linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Vladimir Oltean <olteanv@gmail.com>
To: Ansuel Smith <ansuelsmth@gmail.com>
Cc: Andrew Lunn <andrew@lunn.ch>,
	Vivien Didelot <vivien.didelot@gmail.com>,
	Florian Fainelli <f.fainelli@gmail.com>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	linux-kernel@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: [net-next PATCH RFC v6 12/16] net: dsa: qca8k: add support for mdio read/write in Ethernet packet
Date: Wed, 15 Dec 2021 11:49:12 +0200	[thread overview]
Message-ID: <20211215094912.gkqq4pfwac7gqeaa@skbuf> (raw)
In-Reply-To: <20211214224409.5770-13-ansuelsmth@gmail.com>

On Tue, Dec 14, 2021 at 11:44:05PM +0100, Ansuel Smith wrote:
> Add qca8k side support for mdio read/write in Ethernet packet.
> qca8k supports some specially crafted Ethernet packet that can be used
> for mdio read/write instead of the legacy method uart/internal mdio.
> This add support for the qca8k side to craft the packet and enqueue it.
> Each port and the qca8k_priv have a special struct to put data in it.
> The completion API is used to wait for the packet to be received back
> with the requested data.
> 
> The various steps are:
> 1. Craft the special packet with the qca hdr set to mdio read/write
>    mode.
> 2. Set the lock in the dedicated mdio struct.
> 3. Reinit the completion.
> 4. Enqueue the packet.
> 5. Wait the packet to be received.
> 6. Use the data set by the tagger to complete the mdio operation.
> 
> If the completion timeouts or the ack value is not true, the legacy
> mdio way is used.
> 
> It has to be considered that in the initial setup mdio is still used and
> mdio is still used until DSA is ready to accept and tag packet.
> 
> tag_proto_connect() is used to fill the required handler for the tagger
> to correctly parse and elaborate the special Ethernet mdio packet.
> 
> Locking is added to qca8k_master_change() to make sure no mdio Ethernet
> are in progress.
> 
> Signed-off-by: Ansuel Smith <ansuelsmth@gmail.com>
> ---
>  drivers/net/dsa/qca8k.c | 192 ++++++++++++++++++++++++++++++++++++++++
>  drivers/net/dsa/qca8k.h |  13 +++
>  2 files changed, 205 insertions(+)
> 
> diff --git a/drivers/net/dsa/qca8k.c b/drivers/net/dsa/qca8k.c
> index f317f527dd6d..b35ba26a0696 100644
> --- a/drivers/net/dsa/qca8k.c
> +++ b/drivers/net/dsa/qca8k.c
> @@ -20,6 +20,7 @@
>  #include <linux/phylink.h>
>  #include <linux/gpio/consumer.h>
>  #include <linux/etherdevice.h>
> +#include <linux/dsa/tag_qca.h>
>  
>  #include "qca8k.h"
>  
> @@ -170,6 +171,158 @@ qca8k_rmw(struct qca8k_priv *priv, u32 reg, u32 mask, u32 write_val)
>  	return regmap_update_bits(priv->regmap, reg, mask, write_val);
>  }
>  
> +static void qca8k_rw_reg_ack_handler(struct dsa_port *dp, struct sk_buff *skb)
> +{
> +	struct qca8k_mdio_hdr_data *mdio_hdr_data;
> +	struct qca8k_priv *priv = dp->ds->priv;
> +	struct mdio_ethhdr *mdio_ethhdr;
> +	u8 len, cmd;
> +
> +	mdio_ethhdr = (struct mdio_ethhdr *)skb_mac_header(skb);
> +	mdio_hdr_data = &priv->mdio_hdr_data;
> +
> +	cmd = FIELD_GET(QCA_HDR_MDIO_CMD, mdio_ethhdr->command);
> +	len = FIELD_GET(QCA_HDR_MDIO_LENGTH, mdio_ethhdr->command);
> +
> +	/* Make sure the seq match the requested packet */
> +	if (mdio_ethhdr->seq == mdio_hdr_data->seq)
> +		mdio_hdr_data->ack = true;
> +
> +	if (cmd == MDIO_READ) {
> +		mdio_hdr_data->data[0] = mdio_ethhdr->mdio_data;
> +
> +		/* Get the rest of the 12 byte of data */
> +		if (len > QCA_HDR_MDIO_DATA1_LEN)
> +			memcpy(mdio_hdr_data->data + 1, skb->data,
> +			       QCA_HDR_MDIO_DATA2_LEN);
> +	}
> +
> +	complete(&mdio_hdr_data->rw_done);
> +}
> +
> +static struct sk_buff *qca8k_alloc_mdio_header(enum mdio_cmd cmd, u32 reg, u32 *val,
> +					       int seq_num, int priority)
> +{
> +	struct mdio_ethhdr *mdio_ethhdr;
> +	struct sk_buff *skb;
> +	u16 hdr;
> +
> +	skb = dev_alloc_skb(QCA_HDR_MDIO_PKG_LEN);

Still no if (!skb) checking... Not only here, but also at the call sites
of this.

