---
id: "CVE-2026-99212"
kind: "vulnerability"
aliases: ["CVE-2026-99212"]
title: "Linux kernel RDS: failed route resolution frees connection id, causing use-after-free"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["infiniband", "linux-kernel", "rdma", "rds", "roce", "runtime"]
not_affected_if: ["no-rdma-hardware", "rds-module-not-loaded"]
published_at: "2026-10-28T11:17:42+00:00"
first_seen_at: "2026-10-28T11:32:37+00:00"
updated_at: "2026-10-28T11:33:48+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-99212", "first_seen_at": "2026-10-28T11:32:37+00:00"}]
---

# Linux kernel RDS: failed route resolution frees connection id, causing use-after-free

When RDS route resolution fails synchronously on the RDMA address-resolved event, the rdma_cm layer destroys the connection's cm id while RDS still holds the pointer. The connection then operates on a freed id until shutdown, which is a use-after-free that can crash the kernel.

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

## Fix

- Update the Linux kernel to a release that includes the upstream commit 'net/rds: don't let the rdma_cm destroy an id RDS still owns on route failure'.

## Timeline

- 2026-10-28T11:32:37+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-99212

## From the sources

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

net/rds: don't let the rdma_cm destroy an id RDS still owns on route failure

rds_rdma_cm_event_handler_cmn() hands the return value of
rdma_resolve_route() straight back to the rdma_cm on
RDMA_CM_EVENT_ADDR_RESOLVED.  A synchronous failure there - -ENOMEM
from the route work allocation, an SA query that cannot be set up, a
RoCE route with no usable device - makes addr_handler() destroy the
id the event was delivered on.  That id is ic->i_cm_id, and nothing
clears the pointer: the connection sits in RDS_CONN_CONNECTING with a
freed id until its shutdown calls rdma_disconnect() and
rdma_destroy_id() on it.

The ROUTE_RESOLVED case already avoids this: rds_ib_cm_initiate_connect()
forces a zero return while ic->i_cm_id == cm_id, as the comment there
explains.  Do the same here - drop the connection and return 0 - so
the id stays RDS's to destroy from the shutdown, and the reconnect
gets a fresh one.

---

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