diff options
Diffstat (limited to 'drivers/net/ethernet/amazon/ena/ena_netdev.c')
-rw-r--r-- | drivers/net/ethernet/amazon/ena/ena_netdev.c | 424 |
1 files changed, 291 insertions, 133 deletions
diff --git a/drivers/net/ethernet/amazon/ena/ena_netdev.c b/drivers/net/ethernet/amazon/ena/ena_netdev.c index a70bb1bb90e7..664e3ed97ea9 100644 --- a/drivers/net/ethernet/amazon/ena/ena_netdev.c +++ b/drivers/net/ethernet/amazon/ena/ena_netdev.c @@ -182,7 +182,7 @@ static void ena_init_io_rings(struct ena_adapter *adapter) ena_init_io_rings_common(adapter, rxr, i); /* TX specific ring state */ - txr->ring_size = adapter->tx_ring_size; + txr->ring_size = adapter->requested_tx_ring_size; txr->tx_max_header_size = ena_dev->tx_max_header_size; txr->tx_mem_queue_type = ena_dev->tx_mem_queue_type; txr->sgl_size = adapter->max_tx_sgl_size; @@ -190,7 +190,7 @@ static void ena_init_io_rings(struct ena_adapter *adapter) ena_com_get_nonadaptive_moderation_interval_tx(ena_dev); /* RX specific ring state */ - rxr->ring_size = adapter->rx_ring_size; + rxr->ring_size = adapter->requested_rx_ring_size; rxr->rx_copybreak = adapter->rx_copybreak; rxr->sgl_size = adapter->max_rx_sgl_size; rxr->smoothed_interval = @@ -224,33 +224,28 @@ static int ena_setup_tx_resources(struct ena_adapter *adapter, int qid) if (!tx_ring->tx_buffer_info) { tx_ring->tx_buffer_info = vzalloc(size); if (!tx_ring->tx_buffer_info) - return -ENOMEM; + goto err_tx_buffer_info; } size = sizeof(u16) * tx_ring->ring_size; - tx_ring->free_tx_ids = vzalloc_node(size, node); - if (!tx_ring->free_tx_ids) { - tx_ring->free_tx_ids = vzalloc(size); - if (!tx_ring->free_tx_ids) { - vfree(tx_ring->tx_buffer_info); - return -ENOMEM; - } + tx_ring->free_ids = vzalloc_node(size, node); + if (!tx_ring->free_ids) { + tx_ring->free_ids = vzalloc(size); + if (!tx_ring->free_ids) + goto err_tx_free_ids; } size = tx_ring->tx_max_header_size; tx_ring->push_buf_intermediate_buf = vzalloc_node(size, node); if (!tx_ring->push_buf_intermediate_buf) { tx_ring->push_buf_intermediate_buf = vzalloc(size); - if (!tx_ring->push_buf_intermediate_buf) { - vfree(tx_ring->tx_buffer_info); - vfree(tx_ring->free_tx_ids); - return -ENOMEM; - } + if (!tx_ring->push_buf_intermediate_buf) + goto err_push_buf_intermediate_buf; } /* Req id ring for TX out of order completions */ for (i = 0; i < tx_ring->ring_size; i++) - tx_ring->free_tx_ids[i] = i; + tx_ring->free_ids[i] = i; /* Reset tx statistics */ memset(&tx_ring->tx_stats, 0x0, sizeof(tx_ring->tx_stats)); @@ -259,6 +254,15 @@ static int ena_setup_tx_resources(struct ena_adapter *adapter, int qid) tx_ring->next_to_clean = 0; tx_ring->cpu = ena_irq->cpu; return 0; + +err_push_buf_intermediate_buf: + vfree(tx_ring->free_ids); + tx_ring->free_ids = NULL; +err_tx_free_ids: + vfree(tx_ring->tx_buffer_info); + tx_ring->tx_buffer_info = NULL; +err_tx_buffer_info: + return -ENOMEM; } /* ena_free_tx_resources - Free I/O Tx Resources per Queue @@ -274,8 +278,8 @@ static void ena_free_tx_resources(struct ena_adapter *adapter, int qid) vfree(tx_ring->tx_buffer_info); tx_ring->tx_buffer_info = NULL; - vfree(tx_ring->free_tx_ids); - tx_ring->free_tx_ids = NULL; + vfree(tx_ring->free_ids); + tx_ring->free_ids = NULL; vfree(tx_ring->push_buf_intermediate_buf); tx_ring->push_buf_intermediate_buf = NULL; @@ -322,7 +326,7 @@ static void ena_free_all_io_tx_resources(struct ena_adapter *adapter) ena_free_tx_resources(adapter, i); } -static inline int validate_rx_req_id(struct ena_ring *rx_ring, u16 req_id) +static int validate_rx_req_id(struct ena_ring *rx_ring, u16 req_id) { if (likely(req_id < rx_ring->ring_size)) return 0; @@ -373,18 +377,19 @@ static int ena_setup_rx_resources(struct ena_adapter *adapter, } size = sizeof(u16) * rx_ring->ring_size; - rx_ring->free_rx_ids = vzalloc_node(size, node); - if (!rx_ring->free_rx_ids) { - rx_ring->free_rx_ids = vzalloc(size); - if (!rx_ring->free_rx_ids) { + rx_ring->free_ids = vzalloc_node(size, node); + if (!rx_ring->free_ids) { + rx_ring->free_ids = vzalloc(size); + if (!rx_ring->free_ids) { vfree(rx_ring->rx_buffer_info); + rx_ring->rx_buffer_info = NULL; return -ENOMEM; } } /* Req id ring for receiving RX pkts out of order */ for (i = 0; i < rx_ring->ring_size; i++) - rx_ring->free_rx_ids[i] = i; + rx_ring->free_ids[i] = i; /* Reset rx statistics */ memset(&rx_ring->rx_stats, 0x0, sizeof(rx_ring->rx_stats)); @@ -410,8 +415,8 @@ static void ena_free_rx_resources(struct ena_adapter *adapter, vfree(rx_ring->rx_buffer_info); rx_ring->rx_buffer_info = NULL; - vfree(rx_ring->free_rx_ids); - rx_ring->free_rx_ids = NULL; + vfree(rx_ring->free_ids); + rx_ring->free_ids = NULL; } /* ena_setup_all_rx_resources - allocate I/O Rx queues resources for all queues @@ -455,7 +460,7 @@ static void ena_free_all_io_rx_resources(struct ena_adapter *adapter) ena_free_rx_resources(adapter, i); } -static inline int ena_alloc_rx_page(struct ena_ring *rx_ring, +static int ena_alloc_rx_page(struct ena_ring *rx_ring, struct ena_rx_buffer *rx_info, gfp_t gfp) { struct ena_com_buf *ena_buf; @@ -526,7 +531,7 @@ static int ena_refill_rx_bufs(struct ena_ring *rx_ring, u32 num) for (i = 0; i < num; i++) { struct ena_rx_buffer *rx_info; - req_id = rx_ring->free_rx_ids[next_to_use]; + req_id = rx_ring->free_ids[next_to_use]; rc = validate_rx_req_id(rx_ring, req_id); if (unlikely(rc < 0)) break; @@ -589,7 +594,6 @@ static void ena_free_rx_bufs(struct ena_adapter *adapter, /* ena_refill_all_rx_bufs - allocate all queues Rx buffers * @adapter: board private structure - * */ static void ena_refill_all_rx_bufs(struct ena_adapter *adapter) { @@ -616,7 +620,7 @@ static void ena_free_all_rx_bufs(struct ena_adapter *adapter) ena_free_rx_bufs(adapter, i); } -static inline void ena_unmap_tx_skb(struct ena_ring *tx_ring, +static void ena_unmap_tx_skb(struct ena_ring *tx_ring, struct ena_tx_buffer *tx_info) { struct ena_com_buf *ena_buf; @@ -792,7 +796,7 @@ static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget) tx_pkts++; total_done += tx_info->tx_descs; - tx_ring->free_tx_ids[next_to_clean] = req_id; + tx_ring->free_ids[next_to_clean] = req_id; next_to_clean = ENA_TX_RING_IDX_NEXT(next_to_clean, tx_ring->ring_size); } @@ -906,7 +910,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, skb_put(skb, len); skb->protocol = eth_type_trans(skb, rx_ring->netdev); - rx_ring->free_rx_ids[*next_to_clean] = req_id; + rx_ring->free_ids[*next_to_clean] = req_id; *next_to_clean = ENA_RX_RING_IDX_ADD(*next_to_clean, descs, rx_ring->ring_size); return skb; @@ -930,7 +934,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, rx_info->page = NULL; - rx_ring->free_rx_ids[*next_to_clean] = req_id; + rx_ring->free_ids[*next_to_clean] = req_id; *next_to_clean = ENA_RX_RING_IDX_NEXT(*next_to_clean, rx_ring->ring_size); @@ -951,7 +955,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring, * @ena_rx_ctx: received packet context/metadata * @skb: skb currently being received and modified */ -static inline void ena_rx_checksum(struct ena_ring *rx_ring, +static void ena_rx_checksum(struct ena_ring *rx_ring, struct ena_com_rx_ctx *ena_rx_ctx, struct sk_buff *skb) { @@ -996,6 +1000,9 @@ static inline void ena_rx_checksum(struct ena_ring *rx_ring, if (likely(ena_rx_ctx->l4_csum_checked)) { skb->ip_summed = CHECKSUM_UNNECESSARY; + u64_stats_update_begin(&rx_ring->syncp); + rx_ring->rx_stats.csum_good++; + u64_stats_update_end(&rx_ring->syncp); } else { u64_stats_update_begin(&rx_ring->syncp); rx_ring->rx_stats.csum_unchecked++; @@ -1083,7 +1090,7 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi, /* exit if we failed to retrieve a buffer */ if (unlikely(!skb)) { for (i = 0; i < ena_rx_ctx.descs; i++) { - rx_ring->free_tx_ids[next_to_clean] = + rx_ring->free_ids[next_to_clean] = rx_ring->ena_bufs[i].req_id; next_to_clean = ENA_RX_RING_IDX_NEXT(next_to_clean, @@ -1148,7 +1155,7 @@ error: return 0; } -inline void ena_adjust_intr_moderation(struct ena_ring *rx_ring, +void ena_adjust_intr_moderation(struct ena_ring *rx_ring, struct ena_ring *tx_ring) { /* We apply adaptive moderation on Rx path only. @@ -1167,7 +1174,7 @@ inline void ena_adjust_intr_moderation(struct ena_ring *rx_ring, rx_ring->per_napi_bytes = 0; } -static inline void ena_unmask_interrupt(struct ena_ring *tx_ring, +static void ena_unmask_interrupt(struct ena_ring *tx_ring, struct ena_ring *rx_ring) { struct ena_eth_io_intr_reg intr_reg; @@ -1187,7 +1194,7 @@ static inline void ena_unmask_interrupt(struct ena_ring *tx_ring, ena_com_unmask_intr(rx_ring->ena_com_io_cq, &intr_reg); } -static inline void ena_update_ring_numa_node(struct ena_ring *tx_ring, +static void ena_update_ring_numa_node(struct ena_ring *tx_ring, struct ena_ring *rx_ring) { int cpu = get_cpu(); @@ -1630,7 +1637,7 @@ static int ena_create_io_tx_queue(struct ena_adapter *adapter, int qid) ctx.qid = ena_qid; ctx.mem_queue_type = ena_dev->tx_mem_queue_type; ctx.msix_vector = msix_vector; - ctx.queue_size = adapter->tx_ring_size; + ctx.queue_size = tx_ring->ring_size; ctx.numa_node = cpu_to_node(tx_ring->cpu); rc = ena_com_create_io_queue(ena_dev, &ctx); @@ -1697,7 +1704,7 @@ static int ena_create_io_rx_queue(struct ena_adapter *adapter, int qid) ctx.direction = ENA_COM_IO_QUEUE_DIRECTION_RX; ctx.mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST; ctx.msix_vector = msix_vector; - ctx.queue_size = adapter->rx_ring_size; + ctx.queue_size = rx_ring->ring_size; ctx.numa_node = cpu_to_node(rx_ring->cpu); rc = ena_com_create_io_queue(ena_dev, &ctx); @@ -1744,6 +1751,112 @@ create_err: return rc; } +static void set_io_rings_size(struct ena_adapter *adapter, + int new_tx_size, int new_rx_size) +{ + int i; + + for (i = 0; i < adapter->num_queues; i++) { + adapter->tx_ring[i].ring_size = new_tx_size; + adapter->rx_ring[i].ring_size = new_rx_size; + } +} + +/* This function allows