CVE-2026-98289

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

af unix: Unify scc index when finalising SCC in unix walk scc().

Commit bfdb01283ee8 ("af unix: Assign a unique index to SCC.") changed Tarjan's algorithm to update lowlink with lowlink, which is called lowpoint (unix vertex.scc index).

unix vertex dead() assumes all vertices in an SCC share the same lowpoint, but this is not always true if an SCC has two or more back edges, depending on the order of DFS.

For example, the graph below has two back edges from B to A and from C to B.

A --> B --> C ^ | ^ | ----' ----'

If DFS walks through A -> B -> C -> B (-> C -> B) -> A (-> B -> A), each index and scc index will be updated as follows.

A --> B --> C C = (3, 3) (index, scc index) B = (2, 2) A = (1, 1)

A ... B ... C C = (3, 2)<-. ^ | B = (2, 2) -' `----' A = (1, 1)

A ... B ... C C = (3, 2) ^ | . . B = (2, 1)<-. `----' .... A = (1, 1) -'

Then, unix vertex dead() thinks that B is passed to another SCC with scc index 2, and the SCC is not garbage-collected.

This does not happen if DFS walks in a different order below or starts from B.

1 3 A --> B --> C ^ | ^ | ----' ----' 2 4

Let's unify scc index across the SCC when finalising it.

Note that updating v->index was previously done in unix scc dead(), when called from unix walk scc(), just to save one loop. Since unix walk scc() now iterates over the SCC anyway, the update is moved back to unix walk scc() and 'fast' argument is dropped.