Date:   Fri, 10 Mar 2023 21:18:32 +0900
From:   Sergey Senozhatsky <senozhatsky@chromium.org>
To:     Ricardo Ribalda <ribalda@chromium.org>
Cc:     Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        linux-kernel@vger.kernel.org, linux-media@vger.kernel.org,
        Sergey Senozhatsky <senozhatsky@chromium.org>,
        "hn.chen" <hn.chen@sunplusit.com>
Subject: Re: [PATCH v7 1/6] media: uvcvideo: Fix underflow addressing on hw
 timestamp
Message-ID: <20230310121832.GH1045797@google.com>
References: <20220920-resend-hwtimestamp-v7-0-cf1d78bb8821@chromium.org>
 <20220920-resend-hwtimestamp-v7-1-cf1d78bb8821@chromium.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220920-resend-hwtimestamp-v7-1-cf1d78bb8821@chromium.org>
Precedence: bulk

On (23/03/10 10:01), Ricardo Ribalda wrote:
> If head is 0, we will be addressing clock->samples[-1], which will
> result in undefined behaviour.

[..]

>  	first = &clock->samples[clock->head];
> -	last = &clock->samples[(clock->head - 1) % clock->size];
> +	last = &clock->samples[(clock->head - 1 + clock->size) % clock->size];

Just for visibility:

Per offline discussion with Ricardo, this should not cause a samples[-1]
access, as all arithmetic operations there `unsigned`

    75ef:       41 8d 46 ff             lea    -0x1(%r14),%eax
    75f3:       31 d2                   xor    %edx,%edx
    75f5:       f7 f1                   div    %ecx
    75f7:       41 89 d5                mov    %edx,%r13d

Regardless of that, Ricardo still has a point, there is a bug in the code.