On this page (7 sections)
This article covers three Linux kernel CVEs from public records: CVE-2017-9986, CVE-2022-49114 and CVE-2023-52784. They affect different kernel components and carry different severity ratings. The research for this story contains no distribution advisory, so affected releases and fixed package versions are not stated here. Check your own distribution's security notice for the authoritative list for your release.
If you run Linux servers, VPS instances or container hosts for websites, the checks and update steps below are worth a few minutes. The sources describe a local attack vector for at least two of these flaws, so they matter most on shared hosting, build servers and any host where untrusted users or code can run.
What the three CVEs are
The descriptions below come from the CVE records and the GitHub advisories: CVE-2017-9986, CVE-2022-49114 and CVE-2023-52784. Treat this as the list of CVEs attached to this story, not as a confirmed list of what any particular vendor notice fixes.
| CVE | Component | Weakness | Score |
|---|---|---|---|
| CVE-2017-9986 | intr function in sound/oss/msnd_pinnacle.c (kernel through 4.11.7) | CWE-125 out-of-bounds read | 7.8 High |
| CVE-2022-49114 | SCSI libfc, fc_exch_abts_resp() | CWE-416 use after free | 8.8 High (CVE record), 7.8 High (GitHub advisory) |
| CVE-2023-52784 | bonding driver, bond_setup_by_slave() | No CWE listed | 5.5 Medium |
Technical details in plain terms
CVE-2017-9986: double fetch in an old sound driver
The record says a local user can cause a denial of service or possibly have unspecified other impact by changing a message queue head pointer between two kernel reads of that value. This is a "double fetch" race: the kernel reads a value, it is changed, and the kernel reads it again and trusts the new value. The record names the intr function in sound/oss/msnd_pinnacle.c and kernels through 4.11.7. Whether a given distribution kernel is affected depends on the vendor notice and on whether that driver is built or loaded.
CVE-2022-49114: use after free in SCSI libfc
In the SCSI libfc code, a reference count is decreased and the object can be freed, but the code keeps using it afterward. The upstream fix returns right after the release call. Use after free bugs can lead to crashes or memory corruption. The GitHub advisory's vector is local, with low privileges required and no user interaction.
CVE-2023-52784: bonding device type change
The record describes a crash where a lapbether device was created on a bonding device with no members, and a non-Ethernet member was then added, forcing the bonding master to change its type. The fix makes bond_setup_by_slave() call dev_close() so that dependent devices are removed. The record lists a medium score of 5.5 and the included trace ends in a kernel panic ("kernel BUG"). The advisory text in the research is truncated and gives no attack vector, so we do not know from these sources whether it is local-only.
Who is realistically exposed
The following is general reasoning, not a finding of the sources. Whether the OSS sound, libfc or bonding code is present depends on your kernel configuration and loaded modules.
- Shared servers where several users or customers can run code, since CVE-2017-9986 is described as a local issue and the GitHub advisory for CVE-2022-49114 lists a local vector.
- Container and CI hosts, where workloads share the host kernel, so a flaw reachable from a local process could matter there.
- Servers that actually use Fibre Channel storage or bonded network interfaces, if the relevant code is built into or loaded by their kernel.
- Single-purpose web servers with no untrusted local users face lower practical risk. An attacker with a foothold through a vulnerable web app could in principle try a kernel flaw next, but the research reports nothing like that for these CVEs.
None of the research reports active exploitation. The GitHub advisories give EPSS scores of 0.362% for CVE-2017-9986 and 0.357% for CVE-2022-49114, both in the 30th percentile, which suggests a low near-term likelihood. That is a statistical estimate, not a guarantee, and the research gives no EPSS figure for CVE-2023-52784.
How to check your exposure
First find out which kernel is running and whether your distribution has a newer one waiting. These commands are read-only. The first prints the running kernel version.
uname -rThis prints the distribution release details, which you need to find the correct security notice for your release.
lsb_release -aThis lists installed kernel image packages and their versions; look for the newest installed version.
dpkg -l 'linux-image*' | grep ^iiThis shows whether an updated kernel package is waiting; any linux-image line means an update is available.
apt list --upgradable 2>/dev/null | grep -i linuxOn RHEL-family systems, this lists installed kernel packages so you can compare them with your vendor's advisory.
rpm -q kernelCompare your installed version with the fixed version in your distribution's security notice for your release. Each distribution publishes its own notices and package versions, so use the one that matches your operating system and release.
What to do now
- Find the security notice from your distribution for these CVEs and note the fixed kernel package versions for your release.
- Take a snapshot or backup of the server before changing the kernel.
- Refresh package lists and install the updated kernel with your package manager.
- Reboot into the new kernel in a planned window.
- Run uname -r after the reboot and confirm it matches the fixed version.
- Repeat for every host, including images and templates used to create new servers.
- Remove accounts and workloads on shared hosts that do not need to be there.
This refreshes package lists and installs pending updates, including kernel packages, on Debian and Ubuntu.
sudo apt update
sudo apt upgradeThis updates all packages, including the kernel, on RHEL-family systems.
sudo dnf upgradeThis restarts the server so the new kernel loads. Services are interrupted until it comes back.
sudo rebootDetection and monitoring afterward
Kernel bugs rarely leave a clear trace, so focus on hygiene and visibility. Watch for unexpected kernel panics or reboots in your logs, since use after free and similar flaw classes can end in a crash. Keep a simple inventory of kernel versions per host so you can see which machines lag behind.
- Search logs for unexplained oops or panic messages before and after patching.
- Review who has shell access and which containers run on each host.
- Subscribe to your distribution's security notices so new kernel updates reach you quickly.
For the outside view of your sites, Webforta Monitor can watch uptime and certificate changes, which helps you notice an unplanned outage after a reboot or crash. It does not detect or fix kernel flaws. For general server upkeep, see small business website security.
What we could not confirm
- Which distribution releases and kernel flavors are affected.
- The fixed package versions for any distribution.
- The attack vector for CVE-2023-52784, and the vector behind the 8.8 score for CVE-2022-49114.
- Any exploitation in the wild. The research shows no CISA KEV entry for these CVEs.
The upstream fix commits for CVE-2022-49114 and CVE-2023-52784 are listed in the CVE references on git.kernel.org, but those pages did not load in our research, so we have not relied on their contents.
Sources
This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.
