linux-kernel - Re: [PATCH RFC net-next v2 16/17] net: stmmac: Move internal PCS init method to stmmac

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]

Message-ID: <Zn7KZG+KDU01APar@shell.armlinux.org.uk>
Date: Fri, 28 Jun 2024 15:36:20 +0100
From: "Russell King (Oracle)" <linux@...linux.org.uk>
To: Serge Semin <fancer.lancer@...il.com>
Cc: Andrew Halaney <ahalaney@...hat.com>,
	Alexandre Torgue <alexandre.torgue@...s.st.com>,
	Jose Abreu <joabreu@...opsys.com>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	Maxime Coquelin <mcoquelin.stm32@...il.com>,
	Alexei Starovoitov <ast@...nel.org>,
	Jesper Dangaard Brouer <hawk@...nel.org>,
	John Fastabend <john.fastabend@...il.com>,
	Daniel Borkmann <daniel@...earbox.net>,
	linux-arm-kernel@...ts.infradead.org,
	linux-stm32@...md-mailman.stormreply.com, bpf@...r.kernel.org,
	netdev@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH RFC net-next v2 16/17] net: stmmac: Move internal PCS
 init method to stmmac_pcs.c

On Mon, Jun 24, 2024 at 04:26:33PM +0300, Serge Semin wrote:
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index 72c2d3e2c121..743d356f6d12 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -950,13 +950,16 @@ static struct phylink_pcs *stmmac_mac_select_pcs(struct phylink_config *config,
>  {
>  	struct stmmac_priv *priv = netdev_priv(to_net_dev(config->dev));
>  
> +	if (priv->hw->pcs)
> +		return &priv->hw->mac_pcs;
> +
>  	if (priv->hw->xpcs)
>  		return &priv->hw->xpcs->pcs;
>  
>  	if (priv->hw->phylink_pcs)
>  		return priv->hw->phylink_pcs;
>  
> -	return stmmac_mac_phylink_select_pcs(priv, interface);
> +	return NULL;

I really really don't like this due to:

1. I spent a long time working out what the priority here should be, and
you've just thrown all that work away by changing it - to something that
I believe is incorrect.

2. I want to eventually see this function checking the interface type
before just handing out a random PCS, and it was my intention to
eventually that into the MACs own select_pcs() methods. Getting rid of
those methods means that the MACs themselves now can't make the
decision which is where that should be.

3. When operating in RGMII "inband" mode, the .pcs_config etc doesn't
make much sense (we're probably accessing registers that don't exist)
and I had plans to split this into a RGMII "PCS" which was just a PCS
that implemented .pcs_get_state(), a stub .pcs_config(), and a separate
fully-featured "SGMII PCS".

So, I would like to eventually see here something like:

	if (priv->hw->xpcs)
		return &priv->hw->xpcs->pcs;

	if (priv->hw->phylink_pcs)
		return priv->hw->phylink_pcs;

	if (!(priv->plat->flags & STMMAC_FLAG_HAS_INTEGRATED_PCS)) {
		if (phy_interface_mode_is_rgmii(priv->plat->mac_interface))
			return &priv->hw->mac_rgmii_pcs;

		if (priv->dma_cap.pcs &&
		    priv->plat->mac_interface == PHY_INTERFACE_MODE_SGMII)
			return &priv->hw->mac_sgmii_pcs;
	}

	return NULL;

> +void dwmac_pcs_init(struct mac_device_info *hw)
> +{
> +	struct stmmac_priv *priv = hw->priv;
> +	int interface = priv->plat->mac_interface;
> +
> +	if (priv->plat->flags & STMMAC_FLAG_HAS_INTEGRATED_PCS)
> +		return;
> +	else if (phy_interface_mode_is_rgmii(interface))
> +		hw->pcs = STMMAC_PCS_RGMII;
> +	else if (priv->dma_cap.pcs && interface == PHY_INTERFACE_MODE_SGMII)
> +		hw->pcs = STMMAC_PCS_SGMII;
> +
> +	hw->mac_pcs.neg_mode = true;
> +}

Please move "hw->mac_pcs.neg_mode = true;" to where the PCS method
functions are implemented - it determines whether the PCS method
functions take the AN mode or the neg mode, and this is a property of
their implementations. It should not be split away from them.

Thanks.

-- 
RMK's Patch system: https://www.armlinux.org.uk/developer/patches/
FTTP is here! 80Mbps down 10Mbps up. Decent connectivity at last!