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:   Tue, 22 Jun 2021 09:44:19 +0300
From:   Dan Carpenter <dan.carpenter@oracle.com>
To:     Zhihao Cheng <chengzhihao1@huawei.com>
Cc:     Colin King <colin.king@canonical.com>,
        Richard Weinberger <richard@nod.at>,
        linux-mtd@lists.infradead.org, kernel-janitors@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] ubifs: Remove a redundant null check on pointer lp
Message-ID: <20210622064419.GJ1901@kadam>
References: <20210621152249.20901-1-colin.king@canonical.com>
 <f2ea606a-4a05-8d14-4380-d96ca0f981a1@huawei.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <f2ea606a-4a05-8d14-4380-d96ca0f981a1@huawei.com>
User-Agent: Mutt/1.9.4 (2018-02-28)
Precedence: bulk

On Tue, Jun 22, 2021 at 10:38:52AM +0800, Zhihao Cheng wrote:
> 在 2021/6/21 23:22, Colin King 写道:
> > From: Colin Ian King <colin.king@canonical.com>
> > 
> > An earlier fix to replace an IS_ERR check on lp with a null check
> > on lp didn't remove a following null check on lp. The second null
> > check is redundant and can be removed.
> > 
> > Addresses-Coverity: ("Logically dead code")
> > Fixes: c770cd5190ba ("ubifs: fix an IS_ERR() vs NULL check")
> > Signed-off-by: Colin Ian King <colin.king@canonical.com>
> > ---
> >   fs/ubifs/gc.c | 2 --
> >   1 file changed, 2 deletions(-)
> > 
> > diff --git a/fs/ubifs/gc.c b/fs/ubifs/gc.c
> > index 7cc22d7317ea..465beea52176 100644
> > --- a/fs/ubifs/gc.c
> > +++ b/fs/ubifs/gc.c
> > @@ -899,8 +899,6 @@ int ubifs_gc_start_commit(struct ubifs_info *c)
> >   			err = -ENOMEM;
> >   			goto out;
> >   		}
> Hi Colin,
> I just found out about it today thanks to your patch. Commit c770cd5190ba
> ("ubifs: fix an IS_ERR() vs NULL check") did import a new problem that
> ubifs_gc_start_commit() may return -ENOMEM while syncing fs.
> I guess ubifs_fast_find_frdi_idx() return NULL pointer is the termination
> condition in while-loop, which means we cannot get a freeable index LEB in
> ubifs_gc_start_commit().

Ugh...  I'm so sorry.  My patch was clearly wrong.  I've tried before to
add a Smatch check which warns about duplicative NULL checks, but I
think this gives me a new idea to try.  Hopefully, it will prevent this
in the future.

Yeah, and it's my check which needs to be deleted, not the other one.

regards,
dan carpenter