieve that asynchronous execution. The check for a too-long hostname only happens during the initialization state. Copying the hostname into the buffer happens in a different state. If setting up the connection takes enough time, the function will return and be executed again when something has changed. The ternary check runs again, but not the hostname-too-long. So if set to do remote resolution with a long enough host name, execution slips through this edge case, and the long hostname is copied into a too-small buffer.

It’s safe to assume that this heap overflow can result in arbitrary code execution. The fix has landed in 8.4.0, after being present for 1,315 days. [Daniel] goes ahead and gets ahead of the inevitable suggestion that Curl should be written in rust or another memory-safe language. Curl was started before those alternatives existed, and there is a very slow effort to move portions of the project to memory-safe languages. And you’re welcome to help out. Continue reading “This Week In Security: Curl Reveal, Rapid Reset DDoS, And Libcue” →

This Week In Security: Looney Tunables, Not A 0-day*, And Curl Warning

October 6, 2023 by 8 Comments

This week starts out with a nifty vulnerability in the glibc dynamic loader. This is an important step in running a binary executable on Linux, as it pulls the list of required shared libraries, and loads those libraries into memory. Glibc also includes a feature to adjust some runtime settings, via the GLIBC_TUNABLES environment variable. That’s where the vulnerability resides, and researchers from Qualsys obviously had a bit of fun in taking inspiration to pick the vulnerability name, “Looney Tunables”.

The problem is memory handling in the sanitizing parser. This function iterates through the environment variable, looking for strings of tunable1=aa, separated by colons. These strings get copied to the sanitized buffer, but the parsing logic goes awry when handling the malformed tunable1=tunable2=AAA. The first equals sign is taken at face value, copying the rest of the string into the buffer. But then the second equals sign is also processed as another key=value pair, leading to a buffer overflow.

The reason this particular overflow is interesting is that if the binary to be run is a Set-User-ID (SUID) root application, the dynamic loader runs as root, too. If the overflow can achieve code execution, then it’s a straightforward privilege escalation. And since we’re talking about it, you know there’s a way to execute code. It turns out, it’s possible to overwrite the pointer to the library search path, which determines where the dynamic loader will look for libraries. Tell it to look first in an attacker-controlled location, and you can easily load a malicious libc.so for instant code execution.

This vulnerability affects many Linux distros, and there’s already a Proof of Concept (PoC) published. So, it’s time to go check for updates for cve-2023-4911. Continue reading “This Week In Security: Looney Tunables, Not A 0-day*, And Curl Warning” →

This Week In Security: Not A Vulnerability, BGP Bug Propogation, And Press Enter To Hack

September 1, 2023 by 6 Comments

Curl was recently notified of a CVE, CVE-2020-19909, rated at a hair-raising 9.8 on the CVSS scale. And PostgreSQL has CVE-2020-21469, clocking in with a 7.5 severity. You may notice something odd about those two vulnerabilities, but I promise the 2020 date is only the tip of the iceberg here.

Let’s start with PostgreSQL. That vulnerability was only present in version 12.2, which released in February of 2020, and was fixed with the 12.3 release in May of that same year. The problem is a stack buffer overflow, which doesn’t seem to enable code execution, but does cause a denial of service situation. To trigger the bug? Repeatedly send the PostgreSQL daemon the SIGHUP signal.

If you’re familiar with Linux signals, that might sound odd. See, the SIGHUP signal technically indicates the end of a user session, but most daemons use it to indicate a restart or reload request. And to send this signal, a user has to have elevated privileges — elevated enough to simply stop the daemon altogether. Put simply, it’s not a security vulnerability, just a minor bug.

And now on to curl — This one is just bizarre. The issue is a integer overflow in the --retry-delay argument, which specifies in seconds how often curl should retry a failing download. The value is multiplied by 1000 to convert to milliseconds, resulting in an overflow for very large values. The result of that overflow? A smaller value for the retry delay.

[Daniel Stenberg] makes the point that this tale is a wonderful demonstration of the brokenness of the CVE system and NVD’s handling of it. And in this case, it’s hard not to see this as negligence. We have to work really hard to construct a theoretical scenario where this bug could actually be exploited. The best I’ve been able to come up with is an online download tool, where the user can specify part of the target name and a timeout. If that tool had a check to ensure that the timeout was large enough to avoid excess traffic, this bug could bypass that check. Should we be assigning CVEs for that sort of convoluted, theoretical attack?

But here’s the thing, that attack scenario should rate something like a CVSS of 4.8 at absolute worst. NVD assigned this a 9.8. There’s no way you can squint at this bug hard enough to legitimately rank it that severe. At the time of writing, the NVD lists this as “UNDERGOING REANALYSIS”.
Continue reading “This Week In Security: Not A Vulnerability, BGP Bug Propogation, And Press Enter To Hack” →

This Week In Security: Bogus CVEs, Bogus PoCs, And Maybe A Bogus Breach

July 7, 2023 by 3 Comments

It appears we have something of a problem. It’s not really a new problem, and shouldn’t be too surprising, but it did pop up again this week: bogus CVEs. Starting out in the security field? What’s the best way to jump-start a career? Getting a CVE find to your name certainly can’t hurt. And as a result, you get very junior security researchers looking for and reporting novel security vulnerabilities of sometimes dubious quality. Sometimes that process looks a lot like slinging reports against the wall to see what sticks. Things brings us to an odd bug report in the OBS Studio project.

A researcher put together a script to look for possible password exposure on Github projects, and it caught a configuration value named “password” in a .ini file, being distributed in the project source. Obvious credential leak in Git source, right? Except for the little detail that it was in the “locale” folder, and the files were named ca-es.ini, ja-jp.ini, and similar. You may be in on the joke by now, but if not, those are translation strings. It wasn’t leaked credentials, it was various translations of the word “password”. This sort of thing happens quite often, and from the viewpoint of a researcher looking at results from an automated tool, it can be challenging to spend enough time with each result to fully understand the code in question. It looks like this case includes a language barrier, making it even harder to clear up the confusion.

