USN-6520-1: Linux kernel (StarFive) vulnerabilities

Read Time:3 Minute, 44 Second

Ivan D Barrera, Christopher Bednarz, Mustafa Ismail, and Shiraz Saleem
discovered that the InfiniBand RDMA driver in the Linux kernel did not
properly check for zero-length STAG or MR registration. A remote attacker
could possibly use this to execute arbitrary code. (CVE-2023-25775)

Yu Hao and Weiteng Chen discovered that the Bluetooth HCI UART driver in
the Linux kernel contained a race condition, leading to a null pointer
dereference vulnerability. A local attacker could use this to cause a
denial of service (system crash). (CVE-2023-31083)

Yu Hao discovered that the UBI driver in the Linux kernel did not properly
check for MTD with zero erasesize during device attachment. A local
privileged attacker could use this to cause a denial of service (system
crash). (CVE-2023-31085)

Lin Ma discovered that the Netlink Transformation (XFRM) subsystem in the
Linux kernel contained a null pointer dereference vulnerability in some
situations. A local privileged attacker could use this to cause a denial of
service (system crash). (CVE-2023-3772)

Chih-Yen Chang discovered that the KSMBD implementation in the Linux kernel
did not properly validate SMB request protocol IDs, leading to a out-of-
bounds read vulnerability. A remote attacker could possibly use this to
cause a denial of service (system crash). (CVE-2023-38430)

Chih-Yen Chang discovered that the KSMBD implementation in the Linux kernel
did not properly validate command payload size, leading to a out-of-bounds
read vulnerability. A remote attacker could possibly use this to cause a
denial of service (system crash). (CVE-2023-38432)

It was discovered that the NFC implementation in the Linux kernel contained
a use-after-free vulnerability when performing peer-to-peer communication
in certain conditions. A privileged attacker could use this to cause a
denial of service (system crash) or possibly expose sensitive information
(kernel memory). (CVE-2023-3863)

Laurence Wit discovered that the KSMBD implementation in the Linux kernel
did not properly validate a buffer size in certain situations, leading to
an out-of-bounds read vulnerability. A remote attacker could use this to
cause a denial of service (system crash) or possibly expose sensitive
information. (CVE-2023-3865)

Laurence Wit discovered that the KSMBD implementation in the Linux kernel
contained a null pointer dereference vulnerability when handling handling
chained requests. A remote attacker could use this to cause a denial of
service (system crash). (CVE-2023-3866)

It was discovered that the KSMBD implementation in the Linux kernel did not
properly handle session setup requests, leading to an out-of-bounds read
vulnerability. A remote attacker could use this to expose sensitive
information. (CVE-2023-3867)

It was discovered that the Siano USB MDTV receiver device driver in the
Linux kernel did not properly handle device initialization failures in
certain situations, leading to a use-after-free vulnerability. A physically
proximate attacker could use this cause a denial of service (system crash).
(CVE-2023-4132)

It was discovered that a race condition existed in the Cypress touchscreen
driver in the Linux kernel during device removal, leading to a use-after-
free vulnerability. A physically proximate attacker could use this to cause
a denial of service (system crash) or possibly execute arbitrary code.
(CVE-2023-4134)

Thelford Williams discovered that the Ceph file system messenger protocol
implementation in the Linux kernel did not properly validate frame segment
length in certain situation, leading to a buffer overflow vulnerability. A
remote attacker could use this to cause a denial of service (system crash)
or possibly execute arbitrary code. (CVE-2023-44466)

Manfred Rudigier discovered that the Intel(R) PCI-Express Gigabit (igb)
Ethernet driver in the Linux kernel did not properly validate received
frames that are larger than the set MTU size, leading to a buffer overflow
vulnerability. An attacker could use this to cause a denial of service
(system crash) or possibly execute arbitrary code. (CVE-2023-45871)

Maxim Levitsky discovered that the KVM nested virtualization (SVM)
implementation for AMD processors in the Linux kernel did not properly
handle x2AVIC MSRs. An attacker in a guest VM could use this to cause a
denial of service (host kernel crash). (CVE-2023-5090)

It was discovered that the SMB network file sharing protocol implementation
in the Linux kernel did not properly handle certain error conditions,
leading to a use-after-free vulnerability. A local attacker could use this
to cause a denial of service (system crash) or possibly execute arbitrary
code. (CVE-2023-5345)

