[PATCH net 4/7] net: dsa: mt7530: set both CPU port interfaces to PHY_INTERFACE_MODE_NA

[Vladimir Oltean]

On Sun, Mar 26, 2023 at 05:08:15PM +0300, arinc9.unal at gmail.com wrote:
> From: Arınç ÜNAL <arinc.unal at arinc9.com>
> 
> Set interfaces of both CPU ports to PHY_INTERFACE_MODE_NA. Either phylink
> or mt7530_setup_port5() on mt7530_setup() will handle the rest.
> 
> This is already being done for port 6, do it for port 5 as well.
> 
> Fixes: 38f790a80560 ("net: dsa: mt7530: Add support for port 5")

This is getting comical.. I think I'm putting too much energy in
trying to understand the hidden meaning of this patch set.

In include/linux/phy.h we have:

typedef enum {
	PHY_INTERFACE_MODE_NA,

In lack of other initializer, the first element of an enum gets the
value 0 in C.

Then, "priv" is allocated by this driver with devm_kzalloc(), which
means that its entire memory is zero-filled. So priv->p5_interface and
priv->p6_interface are already set to 0, or PHY_INTERFACE_MODE_NA.

There is no code path between the devm_kzalloc() and the position in
mt7530_setup() that would change the value of priv->p5_interface or
priv->p6_interface from their value of 0 (PHY_INTERFACE_MODE_NA).
For example, mt753x_phylink_mac_config() can only be called from
phylink, after dsa_port_phylink_create() was called. But
dsa_port_phylink_create() comes later than ds->ops->setup() - one comes
from dsa_tree_setup_ports(), and the other from dsa_tree_setup_switches().

The movement of the priv->p6_interface assignment with a few lines
earlier does not change anything relative to the other call sites which
assign different values to priv->p6_interface, so there isn't any
functional change there, either.

So this patch is putting 0 into a variable containing 0, and claiming,
through the presence of the Fixes: tag and the submission to the "net"
tree, that it is a bug fix which should be backported to "stable".

Can it be that you are abusing the meaning of a "bug fix", and that I'm
trying too hard to take this patch set seriously?

> Tested-by: Arınç ÜNAL <arinc.unal at arinc9.com>
> Signed-off-by: Arınç ÜNAL <arinc.unal at arinc9.com>
> ---
>  drivers/net/dsa/mt7530.c | 8 ++++++--
>  1 file changed, 6 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/dsa/mt7530.c b/drivers/net/dsa/mt7530.c
> index 6d33c1050458..3deebdcfeedf 100644
> --- a/drivers/net/dsa/mt7530.c
> +++ b/drivers/net/dsa/mt7530.c
> @@ -2203,14 +2203,18 @@ mt7530_setup(struct dsa_switch *ds)
>  		mt7530_rmw(priv, MT7530_TRGMII_RD(i),
>  			   RD_TAP_MASK, RD_TAP(16));
>  
> +	/* Let phylink decide the interface later. If port 5 is used for phy
> +	 * muxing, its interface will be handled without involving phylink.
> +	 */
> +	priv->p5_interface = PHY_INTERFACE_MODE_NA;
> +	priv->p6_interface = PHY_INTERFACE_MODE_NA;
> +
>  	/* Enable port 6 */
>  	val = mt7530_read(priv, MT7530_MHWTRAP);
>  	val &= ~MHWTRAP_P6_DIS & ~MHWTRAP_PHY_ACCESS;
>  	val |= MHWTRAP_MANUAL;
>  	mt7530_write(priv, MT7530_MHWTRAP, val);
>  
> -	priv->p6_interface = PHY_INTERFACE_MODE_NA;
> -
>  	/* Enable and reset MIB counters */
>  	mt7530_mib_reset(ds);
>  
> -- 
> 2.37.2
>