Merge branch 'sja1110-dsa-tagging'

Vladimir Oltean says:

====================
DSA tagging driver for NXP SJA1110

This series adds support for tagging data and control packets on the new
NXP SJA1110 switch (supported by the sja1105 driver). Up to this point
it used the sja1105 driver, which allowed it to send data packets, but
not PDUs as those required by STP and PTP.

To accommodate this new tagger which has both a header and a trailer, we
need to refactor the entire DSA tagging scheme, to replace the "overhead"
concept with separate "needed_headroom" and "needed_tailroom" concepts,
so that SJA1110 can declare its need for both.

There is also some consolidation work for the receive path of tag_8021q
and its callers (sja1105 and ocelot-8021q).

Changes in v3:
Rebase in front of the "Port the SJA1105 DSA driver to XPCS" series
which seems to have stalled for now.

Changes in v2:
Export the dsa_8021q_rcv and sja1110_process_meta_tstamp symbols to
avoid build errors as modules.
====================

Signed-off-by: David S. Miller <[email protected]>

Author: davem330

Documentation/networking/dsa/dsa.rst

@@ -93,14 +93,15 @@ A tagging protocol may tag all packets with switch tags of the same length, or
 the tag length might vary (for example packets with PTP timestamps might
 require an extended switch tag, or there might be one tag length on TX and a
 different one on RX). Either way, the tagging protocol driver must populate the
-``struct dsa_device_ops::overhead`` with the length in octets of the longest
-switch frame header. The DSA framework will automatically adjust the MTU of the
-master interface to accomodate for this extra size in order for DSA user ports
-to support the standard MTU (L2 payload length) of 1500 octets. The ``overhead``
-is also used to request from the network stack, on a best-effort basis, the
-allocation of packets with a ``needed_headroom`` or ``needed_tailroom``
-sufficient such that the act of pushing the switch tag on transmission of a
-packet does not cause it to reallocate due to lack of memory.
+``struct dsa_device_ops::needed_headroom`` and/or ``struct dsa_device_ops::needed_tailroom``
+with the length in octets of the longest switch frame header/trailer. The DSA
+framework will automatically adjust the MTU of the master interface to
+accommodate for this extra size in order for DSA user ports to support the
+standard MTU (L2 payload length) of 1500 octets. The ``needed_headroom`` and
+``needed_tailroom`` properties are also used to request from the network stack,
+on a best-effort basis, the allocation of packets with enough extra space such
+that the act of pushing the switch tag on transmission of a packet does not
+cause it to reallocate due to lack of memory.
 
 Even though applications are not expected to parse DSA-specific frame headers,
 the format on the wire of the tagging protocol represents an Application Binary
@@ -169,8 +170,8 @@ The job of this method is to prepare the skb in a way that the switch will
 understand what egress port the packet is for (and not deliver it towards other
 ports). Typically this is fulfilled by pushing a frame header. Checking for
 insufficient size in the skb headroom or tailroom is unnecessary provided that
-the ``overhead`` and ``tail_tag`` properties were filled out properly, because
-DSA ensures there is enough space before calling this method.
+the ``needed_headroom`` and ``needed_tailroom`` properties were filled out
+properly, because DSA ensures there is enough space before calling this method.
 
 The reception of a packet goes through the tagger's ``rcv`` function. The
 passed ``struct sk_buff *skb`` has ``skb->data`` pointing at

drivers/net/dsa/sja1105/sja1105.h

@@ -109,6 +109,8 @@ struct sja1105_info {
 	int num_cbs_shapers;
 	int max_frame_mem;
 	int num_ports;
+	bool multiple_cascade_ports;
+	enum dsa_tag_protocol tag_proto;
 	const struct sja1105_dynamic_table_ops *dyn_ops;
 	const struct sja1105_table_ops *static_ops;
 	const struct sja1105_regs *regs;
@@ -128,6 +130,8 @@ struct sja1105_info {
 			   const unsigned char *addr, u16 vid);
 	void (*ptp_cmd_packing)(u8 *buf, struct sja1105_ptp_cmd *cmd,
 				enum packing_op op);
+	bool (*rxtstamp)(struct dsa_switch *ds, int port, struct sk_buff *skb);
+	void (*txtstamp)(struct dsa_switch *ds, int port, struct sk_buff *skb);
 	int (*clocking_setup)(struct sja1105_private *priv);
 	const char *name;
 	bool supports_mii[SJA1105_MAX_NUM_PORTS];

