Merge branch 'ena-next'

Arthur Kiyanovski says:

====================
Improving performance and reducing latencies, by using latest capabilities exposed in ENA device

This patchset introduces the following:
1. A new placement policy of Tx headers and descriptors, which takes
advantage of an option to place headers + descriptors in device memory
space. This is sometimes referred to as LLQ - low latency queue.
The patch set defines the admin capability, maps the device memory as
write-combined, and adds a mode in transmit datapath to do header +
descriptor placement on the device.
2. Support for RX checksum offloading
3. Miscelaneous small improvements and code cleanups

Note: V1 of this patchset was created as if patches e2a322a0 248ab773
from net were applied to net-next before applying the patchset. This V2
version does not assume this, and should be applyed directly on net-next
without the aformentioned patches.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

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

@@ -37,6 +37,7 @@
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/gfp.h>
+#include <linux/io.h>
 #include <linux/sched.h>
 #include <linux/sizes.h>
 #include <linux/spinlock.h>
@@ -108,6 +109,14 @@ enum ena_intr_moder_level {
 	ENA_INTR_MAX_NUM_OF_LEVELS,
 };
 
+struct ena_llq_configurations {
+	enum ena_admin_llq_header_location llq_header_location;
+	enum ena_admin_llq_ring_entry_size llq_ring_entry_size;
+	enum ena_admin_llq_stride_ctrl  llq_stride_ctrl;
+	enum ena_admin_llq_num_descs_before_header llq_num_decs_before_header;
+	u16 llq_ring_entry_size_value;
+};
+
 struct ena_intr_moder_entry {
 	unsigned int intr_moder_interval;
 	unsigned int pkts_per_interval;
@@ -142,6 +151,15 @@ struct ena_com_tx_meta {
 	u16 l4_hdr_len; /* In words */
 };
 
+struct ena_com_llq_info {
+	u16 header_location_ctrl;
+	u16 desc_stride_ctrl;
+	u16 desc_list_entry_size_ctrl;
+	u16 desc_list_entry_size;
+	u16 descs_num_before_header;
+	u16 descs_per_entry;
+};
+
 struct ena_com_io_cq {
 	struct ena_com_io_desc_addr cdesc_addr;
 
@@ -179,6 +197,20 @@ struct ena_com_io_cq {
 
 } ____cacheline_aligned;
 
+struct ena_com_io_bounce_buffer_control {
+	u8 *base_buffer;
+	u16 next_to_use;
+	u16 buffer_size;
+	u16 buffers_num;  /* Must be a power of 2 */
+};
+
+/* This struct is to keep tracking the current location of the next llq entry */
+struct ena_com_llq_pkt_ctrl {
+	u8 *curr_bounce_buf;
+	u16 idx;
+	u16 descs_left_in_line;
+};
+
 struct ena_com_io_sq {
 	struct ena_com_io_desc_addr desc_addr;
 
@@ -190,6 +222,9 @@ struct ena_com_io_sq {
 
 	u32 msix_vector;
 	struct ena_com_tx_meta cached_tx_meta;
+	struct ena_com_llq_info llq_info;
+	struct ena_com_llq_pkt_ctrl llq_buf_ctrl;
+	struct ena_com_io_bounce_buffer_control bounce_buf_ctrl;
 
 	u16 q_depth;
 	u16 qid;
@@ -197,6 +232,7 @@ struct ena_com_io_sq {
 	u16 idx;
 	u16 tail;
 	u16 next_to_comp;
+	u16 llq_last_copy_tail;
 	u32 tx_max_header_size;
 	u8 phase;
 	u8 desc_entry_size;
@@ -334,6 +370,8 @@ struct ena_com_dev {
 	u16 intr_delay_resolution;
 	u32 intr_moder_tx_interval;
 	struct ena_intr_moder_entry *intr_moder_tbl;
+
+	struct ena_com_llq_info llq_info;
 };
 
 struct ena_com_dev_get_features_ctx {
@@ -342,6 +380,7 @@ struct ena_com_dev_get_features_ctx {
 	struct ena_admin_feature_aenq_desc aenq;
 	struct ena_admin_feature_offload_desc offload;
 	struct ena_admin_ena_hw_hints hw_hints;
+	struct ena_admin_feature_llq_desc llq;
 };
 
