diff mbox series

[v2,net-next,2/2] net: airoha: Introduce ethernet support for EN7581 SoC

Message ID f146a6f58492394a77f7d159f3c650a268fbe489.1718696209.git.lorenzo@kernel.org (mailing list archive)
State Not Applicable, archived
Headers show
Series Introduce EN7581 ethernet support | expand

Commit Message

Lorenzo Bianconi June 18, 2024, 7:49 a.m. UTC
Add airoha_eth driver in order to introduce ethernet support for
Airoha EN7581 SoC available on EN7581 development board (en7581-evb).
en7581-evb networking architecture is composed by airoha_eth as mac
controller (cpu port) and a mt7530 dsa based switch.
EN7581 mac controller is mainly composed by Frame Engine (FE) and
QoS-DMA (QDMA) modules. FE is used for traffic offloading (just basic
functionalities are supported now) while QDMA is used for DMA operation
and QOS functionalities between mac layer and the dsa switch (hw QoS is
not available yet and it will be added in the future).
Currently only hw lan features are available, hw wan will be added with
subsequent patches.

Tested-by: Benjamin Larsson <benjamin.larsson@genexis.eu>
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
---
 MAINTAINERS                                |   10 +
 drivers/net/ethernet/mediatek/Kconfig      |   11 +-
 drivers/net/ethernet/mediatek/Makefile     |    1 +
 drivers/net/ethernet/mediatek/airoha_eth.c | 1804 ++++++++++++++++++++
 drivers/net/ethernet/mediatek/airoha_eth.h |  793 +++++++++
 5 files changed, 2618 insertions(+), 1 deletion(-)
 create mode 100644 drivers/net/ethernet/mediatek/airoha_eth.c
 create mode 100644 drivers/net/ethernet/mediatek/airoha_eth.h

Comments

Andrew Lunn June 18, 2024, 1:46 p.m. UTC | #1
> +static void airoha_fe_maccr_init(struct airoha_eth *eth)
> +{

...

> +	airoha_fe_wr(eth, REG_FE_VIP_PATN(11), 0xc057); /* PPP->IPv6CP (0xc057) */

include/uapi/linux/ppp_defs.h
#define PPP_IPV6CP      0x8057  /* IPv6 Control Protocol */

Are these the same thing? Why is there one bit difference?


> +	airoha_fe_wr(eth, REG_FE_VIP_PATN(17), 0x1ae0);
> +	airoha_fe_wr(eth, REG_FE_VIP_EN(17),
> +		     PATN_FCPU_EN_MASK | PATN_SP_EN_MASK |
> +		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);
> +
> +	airoha_fe_wr(eth, REG_FE_VIP_PATN(18), 0x1ae00000);
> +	airoha_fe_wr(eth, REG_FE_VIP_EN(18),
> +		     PATN_FCPU_EN_MASK | PATN_DP_EN_MASK |
> +		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);

> +	airoha_fe_wr(eth, REG_FE_VIP_PATN(22), 0xaaaa);

Please add a comment what these match.

> +static int airoha_dev_change_mtu(struct net_device *dev, int new_mtu)
> +{
> +	dev->mtu = new_mtu;
> +
> +	return 0;
> +}
> +

I don't think this is needed. Look at __dev_set_mtu().

    Andrew

---
pw-bot: cr
Lorenzo Bianconi June 19, 2024, 7:19 a.m. UTC | #2
> > +static void airoha_fe_maccr_init(struct airoha_eth *eth)
> > +{
> 
> ...
> 
> > +	airoha_fe_wr(eth, REG_FE_VIP_PATN(11), 0xc057); /* PPP->IPv6CP (0xc057) */
> 
> include/uapi/linux/ppp_defs.h
> #define PPP_IPV6CP      0x8057  /* IPv6 Control Protocol */
> 
> Are these the same thing? Why is there one bit difference?

ack, I guess there is a typo here.

> 
> 
> > +	airoha_fe_wr(eth, REG_FE_VIP_PATN(17), 0x1ae0);
> > +	airoha_fe_wr(eth, REG_FE_VIP_EN(17),
> > +		     PATN_FCPU_EN_MASK | PATN_SP_EN_MASK |
> > +		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);
> > +
> > +	airoha_fe_wr(eth, REG_FE_VIP_PATN(18), 0x1ae00000);
> > +	airoha_fe_wr(eth, REG_FE_VIP_EN(18),
> > +		     PATN_FCPU_EN_MASK | PATN_DP_EN_MASK |
> > +		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);
> 
> > +	airoha_fe_wr(eth, REG_FE_VIP_PATN(22), 0xaaaa);
> 
> Please add a comment what these match.

ack, I will look into them.

> 
> > +static int airoha_dev_change_mtu(struct net_device *dev, int new_mtu)
> > +{
> > +	dev->mtu = new_mtu;
> > +
> > +	return 0;
> > +}
> > +
> 
> I don't think this is needed. Look at __dev_set_mtu().

ack, I will get rid of it in v3.

Regards,
Lorenzo

> 
>     Andrew
> 
> ---
> pw-bot: cr
kernel test robot June 24, 2024, 3:58 p.m. UTC | #3
Hi Lorenzo,

kernel test robot noticed the following build warnings:

[auto build test WARNING on net-next/main]

url:    https://github.com/intel-lab-lkp/linux/commits/Lorenzo-Bianconi/dt-bindings-net-airoha-Add-EN7581-ethernet-controller/20240618-182217
base:   net-next/main
patch link:    https://lore.kernel.org/r/f146a6f58492394a77f7d159f3c650a268fbe489.1718696209.git.lorenzo%40kernel.org
patch subject: [PATCH v2 net-next 2/2] net: airoha: Introduce ethernet support for EN7581 SoC
config: arm64-randconfig-r121-20240623 (https://download.01.org/0day-ci/archive/20240624/202406242337.21CXNZvT-lkp@intel.com/config)
compiler: clang version 19.0.0git (https://github.com/llvm/llvm-project ad79a14c9e5ec4a369eed4adf567c22cc029863f)
reproduce: (https://download.01.org/0day-ci/archive/20240624/202406242337.21CXNZvT-lkp@intel.com/reproduce)

If you fix the issue in a separate patch/commit (i.e. not just a new version of
the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <lkp@intel.com>
| Closes: https://lore.kernel.org/oe-kbuild-all/202406242337.21CXNZvT-lkp@intel.com/

sparse warnings: (new ones prefixed by >>)
>> drivers/net/ethernet/mediatek/airoha_eth.c:1328:45: sparse: sparse: incorrect type in assignment (different base types) @@     expected restricted __sum16 [usertype] check @@     got restricted __be16 [usertype] @@
   drivers/net/ethernet/mediatek/airoha_eth.c:1328:45: sparse:     expected restricted __sum16 [usertype] check
   drivers/net/ethernet/mediatek/airoha_eth.c:1328:45: sparse:     got restricted __be16 [usertype]
>> drivers/net/ethernet/mediatek/airoha_eth.c:1787:27: sparse: sparse: symbol 'of_airoha_match' was not declared. Should it be static?

vim +1328 drivers/net/ethernet/mediatek/airoha_eth.c

  1303	
  1304	static netdev_tx_t airoha_dev_xmit(struct sk_buff *skb,
  1305					   struct net_device *dev)
  1306	{
  1307		struct skb_shared_info *sinfo = skb_shinfo(skb);
  1308		struct airoha_eth *eth = netdev_priv(dev);
  1309		int i, qid = skb_get_queue_mapping(skb);
  1310		u32 nr_frags, msg0 = 0, msg1;
  1311		u32 len = skb_headlen(skb);
  1312		struct netdev_queue *txq;
  1313		struct airoha_queue *q;
  1314		void *data = skb->data;
  1315		u16 index;
  1316	
  1317		if (skb->ip_summed == CHECKSUM_PARTIAL)
  1318			msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TCO_MASK, 1) |
  1319				FIELD_PREP(QDMA_ETH_TXMSG_UCO_MASK, 1) |
  1320				FIELD_PREP(QDMA_ETH_TXMSG_ICO_MASK, 1);
  1321	
  1322		/* TSO: fill MSS info in tcp checksum field */
  1323		if (skb_is_gso(skb)) {
  1324			if (skb_cow_head(skb, 0))
  1325				goto error;
  1326	
  1327			if (sinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)) {
> 1328				tcp_hdr(skb)->check = cpu_to_be16(sinfo->gso_size);
  1329				msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TSO_MASK, 1);
  1330			}
  1331		}
  1332	
  1333		msg1 = FIELD_PREP(QDMA_ETH_TXMSG_FPORT_MASK, DPORT_GDM1) |
  1334		       FIELD_PREP(QDMA_ETH_TXMSG_METER_MASK, 0x7f);
  1335	
  1336		if (WARN_ON_ONCE(qid >= ARRAY_SIZE(eth->q_tx)))
  1337			qid = 0;
  1338	
  1339		q = &eth->q_tx[qid];
  1340		if (WARN_ON_ONCE(!q->ndesc))
  1341			goto error;
  1342	
  1343		spin_lock_bh(&q->lock);
  1344	
  1345		nr_frags = 1 + sinfo->nr_frags;
  1346		if (q->queued + nr_frags > q->ndesc) {
  1347			/* not enough space in the queue */
  1348			spin_unlock_bh(&q->lock);
  1349			return NETDEV_TX_BUSY;
  1350		}
  1351	
  1352		index = q->head;
  1353		for (i = 0; i < nr_frags; i++) {
  1354			struct airoha_qdma_desc *desc = &q->desc[index];
  1355			struct airoha_queue_entry *e = &q->entry[index];
  1356			skb_frag_t *frag = &sinfo->frags[i];
  1357			dma_addr_t addr;
  1358			u32 val;
  1359	
  1360			addr = dma_map_single(dev->dev.parent, data, len,
  1361					      DMA_TO_DEVICE);
  1362			if (unlikely(dma_mapping_error(dev->dev.parent, addr)))
  1363				goto error_unmap;
  1364	
  1365			index = (index + 1) % q->ndesc;
  1366	
  1367			val = FIELD_PREP(QDMA_DESC_LEN_MASK, len);
  1368			if (i < nr_frags - 1)
  1369				val |= FIELD_PREP(QDMA_DESC_MORE_MASK, 1);
  1370			WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
  1371			WRITE_ONCE(desc->addr, cpu_to_le32(addr));
  1372			val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, index);
  1373			WRITE_ONCE(desc->data, cpu_to_le32(val));
  1374			WRITE_ONCE(desc->msg0, cpu_to_le32(msg0));
  1375			WRITE_ONCE(desc->msg1, cpu_to_le32(msg1));
  1376			WRITE_ONCE(desc->msg2, cpu_to_le32(0xffff));
  1377	
  1378			e->skb = i ? NULL : skb;
  1379			e->dma_addr = addr;
  1380			e->dma_len = len;
  1381	
  1382			wmb();
  1383			airoha_qdma_rmw(eth, REG_TX_CPU_IDX(qid), TX_RING_CPU_IDX_MASK,
  1384					FIELD_PREP(TX_RING_CPU_IDX_MASK, index));
  1385	
  1386			data = skb_frag_address(frag);
  1387			len = skb_frag_size(frag);
  1388		}
  1389	
  1390		q->head = index;
  1391		q->queued += i;
  1392	
  1393		txq = netdev_get_tx_queue(dev, qid);
  1394		netdev_tx_sent_queue(txq, skb->len);
  1395		skb_tx_timestamp(skb);
  1396	
  1397		if (q->ndesc - q->queued < q->free_thr)
  1398			netif_tx_stop_queue(txq);
  1399	
  1400		spin_unlock_bh(&q->lock);
  1401	
  1402		return NETDEV_TX_OK;
  1403	
  1404	error_unmap:
  1405		for (i--; i >= 0; i++)
  1406			dma_unmap_single(dev->dev.parent, q->entry[i].dma_addr,
  1407					 q->entry[i].dma_len, DMA_TO_DEVICE);
  1408	
  1409		spin_unlock_bh(&q->lock);
  1410	error:
  1411		dev_kfree_skb_any(skb);
  1412		dev->stats.tx_dropped++;
  1413	
  1414		return NETDEV_TX_OK;
  1415	}
  1416
Arnd Bergmann June 24, 2024, 8:05 p.m. UTC | #4
On Tue, Jun 18, 2024, at 09:49, Lorenzo Bianconi wrote:
> Add airoha_eth driver in order to introduce ethernet support for
> Airoha EN7581 SoC available on EN7581 development board (en7581-evb).
> en7581-evb networking architecture is composed by airoha_eth as mac
> controller (cpu port) and a mt7530 dsa based switch.
> EN7581 mac controller is mainly composed by Frame Engine (FE) and
> QoS-DMA (QDMA) modules. FE is used for traffic offloading (just basic
> functionalities are supported now) while QDMA is used for DMA operation
> and QOS functionalities between mac layer and the dsa switch (hw QoS is
> not available yet and it will be added in the future).
> Currently only hw lan features are available, hw wan will be added with
> subsequent patches.
>
> Tested-by: Benjamin Larsson <benjamin.larsson@genexis.eu>
> Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>

I noticed a few small things that you may want to improve:

> +static void airoha_qdma_set_irqmask(struct airoha_eth *eth, int index,
> +				    u32 clear, u32 set)
> +{
> +	unsigned long flags;
> +
> +	if (WARN_ON_ONCE(index >= ARRAY_SIZE(eth->irqmask)))
> +		return;
> +
> +	spin_lock_irqsave(&eth->irq_lock, flags);
> +
> +	eth->irqmask[index] &= ~clear;
> +	eth->irqmask[index] |= set;
> +	airoha_qdma_wr(eth, REG_INT_ENABLE(index), eth->irqmask[index]);
> +
> +	spin_unlock_irqrestore(&eth->irq_lock, flags);
> +}

spin_lock_irqsave() is fairly expensive here, and it doesn't
actually protect the register write since that is posted
and can leak out of the spinlock.

You can probably just remove the lock and instead do the mask
with atomic_cmpxchg() here.

> +
> +		dma_sync_single_for_device(dev, e->dma_addr, e->dma_len, dir);
> +
> +		val = FIELD_PREP(QDMA_DESC_LEN_MASK, e->dma_len);
> +		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
> +		WRITE_ONCE(desc->addr, cpu_to_le32(e->dma_addr));
> +		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, q->head);
> +		WRITE_ONCE(desc->data, cpu_to_le32(val));
> +		WRITE_ONCE(desc->msg0, 0);
> +		WRITE_ONCE(desc->msg1, 0);
> +		WRITE_ONCE(desc->msg2, 0);
> +		WRITE_ONCE(desc->msg3, 0);
> +
> +		wmb();
> +		airoha_qdma_rmw(eth, REG_RX_CPU_IDX(qid), RX_RING_CPU_IDX_MASK,
> +				FIELD_PREP(RX_RING_CPU_IDX_MASK, q->head));

The wmb() between the descriptor write and the MMIO does nothing
and can probably just be removed here, a writel() already contains
all the barriers you need to make DMA memory visible before the
MMIO write.

If there is a chance that the device might read the descriptor
while it is being updated by you have not written to the
register, there should be a barrier before the last store to
the descriptor that sets a 'valid' bit. That one can be a
cheaper dma_wmb() then.

> +static irqreturn_t airoha_irq_handler(int irq, void *dev_instance)
> +{
> +	struct airoha_eth *eth = dev_instance;
> +	u32 intr[ARRAY_SIZE(eth->irqmask)];
> +	int i;
> +
> +	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++) {
> +		intr[i] = airoha_qdma_rr(eth, REG_INT_STATUS(i));
> +		intr[i] &= eth->irqmask[i];
> +		airoha_qdma_wr(eth, REG_INT_STATUS(i), intr[i]);
> +	}

This looks like you send an interrupt Ack to each
interrupt in order to re-arm it, but then you disable
it again. Would it be possible to leave the interrupt enabled
but defer the Ack until the napi poll function is completed?

