[PATCH net-next v3 00/16] Introduce flowtable hw offloading in airoha_eth driver

[PATCH net-next v3 00/16] Introduce flowtable hw offloading in airoha_eth driver

[Lorenzo Bianconi]

Introduce netfilter flowtable integration in airoha_eth driver to
offload 5-tuple flower rules learned by the PPE module if the user
accelerates them using a nft configuration similar to the one reported
below:

table inet filter {
	flowtable ft {
		hook ingress priority filter
		devices = { lan1, lan2, lan3, lan4, eth1 }
		flags offload;
	}
	chain forward {
		type filter hook forward priority filter; policy accept;
		meta l4proto { tcp, udp } flow add @ft
	}
}

Packet Processor Engine (PPE) module available on EN7581 SoC populates
the PPE table with 5-tuples flower rules learned from traffic forwarded
between the GDM ports connected to the Packet Switch Engine (PSE) module.
airoha_eth driver configures and collects data from the PPE module via a
Network Processor Unit (NPU) RISC-V module available on the EN7581 SoC.
Move airoha_eth driver in a dedicated folder
(drivers/net/ethernet/airoha).

---
Changes in v3:
- Fix TSO support for header cloned skbs
- Do not use skb_pull_rcsum() in airoha_get_dsa_tag()
- Fix head lean computation after running airoha_get_dsa_tag() in
  airoha_dev_xmit()
- Link to v2: https://lore.kernel.org/r/20250207-airoha-en7581-flowtable-offload-v2-0-3a2239692a67@kernel.org

Changes in v2:
- Add airoha-npu document binding
- Enable Rx SPTAG on MT7530 dsa switch for EN7581 SoC.
- Fix warnings in airoha_npu_run_firmware()
- Fix sparse warnings
- Link to v1: https://lore.kernel.org/r/20250205-airoha-en7581-flowtable-offload-v1-0-d362cfa97b01@kernel.org

---
Lorenzo Bianconi (16):
      net: airoha: Fix TSO support for header cloned skbs
      net: airoha: Move airoha_eth driver in a dedicated folder
      net: airoha: Move definitions in airoha_eth.h
      net: airoha: Move reg/write utility routines in airoha_eth.h
      net: airoha: Move register definitions in airoha_regs.h
      net: airoha: Move DSA tag in DMA descriptor
      net: dsa: mt7530: Enable Rx sptag for EN7581 SoC
      net: airoha: Enable support for multiple net_devices
      net: airoha: Move REG_GDM_FWD_CFG() initialization in airoha_dev_init()
      net: airoha: Rename airoha_set_gdm_port_fwd_cfg() in airoha_set_vip_for_gdm_port()
      dt-bindings: arm: airoha: Add the NPU node for EN7581 SoC
      dt-bindings: net: airoha: Add airoha,npu phandle property
      net: airoha: Introduce PPE initialization via NPU
      net: airoha: Introduce flowtable offload support
      net: airoha: Add loopback support for GDM2
      net: airoha: Introduce PPE debugfs support

 .../devicetree/bindings/arm/airoha,en7581-npu.yaml |   71 ++
 .../devicetree/bindings/net/airoha,en7581-eth.yaml |   10 +
 drivers/net/dsa/mt7530.c                           |    5 +
 drivers/net/dsa/mt7530.h                           |    4 +
 drivers/net/ethernet/Kconfig                       |    2 +
 drivers/net/ethernet/Makefile                      |    1 +
 drivers/net/ethernet/airoha/Kconfig                |   23 +
 drivers/net/ethernet/airoha/Makefile               |    9 +
 .../net/ethernet/{mediatek => airoha}/airoha_eth.c | 1284 +++++---------------
 drivers/net/ethernet/airoha/airoha_eth.h           |  626 ++++++++++
 drivers/net/ethernet/airoha/airoha_npu.c           |  501 ++++++++
 drivers/net/ethernet/airoha/airoha_ppe.c           |  834 +++++++++++++
 drivers/net/ethernet/airoha/airoha_ppe_debugfs.c   |  175 +++
 drivers/net/ethernet/airoha/airoha_regs.h          |  798 ++++++++++++
 drivers/net/ethernet/mediatek/Kconfig              |    8 -
 drivers/net/ethernet/mediatek/Makefile             |    1 -
 16 files changed, 3344 insertions(+), 1008 deletions(-)
---
base-commit: acdefab0dcbc3833b5a734ab80d792bb778517a0
change-id: 20250205-airoha-en7581-flowtable-offload-e3a11b3b34ad

Best regards,
-- 
Lorenzo Bianconi <lorenzo@kernel.org>

[PATCH net-next v3 01/16] net: airoha: Fix TSO support for header cloned skbs

[Lorenzo Bianconi]

For GSO packets, skb_cow_head() will reallocate the skb for TSO header
cloned skbs in airoha_dev_xmit(). For this reason, sinfo pointer can be
no more valid. Fix the issue relying on skb_shinfo() macro directly in
airoha_dev_xmit().
This is not a user visible issue since we can't currently enable TSO for
DSA user ports since we are missing to initialize net_device
vlan_features field.

Fixes: 23020f049327 ("net: airoha: Introduce ethernet support for EN7581 SoC")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/mediatek/airoha_eth.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/mediatek/airoha_eth.c b/drivers/net/ethernet/mediatek/airoha_eth.c
index 09f448f291240257c5748725848ede231c502fbd..aa5f220ddbcf9ca5bee1173114294cb3aec701c9 100644
--- a/drivers/net/ethernet/mediatek/airoha_eth.c
+++ b/drivers/net/ethernet/mediatek/airoha_eth.c
@@ -2556,11 +2556,10 @@ static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb,
 static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 				   struct net_device *dev)
 {
-	struct skb_shared_info *sinfo = skb_shinfo(skb);
 	struct airoha_gdm_port *port = netdev_priv(dev);
+	u32 nr_frags = 1 + skb_shinfo(skb)->nr_frags;
 	u32 msg0, msg1, len = skb_headlen(skb);
 	struct airoha_qdma *qdma = port->qdma;
-	u32 nr_frags = 1 + sinfo->nr_frags;
 	struct netdev_queue *txq;
 	struct airoha_queue *q;
 	void *data = skb->data;
@@ -2583,8 +2582,9 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 		if (skb_cow_head(skb, 0))
 			goto error;
 
-		if (sinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)) {
-			__be16 csum = cpu_to_be16(sinfo->gso_size);
+		if (skb_shinfo(skb)->gso_type & (SKB_GSO_TCPV4 |
+						 SKB_GSO_TCPV6)) {
+			__be16 csum = cpu_to_be16(skb_shinfo(skb)->gso_size);
 
 			tcp_hdr(skb)->check = (__force __sum16)csum;
 			msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TSO_MASK, 1);
@@ -2613,7 +2613,7 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 	for (i = 0; i < nr_frags; i++) {
 		struct airoha_qdma_desc *desc = &q->desc[index];
 		struct airoha_queue_entry *e = &q->entry[index];
-		skb_frag_t *frag = &sinfo->frags[i];
+		skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
 		dma_addr_t addr;
 		u32 val;
 

-- 
2.48.1

[PATCH net-next v3 02/16] net: airoha: Move airoha_eth driver in a dedicated folder

[Lorenzo Bianconi]

The airoha_eth driver has no codebase shared with mtk_eth_soc one.
Moreover, the upcoming features (flowtable hw offloading, PCS, ..) will
not reuse any code from MediaTek driver. Move the Airoha driver in a
dedicated folder.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/Kconfig                           |  2 ++
 drivers/net/ethernet/Makefile                          |  1 +
 drivers/net/ethernet/airoha/Kconfig                    | 18 ++++++++++++++++++
 drivers/net/ethernet/airoha/Makefile                   |  6 ++++++
 drivers/net/ethernet/{mediatek => airoha}/airoha_eth.c |  0
 drivers/net/ethernet/mediatek/Kconfig                  |  8 --------
 drivers/net/ethernet/mediatek/Makefile                 |  1 -
 7 files changed, 27 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/Kconfig b/drivers/net/ethernet/Kconfig
index 977b42bc1e8c1e8804eb7fafa9ed85252d956cad..7941983d21e9e84cbd78241d5c1d48c95e50a8e4 100644
--- a/drivers/net/ethernet/Kconfig
+++ b/drivers/net/ethernet/Kconfig
@@ -20,6 +20,8 @@ source "drivers/net/ethernet/actions/Kconfig"
 source "drivers/net/ethernet/adaptec/Kconfig"
 source "drivers/net/ethernet/aeroflex/Kconfig"
 source "drivers/net/ethernet/agere/Kconfig"
+source "drivers/net/ethernet/airoha/Kconfig"
+source "drivers/net/ethernet/mellanox/Kconfig"
 source "drivers/net/ethernet/alacritech/Kconfig"
 source "drivers/net/ethernet/allwinner/Kconfig"
 source "drivers/net/ethernet/alteon/Kconfig"
diff --git a/drivers/net/ethernet/Makefile b/drivers/net/ethernet/Makefile
index 99fa180dedb80555e64b0fbcd7767044262cf432..67182339469a0d8337cc4e92aa51e498c615156d 100644
--- a/drivers/net/ethernet/Makefile
+++ b/drivers/net/ethernet/Makefile
@@ -10,6 +10,7 @@ obj-$(CONFIG_NET_VENDOR_ADAPTEC) += adaptec/
 obj-$(CONFIG_GRETH) += aeroflex/
 obj-$(CONFIG_NET_VENDOR_ADI) += adi/
 obj-$(CONFIG_NET_VENDOR_AGERE) += agere/
+obj-$(CONFIG_NET_VENDOR_AIROHA) += airoha/
 obj-$(CONFIG_NET_VENDOR_ALACRITECH) += alacritech/
 obj-$(CONFIG_NET_VENDOR_ALLWINNER) += allwinner/
 obj-$(CONFIG_NET_VENDOR_ALTEON) += alteon/
diff --git a/drivers/net/ethernet/airoha/Kconfig b/drivers/net/ethernet/airoha/Kconfig
new file mode 100644
index 0000000000000000000000000000000000000000..b6a131845f13b23a12464cfc281e3abe5699389f
--- /dev/null
+++ b/drivers/net/ethernet/airoha/Kconfig
@@ -0,0 +1,18 @@
+# SPDX-License-Identifier: GPL-2.0-only
+config NET_VENDOR_AIROHA
+	bool "Airoha devices"
+	depends on ARCH_AIROHA || COMPILE_TEST
+	help
+	  If you have a Airoha SoC with ethernet, say Y.
+
+if NET_VENDOR_AIROHA
+
+config NET_AIROHA
+	tristate "Airoha SoC Gigabit Ethernet support"
+	depends on NET_DSA || !NET_DSA
+	select PAGE_POOL
+	help
+	  This driver supports the gigabit ethernet MACs in the
+	  Airoha SoC family.
+
+endif #NET_VENDOR_AIROHA
diff --git a/drivers/net/ethernet/airoha/Makefile b/drivers/net/ethernet/airoha/Makefile
new file mode 100644
index 0000000000000000000000000000000000000000..73a6f3680a4c4ce92ee785d83b905d76a63421df
--- /dev/null
+++ b/drivers/net/ethernet/airoha/Makefile
@@ -0,0 +1,6 @@
+# SPDX-License-Identifier: GPL-2.0-only
+#
+# Airoha for the Mediatek SoCs built-in ethernet macs
+#
+
+obj-$(CONFIG_NET_AIROHA) += airoha_eth.o
diff --git a/drivers/net/ethernet/mediatek/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
similarity index 100%
rename from drivers/net/ethernet/mediatek/airoha_eth.c
rename to drivers/net/ethernet/airoha/airoha_eth.c
diff --git a/drivers/net/ethernet/mediatek/Kconfig b/drivers/net/ethernet/mediatek/Kconfig
index 95c4405b7d7bee53b964243480a0c173b555da56..7bfd3f230ff50739b3fc6103cd5d0e57ab8f70e1 100644
--- a/drivers/net/ethernet/mediatek/Kconfig
+++ b/drivers/net/ethernet/mediatek/Kconfig
@@ -7,14 +7,6 @@ config NET_VENDOR_MEDIATEK
 
 if NET_VENDOR_MEDIATEK
 
-config NET_AIROHA
-	tristate "Airoha SoC Gigabit Ethernet support"
-	depends on NET_DSA || !NET_DSA
-	select PAGE_POOL
-	help
-	  This driver supports the gigabit ethernet MACs in the
-	  Airoha SoC family.
-
 config NET_MEDIATEK_SOC_WED
 	depends on ARCH_MEDIATEK || COMPILE_TEST
 	def_bool NET_MEDIATEK_SOC != n
diff --git a/drivers/net/ethernet/mediatek/Makefile b/drivers/net/ethernet/mediatek/Makefile
index ddbb7f4a516caccf5eef7140de1872e9b35e3471..03e008fbc859b35067682f8640dab05ccce6caf7 100644
--- a/drivers/net/ethernet/mediatek/Makefile
+++ b/drivers/net/ethernet/mediatek/Makefile
@@ -11,4 +11,3 @@ mtk_eth-$(CONFIG_NET_MEDIATEK_SOC_WED) += mtk_wed_debugfs.o
 endif
 obj-$(CONFIG_NET_MEDIATEK_SOC_WED) += mtk_wed_ops.o
 obj-$(CONFIG_NET_MEDIATEK_STAR_EMAC) += mtk_star_emac.o
-obj-$(CONFIG_NET_AIROHA) += airoha_eth.o

-- 
2.48.1

[PATCH net-next v3 03/16] net: airoha: Move definitions in airoha_eth.h

[Lorenzo Bianconi]

Move common airoha_eth definitions in airoha_eth.h in order to reuse
them for Packet Processor Engine (PPE) codebase.
PPE module is used to enable support for flowtable hw offloading in
airoha_eth driver.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 240 +----------------------------
 drivers/net/ethernet/airoha/airoha_eth.h | 251 +++++++++++++++++++++++++++++++
 2 files changed, 252 insertions(+), 239 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index aa5f220ddbcf9ca5bee1173114294cb3aec701c9..0048a5665576afaf532778f0bd96be8b07d29703 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -3,14 +3,9 @@
  * Copyright (c) 2024 AIROHA Inc
  * Author: Lorenzo Bianconi <lorenzo@kernel.org>
  */
-#include <linux/etherdevice.h>
-#include <linux/iopoll.h>
-#include <linux/kernel.h>
-#include <linux/netdevice.h>
 #include <linux/of.h>
 #include <linux/of_net.h>
 #include <linux/platform_device.h>
-#include <linux/reset.h>
 #include <linux/tcp.h>
 #include <linux/u64_stats_sync.h>
 #include <net/dsa.h>
@@ -18,35 +13,7 @@
 #include <net/pkt_cls.h>
 #include <uapi/linux/ppp_defs.h>
 
-#define AIROHA_MAX_NUM_GDM_PORTS	1
-#define AIROHA_MAX_NUM_QDMA		2
-#define AIROHA_MAX_NUM_RSTS		3
-#define AIROHA_MAX_NUM_XSI_RSTS		5
-#define AIROHA_MAX_MTU			2000
-#define AIROHA_MAX_PACKET_SIZE		2048
-#define AIROHA_NUM_QOS_CHANNELS		4
-#define AIROHA_NUM_QOS_QUEUES		8
-#define AIROHA_NUM_TX_RING		32
-#define AIROHA_NUM_RX_RING		32
-#define AIROHA_NUM_NETDEV_TX_RINGS	(AIROHA_NUM_TX_RING + \
-					 AIROHA_NUM_QOS_CHANNELS)
-#define AIROHA_FE_MC_MAX_VLAN_TABLE	64
-#define AIROHA_FE_MC_MAX_VLAN_PORT	16
-#define AIROHA_NUM_TX_IRQ		2
-#define HW_DSCP_NUM			2048
-#define IRQ_QUEUE_LEN(_n)		((_n) ? 1024 : 2048)
-#define TX_DSCP_NUM			1024
-#define RX_DSCP_NUM(_n)			\
-	((_n) ==  2 ? 128 :		\
-	 (_n) == 11 ? 128 :		\
-	 (_n) == 15 ? 128 :		\
-	 (_n) ==  0 ? 1024 : 16)
-
-#define PSE_RSV_PAGES			128
-#define PSE_QUEUE_RSV_PAGES		64
-
-#define QDMA_METER_IDX(_n)		((_n) & 0xff)
-#define QDMA_METER_GROUP(_n)		(((_n) >> 8) & 0x3)
+#include "airoha_eth.h"
 
 /* FE */
 #define PSE_BASE			0x0100
@@ -706,211 +673,6 @@ struct airoha_qdma_fwd_desc {
 	__le32 rsv1;
 };
 
-enum {
-	QDMA_INT_REG_IDX0,
-	QDMA_INT_REG_IDX1,
-	QDMA_INT_REG_IDX2,
-	QDMA_INT_REG_IDX3,
-	QDMA_INT_REG_IDX4,
-	QDMA_INT_REG_MAX
-};
-
-enum {
-	XSI_PCIE0_PORT,
-	XSI_PCIE1_PORT,
-	XSI_USB_PORT,
-	XSI_AE_PORT,
-	XSI_ETH_PORT,
-};
-
-enum {
-	XSI_PCIE0_VIP_PORT_MASK	= BIT(22),
-	XSI_PCIE1_VIP_PORT_MASK	= BIT(23),
-	XSI_USB_VIP_PORT_MASK	= BIT(25),
-	XSI_ETH_VIP_PORT_MASK	= BIT(24),
-};
-
-enum {
-	DEV_STATE_INITIALIZED,
-};
-
-enum {
-	CDM_CRSN_QSEL_Q1 = 1,
-	CDM_CRSN_QSEL_Q5 = 5,
-	CDM_CRSN_QSEL_Q6 = 6,
-	CDM_CRSN_QSEL_Q15 = 15,
-};
-
-enum {
-	CRSN_08 = 0x8,
-	CRSN_21 = 0x15, /* KA */
-	CRSN_22 = 0x16, /* hit bind and force route to CPU */
-	CRSN_24 = 0x18,
-	CRSN_25 = 0x19,
-};
-
-enum {
-	FE_PSE_PORT_CDM1,
-	FE_PSE_PORT_GDM1,
-	FE_PSE_PORT_GDM2,
-	FE_PSE_PORT_GDM3,
-	FE_PSE_PORT_PPE1,
-	FE_PSE_PORT_CDM2,
-	FE_PSE_PORT_CDM3,
-	FE_PSE_PORT_CDM4,
-	FE_PSE_PORT_PPE2,
-	FE_PSE_PORT_GDM4,
-	FE_PSE_PORT_CDM5,
-	FE_PSE_PORT_DROP = 0xf,
-};
-
-enum tx_sched_mode {
-	TC_SCH_WRR8,
-	TC_SCH_SP,
-	TC_SCH_WRR7,
-	TC_SCH_WRR6,
-	TC_SCH_WRR5,
-	TC_SCH_WRR4,
-	TC_SCH_WRR3,
-	TC_SCH_WRR2,
-};
-
-enum trtcm_param_type {
-	TRTCM_MISC_MODE, /* meter_en, pps_mode, tick_sel */
-	TRTCM_TOKEN_RATE_MODE,
-	TRTCM_BUCKETSIZE_SHIFT_MODE,
-	TRTCM_BUCKET_COUNTER_MODE,
-};
-
-enum trtcm_mode_type {
-	TRTCM_COMMIT_MODE,
-	TRTCM_PEAK_MODE,
-};
-
-enum trtcm_param {
-	TRTCM_TICK_SEL = BIT(0),
-	TRTCM_PKT_MODE = BIT(1),
-	TRTCM_METER_MODE = BIT(2),
-};
-
-#define MIN_TOKEN_SIZE				4096
-#define MAX_TOKEN_SIZE_OFFSET			17
-#define TRTCM_TOKEN_RATE_MASK			GENMASK(23, 6)
-#define TRTCM_TOKEN_RATE_FRACTION_MASK		GENMASK(5, 0)
-
-struct airoha_queue_entry {
-	union {
-		void *buf;
-		struct sk_buff *skb;
-	};
-	dma_addr_t dma_addr;
-	u16 dma_len;
-};
-
-struct airoha_queue {
-	struct airoha_qdma *qdma;
-
-	/* protect concurrent queue accesses */
-	spinlock_t lock;
-	struct airoha_queue_entry *entry;
-	struct airoha_qdma_desc *desc;
-	u16 head;
-	u16 tail;
-
-	int queued;
-	int ndesc;
-	int free_thr;
-	int buf_size;
-
-	struct napi_struct napi;
-	struct page_pool *page_pool;
-};
-
-struct airoha_tx_irq_queue {
-	struct airoha_qdma *qdma;
-
-	struct napi_struct napi;
-
-	int size;
-	u32 *q;
-};
-
-struct airoha_hw_stats {
-	/* protect concurrent hw_stats accesses */
-	spinlock_t lock;
-	struct u64_stats_sync syncp;
-
-	/* get_stats64 */
-	u64 rx_ok_pkts;
-	u64 tx_ok_pkts;
-	u64 rx_ok_bytes;
-	u64 tx_ok_bytes;
-	u64 rx_multicast;
-	u64 rx_errors;
-	u64 rx_drops;
-	u64 tx_drops;
-	u64 rx_crc_error;
-	u64 rx_over_errors;
-	/* ethtool stats */
-	u64 tx_broadcast;
-	u64 tx_multicast;
-	u64 tx_len[7];
-	u64 rx_broadcast;
-	u64 rx_fragment;
-	u64 rx_jabber;
-	u64 rx_len[7];
-};
-
-struct airoha_qdma {
-	struct airoha_eth *eth;
-	void __iomem *regs;
-
-	/* protect concurrent irqmask accesses */
-	spinlock_t irq_lock;
-	u32 irqmask[QDMA_INT_REG_MAX];
-	int irq;
-
-	struct airoha_tx_irq_queue q_tx_irq[AIROHA_NUM_TX_IRQ];
-
-	struct airoha_queue q_tx[AIROHA_NUM_TX_RING];
-	struct airoha_queue q_rx[AIROHA_NUM_RX_RING];
-
-	/* descriptor and packet buffers for qdma hw forward */
-	struct {
-		void *desc;
-		void *q;
-	} hfwd;
-};
-
-struct airoha_gdm_port {
-	struct airoha_qdma *qdma;
-	struct net_device *dev;
-	int id;
-
-	struct airoha_hw_stats stats;
-
-	DECLARE_BITMAP(qos_sq_bmap, AIROHA_NUM_QOS_CHANNELS);
-
-	/* qos stats counters */
-	u64 cpu_tx_packets;
-	u64 fwd_tx_packets;
-};
-
-struct airoha_eth {
-	struct device *dev;
-
-	unsigned long state;
-	void __iomem *fe_regs;
-
-	struct reset_control_bulk_data rsts[AIROHA_MAX_NUM_RSTS];
-	struct reset_control_bulk_data xsi_rsts[AIROHA_MAX_NUM_XSI_RSTS];
-
-	struct net_device *napi_dev;
-
-	struct airoha_qdma qdma[AIROHA_MAX_NUM_QDMA];
-	struct airoha_gdm_port *ports[AIROHA_MAX_NUM_GDM_PORTS];
-};
-
 static u32 airoha_rr(void __iomem *base, u32 offset)
 {
 	return readl(base + offset);
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
new file mode 100644
index 0000000000000000000000000000000000000000..3310e0cf85f1d240e95404a0c15703e5f6be26bc
--- /dev/null
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -0,0 +1,251 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2024 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#ifndef AIROHA_ETH_H
+#define AIROHA_ETH_H
+
+#include <linux/etherdevice.h>
+#include <linux/iopoll.h>
+#include <linux/kernel.h>
+#include <linux/netdevice.h>
+#include <linux/reset.h>
+
+#define AIROHA_MAX_NUM_GDM_PORTS	1
+#define AIROHA_MAX_NUM_QDMA		2
+#define AIROHA_MAX_NUM_RSTS		3
+#define AIROHA_MAX_NUM_XSI_RSTS		5
+#define AIROHA_MAX_MTU			2000
+#define AIROHA_MAX_PACKET_SIZE		2048
+#define AIROHA_NUM_QOS_CHANNELS		4
+#define AIROHA_NUM_QOS_QUEUES		8
+#define AIROHA_NUM_TX_RING		32
+#define AIROHA_NUM_RX_RING		32
+#define AIROHA_NUM_NETDEV_TX_RINGS	(AIROHA_NUM_TX_RING + \
+					 AIROHA_NUM_QOS_CHANNELS)
+#define AIROHA_FE_MC_MAX_VLAN_TABLE	64
+#define AIROHA_FE_MC_MAX_VLAN_PORT	16
+#define AIROHA_NUM_TX_IRQ		2
+#define HW_DSCP_NUM			2048
+#define IRQ_QUEUE_LEN(_n)		((_n) ? 1024 : 2048)
+#define TX_DSCP_NUM			1024
+#define RX_DSCP_NUM(_n)			\
+	((_n) ==  2 ? 128 :		\
+	 (_n) == 11 ? 128 :		\
+	 (_n) == 15 ? 128 :		\
+	 (_n) ==  0 ? 1024 : 16)
+
+#define PSE_RSV_PAGES			128
+#define PSE_QUEUE_RSV_PAGES		64
+
+#define QDMA_METER_IDX(_n)		((_n) & 0xff)
+#define QDMA_METER_GROUP(_n)		(((_n) >> 8) & 0x3)
+
+enum {
+	QDMA_INT_REG_IDX0,
+	QDMA_INT_REG_IDX1,
+	QDMA_INT_REG_IDX2,
+	QDMA_INT_REG_IDX3,
+	QDMA_INT_REG_IDX4,
+	QDMA_INT_REG_MAX
+};
+
+enum {
+	XSI_PCIE0_PORT,
+	XSI_PCIE1_PORT,
+	XSI_USB_PORT,
+	XSI_AE_PORT,
+	XSI_ETH_PORT,
+};
+
+enum {
+	XSI_PCIE0_VIP_PORT_MASK	= BIT(22),
+	XSI_PCIE1_VIP_PORT_MASK	= BIT(23),
+	XSI_USB_VIP_PORT_MASK	= BIT(25),
+	XSI_ETH_VIP_PORT_MASK	= BIT(24),
+};
+
+enum {
+	DEV_STATE_INITIALIZED,
+};
+
+enum {
+	CDM_CRSN_QSEL_Q1 = 1,
+	CDM_CRSN_QSEL_Q5 = 5,
+	CDM_CRSN_QSEL_Q6 = 6,
+	CDM_CRSN_QSEL_Q15 = 15,
+};
+
+enum {
+	CRSN_08 = 0x8,
+	CRSN_21 = 0x15, /* KA */
+	CRSN_22 = 0x16, /* hit bind and force route to CPU */
+	CRSN_24 = 0x18,
+	CRSN_25 = 0x19,
+};
+
+enum {
+	FE_PSE_PORT_CDM1,
+	FE_PSE_PORT_GDM1,
+	FE_PSE_PORT_GDM2,
+	FE_PSE_PORT_GDM3,
+	FE_PSE_PORT_PPE1,
+	FE_PSE_PORT_CDM2,
+	FE_PSE_PORT_CDM3,
+	FE_PSE_PORT_CDM4,
+	FE_PSE_PORT_PPE2,
+	FE_PSE_PORT_GDM4,
+	FE_PSE_PORT_CDM5,
+	FE_PSE_PORT_DROP = 0xf,
+};
+
+enum tx_sched_mode {
+	TC_SCH_WRR8,
+	TC_SCH_SP,
+	TC_SCH_WRR7,
+	TC_SCH_WRR6,
+	TC_SCH_WRR5,
+	TC_SCH_WRR4,
+	TC_SCH_WRR3,
+	TC_SCH_WRR2,
+};
+
+enum trtcm_param_type {
+	TRTCM_MISC_MODE, /* meter_en, pps_mode, tick_sel */
+	TRTCM_TOKEN_RATE_MODE,
+	TRTCM_BUCKETSIZE_SHIFT_MODE,
+	TRTCM_BUCKET_COUNTER_MODE,
+};
+
+enum trtcm_mode_type {
+	TRTCM_COMMIT_MODE,
+	TRTCM_PEAK_MODE,
+};
+
+enum trtcm_param {
+	TRTCM_TICK_SEL = BIT(0),
+	TRTCM_PKT_MODE = BIT(1),
+	TRTCM_METER_MODE = BIT(2),
+};
+
+#define MIN_TOKEN_SIZE				4096
+#define MAX_TOKEN_SIZE_OFFSET			17
+#define TRTCM_TOKEN_RATE_MASK			GENMASK(23, 6)
+#define TRTCM_TOKEN_RATE_FRACTION_MASK		GENMASK(5, 0)
+
+struct airoha_queue_entry {
+	union {
+		void *buf;
+		struct sk_buff *skb;
+	};
+	dma_addr_t dma_addr;
+	u16 dma_len;
+};
+
+struct airoha_queue {
+	struct airoha_qdma *qdma;
+
+	/* protect concurrent queue accesses */
+	spinlock_t lock;
+	struct airoha_queue_entry *entry;
+	struct airoha_qdma_desc *desc;
+	u16 head;
+	u16 tail;
+
+	int queued;
+	int ndesc;
+	int free_thr;
+	int buf_size;
+
+	struct napi_struct napi;
+	struct page_pool *page_pool;
+};
+
+struct airoha_tx_irq_queue {
+	struct airoha_qdma *qdma;
+
+	struct napi_struct napi;
+
+	int size;
+	u32 *q;
+};
+
+struct airoha_hw_stats {
+	/* protect concurrent hw_stats accesses */
+	spinlock_t lock;
+	struct u64_stats_sync syncp;
+
+	/* get_stats64 */
+	u64 rx_ok_pkts;
+	u64 tx_ok_pkts;
+	u64 rx_ok_bytes;
+	u64 tx_ok_bytes;
+	u64 rx_multicast;
+	u64 rx_errors;
+	u64 rx_drops;
+	u64 tx_drops;
+	u64 rx_crc_error;
+	u64 rx_over_errors;
+	/* ethtool stats */
+	u64 tx_broadcast;
+	u64 tx_multicast;
+	u64 tx_len[7];
+	u64 rx_broadcast;
+	u64 rx_fragment;
+	u64 rx_jabber;
+	u64 rx_len[7];
+};
+
+struct airoha_qdma {
+	struct airoha_eth *eth;
+	void __iomem *regs;
+
+	/* protect concurrent irqmask accesses */
+	spinlock_t irq_lock;
+	u32 irqmask[QDMA_INT_REG_MAX];
+	int irq;
+
+	struct airoha_tx_irq_queue q_tx_irq[AIROHA_NUM_TX_IRQ];
+
+	struct airoha_queue q_tx[AIROHA_NUM_TX_RING];
+	struct airoha_queue q_rx[AIROHA_NUM_RX_RING];
+
+	/* descriptor and packet buffers for qdma hw forward */
+	struct {
+		void *desc;
+		void *q;
+	} hfwd;
+};
+
+struct airoha_gdm_port {
+	struct airoha_qdma *qdma;
+	struct net_device *dev;
+	int id;
+
+	struct airoha_hw_stats stats;
+
+	DECLARE_BITMAP(qos_sq_bmap, AIROHA_NUM_QOS_CHANNELS);
+
+	/* qos stats counters */
+	u64 cpu_tx_packets;
+	u64 fwd_tx_packets;
+};
+
+struct airoha_eth {
+	struct device *dev;
+
+	unsigned long state;
+	void __iomem *fe_regs;
+
+	struct reset_control_bulk_data rsts[AIROHA_MAX_NUM_RSTS];
+	struct reset_control_bulk_data xsi_rsts[AIROHA_MAX_NUM_XSI_RSTS];
+
+	struct net_device *napi_dev;
+
+	struct airoha_qdma qdma[AIROHA_MAX_NUM_QDMA];
+	struct airoha_gdm_port *ports[AIROHA_MAX_NUM_GDM_PORTS];
+};
+
+#endif /* AIROHA_ETH_H */

-- 
2.48.1

[PATCH net-next v3 04/16] net: airoha: Move reg/write utility routines in airoha_eth.h

[Lorenzo Bianconi]

This is a preliminary patch to introduce flowtable hw offloading
support for airoha_eth driver.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 28 +++-------------------------
 drivers/net/ethernet/airoha/airoha_eth.h | 26 ++++++++++++++++++++++++++
 2 files changed, 29 insertions(+), 25 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 0048a5665576afaf532778f0bd96be8b07d29703..1c6fb7b9ccbbaec846643343e0347a1e948a575f 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -673,17 +673,17 @@ struct airoha_qdma_fwd_desc {
 	__le32 rsv1;
 };
 
-static u32 airoha_rr(void __iomem *base, u32 offset)
+u32 airoha_rr(void __iomem *base, u32 offset)
 {
 	return readl(base + offset);
 }
 
-static void airoha_wr(void __iomem *base, u32 offset, u32 val)
+void airoha_wr(void __iomem *base, u32 offset, u32 val)
 {
 	writel(val, base + offset);
 }
 
-static u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val)
+u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val)
 {
 	val |= (airoha_rr(base, offset) & ~mask);
 	airoha_wr(base, offset, val);
@@ -691,28 +691,6 @@ static u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val)
 	return val;
 }
 
-#define airoha_fe_rr(eth, offset)				\
-	airoha_rr((eth)->fe_regs, (offset))
-#define airoha_fe_wr(eth, offset, val)				\
-	airoha_wr((eth)->fe_regs, (offset), (val))
-#define airoha_fe_rmw(eth, offset, mask, val)			\
-	airoha_rmw((eth)->fe_regs, (offset), (mask), (val))
-#define airoha_fe_set(eth, offset, val)				\
-	airoha_rmw((eth)->fe_regs, (offset), 0, (val))
-#define airoha_fe_clear(eth, offset, val)			\
-	airoha_rmw((eth)->fe_regs, (offset), (val), 0)
-
-#define airoha_qdma_rr(qdma, offset)				\
-	airoha_rr((qdma)->regs, (offset))
-#define airoha_qdma_wr(qdma, offset, val)			\
-	airoha_wr((qdma)->regs, (offset), (val))
-#define airoha_qdma_rmw(qdma, offset, mask, val)		\
-	airoha_rmw((qdma)->regs, (offset), (mask), (val))
-#define airoha_qdma_set(qdma, offset, val)			\
-	airoha_rmw((qdma)->regs, (offset), 0, (val))
-#define airoha_qdma_clear(qdma, offset, val)			\
-	airoha_rmw((qdma)->regs, (offset), (val), 0)
-
 static void airoha_qdma_set_irqmask(struct airoha_qdma *qdma, int index,
 				    u32 clear, u32 set)
 {
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index 3310e0cf85f1d240e95404a0c15703e5f6be26bc..743aaf10235fe09fb2a91b491f4b25064ed8319b 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -248,4 +248,30 @@ struct airoha_eth {
 	struct airoha_gdm_port *ports[AIROHA_MAX_NUM_GDM_PORTS];
 };
 
+u32 airoha_rr(void __iomem *base, u32 offset);
+void airoha_wr(void __iomem *base, u32 offset, u32 val);
+u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val);
+
+#define airoha_fe_rr(eth, offset)				\
+	airoha_rr((eth)->fe_regs, (offset))
+#define airoha_fe_wr(eth, offset, val)				\
+	airoha_wr((eth)->fe_regs, (offset), (val))
+#define airoha_fe_rmw(eth, offset, mask, val)			\
+	airoha_rmw((eth)->fe_regs, (offset), (mask), (val))
+#define airoha_fe_set(eth, offset, val)				\
+	airoha_rmw((eth)->fe_regs, (offset), 0, (val))
+#define airoha_fe_clear(eth, offset, val)			\
+	airoha_rmw((eth)->fe_regs, (offset), (val), 0)
+
+#define airoha_qdma_rr(qdma, offset)				\
+	airoha_rr((qdma)->regs, (offset))
+#define airoha_qdma_wr(qdma, offset, val)			\
+	airoha_wr((qdma)->regs, (offset), (val))
+#define airoha_qdma_rmw(qdma, offset, mask, val)		\
+	airoha_rmw((qdma)->regs, (offset), (mask), (val))
+#define airoha_qdma_set(qdma, offset, val)			\
+	airoha_rmw((qdma)->regs, (offset), 0, (val))
+#define airoha_qdma_clear(qdma, offset, val)			\
+	airoha_rmw((qdma)->regs, (offset), (val), 0)
+
 #endif /* AIROHA_ETH_H */

-- 
2.48.1

[PATCH net-next v3 05/16] net: airoha: Move register definitions in airoha_regs.h

[Lorenzo Bianconi]

