Merge branch 'Traffic-support-for-dsa_8021q-in-vlan_filtering-1-mode'

Vladimir Oltean says:

====================
Traffic support for dsa_8021q in vlan_filtering=1 mode

This series is an attempt to support as much as possible in terms of
traffic I/O from the network stack with the only dsa_8021q user thus
far, sja1105.

The hardware doesn't support pushing a second VLAN tag to packets that
are already tagged, so our only option is to combine the dsa_8021q with
the user tag into a single tag and decode that on the CPU.

The assumption is that there is a type of use cases for which 7 VLANs
per port are more than sufficient, and that there's another type of use
cases where the full 4096 entries are barely enough. Those use cases are
very different from one another, so I prefer trying to give both the
best experience by creating this best_effort_vlan_filtering knob to
select the mode in which they want to operate in.

v2 was submitted here:
https://patchwork.ozlabs.org/project/netdev/cover/20200511135338.20263-1-olteanv@gmail.com/

v1 was submitted here:
https://patchwork.ozlabs.org/project/netdev/cover/20200510164255.19322-1-olteanv@gmail.com/

Changes in v3:
Patch 01/15:
- Rename again to configure_vlan_while_not_filtering, and add a helper
  function for skipping VLAN configuration.
Patch 03/15:
- Remove sja1105_can_use_vlan_as_tags from driver code.
Patch 06/15:
- Adapt sja1105 driver to the second variable name change.
Patch 08/15:
- Provide an implementation of sja1105_can_use_vlan_as_tags as part of
  the tagger and not as part of the switch driver. So we have to look at
  the skb only, and not at the VLAN awareness state.

Changes in v2:
Patch 01/15:
- Rename variable from vlan_bridge_vtu to configure_vlans_while_disabled.
Patch 03/15:
- Be much more thorough, and make sure that things like virtual links
  and FDB operations still work properly.
Patch 05/15:
- Free the vlan lists on teardown.
- Simplify sja1105_classify_vlan: only look at priv->expect_dsa_8021q.
- Keep vid 1 in the list of dsa_8021q VLANs, to make sure that untagged
  packets transmitted from the stack, like PTP, continue to work in
  VLAN-unaware mode.
Patch 06/15:
- Adapt to vlan_bridge_vtu variable name change.
Patch 11/15:
- In sja1105_best_effort_vlan_filtering_set, get the vlan_filtering
  value of each port instead of just one time for port 0. Normally this
  shouldn't matter, but it avoids issues when port 0 is disabled in
  device tree.
Patch 14/14:
- Only do anything in sja1105_build_subvlans and in
  sja1105_build_crosschip_subvlans when operating in
  SJA1105_VLAN_BEST_EFFORT state. This avoids installing VLAN retagging
  rules in unaware mode, which would cost us a penalty in terms of
  usable frame memory.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Documentation/networking/devlink-params-sja1105.txt

@@ -0,0 +1,27 @@
+best_effort_vlan_filtering
+			[DEVICE, DRIVER-SPECIFIC]
+			Allow plain ETH_P_8021Q headers to be used as DSA tags.
+			Benefits:
+			- Can terminate untagged traffic over switch net
+			  devices even when enslaved to a bridge with
+			  vlan_filtering=1.
+			- Can terminate VLAN-tagged traffic over switch net
+			  devices even when enslaved to a bridge with
+			  vlan_filtering=1, with some constraints (no more than
+			  7 non-pvid VLANs per user port).
+			- Can do QoS based on VLAN PCP and VLAN membership
+			  admission control for autonomously forwarded frames
+			  (regardless of whether they can be terminated on the
+			  CPU or not).
+			Drawbacks:
+			- User cannot use VLANs in range 1024-3071. If the
+			  switch receives frames with such VIDs, it will
+			  misinterpret them as DSA tags.
+			- Switch uses Shared VLAN Learning (FDB lookup uses
+			  only DMAC as key).
+			- When VLANs span cross-chip topologies, the total
+			  number of permitted VLANs may be less than 7 per
+			  port, due to a maximum number of 32 VLAN retagging
+			  rules per switch.
+			Configuration mode: runtime
+			Type: bool.

