Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Fri, 3 Sep 2021 05:38:34 +0100
From:   Christoph Hellwig <hch@infradead.org>
To:     Christian L?hle <CLoehle@hyperstone.com>
Cc:     "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-scsi@vger.kernel.org" <linux-scsi@vger.kernel.org>,
        "martin.petersen@oracle.com" <martin.petersen@oracle.com>,
        "jejb@linux.ibm.com" <jejb@linux.ibm.com>,
        "stable@vger.kernel.org" <stable@vger.kernel.org>
Subject: Re: [PATCH] sd: sd_open: prevent device removal during sd_open
Message-ID: <YTGmyp0kzhVsuk5K@infradead.org>
References: <98bfca4cbaa24462994bcb533d365414@hyperstone.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <98bfca4cbaa24462994bcb533d365414@hyperstone.com>
Precedence: bulk

On Thu, Sep 02, 2021 at 01:57:13PM +0000, Christian L?hle wrote:
> sd and parent devices must not be removed as sd_open checks for events
> 
> sd_need_revalidate and sd_revalidate_disk traverse the device path
> to check for event changes. If during this, e.g. the scsi host is being
> removed and its resources freed, this traversal crashes.
> Locking with scan_mutex for just a scsi disk open may seem blunt, but there
> does not seem to be a more granular option. Also opening /dev/sdX directly
> happens rarely enough that this shouldn't cause any issues.

Can you please root cause how the device could not be valid, as that
should not happen?

>
> The issue occurred on an older kernel with the following trace:
> stack segment: 0000 [#1] PREEMPT SMP PTI
> CPU: 1 PID: 121457 Comm: python3 Not tainted 4.14.238hyLinux #1

.. preferably with a current mainline kernel as things changed a lot
in this area.