Move common airoha_eth register definitions in airoha_regs.h in order
to reuse them for Packet Processor Engine (PPE) codebase.
PPE module is used to enable support for flowtable hw offloading in
airoha_eth driver.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c  | 659 +----------------------------
 drivers/net/ethernet/airoha/airoha_regs.h | 670 ++++++++++++++++++++++++++++++
 2 files changed, 671 insertions(+), 658 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 1c6fb7b9ccbbaec846643343e0347a1e948a575f..b79556f1b4951c687aa89bc5839fc9405581a6c3 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -13,666 +13,9 @@
 #include <net/pkt_cls.h>
 #include <uapi/linux/ppp_defs.h>
 
+#include "airoha_regs.h"
 #include "airoha_eth.h"
 
-/* FE */
-#define PSE_BASE			0x0100
-#define CSR_IFC_BASE			0x0200
-#define CDM1_BASE			0x0400
-#define GDM1_BASE			0x0500
-#define PPE1_BASE			0x0c00
-
-#define CDM2_BASE			0x1400
-#define GDM2_BASE			0x1500
-
-#define GDM3_BASE			0x1100
-#define GDM4_BASE			0x2500
-
-#define GDM_BASE(_n)			\
-	((_n) == 4 ? GDM4_BASE :	\
-	 (_n) == 3 ? GDM3_BASE :	\
-	 (_n) == 2 ? GDM2_BASE : GDM1_BASE)
-
-#define REG_FE_DMA_GLO_CFG		0x0000
-#define FE_DMA_GLO_L2_SPACE_MASK	GENMASK(7, 4)
-#define FE_DMA_GLO_PG_SZ_MASK		BIT(3)
-
-#define REG_FE_RST_GLO_CFG		0x0004
-#define FE_RST_GDM4_MBI_ARB_MASK	BIT(3)
-#define FE_RST_GDM3_MBI_ARB_MASK	BIT(2)
-#define FE_RST_CORE_MASK		BIT(0)
-
-#define REG_FE_WAN_MAC_H		0x0030
-#define REG_FE_LAN_MAC_H		0x0040
-
-#define REG_FE_MAC_LMIN(_n)		((_n) + 0x04)
-#define REG_FE_MAC_LMAX(_n)		((_n) + 0x08)
-
-#define REG_FE_CDM1_OQ_MAP0		0x0050
-#define REG_FE_CDM1_OQ_MAP1		0x0054
-#define REG_FE_CDM1_OQ_MAP2		0x0058
-#define REG_FE_CDM1_OQ_MAP3		0x005c
-
-#define REG_FE_PCE_CFG			0x0070
-#define PCE_DPI_EN_MASK			BIT(2)
-#define PCE_KA_EN_MASK			BIT(1)
-#define PCE_MC_EN_MASK			BIT(0)
-
-#define REG_FE_PSE_QUEUE_CFG_WR		0x0080
-#define PSE_CFG_PORT_ID_MASK		GENMASK(27, 24)
-#define PSE_CFG_QUEUE_ID_MASK		GENMASK(20, 16)
-#define PSE_CFG_WR_EN_MASK		BIT(8)
-#define PSE_CFG_OQRSV_SEL_MASK		BIT(0)
-
-#define REG_FE_PSE_QUEUE_CFG_VAL	0x0084
-#define PSE_CFG_OQ_RSV_MASK		GENMASK(13, 0)
-
-#define PSE_FQ_CFG			0x008c
-#define PSE_FQ_LIMIT_MASK		GENMASK(14, 0)
-
-#define REG_FE_PSE_BUF_SET		0x0090
-#define PSE_SHARE_USED_LTHD_MASK	GENMASK(31, 16)
-#define PSE_ALLRSV_MASK			GENMASK(14, 0)
-
-#define REG_PSE_SHARE_USED_THD		0x0094
-#define PSE_SHARE_USED_MTHD_MASK	GENMASK(31, 16)
-#define PSE_SHARE_USED_HTHD_MASK	GENMASK(15, 0)
-
-#define REG_GDM_MISC_CFG		0x0148
-#define GDM2_RDM_ACK_WAIT_PREF_MASK	BIT(9)
-#define GDM2_CHN_VLD_MODE_MASK		BIT(5)
-
-#define REG_FE_CSR_IFC_CFG		CSR_IFC_BASE
-#define FE_IFC_EN_MASK			BIT(0)
-
-#define REG_FE_VIP_PORT_EN		0x01f0
-#define REG_FE_IFC_PORT_EN		0x01f4
-
-#define REG_PSE_IQ_REV1			(PSE_BASE + 0x08)
-#define PSE_IQ_RES1_P2_MASK		GENMASK(23, 16)
-
-#define REG_PSE_IQ_REV2			(PSE_BASE + 0x0c)
-#define PSE_IQ_RES2_P5_MASK		GENMASK(15, 8)
-#define PSE_IQ_RES2_P4_MASK		GENMASK(7, 0)
-
-#define REG_FE_VIP_EN(_n)		(0x0300 + ((_n) << 3))
-#define PATN_FCPU_EN_MASK		BIT(7)
-#define PATN_SWP_EN_MASK		BIT(6)
-#define PATN_DP_EN_MASK			BIT(5)
-#define PATN_SP_EN_MASK			BIT(4)
-#define PATN_TYPE_MASK			GENMASK(3, 1)
-#define PATN_EN_MASK			BIT(0)
-
-#define REG_FE_VIP_PATN(_n)		(0x0304 + ((_n) << 3))
-#define PATN_DP_MASK			GENMASK(31, 16)
-#define PATN_SP_MASK			GENMASK(15, 0)
-
-#define REG_CDM1_VLAN_CTRL		CDM1_BASE
-#define CDM1_VLAN_MASK			GENMASK(31, 16)
-
-#define REG_CDM1_FWD_CFG		(CDM1_BASE + 0x08)
-#define CDM1_VIP_QSEL_MASK		GENMASK(24, 20)
-
-#define REG_CDM1_CRSN_QSEL(_n)		(CDM1_BASE + 0x10 + ((_n) << 2))
-#define CDM1_CRSN_QSEL_REASON_MASK(_n)	\
-	GENMASK(4 + (((_n) % 4) << 3),	(((_n) % 4) << 3))
-
-#define REG_CDM2_FWD_CFG		(CDM2_BASE + 0x08)
-#define CDM2_OAM_QSEL_MASK		GENMASK(31, 27)
-#define CDM2_VIP_QSEL_MASK		GENMASK(24, 20)
-
-#define REG_CDM2_CRSN_QSEL(_n)		(CDM2_BASE + 0x10 + ((_n) << 2))
-#define CDM2_CRSN_QSEL_REASON_MASK(_n)	\
-	GENMASK(4 + (((_n) % 4) << 3),	(((_n) % 4) << 3))
-
-#define REG_GDM_FWD_CFG(_n)		GDM_BASE(_n)
-#define GDM_DROP_CRC_ERR		BIT(23)
-#define GDM_IP4_CKSUM			BIT(22)
-#define GDM_TCP_CKSUM			BIT(21)
-#define GDM_UDP_CKSUM			BIT(20)
-#define GDM_UCFQ_MASK			GENMASK(15, 12)
-#define GDM_BCFQ_MASK			GENMASK(11, 8)
-#define GDM_MCFQ_MASK			GENMASK(7, 4)
-#define GDM_OCFQ_MASK			GENMASK(3, 0)
-
-#define REG_GDM_INGRESS_CFG(_n)		(GDM_BASE(_n) + 0x10)
-#define GDM_INGRESS_FC_EN_MASK		BIT(1)
-#define GDM_STAG_EN_MASK		BIT(0)
-
-#define REG_GDM_LEN_CFG(_n)		(GDM_BASE(_n) + 0x14)
-#define GDM_SHORT_LEN_MASK		GENMASK(13, 0)
-#define GDM_LONG_LEN_MASK		GENMASK(29, 16)
-
-#define REG_FE_CPORT_CFG		(GDM1_BASE + 0x40)
-#define FE_CPORT_PAD			BIT(26)
-#define FE_CPORT_PORT_XFC_MASK		BIT(25)
-#define FE_CPORT_QUEUE_XFC_MASK		BIT(24)
-
-#define REG_FE_GDM_MIB_CLEAR(_n)	(GDM_BASE(_n) + 0xf0)
-#define FE_GDM_MIB_RX_CLEAR_MASK	BIT(1)
-#define FE_GDM_MIB_TX_CLEAR_MASK	BIT(0)
-
-#define REG_FE_GDM1_MIB_CFG		(GDM1_BASE + 0xf4)
-#define FE_STRICT_RFC2819_MODE_MASK	BIT(31)
-#define FE_GDM1_TX_MIB_SPLIT_EN_MASK	BIT(17)
-#define FE_GDM1_RX_MIB_SPLIT_EN_MASK	BIT(16)
-#define FE_TX_MIB_ID_MASK		GENMASK(15, 8)
-#define FE_RX_MIB_ID_MASK		GENMASK(7, 0)
-
-#define REG_FE_GDM_TX_OK_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x104)
-#define REG_FE_GDM_TX_OK_BYTE_CNT_L(_n)		(GDM_BASE(_n) + 0x10c)
-#define REG_FE_GDM_TX_ETH_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x110)
-#define REG_FE_GDM_TX_ETH_BYTE_CNT_L(_n)	(GDM_BASE(_n) + 0x114)
-#define REG_FE_GDM_TX_ETH_DROP_CNT(_n)		(GDM_BASE(_n) + 0x118)
-#define REG_FE_GDM_TX_ETH_BC_CNT(_n)		(GDM_BASE(_n) + 0x11c)
-#define REG_FE_GDM_TX_ETH_MC_CNT(_n)		(GDM_BASE(_n) + 0x120)
-#define REG_FE_GDM_TX_ETH_RUNT_CNT(_n)		(GDM_BASE(_n) + 0x124)
-#define REG_FE_GDM_TX_ETH_LONG_CNT(_n)		(GDM_BASE(_n) + 0x128)
-#define REG_FE_GDM_TX_ETH_E64_CNT_L(_n)		(GDM_BASE(_n) + 0x12c)
-#define REG_FE_GDM_TX_ETH_L64_CNT_L(_n)		(GDM_BASE(_n) + 0x130)
-#define REG_FE_GDM_TX_ETH_L127_CNT_L(_n)	(GDM_BASE(_n) + 0x134)
-#define REG_FE_GDM_TX_ETH_L255_CNT_L(_n)	(GDM_BASE(_n) + 0x138)
-#define REG_FE_GDM_TX_ETH_L511_CNT_L(_n)	(GDM_BASE(_n) + 0x13c)
-#define REG_FE_GDM_TX_ETH_L1023_CNT_L(_n)	(GDM_BASE(_n) + 0x140)
-
-#define REG_FE_GDM_RX_OK_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x148)
-#define REG_FE_GDM_RX_FC_DROP_CNT(_n)		(GDM_BASE(_n) + 0x14c)
-#define REG_FE_GDM_RX_RC_DROP_CNT(_n)		(GDM_BASE(_n) + 0x150)
-#define REG_FE_GDM_RX_OVERFLOW_DROP_CNT(_n)	(GDM_BASE(_n) + 0x154)
-#define REG_FE_GDM_RX_ERROR_DROP_CNT(_n)	(GDM_BASE(_n) + 0x158)
-#define REG_FE_GDM_RX_OK_BYTE_CNT_L(_n)		(GDM_BASE(_n) + 0x15c)
-#define REG_FE_GDM_RX_ETH_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x160)
-#define REG_FE_GDM_RX_ETH_BYTE_CNT_L(_n)	(GDM_BASE(_n) + 0x164)
-#define REG_FE_GDM_RX_ETH_DROP_CNT(_n)		(GDM_BASE(_n) + 0x168)
-#define REG_FE_GDM_RX_ETH_BC_CNT(_n)		(GDM_BASE(_n) + 0x16c)
-#define REG_FE_GDM_RX_ETH_MC_CNT(_n)		(GDM_BASE(_n) + 0x170)
-#define REG_FE_GDM_RX_ETH_CRC_ERR_CNT(_n)	(GDM_BASE(_n) + 0x174)
-#define REG_FE_GDM_RX_ETH_FRAG_CNT(_n)		(GDM_BASE(_n) + 0x178)
-#define REG_FE_GDM_RX_ETH_JABBER_CNT(_n)	(GDM_BASE(_n) + 0x17c)
-#define REG_FE_GDM_RX_ETH_RUNT_CNT(_n)		(GDM_BASE(_n) + 0x180)
-#define REG_FE_GDM_RX_ETH_LONG_CNT(_n)		(GDM_BASE(_n) + 0x184)
-#define REG_FE_GDM_RX_ETH_E64_CNT_L(_n)		(GDM_BASE(_n) + 0x188)
-#define REG_FE_GDM_RX_ETH_L64_CNT_L(_n)		(GDM_BASE(_n) + 0x18c)
-#define REG_FE_GDM_RX_ETH_L127_CNT_L(_n)	(GDM_BASE(_n) + 0x190)
-#define REG_FE_GDM_RX_ETH_L255_CNT_L(_n)	(GDM_BASE(_n) + 0x194)
-#define REG_FE_GDM_RX_ETH_L511_CNT_L(_n)	(GDM_BASE(_n) + 0x198)
-#define REG_FE_GDM_RX_ETH_L1023_CNT_L(_n)	(GDM_BASE(_n) + 0x19c)
-
-#define REG_PPE1_TB_HASH_CFG		(PPE1_BASE + 0x250)
-#define PPE1_SRAM_TABLE_EN_MASK		BIT(0)
-#define PPE1_SRAM_HASH1_EN_MASK		BIT(8)
-#define PPE1_DRAM_TABLE_EN_MASK		BIT(16)
-#define PPE1_DRAM_HASH1_EN_MASK		BIT(24)
-
-#define REG_FE_GDM_TX_OK_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x280)
-#define REG_FE_GDM_TX_OK_BYTE_CNT_H(_n)		(GDM_BASE(_n) + 0x284)
-#define REG_FE_GDM_TX_ETH_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x288)
-#define REG_FE_GDM_TX_ETH_BYTE_CNT_H(_n)	(GDM_BASE(_n) + 0x28c)
-
-#define REG_FE_GDM_RX_OK_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x290)
-#define REG_FE_GDM_RX_OK_BYTE_CNT_H(_n)		(GDM_BASE(_n) + 0x294)
-#define REG_FE_GDM_RX_ETH_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x298)
-#define REG_FE_GDM_RX_ETH_BYTE_CNT_H(_n)	(GDM_BASE(_n) + 0x29c)
-#define REG_FE_GDM_TX_ETH_E64_CNT_H(_n)		(GDM_BASE(_n) + 0x2b8)
-#define REG_FE_GDM_TX_ETH_L64_CNT_H(_n)		(GDM_BASE(_n) + 0x2bc)
-#define REG_FE_GDM_TX_ETH_L127_CNT_H(_n)	(GDM_BASE(_n) + 0x2c0)
-#define REG_FE_GDM_TX_ETH_L255_CNT_H(_n)	(GDM_BASE(_n) + 0x2c4)
-#define REG_FE_GDM_TX_ETH_L511_CNT_H(_n)	(GDM_BASE(_n) + 0x2c8)
-#define REG_FE_GDM_TX_ETH_L1023_CNT_H(_n)	(GDM_BASE(_n) + 0x2cc)
-#define REG_FE_GDM_RX_ETH_E64_CNT_H(_n)		(GDM_BASE(_n) + 0x2e8)
-#define REG_FE_GDM_RX_ETH_L64_CNT_H(_n)		(GDM_BASE(_n) + 0x2ec)
-#define REG_FE_GDM_RX_ETH_L127_CNT_H(_n)	(GDM_BASE(_n) + 0x2f0)
-#define REG_FE_GDM_RX_ETH_L255_CNT_H(_n)	(GDM_BASE(_n) + 0x2f4)
-#define REG_FE_GDM_RX_ETH_L511_CNT_H(_n)	(GDM_BASE(_n) + 0x2f8)
-#define REG_FE_GDM_RX_ETH_L1023_CNT_H(_n)	(GDM_BASE(_n) + 0x2fc)
-
-#define REG_GDM2_CHN_RLS		(GDM2_BASE + 0x20)
-#define MBI_RX_AGE_SEL_MASK		GENMASK(26, 25)
-#define MBI_TX_AGE_SEL_MASK		GENMASK(18, 17)
-
-#define REG_GDM3_FWD_CFG		GDM3_BASE
-#define GDM3_PAD_EN_MASK		BIT(28)
-
-#define REG_GDM4_FWD_CFG		GDM4_BASE
-#define GDM4_PAD_EN_MASK		BIT(28)
-#define GDM4_SPORT_OFFSET0_MASK		GENMASK(11, 8)
-
-#define REG_GDM4_SRC_PORT_SET		(GDM4_BASE + 0x23c)
-#define GDM4_SPORT_OFF2_MASK		GENMASK(19, 16)
-#define GDM4_SPORT_OFF1_MASK		GENMASK(15, 12)
-#define GDM4_SPORT_OFF0_MASK		GENMASK(11, 8)
-
-#define REG_IP_FRAG_FP			0x2010
-#define IP_ASSEMBLE_PORT_MASK		GENMASK(24, 21)
-#define IP_ASSEMBLE_NBQ_MASK		GENMASK(20, 16)
-#define IP_FRAGMENT_PORT_MASK		GENMASK(8, 5)
-#define IP_FRAGMENT_NBQ_MASK		GENMASK(4, 0)
-
-#define REG_MC_VLAN_EN			0x2100
-#define MC_VLAN_EN_MASK			BIT(0)
-
-#define REG_MC_VLAN_CFG			0x2104
-#define MC_VLAN_CFG_CMD_DONE_MASK	BIT(31)
-#define MC_VLAN_CFG_TABLE_ID_MASK	GENMASK(21, 16)
-#define MC_VLAN_CFG_PORT_ID_MASK	GENMASK(11, 8)
-#define MC_VLAN_CFG_TABLE_SEL_MASK	BIT(4)
-#define MC_VLAN_CFG_RW_MASK		BIT(0)
-
-#define REG_MC_VLAN_DATA		0x2108
-
-#define REG_CDM5_RX_OQ1_DROP_CNT	0x29d4
-
-/* QDMA */
-#define REG_QDMA_GLOBAL_CFG			0x0004
-#define GLOBAL_CFG_RX_2B_OFFSET_MASK		BIT(31)
-#define GLOBAL_CFG_DMA_PREFERENCE_MASK		GENMASK(30, 29)
-#define GLOBAL_CFG_CPU_TXR_RR_MASK		BIT(28)
-#define GLOBAL_CFG_DSCP_BYTE_SWAP_MASK		BIT(27)
-#define GLOBAL_CFG_PAYLOAD_BYTE_SWAP_MASK	BIT(26)
-#define GLOBAL_CFG_MULTICAST_MODIFY_FP_MASK	BIT(25)
-#define GLOBAL_CFG_OAM_MODIFY_MASK		BIT(24)
-#define GLOBAL_CFG_RESET_MASK			BIT(23)
-#define GLOBAL_CFG_RESET_DONE_MASK		BIT(22)
-#define GLOBAL_CFG_MULTICAST_EN_MASK		BIT(21)
-#define GLOBAL_CFG_IRQ1_EN_MASK			BIT(20)
-#define GLOBAL_CFG_IRQ0_EN_MASK			BIT(19)
-#define GLOBAL_CFG_LOOPCNT_EN_MASK		BIT(18)
-#define GLOBAL_CFG_RD_BYPASS_WR_MASK		BIT(17)
-#define GLOBAL_CFG_QDMA_LOOPBACK_MASK		BIT(16)
-#define GLOBAL_CFG_LPBK_RXQ_SEL_MASK		GENMASK(13, 8)
-#define GLOBAL_CFG_CHECK_DONE_MASK		BIT(7)
-#define GLOBAL_CFG_TX_WB_DONE_MASK		BIT(6)
-#define GLOBAL_CFG_MAX_ISSUE_NUM_MASK		GENMASK(5, 4)
-#define GLOBAL_CFG_RX_DMA_BUSY_MASK		BIT(3)
-#define GLOBAL_CFG_RX_DMA_EN_MASK		BIT(2)
-#define GLOBAL_CFG_TX_DMA_BUSY_MASK		BIT(1)
-#define GLOBAL_CFG_TX_DMA_EN_MASK		BIT(0)
-
-#define REG_FWD_DSCP_BASE			0x0010
-#define REG_FWD_BUF_BASE			0x0014
-
-#define REG_HW_FWD_DSCP_CFG			0x0018
-#define HW_FWD_DSCP_PAYLOAD_SIZE_MASK		GENMASK(29, 28)
-#define HW_FWD_DSCP_SCATTER_LEN_MASK		GENMASK(17, 16)
-#define HW_FWD_DSCP_MIN_SCATTER_LEN_MASK	GENMASK(15, 0)
-
-#define REG_INT_STATUS(_n)		\
-	(((_n) == 4) ? 0x0730 :		\
-	 ((_n) == 3) ? 0x0724 :		\
-	 ((_n) == 2) ? 0x0720 :		\
-	 ((_n) == 1) ? 0x0024 : 0x0020)
-
-#define REG_INT_ENABLE(_n)		\
-	(((_n) == 4) ? 0x0750 :		\
-	 ((_n) == 3) ? 0x0744 :		\
-	 ((_n) == 2) ? 0x0740 :		\
-	 ((_n) == 1) ? 0x002c : 0x0028)
-
-/* QDMA_CSR_INT_ENABLE1 */
-#define RX15_COHERENT_INT_MASK		BIT(31)
-#define RX14_COHERENT_INT_MASK		BIT(30)
-#define RX13_COHERENT_INT_MASK		BIT(29)
-#define RX12_COHERENT_INT_MASK		BIT(28)
-#define RX11_COHERENT_INT_MASK		BIT(27)
-#define RX10_COHERENT_INT_MASK		BIT(26)
-#define RX9_COHERENT_INT_MASK		BIT(25)
-#define RX8_COHERENT_INT_MASK		BIT(24)
-#define RX7_COHERENT_INT_MASK		BIT(23)
-#define RX6_COHERENT_INT_MASK		BIT(22)
-#define RX5_COHERENT_INT_MASK		BIT(21)
-#define RX4_COHERENT_INT_MASK		BIT(20)
-#define RX3_COHERENT_INT_MASK		BIT(19)
-#define RX2_COHERENT_INT_MASK		BIT(18)
-#define RX1_COHERENT_INT_MASK		BIT(17)
-#define RX0_COHERENT_INT_MASK		BIT(16)
-#define TX7_COHERENT_INT_MASK		BIT(15)
-#define TX6_COHERENT_INT_MASK		BIT(14)
-#define TX5_COHERENT_INT_MASK		BIT(13)
-#define TX4_COHERENT_INT_MASK		BIT(12)
-#define TX3_COHERENT_INT_MASK		BIT(11)
-#define TX2_COHERENT_INT_MASK		BIT(10)
-#define TX1_COHERENT_INT_MASK		BIT(9)
-#define TX0_COHERENT_INT_MASK		BIT(8)
-#define CNT_OVER_FLOW_INT_MASK		BIT(7)
-#define IRQ1_FULL_INT_MASK		BIT(5)
-#define IRQ1_INT_MASK			BIT(4)
-#define HWFWD_DSCP_LOW_INT_MASK		BIT(3)
-#define HWFWD_DSCP_EMPTY_INT_MASK	BIT(2)
-#define IRQ0_FULL_INT_MASK		BIT(1)
-#define IRQ0_INT_MASK			BIT(0)
-
-#define TX_DONE_INT_MASK(_n)					\
-	((_n) ? IRQ1_INT_MASK | IRQ1_FULL_INT_MASK		\
-	      : IRQ0_INT_MASK | IRQ0_FULL_INT_MASK)
-
-#define INT_TX_MASK						\
-	(IRQ1_INT_MASK | IRQ1_FULL_INT_MASK |			\
-	 IRQ0_INT_MASK | IRQ0_FULL_INT_MASK)
-
-#define INT_IDX0_MASK						\
-	(TX0_COHERENT_INT_MASK | TX1_COHERENT_INT_MASK |	\
-	 TX2_COHERENT_INT_MASK | TX3_COHERENT_INT_MASK |	\
-	 TX4_COHERENT_INT_MASK | TX5_COHERENT_INT_MASK |	\
-	 TX6_COHERENT_INT_MASK | TX7_COHERENT_INT_MASK |	\
-	 RX0_COHERENT_INT_MASK | RX1_COHERENT_INT_MASK |	\
-	 RX2_COHERENT_INT_MASK | RX3_COHERENT_INT_MASK |	\
-	 RX4_COHERENT_INT_MASK | RX7_COHERENT_INT_MASK |	\
-	 RX8_COHERENT_INT_MASK | RX9_COHERENT_INT_MASK |	\
-	 RX15_COHERENT_INT_MASK | INT_TX_MASK)
-
-/* QDMA_CSR_INT_ENABLE2 */
-#define RX15_NO_CPU_DSCP_INT_MASK	BIT(31)
-#define RX14_NO_CPU_DSCP_INT_MASK	BIT(30)
-#define RX13_NO_CPU_DSCP_INT_MASK	BIT(29)
-#define RX12_NO_CPU_DSCP_INT_MASK	BIT(28)
-#define RX11_NO_CPU_DSCP_INT_MASK	BIT(27)
-#define RX10_NO_CPU_DSCP_INT_MASK	BIT(26)
-#define RX9_NO_CPU_DSCP_INT_MASK	BIT(25)
-#define RX8_NO_CPU_DSCP_INT_MASK	BIT(24)
-#define RX7_NO_CPU_DSCP_INT_MASK	BIT(23)
-#define RX6_NO_CPU_DSCP_INT_MASK	BIT(22)
-#define RX5_NO_CPU_DSCP_INT_MASK	BIT(21)
-#define RX4_NO_CPU_DSCP_INT_MASK	BIT(20)
-#define RX3_NO_CPU_DSCP_INT_MASK	BIT(19)
-#define RX2_NO_CPU_DSCP_INT_MASK	BIT(18)
-#define RX1_NO_CPU_DSCP_INT_MASK	BIT(17)
-#define RX0_NO_CPU_DSCP_INT_MASK	BIT(16)
-#define RX15_DONE_INT_MASK		BIT(15)
-#define RX14_DONE_INT_MASK		BIT(14)
-#define RX13_DONE_INT_MASK		BIT(13)
-#define RX12_DONE_INT_MASK		BIT(12)
-#define RX11_DONE_INT_MASK		BIT(11)
-#define RX10_DONE_INT_MASK		BIT(10)
-#define RX9_DONE_INT_MASK		BIT(9)
-#define RX8_DONE_INT_MASK		BIT(8)
-#define RX7_DONE_INT_MASK		BIT(7)
-#define RX6_DONE_INT_MASK		BIT(6)
-#define RX5_DONE_INT_MASK		BIT(5)
-#define RX4_DONE_INT_MASK		BIT(4)
-#define RX3_DONE_INT_MASK		BIT(3)
-#define RX2_DONE_INT_MASK		BIT(2)
-#define RX1_DONE_INT_MASK		BIT(1)
-#define RX0_DONE_INT_MASK		BIT(0)
-
-#define RX_DONE_INT_MASK					\
-	(RX0_DONE_INT_MASK | RX1_DONE_INT_MASK |		\
-	 RX2_DONE_INT_MASK | RX3_DONE_INT_MASK |		\
-	 RX4_DONE_INT_MASK | RX7_DONE_INT_MASK |		\
-	 RX8_DONE_INT_MASK | RX9_DONE_INT_MASK |		\
-	 RX15_DONE_INT_MASK)
-#define INT_IDX1_MASK						\
-	(RX_DONE_INT_MASK |					\
-	 RX0_NO_CPU_DSCP_INT_MASK | RX1_NO_CPU_DSCP_INT_MASK |	\
-	 RX2_NO_CPU_DSCP_INT_MASK | RX3_NO_CPU_DSCP_INT_MASK |	\
-	 RX4_NO_CPU_DSCP_INT_MASK | RX7_NO_CPU_DSCP_INT_MASK |	\
-	 RX8_NO_CPU_DSCP_INT_MASK | RX9_NO_CPU_DSCP_INT_MASK |	\
-	 RX15_NO_CPU_DSCP_INT_MASK)
-
-/* QDMA_CSR_INT_ENABLE5 */
-#define TX31_COHERENT_INT_MASK		BIT(31)
-#define TX30_COHERENT_INT_MASK		BIT(30)
-#define TX29_COHERENT_INT_MASK		BIT(29)
-#define TX28_COHERENT_INT_MASK		BIT(28)
-#define TX27_COHERENT_INT_MASK		BIT(27)
-#define TX26_COHERENT_INT_MASK		BIT(26)
-#define TX25_COHERENT_INT_MASK		BIT(25)
-#define TX24_COHERENT_INT_MASK		BIT(24)
-#define TX23_COHERENT_INT_MASK		BIT(23)
-#define TX22_COHERENT_INT_MASK		BIT(22)
-#define TX21_COHERENT_INT_MASK		BIT(21)
-#define TX20_COHERENT_INT_MASK		BIT(20)
-#define TX19_COHERENT_INT_MASK		BIT(19)
-#define TX18_COHERENT_INT_MASK		BIT(18)
-#define TX17_COHERENT_INT_MASK		BIT(17)
-#define TX16_COHERENT_INT_MASK		BIT(16)
-#define TX15_COHERENT_INT_MASK		BIT(15)
-#define TX14_COHERENT_INT_MASK		BIT(14)
-#define TX13_COHERENT_INT_MASK		BIT(13)
-#define TX12_COHERENT_INT_MASK		BIT(12)
-#define TX11_COHERENT_INT_MASK		BIT(11)
-#define TX10_COHERENT_INT_MASK		BIT(10)
-#define TX9_COHERENT_INT_MASK		BIT(9)
-#define TX8_COHERENT_INT_MASK		BIT(8)
-
-#define INT_IDX4_MASK						\
-	(TX8_COHERENT_INT_MASK | TX9_COHERENT_INT_MASK |	\
-	 TX10_COHERENT_INT_MASK | TX11_COHERENT_INT_MASK |	\
-	 TX12_COHERENT_INT_MASK | TX13_COHERENT_INT_MASK |	\
-	 TX14_COHERENT_INT_MASK | TX15_COHERENT_INT_MASK |	\
-	 TX16_COHERENT_INT_MASK | TX17_COHERENT_INT_MASK |	\
-	 TX18_COHERENT_INT_MASK | TX19_COHERENT_INT_MASK |	\
-	 TX20_COHERENT_INT_MASK | TX21_COHERENT_INT_MASK |	\
-	 TX22_COHERENT_INT_MASK | TX23_COHERENT_INT_MASK |	\
-	 TX24_COHERENT_INT_MASK | TX25_COHERENT_INT_MASK |	\
-	 TX26_COHERENT_INT_MASK | TX27_COHERENT_INT_MASK |	\
-	 TX28_COHERENT_INT_MASK | TX29_COHERENT_INT_MASK |	\
-	 TX30_COHERENT_INT_MASK | TX31_COHERENT_INT_MASK)
-
-#define REG_TX_IRQ_BASE(_n)		((_n) ? 0x0048 : 0x0050)
-
-#define REG_TX_IRQ_CFG(_n)		((_n) ? 0x004c : 0x0054)
-#define TX_IRQ_THR_MASK			GENMASK(27, 16)
-#define TX_IRQ_DEPTH_MASK		GENMASK(11, 0)
-
-#define REG_IRQ_CLEAR_LEN(_n)		((_n) ? 0x0064 : 0x0058)
-#define IRQ_CLEAR_LEN_MASK		GENMASK(7, 0)
-
-#define REG_IRQ_STATUS(_n)		((_n) ? 0x0068 : 0x005c)
-#define IRQ_ENTRY_LEN_MASK		GENMASK(27, 16)
-#define IRQ_HEAD_IDX_MASK		GENMASK(11, 0)
-
-#define REG_TX_RING_BASE(_n)	\
-	(((_n) < 8) ? 0x0100 + ((_n) << 5) : 0x0b00 + (((_n) - 8) << 5))
-
-#define REG_TX_RING_BLOCKING(_n)	\
-	(((_n) < 8) ? 0x0104 + ((_n) << 5) : 0x0b04 + (((_n) - 8) << 5))
-
-#define TX_RING_IRQ_BLOCKING_MAP_MASK			BIT(6)
-#define TX_RING_IRQ_BLOCKING_CFG_MASK			BIT(4)
-#define TX_RING_IRQ_BLOCKING_TX_DROP_EN_MASK		BIT(2)
-#define TX_RING_IRQ_BLOCKING_MAX_TH_TXRING_EN_MASK	BIT(1)
-#define TX_RING_IRQ_BLOCKING_MIN_TH_TXRING_EN_MASK	BIT(0)
-
-#define REG_TX_CPU_IDX(_n)	\
-	(((_n) < 8) ? 0x0108 + ((_n) << 5) : 0x0b08 + (((_n) - 8) << 5))
-
-#define TX_RING_CPU_IDX_MASK		GENMASK(15, 0)
-
-#define REG_TX_DMA_IDX(_n)	\
-	(((_n) < 8) ? 0x010c + ((_n) << 5) : 0x0b0c + (((_n) - 8) << 5))
-
-#define TX_RING_DMA_IDX_MASK		GENMASK(15, 0)
-
-#define IRQ_RING_IDX_MASK		GENMASK(20, 16)
-#define IRQ_DESC_IDX_MASK		GENMASK(15, 0)
-
-#define REG_RX_RING_BASE(_n)	\
-	(((_n) < 16) ? 0x0200 + ((_n) << 5) : 0x0e00 + (((_n) - 16) << 5))
-
-#define REG_RX_RING_SIZE(_n)	\
-	(((_n) < 16) ? 0x0204 + ((_n) << 5) : 0x0e04 + (((_n) - 16) << 5))
-
-#define RX_RING_THR_MASK		GENMASK(31, 16)
-#define RX_RING_SIZE_MASK		GENMASK(15, 0)
-
-#define REG_RX_CPU_IDX(_n)	\
-	(((_n) < 16) ? 0x0208 + ((_n) << 5) : 0x0e08 + (((_n) - 16) << 5))
-
-#define RX_RING_CPU_IDX_MASK		GENMASK(15, 0)
-
-#define REG_RX_DMA_IDX(_n)	\
-	(((_n) < 16) ? 0x020c + ((_n) << 5) : 0x0e0c + (((_n) - 16) << 5))
-
-#define REG_RX_DELAY_INT_IDX(_n)	\
-	(((_n) < 16) ? 0x0210 + ((_n) << 5) : 0x0e10 + (((_n) - 16) << 5))
-
-#define RX_DELAY_INT_MASK		GENMASK(15, 0)
-
-#define RX_RING_DMA_IDX_MASK		GENMASK(15, 0)
-
-#define REG_INGRESS_TRTCM_CFG		0x0070
-#define INGRESS_TRTCM_EN_MASK		BIT(31)
-#define INGRESS_TRTCM_MODE_MASK		BIT(30)
-#define INGRESS_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
-#define INGRESS_FAST_TICK_MASK		GENMASK(15, 0)
-
-#define REG_QUEUE_CLOSE_CFG(_n)		(0x00a0 + ((_n) & 0xfc))
-#define TXQ_DISABLE_CHAN_QUEUE_MASK(_n, _m)	BIT((_m) + (((_n) & 0x3) << 3))
-
-#define REG_TXQ_DIS_CFG_BASE(_n)	((_n) ? 0x20a0 : 0x00a0)
-#define REG_TXQ_DIS_CFG(_n, _m)		(REG_TXQ_DIS_CFG_BASE((_n)) + (_m) << 2)
-
-#define REG_CNTR_CFG(_n)		(0x0400 + ((_n) << 3))
-#define CNTR_EN_MASK			BIT(31)
-#define CNTR_ALL_CHAN_EN_MASK		BIT(30)
-#define CNTR_ALL_QUEUE_EN_MASK		BIT(29)
-#define CNTR_ALL_DSCP_RING_EN_MASK	BIT(28)
-#define CNTR_SRC_MASK			GENMASK(27, 24)
-#define CNTR_DSCP_RING_MASK		GENMASK(20, 16)
-#define CNTR_CHAN_MASK			GENMASK(7, 3)
-#define CNTR_QUEUE_MASK			GENMASK(2, 0)
-
-#define REG_CNTR_VAL(_n)		(0x0404 + ((_n) << 3))
-
-#define REG_LMGR_INIT_CFG		0x1000
-#define LMGR_INIT_START			BIT(31)
-#define LMGR_SRAM_MODE_MASK		BIT(30)
-#define HW_FWD_PKTSIZE_OVERHEAD_MASK	GENMASK(27, 20)
-#define HW_FWD_DESC_NUM_MASK		GENMASK(16, 0)
-
-#define REG_FWD_DSCP_LOW_THR		0x1004
-#define FWD_DSCP_LOW_THR_MASK		GENMASK(17, 0)
-
-#define REG_EGRESS_RATE_METER_CFG		0x100c
-#define EGRESS_RATE_METER_EN_MASK		BIT(31)
-#define EGRESS_RATE_METER_EQ_RATE_EN_MASK	BIT(17)
-#define EGRESS_RATE_METER_WINDOW_SZ_MASK	GENMASK(16, 12)
-#define EGRESS_RATE_METER_TIMESLICE_MASK	GENMASK(10, 0)
-
-#define REG_EGRESS_TRTCM_CFG		0x1010
-#define EGRESS_TRTCM_EN_MASK		BIT(31)
-#define EGRESS_TRTCM_MODE_MASK		BIT(30)
-#define EGRESS_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
-#define EGRESS_FAST_TICK_MASK		GENMASK(15, 0)
-
-#define TRTCM_PARAM_RW_MASK		BIT(31)
-#define TRTCM_PARAM_RW_DONE_MASK	BIT(30)
-#define TRTCM_PARAM_TYPE_MASK		GENMASK(29, 28)
-#define TRTCM_METER_GROUP_MASK		GENMASK(27, 26)
-#define TRTCM_PARAM_INDEX_MASK		GENMASK(23, 17)
-#define TRTCM_PARAM_RATE_TYPE_MASK	BIT(16)
-
-#define REG_TRTCM_CFG_PARAM(_n)		((_n) + 0x4)
-#define REG_TRTCM_DATA_LOW(_n)		((_n) + 0x8)
-#define REG_TRTCM_DATA_HIGH(_n)		((_n) + 0xc)
-
-#define REG_TXWRR_MODE_CFG		0x1020
-#define TWRR_WEIGHT_SCALE_MASK		BIT(31)
-#define TWRR_WEIGHT_BASE_MASK		BIT(3)
-
-#define REG_TXWRR_WEIGHT_CFG		0x1024
-#define TWRR_RW_CMD_MASK		BIT(31)
-#define TWRR_RW_CMD_DONE		BIT(30)
-#define TWRR_CHAN_IDX_MASK		GENMASK(23, 19)
-#define TWRR_QUEUE_IDX_MASK		GENMASK(18, 16)
-#define TWRR_VALUE_MASK			GENMASK(15, 0)
-
-#define REG_PSE_BUF_USAGE_CFG		0x1028
-#define PSE_BUF_ESTIMATE_EN_MASK	BIT(29)
-
-#define REG_CHAN_QOS_MODE(_n)		(0x1040 + ((_n) << 2))
-#define CHAN_QOS_MODE_MASK(_n)		GENMASK(2 + ((_n) << 2), (_n) << 2)
-
-#define REG_GLB_TRTCM_CFG		0x1080
-#define GLB_TRTCM_EN_MASK		BIT(31)
-#define GLB_TRTCM_MODE_MASK		BIT(30)
-#define GLB_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
-#define GLB_FAST_TICK_MASK		GENMASK(15, 0)
-
-#define REG_TXQ_CNGST_CFG		0x10a0
-#define TXQ_CNGST_DROP_EN		BIT(31)
-#define TXQ_CNGST_DEI_DROP_EN		BIT(30)
-
-#define REG_SLA_TRTCM_CFG		0x1150
-#define SLA_TRTCM_EN_MASK		BIT(31)
-#define SLA_TRTCM_MODE_MASK		BIT(30)
-#define SLA_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
-#define SLA_FAST_TICK_MASK		GENMASK(15, 0)
-
-/* CTRL */
-#define QDMA_DESC_DONE_MASK		BIT(31)
-#define QDMA_DESC_DROP_MASK		BIT(30) /* tx: drop - rx: overflow */
-#define QDMA_DESC_MORE_MASK		BIT(29) /* more SG elements */
-#define QDMA_DESC_DEI_MASK		BIT(25)
-#define QDMA_DESC_NO_DROP_MASK		BIT(24)
-#define QDMA_DESC_LEN_MASK		GENMASK(15, 0)
-/* DATA */
-#define QDMA_DESC_NEXT_ID_MASK		GENMASK(15, 0)
-/* TX MSG0 */
-#define QDMA_ETH_TXMSG_MIC_IDX_MASK	BIT(30)
-#define QDMA_ETH_TXMSG_SP_TAG_MASK	GENMASK(29, 14)
-#define QDMA_ETH_TXMSG_ICO_MASK		BIT(13)
-#define QDMA_ETH_TXMSG_UCO_MASK		BIT(12)
-#define QDMA_ETH_TXMSG_TCO_MASK		BIT(11)
-#define QDMA_ETH_TXMSG_TSO_MASK		BIT(10)
-#define QDMA_ETH_TXMSG_FAST_MASK	BIT(9)
-#define QDMA_ETH_TXMSG_OAM_MASK		BIT(8)
-#define QDMA_ETH_TXMSG_CHAN_MASK	GENMASK(7, 3)
-#define QDMA_ETH_TXMSG_QUEUE_MASK	GENMASK(2, 0)
-/* TX MSG1 */
-#define QDMA_ETH_TXMSG_NO_DROP		BIT(31)
-#define QDMA_ETH_TXMSG_METER_MASK	GENMASK(30, 24)	/* 0x7f no meters */
-#define QDMA_ETH_TXMSG_FPORT_MASK	GENMASK(23, 20)
-#define QDMA_ETH_TXMSG_NBOQ_MASK	GENMASK(19, 15)
-#define QDMA_ETH_TXMSG_HWF_MASK		BIT(14)
-#define QDMA_ETH_TXMSG_HOP_MASK		BIT(13)
-#define QDMA_ETH_TXMSG_PTP_MASK		BIT(12)
-#define QDMA_ETH_TXMSG_ACNT_G1_MASK	GENMASK(10, 6)	/* 0x1f do not count */
-#define QDMA_ETH_TXMSG_ACNT_G0_MASK	GENMASK(5, 0)	/* 0x3f do not count */
-
-/* RX MSG1 */
-#define QDMA_ETH_RXMSG_DEI_MASK		BIT(31)
-#define QDMA_ETH_RXMSG_IP6_MASK		BIT(30)
-#define QDMA_ETH_RXMSG_IP4_MASK		BIT(29)
-#define QDMA_ETH_RXMSG_IP4F_MASK	BIT(28)
-#define QDMA_ETH_RXMSG_L4_VALID_MASK	BIT(27)
-#define QDMA_ETH_RXMSG_L4F_MASK		BIT(26)
-#define QDMA_ETH_RXMSG_SPORT_MASK	GENMASK(25, 21)
-#define QDMA_ETH_RXMSG_CRSN_MASK	GENMASK(20, 16)
-#define QDMA_ETH_RXMSG_PPE_ENTRY_MASK	GENMASK(15, 0)
-
-struct airoha_qdma_desc {
-	__le32 rsv;
-	__le32 ctrl;
-	__le32 addr;
-	__le32 data;
-	__le32 msg0;
-	__le32 msg1;
-	__le32 msg2;
-	__le32 msg3;
-};
-
-/* CTRL0 */
-#define QDMA_FWD_DESC_CTX_MASK		BIT(31)
-#define QDMA_FWD_DESC_RING_MASK		GENMASK(30, 28)
-#define QDMA_FWD_DESC_IDX_MASK		GENMASK(27, 16)
-#define QDMA_FWD_DESC_LEN_MASK		GENMASK(15, 0)
-/* CTRL1 */
-#define QDMA_FWD_DESC_FIRST_IDX_MASK	GENMASK(15, 0)
-/* CTRL2 */
-#define QDMA_FWD_DESC_MORE_PKT_NUM_MASK	GENMASK(2, 0)
-
-struct airoha_qdma_fwd_desc {
-	__le32 addr;
-	__le32 ctrl0;