> +	if (!test_bit(DEV_STATE_INITIALIZED, &eth->state))
> +		return IRQ_NONE;
> +
> +	if (intr[1] & RX_DONE_INT_MASK) {
> +		airoha_qdma_irq_disable(eth, QDMA_INT_REG_IDX1,
> +					RX_DONE_INT_MASK);
> +		airoha_qdma_for_each_q_rx(eth, i) {
> +			if (intr[1] & BIT(i))
> +				napi_schedule(&eth->q_rx[i].napi);
> +		}
> +	}

Something seems wrong here, but that's probably just me
misunderstanding the design: if all queues are signaled
through the same interrupt handler, and you then do
napi_schedule() for each queue, I would expect them to
just all get run on the same CPU.

If you have separate queues, doesn't that mean you also need
separate irq numbers here so they can be distributed to the
available CPUs?

> +		val = FIELD_PREP(QDMA_DESC_LEN_MASK, len);
> +		if (i < nr_frags - 1)
> +			val |= FIELD_PREP(QDMA_DESC_MORE_MASK, 1);
> +		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
> +		WRITE_ONCE(desc->addr, cpu_to_le32(addr));
> +		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, index);
> +		WRITE_ONCE(desc->data, cpu_to_le32(val));
> +		WRITE_ONCE(desc->msg0, cpu_to_le32(msg0));
> +		WRITE_ONCE(desc->msg1, cpu_to_le32(msg1));
> +		WRITE_ONCE(desc->msg2, cpu_to_le32(0xffff));
> +
> +		e->skb = i ? NULL : skb;
> +		e->dma_addr = addr;
> +		e->dma_len = len;
> +
> +		wmb();
> +		airoha_qdma_rmw(eth, REG_TX_CPU_IDX(qid), TX_RING_CPU_IDX_MASK,
> +				FIELD_PREP(TX_RING_CPU_IDX_MASK, index));

Same as above with the wmb().

> +static int airoha_rx_queues_show(struct seq_file *s, void *data)
> +{
> +	struct airoha_eth *eth = s->private;
> +	int i;
> +
...
> +static int airoha_xmit_queues_show(struct seq_file *s, void *data)
> +{
> +	struct airoha_eth *eth = s->private;
> +	int i;
> +

Isn't this information available through ethtool already?

> b/drivers/net/ethernet/mediatek/airoha_eth.h
> new file mode 100644
> index 000000000000..fcd684e1418a
> --- /dev/null
> +++ b/drivers/net/ethernet/mediatek/airoha_eth.h
> @@ -0,0 +1,793 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2024 Lorenzo Bianconi <lorenzo@kernel.org>
> + */
> +
> +#define AIROHA_MAX_NUM_RSTS		3
> +#define AIROHA_MAX_NUM_XSI_RSTS		4

If your driver only has a single .c file, I would suggest moving all the
contents of the .h file into that as well for better readability.

      Arnd
Lorenzo Bianconi June 28, 2024, 5:27 p.m. UTC | #5
> On Tue, Jun 18, 2024, at 09:49, Lorenzo Bianconi wrote:
> > Add airoha_eth driver in order to introduce ethernet support for
> > Airoha EN7581 SoC available on EN7581 development board (en7581-evb).
> > en7581-evb networking architecture is composed by airoha_eth as mac
> > controller (cpu port) and a mt7530 dsa based switch.
> > EN7581 mac controller is mainly composed by Frame Engine (FE) and
> > QoS-DMA (QDMA) modules. FE is used for traffic offloading (just basic
> > functionalities are supported now) while QDMA is used for DMA operation
> > and QOS functionalities between mac layer and the dsa switch (hw QoS is
> > not available yet and it will be added in the future).
> > Currently only hw lan features are available, hw wan will be added with
> > subsequent patches.
> >
> > Tested-by: Benjamin Larsson <benjamin.larsson@genexis.eu>
> > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> 
> I noticed a few small things that you may want to improve:

Hi Arnd,

thx for the review.

> 
> > +static void airoha_qdma_set_irqmask(struct airoha_eth *eth, int index,
> > +				    u32 clear, u32 set)
> > +{
> > +	unsigned long flags;
> > +
> > +	if (WARN_ON_ONCE(index >= ARRAY_SIZE(eth->irqmask)))
> > +		return;
> > +
> > +	spin_lock_irqsave(&eth->irq_lock, flags);
> > +
> > +	eth->irqmask[index] &= ~clear;
> > +	eth->irqmask[index] |= set;
> > +	airoha_qdma_wr(eth, REG_INT_ENABLE(index), eth->irqmask[index]);
> > +
> > +	spin_unlock_irqrestore(&eth->irq_lock, flags);
> > +}
> 
> spin_lock_irqsave() is fairly expensive here, and it doesn't
> actually protect the register write since that is posted
> and can leak out of the spinlock.
> 
> You can probably just remove the lock and instead do the mask
> with atomic_cmpxchg() here.

I did not get what you mean here. I guess the spin_lock is used to avoid
concurrent irq registers updates from user/bh context or irq handler.
Am I missing something?

> 
> > +
> > +		dma_sync_single_for_device(dev, e->dma_addr, e->dma_len, dir);
> > +
> > +		val = FIELD_PREP(QDMA_DESC_LEN_MASK, e->dma_len);
> > +		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
> > +		WRITE_ONCE(desc->addr, cpu_to_le32(e->dma_addr));
> > +		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, q->head);
> > +		WRITE_ONCE(desc->data, cpu_to_le32(val));
> > +		WRITE_ONCE(desc->msg0, 0);
> > +		WRITE_ONCE(desc->msg1, 0);
> > +		WRITE_ONCE(desc->msg2, 0);
> > +		WRITE_ONCE(desc->msg3, 0);
> > +
> > +		wmb();
> > +		airoha_qdma_rmw(eth, REG_RX_CPU_IDX(qid), RX_RING_CPU_IDX_MASK,
> > +				FIELD_PREP(RX_RING_CPU_IDX_MASK, q->head));
> 
> The wmb() between the descriptor write and the MMIO does nothing
> and can probably just be removed here, a writel() already contains
> all the barriers you need to make DMA memory visible before the
> MMIO write.
> 
> If there is a chance that the device might read the descriptor
> while it is being updated by you have not written to the
> register, there should be a barrier before the last store to
> the descriptor that sets a 'valid' bit. That one can be a
> cheaper dma_wmb() then.

ack, I will fix it in v4.

> 
> > +static irqreturn_t airoha_irq_handler(int irq, void *dev_instance)
> > +{
> > +	struct airoha_eth *eth = dev_instance;
> > +	u32 intr[ARRAY_SIZE(eth->irqmask)];
> > +	int i;
> > +
> > +	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++) {
> > +		intr[i] = airoha_qdma_rr(eth, REG_INT_STATUS(i));
> > +		intr[i] &= eth->irqmask[i];
> > +		airoha_qdma_wr(eth, REG_INT_STATUS(i), intr[i]);
> > +	}
> 
> This looks like you send an interrupt Ack to each
> interrupt in order to re-arm it, but then you disable
> it again. Would it be possible to leave the interrupt enabled
> but defer the Ack until the napi poll function is completed?

I guess doing so we are not using NAPIs as expected since they are
supposed to run with interrupt disabled. Agree?

> 
> > +	if (!test_bit(DEV_STATE_INITIALIZED, &eth->state))
> > +		return IRQ_NONE;
> > +
> > +	if (intr[1] & RX_DONE_INT_MASK) {
> > +		airoha_qdma_irq_disable(eth, QDMA_INT_REG_IDX1,
> > +					RX_DONE_INT_MASK);
> > +		airoha_qdma_for_each_q_rx(eth, i) {
> > +			if (intr[1] & BIT(i))
> > +				napi_schedule(&eth->q_rx[i].napi);
> > +		}
> > +	}
> 
> Something seems wrong here, but that's probably just me
> misunderstanding the design: if all queues are signaled
> through the same interrupt handler, and you then do
> napi_schedule() for each queue, I would expect them to
> just all get run on the same CPU.
> 
> If you have separate queues, doesn't that mean you also need
> separate irq numbers here so they can be distributed to the
> available CPUs?

Actually I missed to mark the NAPI as threaded. Doing so, even if we have a
single irq line shared by all Rx queues, the scheduler can run the NAPIs in
parallel on different CPUs. I will fix it in v4.

> 
> > +		val = FIELD_PREP(QDMA_DESC_LEN_MASK, len);
> > +		if (i < nr_frags - 1)
> > +			val |= FIELD_PREP(QDMA_DESC_MORE_MASK, 1);
> > +		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
> > +		WRITE_ONCE(desc->addr, cpu_to_le32(addr));
> > +		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, index);
> > +		WRITE_ONCE(desc->data, cpu_to_le32(val));
> > +		WRITE_ONCE(desc->msg0, cpu_to_le32(msg0));
> > +		WRITE_ONCE(desc->msg1, cpu_to_le32(msg1));
> > +		WRITE_ONCE(desc->msg2, cpu_to_le32(0xffff));
> > +
> > +		e->skb = i ? NULL : skb;
> > +		e->dma_addr = addr;
> > +		e->dma_len = len;
> > +
> > +		wmb();
> > +		airoha_qdma_rmw(eth, REG_TX_CPU_IDX(qid), TX_RING_CPU_IDX_MASK,
> > +				FIELD_PREP(TX_RING_CPU_IDX_MASK, index));
> 
> Same as above with the wmb().

ack, I will fix it in v4.

> 
> > +static int airoha_rx_queues_show(struct seq_file *s, void *data)
> > +{
> > +	struct airoha_eth *eth = s->private;
> > +	int i;
> > +
> ...
> > +static int airoha_xmit_queues_show(struct seq_file *s, void *data)
> > +{
> > +	struct airoha_eth *eth = s->private;
> > +	int i;
> > +
> 
> Isn't this information available through ethtool already?

I guess we can get rid of this debugfs since it was just for debugging.

> 
> > b/drivers/net/ethernet/mediatek/airoha_eth.h
> > new file mode 100644
> > index 000000000000..fcd684e1418a
> > --- /dev/null
> > +++ b/drivers/net/ethernet/mediatek/airoha_eth.h
> > @@ -0,0 +1,793 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Copyright (C) 2024 Lorenzo Bianconi <lorenzo@kernel.org>
> > + */
> > +
> > +#define AIROHA_MAX_NUM_RSTS		3
> > +#define AIROHA_MAX_NUM_XSI_RSTS		4
> 
> If your driver only has a single .c file, I would suggest moving all the
> contents of the .h file into that as well for better readability.

I do not have a strong opinion about it but since we will extend the driver
in the future, keeping .c and .h in different files, seems a bit more tidy.
What do you think?

Regards,
Lorenzo

> 
>       Arnd
Arnd Bergmann June 28, 2024, 7:34 p.m. UTC | #6
On Fri, Jun 28, 2024, at 19:27, Lorenzo Bianconi wrote:
>> > +static void airoha_qdma_set_irqmask(struct airoha_eth *eth, int index,
>> > +				    u32 clear, u32 set)
>> > +{
>> > +	unsigned long flags;
>> > +
>> > +	if (WARN_ON_ONCE(index >= ARRAY_SIZE(eth->irqmask)))
>> > +		return;
>> > +
>> > +	spin_lock_irqsave(&eth->irq_lock, flags);
>> > +
>> > +	eth->irqmask[index] &= ~clear;
>> > +	eth->irqmask[index] |= set;
>> > +	airoha_qdma_wr(eth, REG_INT_ENABLE(index), eth->irqmask[index]);
>> > +
>> > +	spin_unlock_irqrestore(&eth->irq_lock, flags);
>> > +}
>> 
>> spin_lock_irqsave() is fairly expensive here, and it doesn't
>> actually protect the register write since that is posted
>> and can leak out of the spinlock.
>> 
>> You can probably just remove the lock and instead do the mask
>> with atomic_cmpxchg() here.
>
> I did not get what you mean here. I guess the spin_lock is used to avoid
> concurrent irq registers updates from user/bh context or irq handler.
> Am I missing something?

What I meant is that the airoha_qdma_wr() doesn't complete
until after the unlock because this is a normal 'posted' ioremap()
mapping.

>> > +static irqreturn_t airoha_irq_handler(int irq, void *dev_instance)
>> > +{
>> > +	struct airoha_eth *eth = dev_instance;
>> > +	u32 intr[ARRAY_SIZE(eth->irqmask)];
>> > +	int i;
>> > +
>> > +	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++) {
>> > +		intr[i] = airoha_qdma_rr(eth, REG_INT_STATUS(i));
>> > +		intr[i] &= eth->irqmask[i];
>> > +		airoha_qdma_wr(eth, REG_INT_STATUS(i), intr[i]);
>> > +	}
>> 
>> This looks like you send an interrupt Ack to each
>> interrupt in order to re-arm it, but then you disable
>> it again. Would it be possible to leave the interrupt enabled
>> but defer the Ack until the napi poll function is completed?
>
> I guess doing so we are not using NAPIs as expected since they are
> supposed to run with interrupt disabled. Agree?

The idea of NAPI is that you don't get the same interrupt
again until all remaining events have been processed.

How this is achieved is device dependent, and it can either
be done by masking the interrupt as you do here, or by
not rearming the interrupt if it gets turned off automatically
by the hardware. My guess is that writing to REG_INT_STATUS(i)
is the rearming here, but the device documentation should
clarify that. It's also possible that this is an Ack that
is required so you don't immediately get another interrupt.

>> > +	if (!test_bit(DEV_STATE_INITIALIZED, &eth->state))
>> > +		return IRQ_NONE;
>> > +
>> > +	if (intr[1] & RX_DONE_INT_MASK) {
>> > +		airoha_qdma_irq_disable(eth, QDMA_INT_REG_IDX1,
>> > +					RX_DONE_INT_MASK);
>> > +		airoha_qdma_for_each_q_rx(eth, i) {
>> > +			if (intr[1] & BIT(i))
>> > +				napi_schedule(&eth->q_rx[i].napi);
>> > +		}
>> > +	}
>> 
>> Something seems wrong here, but that's probably just me
>> misunderstanding the design: if all queues are signaled
>> through the same interrupt handler, and you then do
>> napi_schedule() for each queue, I would expect them to
>> just all get run on the same CPU.
>> 
>> If you have separate queues, doesn't that mean you also need
>> separate irq numbers here so they can be distributed to the
>> available CPUs?
>
> Actually I missed to mark the NAPI as threaded. Doing so, even if we have a
> single irq line shared by all Rx queues, the scheduler can run the NAPIs in
> parallel on different CPUs. I will fix it in v4.

Ok. It's a bit disappointing that the hardware integration
messed this up by not having multiple IRQ lines because
that adds a lot of latency, but I guess there is not much else
you can do about it.

>> > b/drivers/net/ethernet/mediatek/airoha_eth.h
>> > new file mode 100644
>> > index 000000000000..fcd684e1418a
>> > --- /dev/null
>> > +++ b/drivers/net/ethernet/mediatek/airoha_eth.h
>> > @@ -0,0 +1,793 @@
>> > +// SPDX-License-Identifier: GPL-2.0
>> > +/*
>> > + * Copyright (C) 2024 Lorenzo Bianconi <lorenzo@kernel.org>
>> > + */
>> > +
>> > +#define AIROHA_MAX_NUM_RSTS		3
>> > +#define AIROHA_MAX_NUM_XSI_RSTS		4
>> 
>> If your driver only has a single .c file, I would suggest moving all the
>> contents of the .h file into that as well for better readability.
>
> I do not have a strong opinion about it but since we will extend the driver
> in the future, keeping .c and .h in different files, seems a bit more tidy.
> What do you think?