Documentation/networking/dsa/sja1105.rst

@@ -66,34 +66,193 @@ reprogrammed with the updated static configuration.
 Traffic support
 ===============
 
-The switches do not support switch tagging in hardware. But they do support
-customizing the TPID by which VLAN traffic is identified as such. The switch
-driver is leveraging ``CONFIG_NET_DSA_TAG_8021Q`` by requesting that special
-VLANs (with a custom TPID of ``ETH_P_EDSA`` instead of ``ETH_P_8021Q``) are
-installed on its ports when not in ``vlan_filtering`` mode. This does not
-interfere with the reception and transmission of real 802.1Q-tagged traffic,
-because the switch does no longer parse those packets as VLAN after the TPID
-change.
-The TPID is restored when ``vlan_filtering`` is requested by the user through
-the bridge layer, and general IP termination becomes no longer possible through
-the switch netdevices in this mode.
-
-The switches have two programmable filters for link-local destination MACs.
+The switches do not have hardware support for DSA tags, except for "slow
+protocols" for switch control as STP and PTP. For these, the switches have two
+programmable filters for link-local destination MACs.
 These are used to trap BPDUs and PTP traffic to the master netdevice, and are
 further used to support STP and 1588 ordinary clock/boundary clock
-functionality.
-
-The following traffic modes are supported over the switch netdevices:
-
-+--------------------+------------+------------------+------------------+
-|                    | Standalone | Bridged with     | Bridged with     |
-|                    | ports      | vlan_filtering 0 | vlan_filtering 1 |
-+====================+============+==================+==================+
-| Regular traffic    |     Yes    |       Yes        |  No (use master) |
-+--------------------+------------+------------------+------------------+
-| Management traffic |     Yes    |       Yes        |       Yes        |
-| (BPDU, PTP)        |            |                  |                  |
-+--------------------+------------+------------------+------------------+
+functionality. For frames trapped to the CPU, source port and switch ID
+information is encoded by the hardware into the frames.
+
+But by leveraging ``CONFIG_NET_DSA_TAG_8021Q`` (a software-defined DSA tagging
+format based on VLANs), general-purpose traffic termination through the network
+stack can be supported under certain circumstances.
+
+Depending on VLAN awareness state, the following operating modes are possible
+with the switch:
+
+- Mode 1 (VLAN-unaware): a port is in this mode when it is used as a standalone
+  net device, or when it is enslaved to a bridge with ``vlan_filtering=0``.
+- Mode 2 (fully VLAN-aware): a port is in this mode when it is enslaved to a
+  bridge with ``vlan_filtering=1``. Access to the entire VLAN range is given to
+  the user through ``bridge vlan`` commands, but general-purpose (anything
+  other than STP, PTP etc) traffic termination is not possible through the
+  switch net devices. The other packets can be still by user space processed
+  through the DSA master interface (similar to ``DSA_TAG_PROTO_NONE``).
+- Mode 3 (best-effort VLAN-aware): a port is in this mode when enslaved to a
+  bridge with ``vlan_filtering=1``, and the devlink property of its parent
+  switch named ``best_effort_vlan_filtering`` is set to ``true``. When
+  configured like this, the range of usable VIDs is reduced (0 to 1023 and 3072
+  to 4094), so is the number of usable VIDs (maximum of 7 non-pvid VLANs per
+  port*), and shared VLAN learning is performed (FDB lookup is done only by
+  DMAC, not also by VID).
+
+To summarize, in each mode, the following types of traffic are supported over
+the switch net devices:
+
++-------------+-----------+--------------+------------+
+|             |   Mode 1  |    Mode 2    |   Mode 3   |
++=============+===========+==============+============+
+|   Regular   |    Yes    |      No      |     Yes    |
+|   traffic   |           | (use master) |            |
++-------------+-----------+--------------+------------+
+| Management  |    Yes    |     Yes      |     Yes    |
+|   traffic   |           |              |            |
+| (BPDU, PTP) |           |              |            |
++-------------+-----------+--------------+------------+
+
+To configure the switch to operate in Mode 3, the following steps can be
+followed::
+
+  ip link add dev br0 type bridge
+  # swp2 operates in Mode 1 now
+  ip link set dev swp2 master br0
+  # swp2 temporarily moves to Mode 2
+  ip link set dev br0 type bridge vlan_filtering 1
+  [   61.204770] sja1105 spi0.1: Reset switch and programmed static config. Reason: VLAN filtering
+  [   61.239944] sja1105 spi0.1: Disabled switch tagging
+  # swp3 now operates in Mode 3
+  devlink dev param set spi/spi0.1 name best_effort_vlan_filtering value true cmode runtime
+  [   64.682927] sja1105 spi0.1: Reset switch and programmed static config. Reason: VLAN filtering