-	__le32 ctrl1;
-	__le32 ctrl2;
-	__le32 msg0;
-	__le32 msg1;
-	__le32 rsv0;
-	__le32 rsv1;
-};
-
 u32 airoha_rr(void __iomem *base, u32 offset)
 {
 	return readl(base + offset);
diff --git a/drivers/net/ethernet/airoha/airoha_regs.h b/drivers/net/ethernet/airoha/airoha_regs.h
new file mode 100644
index 0000000000000000000000000000000000000000..7c9dadb348834cb5a856760abe45e8221d6fd700
--- /dev/null
+++ b/drivers/net/ethernet/airoha/airoha_regs.h
@@ -0,0 +1,670 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2024 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#ifndef AIROHA_REGS_H
+#define AIROHA_REGS_H
+
+#include <linux/types.h>
+
+/* FE */
+#define PSE_BASE			0x0100
+#define CSR_IFC_BASE			0x0200
+#define CDM1_BASE			0x0400
+#define GDM1_BASE			0x0500
+#define PPE1_BASE			0x0c00
+
+#define CDM2_BASE			0x1400
+#define GDM2_BASE			0x1500
+
+#define GDM3_BASE			0x1100
+#define GDM4_BASE			0x2500
+
+#define GDM_BASE(_n)			\
+	((_n) == 4 ? GDM4_BASE :	\
+	 (_n) == 3 ? GDM3_BASE :	\
+	 (_n) == 2 ? GDM2_BASE : GDM1_BASE)
+
+#define REG_FE_DMA_GLO_CFG		0x0000
+#define FE_DMA_GLO_L2_SPACE_MASK	GENMASK(7, 4)
+#define FE_DMA_GLO_PG_SZ_MASK		BIT(3)
+
+#define REG_FE_RST_GLO_CFG		0x0004
+#define FE_RST_GDM4_MBI_ARB_MASK	BIT(3)
+#define FE_RST_GDM3_MBI_ARB_MASK	BIT(2)
+#define FE_RST_CORE_MASK		BIT(0)
+
+#define REG_FE_WAN_MAC_H		0x0030
+#define REG_FE_LAN_MAC_H		0x0040
+
+#define REG_FE_MAC_LMIN(_n)		((_n) + 0x04)
+#define REG_FE_MAC_LMAX(_n)		((_n) + 0x08)
+
+#define REG_FE_CDM1_OQ_MAP0		0x0050
+#define REG_FE_CDM1_OQ_MAP1		0x0054
+#define REG_FE_CDM1_OQ_MAP2		0x0058
+#define REG_FE_CDM1_OQ_MAP3		0x005c
+
+#define REG_FE_PCE_CFG			0x0070
+#define PCE_DPI_EN_MASK			BIT(2)
+#define PCE_KA_EN_MASK			BIT(1)
+#define PCE_MC_EN_MASK			BIT(0)
+
+#define REG_FE_PSE_QUEUE_CFG_WR		0x0080
+#define PSE_CFG_PORT_ID_MASK		GENMASK(27, 24)
+#define PSE_CFG_QUEUE_ID_MASK		GENMASK(20, 16)
+#define PSE_CFG_WR_EN_MASK		BIT(8)
+#define PSE_CFG_OQRSV_SEL_MASK		BIT(0)
+
+#define REG_FE_PSE_QUEUE_CFG_VAL	0x0084
+#define PSE_CFG_OQ_RSV_MASK		GENMASK(13, 0)
+
+#define PSE_FQ_CFG			0x008c
+#define PSE_FQ_LIMIT_MASK		GENMASK(14, 0)
+
+#define REG_FE_PSE_BUF_SET		0x0090
+#define PSE_SHARE_USED_LTHD_MASK	GENMASK(31, 16)
+#define PSE_ALLRSV_MASK			GENMASK(14, 0)
+
+#define REG_PSE_SHARE_USED_THD		0x0094
+#define PSE_SHARE_USED_MTHD_MASK	GENMASK(31, 16)
+#define PSE_SHARE_USED_HTHD_MASK	GENMASK(15, 0)
+
+#define REG_GDM_MISC_CFG		0x0148
+#define GDM2_RDM_ACK_WAIT_PREF_MASK	BIT(9)
+#define GDM2_CHN_VLD_MODE_MASK		BIT(5)
+
+#define REG_FE_CSR_IFC_CFG		CSR_IFC_BASE
+#define FE_IFC_EN_MASK			BIT(0)
+
+#define REG_FE_VIP_PORT_EN		0x01f0
+#define REG_FE_IFC_PORT_EN		0x01f4
+
+#define REG_PSE_IQ_REV1			(PSE_BASE + 0x08)
+#define PSE_IQ_RES1_P2_MASK		GENMASK(23, 16)
+
+#define REG_PSE_IQ_REV2			(PSE_BASE + 0x0c)
+#define PSE_IQ_RES2_P5_MASK		GENMASK(15, 8)
+#define PSE_IQ_RES2_P4_MASK		GENMASK(7, 0)
+
+#define REG_FE_VIP_EN(_n)		(0x0300 + ((_n) << 3))
+#define PATN_FCPU_EN_MASK		BIT(7)
+#define PATN_SWP_EN_MASK		BIT(6)
+#define PATN_DP_EN_MASK			BIT(5)
+#define PATN_SP_EN_MASK			BIT(4)
+#define PATN_TYPE_MASK			GENMASK(3, 1)
+#define PATN_EN_MASK			BIT(0)
+
+#define REG_FE_VIP_PATN(_n)		(0x0304 + ((_n) << 3))
+#define PATN_DP_MASK			GENMASK(31, 16)
+#define PATN_SP_MASK			GENMASK(15, 0)
+
+#define REG_CDM1_VLAN_CTRL		CDM1_BASE
+#define CDM1_VLAN_MASK			GENMASK(31, 16)
+
+#define REG_CDM1_FWD_CFG		(CDM1_BASE + 0x08)
+#define CDM1_VIP_QSEL_MASK		GENMASK(24, 20)
+
+#define REG_CDM1_CRSN_QSEL(_n)		(CDM1_BASE + 0x10 + ((_n) << 2))
+#define CDM1_CRSN_QSEL_REASON_MASK(_n)	\
+	GENMASK(4 + (((_n) % 4) << 3),	(((_n) % 4) << 3))
+
+#define REG_CDM2_FWD_CFG		(CDM2_BASE + 0x08)
+#define CDM2_OAM_QSEL_MASK		GENMASK(31, 27)
+#define CDM2_VIP_QSEL_MASK		GENMASK(24, 20)
+
+#define REG_CDM2_CRSN_QSEL(_n)		(CDM2_BASE + 0x10 + ((_n) << 2))
+#define CDM2_CRSN_QSEL_REASON_MASK(_n)	\
+	GENMASK(4 + (((_n) % 4) << 3),	(((_n) % 4) << 3))
+
+#define REG_GDM_FWD_CFG(_n)		GDM_BASE(_n)
+#define GDM_DROP_CRC_ERR		BIT(23)
+#define GDM_IP4_CKSUM			BIT(22)
+#define GDM_TCP_CKSUM			BIT(21)
+#define GDM_UDP_CKSUM			BIT(20)
+#define GDM_UCFQ_MASK			GENMASK(15, 12)
+#define GDM_BCFQ_MASK			GENMASK(11, 8)
+#define GDM_MCFQ_MASK			GENMASK(7, 4)
+#define GDM_OCFQ_MASK			GENMASK(3, 0)
+
+#define REG_GDM_INGRESS_CFG(_n)		(GDM_BASE(_n) + 0x10)
+#define GDM_INGRESS_FC_EN_MASK		BIT(1)
+#define GDM_STAG_EN_MASK		BIT(0)
+
+#define REG_GDM_LEN_CFG(_n)		(GDM_BASE(_n) + 0x14)
+#define GDM_SHORT_LEN_MASK		GENMASK(13, 0)
+#define GDM_LONG_LEN_MASK		GENMASK(29, 16)
+
+#define REG_FE_CPORT_CFG		(GDM1_BASE + 0x40)
+#define FE_CPORT_PAD			BIT(26)
+#define FE_CPORT_PORT_XFC_MASK		BIT(25)
+#define FE_CPORT_QUEUE_XFC_MASK		BIT(24)
+
+#define REG_FE_GDM_MIB_CLEAR(_n)	(GDM_BASE(_n) + 0xf0)
+#define FE_GDM_MIB_RX_CLEAR_MASK	BIT(1)
+#define FE_GDM_MIB_TX_CLEAR_MASK	BIT(0)
+
+#define REG_FE_GDM1_MIB_CFG		(GDM1_BASE + 0xf4)
+#define FE_STRICT_RFC2819_MODE_MASK	BIT(31)
+#define FE_GDM1_TX_MIB_SPLIT_EN_MASK	BIT(17)
+#define FE_GDM1_RX_MIB_SPLIT_EN_MASK	BIT(16)
+#define FE_TX_MIB_ID_MASK		GENMASK(15, 8)
+#define FE_RX_MIB_ID_MASK		GENMASK(7, 0)
+
+#define REG_FE_GDM_TX_OK_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x104)
+#define REG_FE_GDM_TX_OK_BYTE_CNT_L(_n)		(GDM_BASE(_n) + 0x10c)
+#define REG_FE_GDM_TX_ETH_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x110)
+#define REG_FE_GDM_TX_ETH_BYTE_CNT_L(_n)	(GDM_BASE(_n) + 0x114)
+#define REG_FE_GDM_TX_ETH_DROP_CNT(_n)		(GDM_BASE(_n) + 0x118)
+#define REG_FE_GDM_TX_ETH_BC_CNT(_n)		(GDM_BASE(_n) + 0x11c)
+#define REG_FE_GDM_TX_ETH_MC_CNT(_n)		(GDM_BASE(_n) + 0x120)
+#define REG_FE_GDM_TX_ETH_RUNT_CNT(_n)		(GDM_BASE(_n) + 0x124)
+#define REG_FE_GDM_TX_ETH_LONG_CNT(_n)		(GDM_BASE(_n) + 0x128)
+#define REG_FE_GDM_TX_ETH_E64_CNT_L(_n)		(GDM_BASE(_n) + 0x12c)
+#define REG_FE_GDM_TX_ETH_L64_CNT_L(_n)		(GDM_BASE(_n) + 0x130)
+#define REG_FE_GDM_TX_ETH_L127_CNT_L(_n)	(GDM_BASE(_n) + 0x134)
+#define REG_FE_GDM_TX_ETH_L255_CNT_L(_n)	(GDM_BASE(_n) + 0x138)
+#define REG_FE_GDM_TX_ETH_L511_CNT_L(_n)	(GDM_BASE(_n) + 0x13c)
+#define REG_FE_GDM_TX_ETH_L1023_CNT_L(_n)	(GDM_BASE(_n) + 0x140)
+
+#define REG_FE_GDM_RX_OK_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x148)
+#define REG_FE_GDM_RX_FC_DROP_CNT(_n)		(GDM_BASE(_n) + 0x14c)
+#define REG_FE_GDM_RX_RC_DROP_CNT(_n)		(GDM_BASE(_n) + 0x150)
+#define REG_FE_GDM_RX_OVERFLOW_DROP_CNT(_n)	(GDM_BASE(_n) + 0x154)
+#define REG_FE_GDM_RX_ERROR_DROP_CNT(_n)	(GDM_BASE(_n) + 0x158)
+#define REG_FE_GDM_RX_OK_BYTE_CNT_L(_n)		(GDM_BASE(_n) + 0x15c)
+#define REG_FE_GDM_RX_ETH_PKT_CNT_L(_n)		(GDM_BASE(_n) + 0x160)
+#define REG_FE_GDM_RX_ETH_BYTE_CNT_L(_n)	(GDM_BASE(_n) + 0x164)
+#define REG_FE_GDM_RX_ETH_DROP_CNT(_n)		(GDM_BASE(_n) + 0x168)
+#define REG_FE_GDM_RX_ETH_BC_CNT(_n)		(GDM_BASE(_n) + 0x16c)
+#define REG_FE_GDM_RX_ETH_MC_CNT(_n)		(GDM_BASE(_n) + 0x170)
+#define REG_FE_GDM_RX_ETH_CRC_ERR_CNT(_n)	(GDM_BASE(_n) + 0x174)
+#define REG_FE_GDM_RX_ETH_FRAG_CNT(_n)		(GDM_BASE(_n) + 0x178)
+#define REG_FE_GDM_RX_ETH_JABBER_CNT(_n)	(GDM_BASE(_n) + 0x17c)
+#define REG_FE_GDM_RX_ETH_RUNT_CNT(_n)		(GDM_BASE(_n) + 0x180)
+#define REG_FE_GDM_RX_ETH_LONG_CNT(_n)		(GDM_BASE(_n) + 0x184)
+#define REG_FE_GDM_RX_ETH_E64_CNT_L(_n)		(GDM_BASE(_n) + 0x188)
+#define REG_FE_GDM_RX_ETH_L64_CNT_L(_n)		(GDM_BASE(_n) + 0x18c)
+#define REG_FE_GDM_RX_ETH_L127_CNT_L(_n)	(GDM_BASE(_n) + 0x190)
+#define REG_FE_GDM_RX_ETH_L255_CNT_L(_n)	(GDM_BASE(_n) + 0x194)
+#define REG_FE_GDM_RX_ETH_L511_CNT_L(_n)	(GDM_BASE(_n) + 0x198)
+#define REG_FE_GDM_RX_ETH_L1023_CNT_L(_n)	(GDM_BASE(_n) + 0x19c)
+
+#define REG_PPE1_TB_HASH_CFG		(PPE1_BASE + 0x250)
+#define PPE1_SRAM_TABLE_EN_MASK		BIT(0)
+#define PPE1_SRAM_HASH1_EN_MASK		BIT(8)
+#define PPE1_DRAM_TABLE_EN_MASK		BIT(16)
+#define PPE1_DRAM_HASH1_EN_MASK		BIT(24)
+
+#define REG_FE_GDM_TX_OK_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x280)
+#define REG_FE_GDM_TX_OK_BYTE_CNT_H(_n)		(GDM_BASE(_n) + 0x284)
+#define REG_FE_GDM_TX_ETH_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x288)
+#define REG_FE_GDM_TX_ETH_BYTE_CNT_H(_n)	(GDM_BASE(_n) + 0x28c)
+
+#define REG_FE_GDM_RX_OK_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x290)
+#define REG_FE_GDM_RX_OK_BYTE_CNT_H(_n)		(GDM_BASE(_n) + 0x294)
+#define REG_FE_GDM_RX_ETH_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x298)
+#define REG_FE_GDM_RX_ETH_BYTE_CNT_H(_n)	(GDM_BASE(_n) + 0x29c)
+#define REG_FE_GDM_TX_ETH_E64_CNT_H(_n)		(GDM_BASE(_n) + 0x2b8)
+#define REG_FE_GDM_TX_ETH_L64_CNT_H(_n)		(GDM_BASE(_n) + 0x2bc)
+#define REG_FE_GDM_TX_ETH_L127_CNT_H(_n)	(GDM_BASE(_n) + 0x2c0)
+#define REG_FE_GDM_TX_ETH_L255_CNT_H(_n)	(GDM_BASE(_n) + 0x2c4)
+#define REG_FE_GDM_TX_ETH_L511_CNT_H(_n)	(GDM_BASE(_n) + 0x2c8)
+#define REG_FE_GDM_TX_ETH_L1023_CNT_H(_n)	(GDM_BASE(_n) + 0x2cc)
+#define REG_FE_GDM_RX_ETH_E64_CNT_H(_n)		(GDM_BASE(_n) + 0x2e8)
+#define REG_FE_GDM_RX_ETH_L64_CNT_H(_n)		(GDM_BASE(_n) + 0x2ec)
+#define REG_FE_GDM_RX_ETH_L127_CNT_H(_n)	(GDM_BASE(_n) + 0x2f0)
+#define REG_FE_GDM_RX_ETH_L255_CNT_H(_n)	(GDM_BASE(_n) + 0x2f4)
+#define REG_FE_GDM_RX_ETH_L511_CNT_H(_n)	(GDM_BASE(_n) + 0x2f8)
+#define REG_FE_GDM_RX_ETH_L1023_CNT_H(_n)	(GDM_BASE(_n) + 0x2fc)
+
+#define REG_GDM2_CHN_RLS		(GDM2_BASE + 0x20)
+#define MBI_RX_AGE_SEL_MASK		GENMASK(26, 25)
+#define MBI_TX_AGE_SEL_MASK		GENMASK(18, 17)
+
+#define REG_GDM3_FWD_CFG		GDM3_BASE
+#define GDM3_PAD_EN_MASK		BIT(28)
+
+#define REG_GDM4_FWD_CFG		GDM4_BASE
+#define GDM4_PAD_EN_MASK		BIT(28)
+#define GDM4_SPORT_OFFSET0_MASK		GENMASK(11, 8)
+
+#define REG_GDM4_SRC_PORT_SET		(GDM4_BASE + 0x23c)
+#define GDM4_SPORT_OFF2_MASK		GENMASK(19, 16)
+#define GDM4_SPORT_OFF1_MASK		GENMASK(15, 12)
+#define GDM4_SPORT_OFF0_MASK		GENMASK(11, 8)
+
+#define REG_IP_FRAG_FP			0x2010
+#define IP_ASSEMBLE_PORT_MASK		GENMASK(24, 21)
+#define IP_ASSEMBLE_NBQ_MASK		GENMASK(20, 16)
+#define IP_FRAGMENT_PORT_MASK		GENMASK(8, 5)
+#define IP_FRAGMENT_NBQ_MASK		GENMASK(4, 0)
+
+#define REG_MC_VLAN_EN			0x2100
+#define MC_VLAN_EN_MASK			BIT(0)
+
+#define REG_MC_VLAN_CFG			0x2104
+#define MC_VLAN_CFG_CMD_DONE_MASK	BIT(31)
+#define MC_VLAN_CFG_TABLE_ID_MASK	GENMASK(21, 16)
+#define MC_VLAN_CFG_PORT_ID_MASK	GENMASK(11, 8)
+#define MC_VLAN_CFG_TABLE_SEL_MASK	BIT(4)
+#define MC_VLAN_CFG_RW_MASK		BIT(0)
+
+#define REG_MC_VLAN_DATA		0x2108
+
+#define REG_CDM5_RX_OQ1_DROP_CNT	0x29d4
+
+/* QDMA */
+#define REG_QDMA_GLOBAL_CFG			0x0004
+#define GLOBAL_CFG_RX_2B_OFFSET_MASK		BIT(31)
+#define GLOBAL_CFG_DMA_PREFERENCE_MASK		GENMASK(30, 29)
+#define GLOBAL_CFG_CPU_TXR_RR_MASK		BIT(28)
+#define GLOBAL_CFG_DSCP_BYTE_SWAP_MASK		BIT(27)
+#define GLOBAL_CFG_PAYLOAD_BYTE_SWAP_MASK	BIT(26)
+#define GLOBAL_CFG_MULTICAST_MODIFY_FP_MASK	BIT(25)
+#define GLOBAL_CFG_OAM_MODIFY_MASK		BIT(24)
+#define GLOBAL_CFG_RESET_MASK			BIT(23)
+#define GLOBAL_CFG_RESET_DONE_MASK		BIT(22)
+#define GLOBAL_CFG_MULTICAST_EN_MASK		BIT(21)
+#define GLOBAL_CFG_IRQ1_EN_MASK			BIT(20)
+#define GLOBAL_CFG_IRQ0_EN_MASK			BIT(19)
+#define GLOBAL_CFG_LOOPCNT_EN_MASK		BIT(18)
+#define GLOBAL_CFG_RD_BYPASS_WR_MASK		BIT(17)
+#define GLOBAL_CFG_QDMA_LOOPBACK_MASK		BIT(16)
+#define GLOBAL_CFG_LPBK_RXQ_SEL_MASK		GENMASK(13, 8)
+#define GLOBAL_CFG_CHECK_DONE_MASK		BIT(7)
+#define GLOBAL_CFG_TX_WB_DONE_MASK		BIT(6)
+#define GLOBAL_CFG_MAX_ISSUE_NUM_MASK		GENMASK(5, 4)
+#define GLOBAL_CFG_RX_DMA_BUSY_MASK		BIT(3)
+#define GLOBAL_CFG_RX_DMA_EN_MASK		BIT(2)
+#define GLOBAL_CFG_TX_DMA_BUSY_MASK		BIT(1)
+#define GLOBAL_CFG_TX_DMA_EN_MASK		BIT(0)
+
+#define REG_FWD_DSCP_BASE			0x0010
+#define REG_FWD_BUF_BASE			0x0014
+
+#define REG_HW_FWD_DSCP_CFG			0x0018
+#define HW_FWD_DSCP_PAYLOAD_SIZE_MASK		GENMASK(29, 28)
+#define HW_FWD_DSCP_SCATTER_LEN_MASK		GENMASK(17, 16)
+#define HW_FWD_DSCP_MIN_SCATTER_LEN_MASK	GENMASK(15, 0)
+
+#define REG_INT_STATUS(_n)		\
+	(((_n) == 4) ? 0x0730 :		\
+	 ((_n) == 3) ? 0x0724 :		\
+	 ((_n) == 2) ? 0x0720 :		\
+	 ((_n) == 1) ? 0x0024 : 0x0020)
+
+#define REG_INT_ENABLE(_n)		\
+	(((_n) == 4) ? 0x0750 :		\
+	 ((_n) == 3) ? 0x0744 :		\
+	 ((_n) == 2) ? 0x0740 :		\
+	 ((_n) == 1) ? 0x002c : 0x0028)
+
+/* QDMA_CSR_INT_ENABLE1 */
+#define RX15_COHERENT_INT_MASK		BIT(31)
+#define RX14_COHERENT_INT_MASK		BIT(30)
+#define RX13_COHERENT_INT_MASK		BIT(29)
+#define RX12_COHERENT_INT_MASK		BIT(28)
+#define RX11_COHERENT_INT_MASK		BIT(27)
+#define RX10_COHERENT_INT_MASK		BIT(26)
+#define RX9_COHERENT_INT_MASK		BIT(25)
+#define RX8_COHERENT_INT_MASK		BIT(24)
+#define RX7_COHERENT_INT_MASK		BIT(23)
+#define RX6_COHERENT_INT_MASK		BIT(22)
+#define RX5_COHERENT_INT_MASK		BIT(21)
+#define RX4_COHERENT_INT_MASK		BIT(20)
+#define RX3_COHERENT_INT_MASK		BIT(19)
+#define RX2_COHERENT_INT_MASK		BIT(18)
+#define RX1_COHERENT_INT_MASK		BIT(17)
+#define RX0_COHERENT_INT_MASK		BIT(16)
+#define TX7_COHERENT_INT_MASK		BIT(15)
+#define TX6_COHERENT_INT_MASK		BIT(14)
+#define TX5_COHERENT_INT_MASK		BIT(13)
+#define TX4_COHERENT_INT_MASK		BIT(12)
+#define TX3_COHERENT_INT_MASK		BIT(11)
+#define TX2_COHERENT_INT_MASK		BIT(10)
+#define TX1_COHERENT_INT_MASK		BIT(9)
+#define TX0_COHERENT_INT_MASK		BIT(8)
+#define CNT_OVER_FLOW_INT_MASK		BIT(7)
+#define IRQ1_FULL_INT_MASK		BIT(5)
+#define IRQ1_INT_MASK			BIT(4)
+#define HWFWD_DSCP_LOW_INT_MASK		BIT(3)
+#define HWFWD_DSCP_EMPTY_INT_MASK	BIT(2)
+#define IRQ0_FULL_INT_MASK		BIT(1)
+#define IRQ0_INT_MASK			BIT(0)
+
+#define TX_DONE_INT_MASK(_n)					\
+	((_n) ? IRQ1_INT_MASK | IRQ1_FULL_INT_MASK		\
+	      : IRQ0_INT_MASK | IRQ0_FULL_INT_MASK)
+
+#define INT_TX_MASK						\
+	(IRQ1_INT_MASK | IRQ1_FULL_INT_MASK |			\
+	 IRQ0_INT_MASK | IRQ0_FULL_INT_MASK)
+
+#define INT_IDX0_MASK						\
+	(TX0_COHERENT_INT_MASK | TX1_COHERENT_INT_MASK |	\
+	 TX2_COHERENT_INT_MASK | TX3_COHERENT_INT_MASK |	\
+	 TX4_COHERENT_INT_MASK | TX5_COHERENT_INT_MASK |	\
+	 TX6_COHERENT_INT_MASK | TX7_COHERENT_INT_MASK |	\
+	 RX0_COHERENT_INT_MASK | RX1_COHERENT_INT_MASK |	\
+	 RX2_COHERENT_INT_MASK | RX3_COHERENT_INT_MASK |	\
+	 RX4_COHERENT_INT_MASK | RX7_COHERENT_INT_MASK |	\
+	 RX8_COHERENT_INT_MASK | RX9_COHERENT_INT_MASK |	\
+	 RX15_COHERENT_INT_MASK | INT_TX_MASK)
+
+/* QDMA_CSR_INT_ENABLE2 */
+#define RX15_NO_CPU_DSCP_INT_MASK	BIT(31)
+#define RX14_NO_CPU_DSCP_INT_MASK	BIT(30)
+#define RX13_NO_CPU_DSCP_INT_MASK	BIT(29)
+#define RX12_NO_CPU_DSCP_INT_MASK	BIT(28)
+#define RX11_NO_CPU_DSCP_INT_MASK	BIT(27)
+#define RX10_NO_CPU_DSCP_INT_MASK	BIT(26)
+#define RX9_NO_CPU_DSCP_INT_MASK	BIT(25)
+#define RX8_NO_CPU_DSCP_INT_MASK	BIT(24)
+#define RX7_NO_CPU_DSCP_INT_MASK	BIT(23)
+#define RX6_NO_CPU_DSCP_INT_MASK	BIT(22)
+#define RX5_NO_CPU_DSCP_INT_MASK	BIT(21)
+#define RX4_NO_CPU_DSCP_INT_MASK	BIT(20)
+#define RX3_NO_CPU_DSCP_INT_MASK	BIT(19)
+#define RX2_NO_CPU_DSCP_INT_MASK	BIT(18)
+#define RX1_NO_CPU_DSCP_INT_MASK	BIT(17)
+#define RX0_NO_CPU_DSCP_INT_MASK	BIT(16)
+#define RX15_DONE_INT_MASK		BIT(15)
+#define RX14_DONE_INT_MASK		BIT(14)
+#define RX13_DONE_INT_MASK		BIT(13)
+#define RX12_DONE_INT_MASK		BIT(12)
+#define RX11_DONE_INT_MASK		BIT(11)
+#define RX10_DONE_INT_MASK		BIT(10)
+#define RX9_DONE_INT_MASK		BIT(9)
+#define RX8_DONE_INT_MASK		BIT(8)
+#define RX7_DONE_INT_MASK		BIT(7)
+#define RX6_DONE_INT_MASK		BIT(6)
+#define RX5_DONE_INT_MASK		BIT(5)
+#define RX4_DONE_INT_MASK		BIT(4)
+#define RX3_DONE_INT_MASK		BIT(3)
+#define RX2_DONE_INT_MASK		BIT(2)
+#define RX1_DONE_INT_MASK		BIT(1)
+#define RX0_DONE_INT_MASK		BIT(0)
+
+#define RX_DONE_INT_MASK					\
+	(RX0_DONE_INT_MASK | RX1_DONE_INT_MASK |		\
+	 RX2_DONE_INT_MASK | RX3_DONE_INT_MASK |		\
+	 RX4_DONE_INT_MASK | RX7_DONE_INT_MASK |		\
+	 RX8_DONE_INT_MASK | RX9_DONE_INT_MASK |		\
+	 RX15_DONE_INT_MASK)
+#define INT_IDX1_MASK						\
+	(RX_DONE_INT_MASK |					\
+	 RX0_NO_CPU_DSCP_INT_MASK | RX1_NO_CPU_DSCP_INT_MASK |	\
+	 RX2_NO_CPU_DSCP_INT_MASK | RX3_NO_CPU_DSCP_INT_MASK |	\
+	 RX4_NO_CPU_DSCP_INT_MASK | RX7_NO_CPU_DSCP_INT_MASK |	\
+	 RX8_NO_CPU_DSCP_INT_MASK | RX9_NO_CPU_DSCP_INT_MASK |	\
+	 RX15_NO_CPU_DSCP_INT_MASK)
+
+/* QDMA_CSR_INT_ENABLE5 */
+#define TX31_COHERENT_INT_MASK		BIT(31)
+#define TX30_COHERENT_INT_MASK		BIT(30)
+#define TX29_COHERENT_INT_MASK		BIT(29)
+#define TX28_COHERENT_INT_MASK		BIT(28)
+#define TX27_COHERENT_INT_MASK		BIT(27)
+#define TX26_COHERENT_INT_MASK		BIT(26)
+#define TX25_COHERENT_INT_MASK		BIT(25)
+#define TX24_COHERENT_INT_MASK		BIT(24)
+#define TX23_COHERENT_INT_MASK		BIT(23)
+#define TX22_COHERENT_INT_MASK		BIT(22)
+#define TX21_COHERENT_INT_MASK		BIT(21)
+#define TX20_COHERENT_INT_MASK		BIT(20)
+#define TX19_COHERENT_INT_MASK		BIT(19)
+#define TX18_COHERENT_INT_MASK		BIT(18)
+#define TX17_COHERENT_INT_MASK		BIT(17)
+#define TX16_COHERENT_INT_MASK		BIT(16)
+#define TX15_COHERENT_INT_MASK		BIT(15)
+#define TX14_COHERENT_INT_MASK		BIT(14)
+#define TX13_COHERENT_INT_MASK		BIT(13)
+#define TX12_COHERENT_INT_MASK		BIT(12)
+#define TX11_COHERENT_INT_MASK		BIT(11)
+#define TX10_COHERENT_INT_MASK		BIT(10)
+#define TX9_COHERENT_INT_MASK		BIT(9)
+#define TX8_COHERENT_INT_MASK		BIT(8)
+
+#define INT_IDX4_MASK						\
+	(TX8_COHERENT_INT_MASK | TX9_COHERENT_INT_MASK |	\
+	 TX10_COHERENT_INT_MASK | TX11_COHERENT_INT_MASK |	\
+	 TX12_COHERENT_INT_MASK | TX13_COHERENT_INT_MASK |	\
+	 TX14_COHERENT_INT_MASK | TX15_COHERENT_INT_MASK |	\
+	 TX16_COHERENT_INT_MASK | TX17_COHERENT_INT_MASK |	\
+	 TX18_COHERENT_INT_MASK | TX19_COHERENT_INT_MASK |	\
+	 TX20_COHERENT_INT_MASK | TX21_COHERENT_INT_MASK |	\
+	 TX22_COHERENT_INT_MASK | TX23_COHERENT_INT_MASK |	\
+	 TX24_COHERENT_INT_MASK | TX25_COHERENT_INT_MASK |	\
+	 TX26_COHERENT_INT_MASK | TX27_COHERENT_INT_MASK |	\
+	 TX28_COHERENT_INT_MASK | TX29_COHERENT_INT_MASK |	\
+	 TX30_COHERENT_INT_MASK | TX31_COHERENT_INT_MASK)
+
+#define REG_TX_IRQ_BASE(_n)		((_n) ? 0x0048 : 0x0050)
+
+#define REG_TX_IRQ_CFG(_n)		((_n) ? 0x004c : 0x0054)
+#define TX_IRQ_THR_MASK			GENMASK(27, 16)
+#define TX_IRQ_DEPTH_MASK		GENMASK(11, 0)
+
+#define REG_IRQ_CLEAR_LEN(_n)		((_n) ? 0x0064 : 0x0058)
+#define IRQ_CLEAR_LEN_MASK		GENMASK(7, 0)
+
+#define REG_IRQ_STATUS(_n)		((_n) ? 0x0068 : 0x005c)
+#define IRQ_ENTRY_LEN_MASK		GENMASK(27, 16)
+#define IRQ_HEAD_IDX_MASK		GENMASK(11, 0)
+
+#define REG_TX_RING_BASE(_n)	\
+	(((_n) < 8) ? 0x0100 + ((_n) << 5) : 0x0b00 + (((_n) - 8) << 5))
+
+#define REG_TX_RING_BLOCKING(_n)	\
+	(((_n) < 8) ? 0x0104 + ((_n) << 5) : 0x0b04 + (((_n) - 8) << 5))
+
+#define TX_RING_IRQ_BLOCKING_MAP_MASK			BIT(6)
+#define TX_RING_IRQ_BLOCKING_CFG_MASK			BIT(4)
+#define TX_RING_IRQ_BLOCKING_TX_DROP_EN_MASK		BIT(2)
+#define TX_RING_IRQ_BLOCKING_MAX_TH_TXRING_EN_MASK	BIT(1)
+#define TX_RING_IRQ_BLOCKING_MIN_TH_TXRING_EN_MASK	BIT(0)
+
+#define REG_TX_CPU_IDX(_n)	\
+	(((_n) < 8) ? 0x0108 + ((_n) << 5) : 0x0b08 + (((_n) - 8) << 5))
+
+#define TX_RING_CPU_IDX_MASK		GENMASK(15, 0)
+
+#define REG_TX_DMA_IDX(_n)	\
+	(((_n) < 8) ? 0x010c + ((_n) << 5) : 0x0b0c + (((_n) - 8) << 5))
+
+#define TX_RING_DMA_IDX_MASK		GENMASK(15, 0)
+
+#define IRQ_RING_IDX_MASK		GENMASK(20, 16)
+#define IRQ_DESC_IDX_MASK		GENMASK(15, 0)
+
+#define REG_RX_RING_BASE(_n)	\
+	(((_n) < 16) ? 0x0200 + ((_n) << 5) : 0x0e00 + (((_n) - 16) << 5))
+
+#define REG_RX_RING_SIZE(_n)	\
+	(((_n) < 16) ? 0x0204 + ((_n) << 5) : 0x0e04 + (((_n) - 16) << 5))
+
+#define RX_RING_THR_MASK		GENMASK(31, 16)
+#define RX_RING_SIZE_MASK		GENMASK(15, 0)
+
+#define REG_RX_CPU_IDX(_n)	\
+	(((_n) < 16) ? 0x0208 + ((_n) << 5) : 0x0e08 + (((_n) - 16) << 5))
+
+#define RX_RING_CPU_IDX_MASK		GENMASK(15, 0)
+
+#define REG_RX_DMA_IDX(_n)	\
+	(((_n) < 16) ? 0x020c + ((_n) << 5) : 0x0e0c + (((_n) - 16) << 5))
+
+#define REG_RX_DELAY_INT_IDX(_n)	\
+	(((_n) < 16) ? 0x0210 + ((_n) << 5) : 0x0e10 + (((_n) - 16) << 5))
+
+#define RX_DELAY_INT_MASK		GENMASK(15, 0)
+
+#define RX_RING_DMA_IDX_MASK		GENMASK(15, 0)
+
+#define REG_INGRESS_TRTCM_CFG		0x0070
+#define INGRESS_TRTCM_EN_MASK		BIT(31)
+#define INGRESS_TRTCM_MODE_MASK		BIT(30)
+#define INGRESS_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
+#define INGRESS_FAST_TICK_MASK		GENMASK(15, 0)
+
+#define REG_QUEUE_CLOSE_CFG(_n)		(0x00a0 + ((_n) & 0xfc))
+#define TXQ_DISABLE_CHAN_QUEUE_MASK(_n, _m)	BIT((_m) + (((_n) & 0x3) << 3))
+
+#define REG_TXQ_DIS_CFG_BASE(_n)	((_n) ? 0x20a0 : 0x00a0)
+#define REG_TXQ_DIS_CFG(_n, _m)		(REG_TXQ_DIS_CFG_BASE((_n)) + (_m) << 2)
+
+#define REG_CNTR_CFG(_n)		(0x0400 + ((_n) << 3))
+#define CNTR_EN_MASK			BIT(31)
+#define CNTR_ALL_CHAN_EN_MASK		BIT(30)
+#define CNTR_ALL_QUEUE_EN_MASK		BIT(29)
+#define CNTR_ALL_DSCP_RING_EN_MASK	BIT(28)
+#define CNTR_SRC_MASK			GENMASK(27, 24)
+#define CNTR_DSCP_RING_MASK		GENMASK(20, 16)
+#define CNTR_CHAN_MASK			GENMASK(7, 3)
+#define CNTR_QUEUE_MASK			GENMASK(2, 0)
+
+#define REG_CNTR_VAL(_n)		(0x0404 + ((_n) << 3))
+
+#define REG_LMGR_INIT_CFG		0x1000
+#define LMGR_INIT_START			BIT(31)
+#define LMGR_SRAM_MODE_MASK		BIT(30)
+#define HW_FWD_PKTSIZE_OVERHEAD_MASK	GENMASK(27, 20)
+#define HW_FWD_DESC_NUM_MASK		GENMASK(16, 0)
+
+#define REG_FWD_DSCP_LOW_THR		0x1004
+#define FWD_DSCP_LOW_THR_MASK		GENMASK(17, 0)
+
+#define REG_EGRESS_RATE_METER_CFG		0x100c
+#define EGRESS_RATE_METER_EN_MASK		BIT(31)
+#define EGRESS_RATE_METER_EQ_RATE_EN_MASK	BIT(17)
+#define EGRESS_RATE_METER_WINDOW_SZ_MASK	GENMASK(16, 12)
+#define EGRESS_RATE_METER_TIMESLICE_MASK	GENMASK(10, 0)
+
+#define REG_EGRESS_TRTCM_CFG		0x1010
+#define EGRESS_TRTCM_EN_MASK		BIT(31)
+#define EGRESS_TRTCM_MODE_MASK		BIT(30)
+#define EGRESS_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
+#define EGRESS_FAST_TICK_MASK		GENMASK(15, 0)
+
+#define TRTCM_PARAM_RW_MASK		BIT(31)
+#define TRTCM_PARAM_RW_DONE_MASK	BIT(30)
+#define TRTCM_PARAM_TYPE_MASK		GENMASK(29, 28)
+#define TRTCM_METER_GROUP_MASK		GENMASK(27, 26)
+#define TRTCM_PARAM_INDEX_MASK		GENMASK(23, 17)
+#define TRTCM_PARAM_RATE_TYPE_MASK	BIT(16)
+
+#define REG_TRTCM_CFG_PARAM(_n)		((_n) + 0x4)
+#define REG_TRTCM_DATA_LOW(_n)		((_n) + 0x8)
+#define REG_TRTCM_DATA_HIGH(_n)		((_n) + 0xc)
+
+#define REG_TXWRR_MODE_CFG		0x1020
+#define TWRR_WEIGHT_SCALE_MASK		BIT(31)
+#define TWRR_WEIGHT_BASE_MASK		BIT(3)
+
+#define REG_TXWRR_WEIGHT_CFG		0x1024
+#define TWRR_RW_CMD_MASK		BIT(31)
+#define TWRR_RW_CMD_DONE		BIT(30)
+#define TWRR_CHAN_IDX_MASK		GENMASK(23, 19)
+#define TWRR_QUEUE_IDX_MASK		GENMASK(18, 16)
+#define TWRR_VALUE_MASK			GENMASK(15, 0)
+
+#define REG_PSE_BUF_USAGE_CFG		0x1028
+#define PSE_BUF_ESTIMATE_EN_MASK	BIT(29)
+
+#define REG_CHAN_QOS_MODE(_n)		(0x1040 + ((_n) << 2))
+#define CHAN_QOS_MODE_MASK(_n)		GENMASK(2 + ((_n) << 2), (_n) << 2)
+
+#define REG_GLB_TRTCM_CFG		0x1080
+#define GLB_TRTCM_EN_MASK		BIT(31)
+#define GLB_TRTCM_MODE_MASK		BIT(30)
+#define GLB_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
+#define GLB_FAST_TICK_MASK		GENMASK(15, 0)
+
+#define REG_TXQ_CNGST_CFG		0x10a0
+#define TXQ_CNGST_DROP_EN		BIT(31)
+#define TXQ_CNGST_DEI_DROP_EN		BIT(30)
+
+#define REG_SLA_TRTCM_CFG		0x1150
+#define SLA_TRTCM_EN_MASK		BIT(31)
+#define SLA_TRTCM_MODE_MASK		BIT(30)
+#define SLA_SLOW_TICK_RATIO_MASK	GENMASK(29, 16)
+#define SLA_FAST_TICK_MASK		GENMASK(15, 0)
+
+/* CTRL */
+#define QDMA_DESC_DONE_MASK		BIT(31)
+#define QDMA_DESC_DROP_MASK		BIT(30) /* tx: drop - rx: overflow */
+#define QDMA_DESC_MORE_MASK		BIT(29) /* more SG elements */
+#define QDMA_DESC_DEI_MASK		BIT(25)
+#define QDMA_DESC_NO_DROP_MASK		BIT(24)
+#define QDMA_DESC_LEN_MASK		GENMASK(15, 0)
+/* DATA */
+#define QDMA_DESC_NEXT_ID_MASK		GENMASK(15, 0)
+/* TX MSG0 */
+#define QDMA_ETH_TXMSG_MIC_IDX_MASK	BIT(30)
+#define QDMA_ETH_TXMSG_SP_TAG_MASK	GENMASK(29, 14)
+#define QDMA_ETH_TXMSG_ICO_MASK		BIT(13)
+#define QDMA_ETH_TXMSG_UCO_MASK		BIT(12)
+#define QDMA_ETH_TXMSG_TCO_MASK		BIT(11)
+#define QDMA_ETH_TXMSG_TSO_MASK		BIT(10)
+#define QDMA_ETH_TXMSG_FAST_MASK	BIT(9)
+#define QDMA_ETH_TXMSG_OAM_MASK		BIT(8)
+#define QDMA_ETH_TXMSG_CHAN_MASK	GENMASK(7, 3)
+#define QDMA_ETH_TXMSG_QUEUE_MASK	GENMASK(2, 0)
+/* TX MSG1 */
+#define QDMA_ETH_TXMSG_NO_DROP		BIT(31)
+#define QDMA_ETH_TXMSG_METER_MASK	GENMASK(30, 24)	/* 0x7f no meters */
+#define QDMA_ETH_TXMSG_FPORT_MASK	GENMASK(23, 20)
+#define QDMA_ETH_TXMSG_NBOQ_MASK	GENMASK(19, 15)
+#define QDMA_ETH_TXMSG_HWF_MASK		BIT(14)
+#define QDMA_ETH_TXMSG_HOP_MASK		BIT(13)
+#define QDMA_ETH_TXMSG_PTP_MASK		BIT(12)
+#define QDMA_ETH_TXMSG_ACNT_G1_MASK	GENMASK(10, 6)	/* 0x1f do not count */
+#define QDMA_ETH_TXMSG_ACNT_G0_MASK	GENMASK(5, 0)	/* 0x3f do not count */
+
+/* RX MSG1 */
+#define QDMA_ETH_RXMSG_DEI_MASK		BIT(31)
+#define QDMA_ETH_RXMSG_IP6_MASK		BIT(30)
+#define QDMA_ETH_RXMSG_IP4_MASK		BIT(29)
+#define QDMA_ETH_RXMSG_IP4F_MASK	BIT(28)
+#define QDMA_ETH_RXMSG_L4_VALID_MASK	BIT(27)
+#define QDMA_ETH_RXMSG_L4F_MASK		BIT(26)
+#define QDMA_ETH_RXMSG_SPORT_MASK	GENMASK(25, 21)
+#define QDMA_ETH_RXMSG_CRSN_MASK	GENMASK(20, 16)
+#define QDMA_ETH_RXMSG_PPE_ENTRY_MASK	GENMASK(15, 0)
+
+struct airoha_qdma_desc {
+	__le32 rsv;
+	__le32 ctrl;
+	__le32 addr;
+	__le32 data;
+	__le32 msg0;
+	__le32 msg1;
+	__le32 msg2;
+	__le32 msg3;
+};
+
+/* CTRL0 */
+#define QDMA_FWD_DESC_CTX_MASK		BIT(31)
+#define QDMA_FWD_DESC_RING_MASK		GENMASK(30, 28)
+#define QDMA_FWD_DESC_IDX_MASK		GENMASK(27, 16)
+#define QDMA_FWD_DESC_LEN_MASK		GENMASK(15, 0)
+/* CTRL1 */
+#define QDMA_FWD_DESC_FIRST_IDX_MASK	GENMASK(15, 0)
+/* CTRL2 */
+#define QDMA_FWD_DESC_MORE_PKT_NUM_MASK	GENMASK(2, 0)
+
+struct airoha_qdma_fwd_desc {
+	__le32 addr;
+	__le32 ctrl0;
+	__le32 ctrl1;
+	__le32 ctrl2;
+	__le32 msg0;
+	__le32 msg1;
+	__le32 rsv0;
+	__le32 rsv1;
+};
+
+#endif /* AIROHA_REGS_H */