queue allocation to backoff when the system is + * low on memory. If there is not enough memory to allocate io queues + * the driver will try to allocate smaller queues. + * + * The backoff algorithm is as follows: + * 1. Try to allocate TX and RX and if successful. + * 1.1. return success + * + * 2. Divide by 2 the size of the larger of RX and TX queues (or both if their size is the same). + * + * 3. If TX or RX is smaller than 256 + * 3.1. return failure. + * 4. else + * 4.1. go back to 1. + */ +static int create_queues_with_size_backoff(struct ena_adapter *adapter) +{ + int rc, cur_rx_ring_size, cur_tx_ring_size; + int new_rx_ring_size, new_tx_ring_size; + + /* current queue sizes might be set to smaller than the requested + * ones due to past queue allocation failures. + */ + set_io_rings_size(adapter, adapter->requested_tx_ring_size, + adapter->requested_rx_ring_size); + + while (1) { + rc = ena_setup_all_tx_resources(adapter); + if (rc) + goto err_setup_tx; + + rc = ena_create_all_io_tx_queues(adapter); + if (rc) + goto err_create_tx_queues; + + rc = ena_setup_all_rx_resources(adapter); + if (rc) + goto err_setup_rx; + + rc = ena_create_all_io_rx_queues(adapter); + if (rc) + goto err_create_rx_queues; + + return 0; + +err_create_rx_queues: + ena_free_all_io_rx_resources(adapter); +err_setup_rx: + ena_destroy_all_tx_queues(adapter); +err_create_tx_queues: + ena_free_all_io_tx_resources(adapter); +err_setup_tx: + if (rc != -ENOMEM) { + netif_err(adapter, ifup, adapter->netdev, + "Queue creation failed with error code %d\n", + rc); + return rc; + } + + cur_tx_ring_size = adapter->tx_ring[0].ring_size; + cur_rx_ring_size = adapter->rx_ring[0].ring_size; + + netif_err(adapter, ifup, adapter->netdev, + "Not enough memory to create queues with sizes TX=%d, RX=%d\n", + cur_tx_ring_size, cur_rx_ring_size); + + new_tx_ring_size = cur_tx_ring_size; + new_rx_ring_size = cur_rx_ring_size; + + /* Decrease the size of the larger queue, or + * decrease both if they are the same size. + */ + if (cur_rx_ring_size <= cur_tx_ring_size) + new_tx_ring_size = cur_tx_ring_size / 2; + if (cur_rx_ring_size >= cur_tx_ring_size) + new_rx_ring_size = cur_rx_ring_size / 2; + + if (new_tx_ring_size < ENA_MIN_RING_SIZE || + new_rx_ring_size < ENA_MIN_RING_SIZE) { + netif_err(adapter, ifup, adapter->netdev, + "Queue creation failed with the smallest possible queue size of %d for both queues. Not retrying with smaller queues\n", + ENA_MIN_RING_SIZE); + return rc; + } + + netif_err(adapter, ifup, adapter->netdev, + "Retrying queue creation with sizes TX=%d, RX=%d\n", + new_tx_ring_size, + new_rx_ring_size); + + set_io_rings_size(adapter, new_tx_ring_size, + new_rx_ring_size); + } +} + static int ena_up(struct ena_adapter *adapter) { int rc, i; @@ -1763,25 +1876,9 @@ static int ena_up(struct ena_adapter *adapter) if (rc) goto err_req_irq; - /* allocate transmit descriptors */ - rc = ena_setup_all_tx_resources(adapter); + rc = create_queues_with_size_backoff(adapter); if (rc) - goto err_setup_tx; - - /* allocate receive descriptors */ - rc = ena_setup_all_rx_resources(adapter); - if (rc) - goto err_setup_rx; - - /* Create TX queues */ - rc = ena_create_all_io_tx_queues(adapter); - if (rc) - goto err_create_tx_queues; - - /* Create RX queues */ - rc = ena_create_all_io_rx_queues(adapter); - if (rc) - goto err_create_rx_queues; + goto err_create_queues_with_backoff; rc = ena_up_complete(adapter); if (rc) @@ -1810,16 +1907,14 @@ static int ena_up(struct ena_adapter *adapter) return rc; err_up: - ena_destroy_all_rx_queues(adapter); -err_create_rx_queues: ena_destroy_all_tx_queues(adapter); -err_create_tx_queues: - ena_free_all_io_rx_resources(adapter); -err_setup_rx: ena_free_all_io_tx_resources(adapter); -err_setup_tx: + ena_destroy_all_rx_queues(adapter); + ena_free_all_io_rx_resources(adapter); +err_create_queues_with_backoff: ena_free_io_irq(adapter); err_req_irq: + ena_del_napi(adapter); return rc; } @@ -1936,6 +2031,20 @@ static int ena_close(struct net_device *netdev) return 0; } +int ena_update_queue_sizes(struct ena_adapter *adapter, + u32 new_tx_size, + u32 new_rx_size) +{ + bool dev_up; + + dev_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags); + ena_close(adapter->netdev); + adapter->requested_tx_ring_size = new_tx_size; + adapter->requested_rx_ring_size = new_rx_size; + ena_init_io_rings(adapter); + return dev_up ? ena_up(adapter) : 0; +} + static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx, struct sk_buff *skb) { u32 mss = skb_shinfo(skb)->gso_size; @@ -2146,7 +2255,7 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) skb_tx_timestamp(skb); next_to_use = tx_ring->next_to_use; - req_id = tx_ring->free_tx_ids[next_to_use]; + req_id = tx_ring->free_ids[next_to_use]; tx_info = &tx_ring->tx_buffer_info[req_id]; tx_info->num_of_bufs = 0; @@ -2166,6 +2275,13 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) /* set flags and meta data */ ena_tx_csum(&ena_tx_ctx, skb); + if (unlikely(ena_com_is_doorbell_needed(tx_ring->ena_com_io_sq, &ena_tx_ctx))) { + netif_dbg(adapter, tx_queued, dev, + "llq tx max burst size of queue %d achieved, writing doorbell to send burst\n", + qid); + ena_com_write_sq_doorbell(tx_ring->ena_com_io_sq); + } + /* prepare the packet's descriptors to dma engine */ rc = ena_com_prepare_tx(tx_ring->ena_com_io_sq, &ena_tx_ctx, &nb_hw_desc); @@ -2236,7 +2352,7 @@ static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev) } } - if (netif_xmit_stopped(txq) || !skb->xmit_more) { + if (netif_xmit_stopped(txq) || !netdev_xmit_more()) { /* trigger the dma engine. ena_com_write_sq_doorbell() * has a mb */ @@ -2258,8 +2374,7 @@ error_drop_packet: } static u16 ena_select_queue(struct net_device *dev, struct sk_buff *skb, - struct net_device *sb_dev, - select_queue_fallback_t fallback) + struct net_device *sb_dev) { u16 qid; /* we suspect that this is good for in--kernel network services that @@ -2269,7 +2384,7 @@ static u16 ena_select_queue(struct net_device *dev, struct sk_buff *skb, if (skb_rx_queue_recorded(skb)) qid = skb_get_rx_queue(skb); else - qid = fallback(dev, skb, NULL); + qid = netdev_pick_tx(dev, skb, NULL); return qid; } @@ -2292,7 +2407,7 @@ static void ena_config_host_info(struct ena_com_dev *ena_dev, host_info->bdf = (pdev->bus->number << 8) | pdev->devfn; host_info->os_type = ENA_ADMIN_OS_LINUX; host_info->kernel_ver = LINUX_VERSION_CODE; - strncpy(host_info->kernel_ver_str, utsname()->version, + strlcpy(host_info->kernel_ver_str, utsname()->version, sizeof(host_info->kernel_ver_str) - 