Merge tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-nextgrafted

Pull networking updates from Jakub Kicinski:
 "Core:

   - Add dedicated kmem_cache for typical/small skb->head, avoid having
     to access struct page at kfree time, and improve memory use.

   - Introduce sysctl to set default RPS configuration for new netdevs.

   - Define Netlink protocol specification format which can be used to
     describe messages used by each family and auto-generate parsers.
     Add tools for generating kernel data structures and uAPI headers.

   - Expose all net/core sysctls inside netns.

   - Remove 4s sleep in netpoll if carrier is instantly detected on
     boot.

   - Add configurable limit of MDB entries per port, and port-vlan.

   - Continue populating drop reasons throughout the stack.

   - Retire a handful of legacy Qdiscs and classifiers.

  Protocols:

   - Support IPv4 big TCP (TSO frames larger than 64kB).

   - Add IP_LOCAL_PORT_RANGE socket option, to control local port range
     on socket by socket basis.

   - Track and report in procfs number of MPTCP sockets used.

   - Support mixing IPv4 and IPv6 flows in the in-kernel MPTCP path
     manager.

   - IPv6: don't check net.ipv6.route.max_size and rely on garbage
     collection to free memory (similarly to IPv4).

   - Support Penultimate Segment Pop (PSP) flavor in SRv6 (RFC8986).

   - ICMP: add per-rate limit counters.

   - Add support for user scanning requests in ieee802154.

   - Remove static WEP support.

   - Support minimal Wi-Fi 7 Extremely High Throughput (EHT) rate
     reporting.

   - WiFi 7 EHT channel puncturing support (client & AP).

  BPF:

   - Add a rbtree data structure following the "next-gen data structure"
     precedent set by recently added linked list, that is, by using
     kfunc + kptr instead of adding a new BPF map type.

   - Expose XDP hints via kfuncs with initial support for RX hash and
     timestamp metadata.

   - Add BPF_F_NO_TUNNEL_KEY extension to bpf_skb_set_tunnel_key to
     better support decap on GRE tunnel devices not operating in collect
     metadata.

   - Improve x86 JIT's codegen for PROBE_MEM runtime error checks.

   - Remove the need for trace_printk_lock for bpf_trace_printk and
     bpf_trace_vprintk helpers.

   - Extend libbpf's bpf_tracing.h support for tracing arguments of
     kprobes/uprobes and syscall as a special case.

   - Significantly reduce the search time for module symbols by
     livepatch and BPF.

   - Enable cpumasks to be used as kptrs, which is useful for tracing
     programs tracking which tasks end up running on which CPUs in
     different time intervals.

   - Add support for BPF trampoline on s390x and riscv64.

   - Add capability to export the XDP features supported by the NIC.

   - Add __bpf_kfunc tag for marking kernel functions as kfuncs.

   - Add cgroup.memory=nobpf kernel parameter option to disable BPF
     memory accounting for container environments.

  Netfilter:

   - Remove the CLUSTERIP target. It has been marked as obsolete for
     years, and we still have WARN splats wrt races of the out-of-band
     /proc interface installed by this target.

   - Add 'destroy' commands to nf_tables. They are identical to the
     existing 'delete' commands, but do not return an error if the
     referenced object (set, chain, rule...) did not exist.

  Driver API:

   - Improve cpumask_local_spread() locality to help NICs set the right
     IRQ affinity on AMD platforms.

   - Separate C22 and C45 MDIO bus transactions more clearly.

   - Introduce new DCB table to control DSCP rewrite on egress.

   - Support configuration of Physical Layer Collision Avoidance (PLCA)
     Reconciliation Sublayer (RS) (802.3cg-2019). Modern version of
     shared medium Ethernet.

   - Support for MAC Merge layer (IEEE 802.3-2018 clause 99). Allowing
     preemption of low priority frames by high priority frames.

   - Add support for controlling MACSec offload using netlink SET.

   - Rework devlink instance refcounts to allow registration and
     de-registration under the instance lock. Split the code into
     multiple files, drop some of the unnecessarily granular locks and
     factor out common parts of netlink operation handling.

   - Add TX frame aggregation parameters (for USB drivers).

   - Add a new attr TCA_EXT_WARN_MSG to report TC (offload) warning
     messages with notifications for debug.

   - Allow offloading of UDP NEW connections via act_ct.

   - Add support for per action HW stats in TC.

   - Support hardware miss to TC action (continue processing in SW from
     a specific point in the action chain).

   - Warn if old Wireless Extension user space interface is used with
     modern cfg80211/mac80211 drivers. Do not support Wireless
     Extensions for Wi-Fi 7 devices at all. Everyone should switch to
     using nl80211 interface instead.

   - Improve the CAN bit timing configuration. Use extack to return
     error messages directly to user space, update the SJW handling,
     including the definition of a new default value that will benefit
     CAN-FD controllers, by increasing their oscillator tolerance.

  New hardware / drivers:

   - Ethernet:
      - nVidia BlueField-3 support (control traffic driver)
      - Ethernet support for imx93 SoCs
      - Motorcomm yt8531 gigabit Ethernet PHY
      - onsemi NCN26000 10BASE-T1S PHY (with support for PLCA)
      - Microchip LAN8841 PHY (incl. cable diagnostics and PTP)
      - Amlogic gxl MDIO mux

   - WiFi:
      - RealTek RTL8188EU (rtl8xxxu)
      - Qualcomm Wi-Fi 7 devices (ath12k)

   - CAN:
      - Renesas R-Car V4H

  Drivers:

   - Bluetooth:
      - Set Per Platform Antenna Gain (PPAG) for Intel controllers.