> +
> +	skb_reset_mac_header(skb);
> +	skb_set_network_header(skb, skb->len);
> +
> +	mdio_ethhdr = skb_push(skb, QCA_HDR_MDIO_HEADER_LEN + QCA_HDR_LEN);
> +
> +	hdr = FIELD_PREP(QCA_HDR_XMIT_VERSION, QCA_HDR_VERSION);
> +	hdr |= FIELD_PREP(QCA_HDR_XMIT_PRIORITY, priority);
> +	hdr |= QCA_HDR_XMIT_FROM_CPU;
> +	hdr |= FIELD_PREP(QCA_HDR_XMIT_DP_BIT, BIT(0));
> +	hdr |= FIELD_PREP(QCA_HDR_XMIT_CONTROL, QCA_HDR_XMIT_TYPE_RW_REG);
> +
> +	mdio_ethhdr->seq = FIELD_PREP(QCA_HDR_MDIO_SEQ_NUM, seq_num);
> +
> +	mdio_ethhdr->command = FIELD_PREP(QCA_HDR_MDIO_ADDR, reg);
> +	mdio_ethhdr->command |= FIELD_PREP(QCA_HDR_MDIO_LENGTH, 4);
> +	mdio_ethhdr->command |= FIELD_PREP(QCA_HDR_MDIO_CMD, cmd);
> +	mdio_ethhdr->command |= FIELD_PREP(QCA_HDR_MDIO_CHECK_CODE, MDIO_CHECK_CODE_VAL);
> +
> +	if (cmd == MDIO_WRITE)
> +		mdio_ethhdr->mdio_data = *val;
> +
> +	mdio_ethhdr->hdr = htons(hdr);
> +
> +	skb_put_zero(skb, QCA_HDR_MDIO_DATA2_LEN);
> +	skb_put_zero(skb, QCA_HDR_MDIO_PADDING_LEN);

Maybe a single call to skb_put_zero, and pass the sum as argument?

> +
> +	return skb;
> +}
> +
> +static int qca8k_read_eth(struct qca8k_priv *priv, u32 reg, u32 *val)
> +{
> +	struct qca8k_mdio_hdr_data *mdio_hdr_data = &priv->mdio_hdr_data;
> +	struct sk_buff *skb;
> +	bool ack;
> +	int ret;
> +
> +	skb = qca8k_alloc_mdio_header(MDIO_READ, reg, NULL, 200, QCA8K_ETHERNET_MDIO_PRIORITY);

You hardcode "seq" to 200? Aren't you supposed to increment it or
something?

> +	skb->dev = (struct net_device *)priv->master;

You access priv->master outside of priv->mdio_hdr_data.mutex from
qca8k_master_change(), that can't be good.

> +
> +	mutex_lock(&mdio_hdr_data->mutex);
> +
> +	reinit_completion(&mdio_hdr_data->rw_done);
> +	mdio_hdr_data->seq = 200;

Why do you rewrite the seq here?