I would still keep it all in one file. If you extend it to multiple
files later, it's easy to move those parts that you actually need to
share into an interface header. Most likely the majority of the current contents won't be needed by other files.

      Arnd
Felix Fietkau June 28, 2024, 7:57 p.m. UTC | #7
On 28.06.24 21:34, Arnd Bergmann wrote:
>>> > +static irqreturn_t airoha_irq_handler(int irq, void *dev_instance)
>>> > +{
>>> > +	struct airoha_eth *eth = dev_instance;
>>> > +	u32 intr[ARRAY_SIZE(eth->irqmask)];
>>> > +	int i;
>>> > +
>>> > +	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++) {
>>> > +		intr[i] = airoha_qdma_rr(eth, REG_INT_STATUS(i));
>>> > +		intr[i] &= eth->irqmask[i];
>>> > +		airoha_qdma_wr(eth, REG_INT_STATUS(i), intr[i]);
>>> > +	}
>>> 
>>> This looks like you send an interrupt Ack to each
>>> interrupt in order to re-arm it, but then you disable
>>> it again. Would it be possible to leave the interrupt enabled
>>> but defer the Ack until the napi poll function is completed?
>>
>> I guess doing so we are not using NAPIs as expected since they are
>> supposed to run with interrupt disabled. Agree?
> 
> The idea of NAPI is that you don't get the same interrupt
> again until all remaining events have been processed.
> 
> How this is achieved is device dependent, and it can either
> be done by masking the interrupt as you do here, or by
> not rearming the interrupt if it gets turned off automatically
> by the hardware. My guess is that writing to REG_INT_STATUS(i)
> is the rearming here, but the device documentation should
> clarify that. It's also possible that this is an Ack that
> is required so you don't immediately get another interrupt.

The interrupt handling of this hardware is pretty much the same as what 
many other devices do: the interrupt status is not automatically cleared 
by the hardware, so the write at the beginning of the interrupt handler 
does that explicitly.
Within the same handler, the interrupt is then masked to ensure that it 
does not fire again until the NAPI poll has completed.

Performing the status write after the poll has completed would be wrong, 
since that leaves open a small race window where events might be missed.

- Felix
diff mbox series

Patch

diff --git a/MAINTAINERS b/MAINTAINERS
index e66b7d4324ae..5fac9e3a8577 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -682,6 +682,16 @@  S:	Supported
 F:	fs/aio.c
 F:	include/linux/*aio*.h
 
+AIROHA ETHERNET DRIVER
+M:	Lorenzo Bianconi <lorenzo@kernel.org>
+L:	linux-arm-kernel@lists.infradead.org (moderated for non-subscribers)
+L:	linux-mediatek@lists.infradead.org (moderated for non-subscribers)
+L:	netdev@vger.kernel.org
+S:	Maintained
+F:	Documentation/devicetree/bindings/net/airoha,en7581.yaml
+F:	drivers/net/ethernet/mediatek/airoha_eth.c
+F:	drivers/net/ethernet/mediatek/airoha_eth.h
+
 AIROHA SPI SNFI DRIVER
 M:	Lorenzo Bianconi <lorenzo@kernel.org>
 M:	Ray Liu <ray.liu@airoha.com>
diff --git a/drivers/net/ethernet/mediatek/Kconfig b/drivers/net/ethernet/mediatek/Kconfig
index da0db417ab69..e16d0ea5d418 100644
--- a/drivers/net/ethernet/mediatek/Kconfig
+++ b/drivers/net/ethernet/mediatek/Kconfig
@@ -1,12 +1,21 @@ 
 # SPDX-License-Identifier: GPL-2.0-only
 config NET_VENDOR_MEDIATEK
 	bool "MediaTek devices"
-	depends on ARCH_MEDIATEK || SOC_MT7621 || SOC_MT7620 || COMPILE_TEST
+	depends on ARCH_MEDIATEK || ARCH_AIROHA || SOC_MT7621 || SOC_MT7620 || COMPILE_TEST
 	help
 	  If you have a Mediatek SoC with ethernet, say Y.
 