   - Ethernet NICs:
      - Intel (1G, igc):
         - support TSN / Qbv / packet scheduling features of i226 model
      - Intel (100G, ice):
         - use GNSS subsystem instead of TTY
         - multi-buffer XDP support
         - extend support for GPIO pins to E823 devices
      - nVidia/Mellanox:
         - update the shared buffer configuration on PFC commands
         - implement PTP adjphase function for HW offset control
         - TC support for Geneve and GRE with VF tunnel offload
         - more efficient crypto key management method
         - multi-port eswitch support
      - Netronome/Corigine:
         - add DCB IEEE support
         - support IPsec offloading for NFP3800
      - Freescale/NXP (enetc):
         - support XDP_REDIRECT for XDP non-linear buffers
         - improve reconfig, avoid link flap and waiting for idle
         - support MAC Merge layer
      - Other NICs:
         - sfc/ef100: add basic devlink support for ef100
         - ionic: rx_push mode operation (writing descriptors via MMIO)
         - bnxt: use the auxiliary bus abstraction for RDMA
         - r8169: disable ASPM and reset bus in case of tx timeout
         - cpsw: support QSGMII mode for J721e CPSW9G
         - cpts: support pulse-per-second output
         - ngbe: add an mdio bus driver
         - usbnet: optimize usbnet_bh() by avoiding unnecessary queuing
         - r8152: handle devices with FW with NCM support
         - amd-xgbe: support 10Mbps, 2.5GbE speeds and rx-adaptation
         - virtio-net: support multi buffer XDP
         - virtio/vsock: replace virtio_vsock_pkt with sk_buff
         - tsnep: XDP support

   - Ethernet high-speed switches:
      - nVidia/Mellanox (mlxsw):
         - add support for latency TLV (in FW control messages)
      - Microchip (sparx5):
         - separate explicit and implicit traffic forwarding rules, make
           the implicit rules always active
         - add support for egress DSCP rewrite
         - IS0 VCAP support (Ingress Classification)
         - IS2 VCAP filters (protos, L3 addrs, L4 ports, flags, ToS
           etc.)
         - ES2 VCAP support (Egress Access Control)
         - support for Per-Stream Filtering and Policing (802.1Q,
           8.6.5.1)

   - Ethernet embedded switches:
      - Marvell (mv88e6xxx):
         - add MAB (port auth) offload support
         - enable PTP receive for mv88e6390
      - NXP (ocelot):
         - support MAC Merge layer
         - support for the the vsc7512 internal copper phys
      - Microchip:
         - lan9303: convert to PHYLINK
         - lan966x: support TC flower filter statistics
         - lan937x: PTP support for KSZ9563/KSZ8563 and LAN937x
         - lan937x: support Credit Based Shaper configuration
         - ksz9477: support Energy Efficient Ethernet
      - other:
         - qca8k: convert to regmap read/write API, use bulk operations
         - rswitch: Improve TX timestamp accuracy

   - Intel WiFi (iwlwifi):
      - EHT (Wi-Fi 7) rate reporting
      - STEP equalizer support: transfer some STEP (connection to radio
        on platforms with integrated wifi) related parameters from the
        BIOS to the firmware.

   - Qualcomm 802.11ax WiFi (ath11k):
      - IPQ5018 support
      - Fine Timing Measurement (FTM) responder role support
      - channel 177 support

   - MediaTek WiFi (mt76):
      - per-PHY LED support
      - mt7996: EHT (Wi-Fi 7) support
      - Wireless Ethernet Dispatch (WED) reset support
      - switch to using page pool allocator

   - RealTek WiFi (rtw89):
      - support new version of Bluetooth co-existance

   - Mobile:
      - rmnet: support TX aggregation"

* tag 'net-next-6.3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (1872 commits)
  page_pool: add a comment explaining the fragment counter usage
  net: ethtool: fix __ethtool_dev_mm_supported() implementation
  ethtool: pse-pd: Fix double word in comments
  xsk: add linux/vmalloc.h to xsk.c
  sefltests: netdevsim: wait for devlink instance after netns removal
  selftest: fib_tests: Always cleanup before exit
  net/mlx5e: Align IPsec ASO result memory to be as required by hardware
  net/mlx5e: TC, Set CT miss to the specific ct action instance
  net/mlx5e: Rename CHAIN_TO_REG to MAPPED_OBJ_TO_REG
  net/mlx5: Refactor tc miss handling to a single function
  net/mlx5: Kconfig: Make tc offload depend on tc skb extension
  net/sched: flower: Support hardware miss to tc action
  net/sched: flower: Move filter handle initialization earlier
  net/sched: cls_api: Support hardware miss to tc action
  net/sched: Rename user cookie and act cookie
  sfc: fix builds without CONFIG_RTC_LIB
  sfc: clean up some inconsistent indentings
  net/mlx4_en: Introduce flexible array to silence overflow warning
  net: lan966x: Fix possible deadlock inside PTP
  net/ulp: Remove redundant ->clone() test in inet_clone_ulp().
  ...