> +	mdio_hdr_data->ack = false;
> +
> +	dev_queue_xmit(skb);
> +
> +	ret = wait_for_completion_timeout(&mdio_hdr_data->rw_done,
> +					  msecs_to_jiffies(QCA8K_ETHERNET_TIMEOUT));
> +
> +	*val = mdio_hdr_data->data[0];
> +	ack = mdio_hdr_data->ack;
> +
> +	mutex_unlock(&mdio_hdr_data->mutex);
> +
> +	if (ret <= 0)
> +		return -ETIMEDOUT;
> +
> +	if (!ack)
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int qca8k_write_eth(struct qca8k_priv *priv, u32 reg, u32 val)
> +{
> +	struct qca8k_mdio_hdr_data *mdio_hdr_data = &priv->mdio_hdr_data;
> +	struct sk_buff *skb;
> +	bool ack;
> +	int ret;
> +
> +	skb = qca8k_alloc_mdio_header(MDIO_WRITE, reg, &val, 200, QCA8K_ETHERNET_MDIO_PRIORITY);
> +	skb->dev = (struct net_device *)priv->master;
> +
> +	mutex_lock(&mdio_hdr_data->mutex);
> +
> +	reinit_completion(&mdio_hdr_data->rw_done);
> +	mdio_hdr_data->ack = false;
> +	mdio_hdr_data->seq = 200;
> +
> +	dev_queue_xmit(skb);
> +
> +	ret = wait_for_completion_timeout(&mdio_hdr_data->rw_done,
> +					  msecs_to_jiffies(QCA8K_ETHERNET_TIMEOUT));
> +
> +	ack = mdio_hdr_data->ack;
> +
> +	mutex_unlock(&mdio_hdr_data->mutex);
> +
> +	if (ret <= 0)
> +		return -ETIMEDOUT;
> +
> +	if (!ack)
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static int
> +qca8k_regmap_update_bits_eth(struct qca8k_priv *priv, u32 reg, u32 mask, u32 write_val)
> +{
> +	u32 val = 0;
> +	int ret;
> +
> +	ret = qca8k_read_eth(priv, reg, &val);
> +	if (ret)
> +		return ret;
> +
> +	val &= ~mask;
> +	val |= write_val;
> +
> +	return qca8k_write_eth(priv, reg, val);
> +}
> +
>  static int
>  qca8k_regmap_read(void *ctx, uint32_t reg, uint32_t *val)
>  {
> @@ -178,6 +331,9 @@ qca8k_regmap_read(void *ctx, uint32_t reg, uint32_t *val)
>  	u16 r1, r2, page;
>  	int ret;
>  
> +	if (priv->master && !qca8k_read_eth(priv, reg, val))
> +		return 0;
> +
>  	qca8k_split_addr(reg, &r1, &r2, &page);
>  
>  	mutex_lock_nested(&bus->mdio_lock, MDIO_MUTEX_NESTED);
> @@ -201,6 +357,9 @@ qca8k_regmap_write(void *ctx, uint32_t reg, uint32_t val)
>  	u16 r1, r2, page;
>  	int ret;
>  
> +	if (priv->master && !qca8k_write_eth(priv, reg, val))
> +		return 0;
> +
>  	qca8k_split_addr(reg, &r1, &r2, &page);
>  
>  	mutex_lock_nested(&bus->mdio_lock, MDIO_MUTEX_NESTED);
> @@ -225,6 +384,10 @@ qca8k_regmap_update_bits(void *ctx, uint32_t reg, uint32_t mask, uint32_t write_
>  	u32 val;
>  	int ret;
>  
> +	if (priv->master &&
> +	    !qca8k_regmap_update_bits_eth(priv, reg, mask, write_val))
> +		return 0;
> +
>  	qca8k_split_addr(reg, &r1, &r2, &page);
>  
>  	mutex_lock_nested(&bus->mdio_lock, MDIO_MUTEX_NESTED);
> @@ -2394,10 +2557,38 @@ qca8k_master_change(struct dsa_switch *ds, const struct net_device *master,
>  	if (dp->index != 0)
>  		return;
>  
> +	mutex_lock(&priv->mdio_hdr_data.mutex);
> +
>  	if (operational)
>  		priv->master = master;
>  	else
>  		priv->master = NULL;
> +
> +	mutex_unlock(&priv->mdio_hdr_data.mutex);
> +}
> +
> +static int qca8k_connect_tag_protocol(struct dsa_switch *ds,
> +				      enum dsa_tag_protocol proto)
> +{
> +	struct qca8k_priv *qca8k_priv = ds->priv;
> +
> +	switch (proto) {
> +	case DSA_TAG_PROTO_QCA:
> +		struct tag_qca_priv *priv;
> +
> +		priv = ds->tagger_data;
> +
> +		mutex_init(&qca8k_priv->mdio_hdr_data.mutex);
> +		init_completion(&qca8k_priv->mdio_hdr_data.rw_done);

I think having these initializations here dilutes the purpose of this
callback. Could you please move these two lines to qca8k_sw_probe()?

> +
> +		priv->rw_reg_ack_handler = qca8k_rw_reg_ack_handler;
> +
> +		break;
> +	default:
> +		return -EOPNOTSUPP;
> +	}
> +
> +	return 0;
>  }
>  
>  static const struct dsa_switch_ops qca8k_switch_ops = {
> @@ -2436,6 +2627,7 @@ static const struct dsa_switch_ops qca8k_switch_ops = {
>  	.port_lag_join		= qca8k_port_lag_join,
>  	.port_lag_leave		= qca8k_port_lag_leave,
>  	.master_state_change	= qca8k_master_change,
> +	.connect_tag_protocol	= qca8k_connect_tag_protocol,
>  };
>  
>  static int qca8k_read_switch_id(struct qca8k_priv *priv)
> diff --git a/drivers/net/dsa/qca8k.h b/drivers/net/dsa/qca8k.h
> index 6edd6adc3063..dbe8c74c9793 100644
> --- a/drivers/net/dsa/qca8k.h
> +++ b/drivers/net/dsa/qca8k.h
> @@ -11,6 +11,10 @@
>  #include <linux/delay.h>
>  #include <linux/regmap.h>
>  #include <linux/gpio.h>
> +#include <linux/dsa/tag_qca.h>
> +
> +#define QCA8K_ETHERNET_MDIO_PRIORITY			7
> +#define QCA8K_ETHERNET_TIMEOUT				100
>  
>  #define QCA8K_NUM_PORTS					7
>  #define QCA8K_NUM_CPU_PORTS				2
> @@ -328,6 +332,14 @@ enum {
>  	QCA8K_CPU_PORT6,
>  };
>  
> +struct qca8k_mdio_hdr_data {