 if NET_VENDOR_MEDIATEK
 
+config NET_AIROHA
+	tristate "Airoha SoC Gigabit Ethernet support"
+	depends on NET_DSA || !NET_DSA
+	select PAGE_POOL
+	select PAGE_POOL_STATS
+	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 03e008fbc859..ddbb7f4a516c 100644
--- a/drivers/net/ethernet/mediatek/Makefile
+++ b/drivers/net/ethernet/mediatek/Makefile
@@ -11,3 +11,4 @@  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
diff --git a/drivers/net/ethernet/mediatek/airoha_eth.c b/drivers/net/ethernet/mediatek/airoha_eth.c
new file mode 100644
index 000000000000..5a2a7c610848
--- /dev/null
+++ b/drivers/net/ethernet/mediatek/airoha_eth.c
@@ -0,0 +1,1804 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (c) 2024 AIROHA Inc
+ * Author: Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+#include <linux/debugfs.h>
+#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 <net/dsa.h>
+#include <net/page_pool/helpers.h>
+#include <uapi/linux/ppp_defs.h>
+#include "airoha_eth.h"
+
+static u32 airoha_rr(void __iomem *base, u32 offset)
+{
+	return readl(base + offset);
+}
+
+static 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)
+{
+	val |= (airoha_rr(base, offset) & ~mask);
+	airoha_wr(base, offset, 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(eth, offset)		airoha_rr((eth)->qdma_regs, (offset))
+#define airoha_qdma_wr(eth, offset, val)	airoha_wr((eth)->qdma_regs, (offset), (val))
+#define airoha_qdma_rmw(eth, offset, mask, val)	airoha_rmw((eth)->qdma_regs, (offset), (mask), (val))
+#define airoha_qdma_set(eth, offset, val)	airoha_rmw((eth)->qdma_regs, (offset), 0, (val))
+#define airoha_qdma_clear(eth, offset, val)	airoha_rmw((eth)->qdma_regs, (offset), (val), 0)
+
+static void airoha_qdma_set_irqmask(struct airoha_eth *eth, int index,
+				    u32 clear, u32 set)
+{
+	unsigned long flags;
+
+	if (WARN_ON_ONCE(index >= ARRAY_SIZE(eth->irqmask)))
+		return;
+
+	spin_lock_irqsave(&eth->irq_lock, flags);
+
+	eth->irqmask[index] &= ~clear;
+	eth->irqmask[index] |= set;
+	airoha_qdma_wr(eth, REG_INT_ENABLE(index), eth->irqmask[index]);
+
+	spin_unlock_irqrestore(&eth->irq_lock, flags);
+}
+
+static void airoha_qdma_irq_enable(struct airoha_eth *eth, int index,
+				   u32 mask)
+{
+	airoha_qdma_set_irqmask(eth, index, 0, mask);
+}
+
+static void airoha_qdma_irq_disable(struct airoha_eth *eth, int index,
+				    u32 mask)
+{
+	airoha_qdma_set_irqmask(eth, index, mask, 0);
+}
+
+static void airoha_set_macaddr(struct airoha_eth *eth, const u8 *addr)
+{
+	u32 val;
+
+	val = (addr[0] << 16) | (addr[1] << 8) | addr[2];
+	airoha_fe_wr(eth, REG_FE_LAN_MAC_H, val);
+
+	val = (addr[3] << 16) | (addr[4] << 8) | addr[5];
+	airoha_fe_wr(eth, REG_FE_LAN_MAC_LMIN, val);
+	airoha_fe_wr(eth, REG_FE_LAN_MAC_LMAX, val);
+}
+
+static void airoha_set_port_fwd_cfg(struct airoha_eth *eth, u32 addr, u32 val)
+{
+	airoha_fe_rmw(eth, addr, GDM1_OCFQ_MASK,
+		      FIELD_PREP(GDM1_OCFQ_MASK, val));
+	airoha_fe_rmw(eth, addr, GDM1_MCFQ_MASK,
+		      FIELD_PREP(GDM1_MCFQ_MASK, val));
+	airoha_fe_rmw(eth, addr, GDM1_BCFQ_MASK,
+		      FIELD_PREP(GDM1_BCFQ_MASK, val));
+	airoha_fe_rmw(eth, addr, GDM1_UCFQ_MASK,
+		      FIELD_PREP(GDM1_UCFQ_MASK, val));
+}
+
+static int airoha_set_gdma_port(struct airoha_eth *eth, int port, bool enable)
+{
+	u32 vip_port, cfg_addr, val = enable ? FE_DP_PPE : FE_DP_DROP;
+
+	switch (port) {
+	case XSI_PCIE0_PORT:
+		vip_port = XSI_PCIE0_VIP_PORT_MASK;
+		cfg_addr = REG_GDM3_FWD_CFG;
+		break;
+	case XSI_PCIE1_PORT:
+		vip_port = XSI_PCIE1_VIP_PORT_MASK;
+		cfg_addr = REG_GDM3_FWD_CFG;
+		break;
+	case XSI_USB_PORT:
+		vip_port = XSI_USB_VIP_PORT_MASK;
+		cfg_addr = REG_GDM4_FWD_CFG;
+		break;
+	case XSI_ETH_PORT:
+		vip_port = XSI_ETH_VIP_PORT_MASK;
+		cfg_addr = REG_GDM4_FWD_CFG;
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	if (enable) {
+		airoha_fe_set(eth, REG_FE_VIP_PORT_EN, vip_port);
+		airoha_fe_set(eth, REG_FE_IFC_PORT_EN, vip_port);
+	} else {
+		airoha_fe_clear(eth, REG_FE_VIP_PORT_EN, vip_port);
+		airoha_fe_clear(eth, REG_FE_IFC_PORT_EN, vip_port);
+	}
+
+	airoha_set_port_fwd_cfg(eth, cfg_addr, val);
+
+	return 0;
+}
+
+static int airoha_set_gdma_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_gdma_port(eth, port_list[i], enable);
+		if (err)
+			goto error;
+	}
+
+	return 0;
+
+error:
+	for (i--; i >= 0; i++)
+		airoha_set_gdma_port(eth, port_list[i], false);
+
+	return err;
+}
+
+static void airoha_fe_maccr_init(struct airoha_eth *eth)
+{
+	airoha_fe_set(eth, REG_GDM1_FWD_CFG,
+		      GDM1_TCP_CKSUM | GDM1_UDP_CKSUM | GDM1_IP4_CKSUM |
+		      GDM1_DROP_CRC_ERR);
+	airoha_set_port_fwd_cfg(eth, REG_GDM1_FWD_CFG, FE_DP_CPU);
+
+	airoha_fe_set(eth, REG_FE_CPORT_CFG, FE_CPORT_PAD);
+	airoha_fe_rmw(eth, REG_CDM1_VLAN_CTRL, CDM1_VLAN_MASK,
+		      FIELD_PREP(CDM1_VLAN_MASK, 0x8100));
+	airoha_fe_rmw(eth, REG_GDM1_LEN_CFG,
+		      GDM1_SHORT_LEN_MASK | GDM1_LONG_LEN_MASK,
+		      FIELD_PREP(GDM1_SHORT_LEN_MASK, 60) |
+		      FIELD_PREP(GDM1_LONG_LEN_MASK, 4004));
+}
+
+static void airoha_fe_vip_setup(struct airoha_eth *eth)
+{
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(3), ETH_P_PPP_DISC);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(3), PATN_FCPU_EN_MASK | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(4), PPP_LCP);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(4),
+		     PATN_FCPU_EN_MASK | FIELD_PREP(PATN_TYPE_MASK, 1) |
+		     PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(6), PPP_IPCP);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(6),
+		     PATN_FCPU_EN_MASK | FIELD_PREP(PATN_TYPE_MASK, 1) |
+		     PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(7), PPP_CHAP);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(7),
+		     PATN_FCPU_EN_MASK | FIELD_PREP(PATN_TYPE_MASK, 1) |
+		     PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(8), 0x43); /* BOOTP (0x43) */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(8),
+		     PATN_FCPU_EN_MASK | PATN_SP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 4) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(9), 0x44); /* BOOTP (0x44) */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(9),
+		     PATN_FCPU_EN_MASK | PATN_SP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 4) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(10), 0x1f401f4); /* ISAKMP */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(10),
+		     PATN_FCPU_EN_MASK | PATN_DP_EN_MASK | PATN_SP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 4) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(11), 0xc057); /* PPP->IPv6CP (0xc057) */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(11),
+		     PATN_FCPU_EN_MASK | FIELD_PREP(PATN_TYPE_MASK, 1) |
+		     PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(12), 0x2220223); /* DHCPv6 */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(12),
+		     PATN_FCPU_EN_MASK | PATN_DP_EN_MASK | PATN_SP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 4) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(17), 0x1ae0);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(17),
+		     PATN_FCPU_EN_MASK | PATN_SP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(18), 0x1ae00000);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(18),
+		     PATN_FCPU_EN_MASK | PATN_DP_EN_MASK |
+		     FIELD_PREP(PATN_TYPE_MASK, 3) | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(19), PPP_PAP);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(19),
+		     PATN_FCPU_EN_MASK | FIELD_PREP(PATN_TYPE_MASK, 1) |
+		     PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(20), 0x893a); /* ETH->ETH_P_1905 (0x893a) */
+	airoha_fe_wr(eth, REG_FE_VIP_EN(20),
+		     PATN_FCPU_EN_MASK | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(21), ETH_P_LLDP);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(21),
+		     PATN_FCPU_EN_MASK | PATN_EN_MASK);
+
+	airoha_fe_wr(eth, REG_FE_VIP_PATN(22), 0xaaaa);
+	airoha_fe_wr(eth, REG_FE_VIP_EN(22),
+		     PATN_FCPU_EN_MASK | PATN_EN_MASK);
+}
+
+static u32 airoha_fe_get_oq_rsv(struct airoha_eth *eth,
+				u32 port, u32 queue)
+{
+	u32 val;
+
+	airoha_fe_rmw(eth, REG_FE_PSE_QUEUE_CFG_WR,
+		      PSE_CFG_PORT_ID_MASK | PSE_CFG_QUEUE_ID_MASK,
+		      FIELD_PREP(PSE_CFG_PORT_ID_MASK, port) |
+		      FIELD_PREP(PSE_CFG_QUEUE_ID_MASK, queue));
+	val = airoha_fe_rr(eth, REG_FE_PSE_QUEUE_CFG_VAL);
+
+	return FIELD_GET(PSE_CFG_OQ_RSV_MASK, val);
+}
+
+static void airoha_fe_set_oq_rsv(struct airoha_eth *eth,
+				 u32 port, u32 queue, u32 val)
+{
+	airoha_fe_rmw(eth, REG_FE_PSE_QUEUE_CFG_VAL, PSE_CFG_OQ_RSV_MASK,
+		      FIELD_PREP(PSE_CFG_OQ_RSV_MASK, val));
+	airoha_fe_rmw(eth, REG_FE_PSE_QUEUE_CFG_WR,
+		      PSE_CFG_PORT_ID_MASK | PSE_CFG_QUEUE_ID_MASK |
+		      PSE_CFG_WR_EN_MASK | PSE_CFG_OQRSV_SEL_MASK,
+		      FIELD_PREP(PSE_CFG_PORT_ID_MASK, port) |
+		      FIELD_PREP(PSE_CFG_QUEUE_ID_MASK, queue) |
+		      PSE_CFG_WR_EN_MASK | PSE_CFG_OQRSV_SEL_MASK);
+}
+
+static int airoha_fe_set_pse_oq_rsv(struct airoha_eth *eth,
+				    u32 port, u32 queue, u32 val)
+{
+	u32 orig_val, tmp, all_rsv, fq_limit;
+	const u32 pse_port_oq_id[] = {
+		PSE_PORT0_QUEUE,
+		PSE_PORT1_QUEUE,
+		PSE_PORT2_QUEUE,
+		PSE_PORT3_QUEUE,
+		PSE_PORT4_QUEUE,
+		PSE_PORT5_QUEUE,
+		PSE_PORT6_QUEUE,
+		PSE_PORT7_QUEUE,
+		PSE_PORT8_QUEUE,
+		PSE_PORT9_QUEUE,
+		PSE_PORT10_QUEUE
+	};
+
+	if (port >= ARRAY_SIZE(pse_port_oq_id))
+		return -EINVAL;
+
+	if (queue >= pse_port_oq_id[port])
+		return -EINVAL;
+
+	airoha_fe_set_oq_rsv(eth, port, queue, val);
+
+	/* modify all rsv */
+	orig_val = airoha_fe_get_oq_rsv(eth, port, queue);
+	tmp = airoha_fe_rr(eth, REG_FE_PSE_BUF_SET);
+	all_rsv = FIELD_GET(PSE_ALLRSV_MASK, tmp);
+	all_rsv += (val - orig_val);
+	airoha_fe_rmw(eth, REG_FE_PSE_BUF_SET, PSE_ALLRSV_MASK,
+		      FIELD_PREP(PSE_ALLRSV_MASK, all_rsv));
+
+	/* modify hthd */
+	tmp = airoha_fe_rr(eth, PSE_FQ_CFG);
+	fq_limit = FIELD_GET(PSE_FQ_LIMIT_MASK, tmp);
+	tmp = fq_limit - all_rsv - 0x20;
+	airoha_fe_rmw(eth, REG_PSE_SHARE_USED_THD,
+		      PSE_SHARE_USED_HTHD_MASK,
+		      FIELD_PREP(PSE_SHARE_USED_HTHD_MASK, tmp));
+
+	tmp = fq_limit - all_rsv - 0x100;
+	airoha_fe_rmw(eth, REG_PSE_SHARE_USED_THD,
+		      PSE_SHARE_USED_MTHD_MASK,
+		      FIELD_PREP(PSE_SHARE_USED_MTHD_MASK, tmp));
+	tmp = (3 * tmp) >> 2;
+	airoha_fe_rmw(eth, REG_FE_PSE_BUF_SET,
+		      PSE_SHARE_USED_LTHD_MASK,
+		      FIELD_PREP(PSE_SHARE_USED_LTHD_MASK, tmp));
+
+	return 0;
+}
+
+static void airoha_fe_oq_rsv_init(struct airoha_eth *eth)
+{
+	int i;
+
+	/* hw misses PPE2 oq rsv */
+	airoha_fe_set(eth, REG_FE_PSE_BUF_SET,
+		      PSE_DEF_RSV_PAGE * PSE_PORT8_QUEUE);
+
+	for (i = 0; i < PSE_PORT0_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 0, i, 0x40);
+	for (i = 0; i < PSE_PORT1_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 1, i, 0x40);
+
+	for (i = 6; i < PSE_PORT2_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 2, i, 0);
+
+	for (i = 0; i < PSE_PORT3_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 3, i, 0x40);
+
+	airoha_fe_set_pse_oq_rsv(eth, 4, 0, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 4, 1, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 4, 2, 0);
+	airoha_fe_set_pse_oq_rsv(eth, 4, 3, 0);
+	airoha_fe_set_pse_oq_rsv(eth, 8, 0, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 8, 1, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 8, 2, 0);
+	airoha_fe_set_pse_oq_rsv(eth, 8, 3, 0);
+
+	for (i = 0; i < PSE_PORT5_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 5, i, 0x40);
+
+	for (i = 0; i < PSE_PORT6_QUEUE - 1; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 6, i, 0);
+
+	for (i = 4; i < PSE_PORT7_QUEUE; i++)
+		airoha_fe_set_pse_oq_rsv(eth, 7, i, 0x40);
+
