RE: [PATCH net-next 09/18] ravb: Factorise ravb_ring_free function

From: Biju Das <biju.das.jz@bp.renesas.com>
To: Sergei Shtylyov <sergei.shtylyov@gmail.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>
Cc: Geert Uytterhoeven <geert+renesas@glider.be>,
	Sergey Shtylyov <s.shtylyov@omprussia.ru>,
	Adam Ford <aford173@gmail.com>, Andrew Lunn <andrew@lunn.ch>,
	Yuusuke Ashizuka <ashiduka@fujitsu.com>,
	Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"linux-renesas-soc@vger.kernel.org" 
	<linux-renesas-soc@vger.kernel.org>,
	Chris Paterson <Chris.Paterson2@renesas.com>,
	Biju Das <biju.das@bp.renesas.com>,
	Prabhakar Mahadev Lad <prabhakar.mahadev-lad.rj@bp.renesas.com>
Subject: RE: [PATCH net-next 09/18] ravb: Factorise ravb_ring_free function
Date: Fri, 30 Jul 2021 06:21:45 +0000	[thread overview]
Message-ID: <OS0PR01MB59223D88215676866ABD813586EC9@OS0PR01MB5922.jpnprd01.prod.outlook.com> (raw)
In-Reply-To: <a0d1bb7e-0e0a-8237-c30a-e4533b5580dd@gmail.com>

Hi Sergei,

Thanks for the feedback.

> Subject: Re: [PATCH net-next 09/18] ravb: Factorise ravb_ring_free
> function
> 
> Hello!
> 
> On 7/22/21 5:13 PM, Biju Das wrote:
> 
> > Extended descriptor support in RX is available for R-Car where as it
> > is a normal descriptor for RZ/G2L. Factorise ravb_ring_free function
> > so that it can support later SoC.
> >
> > Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
> > Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
> > ---
> >  drivers/net/ethernet/renesas/ravb.h      |  5 +++
> >  drivers/net/ethernet/renesas/ravb_main.c | 49
> > ++++++++++++++++--------
> >  2 files changed, 37 insertions(+), 17 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/renesas/ravb.h
> > b/drivers/net/ethernet/renesas/ravb.h
> > index a474ed68db22..3a9cf6e8671a 100644
> > --- a/drivers/net/ethernet/renesas/ravb.h
> > +++ b/drivers/net/ethernet/renesas/ravb.h
> > @@ -988,7 +988,12 @@ enum ravb_chip_id {
> >  	RCAR_GEN3,
> >  };
> >
> > +struct ravb_ops {
> > +	void (*ring_free)(struct net_device *ndev, int q);
> 
>    Hmm, why not store it right in the *struct* ravb_drv_data?

OK.

> 
> > +};
> > +
> >  struct ravb_drv_data {
> > +	const struct ravb_ops *ravb_ops;
> >  	netdev_features_t net_features;
> >  	netdev_features_t net_hw_features;
> >  	const char (*gstrings_stats)[ETH_GSTRING_LEN];
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c
> > b/drivers/net/ethernet/renesas/ravb_main.c
> > index 4ef2565534d2..a3b8b243fd54 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -247,30 +247,39 @@ static int ravb_tx_free(struct net_device *ndev,
> > int q, bool free_txed_only)  }
> >
> >  /* Free skb's and DMA buffers for Ethernet AVB */ -static void
> > ravb_ring_free(struct net_device *ndev, int q)
> > +static void ravb_ring_free_rx(struct net_device *ndev, int q)
> 
>    How about ravb_rx_ring_free() instead?
Agreed.

> 
> >  {
> >  	struct ravb_private *priv = netdev_priv(ndev);
> > -	int num_tx_desc = priv->num_tx_desc;
> >  	int ring_size;
> >  	int i;
> >
> > -	if (priv->rx_ring[q]) {
> > -		for (i = 0; i < priv->num_rx_ring[q]; i++) {
> > -			struct ravb_ex_rx_desc *desc = &priv->rx_ring[q][i];
> > +	for (i = 0; i < priv->num_rx_ring[q]; i++) {
> > +		struct ravb_ex_rx_desc *desc = &priv->rx_ring[q][i];
> >
> > -			if (!dma_mapping_error(ndev->dev.parent,
> > -					       le32_to_cpu(desc->dptr)))
> > -				dma_unmap_single(ndev->dev.parent,
> > -						 le32_to_cpu(desc->dptr),
> > -						 RX_BUF_SZ,
> > -						 DMA_FROM_DEVICE);
> > -		}
> > -		ring_size = sizeof(struct ravb_ex_rx_desc) *
> > -			    (priv->num_rx_ring[q] + 1);
> > -		dma_free_coherent(ndev->dev.parent, ring_size, priv-
> >rx_ring[q],
> > -				  priv->rx_desc_dma[q]);
> > -		priv->rx_ring[q] = NULL;
> > +		if (!dma_mapping_error(ndev->dev.parent,
> > +				       le32_to_cpu(desc->dptr)))
> > +			dma_unmap_single(ndev->dev.parent,
> > +					 le32_to_cpu(desc->dptr),
> > +					 RX_BUF_SZ,
> > +					 DMA_FROM_DEVICE);
> >  	}
> > +	ring_size = sizeof(struct ravb_ex_rx_desc) *
> > +		    (priv->num_rx_ring[q] + 1);
> > +	dma_free_coherent(ndev->dev.parent, ring_size, priv->rx_ring[q],
> > +			  priv->rx_desc_dma[q]);
> > +	priv->rx_ring[q] = NULL;
> 
>    Couldn't this be moved into the new ravb_ring_free(), like the initial
> NULL check?

For RZ/G2L, it is priv->rgeth_rx_ring, that is the reason.

I can move the initial NULL check here, so the generic ravb_ring_free does not differentiate between
priv->rx_ring and priv->rgeth_rx_ring. See below.

> 
> > +}
> > +
> > +static void ravb_ring_free(struct net_device *ndev, int q) {
> > +	struct ravb_private *priv = netdev_priv(ndev);
> > +	const struct ravb_drv_data *info = priv->info;
> > +	int num_tx_desc = priv->num_tx_desc;
> > +	int ring_size;
> > +	int i;
> > +
> > +	if (priv->rx_ring[q])
> > +		info->ravb_ops->ring_free(ndev, q);
> 
>    ... here?

     It will be just  info->ravb_ops->ring_free(ndev, q);
And NULL check will be handled respective rx helper function.

What do you think?

Cheers,
Biju

> 
> [...]
> 
> MBR, Sergei