Merge branch 'ena-updates'

Shay Agroskin says:

====================
se build_skb and reorganize some code in ENA

this patchset introduces several changes:

- Use build_skb() on RX side.
  This allows to ensure that the headers are in the linear part

- Batch some code into functions and remove some of the code to make it more
  readable and less error prone

- Fix RST format and outdated description in ENA documentation

- Improve cache alignment in the code
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Documentation/networking/device_drivers/ethernet/amazon/ena.rst

@@ -11,12 +11,12 @@ ENA is a networking interface designed to make good use of modern CPU
 features and system architectures.
 
 The ENA device exposes a lightweight management interface with a
-minimal set of memory mapped registers and extendable command set
+minimal set of memory mapped registers and extendible command set
 through an Admin Queue.
 
 The driver supports a range of ENA devices, is link-speed independent
-(i.e., the same driver is used for 10GbE, 25GbE, 40GbE, etc.), and has
-a negotiated and extendable feature set.
+(i.e., the same driver is used for 10GbE, 25GbE, 40GbE, etc), and has
+a negotiated and extendible feature set.
 
 Some ENA devices support SR-IOV. This driver is used for both the
 SR-IOV Physical Function (PF) and Virtual Function (VF) devices.
@@ -27,9 +27,9 @@ is advertised by the device via the Admin Queue), a dedicated MSI-X
 interrupt vector per Tx/Rx queue pair, adaptive interrupt moderation,
 and CPU cacheline optimized data placement.
 
-The ENA driver supports industry standard TCP/IP offload features such
-as checksum offload and TCP transmit segmentation offload (TSO).
-Receive-side scaling (RSS) is supported for multi-core scaling.
+The ENA driver supports industry standard TCP/IP offload features such as
+checksum offload. Receive-side scaling (RSS) is supported for multi-core
+scaling.
 
 The ENA driver and its corresponding devices implement health
 monitoring mechanisms such as watchdog, enabling the device and driver
@@ -38,7 +38,6 @@ debug logs.
 
 Some of the ENA devices support a working mode called Low-latency
 Queue (LLQ), which saves several more microseconds.
-
 ENA Source Code Directory Structure
 ===================================
 
@@ -53,7 +52,6 @@ ena_eth_io_defs.h   Definition of ENA data path interface.
 ena_common_defs.h   Common definitions for ena_com layer.
 ena_regs_defs.h     Definition of ENA PCI memory-mapped (MMIO) registers.
 ena_netdev.[ch]     Main Linux kernel driver.
-ena_syfsfs.[ch]     Sysfs files.
 ena_ethtool.c       ethtool callbacks.
 ena_pci_id_tbl.h    Supported device IDs.
 =================   ======================================================
@@ -69,7 +67,7 @@ ENA management interface is exposed by means of:
 - Asynchronous Event Notification Queue (AENQ)
 
 ENA device MMIO Registers are accessed only during driver
-initialization and are not involved in further normal device
+initialization and are not used during further normal device
 operation.
 
 AQ is used for submitting management commands, and the
@@ -100,28 +98,27 @@ group may have multiple syndromes, as shown below
 
 The events are:
 
-	====================	===============
-	Group			Syndrome
-	====================	===============
-	Link state change	**X**
-	Fatal error		**X**
-	Notification		Suspend traffic
-	Notification		Resume traffic
-	Keep-Alive		**X**
-	====================	===============
+====================    ===============
+Group                   Syndrome
+====================    ===============
+Link state change       **X**
+Fatal error             **X**
+Notification            Suspend traffic
+Notification            Resume traffic
+Keep-Alive              **X**
+====================    ===============
 
 ACQ and AENQ share the same MSI-X vector.
 
-Keep-Alive is a special mechanism that allows monitoring of the
-device's health. The driver maintains a watchdog (WD) handler which,
-if fired, logs the current state and statistics then resets and
-restarts the ENA device and driver. A Keep-Alive event is delivered by
-the device every second. The driver re-arms the WD upon reception of a
-Keep-Alive event. A missed Keep-Alive event causes the WD handler to
-fire.
+Keep-Alive is a special mechanism that allows monitoring the device's health.
+A Keep-Alive event is delivered by the device every second.
+The driver maintains a watchdog (WD) handler which logs the current state and
+statistics. If the keep-alive events aren't delivered as expected the WD resets
+the device and the driver.
 
 Data Path Interface
 ===================
+
 I/O operations are based on Tx and Rx Submission Queues (Tx SQ and Rx
 SQ correspondingly). Each SQ has a completion queue (CQ) associated
 with it.
@@ -131,26 +128,24 @@ physical memory.
 
 The ENA driver supports two Queue Operation modes for Tx SQs:
 
-- Regular mode
-
-  * In this mode the Tx SQs reside in the host's memory. The ENA
+- **Regular mode:**
+  In this mode the Tx SQs reside in the host's memory. The ENA
   device fetches the ENA Tx descriptors and packet data from host
   memory.
 
-- Low Latency Queue (LLQ) mode or "push-mode".
-
-  * In this mode the driver pushes the transmit descriptors and the
+- **Low Latency Queue (LLQ) mode or "push-mode":**
+  In this mode the driver pushes the transmit descriptors and the
   first 128 bytes of the packet directly to the ENA device memory
   space. The rest of the packet payload is fetched by the
   device. For this operation mode, the driver uses a dedicated PCI
   device memory BAR, which is mapped with write-combine capability.
 
