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:   Alexander Lobakin <alexandr.lobakin@intel.com>
To:     Andrew Lunn <andrew@lunn.ch>
Cc:     Alexander Lobakin <alexandr.lobakin@intel.com>,
        Maxim Korotkov <korotkov.maxim.s@gmail.com>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Guangbin Huang <huangguangbin2@huawei.com>,
        Tom Rix <trix@redhat.com>, Marco Bonelli <marco@mebeim.net>,
        Edward Cree <ecree@solarflare.com>, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, lvc-project@linuxtesting.org
Subject: Re: [PATCH] ethtool: avoiding integer overflow in ethtool_phys_id()
Date:   Mon, 21 Nov 2022 16:03:14 +0100
Message-Id: <20221121150314.393682-1-alexandr.lobakin@intel.com>
In-Reply-To: <Y3uGyrxeSbajJqpr@lunn.ch>
References: <20221121075618.15877-1-korotkov.maxim.s@gmail.com> <Y3uGyrxeSbajJqpr@lunn.ch>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Andrew Lunn <andrew@lunn.ch>
Date: Mon, 21 Nov 2022 15:10:18 +0100

> On Mon, Nov 21, 2022 at 10:56:18AM +0300, Maxim Korotkov wrote:
> > The value of an arithmetic expression "n * id.data" is subject
> > to possible overflow due to a failure to cast operands to a larger data
> > type before performing arithmetic. Added cast of first operand to u64
> > for avoiding overflow.
> > 
> > Found by Linux Verification Center (linuxtesting.org) with SVACE.
> > 
> > Fixes: 2adc6edcaec0 ("ethtool: fix error handling in ethtool_phys_id")
> > Signed-off-by: Maxim Korotkov <korotkov.maxim.s@gmail.com>
> > ---
> >  net/ethtool/ioctl.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/net/ethtool/ioctl.c b/net/ethtool/ioctl.c
> > index 6a7308de192d..cf87e53c2e74 100644
> > --- a/net/ethtool/ioctl.c
> > +++ b/net/ethtool/ioctl.c
> > @@ -2007,7 +2007,7 @@ static int ethtool_phys_id(struct net_device *dev, void __user *useraddr)
> >  	} else {
> >  		/* Driver expects to be called at twice the frequency in rc */
> >  		int n = rc * 2, interval = HZ / n;
> > -		u64 count = n * id.data, i = 0;
> > +		u64 count = (u64)n * id.data, i = 0;
> 
> 
> How about moving the code around a bit, change n to a u64 and drop the
> cast? Does this look correct?
> 
> 		int interval = HZ / rc / 2;
> 		u64 n = rc * 2;
> 		u64 count = n * id.data;
> 
> 		i = 0;
> 
> I just don't like casts, they suggest the underlying types are wrong,
> so should fix that, not add a cast.

This particular one is absolutely fine. When you want to multiply
u32 by u32, you always need a cast, otherwise the result will be
truncated. mul_u32_u32() does it the very same way[0].

> 
> 	Andrew
> 

[0] https://elixir.bootlin.com/linux/v6.1-rc6/source/include/linux/math64.h#L153

Thanks,
Olek