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;
Date:   Wed, 7 Jun 2023 10:09:02 +0200
From:   Miquel Raynal <miquel.raynal@bootlin.com>
To:     William Zhang <william.zhang@broadcom.com>
Cc:     Broadcom Kernel List <bcm-kernel-feedback-list@broadcom.com>,
        Linux MTD List <linux-mtd@lists.infradead.org>,
        f.fainelli@gmail.com, rafal@milecki.pl, kursad.oney@broadcom.com,
        joel.peshkin@broadcom.com, computersforpeace@gmail.com,
        anand.gore@broadcom.com, dregan@mail.com, kamal.dasu@broadcom.com,
        tomer.yacoby@broadcom.com, dan.beygelman@broadcom.com,
        Florian Fainelli <florian.fainelli@broadcom.com>,
        linux-kernel@vger.kernel.org,
        Vignesh Raghavendra <vigneshr@ti.com>,
        Richard Weinberger <richard@nod.at>,
        Kamal Dasu <kdasu.kdev@gmail.com>
Subject: Re: [PATCH 04/12] mtd: rawnand: brcmnand: Fix potential
 out-of-bounds access in oob write
Message-ID: <20230607100902.4df2bc27@xps-13>
In-Reply-To: <20230606231252.94838-5-william.zhang@broadcom.com>
References: <20230606231252.94838-1-william.zhang@broadcom.com>
        <20230606231252.94838-5-william.zhang@broadcom.com>
Organization: Bootlin
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

Hi William,

william.zhang@broadcom.com wrote on Tue,  6 Jun 2023 16:12:44 -0700:

> When the oob buffer length is not in multiple of words, the oob write
> function does out-of-bounds read on the oob source buffer at the last
> iteration. Fix that by always checking length limit on the oob buffer
> read and fill with 0xff when reaching the end of the buffer to the oob
> registers.
>=20
> Fixes: 27c5b17cd1b1 ("mtd: nand: add NAND driver "library" for Broadcom S=
TB NAND controller")
> Signed-off-by: William Zhang <william.zhang@broadcom.com>
> Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
> ---
>=20
>  drivers/mtd/nand/raw/brcmnand/brcmnand.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
>=20
> diff --git a/drivers/mtd/nand/raw/brcmnand/brcmnand.c b/drivers/mtd/nand/=
raw/brcmnand/brcmnand.c
> index 20832857c4aa..d920e88c7f5b 100644
> --- a/drivers/mtd/nand/raw/brcmnand/brcmnand.c
> +++ b/drivers/mtd/nand/raw/brcmnand/brcmnand.c
> @@ -1486,10 +1486,10 @@ static int write_oob_to_regs(struct brcmnand_cont=
roller *ctrl, int i,
> =20
>  	for (j =3D 0; j < tbytes; j +=3D 4)
>  		oob_reg_write(ctrl, j,
> -				(oob[j + 0] << 24) |
> -				(oob[j + 1] << 16) |
> -				(oob[j + 2] <<  8) |
> -				(oob[j + 3] <<  0));
> +				(((j < tbytes) ? oob[j] : 0xff) << 24) |
> +				(((j + 1 < tbytes) ? oob[j + 1] : 0xff) << 16) |
> +				(((j + 2 < tbytes) ? oob[j + 2] : 0xff) << 8) |
> +				((j + 3 < tbytes) ? oob[j + 3] : 0xff));

This is a lot of additional operations which most of the time are not
relevant. I would instead got for one less iteration in the for loop
when there is unaligned data, and then dedicated if/else to fill the
missing bytes.

>  	return tbytes;
>  }
> =20


Thanks,
Miqu=C3=A8l