-- 
2.48.1

[PATCH net-next v3 06/16] net: airoha: Move DSA tag in DMA descriptor

[Lorenzo Bianconi]

Packet Processor Engine (PPE) module reads DSA tags from the DMA descriptor
and requires untagged DSA packets to properly parse them. Move DSA tag
in the DMA descriptor on TX side and read DSA tag from DMA descriptor
on RX side. In order to avoid skb reallocation, store tag in skb_dst on
RX side.
This is a preliminary patch to enable netfilter flowtable hw offloading
on EN7581 SoC.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c  | 123 ++++++++++++++++++++++++++++--
 drivers/net/ethernet/airoha/airoha_eth.h  |   7 ++
 drivers/net/ethernet/airoha/airoha_regs.h |   2 +
 3 files changed, 126 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index b79556f1b4951c687aa89bc5839fc9405581a6c3..4f45db86d8d8d6b7a13d56a9315773f8685a09f6 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -9,6 +9,7 @@
 #include <linux/tcp.h>
 #include <linux/u64_stats_sync.h>
 #include <net/dsa.h>
+#include <net/dst_metadata.h>
 #include <net/page_pool/helpers.h>
 #include <net/pkt_cls.h>
 #include <uapi/linux/ppp_defs.h>
@@ -656,6 +657,7 @@ static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
 		struct airoha_qdma_desc *desc = &q->desc[q->tail];
 		dma_addr_t dma_addr = le32_to_cpu(desc->addr);
 		u32 desc_ctrl = le32_to_cpu(desc->ctrl);
+		struct airoha_gdm_port *port;
 		struct sk_buff *skb;
 		int len, p;
 
@@ -683,6 +685,7 @@ static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
 			continue;
 		}
 
+		port = eth->ports[p];
 		skb = napi_build_skb(e->buf, q->buf_size);
 		if (!skb) {
 			page_pool_put_full_page(q->page_pool,
@@ -694,10 +697,26 @@ static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
 		skb_reserve(skb, 2);
 		__skb_put(skb, len);
 		skb_mark_for_recycle(skb);
-		skb->dev = eth->ports[p]->dev;
+		skb->dev = port->dev;
 		skb->protocol = eth_type_trans(skb, skb->dev);
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 		skb_record_rx_queue(skb, qid);
+
+		if (netdev_uses_dsa(port->dev)) {
+			/* PPE module requires untagged packets to work
+			 * properly and it provides DSA port index via the
+			 * DMA descriptor. Report DSA tag to the DSA stack
+			 * via skb dst info.
+			 */
+			u32 sptag = FIELD_GET(QDMA_ETH_RXMSG_SPTAG,
+					      le32_to_cpu(desc->msg0));
+
+			if (sptag < ARRAY_SIZE(port->dsa_meta) &&
+			    port->dsa_meta[sptag])
+				skb_dst_set_noref(skb,
+						  &port->dsa_meta[sptag]->dst);
+		}
+
 		napi_gro_receive(&q->napi, skb);
 
 		done++;
@@ -1636,25 +1655,74 @@ static u16 airoha_dev_select_queue(struct net_device *dev, struct sk_buff *skb,
 	return queue < dev->num_tx_queues ? queue : 0;
 }
 
+static u32 airoha_get_dsa_tag(struct sk_buff *skb, struct net_device *dev)
+{
+#if IS_ENABLED(CONFIG_NET_DSA)
+	struct ethhdr *ehdr;
+	struct dsa_port *dp;
+	u8 xmit_tpid;
+	u16 tag;
+
+	if (!netdev_uses_dsa(dev))
+		return 0;
+
+	dp = dev->dsa_ptr;
+	if (IS_ERR(dp))
+		return 0;
+
+	if (dp->tag_ops->proto != DSA_TAG_PROTO_MTK)
+		return 0;
+
+	if (skb_ensure_writable(skb, ETH_HLEN))
+		return 0;
+
+	ehdr = (struct ethhdr *)skb->data;
+	tag = be16_to_cpu(ehdr->h_proto);
+	xmit_tpid = tag >> 8;
+
+	switch (xmit_tpid) {
+	case MTK_HDR_XMIT_TAGGED_TPID_8100:
+		ehdr->h_proto = cpu_to_be16(ETH_P_8021Q);
+		break;
+	case MTK_HDR_XMIT_TAGGED_TPID_88A8:
+		ehdr->h_proto = cpu_to_be16(ETH_P_8021AD);
+		break;
+	default:
+		/* PPE module requires untagged DSA packets to work properly,
+		 * so move DSA tag to DMA descriptor.
+		 */
+		memmove(skb->data + MTK_HDR_LEN, skb->data, 2 * ETH_ALEN);
+		__skb_pull(skb, MTK_HDR_LEN);
+		break;
+	}
+
+	return tag;
+#else
+	return 0;
+#endif
+}
+
 static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 				   struct net_device *dev)
 {
 	struct airoha_gdm_port *port = netdev_priv(dev);
-	u32 nr_frags = 1 + skb_shinfo(skb)->nr_frags;
-	u32 msg0, msg1, len = skb_headlen(skb);
 	struct airoha_qdma *qdma = port->qdma;
+	u32 nr_frags, tag, msg0, msg1, len;
 	struct netdev_queue *txq;
 	struct airoha_queue *q;
-	void *data = skb->data;
+	void *data;
 	int i, qid;
 	u16 index;
 	u8 fport;
 
 	qid = skb_get_queue_mapping(skb) % ARRAY_SIZE(qdma->q_tx);
+	tag = airoha_get_dsa_tag(skb, dev);
+
 	msg0 = FIELD_PREP(QDMA_ETH_TXMSG_CHAN_MASK,
 			  qid / AIROHA_NUM_QOS_QUEUES) |
 	       FIELD_PREP(QDMA_ETH_TXMSG_QUEUE_MASK,
-			  qid % AIROHA_NUM_QOS_QUEUES);
+			  qid % AIROHA_NUM_QOS_QUEUES) |
+	       FIELD_PREP(QDMA_ETH_TXMSG_SP_TAG_MASK, tag);
 	if (skb->ip_summed == CHECKSUM_PARTIAL)
 		msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TCO_MASK, 1) |
 			FIELD_PREP(QDMA_ETH_TXMSG_UCO_MASK, 1) |
@@ -1685,6 +1753,8 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 	spin_lock_bh(&q->lock);
 
 	txq = netdev_get_tx_queue(dev, qid);
+	nr_frags = 1 + skb_shinfo(skb)->nr_frags;
+
 	if (q->queued + nr_frags > q->ndesc) {
 		/* not enough space in the queue */
 		netif_tx_stop_queue(txq);
@@ -1692,7 +1762,10 @@ static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
 		return NETDEV_TX_BUSY;
 	}
 
+	len = skb_headlen(skb);
+	data = skb->data;
 	index = q->head;
+
 	for (i = 0; i < nr_frags; i++) {
 		struct airoha_qdma_desc *desc = &q->desc[index];
 		struct airoha_queue_entry *e = &q->entry[index];
@@ -2226,6 +2299,37 @@ static const struct ethtool_ops airoha_ethtool_ops = {
 	.get_rmon_stats		= airoha_ethtool_get_rmon_stats,
 };
 
+static int airoha_metadata_dst_alloc(struct airoha_gdm_port *port)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(port->dsa_meta); i++) {
+		struct metadata_dst *md_dst;
+
+		md_dst = metadata_dst_alloc(0, METADATA_HW_PORT_MUX,
+					    GFP_KERNEL);
+		if (!md_dst)
+			return -ENOMEM;
+
+		md_dst->u.port_info.port_id = i;
+		port->dsa_meta[i] = md_dst;
+	}
+
+	return 0;
+}
+
+static void airoha_metadata_dst_free(struct airoha_gdm_port *port)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(port->dsa_meta); i++) {
+		if (!port->dsa_meta[i])
+			continue;
+
+		metadata_dst_free(port->dsa_meta[i]);
+	}
+}
+
 static int airoha_alloc_gdm_port(struct airoha_eth *eth, struct device_node *np)
 {
 	const __be32 *id_ptr = of_get_property(np, "reg", NULL);
@@ -2298,6 +2402,10 @@ static int airoha_alloc_gdm_port(struct airoha_eth *eth, struct device_node *np)
 	port->id = id;
 	eth->ports[index] = port;
 
+	err = airoha_metadata_dst_alloc(port);
+	if (err)
+		return err;
+
 	return register_netdev(dev);
 }
 
@@ -2390,8 +2498,10 @@ static int airoha_probe(struct platform_device *pdev)
 	for (i = 0; i < ARRAY_SIZE(eth->ports); i++) {
 		struct airoha_gdm_port *port = eth->ports[i];
 
-		if (port && port->dev->reg_state == NETREG_REGISTERED)
+		if (port && port->dev->reg_state == NETREG_REGISTERED) {
+			airoha_metadata_dst_free(port);
 			unregister_netdev(port->dev);
+		}
 	}
 	free_netdev(eth->napi_dev);
 	platform_set_drvdata(pdev, NULL);
@@ -2416,6 +2526,7 @@ static void airoha_remove(struct platform_device *pdev)
 			continue;
 
 		airoha_dev_stop(port->dev);
+		airoha_metadata_dst_free(port);
 		unregister_netdev(port->dev);
 	}
 	free_netdev(eth->napi_dev);
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index 743aaf10235fe09fb2a91b491f4b25064ed8319b..fee6c10eaedfd30207205b6557e856091fd45d7e 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -15,6 +15,7 @@
 
 #define AIROHA_MAX_NUM_GDM_PORTS	1
 #define AIROHA_MAX_NUM_QDMA		2
+#define AIROHA_MAX_DSA_PORTS		7
 #define AIROHA_MAX_NUM_RSTS		3
 #define AIROHA_MAX_NUM_XSI_RSTS		5
 #define AIROHA_MAX_MTU			2000
@@ -43,6 +44,10 @@
 #define QDMA_METER_IDX(_n)		((_n) & 0xff)
 #define QDMA_METER_GROUP(_n)		(((_n) >> 8) & 0x3)
 
+#define MTK_HDR_LEN			4
+#define MTK_HDR_XMIT_TAGGED_TPID_8100	1
+#define MTK_HDR_XMIT_TAGGED_TPID_88A8	2
+
 enum {
 	QDMA_INT_REG_IDX0,
 	QDMA_INT_REG_IDX1,
@@ -231,6 +236,8 @@ struct airoha_gdm_port {
 	/* qos stats counters */
 	u64 cpu_tx_packets;
 	u64 fwd_tx_packets;
+
+	struct metadata_dst *dsa_meta[AIROHA_MAX_DSA_PORTS];
 };
 
 struct airoha_eth {
diff --git a/drivers/net/ethernet/airoha/airoha_regs.h b/drivers/net/ethernet/airoha/airoha_regs.h
index 7c9dadb348834cb5a856760abe45e8221d6fd700..e467dd81ff44a9ad560226cab42b7431812f5fb9 100644
--- a/drivers/net/ethernet/airoha/airoha_regs.h
+++ b/drivers/net/ethernet/airoha/airoha_regs.h
@@ -624,6 +624,8 @@
 #define QDMA_ETH_TXMSG_ACNT_G1_MASK	GENMASK(10, 6)	/* 0x1f do not count */
 #define QDMA_ETH_TXMSG_ACNT_G0_MASK	GENMASK(5, 0)	/* 0x3f do not count */
 
+/* RX MSG0 */
+#define QDMA_ETH_RXMSG_SPTAG		GENMASK(21, 14)
 /* RX MSG1 */
 #define QDMA_ETH_RXMSG_DEI_MASK		BIT(31)
 #define QDMA_ETH_RXMSG_IP6_MASK		BIT(30)

-- 
2.48.1

[PATCH net-next v3 07/16] net: dsa: mt7530: Enable Rx sptag for EN7581 SoC

[Lorenzo Bianconi]

Packet Processor Engine (PPE) module used for hw acceleration on EN7581
mac block, in order to properly parse packets, requires DSA untagged
packets on TX side and read DSA tag from DMA descriptor on RX side.
For this reason, enable RX Special Tag (SPTAG) for EN7581 SoC.
This is a preliminary patch to enable netfilter flowtable hw offloading
on EN7581 SoC.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/dsa/mt7530.c | 5 +++++
 drivers/net/dsa/mt7530.h | 4 ++++
 2 files changed, 9 insertions(+)

diff --git a/drivers/net/dsa/mt7530.c b/drivers/net/dsa/mt7530.c
index 1c83af805209cae40c56138fa8f72261e396f58c..eec8ba9d68088f1dbb2a774a32d3d60af1a9784c 100644
--- a/drivers/net/dsa/mt7530.c
+++ b/drivers/net/dsa/mt7530.c
@@ -2586,6 +2586,11 @@ mt7531_setup_common(struct dsa_switch *ds)
 	/* Allow mirroring frames received on the local port (monitor port). */
 	mt7530_set(priv, MT753X_AGC, LOCAL_EN);
 
+	/* Enable Special Tag for rx frames */
+	if (priv->id == ID_EN7581)
+		mt7530_write(priv, MT753X_CPORT_SPTAG_CFG,
+			     CPORT_SW2FE_STAG_EN | CPORT_FE2SW_STAG_EN);
+
 	/* Flush the FDB table */
 	ret = mt7530_fdb_cmd(priv, MT7530_FDB_FLUSH, NULL);
 	if (ret < 0)
diff --git a/drivers/net/dsa/mt7530.h b/drivers/net/dsa/mt7530.h
index 448200689f492dcb73ef056d7284090c1c662e67..349d72a35771f35d478244ab29be1801b3466a5f 100644
--- a/drivers/net/dsa/mt7530.h
+++ b/drivers/net/dsa/mt7530.h
@@ -627,6 +627,10 @@ enum mt7531_xtal_fsel {
 #define  MT7531_GPIO12_RG_RXD3_MASK	GENMASK(19, 16)
 #define  MT7531_EXT_P_MDIO_12		(2 << 16)
 
+#define MT753X_CPORT_SPTAG_CFG		0x7c10
+#define  CPORT_SW2FE_STAG_EN		BIT(1)
+#define  CPORT_FE2SW_STAG_EN		BIT(0)
+
 /* Registers for LED GPIO control (MT7530 only)
  * All registers follow this pattern:
  * [ 2: 0]  port 0

-- 
2.48.1

[PATCH net-next v3 08/16] net: airoha: Enable support for multiple net_devices

[Lorenzo Bianconi]

In the current codebase airoha_eth driver supports just a single
net_device connected to the Packet Switch Engine (PSE) lan port (GDM1).
As shown in commit 23020f049327 ("net: airoha: Introduce ethernet
support for EN7581 SoC"), PSE can switch packets between four GDM ports.
Enable the capability to create a net_device for each GDM port of the
PSE module. Moreover, since the QDMA blocks can be shared between
net_devices, do not stop TX/RX DMA in airoha_dev_stop() if there are
active net_devices for this QDMA block.
This is a preliminary patch to enable flowtable hw offloading for EN7581
SoC.

Co-developed-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 35 +++++++++++++++++++-------------
 drivers/net/ethernet/airoha/airoha_eth.h |  4 +++-
 2 files changed, 24 insertions(+), 15 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 4f45db86d8d8d6b7a13d56a9315773f8685a09f6..513914da8503c1162b0f1b4fcca57434385fa4d1 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -1562,6 +1562,7 @@ static int airoha_dev_open(struct net_device *dev)
 	airoha_qdma_set(qdma, REG_QDMA_GLOBAL_CFG,
 			GLOBAL_CFG_TX_DMA_EN_MASK |
 			GLOBAL_CFG_RX_DMA_EN_MASK);
+	atomic_inc(&qdma->users);
 
 	return 0;
 }
@@ -1577,16 +1578,20 @@ static int airoha_dev_stop(struct net_device *dev)
 	if (err)
 		return err;
 
-	airoha_qdma_clear(qdma, REG_QDMA_GLOBAL_CFG,
-			  GLOBAL_CFG_TX_DMA_EN_MASK |
-			  GLOBAL_CFG_RX_DMA_EN_MASK);
+	for (i = 0; i < ARRAY_SIZE(qdma->q_tx); i++)
+		netdev_tx_reset_subqueue(dev, i);
 
-	for (i = 0; i < ARRAY_SIZE(qdma->q_tx); i++) {
-		if (!qdma->q_tx[i].ndesc)
-			continue;
+	if (atomic_dec_and_test(&qdma->users)) {
+		airoha_qdma_clear(qdma, REG_QDMA_GLOBAL_CFG,
+				  GLOBAL_CFG_TX_DMA_EN_MASK |
+				  GLOBAL_CFG_RX_DMA_EN_MASK);
 
-		airoha_qdma_cleanup_tx_queue(&qdma->q_tx[i]);
-		netdev_tx_reset_subqueue(dev, i);
+		for (i = 0; i < ARRAY_SIZE(qdma->q_tx); i++) {
+			if (!qdma->q_tx[i].ndesc)
+				continue;
+
+			airoha_qdma_cleanup_tx_queue(&qdma->q_tx[i]);
+		}
 	}
 
 	return 0;
@@ -2330,13 +2335,14 @@ static void airoha_metadata_dst_free(struct airoha_gdm_port *port)
 	}
 }
 
-static int airoha_alloc_gdm_port(struct airoha_eth *eth, struct device_node *np)
+static int airoha_alloc_gdm_port(struct airoha_eth *eth,
+				 struct device_node *np, int index)
 {
 	const __be32 *id_ptr = of_get_property(np, "reg", NULL);
 	struct airoha_gdm_port *port;
 	struct airoha_qdma *qdma;
 	struct net_device *dev;
-	int err, index;
+	int err, p;
 	u32 id;
 
 	if (!id_ptr) {
@@ -2345,14 +2351,14 @@ static int airoha_alloc_gdm_port(struct airoha_eth *eth, struct device_node *np)
 	}
 
 	id = be32_to_cpup(id_ptr);
-	index = id - 1;
+	p = id - 1;
 
 	if (!id || id > ARRAY_SIZE(eth->ports)) {
 		dev_err(eth->dev, "invalid gdm port id: %d\n", id);
 		return -EINVAL;
 	}
 
-	if (eth->ports[index]) {
+	if (eth->ports[p]) {
 		dev_err(eth->dev, "duplicate gdm port id: %d\n", id);
 		return -EINVAL;
 	}
@@ -2400,7 +2406,7 @@ static int airoha_alloc_gdm_port(struct airoha_eth *eth, struct device_node *np)
 	port->qdma = qdma;
 	port->dev = dev;
 	port->id = id;
-	eth->ports[index] = port;
+	eth->ports[p] = port;
 
 	err = airoha_metadata_dst_alloc(port);
 	if (err)
@@ -2472,6 +2478,7 @@ static int airoha_probe(struct platform_device *pdev)
 	for (i = 0; i < ARRAY_SIZE(eth->qdma); i++)
 		airoha_qdma_start_napi(&eth->qdma[i]);
 
+	i = 0;
 	for_each_child_of_node(pdev->dev.of_node, np) {
 		if (!of_device_is_compatible(np, "airoha,eth-mac"))
 			continue;
@@ -2479,7 +2486,7 @@ static int airoha_probe(struct platform_device *pdev)
 		if (!of_device_is_available(np))
 			continue;
 
-		err = airoha_alloc_gdm_port(eth, np);
+		err = airoha_alloc_gdm_port(eth, np, i++);
 		if (err) {
 			of_node_put(np);
 			goto error_napi_stop;
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index fee6c10eaedfd30207205b6557e856091fd45d7e..74bdfd9e8d2fb3706f5ec6a4e17fe07fbcb38c3d 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -13,7 +13,7 @@
 #include <linux/netdevice.h>
 #include <linux/reset.h>
 
-#define AIROHA_MAX_NUM_GDM_PORTS	1
+#define AIROHA_MAX_NUM_GDM_PORTS	4
 #define AIROHA_MAX_NUM_QDMA		2
 #define AIROHA_MAX_DSA_PORTS		7
 #define AIROHA_MAX_NUM_RSTS		3
@@ -212,6 +212,8 @@ struct airoha_qdma {
 	u32 irqmask[QDMA_INT_REG_MAX];
 	int irq;
 
+	atomic_t users;
+
 	struct airoha_tx_irq_queue q_tx_irq[AIROHA_NUM_TX_IRQ];
 
 	struct airoha_queue q_tx[AIROHA_NUM_TX_RING];

-- 
2.48.1

[PATCH net-next v3 09/16] net: airoha: Move REG_GDM_FWD_CFG() initialization in airoha_dev_init()

[Lorenzo Bianconi]