Read More

llhttp-9.1.3-1.fc38 python-aiohttp-3.8.6-1.fc38 uxplay-1.66-2.fc38

Read Time:1 Minute, 28 Second

FEDORA-2023-bc1f081ca0

Packages in this update:

llhttp-9.1.3-1.fc38
python-aiohttp-3.8.6-1.fc38
uxplay-1.66-2.fc38

Update description:

Security fix for CVE-2023-47627

https://pagure.io/fesco/issue/3106

python-aiohttp 3.8.6 (2023-10-07)

https://github.com/aio-libs/aiohttp/blob/v3.8.6/CHANGES.rst#386-2023-10-07

Security bugfixes

Upgraded llhttp to v9.1.3: https://github.com/aio-libs/aiohttp/security/advisories/GHSA-pjjw-qhg8-p2p9
Updated Python parser to comply with RFCs 9110/9112: https://github.com/aio-libs/aiohttp/security/advisories/GHSA-gfw2-4jvh-wgfg

Deprecation

Added fallback_charset_resolver parameter in ClientSession to allow a user-supplied character set detection function. Character set detection will no longer be included in 3.9 as a default. If this feature is needed, please use fallback_charset_resolver.

Features

Enabled lenient response parsing for more flexible parsing in the client (this should resolve some regressions when dealing with badly formatted HTTP responses).

Bugfixes

Fixed PermissionError when .netrc is unreadable due to permissions.
Fixed output of parsing errors pointing to a n.
Fixed GunicornWebWorker max_requests_jitter not working.
Fixed sorting in filter_cookies to use cookie with longest path.
Fixed display of BadStatusLine messages from llhttp.

llhttp 9.1.3

Fixes

Restart the parser on HTTP 100
Fix chunk extensions quoted-string value parsing
Fix lenient_flags truncated on reset
Fix chunk extensions’ parameters parsing when more then one name-value pair provided

llhttp 9.1.2

What’s Changed

Fix HTTP 1xx handling

llhttp 9.1.1

What’s Changed

feat: Expose new lenient methods

llhttp 9.1.0

What’s Changed

New lenient flag to make CR completely optional
New lenient flag to have spaces after chunk header

Read More

llhttp-9.1.3-1.fc39 python-aiohttp-3.8.6-1.fc39 uxplay-1.66-2.fc39

Read Time:1 Minute, 28 Second

FEDORA-2023-5130a73b00

Packages in this update:

llhttp-9.1.3-1.fc39
python-aiohttp-3.8.6-1.fc39
uxplay-1.66-2.fc39

Update description:

Security fix for CVE-2023-47627

https://pagure.io/fesco/issue/3106

python-aiohttp 3.8.6 (2023-10-07)

https://github.com/aio-libs/aiohttp/blob/v3.8.6/CHANGES.rst#386-2023-10-07

Security bugfixes

Upgraded llhttp to v9.1.3: https://github.com/aio-libs/aiohttp/security/advisories/GHSA-pjjw-qhg8-p2p9
Updated Python parser to comply with RFCs 9110/9112: https://github.com/aio-libs/aiohttp/security/advisories/GHSA-gfw2-4jvh-wgfg

Deprecation

Added fallback_charset_resolver parameter in ClientSession to allow a user-supplied character set detection function. Character set detection will no longer be included in 3.9 as a default. If this feature is needed, please use fallback_charset_resolver.

Features

Enabled lenient response parsing for more flexible parsing in the client (this should resolve some regressions when dealing with badly formatted HTTP responses).

Bugfixes

Fixed PermissionError when .netrc is unreadable due to permissions.
Fixed output of parsing errors pointing to a n.
Fixed GunicornWebWorker max_requests_jitter not working.
Fixed sorting in filter_cookies to use cookie with longest path.
Fixed display of BadStatusLine messages from llhttp.

llhttp 9.1.3

Fixes

Restart the parser on HTTP 100
Fix chunk extensions quoted-string value parsing
Fix lenient_flags truncated on reset
Fix chunk extensions’ parameters parsing when more then one name-value pair provided

llhttp 9.1.2

What’s Changed

Fix HTTP 1xx handling

llhttp 9.1.1

What’s Changed

feat: Expose new lenient methods

llhttp 9.1.0

What’s Changed

New lenient flag to make CR completely optional
New lenient flag to have spaces after chunk header

Read More

USN-6518-1: AFFLIB vulnerability

Read Time:14 Second

Luis Rocha discovered that AFFLIB incorrectly handled certain input files.
If a user or automated system were tricked into processing a specially
crafted AFF image file, a remote attacker could possibly use this issue
to cause a denial of service via application crash. (CVE-2018-8050)

Read More

ID Theft Service Resold Access to USInfoSearch Data

Read Time:9 Minute, 7 Second

One of the cybercrime underground’s more active sellers of Social Security numbers, background and credit reports has been pulling data from hacked accounts at the U.S. consumer data broker USinfoSearch, KrebsOnSecurity has learned.