+	airoha_fe_set_pse_oq_rsv(eth, 9, 0, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 9, 1, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 10, 0, 0x40);
+	airoha_fe_set_pse_oq_rsv(eth, 10, 1, 0x40);
+}
+
+static int airoha_fe_mc_vlan_clear(struct airoha_eth *eth)
+{
+	int i;
+
+	for (i = 0; i < AIROHA_FE_MC_MAX_VLAN_TABLE; i++) {
+		int err, j;
+		u32 val;
+
+		airoha_fe_wr(eth, REG_MC_VLAN_DATA, 0x0);
+
+		val = FIELD_PREP(MC_VLAN_CFG_TABLE_ID_MASK, i) |
+		      MC_VLAN_CFG_TABLE_SEL_MASK | MC_VLAN_CFG_RW_MASK;
+		airoha_fe_wr(eth, REG_MC_VLAN_CFG, val);
+		err = read_poll_timeout(airoha_fe_rr, val,
+					val & MC_VLAN_CFG_CMD_DONE_MASK,
+					USEC_PER_MSEC, 5 * USEC_PER_MSEC,
+					false, eth, REG_MC_VLAN_CFG);
+		if (err)
+			return err;
+
+		for (j = 0; j < AIROHA_FE_MC_MAX_VLAN_PORT; j++) {
+			airoha_fe_wr(eth, REG_MC_VLAN_DATA, 0x0);
+
+			val = FIELD_PREP(MC_VLAN_CFG_TABLE_ID_MASK, i) |
+			      FIELD_PREP(MC_VLAN_CFG_PORT_ID_MASK, j) |
+			      MC_VLAN_CFG_RW_MASK;
+			airoha_fe_wr(eth, REG_MC_VLAN_CFG, val);
+			err = read_poll_timeout(airoha_fe_rr, val,
+						val & MC_VLAN_CFG_CMD_DONE_MASK,
+						USEC_PER_MSEC, 5 * USEC_PER_MSEC,
+						false, eth, REG_MC_VLAN_CFG);
+			if (err)
+				return err;
+		}
+	}
+
+	return 0;
+}
+
+static void airoha_fe_crsn_qsel_init(struct airoha_eth *eth)
+{
+	/* CDM1_CRSN_QSEL */
+	airoha_fe_rmw(eth, REG_CDM1_CRSN_QSEL(CRSN_22 >> 2),
+		      CDM1_CRSN_QSEL_REASON_MASK(CRSN_22),
+		      FIELD_PREP(CDM1_CRSN_QSEL_REASON_MASK(CRSN_22),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM1_CRSN_QSEL(CRSN_08 >> 2),
+		      CDM1_CRSN_QSEL_REASON_MASK(CRSN_08),
+		      FIELD_PREP(CDM1_CRSN_QSEL_REASON_MASK(CRSN_08),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM1_CRSN_QSEL(CRSN_21 >> 2),
+		      CDM1_CRSN_QSEL_REASON_MASK(CRSN_21),
+		      FIELD_PREP(CDM1_CRSN_QSEL_REASON_MASK(CRSN_21),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM1_CRSN_QSEL(CRSN_24 >> 2),
+		      CDM1_CRSN_QSEL_REASON_MASK(CRSN_24),
+		      FIELD_PREP(CDM1_CRSN_QSEL_REASON_MASK(CRSN_24),
+				 CDM_CRSN_QSEL_Q6));
+	airoha_fe_rmw(eth, REG_CDM1_CRSN_QSEL(CRSN_25 >> 2),
+		      CDM1_CRSN_QSEL_REASON_MASK(CRSN_25),
+		      FIELD_PREP(CDM1_CRSN_QSEL_REASON_MASK(CRSN_25),
+				 CDM_CRSN_QSEL_Q1));
+	/* CDM2_CRSN_QSEL */
+	airoha_fe_rmw(eth, REG_CDM2_CRSN_QSEL(CRSN_08 >> 2),
+		      CDM2_CRSN_QSEL_REASON_MASK(CRSN_08),
+		      FIELD_PREP(CDM2_CRSN_QSEL_REASON_MASK(CRSN_08),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM2_CRSN_QSEL(CRSN_21 >> 2),
+		      CDM2_CRSN_QSEL_REASON_MASK(CRSN_21),
+		      FIELD_PREP(CDM2_CRSN_QSEL_REASON_MASK(CRSN_21),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM2_CRSN_QSEL(CRSN_22 >> 2),
+		      CDM2_CRSN_QSEL_REASON_MASK(CRSN_22),
+		      FIELD_PREP(CDM2_CRSN_QSEL_REASON_MASK(CRSN_22),
+				 CDM_CRSN_QSEL_Q1));
+	airoha_fe_rmw(eth, REG_CDM2_CRSN_QSEL(CRSN_24 >> 2),
+		      CDM2_CRSN_QSEL_REASON_MASK(CRSN_24),
+		      FIELD_PREP(CDM2_CRSN_QSEL_REASON_MASK(CRSN_24),
+				 CDM_CRSN_QSEL_Q6));
+	airoha_fe_rmw(eth, REG_CDM2_CRSN_QSEL(CRSN_25 >> 2),
+		      CDM2_CRSN_QSEL_REASON_MASK(CRSN_25),
+		      FIELD_PREP(CDM2_CRSN_QSEL_REASON_MASK(CRSN_25),
+				 CDM_CRSN_QSEL_Q1));
+}
+
+static int airoha_fe_init(struct airoha_eth *eth)
+{
+	airoha_fe_maccr_init(eth);
+
+	/* PSE IQ reserve */
+	airoha_fe_rmw(eth, REG_PSE_IQ_REV1, PSE_IQ_RES1_P2_MASK,
+		      FIELD_PREP(PSE_IQ_RES1_P2_MASK, 0x10));
+	airoha_fe_rmw(eth, REG_PSE_IQ_REV2,
+		      PSE_IQ_RES2_P5_MASK | PSE_IQ_RES2_P4_MASK,
+		      FIELD_PREP(PSE_IQ_RES2_P5_MASK, 0x40) |
+		      FIELD_PREP(PSE_IQ_RES2_P4_MASK, 0x34));
+
+	/* enable FE copy engine for MC/KA/DPI */
+	airoha_fe_wr(eth, REG_FE_PCE_CFG,
+		     PCE_DPI_EN_MASK | PCE_KA_EN_MASK | PCE_MC_EN_MASK);
+	/* set vip queue selection to ring 1 */
+	airoha_fe_rmw(eth, REG_CDM1_FWD_CFG, CDM1_VIP_QSEL_MASK,
+		      FIELD_PREP(CDM1_VIP_QSEL_MASK, 0x4));
+	airoha_fe_rmw(eth, REG_CDM2_FWD_CFG, CDM2_VIP_QSEL_MASK,
+		      FIELD_PREP(CDM2_VIP_QSEL_MASK, 0x4));
+	/* set GDM4 source interface offset to 8 */
+	airoha_fe_rmw(eth, REG_GDM4_SRC_PORT_SET,
+		      GDM4_SPORT_OFF2_MASK |
+		      GDM4_SPORT_OFF1_MASK |
+		      GDM4_SPORT_OFF0_MASK,
+		      FIELD_PREP(GDM4_SPORT_OFF2_MASK, 8) |
+		      FIELD_PREP(GDM4_SPORT_OFF1_MASK, 8) |
+		      FIELD_PREP(GDM4_SPORT_OFF0_MASK, 8));
+
+	/* set PSE Page as 128B */
+	airoha_fe_rmw(eth, REG_FE_DMA_GLO_CFG,
+		      FE_DMA_GLO_L2_SPACE_MASK | FE_DMA_GLO_PG_SZ_MASK,
+		      FIELD_PREP(FE_DMA_GLO_L2_SPACE_MASK, 2) |
+		      FE_DMA_GLO_PG_SZ_MASK);
+	airoha_fe_wr(eth, REG_FE_RST_GLO_CFG,
+		     FE_RST_CORE_MASK | FE_RST_GDM3_MBI_ARB_MASK |
+		     FE_RST_GDM4_MBI_ARB_MASK);
+	usleep_range(1000, 2000);
+
+	/* connect RxRing1 and RxRing15 to PSE Port0 OQ-1
+	 * connect other rings to PSE Port0 OQ-0
+	 */
+	airoha_fe_wr(eth, REG_FE_CDM1_OQ_MAP0, BIT(4));
+	airoha_fe_wr(eth, REG_FE_CDM1_OQ_MAP1, BIT(28));
+	airoha_fe_wr(eth, REG_FE_CDM1_OQ_MAP2, BIT(4));
+	airoha_fe_wr(eth, REG_FE_CDM1_OQ_MAP3, BIT(28));
+
+	airoha_fe_vip_setup(eth);
+	airoha_fe_oq_rsv_init(eth);
+
+	airoha_fe_set(eth, REG_GDM_MISC_CFG,
+		      GDM2_RDM_ACK_WAIT_PREF_MASK |
+		      GDM2_CHN_VLD_MODE_MASK);
+	airoha_fe_rmw(eth, REG_CDM2_FWD_CFG, CDM2_OAM_QSEL_MASK, 15);
+
+	/* init fragment and assemble Force Port */
+	/* NPU Core-3, NPU Bridge Channel-3 */
+	airoha_fe_rmw(eth, REG_IP_FRAG_FP,
+		      IP_FRAGMENT_PORT_MASK | IP_FRAGMENT_NBQ_MASK,
+		      FIELD_PREP(IP_FRAGMENT_PORT_MASK, 6) |
+		      FIELD_PREP(IP_FRAGMENT_NBQ_MASK, 3));
+	/* QDMA LAN, RX Ring-22 */
+	airoha_fe_rmw(eth, REG_IP_FRAG_FP,
+		      IP_ASSEMBLE_PORT_MASK | IP_ASSEMBLE_NBQ_MASK,
+		      FIELD_PREP(IP_ASSEMBLE_PORT_MASK, 0) |
+		      FIELD_PREP(IP_ASSEMBLE_NBQ_MASK, 22));
+
+	airoha_fe_set(eth, REG_GDM3_FWD_CFG, GDM3_PAD_EN_MASK);
+	airoha_fe_set(eth, REG_GDM4_FWD_CFG, GDM4_PAD_EN_MASK);
+
+	airoha_fe_crsn_qsel_init(eth);
+
+	airoha_fe_clear(eth, REG_FE_CPORT_CFG, FE_CPORT_QUEUE_XFC_MASK);
+	airoha_fe_set(eth, REG_FE_CPORT_CFG, FE_CPORT_PORT_XFC_MASK);
+
+	/* default aging mode for mbi unlock issue */
+	airoha_fe_rmw(eth, REG_GDM2_CHN_RLS,
+		      MBI_RX_AGE_SEL_MASK | MBI_TX_AGE_SEL_MASK,
+		      FIELD_PREP(MBI_RX_AGE_SEL_MASK, 3) |
+		      FIELD_PREP(MBI_TX_AGE_SEL_MASK, 3));
+
+	/* disable IFC by default */
+	airoha_fe_clear(eth, REG_FE_CSR_IFC_CFG, FE_IFC_EN_MASK);
+
+	/* enable 1:N vlan action, init vlan table */
+	airoha_fe_set(eth, REG_MC_VLAN_EN, MC_VLAN_EN_MASK);
+
+	return airoha_fe_mc_vlan_clear(eth);
+}
+
+static int airoha_qdma_fill_rx_queue(struct airoha_queue *q)
+{
+	enum dma_data_direction dir = page_pool_get_dma_dir(q->page_pool);
+	struct airoha_eth *eth = q->eth;
+	int qid, nframes = 0;
+	struct device *dev;
+
+	dev = eth->net_dev->dev.parent;
+	qid = q - &eth->q_rx[0];
+
+	while (q->queued < q->ndesc - 1) {
+		struct airoha_queue_entry *e = &q->entry[q->head];
+		struct airoha_qdma_desc *desc = &q->desc[q->head];
+		struct page *page;
+		int offset;
+		u32 val;
+
+		page = page_pool_dev_alloc_frag(q->page_pool, &offset,
+						q->buf_size);
+		if (!page)
+			break;
+
+		q->head = (q->head + 1) % q->ndesc;
+		q->queued++;
+		nframes++;
+
+		e->buf = page_address(page) + offset;
+		e->dma_addr = page_pool_get_dma_addr(page) + offset;
+		e->dma_len = SKB_WITH_OVERHEAD(q->buf_size);
+
+		dma_sync_single_for_device(dev, e->dma_addr, e->dma_len, dir);
+
+		val = FIELD_PREP(QDMA_DESC_LEN_MASK, e->dma_len);
+		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
+		WRITE_ONCE(desc->addr, cpu_to_le32(e->dma_addr));
+		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, q->head);
+		WRITE_ONCE(desc->data, cpu_to_le32(val));
+		WRITE_ONCE(desc->msg0, 0);
+		WRITE_ONCE(desc->msg1, 0);
+		WRITE_ONCE(desc->msg2, 0);
+		WRITE_ONCE(desc->msg3, 0);
+
+		wmb();
+		airoha_qdma_rmw(eth, REG_RX_CPU_IDX(qid), RX_RING_CPU_IDX_MASK,
+				FIELD_PREP(RX_RING_CPU_IDX_MASK, q->head));
+	}
+
+	return nframes;
+}
+
+static int airoha_qdma_rx_process(struct airoha_queue *q, int budget)
+{
+	enum dma_data_direction dir = page_pool_get_dma_dir(q->page_pool);
+	struct airoha_eth *eth = q->eth;
+	struct device *dev;
+	int done = 0, qid;
+
+	dev = eth->net_dev->dev.parent;
+	qid = q - &eth->q_rx[0];
+
+	while (done < budget) {
+		struct airoha_queue_entry *e = &q->entry[q->tail];
+		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 sk_buff *skb;
+		int len;
+
+		if (!(desc_ctrl & QDMA_DESC_DONE_MASK))
+			break;
+
+		if (!dma_addr)
+			break;
+
+		len = FIELD_GET(QDMA_DESC_LEN_MASK, desc_ctrl);
+		if (!len)
+			break;
+
+		q->tail = (q->tail + 1) % q->ndesc;
+		q->queued--;
+
+		dma_sync_single_for_cpu(dev, dma_addr,
+					SKB_WITH_OVERHEAD(q->buf_size), dir);
+
+		skb = napi_build_skb(e->buf, q->buf_size);
+		if (!skb) {
+			page_pool_put_full_page(q->page_pool,
+						virt_to_head_page(e->buf),
+						true);
+			break;
+		}
+
+		skb_reserve(skb, 2);
+		__skb_put(skb, len);
+
+		skb_mark_for_recycle(skb);
+		skb->dev = eth->net_dev;
+		skb->protocol = eth_type_trans(skb, eth->net_dev);
+		skb->ip_summed = CHECKSUM_UNNECESSARY;
+		skb_record_rx_queue(skb, qid);
+		napi_gro_receive(&q->napi, skb);
+
+		done++;
+	}
+	airoha_qdma_fill_rx_queue(q);
+
+	return done;
+}
+
+static int airoha_qdma_rx_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct airoha_queue *q = container_of(napi, struct airoha_queue, napi);
+	struct airoha_eth *eth = q->eth;
+	int cur, done = 0;
+
+	do {
+		cur = airoha_qdma_rx_process(q, budget - done);
+		done += cur;
+	} while (cur && done < budget);
+
+	if (done < budget && napi_complete(napi))
+		airoha_qdma_irq_enable(eth, QDMA_INT_REG_IDX1,
+				       RX_DONE_INT_MASK);
+
+	return done;
+}
+
+static int airoha_qdma_init_rx_queue(struct airoha_eth *eth,
+				     struct airoha_queue *q, int ndesc)
+{
+	struct device *dev = eth->net_dev->dev.parent;
+	struct page_pool_params pp_params = {
+		.order = 0,
+		.pool_size = 256,
+		.flags = PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV,
+		.dma_dir = DMA_FROM_DEVICE,
+		.max_len = PAGE_SIZE,
+		.nid = NUMA_NO_NODE,
+		.dev = dev,
+		.napi = &q->napi,
+	};
+	int qid = q - &eth->q_rx[0], thr;
+	dma_addr_t dma_addr;
+
+	q->buf_size = PAGE_SIZE / 2;
+	q->ndesc = ndesc;
+	q->eth = eth;
+
+	q->entry = devm_kzalloc(dev, q->ndesc * sizeof(*q->entry),
+				GFP_KERNEL);
+	if (!q->entry)
+		return -ENOMEM;
+
+	q->page_pool = page_pool_create(&pp_params);
+	if (IS_ERR(q->page_pool)) {
+		int err = PTR_ERR(q->page_pool);
+
+		q->page_pool = NULL;
+		return err;
+	}
+
+	q->desc = dmam_alloc_coherent(dev, q->ndesc * sizeof(*q->desc),
+				      &dma_addr, GFP_KERNEL);
+	if (!q->desc)
+		return -ENOMEM;
+
+	netif_napi_add(eth->net_dev, &q->napi, airoha_qdma_rx_napi_poll);
+
+	airoha_qdma_wr(eth, REG_RX_RING_BASE(qid), dma_addr);
+	airoha_qdma_rmw(eth, REG_RX_RING_SIZE(qid), RX_RING_SIZE_MASK,
+			FIELD_PREP(RX_RING_SIZE_MASK, ndesc));
+
+	thr = clamp(ndesc >> 3, 1, 32);
+	airoha_qdma_rmw(eth, REG_RX_RING_SIZE(qid), RX_RING_THR_MASK,
+			FIELD_PREP(RX_RING_THR_MASK, thr));
+	airoha_qdma_rmw(eth, REG_RX_DMA_IDX(qid), RX_RING_DMA_IDX_MASK,
+			FIELD_PREP(RX_RING_DMA_IDX_MASK, q->head));
+
+	airoha_qdma_fill_rx_queue(q);
+
+	return 0;
+}
+
+static void airoha_qdma_clenaup_rx_queue(struct airoha_queue *q)
+{
+	enum dma_data_direction dir = page_pool_get_dma_dir(q->page_pool);
+	struct airoha_eth *eth = q->eth;
+	struct device *dev;
+
+	dev = eth->net_dev->dev.parent;
+
+	while (q->queued) {
+		struct airoha_queue_entry *e = &q->entry[q->tail];
+		struct page *page = virt_to_head_page(e->buf);
+
+		dma_sync_single_for_cpu(dev, e->dma_addr, e->dma_len, dir);
+		page_pool_put_full_page(q->page_pool, page, false);
+		q->tail = (q->tail + 1) % q->ndesc;
+		q->queued--;
+	}
+}
+
+static int airoha_qdma_init_rx(struct airoha_eth *eth)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(eth->q_rx); i++) {
+		int err;
+
