Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   "Fabio M. De Francesco" <fmdefrancesco@gmail.com>
To:     Dan Carpenter <dan.carpenter@oracle.com>
Cc:     Wenchao Hao <haowenchao@huawei.com>,
        syzkaller-bugs@googlegroups.com, axboe@kernel.dk,
        jejb@linux.ibm.com, linux-block@vger.kernel.org,
        linux-kernel@vger.kernel.org, linux-scsi@vger.kernel.org,
        martin.petersen@oracle.com,
        syzbot+f08c77040fa163a75a46@syzkaller.appspotmail.com,
        linfeilong@huawei.com
Subject: Re: [PATCH] scsi: sd: call device_del() if device_add_disk() fails
Date:   Thu, 31 Mar 2022 19:21:13 +0200
Message-ID: <3427592.iIbC2pHGDl@leap>
In-Reply-To: <20220331162416.GI12805@kadam>
References: <20220329154948.10350-1-fmdefrancesco@gmail.com> <1787706.atdPhlSkOF@leap> <20220331162416.GI12805@kadam>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On gioved? 31 marzo 2022 18:24:16 CEST Dan Carpenter wrote:
> On Thu, Mar 31, 2022 at 06:14:27PM +0200, Fabio M. De Francesco wrote:
> > On gioved? 31 marzo 2022 15:42:10 CEST Dan Carpenter wrote:
> > > Wenchao Hao, what you're saying makes a lot of sense but it raises a lot
> > > of questions in turn.
> > > 
> > > Fabio, did you test your patch?
> > 
> > Yes, I did, Dan. I tested it the usual way with the "#syz test:" command.
> > Obviously I have not the hardware to test code on it.
> > 
> 
> Yeah.  What a nightmare.  You posted a link to the first test.  It said
> passed but definitely introduced some use after frees but how was anyone
> supposed to know?

Maybe that a "spare-time Linux developer" like me should leave these 
kinds of bug fixes to more experienced people. But we should also note 
that I tried two or three different patches and _all_ of them passed
the tests. 

> 
> No way we would have figured this out.

I think that something should change about the way Syzbot tests patches 
and about how it provides the results. The other four or five bugs that 
I have fixed were based mainly to the fact that they passed the Syzbot 
tests. 

Perhaps I've been lucky but my patches were good and they were merged. 

However, I began to trust Syzbot too much. This is not how I should 
approach and try to solve bugs.

> I'm working to make Smatch
> understand device_put() better but this one is way difficult.
> 
> Sorry that you went through this.

Please don't be sorry :)

Believe me when I say that I cannot explain how many things I have 
learned during these days while working on this issue. I see no 
problems at all but only opportunities for learning.

Thank you very much!

Fabio M. De Francesco

> 
> regards,
> dan carpenter
> 
>