Since at least February 2023, a service advertised on Telegram called USiSLookups has operated an automated bot that allows anyone to look up the SSN or background report on virtually any American. For prices ranging from $8 to $40 and payable via virtual currency, the bot will return detailed consumer background reports automatically in just a few moments.

USiSLookups is the project of a cybercriminal who uses the nicknames JackieChan/USInfoSearch, and the Telegram channel for this service includes a small number of sample background reports, including that of President Joe Biden, and podcaster Joe Rogan. The data in those reports includes the subject’s date of birth, address, previous addresses, previous phone numbers and employers, known relatives and associates, and driver’s license information.

JackieChan’s service abuses the name and trademarks of Columbus, OH based data broker USinfoSearch, whose website says it provides “identity and background information to assist with risk management, fraud prevention, identity and age verification, skip tracing, and more.”

“We specialize in non-FCRA data from numerous proprietary sources to deliver the information you need, when you need it,” the company’s website explains. “Our services include API-based access for those integrating data into their product or application, as well as bulk and batch processing of records to suit every client.”

As luck would have it, my report was also listed in the Telegram channel for this identity fraud service, presumably as a teaser for would-be customers. On October 19, 2023, KrebsOnSecurity shared a copy of this file with the real USinfoSearch, along with a request for information about the provenance of the data.

USinfoSearch said it would investigate the report, which appears to have been obtained on or before June 30, 2023. On Nov. 9, 2023, Scott Hostettler, general manager of USinfoSearch parent Martin Data LLC shared a written statement about their investigation that suggested the ID theft service was trying to pass off someone else’s consumer data as coming from USinfoSearch:

Regarding the Telegram incident, we understand the importance of protecting sensitive information and upholding the trust of our users is our top priority. Any allegation that we have provided data to criminals is in direct opposition to our fundamental principles and the protective measures we have established and continually monitor to prevent any unauthorized disclosure. Because Martin Data has a reputation for high-quality data, thieves may steal data from other sources and then disguise it as ours. While we implement appropriate safeguards to guarantee that our data is only accessible by those who are legally permitted, unauthorized parties will continue to try to access our data. Thankfully, the requirements needed to pass our credentialing process is tough even for established honest companies.

USinfoSearch’s statement did not address any questions put to the company, such as whether it requires multi-factor authentication for customer accounts, or whether my report had actually come from USinfoSearch’s systems.

After much badgering, on Nov. 21 Hostettler acknowledged that the USinfoSearch identity fraud service on Telegram was in fact pulling data from an account belonging to a vetted USinfoSearch client.

“I do know 100% that my company did not give access to the group who created the bots, but they did gain access to a client,” Hostettler said of the Telegram-based identity fraud service. “I apologize for any inconvenience this has caused.”

Hostettler said USinfoSearch heavily vets any new potential clients, and that all users are required to undergo a background check and provide certain documents. Even so, he said, several fraudsters each month present themselves as credible business owners or C-level executives during the credentialing process, completing the application and providing the necessary documentation to open a new account.

“The level of skill and craftsmanship demonstrated in the creation of these supporting documents is incredible,” Hostettler said. “The numerous licenses provided appear to be exact replicas of the original document. Fortunately, I’ve discovered several methods of verification that do not rely solely on those documents to catch the fraudsters.”

“These people are unrelenting, and they act without regard for the consequences,” Hostettler continued. “After I deny their access, they will contact us again within the week using the same credentials. In the past, I’ve notified both the individual whose identity is being used fraudulently and the local police. Both are hesitant to act because nothing can be done to the offender if they are not apprehended. That is where most attention is needed.”

SIM SWAPPER’S DELIGHT

JackieChan is most active on Telegram channels focused on “SIM swapping,” which involves bribing or tricking mobile phone company employees into redirecting a target’s phone number to a device the attackers control. SIM swapping allows crooks to temporarily intercept the target’s text messages and phone calls, including any links or one-time codes for authentication that are delivered via SMS.

Reached on Telegram, JackieChan said most of his clients hail from the criminal SIM swapping world, and that the bulk of his customers use his service via an application programming interface (API) that allows customers to integrate the lookup service with other web-based services, databases, or applications.

“Sim channels is where I get most of my customers,” JackieChan told KrebsOnSecurity. “I’m averaging around 100 lookups per day on the [Telegram] bot, and around 400 per day on the API.”