1); host_info->os_dist = 0; strncpy(host_info->os_dist_str, utsname()->release, @@ -2442,13 +2557,6 @@ static int ena_device_validate_params(struct ena_adapter *adapter, return -EINVAL; } - if ((get_feat_ctx->max_queues.max_cq_num < adapter->num_queues) || - (get_feat_ctx->max_queues.max_sq_num < adapter->num_queues)) { - netif_err(adapter, drv, netdev, - "Error, device doesn't support enough queues\n"); - return -EINVAL; - } - if (get_feat_ctx->dev_attr.max_mtu < netdev->mtu) { netif_err(adapter, drv, netdev, "Error, device max mtu is smaller than netdev MTU\n"); @@ -2663,11 +2771,6 @@ static int ena_restore_device(struct ena_adapter *adapter) goto err_device_destroy; } - clear_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags); - /* Make sure we don't have a race with AENQ Links state handler */ - if (test_bit(ENA_FLAG_LINK_UP, &adapter->flags)) - netif_carrier_on(adapter->netdev); - rc = ena_enable_msix_and_set_admin_interrupts(adapter, adapter->num_queues); if (rc) { @@ -2684,6 +2787,11 @@ static int ena_restore_device(struct ena_adapter *adapter) } set_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags); + + clear_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags); + if (test_bit(ENA_FLAG_LINK_UP, &adapter->flags)) + netif_carrier_on(adapter->netdev); + mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ)); dev_err(&pdev->dev, "Device reset completed successfully, Driver info: %s\n", @@ -3022,18 +3130,32 @@ static int ena_calc_io_queue_num(struct pci_dev *pdev, struct ena_com_dev *ena_dev, struct ena_com_dev_get_features_ctx *get_feat_ctx) { - int io_sq_num, io_queue_num; + int io_tx_sq_num, io_tx_cq_num, io_rx_num, io_queue_num; - /* In case of LLQ use the llq number in the get feature cmd */ + if (ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) { + struct ena_admin_queue_ext_feature_fields *max_queue_ext = + &get_feat_ctx->max_queue_ext.max_queue_ext; + io_rx_num = min_t(int, max_queue_ext->max_rx_sq_num, + max_queue_ext->max_rx_cq_num); + + io_tx_sq_num = max_queue_ext->max_tx_sq_num; + io_tx_cq_num = max_queue_ext->max_tx_cq_num; + } else { + struct ena_admin_queue_feature_desc *max_queues = + &get_feat_ctx->max_queues; + io_tx_sq_num = max_queues->max_sq_num; + io_tx_cq_num = max_queues->max_cq_num; + io_rx_num = min_t(int, io_tx_sq_num, io_tx_cq_num); + } + + /* In case of LLQ use the llq fields for the tx SQ/CQ */ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) - io_sq_num = get_feat_ctx->llq.max_llq_num; - else - io_sq_num = get_feat_ctx->max_queues.max_sq_num; + io_tx_sq_num = get_feat_ctx->llq.max_llq_num; io_queue_num = min_t(int, num_online_cpus(), ENA_MAX_NUM_IO_QUEUES); - io_queue_num = min_t(int, io_queue_num, io_sq_num); - io_queue_num = min_t(int, io_queue_num, - get_feat_ctx->max_queues.max_cq_num); + io_queue_num = min_t(int, io_queue_num, io_rx_num); + io_queue_num = min_t(int, io_queue_num, io_tx_sq_num); + io_queue_num = min_t(int, io_queue_num, io_tx_cq_num); /* 1 IRQ for for mgmnt and 1 IRQs for each IO direction */ io_queue_num = min_t(int, io_queue_num, pci_msix_vec_count(pdev) - 1); if (unlikely(!io_queue_num)) { @@ -3207,7 +3329,7 @@ static void ena_release_bars(struct ena_com_dev *ena_dev, struct pci_dev *pdev) pci_release_selected_regions(pdev, release_bars); } -static