drivers/net/dsa/sja1105/sja1105_main.c

@@ -654,14 +654,6 @@ static int sja1105_init_general_params(struct sja1105_private *priv)
 		.host_port = priv->ds->num_ports,
 		/* Default to an invalid value */
 		.mirr_port = priv->ds->num_ports,
-		/* Link-local traffic received on casc_port will be forwarded
-		 * to host_port without embedding the source port and device ID
-		 * info in the destination MAC address (presumably because it
-		 * is a cascaded port and a downstream SJA switch already did
-		 * that). Default to an invalid port (to disable the feature)
-		 * and overwrite this if we find any DSA (cascaded) ports.
-		 */
-		.casc_port = priv->ds->num_ports,
 		/* No TTEthernet */
 		.vllupformat = SJA1105_VL_FORMAT_PSFP,
 		.vlmarker = 0,
@@ -673,7 +665,12 @@ static int sja1105_init_general_params(struct sja1105_private *priv)
 		 */
 		.tpid = ETH_P_SJA1105,
 		.tpid2 = ETH_P_SJA1105,
+		/* Enable the TTEthernet engine on SJA1110 */
+		.tte_en = true,
+		/* Set up the EtherType for control packets on SJA1110 */
+		.header_type = ETH_P_SJA1110,
 	};
+	struct sja1105_general_params_entry *general_params;
 	struct dsa_switch *ds = priv->ds;
 	struct sja1105_table *table;
 	int port;
@@ -699,12 +696,26 @@ static int sja1105_init_general_params(struct sja1105_private *priv)
 
 	table->entry_count = table->ops->max_entry_count;
 
+	general_params = table->entries;
+
 	/* This table only has a single entry */
-	((struct sja1105_general_params_entry *)table->entries)[0] =
-				default_general_params;
+	general_params[0] = default_general_params;
 
 	sja1110_select_tdmaconfigidx(priv);
 
+	/* Link-local traffic received on casc_port will be forwarded
+	 * to host_port without embedding the source port and device ID
+	 * info in the destination MAC address, and no RX timestamps will be
+	 * taken either (presumably because it is a cascaded port and a
+	 * downstream SJA switch already did that).
+	 * To disable the feature, we need to do different things depending on
+	 * switch generation. On SJA1105 we need to set an invalid port, while
+	 * on SJA1110 which support multiple cascaded ports, this field is a
+	 * bitmask so it must be left zero.
+	 */
+	if (!priv->info->multiple_cascade_ports)
+		general_params->casc_port = ds->num_ports;
+
 	return 0;
 }
 
@@ -2165,7 +2176,9 @@ static enum dsa_tag_protocol
 sja1105_get_tag_protocol(struct dsa_switch *ds, int port,
 			 enum dsa_tag_protocol mp)
 {
-	return DSA_TAG_PROTO_SJA1105;
+	struct sja1105_private *priv = ds->priv;
+
+	return priv->info->tag_proto;
 }
 
 static int sja1105_find_free_subvlan(u16 *subvlan_map, bool pvid)

drivers/net/dsa/sja1105/sja1105_ptp.c

@@ -79,6 +79,7 @@ static int sja1105_change_rxtstamping(struct sja1105_private *priv,
 		priv->tagger_data.stampable_skb = NULL;
 	}
 	ptp_cancel_worker_sync(ptp_data->clock);
+	skb_queue_purge(&ptp_data->skb_txtstamp_queue);
 	skb_queue_purge(&ptp_data->skb_rxtstamp_queue);
 
 	return sja1105_static_config_reload(priv, SJA1105_RX_HWTSTAMPING);
@@ -397,7 +398,7 @@ static long sja1105_rxtstamp_work(struct ptp_clock_info *ptp)
 