Move REG_GDM_FWD_CFG() register initialization in airoha_dev_init
routine. Moreover, always send traffic PPE module in order to be
processed by hw accelerator.
This is a preliminary patch to enable netfilter flowtable hw offloading
on EN7581 SoC.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 14 ++++----------
 1 file changed, 4 insertions(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 513914da8503c1162b0f1b4fcca57434385fa4d1..6c899358c086e6eb1de3ed25f625e48db129888f 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -107,25 +107,20 @@ static void airoha_set_gdm_port_fwd_cfg(struct airoha_eth *eth, u32 addr,
 
 static int airoha_set_gdm_port(struct airoha_eth *eth, int port, bool enable)
 {
-	u32 val = enable ? FE_PSE_PORT_PPE1 : FE_PSE_PORT_DROP;
-	u32 vip_port, cfg_addr;
+	u32 vip_port;
 
 	switch (port) {
 	case XSI_PCIE0_PORT:
 		vip_port = XSI_PCIE0_VIP_PORT_MASK;
-		cfg_addr = REG_GDM_FWD_CFG(3);
 		break;
 	case XSI_PCIE1_PORT:
 		vip_port = XSI_PCIE1_VIP_PORT_MASK;
-		cfg_addr = REG_GDM_FWD_CFG(3);
 		break;
 	case XSI_USB_PORT:
 		vip_port = XSI_USB_VIP_PORT_MASK;
-		cfg_addr = REG_GDM_FWD_CFG(4);
 		break;
 	case XSI_ETH_PORT:
 		vip_port = XSI_ETH_VIP_PORT_MASK;
-		cfg_addr = REG_GDM_FWD_CFG(4);
 		break;
 	default:
 		return -EINVAL;
@@ -139,8 +134,6 @@ static int airoha_set_gdm_port(struct airoha_eth *eth, int port, bool enable)
 		airoha_fe_clear(eth, REG_FE_IFC_PORT_EN, vip_port);
 	}
 
-	airoha_set_gdm_port_fwd_cfg(eth, cfg_addr, val);
-
 	return 0;
 }
 
@@ -177,8 +170,6 @@ static void airoha_fe_maccr_init(struct airoha_eth *eth)
 		airoha_fe_set(eth, REG_GDM_FWD_CFG(p),
 			      GDM_TCP_CKSUM | GDM_UDP_CKSUM | GDM_IP4_CKSUM |
 			      GDM_DROP_CRC_ERR);
-		airoha_set_gdm_port_fwd_cfg(eth, REG_GDM_FWD_CFG(p),
-					    FE_PSE_PORT_CDM1);
 		airoha_fe_rmw(eth, REG_GDM_LEN_CFG(p),
 			      GDM_SHORT_LEN_MASK | GDM_LONG_LEN_MASK,
 			      FIELD_PREP(GDM_SHORT_LEN_MASK, 60) |
@@ -1614,8 +1605,11 @@ static int airoha_dev_set_macaddr(struct net_device *dev, void *p)
 static int airoha_dev_init(struct net_device *dev)
 {
 	struct airoha_gdm_port *port = netdev_priv(dev);
+	struct airoha_eth *eth = port->qdma->eth;
 
 	airoha_set_macaddr(port, dev->dev_addr);
+	airoha_set_gdm_port_fwd_cfg(eth, REG_GDM_FWD_CFG(port->id),
+				    FE_PSE_PORT_PPE1);
 
 	return 0;
 }

-- 
2.48.1

[PATCH net-next v3 10/16] net: airoha: Rename airoha_set_gdm_port_fwd_cfg() in airoha_set_vip_for_gdm_port()

[Lorenzo Bianconi]

Rename airoha_set_gdm_port() in airoha_set_vip_for_gdm_port().
Get rid of airoha_set_gdm_ports routine.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c | 49 +++++++-------------------------
 drivers/net/ethernet/airoha/airoha_eth.h |  8 ------
 2 files changed, 11 insertions(+), 46 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 6c899358c086e6eb1de3ed25f625e48db129888f..31e5f0368faa13a120ba01f7413cf5c23761c143 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -105,25 +105,23 @@ static void airoha_set_gdm_port_fwd_cfg(struct airoha_eth *eth, u32 addr,
 		      FIELD_PREP(GDM_UCFQ_MASK, val));
 }
 
-static int airoha_set_gdm_port(struct airoha_eth *eth, int port, bool enable)
+static int airoha_set_vip_for_gdm_port(struct airoha_gdm_port *port,
+				       bool enable)
 {
+	struct airoha_eth *eth = port->qdma->eth;
 	u32 vip_port;
 
-	switch (port) {
-	case XSI_PCIE0_PORT:
+	switch (port->id) {
+	case 3:
+		/* FIXME: handle XSI_PCIE1_PORT */
 		vip_port = XSI_PCIE0_VIP_PORT_MASK;
 		break;
-	case XSI_PCIE1_PORT:
-		vip_port = XSI_PCIE1_VIP_PORT_MASK;
-		break;
-	case XSI_USB_PORT:
-		vip_port = XSI_USB_VIP_PORT_MASK;
-		break;
-	case XSI_ETH_PORT:
+	case 4:
+		/* FIXME: handle XSI_USB_PORT */
 		vip_port = XSI_ETH_VIP_PORT_MASK;
 		break;
 	default:
-		return -EINVAL;
+		return 0;
 	}
 
 	if (enable) {
@@ -137,31 +135,6 @@ static int airoha_set_gdm_port(struct airoha_eth *eth, int port, bool enable)
 	return 0;
 }
 
-static int airoha_set_gdm_ports(struct airoha_eth *eth, bool enable)
-{
-	const int port_list[] = {
-		XSI_PCIE0_PORT,
-		XSI_PCIE1_PORT,
-		XSI_USB_PORT,
-		XSI_ETH_PORT
-	};
-	int i, err;
-
-	for (i = 0; i < ARRAY_SIZE(port_list); i++) {
-		err = airoha_set_gdm_port(eth, port_list[i], enable);
-		if (err)
-			goto error;
-	}
-
-	return 0;
-
-error:
-	for (i--; i >= 0; i--)
-		airoha_set_gdm_port(eth, port_list[i], false);
-
-	return err;
-}
-
 static void airoha_fe_maccr_init(struct airoha_eth *eth)
 {
 	int p;
@@ -1539,7 +1512,7 @@ static int airoha_dev_open(struct net_device *dev)
 	int err;
 
 	netif_tx_start_all_queues(dev);
-	err = airoha_set_gdm_ports(qdma->eth, true);
+	err = airoha_set_vip_for_gdm_port(port, true);
 	if (err)
 		return err;
 
@@ -1565,7 +1538,7 @@ static int airoha_dev_stop(struct net_device *dev)
 	int i, err;
 
 	netif_tx_disable(dev);
-	err = airoha_set_gdm_ports(qdma->eth, false);
+	err = airoha_set_vip_for_gdm_port(port, false);
 	if (err)
 		return err;
 
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index 74bdfd9e8d2fb3706f5ec6a4e17fe07fbcb38c3d..44834227a58982d4491f3d8174b9e0bea542f785 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -57,14 +57,6 @@ enum {
 	QDMA_INT_REG_MAX
 };
 
-enum {
-	XSI_PCIE0_PORT,
-	XSI_PCIE1_PORT,
-	XSI_USB_PORT,
-	XSI_AE_PORT,
-	XSI_ETH_PORT,
-};
-
 enum {
 	XSI_PCIE0_VIP_PORT_MASK	= BIT(22),
 	XSI_PCIE1_VIP_PORT_MASK	= BIT(23),

-- 
2.48.1

[PATCH net-next v3 11/16] dt-bindings: arm: airoha: Add the NPU node for EN7581 SoC

[Lorenzo Bianconi]

This patch adds the NPU document binding for EN7581 SoC.
The Airoha Network Processor Unit (NPU) provides a configuration interface
to implement wired and wireless hardware flow offloading programming Packet
Processor Engine (PPE) flow table.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 .../devicetree/bindings/arm/airoha,en7581-npu.yaml | 71 ++++++++++++++++++++++
 1 file changed, 71 insertions(+)

diff --git a/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
new file mode 100644
index 0000000000000000000000000000000000000000..a5bcfa299e7cd54f51e70f7ded113f1efcd3e8b7
--- /dev/null
+++ b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
@@ -0,0 +1,71 @@
+# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/arm/airoha,en7581-npu.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Airoha Network Processor Unit for EN7581 SoC
+
+maintainers:
+  - Lorenzo Bianconi <lorenzo@kernel.org>
+
+description:
+  The Airoha Network Processor Unit (NPU) provides a configuration interface
+  to implement wired and wireless hardware flow offloading programming Packet
+  Processor Engine (PPE) flow table.
+
+properties:
+  compatible:
+    items:
+      - enum:
+          - airoha,en7581-npu
+      - const: syscon
+
+  reg:
+    maxItems: 1
+
+  interrupts:
+    maxItems: 15
+
+  memory-region:
+    maxItems: 1
+    description:
+      Phandle to the node describing memory used to store NPU firmware binary.
+
+required:
+  - compatible
+  - reg
+  - interrupts
+  - memory-region
+
+additionalProperties: false
+
+examples:
+  - |
+    #include <dt-bindings/interrupt-controller/arm-gic.h>
+    #include <dt-bindings/interrupt-controller/irq.h>
+    soc {
+      #address-cells = <2>;
+      #size-cells = <2>;
+
+      npu@1e900000 {
+        compatible = "airoha,en7581-npu", "syscon";
+        reg = <0 0x1e900000 0 0x313000>;
+        interrupts = <GIC_SPI 118 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 119 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 120 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 121 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 122 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 123 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 125 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 103 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 109 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 113 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 117 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 134 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 135 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 136 IRQ_TYPE_LEVEL_HIGH>,
+                     <GIC_SPI 137 IRQ_TYPE_LEVEL_HIGH>;
+        memory-region = <&npu_binary>;
+      };
+    };

-- 
2.48.1

Re: [PATCH net-next v3 11/16] dt-bindings: arm: airoha: Add the NPU node for EN7581 SoC

[Krzysztof Kozlowski]

On Sun, Feb 09, 2025 at 01:09:04PM +0100, Lorenzo Bianconi wrote:
> This patch adds the NPU document binding for EN7581 SoC.
> The Airoha Network Processor Unit (NPU) provides a configuration interface
> to implement wired and wireless hardware flow offloading programming Packet
> Processor Engine (PPE) flow table.
> 
> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> ---
>  .../devicetree/bindings/arm/airoha,en7581-npu.yaml | 71 ++++++++++++++++++++++
>  1 file changed, 71 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
> new file mode 100644
> index 0000000000000000000000000000000000000000..a5bcfa299e7cd54f51e70f7ded113f1efcd3e8b7
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml

arm is for top-level nodes, this has to go to proper directory or as
last-resort to the soc.

> @@ -0,0 +1,71 @@
> +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/arm/airoha,en7581-npu.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: Airoha Network Processor Unit for EN7581 SoC
> +
> +maintainers:
> +  - Lorenzo Bianconi <lorenzo@kernel.org>
> +
> +description:
> +  The Airoha Network Processor Unit (NPU) provides a configuration interface
> +  to implement wired and wireless hardware flow offloading programming Packet
> +  Processor Engine (PPE) flow table.

Sounds like network device, so maybe net?

> +
> +properties:
> +  compatible:
> +    items:
> +      - enum:
> +          - airoha,en7581-npu
> +      - const: syscon
> +
> +  reg:
> +    maxItems: 1
> +
> +  interrupts:
> +    maxItems: 15

You need to list the items.

> +
> +  memory-region:
> +    maxItems: 1
> +    description:
> +      Phandle to the node describing memory used to store NPU firmware binary.

s/Phandle to the node describing//

Best regards,
Krzysztof

Re: [PATCH net-next v3 11/16] dt-bindings: arm: airoha: Add the NPU node for EN7581 SoC

[Lorenzo Bianconi]

[-- Attachment #1: Type: text/plain, Size: 2364 bytes --]

On Feb 11, Krzysztof Kozlowski wrote:
> On Sun, Feb 09, 2025 at 01:09:04PM +0100, Lorenzo Bianconi wrote:
> > This patch adds the NPU document binding for EN7581 SoC.
> > The Airoha Network Processor Unit (NPU) provides a configuration interface
> > to implement wired and wireless hardware flow offloading programming Packet
> > Processor Engine (PPE) flow table.
> > 
> > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> > ---
> >  .../devicetree/bindings/arm/airoha,en7581-npu.yaml | 71 ++++++++++++++++++++++
> >  1 file changed, 71 insertions(+)
> > 
> > diff --git a/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
> > new file mode 100644
> > index 0000000000000000000000000000000000000000..a5bcfa299e7cd54f51e70f7ded113f1efcd3e8b7
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
> 
> arm is for top-level nodes, this has to go to proper directory or as
> last-resort to the soc.
> 
> > @@ -0,0 +1,71 @@
> > +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
> > +%YAML 1.2
> > +---
> > +$id: http://devicetree.org/schemas/arm/airoha,en7581-npu.yaml#
> > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > +
> > +title: Airoha Network Processor Unit for EN7581 SoC
> > +
> > +maintainers:
> > +  - Lorenzo Bianconi <lorenzo@kernel.org>
> > +
> > +description:
> > +  The Airoha Network Processor Unit (NPU) provides a configuration interface
> > +  to implement wired and wireless hardware flow offloading programming Packet
> > +  Processor Engine (PPE) flow table.
> 
> Sounds like network device, so maybe net?

yes. Do you mean to move it in Documentation/devicetree/bindings/net/ ?

> 
> > +
> > +properties:
> > +  compatible:
> > +    items:
> > +      - enum:
> > +          - airoha,en7581-npu
> > +      - const: syscon
> > +
> > +  reg:
> > +    maxItems: 1
> > +
> > +  interrupts:
> > +    maxItems: 15
> 
> You need to list the items.

ack, I will fix it.

> 
> > +
> > +  memory-region:
> > +    maxItems: 1
> > +    description:
> > +      Phandle to the node describing memory used to store NPU firmware binary.
> 
> s/Phandle to the node describing//

ack, I will fix it.

Regards,
Lorenzo

> 
> Best regards,
> Krzysztof
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH net-next v3 11/16] dt-bindings: arm: airoha: Add the NPU node for EN7581 SoC

[Krzysztof Kozlowski]

On Tue, Feb 11, 2025 at 05:32:51PM +0100, Lorenzo Bianconi wrote:
> On Feb 11, Krzysztof Kozlowski wrote:
> > On Sun, Feb 09, 2025 at 01:09:04PM +0100, Lorenzo Bianconi wrote:
> > > This patch adds the NPU document binding for EN7581 SoC.
> > > The Airoha Network Processor Unit (NPU) provides a configuration interface
> > > to implement wired and wireless hardware flow offloading programming Packet
> > > Processor Engine (PPE) flow table.
> > > 
> > > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> > > ---
> > >  .../devicetree/bindings/arm/airoha,en7581-npu.yaml | 71 ++++++++++++++++++++++
> > >  1 file changed, 71 insertions(+)
> > > 
> > > diff --git a/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
> > > new file mode 100644
> > > index 0000000000000000000000000000000000000000..a5bcfa299e7cd54f51e70f7ded113f1efcd3e8b7
> > > --- /dev/null
> > > +++ b/Documentation/devicetree/bindings/arm/airoha,en7581-npu.yaml
> > 
> > arm is for top-level nodes, this has to go to proper directory or as
> > last-resort to the soc.
> > 
> > > @@ -0,0 +1,71 @@
> > > +# SPDX-License-Identifier: (GPL-2.0 OR BSD-2-Clause)
> > > +%YAML 1.2
> > > +---
> > > +$id: http://devicetree.org/schemas/arm/airoha,en7581-npu.yaml#
> > > +$schema: http://devicetree.org/meta-schemas/core.yaml#
> > > +
> > > +title: Airoha Network Processor Unit for EN7581 SoC
> > > +
> > > +maintainers:
> > > +  - Lorenzo Bianconi <lorenzo@kernel.org>
> > > +
> > > +description:
> > > +  The Airoha Network Processor Unit (NPU) provides a configuration interface
> > > +  to implement wired and wireless hardware flow offloading programming Packet
> > > +  Processor Engine (PPE) flow table.
> > 
> > Sounds like network device, so maybe net?
> 
> yes. Do you mean to move it in Documentation/devicetree/bindings/net/ ?


Yes... and no, because after second look at your driver it looks more
like a mailbox. So basically I don't know. I usually hope contributors
know better. :)

If this is onlt mailbox provider, then should be placed in mailbox. If
this is much more (including mailbox) but main function is network, then
could be in net. So it all depends...

Best regards,
Krzysztof

[PATCH net-next v3 12/16] dt-bindings: net: airoha: Add airoha,npu phandle property

[Lorenzo Bianconi]

Introduce the airoha,npu property for the NPU node available on
EN7581 SoC. The airoha Network Processor Unit (NPU) is used to
offload network traffic forwarded between Packet Switch Engine
(PSE) ports.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml b/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
index c578637c5826db4bf470a4d01ac6f3133976ae1a..0fdd1126541774acacc783d98e4c089b2d2b85e2 100644
--- a/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
+++ b/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
@@ -63,6 +63,14 @@ properties:
   "#size-cells":
     const: 0
 
+  airoha,npu:
+    $ref: /schemas/types.yaml#/definitions/phandle
+    description:
+      Phandle to the node used to configure the NPU module.
+      The Airoha Network Processor Unit (NPU) provides a configuration
+      interface to implement hardware flow offloading programming Packet
+      Processor Engine (PPE) flow table.
+
 patternProperties:
   "^ethernet@[1-4]$":
     type: object
@@ -132,6 +140,8 @@ examples:
                      <GIC_SPI 49 IRQ_TYPE_LEVEL_HIGH>,
                      <GIC_SPI 64 IRQ_TYPE_LEVEL_HIGH>;
 
+        airoha,npu = <&npu>;
+
         #address-cells = <1>;
         #size-cells = <0>;
 

-- 
2.48.1

Re: [PATCH net-next v3 12/16] dt-bindings: net: airoha: Add airoha,npu phandle property

[Krzysztof Kozlowski]

On Sun, Feb 09, 2025 at 01:09:05PM +0100, Lorenzo Bianconi wrote:
> Introduce the airoha,npu property for the NPU node available on
> EN7581 SoC. The airoha Network Processor Unit (NPU) is used to
> offload network traffic forwarded between Packet Switch Engine
> (PSE) ports.
> 
> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> ---
>  Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml b/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
> index c578637c5826db4bf470a4d01ac6f3133976ae1a..0fdd1126541774acacc783d98e4c089b2d2b85e2 100644
> --- a/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
> +++ b/Documentation/devicetree/bindings/net/airoha,en7581-eth.yaml
> @@ -63,6 +63,14 @@ properties:
>    "#size-cells":
>      const: 0
>  
> +  airoha,npu:
> +    $ref: /schemas/types.yaml#/definitions/phandle
> +    description:
> +      Phandle to the node used to configure the NPU module.
> +      The Airoha Network Processor Unit (NPU) provides a configuration
> +      interface to implement hardware flow offloading programming Packet
> +      Processor Engine (PPE) flow table.

I see Conor had here objections, but it looks fine to me. So unless
Conor naks it:

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>

Best regards,
Krzysztof

[PATCH net-next v3 13/16] net: airoha: Introduce PPE initialization via NPU

[Lorenzo Bianconi]

Packet Processor Engine (PPE) module available on EN7581 SoC populates
the PPE table with 5-tuples flower rules learned from traffic forwarded
between the GDM ports connected to the Packet Switch Engine (PSE) module.
The airoha_eth driver can enable hw acceleration of learned 5-tuples
rules if the user configure them in netfilter flowtable (netfilter
flowtable support will be added with subsequent patches).
airoha_eth driver configures and collects data from the PPE module via a
Network Processor Unit (NPU) RISC-V module available on the EN7581 SoC.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/Kconfig       |   5 +
 drivers/net/ethernet/airoha/Makefile      |   4 +-
 drivers/net/ethernet/airoha/airoha_eth.c  |   7 +-
 drivers/net/ethernet/airoha/airoha_eth.h  |  89 ++++++
 drivers/net/ethernet/airoha/airoha_npu.c  | 451 ++++++++++++++++++++++++++++++
 drivers/net/ethernet/airoha/airoha_ppe.c  | 150 ++++++++++
 drivers/net/ethernet/airoha/airoha_regs.h | 107 ++++++-
 7 files changed, 804 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/airoha/Kconfig b/drivers/net/ethernet/airoha/Kconfig
index b6a131845f13b23a12464cfc281e3abe5699389f..c5445431595f9d2098cee9be8683b71ca1da1ea4 100644
--- a/drivers/net/ethernet/airoha/Kconfig
+++ b/drivers/net/ethernet/airoha/Kconfig
@@ -7,10 +7,15 @@ config NET_VENDOR_AIROHA
 
 if NET_VENDOR_AIROHA
 
+config NET_AIROHA_NPU
+	depends on ARCH_AIROHA || COMPILE_TEST
+	def_bool NET_AIROHA != n
+
 config NET_AIROHA
 	tristate "Airoha SoC Gigabit Ethernet support"
 	depends on NET_DSA || !NET_DSA
 	select PAGE_POOL
+	select WANT_DEV_COREDUMP
 	help
 	  This driver supports the gigabit ethernet MACs in the
 	  Airoha SoC family.
diff --git a/drivers/net/ethernet/airoha/Makefile b/drivers/net/ethernet/airoha/Makefile
index 73a6f3680a4c4ce92ee785d83b905d76a63421df..50028cfc3e3e04efbdd353b1bd65f46b488637d8 100644
--- a/drivers/net/ethernet/airoha/Makefile
+++ b/drivers/net/ethernet/airoha/Makefile
@@ -3,4 +3,6 @@
 # Airoha for the Mediatek SoCs built-in ethernet macs
 #
 
-obj-$(CONFIG_NET_AIROHA) += airoha_eth.o
+obj-$(CONFIG_NET_AIROHA) += airoha-eth.o
+airoha-eth-y := airoha_eth.o airoha_ppe.o
+airoha-eth-$(CONFIG_NET_AIROHA_NPU) += airoha_npu.o
diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 31e5f0368faa13a120ba01f7413cf5c23761c143..bc8619748f0ac4fdffd78c37e36c4a021514d5fd 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -3,9 +3,6 @@
  * Copyright (c) 2024 AIROHA Inc
  * Author: Lorenzo Bianconi <lorenzo@kernel.org>
  */
-#include <linux/of.h>
-#include <linux/of_net.h>
-#include <linux/platform_device.h>
 #include <linux/tcp.h>
 #include <linux/u64_stats_sync.h>
 #include <net/dsa.h>
@@ -1291,6 +1288,9 @@ static int airoha_hw_init(struct platform_device *pdev,
 		return err;
 
 	msleep(20);
+	if (airoha_ppe_init(eth))
+		dev_err(eth->dev, "ppe initialization failed\n");
+
 	err = airoha_fe_init(eth);
 	if (err)
 		return err;
@@ -2505,6 +2505,7 @@ static void airoha_remove(struct platform_device *pdev)
 	}
 	free_netdev(eth->napi_dev);
 
+	airoha_ppe_deinit(eth);
 	platform_set_drvdata(pdev, NULL);
 }
 
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index 44834227a58982d4491f3d8174b9e0bea542f785..d882b8c74c713209d15603311e4837f1502bbbfd 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -11,8 +11,13 @@
 #include <linux/iopoll.h>
 #include <linux/kernel.h>
 #include <linux/netdevice.h>
+#include <linux/of.h>
+#include <linux/of_net.h>
+#include <linux/of_platform.h>
+#include <linux/platform_device.h>
 #include <linux/reset.h>
 
+#define AIROHA_NPU_NUM_CORES		8
 #define AIROHA_MAX_NUM_GDM_PORTS	4
 #define AIROHA_MAX_NUM_QDMA		2
 #define AIROHA_MAX_DSA_PORTS		7
@@ -44,6 +49,15 @@
 #define QDMA_METER_IDX(_n)		((_n) & 0xff)
 #define QDMA_METER_GROUP(_n)		(((_n) >> 8) & 0x3)
 
+#define PPE_NUM				2
+#define PPE1_SRAM_NUM_ENTRIES		(8 * 1024)
+#define PPE_SRAM_NUM_ENTRIES		(2 * PPE1_SRAM_NUM_ENTRIES)
+#define PPE_DRAM_NUM_ENTRIES		(16 * 1024)
+#define PPE_NUM_ENTRIES			(PPE_SRAM_NUM_ENTRIES + PPE_DRAM_NUM_ENTRIES)
+#define PPE_HASH_MASK			(PPE_NUM_ENTRIES - 1)
+#define PPE_ENTRY_SIZE			80
+#define PPE_RAM_NUM_ENTRIES_SHIFT(_n)	(__ffs((_n) >> 10))
+
 #define MTK_HDR_LEN			4
 #define MTK_HDR_XMIT_TAGGED_TPID_8100	1
 #define MTK_HDR_XMIT_TAGGED_TPID_88A8	2
@@ -195,6 +209,10 @@ struct airoha_hw_stats {
 	u64 rx_len[7];
 };
 
+struct airoha_foe_entry {
+	u8 data[PPE_ENTRY_SIZE];
+};
+
 struct airoha_qdma {
 	struct airoha_eth *eth;
 	void __iomem *regs;
@@ -234,12 +252,36 @@ struct airoha_gdm_port {
 	struct metadata_dst *dsa_meta[AIROHA_MAX_DSA_PORTS];
 };
 
+struct airoha_npu {
+	struct platform_device *pdev;
+	struct device_node *np;
+
+	void __iomem *base;
+
+	struct airoha_npu_core {
+		struct airoha_npu *npu;
+		/* protect concurrent npu memory accesses */
+		spinlock_t lock;
+		struct work_struct wdt_work;
+	} cores[AIROHA_NPU_NUM_CORES];
+};
+
+struct airoha_ppe {
+	struct airoha_eth *eth;
+
+	void *foe;
+	dma_addr_t foe_dma;
+};
+
 struct airoha_eth {
 	struct device *dev;
 
 	unsigned long state;
 	void __iomem *fe_regs;
 
+	struct airoha_npu *npu;
+	struct airoha_ppe *ppe;
+
 	struct reset_control_bulk_data rsts[AIROHA_MAX_NUM_RSTS];
 	struct reset_control_bulk_data xsi_rsts[AIROHA_MAX_NUM_XSI_RSTS];
 
@@ -275,4 +317,51 @@ u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val);
 #define airoha_qdma_clear(qdma, offset, val)			\
 	airoha_rmw((qdma)->regs, (offset), (val), 0)
 