+		if (!(RX_DONE_INT_MASK & BIT(i))) {
+			/* rx-queue not binded to irq */
+			continue;
+		}
+
+		err = airoha_qdma_init_rx_queue(eth, &eth->q_rx[i],
+						RX_DSCP_NUM(i));
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static int airoha_qdma_tx_napi_poll(struct napi_struct *napi, int budget)
+{
+	struct airoha_tx_irq_queue *irq_q;
+	struct airoha_eth *eth;
+	struct device *dev;
+	int id, done = 0;
+
+	irq_q = container_of(napi, struct airoha_tx_irq_queue, napi);
+	eth = irq_q->eth;
+	id = irq_q - &eth->q_tx_irq[0];
+	dev = eth->net_dev->dev.parent;
+
+	while (irq_q->queued > 0 && done < budget) {
+		u32 qid, last, val = irq_q->q[irq_q->head];
+		u32 n_pkts = 0, n_bytes = 0;
+		struct netdev_queue *txq;
+		struct airoha_queue *q;
+
+		if (val == 0xff)
+			break;
+
+		irq_q->q[irq_q->head] = 0xff; /* mark as done */
+		irq_q->head = (irq_q->head + 1) % irq_q->size;
+		irq_q->queued--;
+		done++;
+
+		last = FIELD_GET(IRQ_DESC_IDX_MASK, val);
+		qid = FIELD_GET(IRQ_RING_IDX_MASK, val);
+
+		if (qid >= ARRAY_SIZE(eth->q_tx))
+			continue;
+
+		q = &eth->q_tx[qid];
+		if (!q->ndesc)
+			continue;
+
+		spin_lock_bh(&q->lock);
+
+		while (q->queued > 0) {
+			struct airoha_qdma_desc *desc = &q->desc[q->tail];
+			struct airoha_queue_entry *e = &q->entry[q->tail];
+			u32 desc_ctrl = le32_to_cpu(desc->ctrl);
+			u16 index = q->tail;
+
+			if (!(desc_ctrl & QDMA_DESC_DONE_MASK) &&
+			    !(desc_ctrl & QDMA_DESC_DROP_MASK))
+				break;
+
+			q->tail = (q->tail + 1) % q->ndesc;
+			q->queued--;
+
+			dma_unmap_single(dev, e->dma_addr, e->dma_len,
+					 DMA_TO_DEVICE);
+
+			if (e->skb) {
+				n_bytes += e->skb->len;
+				n_pkts++;
+			}
+
+			dev_kfree_skb_any(e->skb);
+			e->skb = NULL;
+
+			WRITE_ONCE(desc->msg0, 0);
+			WRITE_ONCE(desc->msg1, 0);
+
+			if (index == last)
+				break;
+		}
+
+		txq = netdev_get_tx_queue(eth->net_dev, qid);
+		netdev_tx_completed_queue(txq, n_pkts, n_bytes);
+
+		if (netif_tx_queue_stopped(txq) &&
+		    q->ndesc - q->queued >= q->free_thr)
+			netif_tx_wake_queue(txq);
+
+		spin_unlock_bh(&q->lock);
+	}
+
+	if (done) {
+		int i, len = done >> 7;
+
+		for (i = 0; i < len; i++)
+			airoha_qdma_rmw(eth, REG_IRQ_CLEAR_LEN(id),
+					IRQ_CLEAR_LEN_MASK, 0x80);
+		airoha_qdma_rmw(eth, REG_IRQ_CLEAR_LEN(id),
+				IRQ_CLEAR_LEN_MASK, (done & 0x7f));
+	}
+
+	if (done < budget && napi_complete(napi))
+		airoha_qdma_irq_enable(eth, QDMA_INT_REG_IDX0,
+				       TX_DONE_INT_MASK(id));
+
+	return done;
+}
+
+static int airoha_qdma_init_tx_queue(struct airoha_eth *eth,
+				     struct airoha_queue *q, int size)
+{
+	struct device *dev = eth->net_dev->dev.parent;
+	int i, qid = q - &eth->q_tx[0];
+	dma_addr_t dma_addr;
+
+	spin_lock_init(&q->lock);
+	q->ndesc = size;
+	q->eth = eth;
+	q->free_thr = MAX_SKB_FRAGS;
+
+	q->entry = devm_kzalloc(dev, q->ndesc * sizeof(*q->entry),
+				GFP_KERNEL);
+	if (!q->entry)
+		return -ENOMEM;
+
+	q->desc = dmam_alloc_coherent(dev, q->ndesc * sizeof(*q->desc),
+				      &dma_addr, GFP_KERNEL);
+	if (!q->desc)
+		return -ENOMEM;
+
+	for (i = 0; i < q->ndesc; i++) {
+		u32 val;
+
+		val = FIELD_PREP(QDMA_DESC_DONE_MASK, 1);
+		WRITE_ONCE(q->desc[i].ctrl, cpu_to_le32(val));
+	}
+
+	airoha_qdma_wr(eth, REG_TX_RING_BASE(qid), dma_addr);
+	airoha_qdma_rmw(eth, REG_TX_CPU_IDX(qid), TX_RING_CPU_IDX_MASK,
+			FIELD_PREP(TX_RING_CPU_IDX_MASK, q->head));
+	airoha_qdma_rmw(eth, REG_TX_DMA_IDX(qid), TX_RING_DMA_IDX_MASK,
+			FIELD_PREP(TX_RING_DMA_IDX_MASK, q->head));
+
+	return 0;
+}
+
+static int airoha_qdma_tx_irq_init(struct airoha_eth *eth,
+				   struct airoha_tx_irq_queue *irq_q,
+				   int size)
+{
+	struct device *dev = eth->net_dev->dev.parent;
+	int id = irq_q - &eth->q_tx_irq[0];
+	dma_addr_t dma_addr;
+
+	netif_napi_add(eth->net_dev, &irq_q->napi, airoha_qdma_tx_napi_poll);
+	irq_q->q = dmam_alloc_coherent(dev, size * sizeof(u32), &dma_addr,
+				       GFP_KERNEL);
+	if (!irq_q->q)
+		return -ENOMEM;
+
+	memset(irq_q->q, 0xff, size * sizeof(u32));
+	irq_q->size = size;
+	irq_q->eth = eth;
+
+	airoha_qdma_wr(eth, REG_TX_IRQ_BASE(id), dma_addr);
+	airoha_qdma_rmw(eth, REG_TX_IRQ_CFG(id), TX_IRQ_DEPTH_MASK,
+			FIELD_PREP(TX_IRQ_DEPTH_MASK, size));
+	airoha_qdma_rmw(eth, REG_TX_IRQ_CFG(id), TX_IRQ_THR_MASK,
+			FIELD_PREP(TX_IRQ_THR_MASK, 1));
+
+	return 0;
+}
+
+static int airoha_qdma_init_tx(struct airoha_eth *eth)
+{
+	int i, err;
+
+	for (i = 0; i < ARRAY_SIZE(eth->q_tx_irq); i++) {
+		err = airoha_qdma_tx_irq_init(eth, &eth->q_tx_irq[i],
+					      IRQ_QUEUE_LEN(i));
+		if (err)
+			return err;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(eth->q_tx); i++) {
+		err = airoha_qdma_init_tx_queue(eth, &eth->q_tx[i],
+						TX_DSCP_NUM);
+		if (err)
+			return err;
+	}
+
+	return 0;
+}
+
+static void airoha_qdma_clenaup_tx_queue(struct airoha_queue *q)
+{
+	struct airoha_eth *eth = q->eth;
+	struct device *dev;
+
+	dev = eth->net_dev->dev.parent;
+
+	spin_lock_bh(&q->lock);
+	while (q->queued) {
+		struct airoha_queue_entry *e = &q->entry[q->tail];
+
+		dma_unmap_single(dev, e->dma_addr, e->dma_len, DMA_TO_DEVICE);
+		dev_kfree_skb_any(e->skb);
+		e->skb = NULL;
+
+		q->tail = (q->tail + 1) % q->ndesc;
+		q->queued--;
+	}
+	spin_unlock_bh(&q->lock);
+}
+
+static int airoha_qdma_init_hfwd_queues(struct airoha_eth *eth)
+{
+	struct device *dev = eth->net_dev->dev.parent;
+	dma_addr_t dma_addr;
+	u32 status;
+	int size;
+
+	size = HW_DSCP_NUM * sizeof(struct airoha_qdma_fwd_desc);
+	eth->hfwd.desc = dmam_alloc_coherent(dev, size, &dma_addr,
+					     GFP_KERNEL);
+	if (!eth->hfwd.desc)
+		return -ENOMEM;
+
+	airoha_qdma_wr(eth, REG_FWD_DSCP_BASE, dma_addr);
+
+	size = AIROHA_MAX_PACKET_SIZE * HW_DSCP_NUM;
+	eth->hfwd.q = dmam_alloc_coherent(dev, size, &dma_addr, GFP_KERNEL);
+	if (!eth->hfwd.q)
+		return -ENOMEM;
+
+	airoha_qdma_wr(eth, REG_FWD_BUF_BASE, dma_addr);
+
+	airoha_qdma_rmw(eth, REG_HW_FWD_DSCP_CFG,
+			HW_FWD_DSCP_PAYLOAD_SIZE_MASK,
+			FIELD_PREP(HW_FWD_DSCP_PAYLOAD_SIZE_MASK, 0));
+	airoha_qdma_rmw(eth, REG_FWD_DSCP_LOW_THR, FWD_DSCP_LOW_THR_MASK,
+			FIELD_PREP(FWD_DSCP_LOW_THR_MASK, 128));
+	airoha_qdma_rmw(eth, REG_LMGR_INIT_CFG,
+			LMGR_INIT_START | LMGR_SRAM_MODE_MASK |
+			HW_FWD_DESC_NUM_MASK,
+			FIELD_PREP(HW_FWD_DESC_NUM_MASK, HW_DSCP_NUM) |
+			LMGR_INIT_START);
+
+	return read_poll_timeout(airoha_qdma_rr, status,
+				 !(status & LMGR_INIT_START), USEC_PER_MSEC,
+				 30 * USEC_PER_MSEC, true, eth,
+				 REG_LMGR_INIT_CFG);
+}
+
+static void airoha_qdma_init_qos(struct airoha_eth *eth)
+{
+	airoha_qdma_clear(eth, REG_TXWRR_MODE_CFG, TWRR_WEIGHT_SCALE_MASK);
+	airoha_qdma_set(eth, REG_TXWRR_MODE_CFG, TWRR_WEIGHT_BASE_MASK);
+
+	airoha_qdma_clear(eth, REG_PSE_BUF_USAGE_CFG,
+			  PSE_BUF_ESTIMATE_EN_MASK);
+
+	airoha_qdma_set(eth, REG_EGRESS_RATE_METER_CFG,
+			EGRESS_RATE_METER_EN_MASK |
+			EGRESS_RATE_METER_EQ_RATE_EN_MASK);
+	/* 2047us x 31 = 63.457ms */
+	airoha_qdma_rmw(eth, REG_EGRESS_RATE_METER_CFG,
+			EGRESS_RATE_METER_WINDOW_SZ_MASK,
+			FIELD_PREP(EGRESS_RATE_METER_WINDOW_SZ_MASK, 0x1f));
+	airoha_qdma_rmw(eth, REG_EGRESS_RATE_METER_CFG,
+			EGRESS_RATE_METER_TIMESLICE_MASK,
+			FIELD_PREP(EGRESS_RATE_METER_TIMESLICE_MASK, 0x7ff));
+
+	/* ratelimit init */
+	airoha_qdma_set(eth, REG_GLB_TRTCM_CFG, GLB_TRTCM_EN_MASK);
+	airoha_qdma_rmw(eth, REG_GLB_TRTCM_CFG, GLB_FAST_TICK_MASK,
+			FIELD_PREP(GLB_FAST_TICK_MASK, 25)); /* fast-tick 25us */
+	airoha_qdma_rmw(eth, REG_GLB_TRTCM_CFG, GLB_SLOW_TICK_RATIO_MASK,
+			FIELD_PREP(GLB_SLOW_TICK_RATIO_MASK, 40));
+
+	airoha_qdma_set(eth, REG_EGRESS_TRTCM_CFG, EGRESS_TRTCM_EN_MASK);
+	airoha_qdma_rmw(eth, REG_EGRESS_TRTCM_CFG, EGRESS_FAST_TICK_MASK,
+			FIELD_PREP(EGRESS_FAST_TICK_MASK, 25));
+	airoha_qdma_rmw(eth, REG_EGRESS_TRTCM_CFG,
+			EGRESS_SLOW_TICK_RATIO_MASK,
+			FIELD_PREP(EGRESS_SLOW_TICK_RATIO_MASK, 40));
+
+	airoha_qdma_set(eth, REG_INGRESS_TRTCM_CFG, INGRESS_TRTCM_EN_MASK);
+	airoha_qdma_clear(eth, REG_INGRESS_TRTCM_CFG,
+			  INGRESS_TRTCM_MODE_MASK);
+	airoha_qdma_rmw(eth, REG_INGRESS_TRTCM_CFG, INGRESS_FAST_TICK_MASK,
+			FIELD_PREP(INGRESS_FAST_TICK_MASK, 125));
+	airoha_qdma_rmw(eth, REG_INGRESS_TRTCM_CFG,
+			INGRESS_SLOW_TICK_RATIO_MASK,
+			FIELD_PREP(INGRESS_SLOW_TICK_RATIO_MASK, 8));
+
+	airoha_qdma_set(eth, REG_SLA_TRTCM_CFG, SLA_TRTCM_EN_MASK);
+	airoha_qdma_rmw(eth, REG_SLA_TRTCM_CFG, SLA_FAST_TICK_MASK,
+			FIELD_PREP(SLA_FAST_TICK_MASK, 25));
+	airoha_qdma_rmw(eth, REG_SLA_TRTCM_CFG, SLA_SLOW_TICK_RATIO_MASK,
+			FIELD_PREP(SLA_SLOW_TICK_RATIO_MASK, 40));
+}
+
+static int airoha_qdma_hw_init(struct airoha_eth *eth)
+{
+	int i;
+
+	/* clear pending irqs */
+	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++)
+		airoha_qdma_wr(eth, REG_INT_STATUS(i), 0xffffffff);
+
+	/* setup irqs */
+	airoha_qdma_irq_enable(eth, QDMA_INT_REG_IDX0, INT_IDX0_MASK);
+	airoha_qdma_irq_enable(eth, QDMA_INT_REG_IDX1, INT_IDX1_MASK);
+	airoha_qdma_irq_enable(eth, QDMA_INT_REG_IDX4, INT_IDX4_MASK);
+
+	/* setup irq binding */
+	airoha_qdma_for_each_q_tx(eth, i) {
+		if (TX_RING_IRQ_BLOCKING_MAP_MASK & BIT(i))
+			airoha_qdma_set(eth, REG_TX_RING_BLOCKING(i),
+					TX_RING_IRQ_BLOCKING_CFG_MASK);
+		else
+			airoha_qdma_clear(eth, REG_TX_RING_BLOCKING(i),
+					  TX_RING_IRQ_BLOCKING_CFG_MASK);
+	}
+
+	airoha_qdma_wr(eth, REG_QDMA_GLOBAL_CFG,
+		       GLOBAL_CFG_RX_2B_OFFSET_MASK |
+		       FIELD_PREP(GLOBAL_CFG_DMA_PREFERENCE_MASK, 3) |
+		       GLOBAL_CFG_CPU_TXR_RR_MASK |
+		       GLOBAL_CFG_PAYLOAD_BYTE_SWAP_MASK |
+		       GLOBAL_CFG_MULTICAST_MODIFY_FP_MASK |
+		       GLOBAL_CFG_MULTICAST_EN_MASK |
+		       GLOBAL_CFG_IRQ0_EN_MASK | GLOBAL_CFG_IRQ1_EN_MASK |
+		       GLOBAL_CFG_TX_WB_DONE_MASK |
+		       FIELD_PREP(GLOBAL_CFG_MAX_ISSUE_NUM_MASK, 2));
+
+	airoha_qdma_init_qos(eth);
+
+	/* disable qdma rx delay interrupt */
+	airoha_qdma_for_each_q_rx(eth, i)
+		airoha_qdma_clear(eth, REG_RX_DELAY_INT_IDX(i),
+				  RX_DELAY_INT_MASK);
+
+	airoha_qdma_set(eth, REG_TXQ_CNGST_CFG,
+			TXQ_CNGST_DROP_EN | TXQ_CNGST_DEI_DROP_EN);
+
+	return 0;
+}
+
+static irqreturn_t airoha_irq_handler(int irq, void *dev_instance)
+{
+	struct airoha_eth *eth = dev_instance;
+	u32 intr[ARRAY_SIZE(eth->irqmask)];
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(eth->irqmask); i++) {
+		intr[i] = airoha_qdma_rr(eth, REG_INT_STATUS(i));
+		intr[i] &= eth->irqmask[i];
+		airoha_qdma_wr(eth, REG_INT_STATUS(i), intr[i]);
+	}
+
+	if (!test_bit(DEV_STATE_INITIALIZED, &eth->state))
+		return IRQ_NONE;
+
+	if (intr[1] & RX_DONE_INT_MASK) {
+		airoha_qdma_irq_disable(eth, QDMA_INT_REG_IDX1,
+					RX_DONE_INT_MASK);
+		airoha_qdma_for_each_q_rx(eth, i) {
+			if (intr[1] & BIT(i))
+				napi_schedule(&eth->q_rx[i].napi);
+		}
+	}
+
+	if (intr[0] & INT_TX_MASK) {
+		for (i = 0; i < ARRAY_SIZE(eth->q_tx_irq); i++) {
+			struct airoha_tx_irq_queue *irq_q = &eth->q_tx_irq[i];
+			u32 status, head;
+
+			if (!(intr[0] & TX_DONE_INT_MASK(i)))
+				continue;
+
+			airoha_qdma_irq_disable(eth, QDMA_INT_REG_IDX0,
+						TX_DONE_INT_MASK(i));
+
+			status = airoha_qdma_rr(eth, REG_IRQ_STATUS(i));
+			head = FIELD_GET(IRQ_HEAD_IDX_MASK, status);
+			irq_q->head = head % irq_q->size;
+			irq_q->queued = FIELD_GET(IRQ_ENTRY_LEN_MASK, status);
+
+			napi_schedule(&eth->q_tx_irq[i].napi);
+		}
+	}
+
+	return IRQ_HANDLED;
+}
+
+static int airoha_qdma_init(struct airoha_eth *eth)
+{
+	struct device *dev = eth->net_dev->dev.parent;
+	int err;
+
+	err = devm_request_irq(dev, eth->irq, airoha_irq_handler,
+			       IRQF_SHARED, KBUILD_MODNAME, eth);
+	if (err)
+		return err;
+
+	err = airoha_qdma_init_rx(eth);
+	if (err)
+		return err;
+
+	err = airoha_qdma_init_tx(eth);
+	if (err)
+		return err;
+
+	err = airoha_qdma_init_hfwd_queues(eth);
+	if (err)
+		return err;
+
+	err = airoha_qdma_hw_init(eth);
+	if (err)
+		return err;
+
+	set_bit(DEV_STATE_INITIALIZED, &eth->state);
+