 		*shwt = (struct skb_shared_hwtstamps) {0};
 
-		ts = SJA1105_SKB_CB(skb)->meta_tstamp;
+		ts = SJA1105_SKB_CB(skb)->tstamp;
 		ts = sja1105_tstamp_reconstruct(ds, ticks, ts);
 
 		shwt->hwtstamp = ns_to_ktime(sja1105_ticks_to_ns(ts));
@@ -413,9 +414,7 @@ static long sja1105_rxtstamp_work(struct ptp_clock_info *ptp)
 	return -1;
 }
 
-/* Called from dsa_skb_defer_rx_timestamp */
-bool sja1105_port_rxtstamp(struct dsa_switch *ds, int port,
-			   struct sk_buff *skb, unsigned int type)
+bool sja1105_rxtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb)
 {
 	struct sja1105_private *priv = ds->priv;
 	struct sja1105_ptp_data *ptp_data = &priv->ptp_data;
@@ -431,6 +430,89 @@ bool sja1105_port_rxtstamp(struct dsa_switch *ds, int port,
 	return true;
 }
 
+bool sja1110_rxtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb)
+{
+	struct skb_shared_hwtstamps *shwt = skb_hwtstamps(skb);
+	u64 ts = SJA1105_SKB_CB(skb)->tstamp;
+
+	*shwt = (struct skb_shared_hwtstamps) {0};
+
+	shwt->hwtstamp = ns_to_ktime(sja1105_ticks_to_ns(ts));
+
+	/* Don't defer */
+	return false;
+}
+
+/* Called from dsa_skb_defer_rx_timestamp */
+bool sja1105_port_rxtstamp(struct dsa_switch *ds, int port,
+			   struct sk_buff *skb, unsigned int type)
+{
+	struct sja1105_private *priv = ds->priv;
+
+	return priv->info->rxtstamp(ds, port, skb);
+}
+
+void sja1110_process_meta_tstamp(struct dsa_switch *ds, int port, u8 ts_id,
+				 enum sja1110_meta_tstamp dir, u64 tstamp)
+{
+	struct sja1105_private *priv = ds->priv;
+	struct sja1105_ptp_data *ptp_data = &priv->ptp_data;
+	struct sk_buff *skb, *skb_tmp, *skb_match = NULL;
+	struct skb_shared_hwtstamps shwt = {0};
+
+	/* We don't care about RX timestamps on the CPU port */
+	if (dir == SJA1110_META_TSTAMP_RX)
+		return;
+
+	spin_lock(&ptp_data->skb_txtstamp_queue.lock);
+
+	skb_queue_walk_safe(&ptp_data->skb_txtstamp_queue, skb, skb_tmp) {
+		if (SJA1105_SKB_CB(skb)->ts_id != ts_id)
+			continue;
+
+		__skb_unlink(skb, &ptp_data->skb_txtstamp_queue);
+		skb_match = skb;
+
+		break;
+	}
+
+	spin_unlock(&ptp_data->skb_txtstamp_queue.lock);
+
+	if (WARN_ON(!skb_match))
+		return;
+
+	shwt.hwtstamp = ns_to_ktime(sja1105_ticks_to_ns(tstamp));
+	skb_complete_tx_timestamp(skb_match, &shwt);
+}
+EXPORT_SYMBOL_GPL(sja1110_process_meta_tstamp);
+
+/* In addition to cloning the skb which is done by the common
+ * sja1105_port_txtstamp, we need to generate a timestamp ID and save the
+ * packet to the TX timestamping queue.
+ */
+void sja1110_txtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb)
+{
+	struct sk_buff *clone = SJA1105_SKB_CB(skb)->clone;
+	struct sja1105_private *priv = ds->priv;
+	struct sja1105_ptp_data *ptp_data = &priv->ptp_data;
+	struct sja1105_port *sp = &priv->ports[port];
+	u8 ts_id;
+
+	skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
+
+	spin_lock(&sp->data->meta_lock);
+
+	ts_id = sp->data->ts_id;
+	/* Deal automatically with 8-bit wraparound */
+	sp->data->ts_id++;
+
+	SJA1105_SKB_CB(clone)->ts_id = ts_id;
+
+	spin_unlock(&sp->data->meta_lock);
+
+	skb_queue_tail(&ptp_data->skb_txtstamp_queue, clone);
+}
+
 /* Called from dsa_skb_tx_timestamp. This callback is just to clone
  * the skb and have it available in SJA1105_SKB_CB in the .port_deferred_xmit
  * callback, where we will timestamp it synchronously.
@@ -449,6 +531,9 @@ void sja1105_port_txtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb)
 		return;
 