1 files changed, 982 insertions, 0 deletions

diff --git a/drivers/net/wireless/mediatek/mt76/dma.c b/drivers/net/wireless/mediatek/mt76/dma.c
new file mode 100644
index 000000000..da281cd1d
--- /dev/null
+++ b/drivers/net/wireless/mediatek/mt76/dma.c
@@ -0,0 +1,982 @@
+// SPDX-License-Identifier: ISC
+/*
+ * Copyright (C) 2016 Felix Fietkau <nbd@nbd.name>
+ */
+
+#include <linux/dma-mapping.h>
+#include "mt76.h"
+#include "dma.h"
+
+#if IS_ENABLED(CONFIG_NET_MEDIATEK_SOC_WED)
+
+#define Q_READ(_dev, _q, _field) ({					\
+	u32 _offset = offsetof(struct mt76_queue_regs, _field);		\
+	u32 _val;							\
+	if ((_q)->flags & MT_QFLAG_WED)					\
+		_val = mtk_wed_device_reg_read(&(_dev)->mmio.wed,	\
+					       ((_q)->wed_regs +	\
+					        _offset));		\
+	else								\
+		_val = readl(&(_q)->regs->_field);			\
+	_val;								\
+})
+
+#define Q_WRITE(_dev, _q, _field, _val)	do {				\
+	u32 _offset = offsetof(struct mt76_queue_regs, _field);		\
+	if ((_q)->flags & MT_QFLAG_WED)					\
+		mtk_wed_device_reg_write(&(_dev)->mmio.wed,		\
+					 ((_q)->wed_regs + _offset),	\
+					 _val);				\
+	else								\
+		writel(_val, &(_q)->regs->_field);			\
+} while (0)
+
+#else
+
+#define Q_READ(_dev, _q, _field)	readl(&(_q)->regs->_field)
+#define Q_WRITE(_dev, _q, _field, _val)	writel(_val, &(_q)->regs->_field)
+
+#endif
+
+static struct mt76_txwi_cache *
+mt76_alloc_txwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t;
+	dma_addr_t addr;
+	u8 *txwi;
+	int size;
+
+	size = L1_CACHE_ALIGN(dev->drv->txwi_size + sizeof(*t));
+	txwi = kzalloc(size, GFP_ATOMIC);
+	if (!txwi)
+		return NULL;
+
+	addr = dma_map_single(dev->dma_dev, txwi, dev->drv->txwi_size,
+			      DMA_TO_DEVICE);
+	t = (struct mt76_txwi_cache *)(txwi + dev->drv->txwi_size);
+	t->dma_addr = addr;
+
+	return t;
+}
+
+static struct mt76_txwi_cache *
+mt76_alloc_rxwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t;
+
+	t = kzalloc(L1_CACHE_ALIGN(sizeof(*t)), GFP_ATOMIC);
+	if (!t)
+		return NULL;
+
+	t->ptr = NULL;
+	return t;
+}
+
+static struct mt76_txwi_cache *
+__mt76_get_txwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t = NULL;
+
+	spin_lock(&dev->lock);
+	if (!list_empty(&dev->txwi_cache)) {
+		t = list_first_entry(&dev->txwi_cache, struct mt76_txwi_cache,
+				     list);
+		list_del(&t->list);
+	}
+	spin_unlock(&dev->lock);
+
+	return t;
+}
+
+static struct mt76_txwi_cache *
+__mt76_get_rxwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t = NULL;
+
+	spin_lock(&dev->wed_lock);
+	if (!list_empty(&dev->rxwi_cache)) {
+		t = list_first_entry(&dev->rxwi_cache, struct mt76_txwi_cache,
+				     list);
+		list_del(&t->list);
+	}
+	spin_unlock(&dev->wed_lock);
+
+	return t;
+}
+
+static struct mt76_txwi_cache *
+mt76_get_txwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t = __mt76_get_txwi(dev);
+
+	if (t)
+		return t;
+
+	return mt76_alloc_txwi(dev);
+}
+
+struct mt76_txwi_cache *
+mt76_get_rxwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t = __mt76_get_rxwi(dev);
+
+	if (t)
+		return t;
+
+	return mt76_alloc_rxwi(dev);
+}
+EXPORT_SYMBOL_GPL(mt76_get_rxwi);
+
+void
+mt76_put_txwi(struct mt76_dev *dev, struct mt76_txwi_cache *t)
+{
+	if (!t)
+		return;
+
+	spin_lock(&dev->lock);
+	list_add(&t->list, &dev->txwi_cache);
+	spin_unlock(&dev->lock);
+}
+EXPORT_SYMBOL_GPL(mt76_put_txwi);
+
+void
+mt76_put_rxwi(struct mt76_dev *dev, struct mt76_txwi_cache *t)
+{
+	if (!t)
+		return;
+
+	spin_lock(&dev->wed_lock);
+	list_add(&t->list, &dev->rxwi_cache);
+	spin_unlock(&dev->wed_lock);
+}
+EXPORT_SYMBOL_GPL(mt76_put_rxwi);
+
+static void
+mt76_free_pending_txwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t;
+
+	local_bh_disable();
+	while ((t = __mt76_get_txwi(dev)) != NULL) {