 struct ena_com_create_io_ctx {
@@ -397,8 +436,6 @@ void ena_com_mmio_reg_read_request_destroy(struct ena_com_dev *ena_dev);
 /* ena_com_admin_init - Init the admin and the async queues
  * @ena_dev: ENA communication layer struct
  * @aenq_handlers: Those handlers to be called upon event.
- * @init_spinlock: Indicate if this method should init the admin spinlock or
- * the spinlock was init before (for example, in a case of FLR).
  *
  * Initialize the admin submission and completion queues.
  * Initialize the asynchronous events notification queues.
@@ -406,8 +443,7 @@ void ena_com_mmio_reg_read_request_destroy(struct ena_com_dev *ena_dev);
  * @return - 0 on success, negative value on failure.
  */
 int ena_com_admin_init(struct ena_com_dev *ena_dev,
-		       struct ena_aenq_handlers *aenq_handlers,
-		       bool init_spinlock);
+		       struct ena_aenq_handlers *aenq_handlers);
 
 /* ena_com_admin_destroy - Destroy the admin and the async events queues.
  * @ena_dev: ENA communication layer struct
@@ -935,6 +971,16 @@ void ena_com_get_intr_moderation_entry(struct ena_com_dev *ena_dev,
 				       enum ena_intr_moder_level level,
 				       struct ena_intr_moder_entry *entry);
 
+/* ena_com_config_dev_mode - Configure the placement policy of the device.
+ * @ena_dev: ENA communication layer struct
+ * @llq_features: LLQ feature descriptor, retrieve via
+ *                ena_com_get_dev_attr_feat.
+ * @ena_llq_config: The default driver LLQ parameters configurations
+ */
+int ena_com_config_dev_mode(struct ena_com_dev *ena_dev,
+			    struct ena_admin_feature_llq_desc *llq_features,
+			    struct ena_llq_configurations *llq_default_config);
+
 static inline bool ena_com_get_adaptive_moderation_enabled(struct ena_com_dev *ena_dev)
 {
 	return ena_dev->adaptive_coalescing;
@@ -1044,4 +1090,21 @@ static inline void ena_com_update_intr_reg(struct ena_eth_io_intr_reg *intr_reg,
 		intr_reg->intr_control |= ENA_ETH_IO_INTR_REG_INTR_UNMASK_MASK;
 }
 
+static inline u8 *ena_com_get_next_bounce_buffer(struct ena_com_io_bounce_buffer_control *bounce_buf_ctrl)
+{
+	u16 size, buffers_num;
+	u8 *buf;
+
+	size = bounce_buf_ctrl->buffer_size;
+	buffers_num = bounce_buf_ctrl->buffers_num;
+
+	buf = bounce_buf_ctrl->base_buffer +
+		(bounce_buf_ctrl->next_to_use++ & (buffers_num - 1)) * size;
+
+	prefetchw(bounce_buf_ctrl->base_buffer +
+		(bounce_buf_ctrl->next_to_use & (buffers_num - 1)) * size);
+
+	return buf;
+}
+
 #endif /* !(ENA_COM) */

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

@@ -32,8 +32,8 @@
 #ifndef _ENA_COMMON_H_
 #define _ENA_COMMON_H_
 
-#define ENA_COMMON_SPEC_VERSION_MAJOR	0 /*  */
-#define ENA_COMMON_SPEC_VERSION_MINOR	10 /*  */
+#define ENA_COMMON_SPEC_VERSION_MAJOR        2
+#define ENA_COMMON_SPEC_VERSION_MINOR        0
 
 /* ENA operates with 48-bit memory addresses. ena_mem_addr_t */
 struct ena_common_mem_addr {

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

@@ -67,6 +67,7 @@ struct ena_com_rx_ctx {
 	enum ena_eth_io_l4_proto_index l4_proto;
 	bool l3_csum_err;
 	bool l4_csum_err;
+	u8 l4_csum_checked;
 	/* fragmented packet */
 	bool frag;
 	u32 hash;
@@ -86,8 +87,6 @@ int ena_com_add_single_rx_desc(struct ena_com_io_sq *io_sq,
 			       struct ena_com_buf *ena_buf,
 			       u16 req_id);
 
-int ena_com_tx_comp_req_id_get(struct ena_com_io_cq *io_cq, u16 *req_id);
-
 bool ena_com_cq_empty(struct ena_com_io_cq *io_cq);
 
 static inline void ena_com_unmask_intr(struct ena_com_io_cq *io_cq,
@@ -96,7 +95,7 @@ static inline void ena_com_unmask_intr(struct ena_com_io_cq *io_cq,
 	writel(intr_reg->intr_control, io_cq->unmask_reg);
 }
 