+  [   64.711925] sja1105 spi0.1: Enabled switch tagging
+  # Cannot use VLANs in range 1024-3071 while in Mode 3.
+  bridge vlan add dev swp2 vid 1025 untagged pvid
+  RTNETLINK answers: Operation not permitted
+  bridge vlan add dev swp2 vid 100
+  bridge vlan add dev swp2 vid 101 untagged
+  bridge vlan
+  port    vlan ids
+  swp5     1 PVID Egress Untagged
+
+  swp2     1 PVID Egress Untagged
+           100
+           101 Egress Untagged
+
+  swp3     1 PVID Egress Untagged
+
+  swp4     1 PVID Egress Untagged
+
+  br0      1 PVID Egress Untagged
+  bridge vlan add dev swp2 vid 102
+  bridge vlan add dev swp2 vid 103
+  bridge vlan add dev swp2 vid 104
+  bridge vlan add dev swp2 vid 105
+  bridge vlan add dev swp2 vid 106
+  bridge vlan add dev swp2 vid 107
+  # Cannot use mode than 7 VLANs per port while in Mode 3.
+  [ 3885.216832] sja1105 spi0.1: No more free subvlans
+
+\* "maximum of 7 non-pvid VLANs per port": Decoding VLAN-tagged packets on the
+CPU in mode 3 is possible through VLAN retagging of packets that go from the
+switch to the CPU. In cross-chip topologies, the port that goes to the CPU
+might also go to other switches. In that case, those other switches will see
+only a retagged packet (which only has meaning for the CPU). So if they are
+interested in this VLAN, they need to apply retagging in the reverse direction,
+to recover the original value from it. This consumes extra hardware resources
+for this switch. There is a maximum of 32 entries in the Retagging Table of
+each switch device.
+
+As an example, consider this cross-chip topology::
+
+  +-------------------------------------------------+
+  | Host SoC                                        |
+  |           +-------------------------+           |
+  |           | DSA master for embedded |           |
+  |           |   switch (non-sja1105)  |           |
+  |  +--------+-------------------------+--------+  |
+  |  |   embedded L2 switch                      |  |
+  |  |                                           |  |
+  |  |   +--------------+     +--------------+   |  |
+  |  |   |DSA master for|     |DSA master for|   |  |
+  |  |   |  SJA1105 1   |     |  SJA1105 2   |   |  |
+  +--+---+--------------+-----+--------------+---+--+
+
+  +-----------------------+ +-----------------------+
+  |   SJA1105 switch 1    | |   SJA1105 switch 2    |
+  +-----+-----+-----+-----+ +-----+-----+-----+-----+
+  |sw1p0|sw1p1|sw1p2|sw1p3| |sw2p0|sw2p1|sw2p2|sw2p3|
+  +-----+-----+-----+-----+ +-----+-----+-----+-----+
+
+To reach the CPU, SJA1105 switch 1 (spi/spi2.1) uses the same port as is uses
+to reach SJA1105 switch 2 (spi/spi2.2), which would be port 4 (not drawn).
+Similarly for SJA1105 switch 2.
+
+Also consider the following commands, that add VLAN 100 to every sja1105 user
+port::
+
+  devlink dev param set spi/spi2.1 name best_effort_vlan_filtering value true cmode runtime
+  devlink dev param set spi/spi2.2 name best_effort_vlan_filtering value true cmode runtime
+  ip link add dev br0 type bridge
+  for port in sw1p0 sw1p1 sw1p2 sw1p3 \
+              sw2p0 sw2p1 sw2p2 sw2p3; do
+      ip link set dev $port master br0
+  done
+  ip link set dev br0 type bridge vlan_filtering 1
+  for port in sw1p0 sw1p1 sw1p2 sw1p3 \
+              sw2p0 sw2p1 sw2p2; do
+      bridge vlan add dev $port vid 100
+  done
+  ip link add link br0 name br0.100 type vlan id 100 && ip link set dev br0.100 up
+  ip addr add 192.168.100.3/24 dev br0.100
+  bridge vlan add dev br0 vid 100 self
+
+  bridge vlan
+  port    vlan ids
+  sw1p0    1 PVID Egress Untagged
+           100
+
+  sw1p1    1 PVID Egress Untagged
+           100
+
+  sw1p2    1 PVID Egress Untagged
+           100
+
+  sw1p3    1 PVID Egress Untagged
+           100
+
+  sw2p0    1 PVID Egress Untagged
+           100
+
+  sw2p1    1 PVID Egress Untagged
+           100
+
+  sw2p2    1 PVID Egress Untagged
+           100
+
+  sw2p3    1 PVID Egress Untagged
+
+  br0      1 PVID Egress Untagged
+           100
+
+SJA1105 switch 1 consumes 1 retagging entry for each VLAN on each user port
+towards the CPU. It also consumes 1 retagging entry for each non-pvid VLAN that
+it is also interested in, which is configured on any port of any neighbor
+switch.
+
+In this case, SJA1105 switch 1 consumes a total of 11 retagging entries, as
+follows:
+- 8 retagging entries for VLANs 1 and 100 installed on its user ports
+  (``sw1p0`` - ``sw1p3``)
+- 3 retagging entries for VLAN 100 installed on the user ports of SJA1105
+  switch 2 (``sw2p0`` - ``sw2p2``), because it also has ports that are
+  interested in it. The VLAN 1 is a pvid on SJA1105 switch 2 and does not need
+  reverse retagging.
+
+SJA1105 switch 2 also consumes 11 retagging entries, but organized as follows:
+- 7 retagging entries for the bridge VLANs on its user ports (``sw2p0`` -
+  ``sw2p3``).
+- 4 retagging entries for VLAN 100 installed on the user ports of SJA1105
+  switch 1 (``sw1p0`` - ``sw1p3``).
 
