Re: [PATCH net v2 2/2] net: stmmac: Consider Tx VLAN offload tag length for maxSDU

[Jakub Kicinski <kuba@kernel.org>]

On Wed, 17 Sep 2025 15:49:20 -0700 Jakub Kicinski wrote:
> On Mon, 15 Sep 2025 16:17:19 +0800 Rohan G Thomas via B4 Relay wrote:
> > From: Rohan G Thomas <rohan.g.thomas@altera.com>
> > 
> > On hardware with Tx VLAN offload enabled, add the VLAN tag
> > length to the skb length before checking the Qbv maxSDU.
> > Add 4 bytes for 802.1Q an add 8 bytes for 802.1AD tagging.
> > 
> > Fixes: c5c3e1bfc9e0 ("net: stmmac: Offload queueMaxSDU from tc-taprio")
> > Signed-off-by: Rohan G Thomas <rohan.g.thomas@altera.com>
> > Reviewed-by: Matthew Gerlach <matthew.gerlach@altera.com>
> > ---
> >  drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 25 ++++++++++++++++-------
> >  1 file changed, 18 insertions(+), 7 deletions(-)
> > 
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > index 8c8ca5999bd8ad369eafa0cd8448a15da55be86b..c06c947ef7764bf40291a556984651f4edd7cb74 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > @@ -4537,6 +4537,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
> >  	bool has_vlan, set_ic;
> >  	int entry, first_tx;
> >  	dma_addr_t des;
> > +	u32 sdu_len;
> >  
> >  	tx_q = &priv->dma_conf.tx_queue[queue];
> >  	txq_stats = &priv->xstats.txq_stats[queue];
> > @@ -4553,13 +4554,6 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
> >  			return stmmac_tso_xmit(skb, dev);
> >  	}
> >  
> > -	if (priv->est && priv->est->enable &&
> > -	    priv->est->max_sdu[queue] &&
> > -	    skb->len > priv->est->max_sdu[queue]){
> > -		priv->xstats.max_sdu_txq_drop[queue]++;
> > -		goto max_sdu_err;
> > -	}
> > -
> >  	if (unlikely(stmmac_tx_avail(priv, queue) < nfrags + 1)) {
> >  		if (!netif_tx_queue_stopped(netdev_get_tx_queue(dev, queue))) {
> >  			netif_tx_stop_queue(netdev_get_tx_queue(priv->dev,
> > @@ -4575,6 +4569,23 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev)
> >  	/* Check if VLAN can be inserted by HW */
> >  	has_vlan = stmmac_vlan_insert(priv, skb, tx_q);
> >  
> > +	sdu_len = skb->len;
> > +	if (has_vlan) {
> > +		/* Add VLAN tag length to sdu length in case of txvlan offload */
> > +		if (priv->dev->features & NETIF_F_HW_VLAN_CTAG_TX)
> > +			sdu_len += VLAN_HLEN;
> > +		if (skb->vlan_proto == htons(ETH_P_8021AD) &&
> > +		    priv->dev->features & NETIF_F_HW_VLAN_STAG_TX)
> > +			sdu_len += VLAN_HLEN;  
> 
> Is the device adding the same VLAN tag twice if the proto is 8021AD?
> It looks like it from the code, but how every strange..
> 
> In any case, it doesn't look like the driver is doing anything with 
> the NETIF_F_HW_VLAN_* flags right? stmmac_vlan_insert() works purely
> off of vlan proto. So I think we should do the same thing here?

I suppose the double tagging depends on the exact SKU but first check
looks unnecessary. Maybe stmmac_vlan_insert() should return the number
of vlans it decided to insert?

> > +	}
> > +
> > +	if (priv->est && priv->est->enable &&
> > +	    priv->est->max_sdu[queue] &&
> > +	    sdu_len > priv->est->max_sdu[queue]) {
> > +		priv->xstats.max_sdu_txq_drop[queue]++;
> > +		goto max_sdu_err;
> > +	}
> > +
> >  	entry = tx_q->cur_tx;
> >  	first_entry = entry;
> >  	WARN_ON(tx_q->tx_skbuff[first_entry]);
> >   
>