+bool airoha_ppe2_is_enabled(struct airoha_eth *eth);
+
+#ifdef CONFIG_NET_AIROHA_NPU
+int airoha_ppe_init(struct airoha_eth *eth);
+void airoha_ppe_deinit(struct airoha_eth *eth);
+struct airoha_npu *airoha_npu_init(struct airoha_eth *eth);
+void airoha_npu_deinit(struct airoha_npu *npu);
+int airoha_npu_ppe_init(struct airoha_npu *npu);
+int airoha_npu_ppe_deinit(struct airoha_npu *npu);
+int airoha_npu_flush_ppe_sram_entries(struct airoha_npu *npu,
+				      struct airoha_ppe *ppe);
+#else
+static inline int airoha_ppe_init(struct airoha_eth *eth)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline void airoha_ppe_deinit(struct airoha_eth *eth)
+{
+}
+
+static inline struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
+{
+	return NULL;
+}
+
+static inline void airoha_npu_deinit(struct airoha_npu *npu)
+{
+}
+
+static inline int airoha_npu_ppe_init(struct airoha_npu *npu)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int airoha_npu_ppe_deinit(struct airoha_npu *npu)
+{
+	return -EOPNOTSUPP;
+}
+
+static inline int airoha_npu_flush_ppe_sram_entries(struct airoha_npu *npu,
+						    struct airoha_ppe *ppe)
+{
+	return -EOPNOTSUPP;
+}
+#endif /* CONFIG_NET_AIROHA_NPU */
+
 #endif /* AIROHA_ETH_H */
diff --git a/drivers/net/ethernet/airoha/airoha_npu.c b/drivers/net/ethernet/airoha/airoha_npu.c
new file mode 100644
index 0000000000000000000000000000000000000000..b912ab4c02166d7f9f1aaacb4b5b6b30740a3b6c
--- /dev/null
+++ b/drivers/net/ethernet/airoha/airoha_npu.c
@@ -0,0 +1,451 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2025 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#include <linux/devcoredump.h>
+#include <linux/firmware.h>
+#include <linux/of_reserved_mem.h>
+
+#include "airoha_eth.h"
+
+#define NPU_EN7581_FIRMWARE_DATA		"airoha/en7581_npu_data.bin"
+#define NPU_EN7581_FIRMWARE_RV32		"airoha/en7581_npu_rv32.bin"
+#define NPU_EN7581_FIRMWARE_RV32_MAX_SIZE	0x200000
+#define NPU_EN7581_FIRMWARE_DATA_MAX_SIZE	0x10000
+#define NPU_DUMP_SIZE				512
+
+#define REG_NPU_LOCAL_SRAM		0x0
+
+#define NPU_PC_BASE_ADDR		0x305000
+#define REG_PC_DBG(_n)			(0x305000 + ((_n) * 0x100))
+
+#define NPU_CLUSTER_BASE_ADDR		0x306000
+
+#define REG_CR_BOOT_TRIGGER		(NPU_CLUSTER_BASE_ADDR + 0x000)
+#define REG_CR_BOOT_CONFIG		(NPU_CLUSTER_BASE_ADDR + 0x004)
+#define REG_CR_BOOT_BASE(_n)		(NPU_CLUSTER_BASE_ADDR + 0x020 + ((_n) << 2))
+
+#define NPU_MBOX_BASE_ADDR		0x30c000
+
+#define REG_CR_MBOX_INT_STATUS		(NPU_MBOX_BASE_ADDR + 0x000)
+#define MBOX_INT_STATUS_MASK		BIT(8)
+
+#define REG_CR_MBOX_INT_MASK(_n)	(NPU_MBOX_BASE_ADDR + 0x004 + ((_n) << 2))
+#define REG_CR_MBQ0_CTRL(_n)		(NPU_MBOX_BASE_ADDR + 0x030 + ((_n) << 2))
+#define REG_CR_MBQ8_CTRL(_n)		(NPU_MBOX_BASE_ADDR + 0x0b0 + ((_n) << 2))
+#define REG_CR_NPU_MIB(_n)		(NPU_MBOX_BASE_ADDR + 0x140 + ((_n) << 2))
+
+#define NPU_TIMER_BASE_ADDR		0x310100
+#define REG_WDT_TIMER_CTRL(_n)		(NPU_TIMER_BASE_ADDR + ((_n) * 0x100))
+#define WDT_EN_MASK			BIT(25)
+#define WDT_INTR_MASK			BIT(21)
+
+enum {
+	NPU_OP_SET = 1,
+	NPU_OP_SET_NO_WAIT,
+	NPU_OP_GET,
+	NPU_OP_GET_NO_WAIT,
+};
+
+enum {
+	NPU_FUNC_WIFI,
+	NPU_FUNC_TUNNEL,
+	NPU_FUNC_NOTIFY,
+	NPU_FUNC_DBA,
+	NPU_FUNC_TR471,
+	NPU_FUNC_PPE,
+};
+
+enum {
+	NPU_MBOX_ERROR,
+	NPU_MBOX_SUCCESS,
+};
+
+enum {
+	PPE_FUNC_SET_WAIT,
+	PPE_FUNC_SET_WAIT_HWNAT_INIT,
+	PPE_FUNC_SET_WAIT_HWNAT_DEINIT,
+	PPE_FUNC_SET_WAIT_API,
+};
+
+enum {
+	PPE2_SRAM_SET_ENTRY,
+	PPE_SRAM_SET_ENTRY,
+	PPE_SRAM_SET_VAL,
+	PPE_SRAM_RESET_VAL,
+};
+
+enum {
+	QDMA_WAN_ETHER = 1,
+	QDMA_WAN_PON_XDSL,
+};
+
+struct npu_mbox_metadata {
+	union {
+		struct {
+			u16 wait_rsp:1;
+			u16 done:1;
+			u16 status:3;
+			u16 static_buf:1;
+			u16 rsv:5;
+			u16 func_id:4;
+		};
+		u16 data;
+	};
+};
+
+#define PPE_TYPE_L2B_IPV4	2
+#define PPE_TYPE_L2B_IPV4_IPV6	3
+
+struct ppe_mbox_data {
+	u32 func_type;
+	u32 func_id;
+	union {
+		struct {
+			u8 cds;
+			u8 xpon_hal_api;
+			u8 wan_xsi;
+			u8 ct_joyme4;
+			int ppe_type;
+			int wan_mode;
+			int wan_sel;
+		} init_info;
+		struct {
+			int func_id;
+			u32 size;
+			u32 data;
+		} set_info;
+	};
+};
+
+static u32 airoha_npu_rr(struct airoha_npu *npu, u32 reg)
+{
+	return readl(npu->base + reg);
+}
+
+static void airoha_npu_wr(struct airoha_npu *npu, u32 reg, u32 val)
+{
+	writel(val, npu->base + reg);
+}
+
+static u32 airoha_npu_rmw(struct airoha_npu *npu, u32 reg, u32 mask, u32 val)
+{
+	val |= airoha_npu_rr(npu, reg) & ~mask;
+	airoha_npu_wr(npu, reg, val);
+
+	return val;
+}
+
+static int airoha_npu_send_msg(struct airoha_npu *npu, int func_id,
+			       void *p, int size)
+{
+	struct device *dev = &npu->pdev->dev;
+	struct npu_mbox_metadata meta = {
+		.wait_rsp = 1,
+		.func_id = func_id,
+	};
+	u16 core = 0; /* FIXME */
+	u32 val, offset = core << 4;
+	dma_addr_t dma_addr;
+	void *addr;
+	int ret;
+
+	addr = kzalloc(size, GFP_ATOMIC | GFP_DMA);
+	if (!addr)
+		return -ENOMEM;
+
+	memcpy(addr, p, size);
+	dma_addr = dma_map_single(dev, addr, size, DMA_TO_DEVICE);
+	ret = dma_mapping_error(dev, dma_addr);
+	if (ret)
+		goto out;
+
+	spin_lock_bh(&npu->cores[core].lock);
+
+	airoha_npu_wr(npu, REG_CR_MBQ0_CTRL(0) + offset, dma_addr);
+	airoha_npu_wr(npu, REG_CR_MBQ0_CTRL(1) + offset, size);
+	val = airoha_npu_rr(npu, REG_CR_MBQ0_CTRL(2) + offset);
+	airoha_npu_wr(npu, REG_CR_MBQ0_CTRL(2) + offset, val + 1);
+	airoha_npu_wr(npu, REG_CR_MBQ0_CTRL(3) + offset, meta.data);
+
+	ret = read_poll_timeout_atomic(airoha_npu_rr, meta.data, meta.done,
+				       100, 100 * MSEC_PER_SEC, false, npu,
+				       REG_CR_MBQ0_CTRL(3) + offset);
+	if (!ret)
+		ret = meta.status == NPU_MBOX_SUCCESS ? 0 : -EINVAL;
+
+	spin_unlock_bh(&npu->cores[core].lock);
+
+	dma_unmap_single(dev, dma_addr, size, DMA_TO_DEVICE);
+out:
+	kfree(addr);
+
+	return ret;
+}
+
+static int airoha_npu_run_firmware(struct airoha_npu *npu, struct reserved_mem *rmem)
+{
+	struct device *dev = &npu->pdev->dev;
+	const struct firmware *fw;
+	void __iomem *addr;
+	int ret;
+
+	ret = request_firmware(&fw, NPU_EN7581_FIRMWARE_RV32, dev);
+	if (ret)
+		return ret;
+
+	if (fw->size > NPU_EN7581_FIRMWARE_RV32_MAX_SIZE) {
+		dev_err(dev, "%s: fw size too overlimit (%zu)\n",
+			NPU_EN7581_FIRMWARE_RV32, fw->size);
+		ret = -E2BIG;
+		goto out;
+	}
+
+	addr = devm_ioremap(dev, rmem->base, rmem->size);
+	if (!addr) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	memcpy_toio(addr, fw->data, fw->size);
+	release_firmware(fw);
+
+	ret = request_firmware(&fw, NPU_EN7581_FIRMWARE_DATA, dev);
+	if (ret)
+		return ret;
+
+	if (fw->size > NPU_EN7581_FIRMWARE_DATA_MAX_SIZE) {
+		dev_err(dev, "%s: fw size too overlimit (%zu)\n",
+			NPU_EN7581_FIRMWARE_DATA, fw->size);
+		ret = -E2BIG;
+		goto out;
+	}
+
+	memcpy_toio(npu->base + REG_NPU_LOCAL_SRAM, fw->data, fw->size);
+out:
+	release_firmware(fw);
+
+	return ret;
+}
+
+static irqreturn_t airoha_npu_mbox_handler(int irq, void *npu_instance)
+{
+	struct airoha_npu *npu = npu_instance;
+	struct npu_mbox_metadata meta;
+
+	/* clear mbox interrupt status */
+	airoha_npu_wr(npu, REG_CR_MBOX_INT_STATUS, MBOX_INT_STATUS_MASK);
+
+	/* acknowledge npu */
+	meta.data = airoha_npu_rr(npu, REG_CR_MBQ8_CTRL(3));
+	meta.status = 0;
+	meta.done = 1;
+	airoha_npu_wr(npu, REG_CR_MBQ8_CTRL(3), meta.data);
+
+	return IRQ_HANDLED;
+}
+
+int airoha_npu_ppe_init(struct airoha_npu *npu)
+{
+	struct ppe_mbox_data ppe_data = {
+		.func_type = NPU_OP_SET,
+		.func_id = PPE_FUNC_SET_WAIT_HWNAT_INIT,
+		.init_info = {
+			.ppe_type = PPE_TYPE_L2B_IPV4_IPV6,
+			.wan_mode = QDMA_WAN_ETHER,
+		},
+	};
+
+	return airoha_npu_send_msg(npu, NPU_FUNC_PPE, &ppe_data,
+				   sizeof(struct ppe_mbox_data));
+}
+
+int airoha_npu_ppe_deinit(struct airoha_npu *npu)
+{
+	struct ppe_mbox_data ppe_data = {
+		.func_type = NPU_OP_SET,
+		.func_id = PPE_FUNC_SET_WAIT_HWNAT_DEINIT,
+	};
+
+	return airoha_npu_send_msg(npu, NPU_FUNC_PPE, &ppe_data,
+				   sizeof(struct ppe_mbox_data));
+}
+
+int airoha_npu_flush_ppe_sram_entries(struct airoha_npu *npu,
+				      struct airoha_ppe *ppe)
+{
+	struct ppe_mbox_data ppe_data = {
+		.func_type = NPU_OP_SET,
+		.func_id = PPE_FUNC_SET_WAIT_API,
+		.set_info = {
+			.func_id = PPE_SRAM_RESET_VAL,
+			.data = ppe->foe_dma,
+			.size = PPE_SRAM_NUM_ENTRIES,
+		},
+	};
+	int i, sram_num_entries = PPE_SRAM_NUM_ENTRIES;
+	struct airoha_foe_entry *hwe = ppe->foe;
+
+	if (airoha_ppe2_is_enabled(ppe->eth))
+		sram_num_entries = sram_num_entries / 2;
+
+	for (i = 0; i < sram_num_entries; i++)
+		memset(&hwe[i], 0, sizeof(*hwe));
+
+	return airoha_npu_send_msg(npu, NPU_FUNC_PPE, &ppe_data,
+				   sizeof(struct ppe_mbox_data));
+}
+
+static void airoha_npu_wdt_work(struct work_struct *work)
+{
+	struct airoha_npu_core *core;
+	struct airoha_npu *npu;
+	void *dump;
+	int c;
+
+	core = container_of(work, struct airoha_npu_core, wdt_work);
+	npu = core->npu;
+
+	dump = vzalloc(NPU_DUMP_SIZE);
+	if (!dump)
+		return;
+
+	c = core - &npu->cores[0];
+	snprintf(dump, NPU_DUMP_SIZE, "PC: %08x SP: %08x LR: %08x\n",
+		 airoha_npu_rr(npu, REG_PC_DBG(c)),
+		 airoha_npu_rr(npu, REG_PC_DBG(c) + 0x4),
+		 airoha_npu_rr(npu, REG_PC_DBG(c) + 0x8));
+
+	dev_coredumpv(&npu->pdev->dev, dump, NPU_DUMP_SIZE, GFP_KERNEL);
+}
+
+static irqreturn_t airoha_npu_wdt_handler(int irq, void *core_instance)
+{
+	struct airoha_npu_core *core = core_instance;
+	struct airoha_npu *npu = core->npu;
+	int c = core - &npu->cores[0];
+	u32 val;
+
+	airoha_npu_rmw(npu, REG_WDT_TIMER_CTRL(c), 0, WDT_INTR_MASK);
+	val = airoha_npu_rr(npu, REG_WDT_TIMER_CTRL(c));
+	if (FIELD_GET(WDT_EN_MASK, val))
+		schedule_work(&core->wdt_work);
+
+	return IRQ_HANDLED;
+}
+
+struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
+{
+	struct reserved_mem *rmem;
+	int i, irq, err = -ENODEV;
+	struct airoha_npu *npu;
+	struct device_node *np;
+
+	npu = devm_kzalloc(eth->dev, sizeof(*npu), GFP_KERNEL);
+	if (!npu)
+		return ERR_PTR(-ENOMEM);
+
+	npu->np = of_parse_phandle(eth->dev->of_node, "airoha,npu", 0);
+	if (!npu->np)
+		return ERR_PTR(-ENODEV);
+
+	npu->pdev = of_find_device_by_node(npu->np);
+	if (!npu->pdev)
+		goto error_of_node_put;
+
+	get_device(&npu->pdev->dev);
+
+	npu->base = devm_platform_ioremap_resource(npu->pdev, 0);
+	if (IS_ERR(npu->base))
+		goto error_put_dev;
+
+	np = of_parse_phandle(npu->np, "memory-region", 0);
+	if (!np)
+		goto error_put_dev;
+
+	rmem = of_reserved_mem_lookup(np);
+	of_node_put(np);
+
+	if (!rmem)
+		goto error_put_dev;
+
+	irq = platform_get_irq(npu->pdev, 0);
+	if (irq < 0) {
+		err = irq;
+		goto error_put_dev;
+	}
+
+	err = devm_request_irq(&npu->pdev->dev, irq, airoha_npu_mbox_handler,
+			       IRQF_SHARED, "airoha-npu-mbox", npu);
+	if (err)
+		goto error_put_dev;
+
+	for (i = 0; i < ARRAY_SIZE(npu->cores); i++) {
+		struct airoha_npu_core *core = &npu->cores[i];
+
+		spin_lock_init(&core->lock);
+		core->npu = npu;
+
+		irq = platform_get_irq(npu->pdev, i + 1);
+		if (irq < 0) {
+			err = irq;
+			goto error_put_dev;
+		}
+
+		err = devm_request_irq(&npu->pdev->dev, irq,
+				       airoha_npu_wdt_handler, IRQF_SHARED,
+				       "airoha-npu-wdt", core);
+		if (err)
+			goto error_put_dev;
+
+		INIT_WORK(&core->wdt_work, airoha_npu_wdt_work);
+	}
+
+	if (dma_set_coherent_mask(&npu->pdev->dev, 0xbfffffff))
+		dev_err(&npu->pdev->dev,
+			"failed coherent DMA configuration\n");
+
+	err = airoha_npu_run_firmware(npu, rmem);
+	if (err)
+		goto error_put_dev;
+
+	airoha_npu_wr(npu, REG_CR_NPU_MIB(10),
+		      rmem->base + NPU_EN7581_FIRMWARE_RV32_MAX_SIZE);
+	airoha_npu_wr(npu, REG_CR_NPU_MIB(11), 0x40000); /* SRAM 256K */
+	airoha_npu_wr(npu, REG_CR_NPU_MIB(12), 0);
+	airoha_npu_wr(npu, REG_CR_NPU_MIB(21), 1);
+	msleep(100);
+
+	/* setting booting address */
+	for (i = 0; i < AIROHA_NPU_NUM_CORES; i++)
+		airoha_npu_wr(npu, REG_CR_BOOT_BASE(i), rmem->base);
+	usleep_range(1000, 2000);
+
+	/* enable NPU cores */
+	/* do not start core3 since it is used for WiFi offloading */
+	airoha_npu_wr(npu, REG_CR_BOOT_CONFIG, 0xf7);
+	airoha_npu_wr(npu, REG_CR_BOOT_TRIGGER, 0x1);
+	msleep(100);
+
+	return npu;
+
+error_put_dev:
+	put_device(&npu->pdev->dev);
+error_of_node_put:
+	of_node_put(npu->np);
+
+	return ERR_PTR(err);
+}
+
+void airoha_npu_deinit(struct airoha_npu *npu)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(npu->cores); i++)
+		cancel_work_sync(&npu->cores[i].wdt_work);
+
+	put_device(&npu->pdev->dev);
+	of_node_put(npu->np);
+}
diff --git a/drivers/net/ethernet/airoha/airoha_ppe.c b/drivers/net/ethernet/airoha/airoha_ppe.c
new file mode 100644
index 0000000000000000000000000000000000000000..b6af9f8fe72d10aae118c2fa2fb7a15dabaa5ebf
--- /dev/null
+++ b/drivers/net/ethernet/airoha/airoha_ppe.c
@@ -0,0 +1,150 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2025 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#include "airoha_regs.h"
+#include "airoha_eth.h"
+
+bool airoha_ppe2_is_enabled(struct airoha_eth *eth)
+{
+	return airoha_fe_rr(eth, REG_PPE_GLO_CFG(1)) & PPE_GLO_CFG_EN_MASK;
+}
+
+static void airoha_ppe_hw_init(struct airoha_ppe *ppe)
+{
+	u32 sram_tb_size, sram_num_entries, dram_num_entries;
+	struct airoha_eth *eth = ppe->eth;
+	int i;
+
+	sram_tb_size = PPE_SRAM_NUM_ENTRIES * sizeof(struct airoha_foe_entry);
+	dram_num_entries = PPE_RAM_NUM_ENTRIES_SHIFT(PPE_DRAM_NUM_ENTRIES);
+
+	for (i = 0; i < PPE_NUM; i++) {
+		int p;
+
+		airoha_fe_wr(eth, REG_PPE_TB_BASE(i),
+			     ppe->foe_dma + sram_tb_size);
+
+		airoha_fe_rmw(eth, REG_PPE_BND_AGE0(i),
+			      PPE_BIND_AGE0_DELTA_NON_L4 |
+			      PPE_BIND_AGE0_DELTA_UDP,
+			      FIELD_PREP(PPE_BIND_AGE0_DELTA_NON_L4, 1) |
+			      FIELD_PREP(PPE_BIND_AGE0_DELTA_UDP, 12));
+		airoha_fe_rmw(eth, REG_PPE_BND_AGE1(i),
+			      PPE_BIND_AGE1_DELTA_TCP_FIN |
+			      PPE_BIND_AGE1_DELTA_TCP,
+			      FIELD_PREP(PPE_BIND_AGE1_DELTA_TCP_FIN, 1) |
+			      FIELD_PREP(PPE_BIND_AGE1_DELTA_TCP, 7));
+
+		airoha_fe_rmw(eth, REG_PPE_TB_HASH_CFG(i),
+			      PPE_SRAM_TABLE_EN_MASK |
+			      PPE_SRAM_HASH1_EN_MASK |
+			      PPE_DRAM_TABLE_EN_MASK |
+			      PPE_SRAM_HASH0_MODE_MASK |
+			      PPE_SRAM_HASH1_MODE_MASK |
+			      PPE_DRAM_HASH0_MODE_MASK |
+			      PPE_DRAM_HASH1_MODE_MASK,
+			      FIELD_PREP(PPE_SRAM_TABLE_EN_MASK, 1) |
+			      FIELD_PREP(PPE_SRAM_HASH1_EN_MASK, 1) |
+			      FIELD_PREP(PPE_SRAM_HASH1_MODE_MASK, 1) |
+			      FIELD_PREP(PPE_DRAM_HASH1_MODE_MASK, 3));
+
+		airoha_fe_rmw(eth, REG_PPE_TB_CFG(i),
+			      PPE_TB_CFG_SEARCH_MISS_MASK |
+			      PPE_TB_ENTRY_SIZE_MASK,
+			      FIELD_PREP(PPE_TB_CFG_SEARCH_MISS_MASK, 3) |
+			      FIELD_PREP(PPE_TB_ENTRY_SIZE_MASK, 0));
+
+		airoha_fe_wr(eth, REG_PPE_HASH_SEED(i), PPE_HASH_SEED);
+
+		for (p = 0; p < ARRAY_SIZE(eth->ports); p++)
+			airoha_fe_rmw(eth, REG_PPE_MTU(i, p),
+				      FP0_EGRESS_MTU_MASK |
+				      FP1_EGRESS_MTU_MASK,
+				      FIELD_PREP(FP0_EGRESS_MTU_MASK,
+						 AIROHA_MAX_MTU) |
+				      FIELD_PREP(FP1_EGRESS_MTU_MASK,
+						 AIROHA_MAX_MTU));
+	}
+
+	if (airoha_ppe2_is_enabled(eth)) {
+		sram_num_entries =
+			PPE_RAM_NUM_ENTRIES_SHIFT(PPE1_SRAM_NUM_ENTRIES);
+		airoha_fe_rmw(eth, REG_PPE_TB_CFG(0),
+			      PPE_SRAM_TB_NUM_ENTRY_MASK |
+			      PPE_DRAM_TB_NUM_ENTRY_MASK,
+			      FIELD_PREP(PPE_SRAM_TB_NUM_ENTRY_MASK,
+					 sram_num_entries) |
+			      FIELD_PREP(PPE_DRAM_TB_NUM_ENTRY_MASK,
+					 dram_num_entries));
+		airoha_fe_rmw(eth, REG_PPE_TB_CFG(1),
+			      PPE_SRAM_TB_NUM_ENTRY_MASK |
+			      PPE_DRAM_TB_NUM_ENTRY_MASK,
+			      FIELD_PREP(PPE_SRAM_TB_NUM_ENTRY_MASK,
+					 sram_num_entries) |
+			      FIELD_PREP(PPE_DRAM_TB_NUM_ENTRY_MASK,
+					 dram_num_entries));
+	} else {
+		sram_num_entries =
+			PPE_RAM_NUM_ENTRIES_SHIFT(PPE_SRAM_NUM_ENTRIES);
+		airoha_fe_rmw(eth, REG_PPE_TB_CFG(0),
+			      PPE_SRAM_TB_NUM_ENTRY_MASK |
+			      PPE_DRAM_TB_NUM_ENTRY_MASK,
+			      FIELD_PREP(PPE_SRAM_TB_NUM_ENTRY_MASK,
+					 sram_num_entries) |
+			      FIELD_PREP(PPE_DRAM_TB_NUM_ENTRY_MASK,
+					 dram_num_entries));
+	}
+}
+
+int airoha_ppe_init(struct airoha_eth *eth)
+{
+	struct airoha_npu *npu;
+	struct airoha_ppe *ppe;
+	int foe_size, err;
+
+	ppe = devm_kzalloc(eth->dev, sizeof(*ppe), GFP_KERNEL);
+	if (!ppe)
+		return -ENOMEM;
+
+	foe_size = PPE_NUM_ENTRIES * sizeof(struct airoha_foe_entry);
+	ppe->foe = dmam_alloc_coherent(eth->dev, foe_size, &ppe->foe_dma,
+				       GFP_KERNEL | __GFP_ZERO);
+	if (!ppe->foe)
+		return -ENOMEM;
+
+	ppe->eth = eth;
+	eth->ppe = ppe;
+
+	npu = airoha_npu_init(eth);
+	if (IS_ERR(npu))
+		return PTR_ERR(npu);
+
+	eth->npu = npu;
+	err = airoha_npu_ppe_init(npu);
+	if (err)
+		goto error;
+
+	airoha_ppe_hw_init(ppe);
+	err = airoha_npu_flush_ppe_sram_entries(npu, ppe);
+	if (err)
+		goto error;
+
+	return 0;
+
+error:
+	airoha_npu_deinit(npu);
+	eth->npu = NULL;
+
+	return err;
+}
+
+void airoha_ppe_deinit(struct airoha_eth *eth)
+{
+	if (eth->npu) {
+		airoha_npu_ppe_deinit(eth->npu);
+		airoha_npu_deinit(eth->npu);
+	}
+}
diff --git a/drivers/net/ethernet/airoha/airoha_regs.h b/drivers/net/ethernet/airoha/airoha_regs.h
index e467dd81ff44a9ad560226cab42b7431812f5fb9..6cc64c60953a3961b7c93dfa75a289a6f7a6599b 100644
--- a/drivers/net/ethernet/airoha/airoha_regs.h
+++ b/drivers/net/ethernet/airoha/airoha_regs.h
@@ -15,6 +15,7 @@
 #define CDM1_BASE			0x0400
 #define GDM1_BASE			0x0500
 #define PPE1_BASE			0x0c00
+#define PPE2_BASE			0x1c00
 
 #define CDM2_BASE			0x1400
 #define GDM2_BASE			0x1500
@@ -36,6 +37,7 @@
 #define FE_RST_GDM3_MBI_ARB_MASK	BIT(2)
 #define FE_RST_CORE_MASK		BIT(0)
 
+#define REG_FE_FOE_TS			0x0010
 #define REG_FE_WAN_MAC_H		0x0030
 #define REG_FE_LAN_MAC_H		0x0040
 
@@ -192,11 +194,106 @@
 #define REG_FE_GDM_RX_ETH_L511_CNT_L(_n)	(GDM_BASE(_n) + 0x198)
 #define REG_FE_GDM_RX_ETH_L1023_CNT_L(_n)	(GDM_BASE(_n) + 0x19c)
 
-#define REG_PPE1_TB_HASH_CFG		(PPE1_BASE + 0x250)
-#define PPE1_SRAM_TABLE_EN_MASK		BIT(0)
-#define PPE1_SRAM_HASH1_EN_MASK		BIT(8)
-#define PPE1_DRAM_TABLE_EN_MASK		BIT(16)
-#define PPE1_DRAM_HASH1_EN_MASK		BIT(24)
+#define REG_PPE_GLO_CFG(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x200)
+#define PPE_GLO_CFG_BUSY_MASK			BIT(31)
+#define PPE_GLO_CFG_FLOW_DROP_UPDATE_MASK	BIT(9)
+#define PPE_GLO_CFG_PSE_HASH_OFS_MASK		BIT(6)
+#define PPE_GLO_CFG_PPE_BSWAP_MASK		BIT(5)
+#define PPE_GLO_CFG_TTL_DROP_MASK		BIT(4)
+#define PPE_GLO_CFG_IP4_CS_DROP_MASK		BIT(3)
+#define PPE_GLO_CFG_IP4_L4_CS_DROP_MASK		BIT(2)
+#define PPE_GLO_CFG_EN_MASK			BIT(0)
+
+#define REG_PPE_PPE_FLOW_CFG(_n)		(((_n) ? PPE2_BASE : PPE1_BASE) + 0x204)
+#define PPE_FLOW_CFG_IP6_HASH_GRE_KEY_MASK	BIT(20)
+#define PPE_FLOW_CFG_IP4_HASH_GRE_KEY_MASK	BIT(19)
+#define PPE_FLOW_CFG_IP4_HASH_FLOW_LABEL_MASK	BIT(18)
+#define PPE_FLOW_CFG_IP4_NAT_FRAG_MASK		BIT(17)
+#define PPE_FLOW_CFG_IP_PROTO_BLACKLIST_MASK	BIT(16)
+#define PPE_FLOW_CFG_IP4_DSLITE_MASK		BIT(14)
+#define PPE_FLOW_CFG_IP4_NAPT_MASK		BIT(13)
+#define PPE_FLOW_CFG_IP4_NAT_MASK		BIT(12)
+#define PPE_FLOW_CFG_IP6_6RD_MASK		BIT(10)
+#define PPE_FLOW_CFG_IP6_5T_ROUTE_MASK		BIT(9)
+#define PPE_FLOW_CFG_IP6_3T_ROUTE_MASK		BIT(8)
+#define PPE_FLOW_CFG_IP4_UDP_FRAG_MASK		BIT(7)
+#define PPE_FLOW_CFG_IP4_TCP_FRAG_MASK		BIT(6)
+
+#define REG_PPE_IP_PROTO_CHK(_n)		(((_n) ? PPE2_BASE : PPE1_BASE) + 0x208)
+#define PPE_IP_PROTO_CHK_IPV4_MASK		GENMASK(15, 0)
+#define PPE_IP_PROTO_CHK_IPV6_MASK		GENMASK(31, 16)
+
+#define REG_PPE_TB_CFG(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x21c)
+#define PPE_SRAM_TB_NUM_ENTRY_MASK		GENMASK(26, 24)
+#define PPE_TB_CFG_KEEPALIVE_MASK		GENMASK(13, 12)
+#define PPE_TB_CFG_AGE_TCP_FIN_MASK		BIT(11)
+#define PPE_TB_CFG_AGE_UDP_MASK			BIT(10)
+#define PPE_TB_CFG_AGE_TCP_MASK			BIT(9)
+#define PPE_TB_CFG_AGE_UNBIND_MASK		BIT(8)
+#define PPE_TB_CFG_AGE_NON_L4_MASK		BIT(7)
+#define PPE_TB_CFG_AGE_PREBIND_MASK		BIT(6)
+#define PPE_TB_CFG_SEARCH_MISS_MASK		GENMASK(5, 4)
+#define PPE_TB_ENTRY_SIZE_MASK			BIT(3)
+#define PPE_DRAM_TB_NUM_ENTRY_MASK		GENMASK(2, 0)
+
+#define REG_PPE_TB_BASE(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x220)
+
+#define REG_PPE_BIND_RATE(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x228)
+#define PPE_BIND_RATE_L2B_BIND_MASK		GENMASK(31, 16)
+#define PPE_BIND_RATE_BIND_MASK			GENMASK(15, 0)
+
+#define REG_PPE_BIND_LIMIT0(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x22c)
+#define PPE_BIND_LIMIT0_HALF_MASK		GENMASK(29, 16)
+#define PPE_BIND_LIMIT0_QUARTER_MASK		GENMASK(13, 0)
+
+#define REG_PPE_BIND_LIMIT1(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x230)
+#define PPE_BIND_LIMIT1_NON_L4_MASK		GENMASK(23, 16)
+#define PPE_BIND_LIMIT1_FULL_MASK		GENMASK(13, 0)
+
+#define REG_PPE_BND_AGE0(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x23c)
+#define PPE_BIND_AGE0_DELTA_NON_L4		GENMASK(30, 16)
+#define PPE_BIND_AGE0_DELTA_UDP			GENMASK(14, 0)
+
+#define REG_PPE_UNBIND_AGE(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x238)
+#define PPE_UNBIND_AGE_MIN_PACKETS_MASK		GENMASK(31, 16)
+#define PPE_UNBIND_AGE_DELTA_MASK		GENMASK(7, 0)
+
+#define REG_PPE_BND_AGE1(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x240)
+#define PPE_BIND_AGE1_DELTA_TCP_FIN		GENMASK(30, 16)
+#define PPE_BIND_AGE1_DELTA_TCP			GENMASK(14, 0)
+
+#define REG_PPE_HASH_SEED(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x244)
+#define PPE_HASH_SEED				0x12345678
+
+#define REG_PPE_DFT_CPORT0(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x248)
+
+#define REG_PPE_DFT_CPORT1(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x24c)
+
+#define REG_PPE_TB_HASH_CFG(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x250)
+#define PPE_DRAM_HASH1_MODE_MASK		GENMASK(31, 28)
+#define PPE_DRAM_HASH1_EN_MASK			BIT(24)
+#define PPE_DRAM_HASH0_MODE_MASK		GENMASK(23, 20)
+#define PPE_DRAM_TABLE_EN_MASK			BIT(16)
+#define PPE_SRAM_HASH1_MODE_MASK		GENMASK(15, 12)
+#define PPE_SRAM_HASH1_EN_MASK			BIT(8)
+#define PPE_SRAM_HASH0_MODE_MASK		GENMASK(7, 4)
+#define PPE_SRAM_TABLE_EN_MASK			BIT(0)
+
+#define REG_PPE_MTU_BASE(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x304)
+#define REG_PPE_MTU(_m, _n)			(REG_PPE_MTU_BASE(_m) + ((_n) << 2))
+#define FP1_EGRESS_MTU_MASK			GENMASK(29, 16)
+#define FP0_EGRESS_MTU_MASK			GENMASK(13, 0)
+
+#define REG_PPE_RAM_CTRL(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x31c)
+#define PPE_SRAM_CTRL_ACK_MASK			BIT(31)
+#define PPE_SRAM_CTRL_DUAL_SUCESS_MASK		BIT(30)
+#define PPE_SRAM_CTRL_ENTRY_MASK		GENMASK(23, 8)
+#define PPE_SRAM_WR_DUAL_DIRECTION_MASK		BIT(2)
+#define PPE_SRAM_CTRL_WR_MASK			BIT(1)
+#define PPE_SRAM_CTRL_REQ_MASK			BIT(0)
+
+#define REG_PPE_RAM_BASE(_n)			(((_n) ? PPE2_BASE : PPE1_BASE) + 0x320)
+#define REG_PPE_RAM_ENTRY(_m, _n)		(REG_PPE_RAM_BASE(_m) + ((_n) << 2))
 
 #define REG_FE_GDM_TX_OK_PKT_CNT_H(_n)		(GDM_BASE(_n) + 0x280)
 #define REG_FE_GDM_TX_OK_BYTE_CNT_H(_n)		(GDM_BASE(_n) + 0x284)

-- 
2.48.1

Re: [PATCH net-next v3 13/16] net: airoha: Introduce PPE initialization via NPU

[Krzysztof Kozlowski]

On Sun, Feb 09, 2025 at 01:09:06PM +0100, Lorenzo Bianconi wrote:
> +static irqreturn_t airoha_npu_wdt_handler(int irq, void *core_instance)
> +{
> +	struct airoha_npu_core *core = core_instance;
> +	struct airoha_npu *npu = core->npu;
> +	int c = core - &npu->cores[0];
> +	u32 val;
> +
> +	airoha_npu_rmw(npu, REG_WDT_TIMER_CTRL(c), 0, WDT_INTR_MASK);
> +	val = airoha_npu_rr(npu, REG_WDT_TIMER_CTRL(c));
> +	if (FIELD_GET(WDT_EN_MASK, val))
> +		schedule_work(&core->wdt_work);
> +
> +	return IRQ_HANDLED;
> +}
> +
> +struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
> +{
> +	struct reserved_mem *rmem;
> +	int i, irq, err = -ENODEV;
> +	struct airoha_npu *npu;
> +	struct device_node *np;
> +
> +	npu = devm_kzalloc(eth->dev, sizeof(*npu), GFP_KERNEL);
> +	if (!npu)
> +		return ERR_PTR(-ENOMEM);
> +
> +	npu->np = of_parse_phandle(eth->dev->of_node, "airoha,npu", 0);
> +	if (!npu->np)
> +		return ERR_PTR(-ENODEV);

Why? The property is not required, so how can missing property fail the
probe?

This is also still unnecessary ABI break without explanation/reasoning.

> +
> +	npu->pdev = of_find_device_by_node(npu->np);
> +	if (!npu->pdev)
> +		goto error_of_node_put;

You should also add device link and probably try_module_get. See
qcom,ice (patch for missing try_module_get is on the lists).

> +
> +	get_device(&npu->pdev->dev);

Why? of_find_device_by_node() does it.

> +
> +	npu->base = devm_platform_ioremap_resource(npu->pdev, 0);
> +	if (IS_ERR(npu->base))
> +		goto error_put_dev;
> +
> +	np = of_parse_phandle(npu->np, "memory-region", 0);
> +	if (!np)
> +		goto error_put_dev;
> +
> +	rmem = of_reserved_mem_lookup(np);
> +	of_node_put(np);
> +
> +	if (!rmem)
> +		goto error_put_dev;
> +
> +	irq = platform_get_irq(npu->pdev, 0);
> +	if (irq < 0) {
> +		err = irq;
> +		goto error_put_dev;
> +	}
> +
> +	err = devm_request_irq(&npu->pdev->dev, irq, airoha_npu_mbox_handler,
> +			       IRQF_SHARED, "airoha-npu-mbox", npu);
> +	if (err)
> +		goto error_put_dev;
> +
> +	for (i = 0; i < ARRAY_SIZE(npu->cores); i++) {
> +		struct airoha_npu_core *core = &npu->cores[i];
> +
> +		spin_lock_init(&core->lock);
> +		core->npu = npu;
> +
> +		irq = platform_get_irq(npu->pdev, i + 1);
> +		if (irq < 0) {
> +			err = irq;
> +			goto error_put_dev;
> +		}

This is all confusing. Why are you requesting IRQs for other - the npu -
device? That device driver is responsible for its interrupts, not you
here. This breaks encapsulation. And what do you do if the other device
starts handling interrupts on its own? This is really unexpected to see
here.

> +
> +		err = devm_request_irq(&npu->pdev->dev, irq,
> +				       airoha_npu_wdt_handler, IRQF_SHARED,
> +				       "airoha-npu-wdt", core);
> +		if (err)
> +			goto error_put_dev;
> +
> +		INIT_WORK(&core->wdt_work, airoha_npu_wdt_work);
> +	}
> +
> +	if (dma_set_coherent_mask(&npu->pdev->dev, 0xbfffffff))
> +		dev_err(&npu->pdev->dev,
> +			"failed coherent DMA configuration\n");
> +
> +	err = airoha_npu_run_firmware(npu, rmem);
> +	if (err)
> +		goto error_put_dev;
> +
> +	airoha_npu_wr(npu, REG_CR_NPU_MIB(10),
> +		      rmem->base + NPU_EN7581_FIRMWARE_RV32_MAX_SIZE);
> +	airoha_npu_wr(npu, REG_CR_NPU_MIB(11), 0x40000); /* SRAM 256K */
> +	airoha_npu_wr(npu, REG_CR_NPU_MIB(12), 0);
> +	airoha_npu_wr(npu, REG_CR_NPU_MIB(21), 1);
> +	msleep(100);
> +
> +	/* setting booting address */
> +	for (i = 0; i < AIROHA_NPU_NUM_CORES; i++)
> +		airoha_npu_wr(npu, REG_CR_BOOT_BASE(i), rmem->base);
> +	usleep_range(1000, 2000);
> +
> +	/* enable NPU cores */
> +	/* do not start core3 since it is used for WiFi offloading */
> +	airoha_npu_wr(npu, REG_CR_BOOT_CONFIG, 0xf7);
> +	airoha_npu_wr(npu, REG_CR_BOOT_TRIGGER, 0x1);
> +	msleep(100);
> +
> +	return npu;
> +
> +error_put_dev:
> +	put_device(&npu->pdev->dev);

Missing platform_device_put()

> +error_of_node_put:
> +	of_node_put(npu->np);
> +
> +	return ERR_PTR(err);
> +}
> +
> +void airoha_npu_deinit(struct airoha_npu *npu)
> +{
> +	int i;
> +
> +	for (i = 0; i < ARRAY_SIZE(npu->cores); i++)
> +		cancel_work_sync(&npu->cores[i].wdt_work);
> +

Leaking device put.

> +	put_device(&npu->pdev->dev);
> +	of_node_put(npu->np);

Best regards,
Krzysztof

Re: [PATCH net-next v3 13/16] net: airoha: Introduce PPE initialization via NPU

[Lorenzo Bianconi]

[-- Attachment #1: Type: text/plain, Size: 5292 bytes --]

> On Sun, Feb 09, 2025 at 01:09:06PM +0100, Lorenzo Bianconi wrote:
> > +static irqreturn_t airoha_npu_wdt_handler(int irq, void *core_instance)
> > +{
> > +	struct airoha_npu_core *core = core_instance;
> > +	struct airoha_npu *npu = core->npu;
> > +	int c = core - &npu->cores[0];
> > +	u32 val;
> > +
> > +	airoha_npu_rmw(npu, REG_WDT_TIMER_CTRL(c), 0, WDT_INTR_MASK);
> > +	val = airoha_npu_rr(npu, REG_WDT_TIMER_CTRL(c));
> > +	if (FIELD_GET(WDT_EN_MASK, val))
> > +		schedule_work(&core->wdt_work);
> > +
> > +	return IRQ_HANDLED;
> > +}
> > +
> > +struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
> > +{
> > +	struct reserved_mem *rmem;
> > +	int i, irq, err = -ENODEV;
> > +	struct airoha_npu *npu;
> > +	struct device_node *np;
> > +
> > +	npu = devm_kzalloc(eth->dev, sizeof(*npu), GFP_KERNEL);
> > +	if (!npu)
> > +		return ERR_PTR(-ENOMEM);
> > +
> > +	npu->np = of_parse_phandle(eth->dev->of_node, "airoha,npu", 0);
> > +	if (!npu->np)
> > +		return ERR_PTR(-ENODEV);
> 
> Why? The property is not required, so how can missing property fail the
> probe?

similar to mtk_wed device, airoha_npu is not modeled as a standalone driver,
but it is part of the airoha_eth driver. If you think it is better, I can
rework it implementing a dedicated driver for it. What do you think?

> 
> This is also still unnecessary ABI break without explanation/reasoning.

At the moment if airoha_npu_init() fails (e.g. if the npu node is not present),
it will not cause any failure in airoha_hw_init() (so in the core ethernet
driver probing).

> 
> > +
> > +	npu->pdev = of_find_device_by_node(npu->np);
> > +	if (!npu->pdev)
> > +		goto error_of_node_put;
> 
> You should also add device link and probably try_module_get. See
> qcom,ice (patch for missing try_module_get is on the lists).

thx for the pointer, I will take a look to it.

> 
> > +
> > +	get_device(&npu->pdev->dev);
> 
> Why? of_find_device_by_node() does it.

ack, I will fix it.

> 
> > +
> > +	npu->base = devm_platform_ioremap_resource(npu->pdev, 0);
> > +	if (IS_ERR(npu->base))
> > +		goto error_put_dev;
> > +
> > +	np = of_parse_phandle(npu->np, "memory-region", 0);
> > +	if (!np)
> > +		goto error_put_dev;
> > +
> > +	rmem = of_reserved_mem_lookup(np);
> > +	of_node_put(np);
> > +
> > +	if (!rmem)
> > +		goto error_put_dev;
> > +
> > +	irq = platform_get_irq(npu->pdev, 0);
> > +	if (irq < 0) {
> > +		err = irq;
> > +		goto error_put_dev;
> > +	}
> > +
> > +	err = devm_request_irq(&npu->pdev->dev, irq, airoha_npu_mbox_handler,
> > +			       IRQF_SHARED, "airoha-npu-mbox", npu);
> > +	if (err)
> > +		goto error_put_dev;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(npu->cores); i++) {
> > +		struct airoha_npu_core *core = &npu->cores[i];
> > +
> > +		spin_lock_init(&core->lock);
> > +		core->npu = npu;
> > +
> > +		irq = platform_get_irq(npu->pdev, i + 1);
> > +		if (irq < 0) {
> > +			err = irq;
> > +			goto error_put_dev;
> > +		}
> 
> This is all confusing. Why are you requesting IRQs for other - the npu -
> device? That device driver is responsible for its interrupts, not you
> here. This breaks encapsulation. And what do you do if the other device
> starts handling interrupts on its own? This is really unexpected to see
> here.

As pointed out above, there is no other driver for airoha_npu at the moment,
but I am fine to implement it.

Regards,
Lorenzo

> 
> > +
> > +		err = devm_request_irq(&npu->pdev->dev, irq,
> > +				       airoha_npu_wdt_handler, IRQF_SHARED,
> > +				       "airoha-npu-wdt", core);
> > +		if (err)
> > +			goto error_put_dev;
> > +
> > +		INIT_WORK(&core->wdt_work, airoha_npu_wdt_work);
> > +	}
> > +
> > +	if (dma_set_coherent_mask(&npu->pdev->dev, 0xbfffffff))
> > +		dev_err(&npu->pdev->dev,
> > +			"failed coherent DMA configuration\n");
> > +
> > +	err = airoha_npu_run_firmware(npu, rmem);
> > +	if (err)
> > +		goto error_put_dev;
> > +
> > +	airoha_npu_wr(npu, REG_CR_NPU_MIB(10),
> > +		      rmem->base + NPU_EN7581_FIRMWARE_RV32_MAX_SIZE);
> > +	airoha_npu_wr(npu, REG_CR_NPU_MIB(11), 0x40000); /* SRAM 256K */
> > +	airoha_npu_wr(npu, REG_CR_NPU_MIB(12), 0);
> > +	airoha_npu_wr(npu, REG_CR_NPU_MIB(21), 1);
> > +	msleep(100);
> > +
> > +	/* setting booting address */
> > +	for (i = 0; i < AIROHA_NPU_NUM_CORES; i++)
> > +		airoha_npu_wr(npu, REG_CR_BOOT_BASE(i), rmem->base);
> > +	usleep_range(1000, 2000);
> > +
> > +	/* enable NPU cores */
> > +	/* do not start core3 since it is used for WiFi offloading */
> > +	airoha_npu_wr(npu, REG_CR_BOOT_CONFIG, 0xf7);
> > +	airoha_npu_wr(npu, REG_CR_BOOT_TRIGGER, 0x1);
> > +	msleep(100);
> > +
> > +	return npu;
> > +
> > +error_put_dev:
> > +	put_device(&npu->pdev->dev);
> 
> Missing platform_device_put()
> 
> > +error_of_node_put:
> > +	of_node_put(npu->np);
> > +
> > +	return ERR_PTR(err);
> > +}
> > +
> > +void airoha_npu_deinit(struct airoha_npu *npu)
> > +{
> > +	int i;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(npu->cores); i++)
> > +		cancel_work_sync(&npu->cores[i].wdt_work);
> > +
> 
> Leaking device put.
> 
> > +	put_device(&npu->pdev->dev);
> > +	of_node_put(npu->np);
> 
> Best regards,
> Krzysztof
> 

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

Re: [PATCH net-next v3 13/16] net: airoha: Introduce PPE initialization via NPU

[Krzysztof Kozlowski]

On 11/02/2025 17:31, Lorenzo Bianconi wrote:
>> On Sun, Feb 09, 2025 at 01:09:06PM +0100, Lorenzo Bianconi wrote:
>>> +static irqreturn_t airoha_npu_wdt_handler(int irq, void *core_instance)
>>> +{
>>> +	struct airoha_npu_core *core = core_instance;
>>> +	struct airoha_npu *npu = core->npu;
>>> +	int c = core - &npu->cores[0];
>>> +	u32 val;
>>> +
>>> +	airoha_npu_rmw(npu, REG_WDT_TIMER_CTRL(c), 0, WDT_INTR_MASK);
>>> +	val = airoha_npu_rr(npu, REG_WDT_TIMER_CTRL(c));
>>> +	if (FIELD_GET(WDT_EN_MASK, val))
>>> +		schedule_work(&core->wdt_work);
>>> +
>>> +	return IRQ_HANDLED;
>>> +}
>>> +
>>> +struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
>>> +{
>>> +	struct reserved_mem *rmem;
>>> +	int i, irq, err = -ENODEV;
>>> +	struct airoha_npu *npu;
>>> +	struct device_node *np;
>>> +
>>> +	npu = devm_kzalloc(eth->dev, sizeof(*npu), GFP_KERNEL);
>>> +	if (!npu)
>>> +		return ERR_PTR(-ENOMEM);
>>> +
>>> +	npu->np = of_parse_phandle(eth->dev->of_node, "airoha,npu", 0);
>>> +	if (!npu->np)
>>> +		return ERR_PTR(-ENODEV);
>>
>> Why? The property is not required, so how can missing property fail the
>> probe?
> 
> similar to mtk_wed device, airoha_npu is not modeled as a standalone driver,
> but it is part of the airoha_eth driver. If you think it is better, I can
> rework it implementing a dedicated driver for it. What do you think?