What do you think about the "qca8k_eth_mgmt_data" name rather than
"mdio_hdr_data"? I don't think this has anything to do with MDIO.

> +	struct completion rw_done;
> +	struct mutex mutex; /* Enforce one mdio read/write at time */
> +	bool ack;
> +	u32 seq;
> +	u32 data[4];
> +};
> +
>  struct qca8k_ports_config {
>  	bool sgmii_rx_clk_falling_edge;
>  	bool sgmii_tx_clk_falling_edge;
> @@ -354,6 +366,7 @@ struct qca8k_priv {
>  	struct gpio_desc *reset_gpio;
>  	unsigned int port_mtu[QCA8K_NUM_PORTS];
>  	const struct net_device *master; /* Track if mdio/mib Ethernet is available */
> +	struct qca8k_mdio_hdr_data mdio_hdr_data;
>  };
>  
>  struct qca8k_mib_desc {
> -- 
> 2.33.1
> 

  reply	other threads:[~2021-12-15  9:49 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-12-14 22:43 [net-next PATCH RFC v6 00/16] Add support for qca8k mdio rw in Ethernet packet Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 01/16] net: dsa: provide switch operations for tracking the master state Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 02/16] net: dsa: stop updating master MTU from master.c Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 03/16] net: dsa: hold rtnl_mutex when calling dsa_master_{setup,teardown} Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 04/16] net: dsa: replay master state events in dsa_tree_{setup,teardown}_master Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 05/16] net: dsa: tag_qca: convert to FIELD macro Ansuel Smith
2021-12-14 22:43 ` [net-next PATCH RFC v6 06/16] net: dsa: tag_qca: move define to include linux/dsa Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 07/16] net: dsa: tag_qca: enable promisc_on_master flag Ansuel Smith
2021-12-15  9:58   ` Vladimir Oltean
2021-12-14 22:44 ` [net-next PATCH RFC v6 08/16] net: dsa: tag_qca: add define for handling mdio Ethernet packet Ansuel Smith
2021-12-15  9:58   ` Vladimir Oltean
2021-12-14 22:44 ` [net-next PATCH RFC v6 09/16] net: dsa: tag_qca: add define for handling MIB packet Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 10/16] net: dsa: tag_qca: add support for handling mdio Ethernet and " Ansuel Smith
2021-12-15  9:54   ` Vladimir Oltean
2021-12-14 22:44 ` [net-next PATCH RFC v6 11/16] net: dsa: qca8k: add tracking state of master port Ansuel Smith
2021-12-15  9:51   ` Vladimir Oltean
2021-12-16  0:00     ` Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 12/16] net: dsa: qca8k: add support for mdio read/write in Ethernet packet Ansuel Smith
2021-12-15  9:49   ` Vladimir Oltean [this message]
2021-12-15 16:53     ` Ansuel Smith
2021-12-15 12:47   ` Vladimir Oltean
2021-12-15 16:56     ` Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 13/16] net: dsa: qca8k: add support for mib autocast " Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 14/16] net: dsa: qca8k: add support for phy read/write with mdio Ethernet Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 15/16] net: dsa: qca8k: move page cache to driver priv Ansuel Smith
2021-12-14 22:44 ` [net-next PATCH RFC v6 16/16] net: dsa: qca8k: cache lo and hi for mdio write Ansuel Smith
2021-12-14 23:34 ` [net-next PATCH RFC v6 00/16] Add support for qca8k mdio rw in Ethernet packet Ansuel Smith
2021-12-15 10:26 ` Vladimir Oltean
2021-12-15 16:34   ` Ansuel Smith
2021-12-16 23:38     ` Vladimir Oltean
2022-01-06 20:56       ` Ansuel Smith
2022-01-07 17:17         ` Vladimir Oltean

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20211215094912.gkqq4pfwac7gqeaa@skbuf \
    --to=olteanv@gmail.com \
    --cc=andrew@lunn.ch \
    --cc=ansuelsmth@gmail.com \
    --cc=davem@davemloft.net \
    --cc=f.fainelli@gmail.com \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=vivien.didelot@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).