Re: [BUG] KASAN: slab-use-after-free in nvmet_fc_tgt_q_put

[Shuangpeng <shuangpeng.kernel@gmail.com>]

Hi Justin,

Thanks for taking a look.

Just to clarify, the commit e8c2f9fdadee7cbc75134dc463c1e0d856d6e5c7
is only the upstream kernel revision that I tested on. I did not mean to suggest
that this commit introduced the bug, or that it is an NVMe-specific commit.

Regarding fcloop, thanks for pointing that out. I understand that nvme-fcloop
is a loopback transport and may not directly represent a deployment with
physical Fibre Channel hardware. I used it because it provides an in-tree way
to trigger the nvmet-fc target path from userspace without requiring hardware.

Best,
Shuangpeng

> On Jun 3, 2026, at 19:22, Justin Tee <justintee8345@gmail.com> wrote:
> 
> Hi Shuangpeng,
> 
> I will look into this and get back to you, but typically the nvme target fcloop module does not indicate a real life issue when utilizing the nvmet-fc stack on actual fibre channel hardware.
> 
> > on commit: e8c2f9fdadee7cbc75134dc463c1e0d856d6e5c7 (May 25 2026)
> Was wondering what this means?  Because it doesn’t look like that is an nvme commit.
> 
> Regards,
> Justin