Whether it is separate or not, does not matter. Behavior in both cases
would be the same so does not answer my question. But below does however:

> 
>>
>> This is also still unnecessary ABI break without explanation/reasoning.
> 
> At the moment if airoha_npu_init() fails (e.g. if the npu node is not present),
> it will not cause any failure in airoha_hw_init() (so in the core ethernet
> driver probing).


Indeed, it will fail airoha_ppe_init() but airoha_ppe_init() is not
fatal. You will have dmesg errors though, so this should be probably
dev_warn.

> 
>>
>>> +
>>> +	npu->pdev = of_find_device_by_node(npu->np);
>>> +	if (!npu->pdev)
>>> +		goto error_of_node_put;
>>
>> You should also add device link and probably try_module_get. See
>> qcom,ice (patch for missing try_module_get is on the lists).
> 
> thx for the pointer, I will take a look to it.
> 
>>
>>> +
>>> +	get_device(&npu->pdev->dev);
>>
>> Why? of_find_device_by_node() does it.
> 
> ack, I will fix it.
> 
>>
>>> +
>>> +	npu->base = devm_platform_ioremap_resource(npu->pdev, 0);
>>> +	if (IS_ERR(npu->base))
>>> +		goto error_put_dev;
>>> +
>>> +	np = of_parse_phandle(npu->np, "memory-region", 0);
>>> +	if (!np)
>>> +		goto error_put_dev;
>>> +
>>> +	rmem = of_reserved_mem_lookup(np);
>>> +	of_node_put(np);
>>> +
>>> +	if (!rmem)
>>> +		goto error_put_dev;
>>> +
>>> +	irq = platform_get_irq(npu->pdev, 0);
>>> +	if (irq < 0) {
>>> +		err = irq;
>>> +		goto error_put_dev;
>>> +	}
>>> +
>>> +	err = devm_request_irq(&npu->pdev->dev, irq, airoha_npu_mbox_handler,
>>> +			       IRQF_SHARED, "airoha-npu-mbox", npu);
>>> +	if (err)
>>> +		goto error_put_dev;
>>> +
>>> +	for (i = 0; i < ARRAY_SIZE(npu->cores); i++) {
>>> +		struct airoha_npu_core *core = &npu->cores[i];
>>> +
>>> +		spin_lock_init(&core->lock);
>>> +		core->npu = npu;
>>> +
>>> +		irq = platform_get_irq(npu->pdev, i + 1);
>>> +		if (irq < 0) {
>>> +			err = irq;
>>> +			goto error_put_dev;
>>> +		}
>>
>> This is all confusing. Why are you requesting IRQs for other - the npu -
>> device? That device driver is responsible for its interrupts, not you
>> here. This breaks encapsulation. And what do you do if the other device
>> starts handling interrupts on its own? This is really unexpected to see
>> here.
> 
> As pointed out above, there is no other driver for airoha_npu at the moment,
> but I am fine to implement it.

I see. The second driver is there - syscon - just provided by MFD core.
It also looks like the NPU is some sort of mailbox, so maybe NPU should
not have been syscon in the first place, but mailbox?



Best regards,
Krzysztof

[PATCH net-next v3 14/16] net: airoha: Introduce flowtable offload support

[Lorenzo Bianconi]

Introduce netfilter flowtable integration in order to allow airoha_eth
driver to offload 5-tuple flower rules learned by the PPE module if the
user accelerates them using a nft configuration similar to the one reported
below:

table inet filter {
	flowtable ft {
		hook ingress priority filter
		devices = { lan1, lan2, lan3, lan4, eth1 }
		flags offload;
	}
	chain forward {
		type filter hook forward priority filter; policy accept;
		meta l4proto { tcp, udp } flow add @ft
	}
}

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c |  56 ++-
 drivers/net/ethernet/airoha/airoha_eth.h | 242 ++++++++++-
 drivers/net/ethernet/airoha/airoha_npu.c |  50 +++
 drivers/net/ethernet/airoha/airoha_ppe.c | 690 ++++++++++++++++++++++++++++++-
 4 files changed, 1029 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index bc8619748f0ac4fdffd78c37e36c4a021514d5fd..46ef80b190e696f8341f262234e2d63e1372e715 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -5,10 +5,8 @@
  */
 #include <linux/tcp.h>
 #include <linux/u64_stats_sync.h>
-#include <net/dsa.h>
 #include <net/dst_metadata.h>
 #include <net/page_pool/helpers.h>
-#include <net/pkt_cls.h>
 #include <uapi/linux/ppp_defs.h>
 
 #include "airoha_regs.h"
@@ -616,6 +614,7 @@ static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
 	while (done < budget) {
 		struct airoha_queue_entry *e = &q->entry[q->tail];
 		struct airoha_qdma_desc *desc = &q->desc[q->tail];
+		u32 hash, reason, msg1 = le32_to_cpu(desc->msg1);
 		dma_addr_t dma_addr = le32_to_cpu(desc->addr);
 		u32 desc_ctrl = le32_to_cpu(desc->ctrl);
 		struct airoha_gdm_port *port;
@@ -678,6 +677,15 @@ static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
 						  &port->dsa_meta[sptag]->dst);
 		}
 
+		hash = FIELD_GET(AIROHA_RXD4_FOE_ENTRY, msg1);
+		if (hash != AIROHA_RXD4_FOE_ENTRY)
+			skb_set_hash(skb, jhash_1word(hash, 0),
+				     PKT_HASH_TYPE_L4);
+
+		reason = FIELD_GET(AIROHA_RXD4_PPE_CPU_REASON, msg1);
+		if (reason == PPE_CPU_REASON_HIT_UNBIND_RATE_REACHED)
+			airoha_ppe_check_skb(eth->ppe, hash);
+
 		napi_gro_receive(&q->napi, skb);
 
 		done++;
@@ -2166,6 +2174,47 @@ static int airoha_tc_htb_alloc_leaf_queue(struct airoha_gdm_port *port,
 	return 0;
 }
 
+static int airoha_dev_setup_tc_block(struct airoha_gdm_port *port,
+				     struct flow_block_offload *f)
+{
+	flow_setup_cb_t *cb = airoha_ppe_setup_tc_block_cb;
+	static LIST_HEAD(block_cb_list);
+	struct flow_block_cb *block_cb;
+
+	if (f->binder_type != FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS)
+		return -EOPNOTSUPP;
+
+	f->driver_block_list = &block_cb_list;
+	switch (f->command) {
+	case FLOW_BLOCK_BIND:
+		block_cb = flow_block_cb_lookup(f->block, cb, port->dev);
+		if (block_cb) {
+			flow_block_cb_incref(block_cb);
+			return 0;
+		}
+		block_cb = flow_block_cb_alloc(cb, port->dev, port->dev, NULL);
+		if (IS_ERR(block_cb))
+			return PTR_ERR(block_cb);
+
+		flow_block_cb_incref(block_cb);
+		flow_block_cb_add(block_cb, f);
+		list_add_tail(&block_cb->driver_list, &block_cb_list);
+		return 0;
+	case FLOW_BLOCK_UNBIND:
+		block_cb = flow_block_cb_lookup(f->block, cb, port->dev);
+		if (!block_cb)
+			return -ENOENT;
+
+		if (!flow_block_cb_decref(block_cb)) {
+			flow_block_cb_remove(block_cb, f);
+			list_del(&block_cb->driver_list);
+		}
+		return 0;
+	default:
+		return -EOPNOTSUPP;
+	}
+}
+
 static void airoha_tc_remove_htb_queue(struct airoha_gdm_port *port, int queue)
 {
 	struct net_device *dev = port->dev;
@@ -2249,6 +2298,9 @@ static int airoha_dev_tc_setup(struct net_device *dev, enum tc_setup_type type,
 		return airoha_tc_setup_qdisc_ets(port, type_data);
 	case TC_SETUP_QDISC_HTB:
 		return airoha_tc_setup_qdisc_htb(port, type_data);
+	case TC_SETUP_BLOCK:
+	case TC_SETUP_FT:
+		return airoha_dev_setup_tc_block(port, type_data);
 	default:
 		return -EOPNOTSUPP;
 	}
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index d882b8c74c713209d15603311e4837f1502bbbfd..c51fa6713be07415a56ba4da2456d738d9a8f2af 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -16,6 +16,8 @@
 #include <linux/of_platform.h>
 #include <linux/platform_device.h>
 #include <linux/reset.h>
+#include <net/dsa.h>
+#include <net/pkt_cls.h>
 
 #define AIROHA_NPU_NUM_CORES		8
 #define AIROHA_MAX_NUM_GDM_PORTS	4
@@ -209,8 +211,224 @@ struct airoha_hw_stats {
 	u64 rx_len[7];
 };
 
+enum {
+	PPE_CPU_REASON_HIT_UNBIND_RATE_REACHED = 0x0f,
+};
+
+enum {
+	AIROHA_FOE_STATE_INVALID,
+	AIROHA_FOE_STATE_UNBIND,
+	AIROHA_FOE_STATE_BIND,
+	AIROHA_FOE_STATE_FIN
+};
+
+enum {
+	PPE_PKT_TYPE_IPV4_HNAPT = 0,
+	PPE_PKT_TYPE_IPV4_ROUTE = 1,
+	PPE_PKT_TYPE_BRIDGE = 2,
+	PPE_PKT_TYPE_IPV4_DSLITE = 3,
+	PPE_PKT_TYPE_IPV6_ROUTE_3T = 4,
+	PPE_PKT_TYPE_IPV6_ROUTE_5T = 5,
+	PPE_PKT_TYPE_IPV6_6RD = 7,
+};
+
+#define AIROHA_FOE_MAC_PPPOE_ID		GENMASK(15, 0)
+#define AIROHA_FOE_MAC_SMAC_ID		GENMASK(20, 16)
+
+struct airoha_foe_mac_info_common {
+	u16 vlan1;
+	u16 etype;
+
+	u32 dest_mac_hi;
+
+	u16 vlan2;
+	u16 dest_mac_lo;
+
+	u32 src_mac_hi;
+};
+
+struct airoha_foe_mac_info {
+	struct airoha_foe_mac_info_common common;
+
+	u16 pppoe_id;
+	u16 src_mac_lo;
+};
+
+#define AIROHA_FOE_IB1_UNBIND_TIMESTAMP		GENMASK(7, 0)
+#define AIROHA_FOE_IB1_UNBIND_PACKETS		GENMASK(23, 8)
+#define AIROHA_FOE_IB1_UNBIND_PREBIND		BIT(24)
+
+#define AIROHA_FOE_IB1_BIND_TIMESTAMP		GENMASK(14, 0)
+#define AIROHA_FOE_IB1_BIND_KEEPALIVE		BIT(15)
+#define AIROHA_FOE_IB1_BIND_VLAN_LAYER		GENMASK(18, 16)
+#define AIROHA_FOE_IB1_BIND_PPPOE		BIT(19)
+#define AIROHA_FOE_IB1_BIND_VLAN_TAG		BIT(20)
+#define AIROHA_FOE_IB1_BIND_PKT_SAMPLE		BIT(21)
+#define AIROHA_FOE_IB1_BIND_CACHE		BIT(22)
+#define AIROHA_FOE_IB1_BIND_TUNNEL_DECAP	BIT(23)
+#define AIROHA_FOE_IB1_BIND_TTL			BIT(24)
+#define AIROHA_FOE_IB1_PACKET_TYPE		GENMASK(27, 25)
+#define AIROHA_FOE_IB1_STATE			GENMASK(29, 28)
+#define AIROHA_FOE_IB1_UDP			BIT(30)
+#define AIROHA_FOE_IB1_STATIC			BIT(31)
+
+#define AIROHA_FOE_IB2_NBQ			GENMASK(4, 0)
+#define AIROHA_FOE_IB2_PSE_PORT			GENMASK(8, 5)
+#define AIROHA_FOE_IB2_PSE_QOS			BIT(9)
+#define AIROHA_FOE_IB2_FAST_PATH		BIT(10)
+#define AIROHA_FOE_IB2_MULTICAST		BIT(11)
+#define AIROHA_FOE_IB2_PCP			BIT(12)
+#define AIROHA_FOE_IB2_PORT_AG			GENMASK(23, 13)
+#define AIROHA_FOE_IB2_DSCP			GENMASK(31, 24)
+
+#define AIROHA_FOE_TUNNEL_ID			GENMASK(5, 0)
+#define AIROHA_FOE_TUNNEL			BIT(6)
+#define AIROHA_FOE_DPI				BIT(7)
+#define AIROHA_FOE_QID				GENMASK(10, 8)
+#define AIROHA_FOE_CHANNEL			GENMASK(15, 11)
+#define AIROHA_FOE_SHAPER_ID			GENMASK(23, 16)
+#define AIROHA_FOE_ACTDP			GENMASK(31, 24)
+
+struct airoha_foe_bridge {
+	u32 dest_mac_hi;
+
+	u16 src_mac_hi;
+	u16 dest_mac_lo;
+
+	u32 src_mac_lo;
+
+	u32 ib2;
+
+	u32 rsv[5];
+
+	u32 data;
+
+	struct airoha_foe_mac_info l2;
+};
+
+struct airoha_foe_ipv4_tuple {
+	u32 src_ip;
+	u32 dest_ip;
+	union {
+		struct {
+			u16 dest_port;
+			u16 src_port;
+		};
+		struct {
+			u8 protocol;
+			u8 _pad[3]; /* fill with 0xa5a5a5 */
+		};
+		u32 ports;
+	};
+};
+
+struct airoha_foe_ipv4 {
+	struct airoha_foe_ipv4_tuple orig_tuple;
+
+	u32 ib2;
+
+	struct airoha_foe_ipv4_tuple new_tuple;
+
+	u32 rsv[2];
+
+	u32 data;
+
+	struct airoha_foe_mac_info l2;
+};
+
+struct airoha_foe_ipv4_dslite {
+	struct airoha_foe_ipv4_tuple ip4;
+
+	u32 ib2;
+
+	u8 flow_label[3];
+	u8 priority;
+
+	u32 rsv[4];
+
+	u32 data;
+
+	struct airoha_foe_mac_info l2;
+};
+
+struct airoha_foe_ipv6 {
+	u32 src_ip[4];
+	u32 dest_ip[4];
+
+	union {
+		struct {
+			u16 dest_port;
+			u16 src_port;
+		};
+		struct {
+			u8 protocol;
+			u8 pad[3];
+		};
+		u32 ports;
+	};
+
+	u32 data;
+
+	u32 ib2;
+
+	struct airoha_foe_mac_info_common l2;
+};
+
 struct airoha_foe_entry {
-	u8 data[PPE_ENTRY_SIZE];
+	union {
+		struct {
+			u32 ib1;
+			union {
+				struct airoha_foe_bridge bridge;
+				struct airoha_foe_ipv4 ipv4;
+				struct airoha_foe_ipv4_dslite dslite;
+				struct airoha_foe_ipv6 ipv6;
+				DECLARE_FLEX_ARRAY(u32, d);
+			};
+		};
+		u8 data[PPE_ENTRY_SIZE];
+	};
+};
+
+struct airoha_flow_data {
+	struct ethhdr eth;
+
+	union {
+		struct {
+			__be32 src_addr;
+			__be32 dst_addr;
+		} v4;
+
+		struct {
+			struct in6_addr src_addr;
+			struct in6_addr dst_addr;
+		} v6;
+	};
+
+	__be16 src_port;
+	__be16 dst_port;
+
+	u16 vlan_in;
+
+	struct {
+		u16 id;
+		__be16 proto;
+		u8 num;
+	} vlan;
+	struct {
+		u16 sid;
+		u8 num;
+	} pppoe;
+};
+
+struct airoha_flow_table_entry {
+	struct hlist_node list;
+
+	struct airoha_foe_entry data;
+	u32 hash;
+
+	struct rhash_head node;
+	unsigned long cookie;
 };
 
 struct airoha_qdma {
@@ -266,11 +484,17 @@ struct airoha_npu {
 	} cores[AIROHA_NPU_NUM_CORES];
 };
 
+#define AIROHA_RXD4_FOE_ENTRY		GENMASK(15, 0)
+#define AIROHA_RXD4_PPE_CPU_REASON	GENMASK(20, 16)
+
 struct airoha_ppe {
 	struct airoha_eth *eth;
 
 	void *foe;
 	dma_addr_t foe_dma;
+
+	struct hlist_head *foe_flow;
+	u16 foe_check_time[PPE_NUM_ENTRIES];
 };
 
 struct airoha_eth {
@@ -281,6 +505,7 @@ struct airoha_eth {
 
 	struct airoha_npu *npu;
 	struct airoha_ppe *ppe;
+	struct rhashtable flow_table;
 
 	struct reset_control_bulk_data rsts[AIROHA_MAX_NUM_RSTS];
 	struct reset_control_bulk_data xsi_rsts[AIROHA_MAX_NUM_XSI_RSTS];
@@ -318,6 +543,12 @@ u32 airoha_rmw(void __iomem *base, u32 offset, u32 mask, u32 val);
 	airoha_rmw((qdma)->regs, (offset), (val), 0)
 
 bool airoha_ppe2_is_enabled(struct airoha_eth *eth);
+void airoha_ppe_check_skb(struct airoha_ppe *ppe, u16 hash);
+int airoha_ppe_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
+				 void *cb_priv);
+u32 airoha_ppe_get_timestamp(struct airoha_ppe *ppe);
+struct airoha_foe_entry *airoha_ppe_foe_get_entry(struct airoha_ppe *ppe,
+						  u32 hash);
 
 #ifdef CONFIG_NET_AIROHA_NPU
 int airoha_ppe_init(struct airoha_eth *eth);
@@ -328,6 +559,8 @@ int airoha_npu_ppe_init(struct airoha_npu *npu);
 int airoha_npu_ppe_deinit(struct airoha_npu *npu);
 int airoha_npu_flush_ppe_sram_entries(struct airoha_npu *npu,
 				      struct airoha_ppe *ppe);
+int airoha_npu_foe_commit_entry(struct airoha_ppe *ppe,
+				struct airoha_foe_entry *e, u32 hash);
 #else
 static inline int airoha_ppe_init(struct airoha_eth *eth)
 {
@@ -362,6 +595,13 @@ static inline int airoha_npu_flush_ppe_sram_entries(struct airoha_npu *npu,
 {
 	return -EOPNOTSUPP;
 }
+
+static inline int airoha_npu_foe_commit_entry(struct airoha_ppe *ppe,
+					      struct airoha_foe_entry *e,
+					      u32 hash)
+{
+	return -EOPNOTSUPP;
+}
 #endif /* CONFIG_NET_AIROHA_NPU */
 
 #endif /* AIROHA_ETH_H */
diff --git a/drivers/net/ethernet/airoha/airoha_npu.c b/drivers/net/ethernet/airoha/airoha_npu.c
index b912ab4c02166d7f9f1aaacb4b5b6b30740a3b6c..f8b2407a5dd2086a1a5da469a70720b8b0a22a7f 100644
--- a/drivers/net/ethernet/airoha/airoha_npu.c
+++ b/drivers/net/ethernet/airoha/airoha_npu.c
@@ -336,6 +336,56 @@ static irqreturn_t airoha_npu_wdt_handler(int irq, void *core_instance)
 	return IRQ_HANDLED;
 }
 
+int airoha_npu_foe_commit_entry(struct airoha_ppe *ppe,
+				struct airoha_foe_entry *e, u32 hash)
+{
+	struct airoha_foe_entry *hwe = ppe->foe + hash * sizeof(*hwe);
+	u16 ts = airoha_ppe_get_timestamp(ppe);
+
+	memcpy(&hwe->d, &e->d, sizeof(*hwe) - sizeof(hwe->ib1));
+	wmb();
+
+	e->ib1 &= ~AIROHA_FOE_IB1_BIND_TIMESTAMP;
+	e->ib1 |= FIELD_PREP(AIROHA_FOE_IB1_BIND_TIMESTAMP, ts);
+	hwe->ib1 = e->ib1;
+
+	if (hash < PPE_SRAM_NUM_ENTRIES) {
+		dma_addr_t addr = ppe->foe_dma + hash * sizeof(*hwe);
+		struct ppe_mbox_data ppe_data = {
+			.func_type = NPU_OP_SET,
+			.func_id = PPE_FUNC_SET_WAIT_API,
+			.set_info = {
+				.data = addr,
+				.size = sizeof(*hwe),
+			},
+		};
+		struct airoha_eth *eth = ppe->eth;
+		bool ppe2;
+		int err;
+
+		ppe2 = airoha_ppe2_is_enabled(ppe->eth) &&
+		       hash >= PPE1_SRAM_NUM_ENTRIES;
+		ppe_data.set_info.func_id = ppe2 ? PPE2_SRAM_SET_ENTRY
+						 : PPE_SRAM_SET_ENTRY;
+
+		err = airoha_npu_send_msg(eth->npu, NPU_FUNC_PPE, &ppe_data,
+					  sizeof(struct ppe_mbox_data));
+		if (err)
+			return err;
+
+		ppe_data.set_info.func_id = PPE_SRAM_SET_VAL;
+		ppe_data.set_info.data = hash;
+		ppe_data.set_info.size = sizeof(u32);
+
+		err = airoha_npu_send_msg(eth->npu, NPU_FUNC_PPE, &ppe_data,
+					  sizeof(struct ppe_mbox_data));
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
 struct airoha_npu *airoha_npu_init(struct airoha_eth *eth)
 {
 	struct reserved_mem *rmem;
diff --git a/drivers/net/ethernet/airoha/airoha_ppe.c b/drivers/net/ethernet/airoha/airoha_ppe.c
index b6af9f8fe72d10aae118c2fa2fb7a15dabaa5ebf..54f63b017bcfb0af840d0996e4b099371fcaa0e6 100644
--- a/drivers/net/ethernet/airoha/airoha_ppe.c
+++ b/drivers/net/ethernet/airoha/airoha_ppe.c
@@ -4,14 +4,678 @@
  * Author: Lorenzo Bianconi <lorenzo@kernel.org>
  */
 
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/rhashtable.h>
+#include <net/ipv6.h>
+
 #include "airoha_regs.h"
 #include "airoha_eth.h"
 
+static DEFINE_MUTEX(flow_offload_mutex);
+static DEFINE_SPINLOCK(ppe_lock);
+
 bool airoha_ppe2_is_enabled(struct airoha_eth *eth)
 {
 	return airoha_fe_rr(eth, REG_PPE_GLO_CFG(1)) & PPE_GLO_CFG_EN_MASK;
 }
 
+static const struct rhashtable_params airoha_flow_table_params = {
+	.head_offset = offsetof(struct airoha_flow_table_entry, node),
+	.key_offset = offsetof(struct airoha_flow_table_entry, cookie),
+	.key_len = sizeof(unsigned long),
+	.automatic_shrinking = true,
+};
+
+u32 airoha_ppe_get_timestamp(struct airoha_ppe *ppe)
+{
+	u16 timestamp = airoha_fe_rr(ppe->eth, REG_FE_FOE_TS);
+
+	return FIELD_GET(AIROHA_FOE_IB1_BIND_TIMESTAMP, timestamp);
+}
+
+static void airoha_ppe_flow_mangle_eth(const struct flow_action_entry *act, void *eth)
+{
+	void *dest = eth + act->mangle.offset;
+	const void *src = &act->mangle.val;
+
+	if (act->mangle.offset > 8)
+		return;
+
+	if (act->mangle.mask == 0xffff) {
+		src += 2;
+		dest += 2;
+	}
+
+	memcpy(dest, src, act->mangle.mask ? 2 : 4);
+}
+
+static int airoha_ppe_flow_mangle_ports(const struct flow_action_entry *act,
+					struct airoha_flow_data *data)
+{
+	u32 val = be32_to_cpu((__force __be32)act->mangle.val);
+
+	switch (act->mangle.offset) {
+	case 0:
+		if ((__force __be32)act->mangle.mask == ~cpu_to_be32(0xffff))
+			data->dst_port = cpu_to_be16(val);
+		else
+			data->src_port = cpu_to_be16(val >> 16);
+		break;
+	case 2:
+		data->dst_port = cpu_to_be16(val);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int airoha_ppe_flow_mangle_ipv4(const struct flow_action_entry *act,
+				       struct airoha_flow_data *data)
+{
+	__be32 *dest;
+
+	switch (act->mangle.offset) {
+	case offsetof(struct iphdr, saddr):
+		dest = &data->v4.src_addr;
+		break;
+	case offsetof(struct iphdr, daddr):
+		dest = &data->v4.dst_addr;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	memcpy(dest, &act->mangle.val, sizeof(u32));
+
+	return 0;
+}
+
+static int airoha_get_dsa_port(struct net_device **dev)
+{
+#if IS_ENABLED(CONFIG_NET_DSA)
+	struct dsa_port *dp = dsa_port_from_netdev(*dev);
+
+	if (IS_ERR(dp))
+		return -ENODEV;
+
+	*dev = dsa_port_to_conduit(dp);
+	return dp->index;
+#else
+	return -ENODEV;
+#endif
+}
+
+static int airoha_ppe_foe_entry_prepare(struct airoha_foe_entry *hwe,
+					struct net_device *dev, int type,
+					int l4proto, u8 *src_mac, u8 *dest_mac)
+{
+	int dsa_port = airoha_get_dsa_port(&dev);
+	struct airoha_foe_mac_info_common *l2;
+	u32 data, ports_pad, val;
+
+	memset(hwe, 0, sizeof(*hwe));
+
+	val = FIELD_PREP(AIROHA_FOE_IB1_STATE, AIROHA_FOE_STATE_BIND) |
+	      FIELD_PREP(AIROHA_FOE_IB1_PACKET_TYPE, type) |
+	      FIELD_PREP(AIROHA_FOE_IB1_UDP, l4proto == IPPROTO_UDP) |
+	      AIROHA_FOE_IB1_BIND_TTL;
+	hwe->ib1 = val;
+
+	val = FIELD_PREP(AIROHA_FOE_IB2_PORT_AG, 0x1f);
+	if (dsa_port >= 0)
+		val |= FIELD_PREP(AIROHA_FOE_IB2_NBQ, dsa_port);
+	if (dev) {
+		struct airoha_gdm_port *port = netdev_priv(dev);
+		u8 pse_port;
+
+		pse_port = port->id == 4 ? FE_PSE_PORT_GDM4 : port->id;
+		val |= FIELD_PREP(AIROHA_FOE_IB2_PSE_PORT, pse_port);
+	}
+
+	/* FIXME: implement QoS support setting pse_port to 2 (loopback)
+	 * for uplink and setting qos bit in ib2
+	 */
+
+	if (is_multicast_ether_addr(dest_mac))
+		val |= AIROHA_FOE_IB2_MULTICAST;
+
+	ports_pad = 0xa5a5a500 | (l4proto & 0xff);
+	if (type == PPE_PKT_TYPE_IPV4_ROUTE)
+		hwe->ipv4.orig_tuple.ports = ports_pad;
+	if (type == PPE_PKT_TYPE_IPV6_ROUTE_3T)
+		hwe->ipv6.ports = ports_pad;
+
+	data = FIELD_PREP(AIROHA_FOE_SHAPER_ID, 0x7f);
+	if (type == PPE_PKT_TYPE_BRIDGE) {
+		hwe->bridge.dest_mac_hi = get_unaligned_be32(dest_mac);
+		hwe->bridge.dest_mac_lo = get_unaligned_be16(dest_mac + 4);
+		hwe->bridge.src_mac_hi = get_unaligned_be16(src_mac);
+		hwe->bridge.src_mac_lo = get_unaligned_be32(src_mac + 2);
+		hwe->bridge.data = data;
+		hwe->bridge.ib2 = val;
+		l2 = &hwe->bridge.l2.common;
+	} else if (type >= PPE_PKT_TYPE_IPV6_ROUTE_3T) {
+		hwe->ipv6.data = data;
+		hwe->ipv6.ib2 = val;
+		l2 = &hwe->ipv6.l2;
+	} else {
+		hwe->ipv4.data = data;
+		hwe->ipv4.ib2 = val;
+		l2 = &hwe->ipv4.l2.common;
+	}
+
+	l2->dest_mac_hi = get_unaligned_be32(dest_mac);
+	l2->dest_mac_lo = get_unaligned_be16(dest_mac + 4);
+	if (type <= PPE_PKT_TYPE_IPV4_DSLITE) {
+		l2->src_mac_hi = get_unaligned_be32(src_mac);
+		hwe->ipv4.l2.src_mac_lo = get_unaligned_be16(src_mac + 4);
+	} else {
+		l2->src_mac_hi = FIELD_PREP(AIROHA_FOE_MAC_SMAC_ID, 0xf);
+	}
+
+	if (dsa_port >= 0)
+		l2->etype = BIT(15) | BIT(dsa_port);
+	else if (type >= PPE_PKT_TYPE_IPV6_ROUTE_3T)
+		l2->etype = ETH_P_IPV6;
+	else
+		l2->etype = ETH_P_IP;
+
+	return 0;
+}
+
+static int airoha_ppe_foe_entry_set_ipv4_tuple(struct airoha_foe_entry *hwe,
+					       struct airoha_flow_data *data,
+					       bool egress)
+{
+	int type = FIELD_GET(AIROHA_FOE_IB1_PACKET_TYPE, hwe->ib1);
+	struct airoha_foe_ipv4_tuple *t;
+
+	switch (type) {
+	case PPE_PKT_TYPE_IPV4_HNAPT:
+		if (egress) {
+			t = &hwe->ipv4.new_tuple;
+			break;
+		}
+		fallthrough;
+	case PPE_PKT_TYPE_IPV4_DSLITE:
+	case PPE_PKT_TYPE_IPV4_ROUTE:
+		t = &hwe->ipv4.orig_tuple;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	t->src_ip = be32_to_cpu(data->v4.src_addr);
+	t->dest_ip = be32_to_cpu(data->v4.dst_addr);
+
+	if (type != PPE_PKT_TYPE_IPV4_ROUTE) {
+		t->src_port = be16_to_cpu(data->src_port);
+		t->dest_port = be16_to_cpu(data->dst_port);
+	}
+
+	return 0;
+}
+
+static int airoha_ppe_foe_entry_set_ipv6_tuple(struct airoha_foe_entry *hwe,
+					       struct airoha_flow_data *data)
+
+{
+	int type = FIELD_GET(AIROHA_FOE_IB1_PACKET_TYPE, hwe->ib1);
+	u32 *src, *dest;
+
+	switch (type) {
+	case PPE_PKT_TYPE_IPV6_ROUTE_5T:
+	case PPE_PKT_TYPE_IPV6_6RD:
+		hwe->ipv6.src_port = be16_to_cpu(data->src_port);
+		hwe->ipv6.dest_port = be16_to_cpu(data->dst_port);
+		fallthrough;
+	case PPE_PKT_TYPE_IPV6_ROUTE_3T:
+		src = hwe->ipv6.src_ip;
+		dest = hwe->ipv6.dest_ip;
+		break;
+	default:
+		WARN_ON_ONCE(1);
+		return -EINVAL;
+	}
+
+	ipv6_addr_be32_to_cpu(src, data->v6.src_addr.s6_addr32);
+	ipv6_addr_be32_to_cpu(dest, data->v6.dst_addr.s6_addr32);
+
+	return 0;
+}
+
+static u32 airoha_ppe_foe_get_entry_hash(struct airoha_foe_entry *hwe)
+{
+	int type = FIELD_GET(AIROHA_FOE_IB1_PACKET_TYPE, hwe->ib1);
+	u32 hash, hv1, hv2, hv3;
+
+	switch (type) {
+	case PPE_PKT_TYPE_IPV4_ROUTE:
+	case PPE_PKT_TYPE_IPV4_HNAPT:
+		hv1 = hwe->ipv4.orig_tuple.ports;
+		hv2 = hwe->ipv4.orig_tuple.dest_ip;
+		hv3 = hwe->ipv4.orig_tuple.src_ip;
+		break;
+	case PPE_PKT_TYPE_IPV6_ROUTE_3T:
+	case PPE_PKT_TYPE_IPV6_ROUTE_5T:
+		hv1 = hwe->ipv6.src_ip[3] ^ hwe->ipv6.dest_ip[3];
+		hv1 ^= hwe->ipv6.ports;
+
+		hv2 = hwe->ipv6.src_ip[2] ^ hwe->ipv6.dest_ip[2];
+		hv2 ^= hwe->ipv6.dest_ip[0];
+
+		hv3 = hwe->ipv6.src_ip[1] ^ hwe->ipv6.dest_ip[1];
+		hv3 ^= hwe->ipv6.src_ip[0];
+		break;
+	case PPE_PKT_TYPE_IPV4_DSLITE:
+	case PPE_PKT_TYPE_IPV6_6RD:
+	default:
+		WARN_ON_ONCE(1);
+		return PPE_HASH_MASK;
+	}
+
+	hash = (hv1 & hv2) | ((~hv1) & hv3);
+	hash = (hash >> 24) | ((hash & 0xffffff) << 8);
+	hash ^= hv1 ^ hv2 ^ hv3;
+	hash ^= hash >> 16;
+	hash &= PPE_NUM_ENTRIES - 1;
+
+	return hash;
+}
+
+struct airoha_foe_entry *airoha_ppe_foe_get_entry(struct airoha_ppe *ppe,
+						  u32 hash)
+{
+	if (hash < PPE_SRAM_NUM_ENTRIES) {
+		u32 *hwe = ppe->foe + hash * sizeof(struct airoha_foe_entry);
+		struct airoha_eth *eth = ppe->eth;
+		bool ppe2;
+		u32 val;
+		int i;
+
+		ppe2 = airoha_ppe2_is_enabled(ppe->eth) &&
+		       hash >= PPE1_SRAM_NUM_ENTRIES;
+		airoha_fe_wr(ppe->eth, REG_PPE_RAM_CTRL(ppe2),
+			     FIELD_PREP(PPE_SRAM_CTRL_ENTRY_MASK, hash) |
+			     PPE_SRAM_CTRL_REQ_MASK);
+		if (read_poll_timeout_atomic(airoha_fe_rr, val,
+					     val & PPE_SRAM_CTRL_ACK_MASK,
+					     10, 100, false, eth,
+					     REG_PPE_RAM_CTRL(ppe2)))
+			return NULL;
+
+		for (i = 0; i < sizeof(struct airoha_foe_entry) / 4; i++)
+			hwe[i] = airoha_fe_rr(eth,
+					      REG_PPE_RAM_ENTRY(ppe2, i));
+	}
+
+	return ppe->foe + hash * sizeof(struct airoha_foe_entry);
+}
+
+static bool airoha_ppe_foe_compare_entry(struct airoha_flow_table_entry *e,
+					 struct airoha_foe_entry *hwe)
+{
+	int type = FIELD_GET(AIROHA_FOE_IB1_PACKET_TYPE, e->data.ib1), len;
+
+	if ((hwe->ib1 ^ e->data.ib1) & AIROHA_FOE_IB1_UDP)
+		return false;
+
+	if (type > PPE_PKT_TYPE_IPV4_DSLITE)
+		len = offsetof(struct airoha_foe_entry, ipv6.data);
+	else
+		len = offsetof(struct airoha_foe_entry, ipv4.ib2);
+
+	return !memcmp(&e->data.d, &hwe->d, len - sizeof(hwe->ib1));
+}
+
+static void airoha_ppe_foe_insert_entry(struct airoha_ppe *ppe, u32 hash)
+{
+	struct airoha_flow_table_entry *e;
+	struct airoha_foe_entry *hwe;
+	struct hlist_node *n;
+	u32 index;
+
+	spin_lock_bh(&ppe_lock);
+
+	hwe = airoha_ppe_foe_get_entry(ppe, hash);
+	if (!hwe)
+		goto unlock;
+
+	if (FIELD_GET(AIROHA_FOE_IB1_STATE, hwe->ib1) == AIROHA_FOE_STATE_BIND)
+		goto unlock;
+
+	index = airoha_ppe_foe_get_entry_hash(hwe);
+	hlist_for_each_entry_safe(e, n, &ppe->foe_flow[index], list) {
+		if (airoha_ppe_foe_compare_entry(e, hwe)) {
+			airoha_npu_foe_commit_entry(ppe, &e->data, hash);
+			e->hash = hash;
+			break;
+		}
+	}
+unlock:
+	spin_unlock_bh(&ppe_lock);
+}
+
+static int airoha_ppe_foe_flow_commit_entry(struct airoha_ppe *ppe,
+					    struct airoha_flow_table_entry *e)
+{
+	u32 hash = airoha_ppe_foe_get_entry_hash(&e->data);
+
+	e->hash = 0xffff;
+
+	spin_lock_bh(&ppe_lock);
+	hlist_add_head(&e->list, &ppe->foe_flow[hash]);
+	spin_unlock_bh(&ppe_lock);
+
+	return 0;
+}
+
+static void airoha_ppe_foe_flow_remove_entry(struct airoha_ppe *ppe,
+					     struct airoha_flow_table_entry *e)
+{
+	spin_lock_bh(&ppe_lock);
+
+	hlist_del_init(&e->list);
+	if (e->hash != 0xffff) {
+		e->data.ib1 &= ~AIROHA_FOE_IB1_STATE;
+		e->data.ib1 |= FIELD_PREP(AIROHA_FOE_IB1_STATE,
+					  AIROHA_FOE_STATE_INVALID);
+		airoha_npu_foe_commit_entry(ppe, &e->data, e->hash);
+		e->hash = 0xffff;
+	}
+
+	spin_unlock_bh(&ppe_lock);
+}
+
+static int airoha_ppe_flow_offload_replace(struct airoha_gdm_port *port,
+					   struct flow_cls_offload *f)
+{
+	struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+	struct airoha_eth *eth = port->qdma->eth;
+	struct airoha_flow_table_entry *e;
+	struct airoha_flow_data data = {};
+	struct net_device *odev = NULL;
+	struct flow_action_entry *act;
+	struct airoha_foe_entry hwe;
+	int err, i, offload_type;
+	u16 addr_type = 0;
+	u8 l4proto = 0;
+
+	if (rhashtable_lookup(&eth->flow_table, &f->cookie,
+			      airoha_flow_table_params))
+		return -EEXIST;
+
+	if (!flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_META))
+		return -EOPNOTSUPP;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+		struct flow_match_control match;
+
+		flow_rule_match_control(rule, &match);
+		addr_type = match.key->addr_type;
+		if (flow_rule_has_control_flags(match.mask->flags,
+						f->common.extack))
+			return -EOPNOTSUPP;
+	} else {
+		return -EOPNOTSUPP;
+	}
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
+		struct flow_match_basic match;
+
+		flow_rule_match_basic(rule, &match);
+		l4proto = match.key->ip_proto;
+	} else {
+		return -EOPNOTSUPP;
+	}
+
+	switch (addr_type) {
+	case 0:
+		offload_type = PPE_PKT_TYPE_BRIDGE;
+		if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
+			struct flow_match_eth_addrs match;
+
+			flow_rule_match_eth_addrs(rule, &match);
+			memcpy(data.eth.h_dest, match.key->dst, ETH_ALEN);
+			memcpy(data.eth.h_source, match.key->src, ETH_ALEN);
+		} else {
+			return -EOPNOTSUPP;
+		}
+
+		if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN)) {
+			struct flow_match_vlan match;
+
+			flow_rule_match_vlan(rule, &match);
+			if (match.key->vlan_tpid != cpu_to_be16(ETH_P_8021Q))
+				return -EOPNOTSUPP;
+
+			data.vlan_in = match.key->vlan_id;
+		}
+		break;
+	case FLOW_DISSECTOR_KEY_IPV4_ADDRS:
+		offload_type = PPE_PKT_TYPE_IPV4_HNAPT;
+		break;
+	case FLOW_DISSECTOR_KEY_IPV6_ADDRS:
+		offload_type = PPE_PKT_TYPE_IPV6_ROUTE_5T;
+		break;
+	default:
+		return -EOPNOTSUPP;
+	}
+
+	flow_action_for_each(i, act, &rule->action) {
+		switch (act->id) {
+		case FLOW_ACTION_MANGLE:
+			if (offload_type == PPE_PKT_TYPE_BRIDGE)
+				return -EOPNOTSUPP;
+
+			if (act->mangle.htype == FLOW_ACT_MANGLE_HDR_TYPE_ETH)
+				airoha_ppe_flow_mangle_eth(act, &data.eth);
+			break;
+		case FLOW_ACTION_REDIRECT:
+			odev = act->dev;
+			break;
+		case FLOW_ACTION_CSUM:
+			break;
+		case FLOW_ACTION_VLAN_PUSH:
+			if (data.vlan.num == 1 ||
+			    act->vlan.proto != htons(ETH_P_8021Q))
+				return -EOPNOTSUPP;
+
+			data.vlan.id = act->vlan.vid;
+			data.vlan.proto = act->vlan.proto;
+			data.vlan.num++;
+			break;
+		case FLOW_ACTION_VLAN_POP:
+			break;
+		case FLOW_ACTION_PPPOE_PUSH:
+			if (data.pppoe.num == 1)
+				return -EOPNOTSUPP;
+
+			data.pppoe.sid = act->pppoe.sid;
+			data.pppoe.num++;
+			break;
+		default:
+			return -EOPNOTSUPP;
+		}
+	}
+
+	if (!is_valid_ether_addr(data.eth.h_source) ||
+	    !is_valid_ether_addr(data.eth.h_dest))
+		return -EINVAL;
+
+	err = airoha_ppe_foe_entry_prepare(&hwe, odev, offload_type, l4proto,
+					   data.eth.h_source, data.eth.h_dest);
+	if (err)
+		return err;
+
+	if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS)) {
+		struct flow_match_ports ports;
+
+		if (offload_type == PPE_PKT_TYPE_BRIDGE)
+			return -EOPNOTSUPP;
+
+		flow_rule_match_ports(rule, &ports);
+		data.src_port = ports.key->src;
+		data.dst_port = ports.key->dst;
+	} else if (offload_type != PPE_PKT_TYPE_BRIDGE) {
+		return -EOPNOTSUPP;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		struct flow_match_ipv4_addrs addrs;
+
+		flow_rule_match_ipv4_addrs(rule, &addrs);
+		data.v4.src_addr = addrs.key->src;
+		data.v4.dst_addr = addrs.key->dst;
+		airoha_ppe_foe_entry_set_ipv4_tuple(&hwe, &data, false);
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+		struct flow_match_ipv6_addrs addrs;
+
+		flow_rule_match_ipv6_addrs(rule, &addrs);
+
+		data.v6.src_addr = addrs.key->src;
+		data.v6.dst_addr = addrs.key->dst;
+		airoha_ppe_foe_entry_set_ipv6_tuple(&hwe, &data);
+	}
+
+	flow_action_for_each(i, act, &rule->action) {
+		if (act->id != FLOW_ACTION_MANGLE)
+			continue;
+
+		if (offload_type == PPE_PKT_TYPE_BRIDGE)
+			return -EOPNOTSUPP;
+
+		switch (act->mangle.htype) {
+		case FLOW_ACT_MANGLE_HDR_TYPE_TCP:
+		case FLOW_ACT_MANGLE_HDR_TYPE_UDP:
+			err = airoha_ppe_flow_mangle_ports(act, &data);
+			break;
+		case FLOW_ACT_MANGLE_HDR_TYPE_IP4:
+			err = airoha_ppe_flow_mangle_ipv4(act, &data);
+			break;
+		case FLOW_ACT_MANGLE_HDR_TYPE_ETH:
+			/* handled earlier */
+			break;
+		default:
+			return -EOPNOTSUPP;
+		}
+
+		if (err)
+			return err;
+	}
+
+	if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+		err = airoha_ppe_foe_entry_set_ipv4_tuple(&hwe, &data, true);
+		if (err)
+			return err;
+	}
+
+	e = kzalloc(sizeof(*e), GFP_KERNEL);
+	if (!e)
+		return -ENOMEM;
+
+	e->cookie = f->cookie;
+	memcpy(&e->data, &hwe, sizeof(e->data));
+
+	err = airoha_ppe_foe_flow_commit_entry(eth->ppe, e);
+	if (err)
+		goto free_entry;
+
+	err = rhashtable_insert_fast(&eth->flow_table, &e->node,
+				     airoha_flow_table_params);
+	if (err < 0)
+		goto remove_foe_entry;
+
+	return 0;
+
+remove_foe_entry:
+	airoha_ppe_foe_flow_remove_entry(eth->ppe, e);
+free_entry:
+	kfree(e);
+
+	return err;
+}
+
+static int airoha_ppe_flow_offload_destroy(struct airoha_gdm_port *port,
+					   struct flow_cls_offload *f)
+{
+	struct airoha_eth *eth = port->qdma->eth;
+	struct airoha_flow_table_entry *e;
+
+	e = rhashtable_lookup(&eth->flow_table, &f->cookie,
+			      airoha_flow_table_params);
+	if (!e)
+		return -ENOENT;
+
+	airoha_ppe_foe_flow_remove_entry(eth->ppe, e);
+	rhashtable_remove_fast(&eth->flow_table, &e->node,
+			       airoha_flow_table_params);
+	kfree(e);
+
+	return 0;
+}
+
+static int airoha_ppe_flow_offload_cmd(struct airoha_gdm_port *port,
+				       struct flow_cls_offload *f)
+{
+	int err = -EOPNOTSUPP;
+
+	mutex_lock(&flow_offload_mutex);
+
+	switch (f->command) {
+	case FLOW_CLS_REPLACE:
+		err = airoha_ppe_flow_offload_replace(port, f);
+		break;
+	case FLOW_CLS_DESTROY:
+		err = airoha_ppe_flow_offload_destroy(port, f);
+		break;
+	default:
+		break;
+	}
+
+	mutex_unlock(&flow_offload_mutex);
+
+	return err;
+}
+
+int airoha_ppe_setup_tc_block_cb(enum tc_setup_type type, void *type_data,
+				 void *cb_priv)
+{
+	struct flow_cls_offload *cls = type_data;
+	struct net_device *dev = cb_priv;
+	struct airoha_gdm_port *port = netdev_priv(dev);
+
+	if (!tc_can_offload(dev) || type != TC_SETUP_CLSFLOWER)
+		return -EOPNOTSUPP;
+
+	if (!port->qdma->eth->npu)
+		return -EOPNOTSUPP;
+
+	return airoha_ppe_flow_offload_cmd(port, cls);
+}
+
+void airoha_ppe_check_skb(struct airoha_ppe *ppe, u16 hash)
+{
+	u16 now, diff;
+
+	if (hash > PPE_HASH_MASK)
+		return;
+
+	now = (u16)jiffies;
+	diff = now - ppe->foe_check_time[hash];
+	if (diff < HZ / 10)
+		return;
+
+	ppe->foe_check_time[hash] = now;
+	airoha_ppe_foe_insert_entry(ppe, hash);
+}
+
 static void airoha_ppe_hw_init(struct airoha_ppe *ppe)
 {
 	u32 sram_tb_size, sram_num_entries, dram_num_entries;
@@ -118,33 +782,47 @@ int airoha_ppe_init(struct airoha_eth *eth)
 	ppe->eth = eth;
 	eth->ppe = ppe;
 