-The Rx SQs support only the regular mode.
-
-Note: Not all ENA devices support LLQ, and this feature is negotiated
+  **Note that** not all ENA devices support LLQ, and this feature is negotiated
   with the device upon initialization. If the ENA device does not
   support LLQ mode, the driver falls back to the regular mode.
 
+The Rx SQs support only the regular mode.
+
 The driver supports multi-queue for both Tx and Rx. This has various
 benefits:
 
@@ -165,6 +160,7 @@ benefits:
 
 Interrupt Modes
 ===============
+
 The driver assigns a single MSI-X vector per queue pair (for both Tx
 and Rx directions). The driver assigns an additional dedicated MSI-X vector
 for management (for ACQ and AENQ).
@@ -190,20 +186,21 @@ unmasked by the driver after NAPI processing is complete.
 
 Interrupt Moderation
 ====================
+
 ENA driver and device can operate in conventional or adaptive interrupt
 moderation mode.
 
-In conventional mode the driver instructs device to postpone interrupt
+**In conventional mode** the driver instructs device to postpone interrupt
 posting according to static interrupt delay value. The interrupt delay
-value can be configured through ethtool(8). The following ethtool
-parameters are supported by the driver: tx-usecs, rx-usecs
+value can be configured through `ethtool(8)`. The following `ethtool`
+parameters are supported by the driver: ``tx-usecs``, ``rx-usecs``
 
-In adaptive interrupt moderation mode the interrupt delay value is
+**In adaptive interrupt** moderation mode the interrupt delay value is
 updated by the driver dynamically and adjusted every NAPI cycle
 according to the traffic nature.
 
-Adaptive coalescing can be switched on/off through ethtool(8)
-adaptive_rx on|off parameter.
+Adaptive coalescing can be switched on/off through `ethtool(8)`'s
+:code:`adaptive_rx on|off` parameter.
 
 More information about Adaptive Interrupt Moderation (DIM) can be found in
 Documentation/networking/net_dim.rst
@@ -214,17 +211,10 @@ The rx_copybreak is initialized by default to ENA_DEFAULT_RX_COPYBREAK
 and can be configured by the ETHTOOL_STUNABLE command of the
 SIOCETHTOOL ioctl.
 
-SKB
-===
-The driver-allocated SKB for frames received from Rx handling using
-NAPI context. The allocation method depends on the size of the packet.
-If the frame length is larger than rx_copybreak, napi_get_frags()
-is used, otherwise netdev_alloc_skb_ip_align() is used, the buffer
-content is copied (by CPU) to the SKB, and the buffer is recycled.
-
 Statistics
 ==========
-The user can obtain ENA device and driver statistics using ethtool.
+
+The user can obtain ENA device and driver statistics using `ethtool`.
 The driver can collect regular or extended statistics (including
 per-queue stats) from the device.
 
@@ -232,22 +222,23 @@ In addition the driver logs the stats to syslog upon device reset.
 
 MTU
 ===
+
 The driver supports an arbitrarily large MTU with a maximum that is
 negotiated with the device. The driver configures MTU using the
 SetFeature command (ENA_ADMIN_MTU property). The user can change MTU
-via ip(8) and similar legacy tools.
+via `ip(8)` and similar legacy tools.
 
 Stateless Offloads
 ==================
+
 The ENA driver supports:
 
-- TSO over IPv4/IPv6
-- TSO with ECN
 - IPv4 header checksum offload
 - TCP/UDP over IPv4/IPv6 checksum offloads
 
 RSS
 ===
+
 - The ENA device supports RSS that allows flexible Rx traffic
   steering.
 - Toeplitz and CRC32 hash functions are supported.
@@ -260,41 +251,42 @@ RSS
   function delivered in the Rx CQ descriptor is set in the received
   SKB.
 - The user can provide a hash key, hash function, and configure the
-  indirection table through ethtool(8).
+  indirection table through `ethtool(8)`.
 
 DATA PATH
 =========
+
 Tx
 --
 
-ena_start_xmit() is called by the stack. This function does the following:
+:code:`ena_start_xmit()` is called by the stack. This function does the following:
 
-- Maps data buffers (skb->data and frags).
-- Populates ena_buf for the push buffer (if the driver and device are
-  in push mode.)
+- Maps data buffers (``skb->data`` and frags).
+- Populates ``ena_buf`` for the push buffer (if the driver and device are
+  in push mode).
 - Prepares ENA bufs for the remaining frags.
-- Allocates a new request ID from the empty req_id ring. The request
+- Allocates a new request ID from the empty ``req_id`` ring. The request
   ID is the index of the packet in the Tx info. This is used for
-  out-of-order TX completions.
+  out-of-order Tx completions.
 - Adds the packet to the proper place in the Tx ring.
-- Calls ena_com_prepare_tx(), an ENA communication layer that converts
-  the ena_bufs to ENA descriptors (and adds meta ENA descriptors as
-  needed.)
+- Calls :code:`ena_com_prepare_tx()`, an ENA communication layer that converts
+  the ``ena_bufs`` to ENA descriptors (and adds meta ENA descriptors as
+  needed).
 