+		dma_unmap_single(dev->dma_dev, t->dma_addr, dev->drv->txwi_size,
+				 DMA_TO_DEVICE);
+		kfree(mt76_get_txwi_ptr(dev, t));
+	}
+	local_bh_enable();
+}
+
+void
+mt76_free_pending_rxwi(struct mt76_dev *dev)
+{
+	struct mt76_txwi_cache *t;
+
+	local_bh_disable();
+	while ((t = __mt76_get_rxwi(dev)) != NULL) {
+		if (t->ptr)
+			mt76_put_page_pool_buf(t->ptr, false);
+		kfree(t);
+	}
+	local_bh_enable();
+}
+EXPORT_SYMBOL_GPL(mt76_free_pending_rxwi);
+
+static void
+mt76_dma_sync_idx(struct mt76_dev *dev, struct mt76_queue *q)
+{
+	Q_WRITE(dev, q, desc_base, q->desc_dma);
+	Q_WRITE(dev, q, ring_size, q->ndesc);
+	q->head = Q_READ(dev, q, dma_idx);
+	q->tail = q->head;
+}
+
+static void
+mt76_dma_queue_reset(struct mt76_dev *dev, struct mt76_queue *q)
+{
+	int i;
+
+	if (!q || !q->ndesc)
+		return;
+
+	/* clear descriptors */
+	for (i = 0; i < q->ndesc; i++)
+		q->desc[i].ctrl = cpu_to_le32(MT_DMA_CTL_DMA_DONE);
+
+	Q_WRITE(dev, q, cpu_idx, 0);
+	Q_WRITE(dev, q, dma_idx, 0);
+	mt76_dma_sync_idx(dev, q);
+}
+
+static int
+mt76_dma_add_rx_buf(struct mt76_dev *dev, struct mt76_queue *q,
+		    struct mt76_queue_buf *buf, void *data)
+{
+	struct mt76_desc *desc = &q->desc[q->head];
+	struct mt76_queue_entry *entry = &q->entry[q->head];
+	struct mt76_txwi_cache *txwi = NULL;
+	u32 buf1 = 0, ctrl;
+	int idx = q->head;
+	int rx_token;
+
+	ctrl = FIELD_PREP(MT_DMA_CTL_SD_LEN0, buf[0].len);
+
+	if (mt76_queue_is_wed_rx(q)) {
+		txwi = mt76_get_rxwi(dev);
+		if (!txwi)
+			return -ENOMEM;
+
+		rx_token = mt76_rx_token_consume(dev, data, txwi, buf->addr);
+		if (rx_token < 0) {
+			mt76_put_rxwi(dev, txwi);
+			return -ENOMEM;
+		}
+
+		buf1 |= FIELD_PREP(MT_DMA_CTL_TOKEN, rx_token);
+		ctrl |= MT_DMA_CTL_TO_HOST;
+	}
+
+	WRITE_ONCE(desc->buf0, cpu_to_le32(buf->addr));
+	WRITE_ONCE(desc->buf1, cpu_to_le32(buf1));
+	WRITE_ONCE(desc->ctrl, cpu_to_le32(ctrl));
+	WRITE_ONCE(desc->info, 0);
+
+	entry->dma_addr[0] = buf->addr;
+	entry->dma_len[0] = buf->len;
+	entry->txwi = txwi;
+	entry->buf = data;
+	entry->wcid = 0xffff;
+	entry->skip_buf1 = true;
+	q->head = (q->head + 1) % q->ndesc;
+	q->queued++;
+
+	return idx;
+}
+
+static int
+mt76_dma_add_buf(struct mt76_dev *dev, struct mt76_queue *q,
+		 struct mt76_queue_buf *buf, int nbufs, u32 info,
+		 struct sk_buff *skb, void *txwi)
+{
+	struct mt76_queue_entry *entry;
+	struct mt76_desc *desc;
+	int i, idx = -1;
+	u32 ctrl, next;
+
+	if (txwi) {
+		q->entry[q->head].txwi = DMA_DUMMY_DATA;
+		q->entry[q->head].skip_buf0 = true;
+	}
+
+	for (i = 0; i < nbufs; i += 2, buf += 2) {
+		u32 buf0 = buf[0].addr, buf1 = 0;
+
+		idx = q->head;
+		next = (q->head + 1) % q->ndesc;
+
+		desc = &q->desc[idx];
+		entry = &q->entry[idx];
+
+		if (buf[0].skip_unmap)
+			entry->skip_buf0 = true;
+		entry->skip_buf1 = i == nbufs - 1;
+
+		entry->dma_addr[0] = buf[0].addr;
+		entry->dma_len[0] = buf[0].len;
+
+		ctrl = FIELD_PREP(MT_DMA_CTL_SD_LEN0, buf[0].len);
+		if (i < nbufs - 1) {
+			entry->dma_addr[1] = buf[1].addr;
+			entry->dma_len[1] = buf[1].len;
+			buf1 = buf[1].addr;
+			ctrl |= FIELD_PREP(MT_DMA_CTL_SD_LEN1, buf[1].len);
+			if (buf[1].skip_unmap)
+				entry->skip_buf1 = true;
+		}
+
+		if (i == nbufs - 1)
+			ctrl |= MT_DMA_CTL_LAST_SEC0;
+		else if (i == nbufs - 2)
+			ctrl |= MT_DMA_CTL_LAST_SEC1;
+
+		WRITE_ONCE(desc->buf0, cpu_to_le32(buf0));
+		WRITE_ONCE(desc->buf1, cpu_to_le32(buf1));
+		WRITE_ONCE(desc->info, cpu_to_le32(info));
+		WRITE_ONCE(desc->ctrl, cpu_to_le32(ctrl));
+
+		q->head = next;
+		q->queued++;
+	}
+
+	q->entry[idx].txwi = txwi;
+	q->entry[idx].skb = skb;
+	q->entry[idx].wcid = 0xffff;
+
+	return idx;
+}
+
+static void
+mt76_dma_tx_cleanup_idx(struct mt76_dev *dev, struct mt76_queue *q, int idx,
+			struct mt76_queue_entry *prev_e)
+{