+	ppe->foe_flow = devm_kzalloc(eth->dev,
+				     PPE_NUM_ENTRIES * sizeof(*ppe->foe_flow),
+				     GFP_KERNEL);
+	if (!ppe->foe_flow)
+		return -ENOMEM;
+
+	err = rhashtable_init(&eth->flow_table, &airoha_flow_table_params);
+	if (err)
+		return err;
+
 	npu = airoha_npu_init(eth);
-	if (IS_ERR(npu))
-		return PTR_ERR(npu);
+	if (IS_ERR(npu)) {
+		err = PTR_ERR(npu);
+		goto error_destroy_flow_table;
+	}
 
 	eth->npu = npu;
 	err = airoha_npu_ppe_init(npu);
 	if (err)
-		goto error;
+		goto error_npu_deinit;
 
 	airoha_ppe_hw_init(ppe);
 	err = airoha_npu_flush_ppe_sram_entries(npu, ppe);
 	if (err)
-		goto error;
+		goto error_npu_deinit;
 
 	return 0;
 
-error:
+error_npu_deinit:
 	airoha_npu_deinit(npu);
 	eth->npu = NULL;
+error_destroy_flow_table:
+	rhashtable_destroy(&eth->flow_table);
 
 	return err;
 }
-
 void airoha_ppe_deinit(struct airoha_eth *eth)
 {
 	if (eth->npu) {
 		airoha_npu_ppe_deinit(eth->npu);
 		airoha_npu_deinit(eth->npu);
 	}
+	rhashtable_destroy(&eth->flow_table);
 }

-- 
2.48.1

[PATCH net-next v3 15/16] net: airoha: Add loopback support for GDM2

[Lorenzo Bianconi]

Enable hw redirection for traffic received on GDM2 port to GDM{3,4}.
This is required to apply Qdisc offloading (HTB or ETS) for traffic to
and from GDM{3,4} port.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/airoha_eth.c  | 69 ++++++++++++++++++++++++++++++-
 drivers/net/ethernet/airoha/airoha_eth.h  |  7 ++++
 drivers/net/ethernet/airoha/airoha_ppe.c  | 13 +++---
 drivers/net/ethernet/airoha/airoha_regs.h | 29 +++++++++++++
 4 files changed, 110 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/airoha/airoha_eth.c b/drivers/net/ethernet/airoha/airoha_eth.c
index 46ef80b190e696f8341f262234e2d63e1372e715..5845e7cda2840c5a4d677e10a01fd836128587af 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.c
+++ b/drivers/net/ethernet/airoha/airoha_eth.c
@@ -1583,14 +1583,79 @@ static int airoha_dev_set_macaddr(struct net_device *dev, void *p)
 	return 0;
 }
 
+static void airhoha_set_gdm2_loopback(struct airoha_gdm_port *port)
+{
+	u32 pse_port = port->id == 3 ? FE_PSE_PORT_GDM3 : FE_PSE_PORT_GDM4;
+	struct airoha_eth *eth = port->qdma->eth;
+	u32 chan = port->id == 3 ? 4 : 0;
+
+	/* Forward the traffic to the proper GDM port */
+	airoha_set_gdm_port_fwd_cfg(eth, REG_GDM_FWD_CFG(2), pse_port);
+	airoha_fe_clear(eth, REG_GDM_FWD_CFG(2), GDM_STRIP_CRC);
+
+	/* Enable GDM2 loopback */
+	airoha_fe_wr(eth, REG_GDM_TXCHN_EN(2), 0xffffffff);
+	airoha_fe_wr(eth, REG_GDM_RXCHN_EN(2), 0xffff);
+	airoha_fe_rmw(eth, REG_GDM_LPBK_CFG(2),
+		      LPBK_CHAN_MASK | LPBK_MODE_MASK | LPBK_EN_MASK,
+		      FIELD_PREP(LPBK_CHAN_MASK, chan) | LPBK_EN_MASK);
+	airoha_fe_rmw(eth, REG_GDM_LEN_CFG(2),
+		      GDM_SHORT_LEN_MASK | GDM_LONG_LEN_MASK,
+		      FIELD_PREP(GDM_SHORT_LEN_MASK, 60) |
+		      FIELD_PREP(GDM_LONG_LEN_MASK, AIROHA_MAX_MTU));
+
+	/* Disable VIP and IFC for GDM2 */
+	airoha_fe_clear(eth, REG_FE_VIP_PORT_EN, BIT(2));
+	airoha_fe_clear(eth, REG_FE_IFC_PORT_EN, BIT(2));
+
+	if (port->id == 3) {
+		/* FIXME: handle XSI_PCE1_PORT */
+		airoha_fe_wr(eth, REG_PPE_DFT_CPORT0(0),  0x5500);
+		airoha_fe_rmw(eth, REG_FE_WAN_PORT,
+			      WAN1_EN_MASK | WAN1_MASK | WAN0_MASK,
+			      FIELD_PREP(WAN0_MASK, HSGMII_LAN_PCIE0_SRCPORT));
+		airoha_fe_rmw(eth,
+			      REG_SP_DFT_CPORT(HSGMII_LAN_PCIE0_SRCPORT >> 3),
+			      SP_CPORT_PCIE0_MASK,
+			      FIELD_PREP(SP_CPORT_PCIE0_MASK,
+					 FE_PSE_PORT_CDM2));
+	} else {
+		/* FIXME: handle XSI_USB_PORT */
+		airoha_fe_rmw(eth, REG_SRC_PORT_FC_MAP6,
+			      FC_ID_OF_SRC_PORT24_MASK,
+			      FIELD_PREP(FC_ID_OF_SRC_PORT24_MASK, 2));
+		airoha_fe_rmw(eth, REG_FE_WAN_PORT,
+			      WAN1_EN_MASK | WAN1_MASK | WAN0_MASK,
+			      FIELD_PREP(WAN0_MASK, HSGMII_LAN_ETH_SRCPORT));
+		airoha_fe_rmw(eth,
+			      REG_SP_DFT_CPORT(HSGMII_LAN_ETH_SRCPORT >> 3),
+			      SP_CPORT_ETH_MASK,
+			      FIELD_PREP(SP_CPORT_ETH_MASK, FE_PSE_PORT_CDM2));
+	}
+}
+
 static int airoha_dev_init(struct net_device *dev)
 {
 	struct airoha_gdm_port *port = netdev_priv(dev);
 	struct airoha_eth *eth = port->qdma->eth;
+	u32 pse_port;
 
 	airoha_set_macaddr(port, dev->dev_addr);
-	airoha_set_gdm_port_fwd_cfg(eth, REG_GDM_FWD_CFG(port->id),
-				    FE_PSE_PORT_PPE1);
+
+	switch (port->id) {
+	case 3:
+	case 4:
+		airhoha_set_gdm2_loopback(port);
+		fallthrough;
+	case 2:
+		pse_port = FE_PSE_PORT_PPE2;
+		break;
+	default:
+		pse_port = FE_PSE_PORT_PPE1;
+		break;
+	}
+
+	airoha_set_gdm_port_fwd_cfg(eth, REG_GDM_FWD_CFG(port->id), pse_port);
 
 	return 0;
 }
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index c51fa6713be07415a56ba4da2456d738d9a8f2af..464d92ad5371a137681b4128cc49a7c1a91ddd19 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -73,6 +73,13 @@ enum {
 	QDMA_INT_REG_MAX
 };
 
+enum {
+	HSGMII_LAN_PCIE0_SRCPORT = 0x16,
+	HSGMII_LAN_PCIE1_SRCPORT,
+	HSGMII_LAN_ETH_SRCPORT,
+	HSGMII_LAN_USB_SRCPORT,
+};
+
 enum {
 	XSI_PCIE0_VIP_PORT_MASK	= BIT(22),
 	XSI_PCIE1_VIP_PORT_MASK	= BIT(23),
diff --git a/drivers/net/ethernet/airoha/airoha_ppe.c b/drivers/net/ethernet/airoha/airoha_ppe.c
index 54f63b017bcfb0af840d0996e4b099371fcaa0e6..52c8a40c353bd8b812ac44cc936eb6721447a67f 100644
--- a/drivers/net/ethernet/airoha/airoha_ppe.c
+++ b/drivers/net/ethernet/airoha/airoha_ppe.c
@@ -124,21 +124,22 @@ static int airoha_ppe_foe_entry_prepare(struct airoha_foe_entry *hwe,
 	      AIROHA_FOE_IB1_BIND_TTL;
 	hwe->ib1 = val;
 
-	val = FIELD_PREP(AIROHA_FOE_IB2_PORT_AG, 0x1f);
+	val = FIELD_PREP(AIROHA_FOE_IB2_PORT_AG, 0x1f) |
+	      AIROHA_FOE_IB2_PSE_QOS;
 	if (dsa_port >= 0)
 		val |= FIELD_PREP(AIROHA_FOE_IB2_NBQ, dsa_port);
+
 	if (dev) {
 		struct airoha_gdm_port *port = netdev_priv(dev);
 		u8 pse_port;
 
-		pse_port = port->id == 4 ? FE_PSE_PORT_GDM4 : port->id;
+		if (dsa_port >= 0)
+			pse_port = port->id == 4 ? FE_PSE_PORT_GDM4 : port->id;
+		else
+			pse_port = 2; /* uplink relies on GDM2 loopback */
 		val |= FIELD_PREP(AIROHA_FOE_IB2_PSE_PORT, pse_port);
 	}
 
-	/* FIXME: implement QoS support setting pse_port to 2 (loopback)
-	 * for uplink and setting qos bit in ib2
-	 */
-
 	if (is_multicast_ether_addr(dest_mac))
 		val |= AIROHA_FOE_IB2_MULTICAST;
 
diff --git a/drivers/net/ethernet/airoha/airoha_regs.h b/drivers/net/ethernet/airoha/airoha_regs.h
index 6cc64c60953a3961b7c93dfa75a289a6f7a6599b..1aa06cdffe2320375e8710d58f2bbb056a330dfd 100644
--- a/drivers/net/ethernet/airoha/airoha_regs.h
+++ b/drivers/net/ethernet/airoha/airoha_regs.h
@@ -38,6 +38,12 @@
 #define FE_RST_CORE_MASK		BIT(0)
 
 #define REG_FE_FOE_TS			0x0010
+
+#define REG_FE_WAN_PORT			0x0024
+#define WAN1_EN_MASK			BIT(16)
+#define WAN1_MASK			GENMASK(12, 8)
+#define WAN0_MASK			GENMASK(4, 0)
+
 #define REG_FE_WAN_MAC_H		0x0030
 #define REG_FE_LAN_MAC_H		0x0040
 
@@ -126,6 +132,7 @@
 #define GDM_IP4_CKSUM			BIT(22)
 #define GDM_TCP_CKSUM			BIT(21)
 #define GDM_UDP_CKSUM			BIT(20)
+#define GDM_STRIP_CRC			BIT(16)
 #define GDM_UCFQ_MASK			GENMASK(15, 12)
 #define GDM_BCFQ_MASK			GENMASK(11, 8)
 #define GDM_MCFQ_MASK			GENMASK(7, 4)
@@ -139,6 +146,16 @@
 #define GDM_SHORT_LEN_MASK		GENMASK(13, 0)
 #define GDM_LONG_LEN_MASK		GENMASK(29, 16)
 
+#define REG_GDM_LPBK_CFG(_n)		(GDM_BASE(_n) + 0x1c)
+#define LPBK_GAP_MASK			GENMASK(31, 24)
+#define LPBK_LEN_MASK			GENMASK(23, 10)
+#define LPBK_CHAN_MASK			GENMASK(8, 4)
+#define LPBK_MODE_MASK			GENMASK(3, 1)
+#define LPBK_EN_MASK			BIT(0)
+
+#define REG_GDM_TXCHN_EN(_n)		(GDM_BASE(_n) + 0x24)
+#define REG_GDM_RXCHN_EN(_n)		(GDM_BASE(_n) + 0x28)
+
 #define REG_FE_CPORT_CFG		(GDM1_BASE + 0x40)
 #define FE_CPORT_PAD			BIT(26)
 #define FE_CPORT_PORT_XFC_MASK		BIT(25)
@@ -351,6 +368,18 @@
 
 #define REG_MC_VLAN_DATA		0x2108
 
+#define REG_SP_DFT_CPORT(_n)		(0x20e0 + ((_n) << 2))
+#define SP_CPORT_PCIE1_MASK		GENMASK(31, 28)
+#define SP_CPORT_PCIE0_MASK		GENMASK(27, 24)
+#define SP_CPORT_USB_MASK		GENMASK(7, 4)
+#define SP_CPORT_ETH_MASK		GENMASK(7, 4)
+
+#define REG_SRC_PORT_FC_MAP6		0x2298
+#define FC_ID_OF_SRC_PORT27_MASK	GENMASK(28, 24)
+#define FC_ID_OF_SRC_PORT26_MASK	GENMASK(20, 16)
+#define FC_ID_OF_SRC_PORT25_MASK	GENMASK(12, 8)
+#define FC_ID_OF_SRC_PORT24_MASK	GENMASK(4, 0)
+
 #define REG_CDM5_RX_OQ1_DROP_CNT	0x29d4
 
 /* QDMA */

-- 
2.48.1

[PATCH net-next v3 16/16] net: airoha: Introduce PPE debugfs support

[Lorenzo Bianconi]

Similar to PPE support for Mediatek devices, introduce PPE debugfs
in order to dump binded and unbinded flows.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 drivers/net/ethernet/airoha/Makefile             |   1 +
 drivers/net/ethernet/airoha/airoha_eth.h         |  12 ++
 drivers/net/ethernet/airoha/airoha_ppe.c         |   5 +
 drivers/net/ethernet/airoha/airoha_ppe_debugfs.c | 175 +++++++++++++++++++++++
 4 files changed, 193 insertions(+)

diff --git a/drivers/net/ethernet/airoha/Makefile b/drivers/net/ethernet/airoha/Makefile
index 50028cfc3e3e04efbdd353b1bd65f46b488637d8..e5321effdbeaeaf7c4ad3c6df9d0261fe12fdd12 100644
--- a/drivers/net/ethernet/airoha/Makefile
+++ b/drivers/net/ethernet/airoha/Makefile
@@ -5,4 +5,5 @@
 
 obj-$(CONFIG_NET_AIROHA) += airoha-eth.o
 airoha-eth-y := airoha_eth.o airoha_ppe.o
+airoha-eth-$(CONFIG_DEBUG_FS) += airoha_ppe_debugfs.o
 airoha-eth-$(CONFIG_NET_AIROHA_NPU) += airoha_npu.o
diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
index 464d92ad5371a137681b4128cc49a7c1a91ddd19..87d2a2a928eaa6680ce3a256eea973c6ce324f13 100644
--- a/drivers/net/ethernet/airoha/airoha_eth.h
+++ b/drivers/net/ethernet/airoha/airoha_eth.h
@@ -7,6 +7,7 @@
 #ifndef AIROHA_ETH_H
 #define AIROHA_ETH_H
 
+#include <linux/debugfs.h>
 #include <linux/etherdevice.h>
 #include <linux/iopoll.h>
 #include <linux/kernel.h>
@@ -502,6 +503,8 @@ struct airoha_ppe {
 
 	struct hlist_head *foe_flow;
 	u16 foe_check_time[PPE_NUM_ENTRIES];
+
+	struct dentry *debugfs_dir;
 };
 
 struct airoha_eth {
@@ -611,4 +614,13 @@ static inline int airoha_npu_foe_commit_entry(struct airoha_ppe *ppe,
 }
 #endif /* CONFIG_NET_AIROHA_NPU */
 
+#if CONFIG_DEBUG_FS
+int airoha_ppe_debugfs_init(struct airoha_ppe *ppe);
+#else
+static inline int airoha_ppe_debugfs_init(struct airoha_ppe *ppe)
+{
+	return 0;
+}
+#endif
+
 #endif /* AIROHA_ETH_H */
diff --git a/drivers/net/ethernet/airoha/airoha_ppe.c b/drivers/net/ethernet/airoha/airoha_ppe.c
index 52c8a40c353bd8b812ac44cc936eb6721447a67f..135906a52f72be54e4ecb32febad2c4670d118b8 100644
--- a/drivers/net/ethernet/airoha/airoha_ppe.c
+++ b/drivers/net/ethernet/airoha/airoha_ppe.c
@@ -809,6 +809,10 @@ int airoha_ppe_init(struct airoha_eth *eth)
 	if (err)
 		goto error_npu_deinit;
 
+	err = airoha_ppe_debugfs_init(ppe);
+	if (err)
+		goto error_npu_deinit;
+
 	return 0;
 
 error_npu_deinit:
@@ -826,4 +830,5 @@ void airoha_ppe_deinit(struct airoha_eth *eth)
 		airoha_npu_deinit(eth->npu);
 	}
 	rhashtable_destroy(&eth->flow_table);
+	debugfs_remove(eth->ppe->debugfs_dir);
 }
diff --git a/drivers/net/ethernet/airoha/airoha_ppe_debugfs.c b/drivers/net/ethernet/airoha/airoha_ppe_debugfs.c
new file mode 100644
index 0000000000000000000000000000000000000000..f79788458903c93ca73d774edaca97f1e08f92a7
--- /dev/null
+++ b/drivers/net/ethernet/airoha/airoha_ppe_debugfs.c
@@ -0,0 +1,175 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2025 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#include "airoha_eth.h"
+
+static void airoha_debugfs_ppe_print_tuple(struct seq_file *m,
+					   void *src_addr, void *dest_addr,
+					   u16 *src_port, u16 *dest_port,
+					   bool ipv6)
+{
+	__be32 n_addr[IPV6_ADDR_WORDS];
+
+	if (ipv6) {
+		ipv6_addr_cpu_to_be32(n_addr, src_addr);
+		seq_printf(m, "%pI6", n_addr);
+	} else {
+		seq_printf(m, "%pI4h", src_addr);
+	}
+	if (src_port)
+		seq_printf(m, ":%d", *src_port);
+
+	seq_puts(m, "->");
+
+	if (ipv6) {
+		ipv6_addr_cpu_to_be32(n_addr, dest_addr);
+		seq_printf(m, "%pI6", n_addr);
+	} else {
+		seq_printf(m, "%pI4h", dest_addr);
+	}
+	if (dest_port)
+		seq_printf(m, ":%d", *dest_port);
+}
+
+static int airoha_ppe_debugfs_foe_show(struct seq_file *m, void *private,
+				       bool bind)
+{
+	static const char *const ppe_type_str[] = {
+		[PPE_PKT_TYPE_IPV4_HNAPT] = "IPv4 5T",
+		[PPE_PKT_TYPE_IPV4_ROUTE] = "IPv4 3T",
+		[PPE_PKT_TYPE_BRIDGE] = "L2B",
+		[PPE_PKT_TYPE_IPV4_DSLITE] = "DS-LITE",
+		[PPE_PKT_TYPE_IPV6_ROUTE_3T] = "IPv6 3T",
+		[PPE_PKT_TYPE_IPV6_ROUTE_5T] = "IPv6 5T",
+		[PPE_PKT_TYPE_IPV6_6RD] = "6RD",
+	};
+	static const char *const ppe_state_str[] = {
+		[AIROHA_FOE_STATE_INVALID] = "INV",
+		[AIROHA_FOE_STATE_UNBIND] = "UNB",
+		[AIROHA_FOE_STATE_BIND] = "BND",
+		[AIROHA_FOE_STATE_FIN] = "FIN",
+	};
+	struct airoha_ppe *ppe = m->private;
+	int i;
+
+	for (i = 0; i < PPE_NUM_ENTRIES; i++) {
+		const char *state_str, *type_str = "UNKNOWN";
+		u16 *src_port = NULL, *dest_port = NULL;
+		struct airoha_foe_mac_info_common *l2;
+		unsigned char h_source[ETH_ALEN] = {};
+		unsigned char h_dest[ETH_ALEN];
+		struct airoha_foe_entry *hwe;
+		u32 type, state, ib2, data;
+		void *src_addr, *dest_addr;
+		bool ipv6 = false;
+
+		hwe = airoha_ppe_foe_get_entry(ppe, i);
+		if (!hwe)
+			continue;
+
+		state = FIELD_GET(AIROHA_FOE_IB1_STATE, hwe->ib1);
+		if (!state)
+			continue;
+
+		if (bind && state != AIROHA_FOE_STATE_BIND)
+			continue;
+
+		state_str = ppe_state_str[state % ARRAY_SIZE(ppe_state_str)];
+		type = FIELD_GET(AIROHA_FOE_IB1_PACKET_TYPE, hwe->ib1);
+		if (type < ARRAY_SIZE(ppe_type_str) && ppe_type_str[type])
+			type_str = ppe_type_str[type];
+
+		seq_printf(m, "%05x %s %7s", i, state_str, type_str);
+
+		switch (type) {
+		case PPE_PKT_TYPE_IPV4_HNAPT:
+		case PPE_PKT_TYPE_IPV4_DSLITE:
+			src_port = &hwe->ipv4.orig_tuple.src_port;
+			dest_port = &hwe->ipv4.orig_tuple.dest_port;
+			fallthrough;
+		case PPE_PKT_TYPE_IPV4_ROUTE:
+			src_addr = &hwe->ipv4.orig_tuple.src_ip;
+			dest_addr = &hwe->ipv4.orig_tuple.dest_ip;
+			break;
+		case PPE_PKT_TYPE_IPV6_ROUTE_5T:
+			src_port = &hwe->ipv6.src_port;
+			dest_port = &hwe->ipv6.dest_port;
+			fallthrough;
+		case PPE_PKT_TYPE_IPV6_ROUTE_3T:
+		case PPE_PKT_TYPE_IPV6_6RD:
+			src_addr = &hwe->ipv6.src_ip;
+			dest_addr = &hwe->ipv6.dest_ip;
+			ipv6 = true;
+			break;
+		}
+
+		seq_puts(m, " orig=");
+		airoha_debugfs_ppe_print_tuple(m, src_addr, dest_addr,
+					       src_port, dest_port, ipv6);
+
+		switch (type) {
+		case PPE_PKT_TYPE_IPV4_HNAPT:
+		case PPE_PKT_TYPE_IPV4_DSLITE:
+			src_port = &hwe->ipv4.new_tuple.src_port;
+			dest_port = &hwe->ipv4.new_tuple.dest_port;
+			fallthrough;
+		case PPE_PKT_TYPE_IPV4_ROUTE:
+			src_addr = &hwe->ipv4.new_tuple.src_ip;
+			dest_addr = &hwe->ipv4.new_tuple.dest_ip;
+			seq_puts(m, " new=");
+			airoha_debugfs_ppe_print_tuple(m, src_addr, dest_addr,
+						       src_port, dest_port,
+						       ipv6);
+			break;
+		}
+
+		if (type >= PPE_PKT_TYPE_IPV6_ROUTE_3T) {
+			data = hwe->ipv6.data;
+			ib2 = hwe->ipv6.ib2;
+			l2 = &hwe->ipv6.l2;
+		} else {
+			data = hwe->ipv4.data;
+			ib2 = hwe->ipv4.ib2;
+			l2 = &hwe->ipv4.l2.common;
+			*((__be16 *)&h_source[4]) =
+				cpu_to_be16(hwe->ipv4.l2.src_mac_lo);
+		}
+
+		*((__be32 *)h_dest) = cpu_to_be32(l2->dest_mac_hi);
+		*((__be16 *)&h_dest[4]) = cpu_to_be16(l2->dest_mac_lo);
+		*((__be32 *)h_source) = cpu_to_be32(l2->src_mac_hi);
+
+		seq_printf(m, " eth=%pM->%pM etype=%04x data=%08x"
+			      " vlan=%d,%d ib1=%08x ib2=%08x\n",
+			   h_source, h_dest, l2->etype, data,
+			   l2->vlan1, l2->vlan2, hwe->ib1, ib2);
+	}
+
+	return 0;
+}
+
+static int airoha_ppe_debugfs_foe_all_show(struct seq_file *m, void *private)
+{
+	return airoha_ppe_debugfs_foe_show(m, private, false);
+}
+DEFINE_SHOW_ATTRIBUTE(airoha_ppe_debugfs_foe_all);
+
+static int airoha_ppe_debugfs_foe_bind_show(struct seq_file *m, void *private)
+{
+	return airoha_ppe_debugfs_foe_show(m, private, true);
+}
+DEFINE_SHOW_ATTRIBUTE(airoha_ppe_debugfs_foe_bind);
+
+int airoha_ppe_debugfs_init(struct airoha_ppe *ppe)
+{
+	ppe->debugfs_dir = debugfs_create_dir("ppe", NULL);
+	debugfs_create_file("entries", 0444, ppe->debugfs_dir, ppe,
+			    &airoha_ppe_debugfs_foe_all_fops);
+	debugfs_create_file("bind", 0444, ppe->debugfs_dir, ppe,
+			    &airoha_ppe_debugfs_foe_bind_fops);
+
+	return 0;
+}

-- 
2.48.1