-static inline int ena_com_sq_empty_space(struct ena_com_io_sq *io_sq)
+static inline int ena_com_free_desc(struct ena_com_io_sq *io_sq)
 {
 	u16 tail, next_to_comp, cnt;
 
@@ -107,11 +106,28 @@ static inline int ena_com_sq_empty_space(struct ena_com_io_sq *io_sq)
 	return io_sq->q_depth - 1 - cnt;
 }
 
-static inline int ena_com_write_sq_doorbell(struct ena_com_io_sq *io_sq)
+/* Check if the submission queue has enough space to hold required_buffers */
+static inline bool ena_com_sq_have_enough_space(struct ena_com_io_sq *io_sq,
+						u16 required_buffers)
 {
-	u16 tail;
+	int temp;
 
-	tail = io_sq->tail;
+	if (io_sq->mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_HOST)
+		return ena_com_free_desc(io_sq) >= required_buffers;
+
+	/* This calculation doesn't need to be 100% accurate. So to reduce
+	 * the calculation overhead just Subtract 2 lines from the free descs
+	 * (one for the header line and one to compensate the devision
+	 * down calculation.
+	 */
+	temp = required_buffers / io_sq->llq_info.descs_per_entry + 2;
+
+	return ena_com_free_desc(io_sq) > temp;
+}
+
+static inline int ena_com_write_sq_doorbell(struct ena_com_io_sq *io_sq)
+{
+	u16 tail = io_sq->tail;
 
 	pr_debug("write submission queue doorbell for queue: %d tail: %d\n",
 		 io_sq->qid, tail);
@@ -159,4 +175,48 @@ static inline void ena_com_comp_ack(struct ena_com_io_sq *io_sq, u16 elem)
 	io_sq->next_to_comp += elem;
 }
 
+static inline void ena_com_cq_inc_head(struct ena_com_io_cq *io_cq)
+{
+	io_cq->head++;
+
+	/* Switch phase bit in case of wrap around */
+	if (unlikely((io_cq->head & (io_cq->q_depth - 1)) == 0))
+		io_cq->phase ^= 1;
+}
+
+static inline int ena_com_tx_comp_req_id_get(struct ena_com_io_cq *io_cq,
+					     u16 *req_id)
+{
+	u8 expected_phase, cdesc_phase;
+	struct ena_eth_io_tx_cdesc *cdesc;
+	u16 masked_head;
+
+	masked_head = io_cq->head & (io_cq->q_depth - 1);
+	expected_phase = io_cq->phase;
+
+	cdesc = (struct ena_eth_io_tx_cdesc *)
+		((uintptr_t)io_cq->cdesc_addr.virt_addr +
+		(masked_head * io_cq->cdesc_entry_size_in_bytes));
+
+	/* When the current completion descriptor phase isn't the same as the
+	 * expected, it mean that the device still didn't update
+	 * this completion.
+	 */
+	cdesc_phase = READ_ONCE(cdesc->flags) & ENA_ETH_IO_TX_CDESC_PHASE_MASK;
+	if (cdesc_phase != expected_phase)
+		return -EAGAIN;
+
+	dma_rmb();
+
+	*req_id = READ_ONCE(cdesc->req_id);
+	if (unlikely(*req_id >= io_cq->q_depth)) {
+		pr_err("Invalid req id %d\n", cdesc->req_id);
+		return -EINVAL;
+	}
+
+	ena_com_cq_inc_head(io_cq);
+
+	return 0;
+}
+
 #endif /* ENA_ETH_COM_H_ */

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

@@ -34,23 +34,16 @@
 
 enum ena_eth_io_l3_proto_index {
 	ENA_ETH_IO_L3_PROTO_UNKNOWN                 = 0,
-
 	ENA_ETH_IO_L3_PROTO_IPV4                    = 8,
-
 	ENA_ETH_IO_L3_PROTO_IPV6                    = 11,
-
 	ENA_ETH_IO_L3_PROTO_FCOE                    = 21,
-
 	ENA_ETH_IO_L3_PROTO_ROCE                    = 22,
 };
 
 enum ena_eth_io_l4_proto_index {
 	ENA_ETH_IO_L4_PROTO_UNKNOWN                 = 0,
-
 	ENA_ETH_IO_L4_PROTO_TCP                     = 12,
-
 	ENA_ETH_IO_L4_PROTO_UDP                     = 13,
-
 	ENA_ETH_IO_L4_PROTO_ROUTEABLE_ROCE          = 23,
 };
 
