netdev - Re: [PATCH net] net: dsa: mv88e6xxx: check for address type in port_db_load

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <YPbThTUcE1QXnZJM@lunn.ch>
Date:   Tue, 20 Jul 2021 15:45:41 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     DENG Qingfang <dqfext@...il.com>
Cc:     Vivien Didelot <vivien.didelot@...il.com>,
        Florian Fainelli <f.fainelli@...il.com>,
        Vladimir Oltean <olteanv@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH net] net: dsa: mv88e6xxx: check for address type in
 port_db_load_purge

On Tue, Jul 20, 2021 at 05:24:26PM +0800, DENG Qingfang wrote:
> The same state value of an ATU entry can mean different states,
> depending on the entry's address type.
> Check for its address type instead of state, to determine if its
> portvec should be overridden.

Please could you expand this description. It is not obvious to me this
change correct.

> 
> Fixes: f72f2fb8fb6b ("net: dsa: mv88e6xxx: override existent unicast portvec in port_fdb_add")
> Signed-off-by: DENG Qingfang <dqfext@...il.com>
> ---
>  drivers/net/dsa/mv88e6xxx/chip.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/net/dsa/mv88e6xxx/chip.c b/drivers/net/dsa/mv88e6xxx/chip.c
> index beb41572d04e..dd4d7fa0da8e 100644
> --- a/drivers/net/dsa/mv88e6xxx/chip.c
> +++ b/drivers/net/dsa/mv88e6xxx/chip.c
> @@ -1741,7 +1741,7 @@ static int mv88e6xxx_port_db_load_purge(struct mv88e6xxx_chip *chip, int port,
>  		if (!entry.portvec)
>  			entry.state = 0;
>  	} else {
> -		if (state == MV88E6XXX_G1_ATU_DATA_STATE_UC_STATIC)
> +		if (is_unicast_ether_addr(addr))
>  			entry.portvec = BIT(port);
>  		else
>  			entry.portvec |= BIT(port);

I agree that MV88E6XXX_G1_ATU_DATA_STATE_UC_STATIC should only be used
with an address which is know to be unicast. If the address is
multicast, it means it should be considered as a management frame, and
given priority override.

But what exactly are we interested in here? The old code looked like
it wanted to match on static unicast entries. Your change will make it
match on static and dynamic unicast? Do we want dynamic as well? Why
is the fix not this?

> -		if (state == MV88E6XXX_G1_ATU_DATA_STATE_UC_STATIC)
> +		if (is_unicast_ether_addr(addr) && state == MV88E6XXX_G1_ATU_DATA_STATE_UC_STATIC)

which fixes the problem you describe in the commit message.

Thanks
	Andrew