Freebsd » Freebsd : Security Vulnerabilities, CVEs,

NFS in a BSD derived codebase, as used in OpenBSD through 7.4 and FreeBSD through 14.0-RELEASE, allows remote attackers to execute arbitrary code via a bug that is unrelated to memory corruption.

The jail(2) system call has not limited a visiblity of allocated TTYs (the kern.ttys sysctl). This gives rise to an information leak about processes outside the current jail. Attacker can get information about TTYs allocated on the host or in other jails. Effectively, the information printed by "pstat -t" may be leaked.

`bhyveload -h <host-path>` may be used to grant loader access to the <host-path> directory tree on the host. Affected versions of bhyveload(8) do not make any attempt to restrict loader's access to <host-path>, allowing the loader to read any file the host user has access to. In the bhyveload(8) model, the host supplies a userboot.so to boot with, but the loader scripts generally come from the guest image. A maliciously crafted script could be used to exfiltrate sensitive data from the host accessible to the user running bhyhveload(8), which is often the system root.

ping reads raw IP packets from the network to process responses in the pr_pack() function. As part of processing a response ping has to reconstruct the IP header, the ICMP header and if present a "quoted packet," which represents the packet that generated an ICMP error. The quoted packet again has an IP header and an ICMP header. The pr_pack() copies received IP and ICMP headers into stack buffers for further processing. In so doing, it fails to take into account the possible presence of IP option headers following the IP header in either the response or the quoted packet. When IP options are present, pr_pack() overflows the destination buffer by up to 40 bytes. The memory safety bugs described above can be triggered by a remote host, causing the ping program to crash. The ping process runs in a capability mode sandbox on all affected versions of FreeBSD and is thus very constrained in how it can interact with the rest of the system at the point where the bug can occur.

The implementation of lib9p's handling of RWALK messages was missing a bounds check needed when unpacking the message contents. The missing check means that the receipt of a specially crafted message will cause lib9p to overwrite unrelated memory. The bug can be triggered by a malicious bhyve guest kernel to overwrite memory in the bhyve(8) process. This could potentially lead to user-mode code execution on the host, subject to bhyve's Capsicum sandbox.

A particular case of memory sharing is mishandled in the virtual memory system. This is very similar to SA-21:08.vm, but with a different root cause. An unprivileged local user process can maintain a mapping of a page after it is freed, allowing that process to read private data belonging to other processes or the kernel.

The aio_aqueue function, used by the lio_listio system call, fails to release a reference to a credential in an error case. An attacker may cause the reference count to overflow, leading to a use after free (UAF).

When dumping core and saving process information, proc_getargv() might return an sbuf which have a sbuf_len() of 0 or -1, which is not properly handled. An out-of-bound read can happen when user constructs a specially crafted ps_string, which in turn can cause the kernel to crash.

The 802.11 beacon handling routine failed to validate the length of an IEEE 802.11s Mesh ID before copying it to a heap-allocated buffer. While a FreeBSD Wi-Fi client is in scanning mode (i.e., not associated with a SSID) a malicious beacon frame may overwrite kernel memory, leading to remote code execution.

The e1000 network adapters permit a variety of modifications to an Ethernet packet when it is being transmitted. These include the insertion of IP and TCP checksums, insertion of an Ethernet VLAN header, and TCP segmentation offload ("TSO"). The e1000 device model uses an on-stack buffer to generate the modified packet header when simulating these modifications on transmitted packets. When checksum offload is requested for a transmitted packet, the e1000 device model used a guest-provided value to specify the checksum offset in the on-stack buffer. The offset was not validated for certain packet types. A misbehaving bhyve guest could overwrite memory in the bhyve process on the host, possibly leading to code execution in the host context. The bhyve process runs in a Capsicum sandbox, which (depending on the FreeBSD version and bhyve configuration) limits the impact of exploiting this issue.

Handlers for *_CFG_PAGE read / write ioctls in the mpr, mps, and mpt drivers allocated a buffer of a caller-specified size, but copied to it a fixed size header. Other heap content would be overwritten if the specified size was too small. Users with access to the mpr, mps or mpt device node may overwrite heap data, potentially resulting in privilege escalation. Note that the device node is only accessible to root and members of the operator group.