@@ -242,9 +235,13 @@ struct ena_eth_io_rx_cdesc_base {
 	 *    checksum error detected, or, the controller didn't
 	 *    validate the checksum. This bit is valid only when
 	 *    l4_proto_idx indicates TCP/UDP packet, and,
-	 *    ipv4_frag is not set
+	 *    ipv4_frag is not set. This bit is valid only when
+	 *    l4_csum_checked below is set.
 	 * 15 : ipv4_frag - Indicates IPv4 fragmented packet
-	 * 23:16 : reserved16
+	 * 16 : l4_csum_checked - L4 checksum was verified
+	 *    (could be OK or error), when cleared the status of
+	 *    checksum is unknown
+	 * 23:17 : reserved17 - MBZ
 	 * 24 : phase
 	 * 25 : l3_csum2 - second checksum engine result
 	 * 26 : first - Indicates first descriptor in
@@ -390,6 +387,8 @@ struct ena_eth_io_numa_node_cfg_reg {
 #define ENA_ETH_IO_RX_CDESC_BASE_L4_CSUM_ERR_MASK           BIT(14)
 #define ENA_ETH_IO_RX_CDESC_BASE_IPV4_FRAG_SHIFT            15
 #define ENA_ETH_IO_RX_CDESC_BASE_IPV4_FRAG_MASK             BIT(15)
+#define ENA_ETH_IO_RX_CDESC_BASE_L4_CSUM_CHECKED_SHIFT      16
+#define ENA_ETH_IO_RX_CDESC_BASE_L4_CSUM_CHECKED_MASK       BIT(16)
 #define ENA_ETH_IO_RX_CDESC_BASE_PHASE_SHIFT                24
 #define ENA_ETH_IO_RX_CDESC_BASE_PHASE_MASK                 BIT(24)
 #define ENA_ETH_IO_RX_CDESC_BASE_L3_CSUM2_SHIFT             25

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

@@ -81,6 +81,7 @@ static const struct ena_stats ena_stats_tx_strings[] = {
 	ENA_STAT_TX_ENTRY(doorbells),
 	ENA_STAT_TX_ENTRY(prepare_ctx_err),
 	ENA_STAT_TX_ENTRY(bad_req_id),
+	ENA_STAT_TX_ENTRY(llq_buffer_copy),
 	ENA_STAT_TX_ENTRY(missed_tx),
 };
 
@@ -96,6 +97,7 @@ static const struct ena_stats ena_stats_rx_strings[] = {
 	ENA_STAT_RX_ENTRY(rx_copybreak_pkt),
 	ENA_STAT_RX_ENTRY(bad_req_id),
 	ENA_STAT_RX_ENTRY(empty_rx_ring),
+	ENA_STAT_RX_ENTRY(csum_unchecked),
 };
 
 static const struct ena_stats ena_stats_ena_com_strings[] = {

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

@@ -43,9 +43,9 @@
 #include "ena_com.h"
 #include "ena_eth_com.h"
 
-#define DRV_MODULE_VER_MAJOR	1
-#define DRV_MODULE_VER_MINOR	5
-#define DRV_MODULE_VER_SUBMINOR 0
+#define DRV_MODULE_VER_MAJOR	2
+#define DRV_MODULE_VER_MINOR	0
+#define DRV_MODULE_VER_SUBMINOR 1
 
 #define DRV_MODULE_NAME		"ena"
 #ifndef DRV_MODULE_VERSION
@@ -61,6 +61,17 @@
 #define ENA_ADMIN_MSIX_VEC		1
 #define ENA_MAX_MSIX_VEC(io_queues)	(ENA_ADMIN_MSIX_VEC + (io_queues))
 
+/* The ENA buffer length fields is 16 bit long. So when PAGE_SIZE == 64kB the
+ * driver passes 0.
+ * Since the max packet size the ENA handles is ~9kB limit the buffer length to
+ * 16kB.
+ */
+#if PAGE_SIZE > SZ_16K
+#define ENA_PAGE_SIZE SZ_16K
+#else
+#define ENA_PAGE_SIZE PAGE_SIZE
+#endif
+
 #define ENA_MIN_MSIX_VEC		2
 
 #define ENA_REG_BAR			0
@@ -70,7 +81,7 @@
 #define ENA_DEFAULT_RING_SIZE	(1024)
 
