Re: [PATCH v4] serial: 8250: fix use-after-free in IRQ chain handling

[Jing Wu]

From: Qiliang Yuan <realwujing@xxxxxxxxx>

On Fri, Jun 12, 2026 at 11:49:51AM +0200, Greg Kroah-Hartman wrote:
> What real systems causes this to happen?  How are you triggering this
> warning to happen?  How was this tested?

The original report is Bugzilla #221579 from Wang Zhaolong.  The bug
triggers on systems with multiple 8250 serial ports sharing an IRQ
(e.g. NAS-like devices).  It can be reproduced by probing/removing
8250 serial ports that share an IRQ.  Wang confirmed off-list that v3
fixes the reproducer on his setup.

> Shouldn't the function be marked as requiring this lock to be held?
> Just putting in this lockdep_assert will not catch the static analysis
> tools :(

Agreed, will add __must_hold(&hash_mutex) in v5.

Thanks,
Qiliang