   * This function also copies the ENA descriptors and the push buffer
-    to the Device memory space (if in push mode.)
+    to the Device memory space (if in push mode).
 
-- Writes doorbell to the ENA device.
+- Writes a doorbell to the ENA device.
 - When the ENA device finishes sending the packet, a completion
   interrupt is raised.
 - The interrupt handler schedules NAPI.
-- The ena_clean_tx_irq() function is called. This function handles the
+- The :code:`ena_clean_tx_irq()` function is called. This function handles the
   completion descriptors generated by the ENA, with a single
   completion descriptor per completed packet.
 
-  * req_id is retrieved from the completion descriptor. The tx_info of
-    the packet is retrieved via the req_id. The data buffers are
-    unmapped and req_id is returned to the empty req_id ring.
+  * ``req_id`` is retrieved from the completion descriptor. The ``tx_info`` of
+    the packet is retrieved via the ``req_id``. The data buffers are
+    unmapped and ``req_id`` is returned to the empty ``req_id`` ring.
   * The function stops when the completion descriptors are completed or
     the budget is reached.
 
@@ -303,12 +295,11 @@ Rx
 
 - When a packet is received from the ENA device.
 - The interrupt handler schedules NAPI.
-- The ena_clean_rx_irq() function is called. This function calls
-  ena_rx_pkt(), an ENA communication layer function, which returns the
-  number of descriptors used for a new unhandled packet, and zero if
+- The :code:`ena_clean_rx_irq()` function is called. This function calls
+  :code:`ena_com_rx_pkt()`, an ENA communication layer function, which returns the
+  number of descriptors used for a new packet, and zero if
   no new packet is found.
-- Then it calls the ena_clean_rx_irq() function.
-- ena_eth_rx_skb() checks packet length:
+- :code:`ena_rx_skb()` checks packet length:
 
   * If the packet is small (len < rx_copybreak), the driver allocates
     a SKB for the new packet, and copies the packet payload into the
@@ -317,9 +308,10 @@ Rx
     - In this way the original data buffer is not passed to the stack
       and is reused for future Rx packets.
 
-  * Otherwise the function unmaps the Rx buffer, then allocates the
-    new SKB structure and hooks the Rx buffer to the SKB frags.
+  * Otherwise the function unmaps the Rx buffer, sets the first
+    descriptor as `skb`'s linear part and the other descriptors as the
+    `skb`'s frags.
 
 - The new SKB is updated with the necessary information (protocol,
-  checksum hw verify result, etc.), and then passed to the network
-  stack, using the NAPI interface function napi_gro_receive().
+  checksum hw verify result, etc), and then passed to the network
+  stack, using the NAPI interface function :code:`napi_gro_receive()`.

drivers/net/ethernet/amazon/ena/ena_admin_defs.h

@@ -1042,8 +1042,6 @@ enum ena_admin_aenq_group {
 };
 
 enum ena_admin_aenq_notification_syndrome {
-	ENA_ADMIN_SUSPEND                           = 0,
-	ENA_ADMIN_RESUME                            = 1,
 	ENA_ADMIN_UPDATE_HINTS                      = 2,
 };
 

drivers/net/ethernet/amazon/ena/ena_com.c

@@ -1979,7 +1979,8 @@ int ena_com_get_dev_attr_feat(struct ena_com_dev *ena_dev,
 		if (rc)
 			return rc;
 
-		if (get_resp.u.max_queue_ext.version != ENA_FEATURE_MAX_QUEUE_EXT_VER)
+		if (get_resp.u.max_queue_ext.version !=
+		    ENA_FEATURE_MAX_QUEUE_EXT_VER)
 			return -EINVAL;
 
 		memcpy(&get_feat_ctx->max_queue_ext, &get_resp.u.max_queue_ext,

drivers/net/ethernet/amazon/ena/ena_eth_com.c

@@ -151,11 +151,14 @@ static int ena_com_close_bounce_buffer(struct ena_com_io_sq *io_sq)
 		return 0;
 
 	/* bounce buffer was used, so write it and get a new one */
-	if (pkt_ctrl->idx) {
+	if (likely(pkt_ctrl->idx)) {
 		rc = ena_com_write_bounce_buffer_to_dev(io_sq,
 							pkt_ctrl->curr_bounce_buf);
-		if (unlikely(rc))
+		if (unlikely(rc)) {
+			netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+				   "Failed to write bounce buffer to device\n");
 			return rc;
+		}
 
 		pkt_ctrl->curr_bounce_buf =
 			ena_com_get_next_bounce_buffer(&io_sq->bounce_buf_ctrl);
@@ -185,8 +188,11 @@ static int ena_com_sq_update_llq_tail(struct ena_com_io_sq *io_sq)
 	if (!pkt_ctrl->descs_left_in_line) {
 		rc = ena_com_write_bounce_buffer_to_dev(io_sq,
 							pkt_ctrl->curr_bounce_buf);
-		if (unlikely(rc))
+		if (unlikely(rc)) {
+			netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+				   "Failed to write bounce buffer to device\n");
 			return rc;
+		}
 
 		pkt_ctrl->curr_bounce_buf =
 			ena_com_get_next_bounce_buffer(&io_sq->bounce_buf_ctrl);
@@ -406,8 +412,11 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 	}
 
