Re: [net-next v14 06/12] net: mtip: Add net_device_ops functions to the L2 switch driver

From: Lukasz Majewski
Date: Wed Jul 09 2025 - 06:11:28 EST


Hi Paolo,

> On 7/1/25 1:49 PM, Lukasz Majewski wrote:
> > +static netdev_tx_t mtip_start_xmit_port(struct sk_buff *skb,
> > + struct net_device *dev,
> > int port) +{
> > + struct mtip_ndev_priv *priv = netdev_priv(dev);
> > + struct switch_enet_private *fep = priv->fep;
> > + unsigned short status;
> > + struct cbd_t *bdp;
> > + void *bufaddr;
> > +
> > + spin_lock(&fep->hw_lock);
>
> mtip_start_xmit_port() runs with BH disabled. The above lock variant
> is inconsistent with what you use in patch 4.

I've looked into the fec_main.c driver. They use for TX path
__netif_tx_lock(nq, cpu); which is a simple spin_lock(). I've followed
the same approach (as _irqsave() seems to be an overkill).

This function (mtip_start_xmit_port()) is call as a callback from:
.ndo_start_xmit (member of struct net_device_ops).

IIRC net core code provides locking on this call anyway.

> Please be sure to run
> tests vs the next iteration with CONFIG_PROVE_LOCKING enabled.

This is already enabled. Locking in this driver is a bit special, as
one uDMA is used for both ports... (unlikely as in fec_main.c).

>
> /P
>




Best regards,

Lukasz Majewski

--

DENX Software Engineering GmbH, Managing Director: Johanna Denk,
Tabea Lutz HRB 165235 Munich, Office: Kirchenstr.5, D-82194
Groebenzell, Germany
Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@xxxxxxx

Attachment: pgpCLw9YKWYMb.pgp
Description: OpenPGP digital signature