inline void set_default_llq_configurations(struct ena_llq_configurations *llq_config) +static void set_default_llq_configurations(struct ena_llq_configurations *llq_config) { llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER; llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B; @@ -3216,36 +3338,70 @@ static inline void set_default_llq_configurations(struct ena_llq_configurations llq_config->llq_ring_entry_size_value = 128; } -static int ena_calc_queue_size(struct pci_dev *pdev, - struct ena_com_dev *ena_dev, - u16 *max_tx_sgl_size, - u16 *max_rx_sgl_size, - struct ena_com_dev_get_features_ctx *get_feat_ctx) +static int ena_calc_queue_size(struct ena_calc_queue_size_ctx *ctx) { - u32 queue_size = ENA_DEFAULT_RING_SIZE; + struct ena_admin_feature_llq_desc *llq = &ctx->get_feat_ctx->llq; + struct ena_com_dev *ena_dev = ctx->ena_dev; + u32 tx_queue_size = ENA_DEFAULT_RING_SIZE; + u32 rx_queue_size = ENA_DEFAULT_RING_SIZE; + u32 max_tx_queue_size; + u32 max_rx_queue_size; - queue_size = min_t(u32, queue_size, - get_feat_ctx->max_queues.max_cq_depth); - queue_size = min_t(u32, queue_size, - get_feat_ctx->max_queues.max_sq_depth); + if (ctx->ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) { + struct ena_admin_queue_ext_feature_fields *max_queue_ext = + &ctx->get_feat_ctx->max_queue_ext.max_queue_ext; + max_rx_queue_size = min_t(u32, max_queue_ext->max_rx_cq_depth, + max_queue_ext->max_rx_sq_depth); + max_tx_queue_size = max_queue_ext->max_tx_cq_depth; - if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) - queue_size = min_t(u32, queue_size, - get_feat_ctx->llq.max_llq_depth); + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + max_tx_queue_size = min_t(u32, max_tx_queue_size, + llq->max_llq_depth); + else + max_tx_queue_size = min_t(u32, max_tx_queue_size, + max_queue_ext->max_tx_sq_depth); - queue_size = rounddown_pow_of_two(queue_size); + ctx->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queue_ext->max_per_packet_tx_descs); + ctx->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queue_ext->max_per_packet_rx_descs); + } else { + struct ena_admin_queue_feature_desc *max_queues = + &ctx->get_feat_ctx->max_queues; + max_rx_queue_size = min_t(u32, max_queues->max_cq_depth, + max_queues->max_sq_depth); + max_tx_queue_size = max_queues->max_cq_depth; + + if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) + max_tx_queue_size = min_t(u32, max_tx_queue_size, + llq->max_llq_depth); + else + max_tx_queue_size = min_t(u32, max_tx_queue_size, + max_queues->max_sq_depth); - if (unlikely(!queue_size)) { - dev_err(&pdev->dev, "Invalid queue size\n"); - return -EFAULT; + ctx->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queues->max_packet_tx_descs); + ctx->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, + max_queues->max_packet_rx_descs); } - *max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, - get_feat_ctx->max_queues.max_packet_tx_descs); - *max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS, - get_feat_ctx->max_queues.max_packet_rx_descs); + max_tx_queue_size = rounddown_pow_of_two(max_tx_queue_size); + max_rx_queue_size = rounddown_pow_of_two(max_rx_queue_size); + + tx_queue_size = clamp_val(tx_queue_size, ENA_MIN_RING_SIZE, + max_tx_queue_size); + rx_queue_size = clamp_val(rx_queue_size, ENA_MIN_RING_SIZE, + max_rx_queue_size); + + tx_queue_size = rounddown_pow_of_two(tx_queue_size); + rx_queue_size = rounddown_pow_of_two(rx_queue_size); + + ctx->max_tx_queue_size = max_tx_queue_size; + ctx->max_rx_queue_size = max_rx_queue_size; + ctx->tx_queue_size = tx_queue_size; + ctx->rx_queue_size = rx_queue_size; - return queue_size; + return 0; } /* ena_probe - Device Initialization Routine @@ -3261,23 +3417,19 @@ static int ena_calc_queue_size(struct pci_dev *pdev, static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { struct ena_com_dev_get_features_ctx get_feat_ctx; - static int version_printed; - struct net_device *netdev; - struct ena_adapter *adapter; + struct ena_calc_queue_size_ctx calc_queue_ctx = { 0 }; struct ena_llq_configurations llq_config; struct ena_com_dev *ena_dev = NULL; - char *queue_type_str; - static int adapters_found; + struct ena_adapter *adapter; int io_queue_num, bars, rc; - int queue_size; - u16 tx_sgl_size = 0; - u16 rx_sgl_size = 0; + struct net_device *netdev; + static int adapters_found; + char *queue_type_str; bool wd_state; dev_dbg(&pdev->dev, "%s\n", __func__); - if (version_printed++ == 0) - dev_info(&pdev->dev, "%s", version); + dev_info_once(&pdev->dev, "%s", version); rc = pci_enable_device_mem(pdev); if (rc) { @@ -3329,20 +3481,25 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) goto err_device_destroy; } + calc_queue_ctx.ena_dev = ena_dev; + calc_queue_ctx.get_feat_ctx = &get_feat_ctx; + calc_queue_ctx.pdev = pdev; + /* initial Tx interrupt delay, Assumes 1 usec granularity. * Updated during device initialization with the real granularity */ ena_dev->intr_moder_tx_interval = ENA_INTR_INITIAL_TX_INTERVAL_USECS; io_queue_num = ena_calc_io_queue_num(pdev, ena_dev, &get_feat_ctx); - queue_size = ena_calc_queue_size(pdev, ena_dev, &tx_sgl_size, - &rx_sgl_size, &get_feat_ctx); - if ((queue_size <= 0) || (io_queue_num <= 0)) { + rc = ena_calc_queue_size(&calc_queue_ctx); + if (rc || io_queue_num <= 0) { rc = -EFAULT; goto err_device_destroy; } - dev_info(&pdev->dev, "creating %d io queues. queue size: %d. LLQ is %s\n", - io_queue_num, queue_size, + dev_info(&pdev->dev, "creating %d io queues. rx queue size: %d tx queue size. %d LLQ is %s\n", + io_queue_num, + calc_queue_ctx.rx_queue_size, + calc_queue_ctx.tx_queue_size, (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) ? "ENABLED" : "DISABLED"); @@ -3368,11 +3525,12 @@ static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent) adapter->msg_enable = netif_msg_init(debug, DEFAULT_MSG_ENABLE); adapter->reset_reason = ENA_REGS_RESET_NORMAL; - adapter->tx_ring_size = queue_size; - adapter->rx_ring_size = queue_size; - - adapter->max_tx_sgl_size = tx_sgl_size; - adapter->max_rx_sgl_size = rx_sgl_size; + adapter->requested_tx_ring_size = calc_queue_ctx.tx_queue_size; + adapter->requested_rx_ring_size = calc_queue_ctx.rx_queue_size; + adapter->max_tx_ring_size = calc_queue_ctx.max_tx_queue_size; + adapter->max_rx_ring_size = calc_queue_ctx.max_rx_queue_size; + adapter->max_tx_sgl_size = calc_queue_ctx.max_tx_sgl_size; + adapter->max_rx_sgl_size = calc_queue_ctx.max_rx_sgl_size; adapter->num_queues = io_queue_num; adapter->last_monitored_tx_qid = 0; |