squid: Improper input validation issues in HTTP Request processing

An issue was discovered in Squid before 4.10. Due to incorrect input validation, it can interpret crafted HTTP requests in unexpected ways to access server resources prohibited by earlier security filters.

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to a denial of service (NULL pointer dereference) via crafted messages written directly to the analysisd UNIX domain socket by a local user.

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to a use-after-free during processing of syscheck formatted msgs (received from authenticated remote agents and delivered to the analysisd processing queue by ossec-remoted).

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to path traversal (with write access) via crafted syscheck messages written directly to the analysisd UNIX domain socket by a local user.

In OSSEC-HIDS 2.7 through 3.5.0, the OS_CleanMSG function in ossec-analysisd doesn't remove or encode terminal control characters or newlines from processed log messages. In many cases, those characters are later logged. Because newlines (\n) are permitted in messages processed by ossec-analysisd, it may be possible to inject nested events into the ossec log. Use of terminal control characters may allow obfuscating events or executing commands when viewed through vulnerable terminal emulators. This may be an unauthenticated remote attack for certain types and origins of logged data.

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to a use-after-free during processing of ossec-alert formatted msgs (received from authenticated remote agents and delivered to the analysisd processing queue by ossec-remoted).

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to an off-by-one heap-based buffer overflow during the cleaning of crafted syslog msgs (received from authenticated remote agents and delivered to the analysisd processing queue by ossec-remoted).

In OSSEC-HIDS 2.7 through 3.5.0, the server component responsible for log analysis (ossec-analysisd) is vulnerable to a heap-based buffer overflow in the rootcheck decoder component via an authenticated client.

JYaml through 1.3 allows remote code execution during deserialization of a malicious payload through the load() function. NOTE: this is a discontinued product.

controllers/page_apply.php in Simplejobscript.com SJS through 1.66 is prone to unauthenticated Remote Code Execution by uploading a PHP script as a resume.

Monstra CMS through 3.0.4 allows remote authenticated users to take over arbitrary user accounts via a modified login parameter to an edit URI, as demonstrated by login=victim to the users/21/edit URI.

Ruckus ZoneFlex R500 104.0.0.0.1347 devices allow an authenticated attacker to execute arbitrary OS commands via the hidden /forms/nslookupHandler form, as demonstrated by the nslookuptarget=|cat${IFS} substring.

The bencoding parser in BitTorrent uTorrent through 3.5.5 (build 45505) misparses nested bencoded dictionaries, which allows a remote attacker to cause a denial of service.

XSS was discovered in the RegistrationMagic plugin 4.6.0.0 for WordPress via the rm_form_id, rm_tr, or form_name parameter.

An issue was discovered in the RegistrationMagic plugin 4.6.0.0 for WordPress. There is SQL injection via the rm_analytics_show_form rm_form_id parameter.

Jenzabar JICS (aka Internet Campus Solution) before 9.0.1 Patch 3, 9.1 before 9.1.2 Patch 2, and 9.2 before 9.2.2 Patch 8 has session cookies that are a deterministic function of the username. There is a hard-coded password to supply a PBKDF feeding into AES to encrypt a username and base64 encode it to a client-side cookie for persistent session authentication. By knowing the key and algorithm, an attacker can select any username, encrypt it, base64 encode it, and save it in their browser with the correct JICSLoginCookie cookie format to impersonate any real user in the JICS database without the need for authenticating (or verifying with MFA if implemented).

In Das U-Boot through 2020.01, a double free has been found in the cmd/gpt.c do_rename_gpt_parts() function. Double freeing may result in a write-what-where condition, allowing an attacker to execute arbitrary code. NOTE: this vulnerablity was introduced when attempting to fix a memory leak identified by static analysis.

Stormshield Network Security 310 3.7.10 devices have an auth/lang.html?rurl= Open Redirect vulnerability on the captive portal. For example, the attacker can use rurl=//example.com instead of rurl=https://example.com in the query string.

The Admin web application in Kinetica 7.0.9.2.20191118151947 does not properly sanitise the input for the function getLogs. This lack of sanitisation could be exploited to allow an authenticated attacker to run remote code on the underlying operating system. The logFile parameter in the getLogs function was used as a variable in a command to read log files; however, due to poor input sanitisation, it was possible to bypass a replacement and break out of the command.

kernel: use-after-free in fs/namei.c

fs/namei.c in the Linux kernel before 5.5 has a may_create_in_sticky use-after-free, which allows local users to cause a denial of service (OOPS) or possibly obtain sensitive information from kernel memory, aka CID-d0cb50185ae9. One attack vector may be an open system call for a UNIX domain socket, if the socket is being moved to a new parent directory and its old parent directory is being removed.