Re: [PATCH -next] serial: core: fix NULL/dangling port_dev on failed re-register

[Greg KH <gregkh@linuxfoundation.org>]

On Thu, Sep 24, 2026 at 09:28:23PM +0800, cuigaosheng wrote:
> Thanks for the review, I have submitted v2 of the patch.

Great, but please do not top-post, you just lost all relevant
information :(

> I have tested on x86_64 linux-next (7.2.0-rc7) with KASAN, failslab and
> fault injection debugfs enabled;
> 
> It takes two unbind rounds to reproduce since
> serial8250_unregister_port() first unregisters and then re-registers
> the port: the fault injection must hit the re-registration to plant
> the stale port_dev, and only the next unbind dereferences it.

What do you mean by this?  What fault injection and why do we care about
that if it can never hit in real life?

> The
> reproducer is a small userspace program that loops over sysfs
> bind/unbind of serial8250 and scans /proc/self/fail-nth from 1 to
> 6000, making the Nth slab allocation on the re-register path fail
> with -ENOMEM:

bind/unbind is not a normal operation that a user can do, and is for
debugging only.  So are you sure this is a real issue?

thanks,

greg k-h