+	return 0;
+}
+
+static int airoha_hw_init(struct airoha_eth *eth)
+{
+	int err;
+
+	/* disable xsi */
+	reset_control_bulk_assert(ARRAY_SIZE(eth->xsi_rsts), eth->xsi_rsts);
+
+	reset_control_bulk_assert(ARRAY_SIZE(eth->rsts), eth->rsts);
+	msleep(20);
+	reset_control_bulk_deassert(ARRAY_SIZE(eth->rsts), eth->rsts);
+	msleep(20);
+
+	err = airoha_fe_init(eth);
+	if (err)
+		return err;
+
+	return airoha_qdma_init(eth);
+}
+
+static void airoha_hw_cleanup(struct airoha_eth *eth)
+{
+	int i;
+
+	airoha_qdma_for_each_q_rx(eth, i) {
+		struct airoha_queue *q = &eth->q_rx[i];
+
+		netif_napi_del(&q->napi);
+		airoha_qdma_clenaup_rx_queue(q);
+		if (q->page_pool)
+			page_pool_destroy(q->page_pool);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(eth->q_tx_irq); i++)
+		netif_napi_del(&eth->q_tx_irq[i].napi);
+
+	airoha_qdma_for_each_q_tx(eth, i)
+		airoha_qdma_clenaup_tx_queue(&eth->q_tx[i]);
+}
+
+static void airoha_qdma_start_napi(struct airoha_eth *eth)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(eth->q_tx_irq); i++)
+		napi_enable(&eth->q_tx_irq[i].napi);
+
+	airoha_qdma_for_each_q_rx(eth, i)
+		napi_enable(&eth->q_rx[i].napi);
+}
+
+static int airoha_dev_open(struct net_device *dev)
+{
+	struct airoha_eth *eth = netdev_priv(dev);
+	int err;
+
+	if (netdev_uses_dsa(dev))
+		airoha_fe_set(eth, REG_GDM1_INGRESS_CFG, GDM1_STAG_EN_MASK);
+	else
+		airoha_fe_clear(eth, REG_GDM1_INGRESS_CFG, GDM1_STAG_EN_MASK);
+
+	netif_tx_start_all_queues(dev);
+	err = airoha_set_gdma_ports(eth, true);
+	if (err)
+		return err;
+
+	airoha_qdma_set(eth, REG_QDMA_GLOBAL_CFG, GLOBAL_CFG_TX_DMA_EN_MASK);
+	airoha_qdma_set(eth, REG_QDMA_GLOBAL_CFG, GLOBAL_CFG_RX_DMA_EN_MASK);
+
+	return 0;
+}
+
+static int airoha_dev_stop(struct net_device *dev)
+{
+	struct airoha_eth *eth = netdev_priv(dev);
+	int err;
+
+	netif_tx_disable(dev);
+	err = airoha_set_gdma_ports(eth, false);
+	if (err)
+		return err;
+
+	airoha_qdma_clear(eth, REG_QDMA_GLOBAL_CFG, GLOBAL_CFG_TX_DMA_EN_MASK);
+	airoha_qdma_clear(eth, REG_QDMA_GLOBAL_CFG, GLOBAL_CFG_RX_DMA_EN_MASK);
+
+	return 0;
+}
+
+static int airoha_dev_set_macaddr(struct net_device *dev, void *p)
+{
+	struct airoha_eth *eth = netdev_priv(dev);
+	int err;
+
+	err = eth_mac_addr(dev, p);
+	if (err)
+		return err;
+
+	airoha_set_macaddr(eth, dev->dev_addr);
+
+	return 0;
+}
+
+static int airoha_dev_init(struct net_device *dev)
+{
+	struct airoha_eth *eth = netdev_priv(dev);
+
+	airoha_set_macaddr(eth, dev->dev_addr);
+
+	return 0;
+}
+
+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_eth *eth = netdev_priv(dev);
+	int i, qid = skb_get_queue_mapping(skb);
+	u32 nr_frags, msg0 = 0, msg1;
+	u32 len = skb_headlen(skb);
+	struct netdev_queue *txq;
+	struct airoha_queue *q;
+	void *data = skb->data;
+	u16 index;
+
+	if (skb->ip_summed == CHECKSUM_PARTIAL)
+		msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TCO_MASK, 1) |
+			FIELD_PREP(QDMA_ETH_TXMSG_UCO_MASK, 1) |
+			FIELD_PREP(QDMA_ETH_TXMSG_ICO_MASK, 1);
+
+	/* TSO: fill MSS info in tcp checksum field */
+	if (skb_is_gso(skb)) {
+		if (skb_cow_head(skb, 0))
+			goto error;
+
+		if (sinfo->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6)) {
+			tcp_hdr(skb)->check = cpu_to_be16(sinfo->gso_size);
+			msg0 |= FIELD_PREP(QDMA_ETH_TXMSG_TSO_MASK, 1);
+		}
+	}
+
+	msg1 = FIELD_PREP(QDMA_ETH_TXMSG_FPORT_MASK, DPORT_GDM1) |
+	       FIELD_PREP(QDMA_ETH_TXMSG_METER_MASK, 0x7f);
+
+	if (WARN_ON_ONCE(qid >= ARRAY_SIZE(eth->q_tx)))
+		qid = 0;
+
+	q = &eth->q_tx[qid];
+	if (WARN_ON_ONCE(!q->ndesc))
+		goto error;
+
+	spin_lock_bh(&q->lock);
+
+	nr_frags = 1 + sinfo->nr_frags;
+	if (q->queued + nr_frags > q->ndesc) {
+		/* not enough space in the queue */
+		spin_unlock_bh(&q->lock);
+		return NETDEV_TX_BUSY;
+	}
+
+	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];
+		skb_frag_t *frag = &sinfo->frags[i];
+		dma_addr_t addr;
+		u32 val;
+
+		addr = dma_map_single(dev->dev.parent, data, len,
+				      DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(dev->dev.parent, addr)))
+			goto error_unmap;
+
+		index = (index + 1) % q->ndesc;
+
+		val = FIELD_PREP(QDMA_DESC_LEN_MASK, len);
+		if (i < nr_frags - 1)
+			val |= FIELD_PREP(QDMA_DESC_MORE_MASK, 1);
+		WRITE_ONCE(desc->ctrl, cpu_to_le32(val));
+		WRITE_ONCE(desc->addr, cpu_to_le32(addr));
+		val = FIELD_PREP(QDMA_DESC_NEXT_ID_MASK, index);
+		WRITE_ONCE(desc->data, cpu_to_le32(val));
+		WRITE_ONCE(desc->msg0, cpu_to_le32(msg0));
+		WRITE_ONCE(desc->msg1, cpu_to_le32(msg1));
+		WRITE_ONCE(desc->msg2, cpu_to_le32(0xffff));
+
+		e->skb = i ? NULL : skb;
+		e->dma_addr = addr;
+		e->dma_len = len;
+
+		wmb();
+		airoha_qdma_rmw(eth, REG_TX_CPU_IDX(qid), TX_RING_CPU_IDX_MASK,
+				FIELD_PREP(TX_RING_CPU_IDX_MASK, index));
+
+		data = skb_frag_address(frag);
+		len = skb_frag_size(frag);
+	}
+
+	q->head = index;
+	q->queued += i;
+
+	txq = netdev_get_tx_queue(dev, qid);
+	netdev_tx_sent_queue(txq, skb->len);
+	skb_tx_timestamp(skb);
+
+	if (q->ndesc - q->queued < q->free_thr)
+		netif_tx_stop_queue(txq);
+
+	spin_unlock_bh(&q->lock);
+
+	return NETDEV_TX_OK;
+
+error_unmap:
+	for (i--; i >= 0; i++)
+		dma_unmap_single(dev->dev.parent, q->entry[i].dma_addr,
+				 q->entry[i].dma_len, DMA_TO_DEVICE);
+
+	spin_unlock_bh(&q->lock);
+error:
+	dev_kfree_skb_any(skb);
+	dev->stats.tx_dropped++;
+
+	return NETDEV_TX_OK;
+}
+
+static int airoha_dev_change_mtu(struct net_device *dev, int new_mtu)
+{
+	dev->mtu = new_mtu;
+
+	return 0;
+}
+
+static const struct airoha_ethtool_stats airoha_hw_stats[] = {
+	{
+		"tx_eth_pkt_cnt",
+		REG_FE_GDM1_TX_ETH_PKT_CNT_L,
+		REG_FE_GDM1_TX_ETH_PKT_CNT_H,
+	}, {
+		"tx_eth_byte_cnt",
+		REG_FE_GDM1_TX_ETH_BYTE_CNT_L,
+		REG_FE_GDM1_TX_ETH_BYTE_CNT_H,
+	}, {
+		"tx_ok_pkt_cnt",
+		REG_FE_GDM1_TX_OK_PKT_CNT_L,
+		REG_FE_GDM1_TX_OK_PKT_CNT_H,
+	}, {
+		"tx_ok_byte_cnt",
+		REG_FE_GDM1_TX_OK_BYTE_CNT_L,
+		REG_FE_GDM1_TX_OK_BYTE_CNT_H,
+	}, {
+		"tx_eth_drop_cnt",
+		REG_FE_GDM1_TX_ETH_DROP_CNT,
+	}, {
+		"tx_eth_bc_cnt",
+		REG_FE_GDM1_TX_ETH_BC_CNT,
+	}, {
+		"tx_eth_mc_cnt",
+		REG_FE_GDM1_TX_ETH_MC_CNT,
+	}, {
+		"tx_eth_lt64_cnt",
+		REG_FE_GDM1_TX_ETH_RUNT_CNT,
+	}, {
+		"tx_eth_eq64_cnt",
+		REG_FE_GDM1_TX_ETH_E64_CNT_L,
+		REG_FE_GDM1_TX_ETH_E64_CNT_H,
+	}, {
+		"tx_eth_65_127_cnt",
+		REG_FE_GDM1_TX_ETH_L64_CNT_L,
+		REG_FE_GDM1_TX_ETH_L64_CNT_H,
+	}, {
+		"tx_eth_128_255_cnt",
+		REG_FE_GDM1_TX_ETH_L127_CNT_L,
+		REG_FE_GDM1_TX_ETH_L127_CNT_H,
+	}, {
+		"tx_eth_256_511_cnt",
+		REG_FE_GDM1_TX_ETH_L255_CNT_L,
+		REG_FE_GDM1_TX_ETH_L255_CNT_H,
+	}, {
+		"tx_eth_512_1023_cnt",
+		REG_FE_GDM1_TX_ETH_L511_CNT_L,
+		REG_FE_GDM1_TX_ETH_L511_CNT_H,
+	}, {
+		"tx_eth_1024_1518_cnt",
+		REG_FE_GDM1_TX_ETH_L1023_CNT_L,
+		REG_FE_GDM1_TX_ETH_L1023_CNT_H,
+	}, {
+		"tx_eth_gt1518_cnt",
+		REG_FE_GDM1_TX_ETH_LONG_CNT,
+	}, {
+		"rx_eth_pkt_cnt",
+		REG_FE_GDM1_RX_ETH_PKT_CNT_L,
+		REG_FE_GDM1_RX_ETH_PKT_CNT_H,
+	}, {
+		"rx_eth_byte_cnt",
+		REG_FE_GDM1_RX_ETH_BYTE_CNT_L,
+		REG_FE_GDM1_RX_ETH_BYTE_CNT_H,
+	}, {
+		"rx_ok_pkt_cnt",
+		REG_FE_GDM1_RX_OK_PKT_CNT_L,
+		REG_FE_GDM1_RX_OK_PKT_CNT_H,
+	}, {
+		"rx_ok_byte_cnt",
+		REG_FE_GDM1_RX_OK_BYTE_CNT_L,
+		REG_FE_GDM1_RX_OK_BYTE_CNT_H,
+	}, {
+		"rx_eth_drop_cnt",
+		REG_FE_GDM1_RX_ETH_DROP_CNT,
+	}, {
+		"rx_eth_bc_cnt",
+		REG_FE_GDM1_RX_ETH_BC_CNT,
+	}, {
+		"rx_eth_mc_cnt",
+		REG_FE_GDM1_RX_ETH_MC_CNT,
+	}, {
+		"rx_eth_crc_drop_cnt",
+		REG_FE_GDM1_RX_ETH_CRC_ERR_CNT,
+	}, {
+		"rx_eth_frag_cnt",
+		REG_FE_GDM1_RX_ETH_FRAG_CNT,
+	}, {
+		"rx_eth_jabber_cnt",
+		REG_FE_GDM1_RX_ETH_JABBER_CNT,
+	}, {
+		"rx_eth_lt64_cnt",
+		REG_FE_GDM1_RX_ETH_RUNT_CNT,
+	}, {
+		"rx_eth_eq64_cnt",
+		REG_FE_GDM1_RX_ETH_E64_CNT_L,
+		REG_FE_GDM1_RX_ETH_E64_CNT_H,
+	}, {
+		"rx_eth_65_127_cnt",
+		REG_FE_GDM1_RX_ETH_L64_CNT_L,
+		REG_FE_GDM1_RX_ETH_L64_CNT_H,
+	}, {
+		"rx_eth_128_255_cnt",
+		REG_FE_GDM1_RX_ETH_L127_CNT_L,
+		REG_FE_GDM1_RX_ETH_L127_CNT_H,
+	}, {
+		"rx_eth_256_511_cnt",
+		REG_FE_GDM1_RX_ETH_L255_CNT_L,
+		REG_FE_GDM1_RX_ETH_L255_CNT_H,
+	}, {
+		"rx_eth_512_1023_cnt",
+		REG_FE_GDM1_RX_ETH_L511_CNT_L,
+		REG_FE_GDM1_RX_ETH_L511_CNT_H,
+	}, {
+		"rx_eth_1024_1518_cnt",
+		REG_FE_GDM1_RX_ETH_L1023_CNT_L,
+		REG_FE_GDM1_RX_ETH_L1023_CNT_H,
+	}, {
+		"rx_eth_gt1518_cnt",
+		REG_FE_GDM1_RX_ETH_LONG_CNT,
+	},
+};
+
+static void airoha_ethtool_get_strings(struct net_device *dev, u32 sset,
+				       u8 *data)
+{
+	int i;
+
+	if (sset != ETH_SS_STATS)
+		return;
+
+	for (i = 0; i < ARRAY_SIZE(airoha_hw_stats); i++) {
+		memcpy(data + i * ETH_GSTRING_LEN,
+		       airoha_hw_stats[i].name, ETH_GSTRING_LEN);
+	}
+
+	data += ETH_GSTRING_LEN * ARRAY_SIZE(airoha_hw_stats);
+	page_pool_ethtool_stats_get_strings(data);
+}
+
+static int airoha_ethtool_get_sset_count(struct net_device *dev, int sset)
+{
+	if (sset != ETH_SS_STATS)
+		return -EOPNOTSUPP;
+
+	return ARRAY_SIZE(airoha_hw_stats) +
+	       page_pool_ethtool_stats_get_count();
+}
+
+static void airoha_ethtool_get_stats(struct net_device *dev,
+				     struct ethtool_stats *stats, u64 *data)
+{
+	struct airoha_eth *eth = netdev_priv(dev);
+	struct page_pool_stats pp_stats = {};
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(airoha_hw_stats); i++) {
+		u32 val;
+
+		if (airoha_hw_stats[i].h_offset) {
+			val = airoha_fe_rr(eth, airoha_hw_stats[i].h_offset);
+			eth->hw_stats[i] += ((u64)val << 32);
+		}
+		val = airoha_fe_rr(eth, airoha_hw_stats[i].l_offset);
+		eth->hw_stats[i] += val;
+
+		*data++ = eth->hw_stats[i];
+	}
+
+	airoha_qdma_for_each_q_rx(eth, i)
+		page_pool_get_stats(eth->q_rx[i].page_pool, &pp_stats);
+	page_pool_ethtool_stats_get(data, &pp_stats);
+
+	/* reset mib counters */
+	airoha_fe_set(eth, REG_FE_GDM1_MIB_CLEAR,
+		      FE_GDM1_MIB_RX_CLEAR_MASK | FE_GDM1_MIB_TX_CLEAR_MASK);
+}
+
+static const struct net_device_ops airoha_netdev_ops = {
+	.ndo_init		= airoha_dev_init,
+	.ndo_open		= airoha_dev_open,
+	.ndo_stop		= airoha_dev_stop,
+	.ndo_start_xmit		= airoha_dev_xmit,
+	.ndo_change_mtu		= airoha_dev_change_mtu,
+	.ndo_set_mac_address	= airoha_dev_set_macaddr,
+};
+
+static const struct ethtool_ops airoha_ethtool_ops = {
+	.get_strings		= airoha_ethtool_get_strings,
+	.get_sset_count		= airoha_ethtool_get_sset_count,
+	.get_ethtool_stats	= airoha_ethtool_get_stats,
+};
+
+static int airoha_rx_queues_show(struct seq_file *s, void *data)
+{
+	struct airoha_eth *eth = s->private;
+	int i;
+
+	seq_puts(s, "     queue | hw-queued |      head |      tail |\n");
+	airoha_qdma_for_each_q_rx(eth, i) {
+		struct airoha_queue *q = &eth->q_rx[i];
+
+		seq_printf(s, " %9d | %9d | %9d | %9d |\n",
+			   i, q->queued, q->head, q->tail);
+	}
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(airoha_rx_queues);
+
+static int airoha_xmit_queues_show(struct seq_file *s, void *data)
+{
+	struct airoha_eth *eth = s->private;
+	int i;
+
+	seq_puts(s, "     queue | hw-queued |      head |      tail |\n");
+	airoha_qdma_for_each_q_tx(eth, i) {
+		struct airoha_queue *q = &eth->q_tx[i];
+
+		seq_printf(s, " %9d | %9d | %9d | %9d |\n",
+			   i, q->queued, q->head, q->tail);
+	}
+
+	return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(airoha_xmit_queues);
+
+static void airoha_register_debugfs(struct airoha_eth *eth)
+{
+	eth->debugfs_dir = debugfs_create_dir(KBUILD_MODNAME, NULL);
+	debugfs_create_file("rx-queues", 0400, eth->debugfs_dir, eth,
+			    &airoha_rx_queues_fops);
+	debugfs_create_file("xmit-queues", 0400, eth->debugfs_dir, eth,
+			    &airoha_xmit_queues_fops);
+}
+
+static int airoha_probe(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct net_device *dev;
+	struct airoha_eth *eth;
+	int err;
+
+	dev = devm_alloc_etherdev_mqs(&pdev->dev, sizeof(*eth),
+				      AIROHA_NUM_TX_RING, AIROHA_NUM_RX_RING);
+	if (!dev) {
+		dev_err(&pdev->dev, "alloc_etherdev failed\n");
+		return -ENOMEM;
+	}
+