 Switching features
 ==================

drivers/net/dsa/sja1105/sja1105.h

@@ -87,6 +87,12 @@ struct sja1105_info {
 	const struct sja1105_dynamic_table_ops *dyn_ops;
 	const struct sja1105_table_ops *static_ops;
 	const struct sja1105_regs *regs;
+	/* Both E/T and P/Q/R/S have quirks when it comes to popping the S-Tag
+	 * from double-tagged frames. E/T will pop it only when it's equal to
+	 * TPID from the General Parameters Table, while P/Q/R/S will only
+	 * pop it when it's equal to TPID2.
+	 */
+	u16 qinq_tpid;
 	int (*reset_cmd)(struct dsa_switch *ds);
 	int (*setup_rgmii_delay)(const void *ctx, int port);
 	/* Prototypes from include/net/dsa.h */
@@ -178,21 +184,40 @@ struct sja1105_flow_block {
 	int num_virtual_links;
 };
 
+struct sja1105_bridge_vlan {
+	struct list_head list;
+	int port;
+	u16 vid;
+	bool pvid;
+	bool untagged;
+};
+
+enum sja1105_vlan_state {
+	SJA1105_VLAN_UNAWARE,
+	SJA1105_VLAN_BEST_EFFORT,
+	SJA1105_VLAN_FILTERING_FULL,
+};
+
 struct sja1105_private {
 	struct sja1105_static_config static_config;
 	bool rgmii_rx_delay[SJA1105_NUM_PORTS];
 	bool rgmii_tx_delay[SJA1105_NUM_PORTS];
+	bool best_effort_vlan_filtering;
 	const struct sja1105_info *info;
 	struct gpio_desc *reset_gpio;
 	struct spi_device *spidev;
 	struct dsa_switch *ds;
+	struct list_head dsa_8021q_vlans;
+	struct list_head bridge_vlans;
 	struct list_head crosschip_links;
 	struct sja1105_flow_block flow_block;
 	struct sja1105_port ports[SJA1105_NUM_PORTS];
 	/* Serializes transmission of management frames so that
 	 * the switch doesn't confuse them with one another.
 	 */
 	struct mutex mgmt_lock;
+	bool expect_dsa_8021q;
+	enum sja1105_vlan_state vlan_state;
 	struct sja1105_tagger_data tagger_data;
 	struct sja1105_ptp_data ptp_data;
 	struct sja1105_tas_data tas_data;
@@ -219,6 +244,8 @@ enum sja1105_reset_reason {
 int sja1105_static_config_reload(struct sja1105_private *priv,
 				 enum sja1105_reset_reason reason);
 
