This update is now available for all supported architectures. For reference the original advisory text follows.
Several vulnerabilities have been discovered in the Linux kernel that may lead to a privilege escalation, denial of service or information leaks.
A race condition was found in the ext4 filesystem implementation. A local user could exploit this to cause a denial of service (filesystem corruption).
Wen Xu from SSLab at Gatech reported that crafted Btrfs volumes could trigger a crash (Oops) and/or out-of-bounds memory access. An attacker able to mount such a volume could use this to cause a denial of service or possibly for privilege escalation.
Mitchell Frank of Cisco discovered that when the IEEE 802.11 (WiFi) stack was used in AP mode with roaming, it would trigger roaming for a newly associated station before the station was authenticated. An attacker within range of the AP could use this to cause a denial of service, either by filling up a switching table or by redirecting traffic away from other stations.
Jungyeon discovered that a crafted filesystem can cause the ext4 implementation to deallocate or reallocate journal blocks. A user permitted to mount filesystems could use this to cause a denial of service (crash), or possibly for privilege escalation.
It was discovered that the ext4 filesystem driver did not safely handle unlinking of an inode that, due to filesystem corruption, already has a link count of 0. An attacker able to mount arbitrary ext4 volumes could use this to cause a denial of service (memory corruption or crash) or possibly for privilege escalation.
Tristan Madani reported a race condition in the blktrace debug facility that could result in a use-after-free. A local user able to trigger removal of block devices could possibly use this to cause a denial of service (crash) or for privilege escalation.
The syzbot tool found that the input subsystem did not fully validate keycode changes, which could result in a heap out-of-bounds write. A local user permitted to access the device node for an input or VT device could possibly use this to cause a denial of service (crash or memory corruption) or for privilege escalation.
Jann Horn reported that the Android ashmem driver did not prevent read-only files from being memory-mapped and then remapped as read-write. However, Android drivers are not enabled in Debian kernel configurations.
Researchers at VU Amsterdam discovered that on some Intel CPUs supporting the RDRAND and RDSEED instructions, part of a random value generated by these instructions may be used in a later speculative execution on any core of the same physical CPU. Depending on how these instructions are used by applications, a local user or VM guest could use this to obtain sensitive information such as cryptographic keys from other users or VMs.
This vulnerability can be mitigated by a microcode update, either as part of system firmware (BIOS) or through the intel-microcode package in Debian's non-free archive section. This kernel update only provides reporting of the vulnerability and the option to disable the mitigation if it is not needed.
Xiumei Mu reported that some network protocols that can run on top of IPv6 would bypass the Transformation (XFRM) layer used by IPsec, IPcomp/IPcomp6, IPIP, and IPv6 Mobility. This could result in disclosure of information over the network, since it would not be encrypted or routed according to the system policy.
Paulo Bonzini discovered that the KVM implementation for Intel processors did not properly handle instruction emulation for L2 guests when nested virtualization is enabled. This could allow an L2 guest to cause privilege escalation, denial of service, or information leaks in the L1 guest.
The Hulk Robot tool found a potential MMIO out-of-bounds access in the vgacon driver. A local user permitted to access a virtual terminal (/dev/tty1 etc.) on a system using the vgacon driver could use this to cause a denial of service (crash or memory corruption) or possibly for privilege escalation.
The syzbot tool found a race condition in the the virtual terminal driver, which could result in a use-after-free. A local user permitted to access a virtual terminal could use this to cause a denial of service (crash or memory corruption) or possibly for privilege escalation.
Jordy Zomer reported an incorrect range check in the floppy driver which could lead to a static out-of-bounds access. A local user permitted to access a floppy drive could use this to cause a denial of service (crash or memory corruption) or possibly for privilege escalation.
It was discovered that the PTP hardware clock subsystem did not properly manage device lifetimes. Removing a PTP hardware clock from the system while a user process was using it could lead to a use-after-free. The security impact of this is unclear.
Dmitry Vyukov reported that the SELinux subsystem did not properly handle validating multiple messages, which could allow a privileged attacker to bypass SELinux netlink restrictions.
It was discovered that the vhost_net driver did not properly validate the type of sockets set as back-ends. A local user permitted to access /dev/vhost-net could use this to cause a stack corruption via crafted system calls, resulting in denial of service (crash) or possibly privilege escalation.
It was discovered that the slcan (serial line CAN) network driver did not fully initialise CAN headers for received packets, resulting in an information leak from the kernel to user-space or over the CAN network.
Entropy Moe reported that the shared memory filesystem (tmpfs) did
not correctly handle an mpol
mount option specifying an empty
node list, leading to a stack-based out-of-bounds write. If user
namespaces are enabled, a local user could use this to cause a
denial of service (crash) or possibly for privilege escalation.
It was discovered that the ov519, stv06xx, and xirlink_cit media drivers did not properly validate USB device descriptors. A physically present user with a specially constructed USB device could use this to cause a denial-of-service (crash) or possibly for privilege escalation.
Piotr Krysiuk discovered a race condition between the umount and pivot_root operations in the filesystem core (vfs). A local user with the CAP_SYS_ADMIN capability in any user namespace could use this to cause a denial of service (crash).
Kyungtae Kim reported a race condition in the USB core that can result in a use-after-free. It is not clear how this can be exploited, but it could result in a denial of service (crash or memory corruption) or privilege escalation.
Tom Hatskevich reported a bug in the mptfusion storage drivers. An ioctl handler fetched a parameter from user memory twice, creating a race condition which could result in incorrect locking of internal data structures. A local user permitted to access /dev/mptctl could use this to cause a denial of service (crash or memory corruption) or for privilege escalation.
It was discovered that the mwifiex WiFi driver did not sufficiently validate scan requests, resulting a potential heap buffer overflow. A local user with CAP_NET_ADMIN capability could use this to cause a denial of service (crash or memory corruption) or possibly for privilege escalation.
It was discovered that the mwifiex WiFi driver did not sufficiently validate WMM parameters received from an access point (AP), resulting a potential heap buffer overflow. A malicious AP could use this to cause a denial of service (crash or memory corruption) or possibly to execute code on a vulnerable system.
It was discovered that the spi-dw SPI host driver did not properly serialise access to its internal state. The security impact of this is unclear, and this driver is not included in Debian's binary packages.
It was discovered that the sg (SCSI generic) driver did not correctly release internal resources in a particular error case. A local user permitted to access an sg device could possibly use this to cause a denial of service (resource exhaustion).
Adam Zabrocki reported a weakness in the signal subsystem's permission checks. A parent process can choose an arbitary signal for a child process to send when it exits, but if the parent has executed a new program then the default SIGCHLD signal is sent. A local user permitted to run a program for several days could bypass this check, execute a setuid program, and then send an arbitrary signal to it. Depending on the setuid programs installed, this could have some security impact.
Kyungtae Kim reported a potential heap out-of-bounds write in the USB gadget subsystem. A local user permitted to write to the gadget configuration filesystem could use this to cause a denial of service (crash or memory corruption) or potentially for privilege escalation.
For Debian 8 Jessie
, these problems have been fixed in version
3.16.84-1.
We recommend that you upgrade your linux packages.
Further information about Debian LTS security advisories, how to apply these updates to your system and frequently asked questions can be found at: https://wiki.debian.org/LTS