+	eth = netdev_priv(dev);
+	eth->net_dev = dev;
+
+	err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(32));
+	if (err) {
+		dev_err(&pdev->dev, "failed configuring DMA mask\n");
+		return err;
+	}
+
+	eth->fe_regs = devm_platform_ioremap_resource_byname(pdev, "fe");
+	if (IS_ERR(eth->fe_regs))
+		return dev_err_probe(&pdev->dev, PTR_ERR(eth->fe_regs),
+				     "failed to iomap fe regs\n");
+
+	eth->qdma_regs = devm_platform_ioremap_resource_byname(pdev, "qdma0");
+	if (IS_ERR(eth->qdma_regs))
+		return dev_err_probe(&pdev->dev, PTR_ERR(eth->qdma_regs),
+				     "failed to iomap qdma regs\n");
+
+	eth->rsts[0].id = "fe";
+	eth->rsts[1].id = "pdma";
+	eth->rsts[2].id = "qdma";
+	err = devm_reset_control_bulk_get_exclusive(&pdev->dev,
+						    ARRAY_SIZE(eth->rsts),
+						    eth->rsts);
+	if (err) {
+		dev_err(&pdev->dev, "failed to get bulk reset lines\n");
+		return err;
+	}
+
+	eth->xsi_rsts[0].id = "xsi-mac";
+	eth->xsi_rsts[1].id = "hsi0-mac";
+	eth->xsi_rsts[2].id = "hsi1-mac";
+	eth->xsi_rsts[3].id = "hsi-mac";
+	err = devm_reset_control_bulk_get_exclusive(&pdev->dev,
+						    ARRAY_SIZE(eth->xsi_rsts),
+						    eth->xsi_rsts);
+	if (err) {
+		dev_err(&pdev->dev, "failed to get bulk xsi reset lines\n");
+		return err;
+	}
+
+	spin_lock_init(&eth->irq_lock);
+	eth->irq = platform_get_irq(pdev, 0);
+	if (eth->irq < 0) {
+		dev_err(&pdev->dev, "failed reading irq line\n");
+		return eth->irq;
+	}
+
+	dev->netdev_ops = &airoha_netdev_ops;
+	dev->ethtool_ops = &airoha_ethtool_ops;
+	dev->max_mtu = AIROHA_MAX_MTU;
+	dev->watchdog_timeo = 5 * HZ;
+	dev->hw_features = NETIF_F_IP_CSUM | NETIF_F_RXCSUM |
+			   NETIF_F_TSO6 | NETIF_F_IPV6_CSUM |
+			   NETIF_F_SG | NETIF_F_TSO;
+	dev->features |= dev->hw_features;
+	dev->dev.of_node = np;
+	dev->irq = eth->irq;
+	SET_NETDEV_DEV(dev, &pdev->dev);
+
+	err = of_get_ethdev_address(np, dev);
+	if (err) {
+		if (err == -EPROBE_DEFER)
+			return err;
+
+		eth_hw_addr_random(dev);
+		dev_info(&pdev->dev, "generated random MAC address %pM\n",
+			 dev->dev_addr);
+	}
+
+	eth->hw_stats = devm_kzalloc(&pdev->dev,
+				     ARRAY_SIZE(airoha_hw_stats) * sizeof(u64),
+				     GFP_KERNEL);
+	if (!eth->hw_stats)
+		return -ENOMEM;
+
+	err = airoha_hw_init(eth);
+	if (err)
+		goto error;
+
+	platform_set_drvdata(pdev, eth);
+	airoha_qdma_start_napi(eth);
+
+	err = register_netdev(dev);
+	if (err) {
+		platform_set_drvdata(pdev, NULL);
+		goto error;
+	}
+
+	airoha_register_debugfs(eth);
+
+	return 0;
+
+error:
+	airoha_hw_cleanup(eth);
+
+	return err;
+}
+
+static void airoha_remove(struct platform_device *pdev)
+{
+	struct airoha_eth *eth = platform_get_drvdata(pdev);
+
+	debugfs_remove(eth->debugfs_dir);
+
+	airoha_dev_stop(eth->net_dev);
+	airoha_hw_cleanup(eth);
+
+	unregister_netdev(eth->net_dev);
+	platform_set_drvdata(pdev, NULL);
+}
+
+const struct of_device_id of_airoha_match[] = {
+	{ .compatible = "airoha,en7581-eth" },
+	{ /* sentinel */ }
+};
+
+static struct platform_driver airoha_driver = {
+	.probe = airoha_probe,
+	.remove_new = airoha_remove,
+	.driver = {
+		.name = KBUILD_MODNAME,
+		.of_match_table = of_airoha_match,
+	},
+};
+module_platform_driver(airoha_driver);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Lorenzo Bianconi <lorenzo@kernel.org>");
+MODULE_DESCRIPTION("Ethernet driver for Airoha SoC");
diff --git a/drivers/net/ethernet/mediatek/airoha_eth.h b/drivers/net/ethernet/mediatek/airoha_eth.h
new file mode 100644
index 000000000000..fcd684e1418a
--- /dev/null
+++ b/drivers/net/ethernet/mediatek/airoha_eth.h
@@ -0,0 +1,793 @@ 
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2024 Lorenzo Bianconi <lorenzo@kernel.org>
+ */
+
+#define AIROHA_MAX_NUM_RSTS		3
+#define AIROHA_MAX_NUM_XSI_RSTS		4
+#define AIROHA_MAX_MTU			2000
+#define AIROHA_MAX_PACKET_SIZE		2048
+#define AIROHA_NUM_TX_RING		32
+#define AIROHA_NUM_RX_RING		32
+#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)
+
+/* 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			0x2400
+
+#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_LAN_MAC_H		0x0040
+#define REG_FE_LAN_MAC_LMIN		0x0044
+#define REG_FE_LAN_MAC_LMAX		0x0048
+
+#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 PSE_DEF_RSV_PAGE		0x80
+
+#define PSE_PORT0_QUEUE			6
+#define PSE_PORT1_QUEUE			6
+#define PSE_PORT2_QUEUE			32
+#define PSE_PORT3_QUEUE			6
+#define PSE_PORT4_QUEUE			4
+#define PSE_PORT5_QUEUE			6
+#define PSE_PORT6_QUEUE			8
+#define PSE_PORT7_QUEUE			10
+#define PSE_PORT8_QUEUE			4
+#define PSE_PORT9_QUEUE			2
+#define PSE_PORT10_QUEUE		2
+#define PSE_PORT11_QUEUE		0
+#define PSE_PORT12_QUEUE		0
+#define PSE_PORT13_QUEUE		0
+#define PSE_PORT14_QUEUE		0
+#define PSE_PORT15_QUEUE		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_GDM1_FWD_CFG		GDM1_BASE
+#define GDM1_DROP_CRC_ERR		BIT(23)
+#define GDM1_IP4_CKSUM			BIT(22)
+#define GDM1_TCP_CKSUM			BIT(21)
+#define GDM1_UDP_CKSUM			BIT(20)
+#define GDM1_UCFQ_MASK			GENMASK(15, 12)
+#define GDM1_BCFQ_MASK			GENMASK(11, 8)
+#define GDM1_MCFQ_MASK			GENMASK(7, 4)
+#define GDM1_OCFQ_MASK			GENMASK(3, 0)
+
+#define REG_GDM1_INGRESS_CFG		(GDM1_BASE + 0x10)
+#define GDM1_INGRESS_FC_EN_MASK		BIT(1)
+#define GDM1_STAG_EN_MASK		BIT(0)
+
+#define REG_GDM1_LEN_CFG		(GDM1_BASE + 0x14)
+#define GDM1_SHORT_LEN_MASK		GENMASK(13, 0)
+#define GDM1_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_GDM1_MIB_CLEAR		(GDM1_BASE + 0xf0)
+#define FE_GDM1_MIB_RX_CLEAR_MASK	BIT(1)
+#define FE_GDM1_MIB_TX_CLEAR_MASK	BIT(1)
+
+#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_GDM1_TX_OK_PKT_CNT_L	(GDM1_BASE + 0x104)
+#define REG_FE_GDM1_TX_OK_BYTE_CNT_L	(GDM1_BASE + 0x10c)
+#define REG_FE_GDM1_TX_ETH_PKT_CNT_L	(GDM1_BASE + 0x110)
+#define REG_FE_GDM1_TX_ETH_BYTE_CNT_L	(GDM1_BASE + 0x114)
+#define REG_FE_GDM1_TX_ETH_DROP_CNT	(GDM1_BASE + 0x118)
+#define REG_FE_GDM1_TX_ETH_BC_CNT	(GDM1_BASE + 0x11c)
+#define REG_FE_GDM1_TX_ETH_MC_CNT	(GDM1_BASE + 0x120)
+#define REG_FE_GDM1_TX_ETH_RUNT_CNT	(GDM1_BASE + 0x124)
+#define REG_FE_GDM1_TX_ETH_LONG_CNT	(GDM1_BASE + 0x128)
+#define REG_FE_GDM1_TX_ETH_E64_CNT_L	(GDM1_BASE + 0x12c)
+#define REG_FE_GDM1_TX_ETH_L64_CNT_L	(GDM1_BASE + 0x130)
+#define REG_FE_GDM1_TX_ETH_L127_CNT_L	(GDM1_BASE + 0x134)
+#define REG_FE_GDM1_TX_ETH_L255_CNT_L	(GDM1_BASE + 0x138)
+#define REG_FE_GDM1_TX_ETH_L511_CNT_L	(GDM1_BASE + 0x13c)
+#define REG_FE_GDM1_TX_ETH_L1023_CNT_L	(GDM1_BASE + 0x140)
+
+#define REG_FE_GDM1_RX_OK_PKT_CNT_L	(GDM1_BASE + 0x148)
+#define REG_FE_GDM1_RX_OK_BYTE_CNT_L	(GDM1_BASE + 0x15c)
+#define REG_FE_GDM1_RX_ETH_PKT_CNT_L	(GDM1_BASE + 0x160)
+#define REG_FE_GDM1_RX_ETH_BYTE_CNT_L	(GDM1_BASE + 0x164)
+#define REG_FE_GDM1_RX_ETH_DROP_CNT	(GDM1_BASE + 0x168)
+#define REG_FE_GDM1_RX_ETH_BC_CNT	(GDM1_BASE + 0x16c)
+#define REG_FE_GDM1_RX_ETH_MC_CNT	(GDM1_BASE + 0x170)
+#define REG_FE_GDM1_RX_ETH_CRC_ERR_CNT	(GDM1_BASE + 0x174)
+#define REG_FE_GDM1_RX_ETH_FRAG_CNT	(GDM1_BASE + 0x178)
+#define REG_FE_GDM1_RX_ETH_JABBER_CNT	(GDM1_BASE + 0x17c)
+#define REG_FE_GDM1_RX_ETH_RUNT_CNT	(GDM1_BASE + 0x180)
+#define REG_FE_GDM1_RX_ETH_LONG_CNT	(GDM1_BASE + 0x184)
+#define REG_FE_GDM1_RX_ETH_E64_CNT_L	(GDM1_BASE + 0x188)
+#define REG_FE_GDM1_RX_ETH_L64_CNT_L	(GDM1_BASE + 0x18c)
+#define REG_FE_GDM1_RX_ETH_L127_CNT_L	(GDM1_BASE + 0x190)
+#define REG_FE_GDM1_RX_ETH_L255_CNT_L	(GDM1_BASE + 0x194)
+#define REG_FE_GDM1_RX_ETH_L511_CNT_L	(GDM1_BASE + 0x198)
+#define REG_FE_GDM1_RX_ETH_L1023_CNT_L	(GDM1_BASE + 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_GDM1_TX_OK_PKT_CNT_H	(GDM1_BASE + 0x280)
+#define REG_FE_GDM1_TX_OK_BYTE_CNT_H	(GDM1_BASE + 0x284)
+#define REG_FE_GDM1_TX_ETH_PKT_CNT_H	(GDM1_BASE + 0x288)
+#define REG_FE_GDM1_TX_ETH_BYTE_CNT_H	(GDM1_BASE + 0x28c)
+
+#define REG_FE_GDM1_RX_OK_PKT_CNT_H	(GDM1_BASE + 0x290)
+#define REG_FE_GDM1_RX_OK_BYTE_CNT_H	(GDM1_BASE + 0x294)
+#define REG_FE_GDM1_RX_ETH_PKT_CNT_H	(GDM1_BASE + 0x298)
+#define REG_FE_GDM1_RX_ETH_BYTE_CNT_H	(GDM1_BASE + 0x29c)
+#define REG_FE_GDM1_TX_ETH_E64_CNT_H	(GDM1_BASE + 0x2B8)
+#define REG_FE_GDM1_TX_ETH_L64_CNT_H	(GDM1_BASE + 0x2Bc)
+#define REG_FE_GDM1_TX_ETH_L127_CNT_H	(GDM1_BASE + 0x2C0)
+#define REG_FE_GDM1_TX_ETH_L255_CNT_H	(GDM1_BASE + 0x2C4)
+#define REG_FE_GDM1_TX_ETH_L511_CNT_H	(GDM1_BASE + 0x2C8)
+#define REG_FE_GDM1_TX_ETH_L1023_CNT_H	(GDM1_BASE + 0x2CC)
+#define REG_FE_GDM1_RX_ETH_E64_CNT_H	(GDM1_BASE + 0x2E8)
+#define REG_FE_GDM1_RX_ETH_L64_CNT_H	(GDM1_BASE + 0x2EC)
+#define REG_FE_GDM1_RX_ETH_L127_CNT_H	(GDM1_BASE + 0x2F0)
+#define REG_FE_GDM1_RX_ETH_L255_CNT_H	(GDM1_BASE + 0x2F4)
+#define REG_FE_GDM1_RX_ETH_L511_CNT_H	(GDM1_BASE + 0x2F8)
+#define REG_FE_GDM1_RX_ETH_L1023_CNT_H	(GDM1_BASE + 0x2FC)
+
+#define REG_GDM2_CHN_RLS		(GDM2_BASE + 0x20)
+#define MBI_RX_AGE_SEL_MASK		GENMASK(18, 17)
+#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 + 0x100)
+#define GDM4_PAD_EN_MASK		BIT(28)
+#define GDM4_SPORT_OFFSET0_MASK		GENMASK(11, 8)
+
+#define REG_GDM4_SRC_PORT_SET		(GDM4_BASE + 0x33c)
+#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 |	\
+	 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_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_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(29)
+#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 REG_TXWRR_MODE_CFG		0x1020
+#define TWRR_WEIGHT_SCALE_MASK		BIT(31)
+#define TWRR_WEIGHT_BASE_MASK		BIT(3)
+
+#define REG_PSE_BUF_USAGE_CFG		0x1028
+#define PSE_BUF_ESTIMATE_EN_MASK	BIT(29)
+
+#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 pkt - 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)
+/* 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)
+/* MSG1 */
+#define QDMA_ETH_TXMSG_NO_DROP		BIT(31)
+#define QDMA_ETH_TXMSG_METER_MASK	GENMASK(30, 24)	/* 0x7f means do not apply 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 means do not count */
+#define QDMA_ETH_TXMSG_ACNT_G0_MASK	GENMASK(5, 0)	/* 0x3f means do not count */
+
+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;
+};
+
+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 airoha_dport {
+	DPORT_PDMA,
+	DPORT_GDM1,
+	DPORT_GDM2,
+	DPORT_GDM3,
+	DPORT_PPE,
+	DPORT_QDMA,
+	DPORT_QDMA_HW,
+	DPORT_DISCARD,
+	DPORT_GDM4 = 9,
+};
+
+enum {
+	FE_DP_CPU,
+	FE_DP_GDM1,
+	FE_DP_GDM2,
+	FE_DP_QDMA1_HWF,
+	FE_DP_GDMA3_HWF = 3,
+	FE_DP_PPE,
+	FE_DP_QDMA2_CPU,
+	FE_DP_QDMA2_HWF,
+	FE_DP_DISCARD,
+	FE_DP_PPE2 = 8,
+	FE_DP_DROP = 15,
+};
+
+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 {
+	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,
+};
+
+struct airoha_queue_entry {
+	union {
+		void *buf;
+		struct sk_buff *skb;
+	};
+	dma_addr_t dma_addr;
+	u16 dma_len;
+};
+
+struct airoha_ethtool_stats {
+	const char name[ETH_GSTRING_LEN];
+	u32 l_offset;
+	u32 h_offset;
+};
+
+struct airoha_queue {
+	struct airoha_eth *eth;
+
+	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_eth *eth;
+
+	struct napi_struct napi;
+	u32 *q;
+
+	int size;
+	int queued;
+	u16 head;
+};
+
+struct airoha_eth {
+	struct net_device *net_dev;
+
+	unsigned long state;
+
+	void __iomem *qdma_regs;
+	void __iomem *fe_regs;
+
+	spinlock_t irq_lock;
+	u32 irqmask[QDMA_INT_REG_MAX];
+	int irq;
+
+	struct reset_control_bulk_data rsts[AIROHA_MAX_NUM_RSTS];
+	struct reset_control_bulk_data xsi_rsts[AIROHA_MAX_NUM_XSI_RSTS];
+
+	struct airoha_queue q_tx[AIROHA_NUM_TX_RING];
+	struct airoha_queue q_rx[AIROHA_NUM_RX_RING];
+
+	struct airoha_tx_irq_queue q_tx_irq[AIROHA_NUM_TX_IRQ];
+
+	/* descriptor and packet buffers for qdma hw forward */
+	struct {
+		void *desc;
+		void *q;
+	} hfwd;
+
+	u64 *hw_stats;
+
+	struct dentry *debugfs_dir;
+};
+
+#define airoha_qdma_for_each_q_rx(eth, i)		\
+	for (i = 0; i < ARRAY_SIZE((eth)->q_rx); i++)	\
+		if ((eth)->q_rx[i].ndesc)
+#define airoha_qdma_for_each_q_tx(eth, i)		\
+	for (i = 0; i < ARRAY_SIZE((eth)->q_tx); i++)	\
+		if ((eth)->q_tx[i].ndesc)