Things took a turn for the worse when a CVE was requested. The CVE Numbering Authority (CNA) that processed the request was MITRE, which issued CVE-2023-34585. It was a completely bogus CVE, and thankfully a more complete explanation from OBS was enough to convince the researcher of his error. That, however, brings us back to CVE-2023-36262, which was published this week. It’s yet another CVE, for the same non-issue, and even pointing at the same GitHub issue where the alleged bug is debunked. There’s multiple fails here, but the biggest disappointment is MITRE, for handing out CVEs twice for the same issue. Shout-out to [Netspooky] on Twitter for spotting this one. Continue reading “This Week In Security: Bogus CVEs, Bogus PoCs, And Maybe A Bogus Breach” →

This Week In Security: IOS Wifi Incantations, Ghosts, And Bad Regex

December 4, 2020 by 8 Comments

I hope everyone had a wonderful Thanksgiving last week. My household celebrated by welcoming a 4th member to the family. My daughter was born on Wednesday morning, November 25th. And thus explains what I did last week instead of writing the normal Hackaday column. Never fear, we shall catch up today, and cover the news that’s fit to be noticed.

iOS Zero-click Wifi Attack

[Ian Beer] of Google’s Project Zero brings us the fruit of his lockdown-induced labors, a spectacular iOS attack. The target of this attack is the kernel code that handles AWDL, an Apple WiFi protocol for adhoc mesh networks between devices. The most notable feature that makes use of AWDL is AirDrop, Apple’s device-to-device file sharing system. Because AWDL is a proprietary protocol, the WiFi hardware can’t do any accelerated processing of packets. A few years back, there was an attack against Broadcom firmware that required a second vulnerability to jump from the WiFi chip to the device CPU. Here, because the protocol is all implemented in Apple’s code, no such pivot is necessary.

And as you’ve likely deduced, there was a vulnerability found. AWDL uses Type-Length-Value (TLV) messages for sending management data. For a security researcher, TLVs are particularly interesting because each data type represents a different code path to attack. One of those data types is a list of MAC addresses, with a maximum of 10. The code that handles it allocates a 60 byte buffer, based on that maximum. The problem is that there isn’t a code path to drop incoming TLVs of that type when they exceed 60 bytes. The remainder is written right past the end of the allocated buffer.

There is more fun to be had, getting to a full exploit, but the details are a bit too much to fully dive in to here. It interesting to note that [Ian] ran into a particular problem: His poking at the target code was triggering unexpected kernel panics. He discovered two separate vulnerabilities, both distinct from the vuln he was trying to exploit.

Finally, this exploit requires the target device to have AWDL enabled, and many won’t. But you can use Bluetooth Low Energy advertisements to trick the target device into believing an Airdrop is coming in from a trusted contact. Once the device enables AWDL to verify the request, the attack can proceed. [Ian] reported his findings to Apple way back in 2019, and this vulnerability was patched in March of 2020.

Via Ars Technica.
Continue reading “This Week In Security: IOS Wifi Incantations, Ghosts, And Bad Regex” →

ESP32 Vulnerability Affects Older Chips

September 24, 2020 by 24 Comments

There is a scene from the movie RED (Retired, Extremely Dangerous) where Bruce Willis encounters a highly-secure door with a constantly changing lock code deep inside the CIA. Knowing the lock would be impossible to break, he simply destroyed the wall next to the door, reached through, and opened the door from the other side. We thought about that when we saw [raelize’s] hack to bypass the ESP32’s security measures.

Before you throw out all your ESP32 spy gadgets, though, be aware that the V3 silicon can be made to prevent the attack. V1 and V2, however, have a flaw that — if you know how to exploit it — renders secure boot and flash encryption almost meaningless.

Continue reading “ESP32 Vulnerability Affects Older Chips” →

This Week In Security: Palo Alto Scores A 10, Cursed Images, VM Escapes, And Malicious Music

July 3, 2020 by 12 Comments

We’ve looked at many vulnerabilities over the years here on Hackaday, but it’s rather rare for a CVE to score a perfect 10 severity. This is reserved for the most severe and exploitable of problems. Palo Alto announced such a vulnerability, CVE-2020-2021, on the 29th. This vulnerability affects Palo Alto devices running PAN-OS that have SAML authentication enabled and a certain validation option disabled. The vulnerability is pre-authentication, but does require access to a service protected by SAML authentication. For example, a Palo Alto device providing a web-based VPN could be vulnerable. The good news is that the vulnerable settings aren’t default, but the bad news is that the official configuration guide recommends the vulnerable settings for certain scenarios, like using a third party authentication service.

The issue is in the Security Assertion Markup Language (SAML) implementation, which is an XML based open standard for authentication. One of the primary use cases for SAML is to provide a Single Sign On (SSO) scheme. The normal deployment of SAML SSO is that a central provider handles the authentication of users, and then asserts to individual services that the connecting user is actually who they claim to be.

The setting needed for this vulnerability to be exploitable is ‘Validate Identity Provider Certificate’ to be disabled. If this option is enabled, the SSO provider must use a CA signed SAML certificates. This doesn’t appear to mean that unsigned SSL certificates would be accepted, and only applies to certificates inside the SAML messages. It seems to be widely accepted that these certificates don’t need to be CA signed. In the official announcement, the vulnerability type is said to be “CWE-347 Improper Verification of Cryptographic Signature”. Continue reading “This Week In Security: Palo Alto Scores A 10, Cursed Images, VM Escapes, And Malicious Music” →