 	if (unlikely(io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV &&
-		     !buffer_to_push))
+		     !buffer_to_push)) {
+		netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+			   "Push header wasn't provided in LLQ mode\n");
 		return -EINVAL;
+	}
 
 	rc = ena_com_write_header_to_bounce(io_sq, buffer_to_push, header_len);
 	if (unlikely(rc))
@@ -423,6 +432,9 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 	/* If the caller doesn't want to send packets */
 	if (unlikely(!num_bufs && !header_len)) {
 		rc = ena_com_close_bounce_buffer(io_sq);
+		if (rc)
+			netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+				   "Failed to write buffers to LLQ\n");
 		*nb_hw_desc = io_sq->tail - start_tail;
 		return rc;
 	}
@@ -482,8 +494,11 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 		/* The first desc share the same desc as the header */
 		if (likely(i != 0)) {
 			rc = ena_com_sq_update_tail(io_sq);
-			if (unlikely(rc))
+			if (unlikely(rc)) {
+				netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+					   "Failed to update sq tail\n");
 				return rc;
+			}
 
 			desc = get_sq_desc(io_sq);
 			if (unlikely(!desc))
@@ -512,8 +527,11 @@ int ena_com_prepare_tx(struct ena_com_io_sq *io_sq,
 	desc->len_ctrl |= ENA_ETH_IO_TX_DESC_LAST_MASK;
 
 	rc = ena_com_sq_update_tail(io_sq);
-	if (unlikely(rc))
+	if (unlikely(rc)) {
+		netdev_err(ena_com_io_sq_to_ena_dev(io_sq)->net_device,
+			   "Failed to update sq tail of the last descriptor\n");
 		return rc;
+	}
 
 	rc = ena_com_close_bounce_buffer(io_sq);
 

drivers/net/ethernet/amazon/ena/ena_ethtool.c

@@ -233,10 +233,13 @@ int ena_get_sset_count(struct net_device *netdev, int sset)
 {
 	struct ena_adapter *adapter = netdev_priv(netdev);
 
-	if (sset != ETH_SS_STATS)
-		return -EOPNOTSUPP;
+	switch (sset) {
+	case ETH_SS_STATS:
+		return ena_get_sw_stats_count(adapter) +
+		       ena_get_hw_stats_count(adapter);
+	}
 
-	return ena_get_sw_stats_count(adapter) + ena_get_hw_stats_count(adapter);
+	return -EOPNOTSUPP;
 }
 
 static void ena_queue_strings(struct ena_adapter *adapter, u8 **data)
@@ -314,10 +317,11 @@ static void ena_get_ethtool_strings(struct net_device *netdev,
 {
 	struct ena_adapter *adapter = netdev_priv(netdev);
 
-	if (sset != ETH_SS_STATS)
-		return;
-
+	switch (sset) {
+	case ETH_SS_STATS:
 		ena_get_strings(adapter, data, adapter->eni_stats_supported);
+		break;
+	}
 }
 
 static int ena_get_link_ksettings(struct net_device *netdev,

drivers/net/ethernet/amazon/ena/ena_netdev.c

@@ -35,9 +35,6 @@ MODULE_LICENSE("GPL");
 
 #define DEFAULT_MSG_ENABLE (NETIF_MSG_DRV | NETIF_MSG_PROBE | NETIF_MSG_IFUP | \
 		NETIF_MSG_TX_DONE | NETIF_MSG_TX_ERR | NETIF_MSG_RX_ERR)
-static int debug = -1;
-module_param(debug, int, 0);
-MODULE_PARM_DESC(debug, "Debug level (0=none,...,16=all)");
 
 static struct ena_aenq_handlers aenq_handlers;
 
@@ -89,6 +86,12 @@ static void ena_increase_stat(u64 *statp, u64 cnt,
 	u64_stats_update_end(syncp);
 }
 
+static void ena_ring_tx_doorbell(struct ena_ring *tx_ring)
+{
+	ena_com_write_sq_doorbell(tx_ring->ena_com_io_sq);
+	ena_increase_stat(&tx_ring->tx_stats.doorbells, 1, &tx_ring->syncp);
+}
+
 static void ena_tx_timeout(struct net_device *dev, unsigned int txqueue)
 {
 	struct ena_adapter *adapter = netdev_priv(dev);
@@ -147,7 +150,7 @@ static int ena_xmit_common(struct net_device *dev,
 		netif_dbg(adapter, tx_queued, dev,
 			  "llq tx max burst size of queue %d achieved, writing doorbell to send burst\n",
 			  ring->qid);
-		ena_com_write_sq_doorbell(ring->ena_com_io_sq);
+		ena_ring_tx_doorbell(ring);
 	}
 
 	/* prepare the packet's descriptors to dma engine */
@@ -197,7 +200,6 @@ static int ena_xdp_io_poll(struct napi_struct *napi, int budget)
 	int ret;
 
 	xdp_ring = ena_napi->xdp_ring;
-	xdp_ring->first_interrupt = ena_napi->first_interrupt;
 
 	xdp_budget = budget;
 
