On this page (6 sections)
Three Linux kernel vulnerabilities have records worth a look if you run servers: a race condition in the rxrpc networking code (CVE-2022-49149), an out-of-bounds array access in the f2fs file system (CVE-2022-49170), and a device-type problem in the bonding driver (CVE-2023-52784). Each is a bug that was fixed upstream in the kernel, and none is listed as exploited in the sources reviewed.
If you administer Linux servers, VPS instances or container hosts, the practical questions are whether your running kernel includes the fixes and whether your system even uses these components. This article explains each flaw in plain terms and shows read-only ways to check.
What the three flaws are
The records describe upstream kernel fixes. Each one is a different weakness class, so it helps to treat them separately.
CVE-2022-49149: rxrpc timer racing with call destruction
According to the GitHub advisory, the rxrpc_call structure uses a timer for timed events. Packet input code running in softirq mode holds only the RCU read lock, so a call can begin to be destroyed at the same moment a packet addressed to it arrives. That can restart a timer that was already stopped, and the timer code may later crash the kernel (an oops) if the memory was freed first. The fix takes a reference on the call and hands it to the timer and then to the work item.
This is a race condition (CWE-362): two pieces of code touch the same object at once without proper synchronization. The practical outcome described is a kernel crash, meaning loss of availability. The scores differ by source. The dataset for this story lists 9.8 critical, while the GitHub Advisory Database lists 4.7 moderate with a local attack vector, high attack complexity and high availability impact only. The research does not say who assigned either score or why they differ, so treat the exact number with caution and judge priority by your own exposure.
CVE-2022-49170: f2fs missing sanity check
The GitHub advisory says the f2fs file system code did not validate curseg->alloc_type, which led to out-of-bounds access on the sbi->block_count[] array. It was reported by Wenqing Liu when mounting and operating on a corrupted file system image, with the report tested on kernel 5.17-rc4 and 5.17-rc6. The advisory scores it 7.8 high with a local attack vector, low complexity and low privileges required. The weakness is improper validation of an array index (CWE-129).
In plain terms, a malformed f2fs image can make the kernel access memory outside an array. The advisory describes the trigger as a corrupted image, so systems that mount f2fs volumes are the ones to think about.
CVE-2023-52784: bonding device type change
Per the GitHub advisory, a lapbether device had been created on a bonding device without members. Adding a member that is not an Ethernet-type device then forced the bonding master to change its type. The advisory's trace, found by syzbot, shows a kernel panic (skb_under_panic). The fix calls dev_close() in bond_setup_by_slave() so linked lapbether devices, or any other devices that assume a physical Ethernet device, are removed. It is scored 5.5 medium. The research gives no attack vector or CWE for this one.
Who is realistically exposed
The advisories themselves point to narrow trigger conditions:
- CVE-2022-49149 and CVE-2022-49170 are scored with a local attack vector in the GitHub advisories, so an attacker generally needs a foothold or an account on the machine. The research gives no attack vector for CVE-2023-52784.
- The f2fs bug concerns mounting and operating on a corrupted f2fs image, so it matters if you use f2fs file systems at all.
- The bonding bug, as described, involves a bond with a lapbether device on top of it and then adding a non-Ethernet member. Most web servers will not have that setup.
How to check your system
Start by finding the running kernel and whether an update is waiting. These commands are read-only and are for Debian and Ubuntu systems.
uname -rdpkg -l 'linux-image*' | grep ^iiapt list --upgradable 2>/dev/null | grep -i linuxTo see whether the affected components are in use, check for loaded modules and mounted file systems. No output means the module is not currently loaded.
lsmod | grep -E 'rxrpc|f2fs|bonding'mount -t f2fsWhat to do now
- Look up your distribution's kernel security notices for these three CVE identifiers and note the fixed kernel package for your release.
- Run uname -r and apt list --upgradable (or your distribution's equivalent) to see where each server stands.
- Schedule a maintenance window, since a kernel update needs a reboot to take effect.
- Install the updated kernel packages with your normal package manager.
- Reboot, then run uname -r again to confirm the new kernel is running.
- Until patched, avoid mounting f2fs images from untrusted sources and limit who has local shell access.
- Repeat for virtual machines, cloud images and any host that pins an older kernel.
sudo apt update && sudo apt upgradesudo rebootDetection and monitoring afterwards
Bugs of this type tend to show up as crashes before anything else. Watch for unexplained reboots, kernel oops or panic messages, and out-of-range warnings from the f2fs code.
sudo dmesg | grep -iE 'oops|panic|ubsan|rxrpc|f2fs'Keep a simple inventory of kernel versions per host so you can spot stragglers after each distribution notice. If your public site depends on these servers, uptime and certificate monitoring helps you notice an unplanned reboot quickly, for example with Webforta Monitor.
What we do not know
- The research does not name any specific distribution notice (including any Ubuntu notice) that covers these CVEs, nor the affected releases or fixed package versions.
- The git.kernel.org patch pages could not be read, so details come from the GitHub advisory text and the CVE records.
- The sources do not report active exploitation, and none of the three CVEs has a CISA Known Exploited Vulnerabilities entry in the data. That can change.
- Severity differs between sources for CVE-2022-49149 (9.8 in the dataset, 4.7 in the GitHub advisory), and the research does not say who assigned either score.
For broader server hygiene, our small business website security guide covers patching routines that fit teams without a dedicated operations staff.
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.