+	struct mt76_queue_entry *e = &q->entry[idx];
+
+	if (!e->skip_buf0)
+		dma_unmap_single(dev->dma_dev, e->dma_addr[0], e->dma_len[0],
+				 DMA_TO_DEVICE);
+
+	if (!e->skip_buf1)
+		dma_unmap_single(dev->dma_dev, e->dma_addr[1], e->dma_len[1],
+				 DMA_TO_DEVICE);
+
+	if (e->txwi == DMA_DUMMY_DATA)
+		e->txwi = NULL;
+
+	if (e->skb == DMA_DUMMY_DATA)
+		e->skb = NULL;
+
+	*prev_e = *e;
+	memset(e, 0, sizeof(*e));
+}
+
+static void
+mt76_dma_kick_queue(struct mt76_dev *dev, struct mt76_queue *q)
+{
+	wmb();
+	Q_WRITE(dev, q, cpu_idx, q->head);
+}
+
+static void
+mt76_dma_tx_cleanup(struct mt76_dev *dev, struct mt76_queue *q, bool flush)
+{
+	struct mt76_queue_entry entry;
+	int last;
+
+	if (!q || !q->ndesc)
+		return;
+
+	spin_lock_bh(&q->cleanup_lock);
+	if (flush)
+		last = -1;
+	else
+		last = Q_READ(dev, q, dma_idx);
+
+	while (q->queued > 0 && q->tail != last) {
+		mt76_dma_tx_cleanup_idx(dev, q, q->tail, &entry);
+		mt76_queue_tx_complete(dev, q, &entry);
+
+		if (entry.txwi) {
+			if (!(dev->drv->drv_flags & MT_DRV_TXWI_NO_FREE))
+				mt76_put_txwi(dev, entry.txwi);
+		}
+
+		if (!flush && q->tail == last)
+			last = Q_READ(dev, q, dma_idx);
+	}
+	spin_unlock_bh(&q->cleanup_lock);
+
+	if (flush) {
+		spin_lock_bh(&q->lock);
+		mt76_dma_sync_idx(dev, q);
+		mt76_dma_kick_queue(dev, q);
+		spin_unlock_bh(&q->lock);
+	}
+
+	if (!q->queued)
+		wake_up(&dev->tx_wait);
+}
+
+static void *
+mt76_dma_get_buf(struct mt76_dev *dev, struct mt76_queue *q, int idx,
+		 int *len, u32 *info, bool *more, bool *drop)
+{
+	struct mt76_queue_entry *e = &q->entry[idx];
+	struct mt76_desc *desc = &q->desc[idx];
+	void *buf;
+
+	if (len) {
+		u32 ctrl = le32_to_cpu(READ_ONCE(desc->ctrl));
+		*len = FIELD_GET(MT_DMA_CTL_SD_LEN0, ctrl);
+		*more = !(ctrl & MT_DMA_CTL_LAST_SEC0);
+	}
+
+	if (info)
+		*info = le32_to_cpu(desc->info);
+
+	if (mt76_queue_is_wed_rx(q)) {
+		u32 token = FIELD_GET(MT_DMA_CTL_TOKEN,
+				      le32_to_cpu(desc->buf1));
+		struct mt76_txwi_cache *t = mt76_rx_token_release(dev, token);
+
+		if (!t)
+			return NULL;
+
+		dma_sync_single_for_cpu(dev->dma_dev, t->dma_addr,
+				SKB_WITH_OVERHEAD(q->buf_size),
+				page_pool_get_dma_dir(q->page_pool));
+
+		buf = t->ptr;
+		t->dma_addr = 0;
+		t->ptr = NULL;
+
+		mt76_put_rxwi(dev, t);
+
+		if (drop) {
+			u32 ctrl = le32_to_cpu(READ_ONCE(desc->ctrl));
+
+			*drop = !!(ctrl & (MT_DMA_CTL_TO_HOST_A |
+					   MT_DMA_CTL_DROP));
+		}
+	} else {
+		buf = e->buf;
+		e->buf = NULL;
+		dma_sync_single_for_cpu(dev->dma_dev, e->dma_addr[0],
+				SKB_WITH_OVERHEAD(q->buf_size),
+				page_pool_get_dma_dir(q->page_pool));
+	}
+
+	return buf;
+}
+
+static void *
+mt76_dma_dequeue(struct mt76_dev *dev, struct mt76_queue *q, bool flush,
+		 int *len, u32 *info, bool *more, bool *drop)
+{
+	int idx = q->tail;
+
+	*more = false;
+	if (!q->queued)
+		return NULL;
+
+	if (flush)
+		q->desc[idx].ctrl |= cpu_to_le32(MT_DMA_CTL_DMA_DONE);
+	else if (!(q->desc[idx].ctrl & cpu_to_le32(MT_DMA_CTL_DMA_DONE)))
+		return NULL;
+
+	q->tail = (q->tail + 1) % q->ndesc;
+	q->queued--;
+
+	return mt76_dma_get_buf(dev, q, idx, len, info, more, drop);
+}
+
+static int
+mt76_dma_tx_queue_skb_raw(struct mt76_dev *dev, struct mt76_queue *q,
+			  struct sk_buff *skb, u32 tx_info)
+{
+	struct mt76_queue_buf buf = {};
+	dma_addr_t addr;
+
+	if (q->queued + 1 >= q->ndesc - 1)
+		goto error;
+
+	addr = dma_map_single(dev->dma_dev, skb->data, skb->len,
+			      DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(dev->dma_dev, addr)))
+		goto error;
+
+	buf.addr = addr;
+	buf.len = skb->len;
+
+	spin_lock_bh(&q->lock);
+	mt76_dma_add_buf(dev, q, &buf, 1, tx_info, skb, NULL);
+	mt76_dma_kick_queue(dev, q);
+	spin_unlock_bh(&q->lock);
+
+	return 0;
+
+error:
+	dev_kfree_skb(skb);
+	return -ENOMEM;
+}
+
+static int
+mt76_dma_tx_queue_skb(struct mt76_dev *dev, struct mt76_queue *q,
+		      enum mt76_txq_id qid, struct sk_buff *skb,
+		      struct mt76_wcid *wcid, struct ieee80211_sta *sta)
+{
+	struct ieee80211_tx_status status = {
+		.sta = sta,
+	};
+	struct mt76_tx_info tx_info = {
+		.skb = skb,
+	};
+	struct ieee80211_hw *hw;
+	int len, n = 0, ret = -ENOMEM;
+	struct mt76_txwi_cache *t;
+	struct sk_buff *iter;
+	dma_addr_t addr;
+	u8 *txwi;
+
+	t = mt76_get_txwi(dev);
+	if (!t)
+		goto free_skb;
+
+	txwi = mt76_get_txwi_ptr(dev, t);
+
+	skb->prev = skb->next = NULL;
+	if (dev->drv->drv_flags & MT_DRV_TX_ALIGNED4_SKBS)
+		mt76_insert_hdr_pad(skb);
+
+	len = skb_headlen(skb);
+	addr = dma_map_single(dev->dma_dev, skb->data, len, DMA_TO_DEVICE);
+	if (unlikely(dma_mapping_error(dev->dma_dev, addr)))
+		goto free;
+
+	tx_info.buf[n].addr = t->dma_addr;
+	tx_info.buf[n++].len = dev->drv->txwi_size;
+	tx_info.buf[n].addr = addr;
+	tx_info.buf[n++].len = len;
+
+	skb_walk_frags(skb, iter) {
+		if (n == ARRAY_SIZE(tx_info.buf))
+			goto unmap;
+
+		addr = dma_map_single(dev->dma_dev, iter->data, iter->len,
+				      DMA_TO_DEVICE);
+		if (unlikely(dma_mapping_error(dev->dma_dev, addr)))
+			goto unmap;
+
+		tx_info.buf[n].addr = addr;
+		tx_info.buf[n++].len = iter->len;
+	}
+	tx_info.nbuf = n;
+
+	if (q->queued + (tx_info.nbuf + 1) / 2 >= q->ndesc - 1) {
+		ret = -ENOMEM;
+		goto unmap;
+	}
+
+	dma_sync_single_for_cpu(dev->dma_dev, t->dma_addr, dev->drv->txwi_size,
+				DMA_TO_DEVICE);
+	ret = dev->drv->tx_prepare_skb(dev, txwi, qid, wcid, sta, &tx_info);
+	dma_sync_single_for_device(dev->dma_dev, t->dma_addr, dev->drv->txwi_size,
+				   DMA_TO_DEVICE);
+	if (ret < 0)
+		goto unmap;
+
+	return mt76_dma_add_buf(dev, q, tx_info.buf, tx_info.nbuf,
+				tx_info.info, tx_info.skb, t);
+
+unmap:
+	for (n--; n > 0; n--)
+		dma_unmap_single(dev->dma_dev, tx_info.buf[n].addr,
+				 tx_info.buf[n].len, DMA_TO_DEVICE);
+
+free:
+#ifdef CONFIG_NL80211_TESTMODE
+	/* fix tx_done accounting on queue overflow */
+	if (mt76_is_testmode_skb(dev, skb, &hw)) {
+		struct mt76_phy *phy = hw->priv;
+
+		if (tx_info.skb == phy->test.tx_skb)
+			phy->test.tx_done--;
+	}
+#endif
+
+	mt76_put_txwi(dev, t);
+
+free_skb:
+	status.skb = tx_info.skb;
+	hw = mt76_tx_status_get_hw(dev, tx_info.skb);
+	ieee80211_tx_status_ext(hw, &status);
+
+	return ret;
+}
+
+static int
+mt76_dma_rx_fill(struct mt76_dev *dev, struct mt76_queue *q,
+		 bool allow_direct)
+{
+	int len = SKB_WITH_OVERHEAD(q->buf_size);
+	int frames = 0;
+
+	if (!q->ndesc)
+		return 0;
+
+	spin_lock_bh(&q->lock);
+
+	while (q->queued < q->ndesc - 1) {
+		enum dma_data_direction dir;
+		struct mt76_queue_buf qbuf;
+		dma_addr_t addr;
+		int offset;
+		void *buf;
+
+		buf = mt76_get_page_pool_buf(q, &offset, q->buf_size);
+		if (!buf)
+			break;
+
+		addr = page_pool_get_dma_addr(virt_to_head_page(buf)) + offset;
+		dir = page_pool_get_dma_dir(q->page_pool);
+		dma_sync_single_for_device(dev->dma_dev, addr, len, dir);
+
+		qbuf.addr = addr + q->buf_offset;
+		qbuf.len = len - q->buf_offset;
+		qbuf.skip_unmap = false;
+		if (mt76_dma_add_rx_buf(dev, q, &qbuf, buf) < 0) {
+			mt76_put_page_pool_buf(buf, allow_direct);
+			break;
+		}
+		frames++;
+	}
+
+	if (frames)
+		mt76_dma_kick_queue(dev, q);
+
+	spin_unlock_bh(&q->lock);
+
+	return frames;
+}
+
+int mt76_dma_wed_setup(struct mt76_dev *dev, struct mt76_queue *q, bool reset)
+{
+#ifdef CONFIG_NET_MEDIATEK_SOC_WED
+	struct mtk_wed_device *wed = &dev->mmio.wed;
+	int ret, type, ring;
+	u8 flags;
+
+	if (!q || !q->ndesc)
+		return -EINVAL;
+
+	flags = q->flags;
+	if (!mtk_wed_device_active(wed))
+		q->flags &= ~MT_QFLAG_WED;
+
+	if (!(q->flags & MT_QFLAG_WED))