@@ -229,6 +231,7 @@ static int ena_xdp_io_poll(struct napi_struct *napi, int budget)
 	xdp_ring->tx_stats.napi_comp += napi_comp_call;
 	xdp_ring->tx_stats.tx_poll++;
 	u64_stats_update_end(&xdp_ring->syncp);
+	xdp_ring->tx_stats.last_napi_jiffies = jiffies;
 
 	return ret;
 }
@@ -316,14 +319,12 @@ static int ena_xdp_xmit_frame(struct ena_ring *xdp_ring,
 			     xdpf->len);
 	if (rc)
 		goto error_unmap_dma;
-	/* trigger the dma engine. ena_com_write_sq_doorbell()
-	 * has a mb
+
+	/* trigger the dma engine. ena_ring_tx_doorbell()
+	 * calls a memory barrier inside it.
 	 */
-	if (flags & XDP_XMIT_FLUSH) {
-		ena_com_write_sq_doorbell(xdp_ring->ena_com_io_sq);
-		ena_increase_stat(&xdp_ring->tx_stats.doorbells, 1,
-				  &xdp_ring->syncp);
-	}
+	if (flags & XDP_XMIT_FLUSH)
+		ena_ring_tx_doorbell(xdp_ring);
 
 	return rc;
 
@@ -364,11 +365,8 @@ static int ena_xdp_xmit(struct net_device *dev, int n,
 	}
 
 	/* Ring doorbell to make device aware of the packets */
-	if (flags & XDP_XMIT_FLUSH) {
-		ena_com_write_sq_doorbell(xdp_ring->ena_com_io_sq);
-		ena_increase_stat(&xdp_ring->tx_stats.doorbells, 1,
-				  &xdp_ring->syncp);
-	}
+	if (flags & XDP_XMIT_FLUSH)
+		ena_ring_tx_doorbell(xdp_ring);
 
 	spin_unlock(&xdp_ring->xdp_tx_lock);
 
@@ -383,7 +381,6 @@ static int ena_xdp_execute(struct ena_ring *rx_ring, struct xdp_buff *xdp)
 	u32 verdict = XDP_PASS;
 	struct xdp_frame *xdpf;
 	u64 *xdp_stat;
-	int qid;
 
 	rcu_read_lock();
 	xdp_prog = READ_ONCE(rx_ring->xdp_bpf_prog);
@@ -404,8 +401,7 @@ static int ena_xdp_execute(struct ena_ring *rx_ring, struct xdp_buff *xdp)
 		}
 
 		/* Find xmit queue */
-		qid = rx_ring->qid + rx_ring->adapter->num_io_queues;
-		xdp_ring = &rx_ring->adapter->tx_ring[qid];
+		xdp_ring = rx_ring->xdp_ring;
 
 		/* The XDP queues are shared between XDP_TX and XDP_REDIRECT */
 		spin_lock(&xdp_ring->xdp_tx_lock);
@@ -532,7 +528,7 @@ static void ena_xdp_exchange_program_rx_in_range(struct ena_adapter *adapter,
 			rx_ring->rx_headroom = XDP_PACKET_HEADROOM;
 		} else {
 			ena_xdp_unregister_rxq_info(rx_ring);
-			rx_ring->rx_headroom = 0;
+			rx_ring->rx_headroom = NET_SKB_PAD;
 		}
 	}
 }
@@ -681,7 +677,6 @@ static void ena_init_io_rings_common(struct ena_adapter *adapter,
 	ring->ena_dev = adapter->ena_dev;
 	ring->per_napi_packets = 0;
 	ring->cpu = 0;
-	ring->first_interrupt = false;
 	ring->no_interrupt_event_cnt = 0;
 	u64_stats_init(&ring->syncp);
 }
@@ -724,7 +719,9 @@ static void ena_init_io_rings(struct ena_adapter *adapter,
 			rxr->smoothed_interval =
 				ena_com_get_nonadaptive_moderation_interval_rx(ena_dev);
 			rxr->empty_rx_queue = 0;
+			rxr->rx_headroom = NET_SKB_PAD;
 			adapter->ena_napi[i].dim.mode = DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+			rxr->xdp_ring = &adapter->tx_ring[i + adapter->num_io_queues];
 		}
 	}
 }
@@ -978,47 +975,65 @@ static void ena_free_all_io_rx_resources(struct ena_adapter *adapter)
 		ena_free_rx_resources(adapter, i);
 }
 
-static int ena_alloc_rx_page(struct ena_ring *rx_ring,
-				    struct ena_rx_buffer *rx_info, gfp_t gfp)
+struct page *ena_alloc_map_page(struct ena_ring *rx_ring, dma_addr_t *dma)
 {
-	int headroom = rx_ring->rx_headroom;
-	struct ena_com_buf *ena_buf;
 	struct page *page;
-	dma_addr_t dma;
-
-	/* restore page offset value in case it has been changed by device */
-	rx_info->page_offset = headroom;
-
-	/* if previous allocated page is not used */
-	if (unlikely(rx_info->page))
-		return 0;
 
-	page = alloc_page(gfp);
-	if (unlikely(!page)) {
+	/* This would allocate the page on the same NUMA node the executing code
+	 * is running on.
+	 */
+	page = dev_alloc_page();
+	if (!page) {
 		ena_increase_stat(&rx_ring->rx_stats.page_alloc_fail, 1,
 				  &rx_ring->syncp);
-		return -ENOMEM;
+		return ERR_PTR(-ENOSPC);
 	}
 