 	SJA1105_SKB_CB(skb)->clone = clone;
+
+	if (priv->info->txtstamp)
+		priv->info->txtstamp(ds, port, skb);
 }
 
 static int sja1105_ptp_reset(struct dsa_switch *ds)
@@ -865,7 +950,10 @@ int sja1105_ptp_clock_register(struct dsa_switch *ds)
 		.n_per_out	= 1,
 	};
 
+	/* Only used on SJA1105 */
 	skb_queue_head_init(&ptp_data->skb_rxtstamp_queue);
+	/* Only used on SJA1110 */
+	skb_queue_head_init(&ptp_data->skb_txtstamp_queue);
 	spin_lock_init(&tagger_data->meta_lock);
 
 	ptp_data->clock = ptp_clock_register(&ptp_data->caps, ds->dev);
@@ -890,6 +978,7 @@ void sja1105_ptp_clock_unregister(struct dsa_switch *ds)
 
 	del_timer_sync(&ptp_data->extts_timer);
 	ptp_cancel_worker_sync(ptp_data->clock);
+	skb_queue_purge(&ptp_data->skb_txtstamp_queue);
 	skb_queue_purge(&ptp_data->skb_rxtstamp_queue);
 	ptp_clock_unregister(ptp_data->clock);
 	ptp_data->clock = NULL;

drivers/net/dsa/sja1105/sja1105_ptp.h

@@ -75,7 +75,12 @@ struct sja1105_ptp_cmd {
 
 struct sja1105_ptp_data {
 	struct timer_list extts_timer;
+	/* Used only on SJA1105 to reconstruct partial timestamps */
 	struct sk_buff_head skb_rxtstamp_queue;
+	/* Used on SJA1110 where meta frames are generated only for
+	 * 2-step TX timestamps
+	 */
+	struct sk_buff_head skb_txtstamp_queue;
 	struct ptp_clock_info caps;
 	struct ptp_clock *clock;
 	struct sja1105_ptp_cmd cmd;
@@ -122,6 +127,10 @@ int __sja1105_ptp_adjtime(struct dsa_switch *ds, s64 delta);
 int sja1105_ptp_commit(struct dsa_switch *ds, struct sja1105_ptp_cmd *cmd,
 		       sja1105_spi_rw_mode_t rw);
 
+bool sja1105_rxtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb);
+bool sja1110_rxtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb);
+void sja1110_txtstamp(struct dsa_switch *ds, int port, struct sk_buff *skb);
+
 #else
 
 struct sja1105_ptp_cmd;
@@ -184,6 +193,10 @@ static inline int sja1105_ptp_commit(struct dsa_switch *ds,
 
 #define sja1105_hwtstamp_set NULL
 
+#define sja1105_rxtstamp NULL
+#define sja1110_rxtstamp NULL
+#define sja1110_txtstamp NULL
+
 #endif /* IS_ENABLED(CONFIG_NET_DSA_SJA1105_PTP) */
 
 #endif /* _SJA1105_PTP_H */