+		return 0;
+
+	type = FIELD_GET(MT_QFLAG_WED_TYPE, q->flags);
+	ring = FIELD_GET(MT_QFLAG_WED_RING, q->flags);
+
+	switch (type) {
+	case MT76_WED_Q_TX:
+		ret = mtk_wed_device_tx_ring_setup(wed, ring, q->regs, reset);
+		if (!ret)
+			q->wed_regs = wed->tx_ring[ring].reg_base;
+		break;
+	case MT76_WED_Q_TXFREE:
+		/* WED txfree queue needs ring to be initialized before setup */
+		q->flags = 0;
+		mt76_dma_queue_reset(dev, q);
+		mt76_dma_rx_fill(dev, q, false);
+		q->flags = flags;
+
+		ret = mtk_wed_device_txfree_ring_setup(wed, q->regs);
+		if (!ret)
+			q->wed_regs = wed->txfree_ring.reg_base;
+		break;
+	case MT76_WED_Q_RX:
+		ret = mtk_wed_device_rx_ring_setup(wed, ring, q->regs, reset);
+		if (!ret)
+			q->wed_regs = wed->rx_ring[ring].reg_base;
+		break;
+	default:
+		ret = -EINVAL;
+	}
+
+	return ret;
+#else
+	return 0;
+#endif
+}
+EXPORT_SYMBOL_GPL(mt76_dma_wed_setup);
+
+static int
+mt76_dma_alloc_queue(struct mt76_dev *dev, struct mt76_queue *q,
+		     int idx, int n_desc, int bufsize,
+		     u32 ring_base)
+{
+	int ret, size;
+
+	spin_lock_init(&q->lock);
+	spin_lock_init(&q->cleanup_lock);
+
+	q->regs = dev->mmio.regs + ring_base + idx * MT_RING_SIZE;
+	q->ndesc = n_desc;
+	q->buf_size = bufsize;
+	q->hw_idx = idx;
+
+	size = q->ndesc * sizeof(struct mt76_desc);
+	q->desc = dmam_alloc_coherent(dev->dma_dev, size, &q->desc_dma, GFP_KERNEL);
+	if (!q->desc)
+		return -ENOMEM;
+
+	size = q->ndesc * sizeof(*q->entry);
+	q->entry = devm_kzalloc(dev->dev, size, GFP_KERNEL);
+	if (!q->entry)
+		return -ENOMEM;
+
+	ret = mt76_create_page_pool(dev, q);
+	if (ret)
+		return ret;
+
+	ret = mt76_dma_wed_setup(dev, q, false);
+	if (ret)
+		return ret;
+
+	if (q->flags != MT_WED_Q_TXFREE)
+		mt76_dma_queue_reset(dev, q);
+
+	return 0;
+}
+
+static void
+mt76_dma_rx_cleanup(struct mt76_dev *dev, struct mt76_queue *q)
+{
+	void *buf;
+	bool more;
+
+	if (!q->ndesc)
+		return;
+
+	spin_lock_bh(&q->lock);
+
+	do {
+		buf = mt76_dma_dequeue(dev, q, true, NULL, NULL, &more, NULL);
+		if (!buf)
+			break;
+
+		mt76_put_page_pool_buf(buf, false);
+	} while (1);
+
+	if (q->rx_head) {
+		dev_kfree_skb(q->rx_head);
+		q->rx_head = NULL;
+	}
+
+	spin_unlock_bh(&q->lock);
+}
+
+static void
+mt76_dma_rx_reset(struct mt76_dev *dev, enum mt76_rxq_id qid)
+{
+	struct mt76_queue *q = &dev->q_rx[qid];
+	int i;
+
+	if (!q->ndesc)
+		return;
+
+	for (i = 0; i < q->ndesc; i++)
+		q->desc[i].ctrl = cpu_to_le32(MT_DMA_CTL_DMA_DONE);
+
+	mt76_dma_rx_cleanup(dev, q);
+
+	/* reset WED rx queues */
+	mt76_dma_wed_setup(dev, q, true);
+	if (q->flags != MT_WED_Q_TXFREE) {
+		mt76_dma_sync_idx(dev, q);
+		mt76_dma_rx_fill(dev, q, false);
+	}
+}
+
+static void
+mt76_add_fragment(struct mt76_dev *dev, struct mt76_queue *q, void *data,
+		  int len, bool more, u32 info)
+{
+	struct sk_buff *skb = q->rx_head;
+	struct skb_shared_info *shinfo = skb_shinfo(skb);
+	int nr_frags = shinfo->nr_frags;
+
+	if (nr_frags < ARRAY_SIZE(shinfo->frags)) {
+		struct page *page = virt_to_head_page(data);
+		int offset = data - page_address(page) + q->buf_offset;
+
+		skb_add_rx_frag(skb, nr_frags, page, offset, len, q->buf_size);
+	} else {
+		mt76_put_page_pool_buf(data, true);
+	}
+
+	if (more)
+		return;
+
+	q->rx_head = NULL;
+	if (nr_frags < ARRAY_SIZE(shinfo->frags))
+		dev->drv->rx_skb(dev, q - dev->q_rx, skb, &info);
+	else
+		dev_kfree_skb(skb);
+}
+
+static int
+mt76_dma_rx_process(struct mt76_dev *dev, struct mt76_queue *q, int budget)
+{
+	int len, data_len, done = 0, dma_idx;
+	struct sk_buff *skb;
+	unsigned char *data;
+	bool check_ddone = false;
+	bool more;
+
+	if (IS_ENABLED(CONFIG_NET_MEDIATEK_SOC_WED) &&
+	    q->flags == MT_WED_Q_TXFREE) {
+		dma_idx = Q_READ(dev, q, dma_idx);
+		check_ddone = true;
+	}