 	/* To enable NIC-side port-mirroring, AKA SPAN port,
 	 * we make the buffer readable from the nic as well
 	 */
-	dma = dma_map_page(rx_ring->dev, page, 0, ENA_PAGE_SIZE,
+	*dma = dma_map_page(rx_ring->dev, page, 0, ENA_PAGE_SIZE,
 			    DMA_BIDIRECTIONAL);
-	if (unlikely(dma_mapping_error(rx_ring->dev, dma))) {
+	if (unlikely(dma_mapping_error(rx_ring->dev, *dma))) {
 		ena_increase_stat(&rx_ring->rx_stats.dma_mapping_err, 1,
 				  &rx_ring->syncp);
-
 		__free_page(page);
-		return -EIO;
+		return ERR_PTR(-EIO);
 	}
+
+	return page;
+}
+
+static int ena_alloc_rx_buffer(struct ena_ring *rx_ring,
+			       struct ena_rx_buffer *rx_info)
+{
+	int headroom = rx_ring->rx_headroom;
+	struct ena_com_buf *ena_buf;
+	struct page *page;
+	dma_addr_t dma;
+	int tailroom;
+
+	/* restore page offset value in case it has been changed by device */
+	rx_info->page_offset = headroom;
+
+	/* if previous allocated page is not used */
+	if (unlikely(rx_info->page))
+		return 0;
+
+	/* We handle DMA here */
+	page = ena_alloc_map_page(rx_ring, &dma);
+	if (unlikely(IS_ERR(page)))
+		return PTR_ERR(page);
+
 	netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
 		  "Allocate page %p, rx_info %p\n", page, rx_info);
 
+	tailroom = SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
+
 	rx_info->page = page;
 	ena_buf = &rx_info->ena_buf;
 	ena_buf->paddr = dma + headroom;
-	ena_buf->len = ENA_PAGE_SIZE - headroom;
+	ena_buf->len = ENA_PAGE_SIZE - headroom - tailroom;
 
 	return 0;
 }
@@ -1065,8 +1080,7 @@ static int ena_refill_rx_bufs(struct ena_ring *rx_ring, u32 num)
 
 		rx_info = &rx_ring->rx_buffer_info[req_id];
 
-		rc = ena_alloc_rx_page(rx_ring, rx_info,
-				       GFP_ATOMIC | __GFP_COMP);
+		rc = ena_alloc_rx_buffer(rx_ring, rx_info);
 		if (unlikely(rc < 0)) {
 			netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev,
 				   "Failed to allocate buffer for rx queue %d\n",
@@ -1384,21 +1398,23 @@ static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget)
 	return tx_pkts;
 }
 
-static struct sk_buff *ena_alloc_skb(struct ena_ring *rx_ring, bool frags)
+static struct sk_buff *ena_alloc_skb(struct ena_ring *rx_ring, void *first_frag)
 {
 	struct sk_buff *skb;
 
-	if (frags)
-		skb = napi_get_frags(rx_ring->napi);
-	else
+	if (!first_frag)
 		skb = netdev_alloc_skb_ip_align(rx_ring->netdev,
 						rx_ring->rx_copybreak);
+	else
+		skb = build_skb(first_frag, ENA_PAGE_SIZE);
 
 	if (unlikely(!skb)) {
 		ena_increase_stat(&rx_ring->rx_stats.skb_alloc_fail, 1,
 				  &rx_ring->syncp);
+
 		netif_dbg(rx_ring->adapter, rx_err, rx_ring->netdev,
-			  "Failed to allocate skb. frags: %d\n", frags);
+			  "Failed to allocate skb. first_frag %s\n",
+			  first_frag ? "provided" : "not provided");
 		return NULL;
 	}
 
@@ -1410,10 +1426,12 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
 				  u32 descs,
 				  u16 *next_to_clean)
 {
-	struct sk_buff *skb;
 	struct ena_rx_buffer *rx_info;
 	u16 len, req_id, buf = 0;
-	void *va;
+	struct sk_buff *skb;
+	void *page_addr;
+	u32 page_offset;
+	void *data_addr;
 
 	len = ena_bufs[buf].len;
 	req_id = ena_bufs[buf].req_id;
@@ -1431,12 +1449,14 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
 		  rx_info, rx_info->page);
 
 	/* save virt address of first buffer */
-	va = page_address(rx_info->page) + rx_info->page_offset;
+	page_addr = page_address(rx_info->page);
+	page_offset = rx_info->page_offset;
+	data_addr = page_addr + page_offset;
 
-	prefetch(va);
+	prefetch(data_addr);
 
 	if (len <= rx_ring->rx_copybreak) {
-		skb = ena_alloc_skb(rx_ring, false);
+		skb = ena_alloc_skb(rx_ring, NULL);
 		if (unlikely(!skb))
 			return NULL;
 
@@ -1449,7 +1469,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
 					dma_unmap_addr(&rx_info->ena_buf, paddr),
 					len,
 					DMA_FROM_DEVICE);
-		skb_copy_to_linear_data(skb, va, len);
+		skb_copy_to_linear_data(skb, data_addr, len);
 		dma_sync_single_for_device(rx_ring->dev,
 					   dma_unmap_addr(&rx_info->ena_buf, paddr),
 					   len,
@@ -1463,16 +1483,18 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
 		return skb;
 	}
 
-	skb = ena_alloc_skb(rx_ring, true);
+	ena_unmap_rx_buff(rx_ring, rx_info);
+
+	skb = ena_alloc_skb(rx_ring, page_addr);
 	if (unlikely(!skb))
 		return NULL;
 