A user-provided integer option was passed to nmreq_copyin() without checking if it would overflow. This insufficient bounds checking could lead to kernel memory corruption. On systems configured to include netmap in their devfs_ruleset, a privileged process running in a jail can affect the host environment.

The total size of the user-provided nmreq to nmreq_copyin() was first computed and then trusted during the copyin. This time-of-check to time-of-use bug could lead to kernel memory corruption. On systems configured to include netmap in their devfs_ruleset, a privileged process running in a jail can affect the host environment.

The bsdinstall installer in FreeBSD 10.x before 10.1 p9, when configuring full disk encrypted ZFS, uses world-readable permissions for the GELI keyfile (/boot/encryption.key), which allows local users to obtain sensitive key information by reading the file.

The setlogin function in FreeBSD 8.4 through 10.1-RC4 does not initialize the buffer used to store the login name, which allows local users to obtain sensitive information from kernel memory via a call to getlogin, which returns the entire buffer.

The sm_close_on_exec function in conf.c in sendmail before 8.14.9 has arguments in the wrong order, and consequently skips setting expected FD_CLOEXEC flags, which allows local users to access unintended high-numbered file descriptors via a custom mail-delivery program.

The ktrace utility in the FreeBSD kernel 8.4 before p11, 9.1 before p14, 9.2 before p7, and 9.3-BETA1 before p1 uses an incorrect page fault kernel trace entry size, which allows local users to obtain sensitive information from kernel memory via a kernel process trace.

crontab.c in crontab in FreeBSD allows local users to determine the existence of arbitrary directories via a command-line argument composed of a directory name concatenated with a directory traversal sequence that leads to the /etc/crontab pathname.

crontab.c in crontab in FreeBSD and Apple Mac OS X allows local users to (1) determine the existence of arbitrary files via a symlink attack on a /tmp/crontab.XXXXXXXXXX temporary file and (2) perform MD5 checksum comparisons on arbitrary pairs of files via two symlink attacks on /tmp/crontab.XXXXXXXXXX temporary files.

The Coda filesystem kernel module, as used in NetBSD and FreeBSD, when Coda is loaded and Venus is running with /coda mounted, allows local users to read sensitive heap memory via a large out_size value in a ViceIoctl struct to a Coda ioctl, which triggers a buffer over-read.

The ptsname function in FreeBSD 6.0 through 7.0-PRERELEASE does not properly verify that a certain portion of a device name is associated with a pty of a user who is calling the pt_chown function, which might allow local users to read data from the pty from another user.

The "internal state tracking" code for the random and urandom devices in FreeBSD 5.5, 6.1 through 6.3, and 7.0 beta 4 allows local users to obtain portions of previously-accessed random values, which could be leveraged to bypass protection mechanisms that rely on secrecy of those values.

The 4BSD process scheduler in the FreeBSD kernel performs scheduling based on CPU billing gathered from periodic process sampling ticks, which allows local users to cause a denial of service (CPU consumption) by performing voluntary nanosecond sleeps that result in the process not being active during a clock interrupt, as described in "Secretly Monopolizing the CPU Without Superuser Privileges."

The ULE process scheduler in the FreeBSD kernel gives preference to "interactive" processes that perform voluntary sleeps, which allows local users to cause a denial of service (CPU consumption), as described in "Secretly Monopolizing the CPU Without Superuser Privileges."

Integer signedness error in the fw_ioctl (FW_IOCTL) function in the FireWire (IEEE-1394) drivers (dev/firewire/fwdev.c) in various BSD kernels, including DragonFlyBSD, FreeBSD 5.5, MidnightBSD 0.1-CURRENT before 20061115, NetBSD-current before 20061116, NetBSD-4 before 20061203, and TrustedBSD, allows local users to read arbitrary memory contents via certain negative values of crom_buf->len in an FW_GCROM command. NOTE: this issue has been labeled as an integer overflow, but it is more like an integer signedness error.