+void sja1105_frame_memory_partitioning(struct sja1105_private *priv);
+
 /* From sja1105_spi.c */
 int sja1105_xfer_buf(const struct sja1105_private *priv,
 		     sja1105_spi_rw_mode_t rw, u64 reg_addr,
@@ -303,6 +330,8 @@ size_t sja1105et_l2_lookup_entry_packing(void *buf, void *entry_ptr,
 					 enum packing_op op);
 size_t sja1105_vlan_lookup_entry_packing(void *buf, void *entry_ptr,
 					 enum packing_op op);
+size_t sja1105_retagging_entry_packing(void *buf, void *entry_ptr,
+				       enum packing_op op);
 size_t sja1105pqrs_mac_config_entry_packing(void *buf, void *entry_ptr,
 					    enum packing_op op);
 size_t sja1105pqrs_avb_params_entry_packing(void *buf, void *entry_ptr,

drivers/net/dsa/sja1105/sja1105_dynamic_config.c

@@ -133,6 +133,9 @@
 #define SJA1105PQRS_SIZE_AVB_PARAMS_DYN_CMD			\
 	(SJA1105_SIZE_DYN_CMD + SJA1105PQRS_SIZE_AVB_PARAMS_ENTRY)
 
+#define SJA1105_SIZE_RETAGGING_DYN_CMD				\
+	(SJA1105_SIZE_DYN_CMD + SJA1105_SIZE_RETAGGING_ENTRY)
+
 #define SJA1105_MAX_DYN_CMD_SIZE				\
 	SJA1105PQRS_SIZE_MAC_CONFIG_DYN_CMD
 
@@ -525,6 +528,20 @@ sja1105pqrs_avb_params_cmd_packing(void *buf, struct sja1105_dyn_cmd *cmd,
 	sja1105_packing(p, &cmd->rdwrset, 29, 29, size, op);
 }
 
+static void
+sja1105_retagging_cmd_packing(void *buf, struct sja1105_dyn_cmd *cmd,
+			      enum packing_op op)
+{
+	u8 *p = buf + SJA1105_SIZE_RETAGGING_ENTRY;
+	const int size = SJA1105_SIZE_DYN_CMD;
+
+	sja1105_packing(p, &cmd->valid,    31, 31, size, op);
+	sja1105_packing(p, &cmd->errors,   30, 30, size, op);
+	sja1105_packing(p, &cmd->valident, 29, 29, size, op);
+	sja1105_packing(p, &cmd->rdwrset,  28, 28, size, op);
+	sja1105_packing(p, &cmd->index,     5,  0, size, op);
+}
+
 #define OP_READ		BIT(0)
 #define OP_WRITE	BIT(1)
 #define OP_DEL		BIT(2)
@@ -606,6 +623,14 @@ struct sja1105_dynamic_table_ops sja1105et_dyn_ops[BLK_IDX_MAX_DYN] = {
 		.packed_size = SJA1105ET_SIZE_GENERAL_PARAMS_DYN_CMD,
 		.addr = 0x34,
 	},
+	[BLK_IDX_RETAGGING] = {
+		.entry_packing = sja1105_retagging_entry_packing,
+		.cmd_packing = sja1105_retagging_cmd_packing,
+		.max_entry_count = SJA1105_MAX_RETAGGING_COUNT,
+		.access = (OP_WRITE | OP_DEL),
+		.packed_size = SJA1105_SIZE_RETAGGING_DYN_CMD,
+		.addr = 0x31,
+	},
 	[BLK_IDX_XMII_PARAMS] = {0},
 };
 
@@ -692,6 +717,14 @@ struct sja1105_dynamic_table_ops sja1105pqrs_dyn_ops[BLK_IDX_MAX_DYN] = {
 		.packed_size = SJA1105ET_SIZE_GENERAL_PARAMS_DYN_CMD,
 		.addr = 0x34,
 	},
+	[BLK_IDX_RETAGGING] = {
+		.entry_packing = sja1105_retagging_entry_packing,
+		.cmd_packing = sja1105_retagging_cmd_packing,
+		.max_entry_count = SJA1105_MAX_RETAGGING_COUNT,
+		.access = (OP_READ | OP_WRITE | OP_DEL),
+		.packed_size = SJA1105_SIZE_RETAGGING_DYN_CMD,
+		.addr = 0x38,
+	},
 	[BLK_IDX_XMII_PARAMS] = {0},
 };