-	do {
-		ena_unmap_rx_buff(rx_ring, rx_info);
-
-		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page,
-				rx_info->page_offset, len, ENA_PAGE_SIZE);
+	/* Populate skb's linear part */
+	skb_reserve(skb, page_offset);
+	skb_put(skb, len);
+	skb->protocol = eth_type_trans(skb, rx_ring->netdev);
 
+	do {
 		netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
 			  "RX skb updated. len %d. data_len %d\n",
 			  skb->len, skb->data_len);
@@ -1491,6 +1513,12 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
 		req_id = ena_bufs[buf].req_id;
 
 		rx_info = &rx_ring->rx_buffer_info[req_id];
+
+		ena_unmap_rx_buff(rx_ring, rx_info);
+
+		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page,
+				rx_info->page_offset, len, ENA_PAGE_SIZE);
+
 	} while (1);
 
 	return skb;
@@ -1703,14 +1731,12 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi,
 
 		skb_record_rx_queue(skb, rx_ring->qid);
 
-		if (rx_ring->ena_bufs[0].len <= rx_ring->rx_copybreak) {
-			total_len += rx_ring->ena_bufs[0].len;
+		if (rx_ring->ena_bufs[0].len <= rx_ring->rx_copybreak)
 			rx_copybreak_pkt++;
-			napi_gro_receive(napi, skb);
-		} else {
+
 		total_len += skb->len;
-			napi_gro_frags(napi);
-		}
+
+		napi_gro_receive(napi, skb);
 
 		res_budget--;
 	} while (likely(res_budget));
@@ -1922,9 +1948,6 @@ static int ena_io_poll(struct napi_struct *napi, int budget)
 	tx_ring = ena_napi->tx_ring;
 	rx_ring = ena_napi->rx_ring;
 
-	tx_ring->first_interrupt = ena_napi->first_interrupt;
-	rx_ring->first_interrupt = ena_napi->first_interrupt;
-
 	tx_budget = tx_ring->ring_size / ENA_TX_POLL_BUDGET_DIVIDER;
 
 	if (!test_bit(ENA_FLAG_DEV_UP, &tx_ring->adapter->flags) ||
@@ -1979,6 +2002,8 @@ static int ena_io_poll(struct napi_struct *napi, int budget)
 	tx_ring->tx_stats.tx_poll++;
 	u64_stats_update_end(&tx_ring->syncp);
 
+	tx_ring->tx_stats.last_napi_jiffies = jiffies;
+
 	return ret;
 }
 
@@ -2003,7 +2028,8 @@ static irqreturn_t ena_intr_msix_io(int irq, void *data)
 {
 	struct ena_napi *ena_napi = data;
 
-	ena_napi->first_interrupt = true;
+	/* Used to check HW health */
+	WRITE_ONCE(ena_napi->first_interrupt, true);
 
 	WRITE_ONCE(ena_napi->interrupts_masked, true);
 	smp_wmb(); /* write interrupts_masked before calling napi */
@@ -3089,14 +3115,11 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev)
 		}
 	}
 
-	if (netif_xmit_stopped(txq) || !netdev_xmit_more()) {
-		/* trigger the dma engine. ena_com_write_sq_doorbell()
-		 * has a mb
+	if (netif_xmit_stopped(txq) || !netdev_xmit_more())
+		/* trigger the dma engine. ena_ring_tx_doorbell()
+		 * calls a memory barrier inside it.
 		 */
-		ena_com_write_sq_doorbell(tx_ring->ena_com_io_sq);
-		ena_increase_stat(&tx_ring->tx_stats.doorbells, 1,
-				  &tx_ring->syncp);
-	}
+		ena_ring_tx_doorbell(tx_ring);
 
 	return NETDEV_TX_OK;
 
@@ -3346,7 +3369,7 @@ static int ena_set_queues_placement_policy(struct pci_dev *pdev,
 
 	llq_feature_mask = 1 << ENA_ADMIN_LLQ;
 	if (!(ena_dev->supported_features & llq_feature_mask)) {
-		dev_err(&pdev->dev,
+		dev_warn(&pdev->dev,
 			"LLQ is not supported Fallback to host mode policy.\n");
 		ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
 		return 0;
@@ -3657,7 +3680,9 @@ static void ena_fw_reset_device(struct work_struct *work)
 static int check_for_rx_interrupt_queue(struct ena_adapter *adapter,
 					struct ena_ring *rx_ring)
 {
-	if (likely(rx_ring->first_interrupt))
+	struct ena_napi *ena_napi = container_of(rx_ring->napi, struct ena_napi, napi);
+
+	if (likely(READ_ONCE(ena_napi->first_interrupt)))
 		return 0;
 
 	if (ena_com_cq_empty(rx_ring->ena_com_io_cq))
@@ -3681,6 +3706,10 @@ static int check_for_rx_interrupt_queue(struct ena_adapter *adapter,
 static int check_missing_comp_in_tx_queue(struct ena_adapter *adapter,
 					  struct ena_ring *tx_ring)
 {
+	struct ena_napi *ena_napi = container_of(tx_ring->napi, struct ena_napi, napi);
+	unsigned int time_since_last_napi;
+	unsigned int missing_tx_comp_to;
+	bool is_tx_comp_time_expired;
 	struct ena_tx_buffer *tx_buf;
 	unsigned long last_jiffies;
 	u32 missed_tx = 0;
@@ -3694,8 +3723,10 @@ static int check_missing_comp_in_tx_queue(struct ena_adapter *adapter,
 			/* no pending Tx at this location */
 			continue;
 
