netdev - Re: [PATCH v2 06/11] net: phylink: Support differing link/interface speed/duplex

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

Date:   Wed, 20 Jul 2022 07:43:59 +0100
From:   "Russell King (Oracle)" <linux@...linux.org.uk>
To:     Sean Anderson <sean.anderson@...o.com>
Cc:     netdev@...r.kernel.org, Andrew Lunn <andrew@...n.ch>,
        Heiner Kallweit <hkallweit1@...il.com>,
        Alexandru Marginean <alexandru.marginean@....com>,
        Paolo Abeni <pabeni@...hat.com>,
        "David S . Miller" <davem@...emloft.net>,
        linux-kernel@...r.kernel.org, Vladimir Oltean <olteanv@...il.com>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>
Subject: Re: [PATCH v2 06/11] net: phylink: Support differing link/interface
 speed/duplex

On Tue, Jul 19, 2022 at 07:49:56PM -0400, Sean Anderson wrote:
> This adds support for cases when the link speed or duplex differs from
> the speed or duplex of the phy interface mode. Such cases can occur when
> some kind of rate adaptation is occurring.
> 
> The following terms are used within this and the following patches. I
> do not believe the meaning of these terms are uncommon or surprising,
> but for maximum clarity I would like to be explicit:
> 
> - Phy interface mode: the protocol used to communicate between the MAC
>   or PCS (if used) and the phy. If no phy is in use, this is the same as
>   the link mode. Each phy interface mode supported by Linux is a member
>   of phy_interface_t.
> - Link mode: the protocol used to communicate between the local phy (or
>   PCS) and the remote phy (or PCS) over the physical medium. Each link
>   mode supported by Linux is a member of ethtool_link_mode_bit_indices.
> - Phy interface mode speed: the speed of unidirectional data transfer
>   over a phy interface mode, including encoding overhead, but excluding
>   protocol and flow-control overhead. The speed of a phy interface mode
>   may vary. For example, SGMII may have a speed of 10, 100, or 1000
>   Mbit/s.
> - Link mode speed: similarly, the speed of unidirectional data transfer
>   over a physical medium, including overhead, but excluding protocol and
>   flow-control overhead. The speed of a link mode is usually fixed, but
>   some exceptional link modes (such as 2BASE-TL) may vary their speed
>   depending on the medium characteristics.
> 
> Before this patch, phylink assumed that the link mode speed was the same
> as the phy interface mode speed. This is typically the case; however,
> some phys have the ability to adapt between differing link mode and phy
> interface mode speeds. To support these phys, this patch removes this
> assumption, and adds a separate variable for link speed. Additionally,
> to support rate adaptation, a MAC may need to have a certain duplex
> (such as half or full). This may be different from the link's duplex. To
> keep track of this distunction, this patch adds another variable to
> track link duplex.

I thought we had decided that using the term "link" in these new members
was a bad idea.

> @@ -925,12 +944,16 @@ static void phylink_mac_pcs_get_state(struct phylink *pl,
>  	linkmode_zero(state->lp_advertising);
>  	state->interface = pl->link_config.interface;
>  	state->an_enabled = pl->link_config.an_enabled;
> -	if  (state->an_enabled) {
> +	if (state->an_enabled) {
> +		state->link_speed = SPEED_UNKNOWN;
> +		state->link_duplex = DUPLEX_UNKNOWN;
>  		state->speed = SPEED_UNKNOWN;
>  		state->duplex = DUPLEX_UNKNOWN;
>  		state->pause = MLO_PAUSE_NONE;
>  	} else {
> -		state->speed =  pl->link_config.speed;
> +		state->link_speed = pl->link_config.link_speed;
> +		state->link_duplex = pl->link_config.link_duplex;
> +		state->speed = pl->link_config.speed;
>  		state->duplex = pl->link_config.duplex;
>  		state->pause = pl->link_config.pause;
>  	}
> @@ -944,6 +967,9 @@ static void phylink_mac_pcs_get_state(struct phylink *pl,
>  		pl->mac_ops->mac_pcs_get_state(pl->config, state);
>  	else
>  		state->link = 0;
> +
> +	state->link_speed = state->speed;
> +	state->link_duplex = state->duplex;

Why do you need to set link_speed and link_duple above if they're always
copied over here?

>  /* The fixed state is... fixed except for the link state,
> @@ -953,10 +979,17 @@ static void phylink_get_fixed_state(struct phylink *pl,
>  				    struct phylink_link_state *state)
>  {
>  	*state = pl->link_config;
> -	if (pl->config->get_fixed_state)
> +	if (pl->config->get_fixed_state) {
>  		pl->config->get_fixed_state(pl->config, state);
> -	else if (pl->link_gpio)
> +		/* FIXME: these should not be updated, but
> +		 * bcm_sf2_sw_fixed_state does it anyway
> +		 */
> +		state->link_speed = state->speed;
> +		state->link_duplex = state->duplex;
> +		phylink_state_fill_speed_duplex(state);

This looks weird. Why copy state->xxx to state->link_xxx and then copy
them back to state->xxx in a helper function?

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 40Mbps down 10Mbps up. Decent connectivity at last!