 #define ENA_TX_WAKEUP_THRESH		(MAX_SKB_FRAGS + 2)
-#define ENA_DEFAULT_RX_COPYBREAK	(128 - NET_IP_ALIGN)
+#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
@@ -95,10 +106,11 @@
  */
 #define ENA_TX_POLL_BUDGET_DIVIDER	4
 
-/* Refill Rx queue when number of available descriptors is below
- * QUEUE_SIZE / ENA_RX_REFILL_THRESH_DIVIDER
+/* Refill Rx queue when number of required descriptors is above
+ * QUEUE_SIZE / ENA_RX_REFILL_THRESH_DIVIDER or ENA_RX_REFILL_THRESH_PACKET
  */
 #define ENA_RX_REFILL_THRESH_DIVIDER	8
+#define ENA_RX_REFILL_THRESH_PACKET	256
 
 /* Number of queues to check for missing queues per timer service */
 #define ENA_MONITORED_TX_QUEUES	4
@@ -151,6 +163,9 @@ struct ena_tx_buffer {
 	/* num of buffers used by this skb */
 	u32 num_of_bufs;
 
+	/* Indicate if bufs[0] map the linear data of the skb. */
+	u8 map_linear_data;
+
 	/* Used for detect missing tx packets to limit the number of prints */
 	u32 print_once;
 	/* Save the last jiffies to detect missing tx packets
@@ -186,6 +201,7 @@ struct ena_stats_tx {
 	u64 tx_poll;
 	u64 doorbells;
 	u64 bad_req_id;
+	u64 llq_buffer_copy;
 	u64 missed_tx;
 };
 
@@ -201,6 +217,7 @@ struct ena_stats_rx {
 	u64 rx_copybreak_pkt;
 	u64 bad_req_id;
 	u64 empty_rx_ring;
+	u64 csum_unchecked;
 };
 
 struct ena_ring {
@@ -257,6 +274,8 @@ struct ena_ring {
 		struct ena_stats_tx tx_stats;
 		struct ena_stats_rx rx_stats;
 	};
+
+	u8 *push_buf_intermediate_buf;
 	int empty_rx_queue;
 } ____cacheline_aligned;
 
@@ -355,15 +374,4 @@ void ena_dump_stats_to_buf(struct ena_adapter *adapter, u8 *buf);
 
 int ena_get_sset_count(struct net_device *netdev, int sset);
 
-/* The ENA buffer length fields is 16 bit long. So when PAGE_SIZE == 64kB the
- * driver passas 0.
- * Since the max packet size the ENA handles is ~9kB limit the buffer length to
- * 16kB.
- */
-#if PAGE_SIZE > SZ_16K
-#define ENA_PAGE_SIZE SZ_16K
-#else
-#define ENA_PAGE_SIZE PAGE_SIZE
-#endif
-
 #endif /* !(ENA_H) */

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

@@ -34,37 +34,25 @@
 
 enum ena_regs_reset_reason_types {
 	ENA_REGS_RESET_NORMAL                       = 0,
-
 	ENA_REGS_RESET_KEEP_ALIVE_TO                = 1,
-
 	ENA_REGS_RESET_ADMIN_TO                     = 2,
-
 	ENA_REGS_RESET_MISS_TX_CMPL                 = 3,
-
 	ENA_REGS_RESET_INV_RX_REQ_ID                = 4,
-
 	ENA_REGS_RESET_INV_TX_REQ_ID                = 5,
-
 	ENA_REGS_RESET_TOO_MANY_RX_DESCS            = 6,
-
 	ENA_REGS_RESET_INIT_ERR                     = 7,
-
 	ENA_REGS_RESET_DRIVER_INVALID_STATE         = 8,
-
 	ENA_REGS_RESET_OS_TRIGGER                   = 9,
-
 	ENA_REGS_RESET_OS_NETDEV_WD                 = 10,
-
 	ENA_REGS_RESET_SHUTDOWN                     = 11,
-
 	ENA_REGS_RESET_USER_TRIGGER                 = 12,
-
 	ENA_REGS_RESET_GENERIC                      = 13,
-
 	ENA_REGS_RESET_MISS_INTERRUPT               = 14,
 };
 
 /* ena_registers offsets */
+
+/* 0 base */
 #define ENA_REGS_VERSION_OFF                                0x0
 #define ENA_REGS_CONTROLLER_VERSION_OFF                     0x4
 #define ENA_REGS_CAPS_OFF                                   0x8