JackieChan claims his USinfoSearch bot on Telegram abuses stolen credentials needed to access an API used by the real USinfoSearch, and that his service was powered by USinfoSearch account credentials that were stolen by malicious software tied to a botnet that he claims to have operated for some time.

This is not the first time USinfoSearch has had trouble with identity thieves masquerading as legitimate customers. In 2013, KrebsOnSecurity broke the news that an identity fraud service in the underground called “SuperGet[.]info” was reselling access to personal and financial data on more than 200 million Americans that was obtained via the big-three credit bureau Experian.

The consumer data resold by Superget was not obtained directly from Experian, but rather via USinfoSearch. At the time, USinfoSearch had a contractual agreement with a California company named Court Ventures, whereby customers of Court Ventures had access to the USinfoSearch data, and vice versa.

When Court Ventures was purchased by Experian in 2012, the proprietor of SuperGet — a Vietnamese hacker named Hieu Minh Ngo who had impersonated an American private investigator — was grandfathered in as a client. The U.S. Secret Service agent who oversaw Ngo’s capture, extradition, prosecution and rehabilitation told KrebsOnSecurity he’s unaware of any other cybercriminal who has caused more material financial harm to more Americans than Ngo.

REAL POLICE, FAKE EDRS

JackieChan also sells access to hacked email accounts belonging to law enforcement personnel in the United States and abroad. Hacked police department emails can come in handy for ID thieves trying to pose as law enforcement officials who wish to purchase consumer data from platforms like USinfoSearch. Hence, Mr. Hostettler’s ongoing battle with fraudsters seeking access to his company’s service.

These police credentials are mainly marketed to criminals seeking fraudulent “Emergency Data Requests,” wherein crooks use compromised government and police department email accounts to rapidly obtain customer account data from mobile providers, ISPs and social media companies.

Normally, these companies will require law enforcement officials to supply a subpoena before turning over customer or user records. But EDRs allow police to bypass that process by attesting that the information sought is related to an urgent matter of life and death, such as an impending suicide or terrorist attack.

In response to an alarming increase in the volume of fraudulent EDRs, many service providers have chosen to require all EDRs be processed through a service called Kodex, which seeks to filter EDRs based on the reputation of the law enforcement entity requesting the information, and other attributes of the requestor.

For example, if you want to send an EDR to Coinbase or Twilio, you’ll first need to have valid law enforcement credentials and create an account at the Kodex online portal at these companies. However, Kodex may still throttle or block any requests from any accounts if they set off certain red flags.

Within their own separate Kodex portals, Twilio can’t see requests submitted to Coinbase, or vice versa. But each can see if a law enforcement entity or individual tied to one of their own requests has ever submitted a request to a different Kodex client, and then drill down further into other data about the submitter, such as Internet address(es) used, and the age of the requestor’s email address.

In August, JackieChan was advertising a working Kodex account for sale on the cybercrime channels, including redacted screenshots of the Kodex account dashboard as proof of access.

Kodex co-founder Matt Donahue told KrebsOnSecurity his company immediately detected that the law enforcement email address used to create the Kodex account pictured in JackieChan’s ad was likely stolen from a police officer in India. One big tipoff, Donahue said, was that the person creating the account did so using an Internet address in Brazil.

“There’s a lot of friction we can put in the way for illegitimate actors,” Donahue said. “We don’t let people use VPNs. In this case we let them in to honeypot them, and that’s how they got that screenshot. But nothing was allowed to be transmitted out from that account.”

Massive amounts of data about you and your personal history are available from USinfoSearch and dozens of other data brokers that acquire and sell “non-FCRA” data — i.e., consumer data that cannot be used for the purposes of determining one’s eligibility for credit, insurance, or employment.

Anyone who works in or adjacent to law enforcement is eligible to apply for access to these data brokers, which often market themselves to police departments and to “skip tracers,” essentially bounty hunters hired to locate others in real life — often on behalf of debt collectors, process servers or a bail bondsman.

There are tens of thousands of police jurisdictions around the world — including roughly 18,000 in the United States alone. And the harsh reality is that all it takes for hackers to apply for access to data brokers (and abuse the EDR process) is illicit access to a single police email account.

The trouble is, compromised credentials to law enforcement email accounts show up for sale with alarming frequency on the Telegram channels where JackieChan and their many clients reside. Indeed, Donahue said Kodex so far this year has identified attempted fake EDRs coming from compromised email accounts for police departments in India, Italy, Thailand and Turkey.

Read More