---
id: "CVE-2026-98960"
kind: "vulnerability"
aliases: ["CVE-2026-98960"]
title: "Linux kernel btrfs: io_uring encoded data leaks on retry, exhausting memory"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["btrfs", "io-uring", "linux-kernel", "runtime", "self-hosted"]
not_affected_if: ["io-uring-disabled", "no-btrfs"]
published_at: "2026-10-23T13:17:17+00:00"
first_seen_at: "2026-10-23T21:57:33+00:00"
updated_at: "2026-10-25T04:57:46+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-98960", "first_seen_at": "2026-10-23T21:57:33+00:00"}]
---

# Linux kernel btrfs: io_uring encoded data leaks on retry, exhausting memory

In the Linux kernel's btrfs filesystem, io_uring encoded read and write requests keep an allocation alive when they return -EAGAIN, so the memory leaks if the request is cancelled or the ring exits before it is reissued. The sources describe a memory leak and do not state an attacker-controlled impact.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Run `uname -r` and compare the kernel version against the fixed release listed in the NVD entry
- Check whether btrfs is in use with `findmnt -t btrfs`
- Check whether io_uring is enabled with `sysctl kernel.io_uring_disabled`

## Fix

- Update the Linux kernel to a release that includes the btrfs fix for CVE-2026-98960; the fixed version is unknown from the sources
- Reboot into the patched kernel after the update

## Timeline

- 2026-10-23T21:57:33+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-98960

## From the sources

In the Linux kernel, the following vulnerability has been resolved:

btrfs: don't stash io_uring encoded data across -EAGAIN

Returning -EAGAIN while leaving btrfs_uring_encoded_data in the cmd PDU
leaks if the request is cancelled or the ring exits before reissue.
io_uring does not free driver PDU allocations on cleanup.

Write: io_queue_sqe() always issues with IO_URING_F_NONBLOCK first, so
return -EAGAIN before allocating and free data on every exit.

Read: free on nowait -EAGAIN too; only -EIOCBQUEUED keeps the allocation
for btrfs_uring_read_finished().

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_