+
+	while (done < budget) {
+		bool drop = false;
+		u32 info;
+
+		if (check_ddone) {
+			if (q->tail == dma_idx)
+				dma_idx = Q_READ(dev, q, dma_idx);
+
+			if (q->tail == dma_idx)
+				break;
+		}
+
+		data = mt76_dma_dequeue(dev, q, false, &len, &info, &more,
+					&drop);
+		if (!data)
+			break;
+
+		if (drop)
+			goto free_frag;
+
+		if (q->rx_head)
+			data_len = q->buf_size;
+		else
+			data_len = SKB_WITH_OVERHEAD(q->buf_size);
+
+		if (data_len < len + q->buf_offset) {
+			dev_kfree_skb(q->rx_head);
+			q->rx_head = NULL;
+			goto free_frag;
+		}
+
+		if (q->rx_head) {
+			mt76_add_fragment(dev, q, data, len, more, info);
+			continue;
+		}
+
+		if (!more && dev->drv->rx_check &&
+		    !(dev->drv->rx_check(dev, data, len)))
+			goto free_frag;
+
+		skb = build_skb(data, q->buf_size);
+		if (!skb)
+			goto free_frag;
+
+		skb_reserve(skb, q->buf_offset);
+		skb_mark_for_recycle(skb);
+
+		*(u32 *)skb->cb = info;
+
+		__skb_put(skb, len);
+		done++;
+
+		if (more) {
+			q->rx_head = skb;
+			continue;
+		}
+
+		dev->drv->rx_skb(dev, q - dev->q_rx, skb, &info);
+		continue;
+
+free_frag:
+		mt76_put_page_pool_buf(data, true);
+	}
+
+	mt76_dma_rx_fill(dev, q, true);
+	return done;
+}
+
+int mt76_dma_rx_poll(struct napi_struct *napi, int budget)
+{
+	struct mt76_dev *dev;
+	int qid, done = 0, cur;
+
+	dev = container_of(napi->dev, struct mt76_dev, napi_dev);
+	qid = napi - dev->napi;
+
+	rcu_read_lock();
+
+	do {
+		cur = mt76_dma_rx_process(dev, &dev->q_rx[qid], budget - done);
+		mt76_rx_poll_complete(dev, qid, napi);
+		done += cur;
+	} while (cur && done < budget);
+
+	rcu_read_unlock();
+
+	if (done < budget && napi_complete(napi))
+		dev->drv->rx_poll_complete(dev, qid);
+
+	return done;
+}
+EXPORT_SYMBOL_GPL(mt76_dma_rx_poll);
+
+static int
+mt76_dma_init(struct mt76_dev *dev,
+	      int (*poll)(struct napi_struct *napi, int budget))
+{
+	int i;
+
+	init_dummy_netdev(&dev->napi_dev);
+	init_dummy_netdev(&dev->tx_napi_dev);
+	snprintf(dev->napi_dev.name, sizeof(dev->napi_dev.name), "%s",
+		 wiphy_name(dev->hw->wiphy));
+	dev->napi_dev.threaded = 1;
+	init_completion(&dev->mmio.wed_reset);
+	init_completion(&dev->mmio.wed_reset_complete);
+
+	mt76_for_each_q_rx(dev, i) {
+		netif_napi_add(&dev->napi_dev, &dev->napi[i], poll);
+		mt76_dma_rx_fill(dev, &dev->q_rx[i], false);
+		napi_enable(&dev->napi[i]);
+	}
+
+	return 0;
+}
+
+static const struct mt76_queue_ops mt76_dma_ops = {
+	.init = mt76_dma_init,
+	.alloc = mt76_dma_alloc_queue,
+	.reset_q = mt76_dma_queue_reset,
+	.tx_queue_skb_raw = mt76_dma_tx_queue_skb_raw,
+	.tx_queue_skb = mt76_dma_tx_queue_skb,
+	.tx_cleanup = mt76_dma_tx_cleanup,
+	.rx_cleanup = mt76_dma_rx_cleanup,
+	.rx_reset = mt76_dma_rx_reset,
+	.kick = mt76_dma_kick_queue,
+};
+
+void mt76_dma_attach(struct mt76_dev *dev)
+{
+	dev->queue_ops = &mt76_dma_ops;
+}
+EXPORT_SYMBOL_GPL(mt76_dma_attach);
+
+void mt76_dma_cleanup(struct mt76_dev *dev)
+{
+	int i;
+
+	mt76_worker_disable(&dev->tx_worker);
+	netif_napi_del(&dev->tx_napi);
+
+	for (i = 0; i < ARRAY_SIZE(dev->phys); i++) {
+		struct mt76_phy *phy = dev->phys[i];
+		int j;
+
+		if (!phy)
+			continue;
+
+		for (j = 0; j < ARRAY_SIZE(phy->q_tx); j++)
+			mt76_dma_tx_cleanup(dev, phy->q_tx[j], true);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(dev->q_mcu); i++)
+		mt76_dma_tx_cleanup(dev, dev->q_mcu[i], true);
+
+	mt76_for_each_q_rx(dev, i) {
+		struct mt76_queue *q = &dev->q_rx[i];
+
+		netif_napi_del(&dev->napi[i]);
+		mt76_dma_rx_cleanup(dev, q);
+
+		page_pool_destroy(q->page_pool);
+	}
+
+	mt76_free_pending_txwi(dev);
+	mt76_free_pending_rxwi(dev);
+
+	if (mtk_wed_device_active(&dev->mmio.wed))
+		mtk_wed_device_detach(&dev->mmio.wed);
+}
+EXPORT_SYMBOL_GPL(mt76_dma_cleanup);