-		if (unlikely(!tx_ring->first_interrupt && time_is_before_jiffies(last_jiffies +
-			     2 * adapter->missing_tx_completion_to))) {
+		is_tx_comp_time_expired = time_is_before_jiffies(last_jiffies +
+			 2 * adapter->missing_tx_completion_to);
+
+		if (unlikely(!READ_ONCE(ena_napi->first_interrupt) && is_tx_comp_time_expired)) {
 			/* If after graceful period interrupt is still not
 			 * received, we schedule a reset
 			 */
@@ -3708,12 +3739,17 @@ static int check_missing_comp_in_tx_queue(struct ena_adapter *adapter,
 			return -EIO;
 		}
 
-		if (unlikely(time_is_before_jiffies(last_jiffies +
-				adapter->missing_tx_completion_to))) {
-			if (!tx_buf->print_once)
+		is_tx_comp_time_expired = time_is_before_jiffies(last_jiffies +
+			adapter->missing_tx_completion_to);
+
+		if (unlikely(is_tx_comp_time_expired)) {
+			if (!tx_buf->print_once) {
+				time_since_last_napi = jiffies_to_usecs(jiffies - tx_ring->tx_stats.last_napi_jiffies);
+				missing_tx_comp_to = jiffies_to_msecs(adapter->missing_tx_completion_to);
 				netif_notice(adapter, tx_err, adapter->netdev,
-					     "Found a Tx that wasn't completed on time, qid %d, index %d.\n",
-					     tx_ring->qid, i);
+					     "Found a Tx that wasn't completed on time, qid %d, index %d. %u usecs have passed since last napi execution. Missing Tx timeout value %u msecs\n",
+					     tx_ring->qid, i, time_since_last_napi, missing_tx_comp_to);
+			}
 
 			tx_buf->print_once = 1;
 			missed_tx++;
@@ -4244,7 +4280,7 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
 	adapter->ena_dev = ena_dev;
 	adapter->netdev = netdev;
 	adapter->pdev = pdev;
-	adapter->msg_enable = netif_msg_init(debug, DEFAULT_MSG_ENABLE);
+	adapter->msg_enable = DEFAULT_MSG_ENABLE;
 
 	ena_dev->net_device = netdev;
 

drivers/net/ethernet/amazon/ena/ena_netdev.h

@@ -55,12 +55,6 @@
 #define ENA_TX_WAKEUP_THRESH		(MAX_SKB_FRAGS + 2)
 #define ENA_DEFAULT_RX_COPYBREAK	(256 - NET_IP_ALIGN)
 
-/* limit the buffer size to 600 bytes to handle MTU changes from very
- * small to very large, in which case the number of buffers per packet
- * could exceed ENA_PKT_MAX_BUFS
- */
-#define ENA_DEFAULT_MIN_RX_BUFF_ALLOC_SIZE 600
-
 #define ENA_MIN_MTU		128
 
 #define ENA_NAME_MAX_LEN	20
@@ -135,12 +129,12 @@ struct ena_irq {
 };
 
 struct ena_napi {
-	struct napi_struct napi ____cacheline_aligned;
+	u8 first_interrupt ____cacheline_aligned;
+	u8 interrupts_masked;
+	struct napi_struct napi;
 	struct ena_ring *tx_ring;
 	struct ena_ring *rx_ring;
 	struct ena_ring *xdp_ring;
-	bool first_interrupt;
-	bool interrupts_masked;
 	u32 qid;
 	struct dim dim;
 };
@@ -212,6 +206,7 @@ struct ena_stats_tx {
 	u64 llq_buffer_copy;
 	u64 missed_tx;
 	u64 unmask_interrupt;
+	u64 last_napi_jiffies;
 };
 
 struct ena_stats_rx {
@@ -259,6 +254,10 @@ struct ena_ring {
 	struct bpf_prog *xdp_bpf_prog;
 	struct xdp_rxq_info xdp_rxq;
 	spinlock_t xdp_tx_lock;	/* synchronize XDP TX/Redirect traffic */
+	/* Used for rx queues only to point to the xdp tx ring, to
+	 * which traffic should be redirected from this rx ring.
+	 */
+	struct ena_ring *xdp_ring;
 
 	u16 next_to_use;
 	u16 next_to_clean;
@@ -271,7 +270,6 @@ struct ena_ring {
 	/* The maximum header length the device can handle */
 	u8 tx_max_header_size;
 
-	bool first_interrupt;
 	bool disable_meta_caching;
 	u16 no_interrupt_event_cnt;
 
@@ -414,11 +412,6 @@ enum ena_xdp_errors_t {
 	ENA_XDP_NO_ENOUGH_QUEUES,
 };
 
-static inline bool ena_xdp_queues_present(struct ena_adapter *adapter)
-{
-	return adapter->xdp_first_ring != 0;
-}
-
 static inline bool ena_xdp_present(struct ena_adapter *adapter)
 {
 	return !!adapter->xdp_bpf_prog;