Subject: Re: [PATCH] of: unittest: Fix out of bounds array access when id < 0
To:     Rob Herring <robh+dt@kernel.org>,
        Yin Xiujiang <yinxiujiang@kylinos.cn>
Cc:     devicetree@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20211221092830.680839-1-yinxiujiang@kylinos.cn>
 <CAL_JsqLtQanbiC4egB8d5quQRZNq0V6q2s+=qhBeypBqLH-pgA@mail.gmail.com>
From:   Frank Rowand <frowand.list@gmail.com>
Message-ID: <cadd5cd5-d2d7-ccbd-50af-deca7b6afd80@gmail.com>
Date:   Wed, 22 Dec 2021 12:06:44 -0600
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.14.0
MIME-Version: 1.0
In-Reply-To: <CAL_JsqLtQanbiC4egB8d5quQRZNq0V6q2s+=qhBeypBqLH-pgA@mail.gmail.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 12/21/21 6:20 AM, Rob Herring wrote:
> On Tue, Dec 21, 2021 at 5:34 AM Yin Xiujiang <yinxiujiang@kylinos.cn> wrote:
>>
>> In of_unittest_untrack_overlay if id is less than 0 then
>> overlay_id_bits will be out of bounds. And it is also mentioned
>> in bugzilla as a bug report:
>> https://bugzilla.kernel.org/show_bug.cgi?id=214867
>>
>> Fix this bug by tiggering WARN_ON()
> 
> If id shouldn't be less than 0, can we make it unsigned instead as
> discussed here[1].
> 
> Rob
> 
> [1] https://lore.kernel.org/all/c474a371-b524-1da8-4a67-e72cf8f2b0f7@gmail.com/
> 

There are problems with changing to unsigned (fixable, but more extensive
code changes).

I think that the implementation of id tracking in unittest.c is overly
fragile, and I would prefer to just re-implement it (see also my reply
to the proposed patch).

Let me take a look at whether I can create an alternate
implementation of id tracking.

-Frank