diff options
Diffstat (limited to 'drivers/net/ethernet/sfc/rx.c')
-rw-r--r-- | drivers/net/ethernet/sfc/rx.c | 592 |
1 files changed, 1 insertions, 591 deletions
diff --git a/drivers/net/ethernet/sfc/rx.c b/drivers/net/ethernet/sfc/rx.c index c29bf862a94c..a2042f16babc 100644 --- a/drivers/net/ethernet/sfc/rx.c +++ b/drivers/net/ethernet/sfc/rx.c @@ -21,6 +21,7 @@ #include <linux/bpf_trace.h> #include "net_driver.h" #include "efx.h" +#include "rx_common.h" #include "filter.h" #include "nic.h" #include "selftest.h" @@ -32,60 +33,13 @@ /* Maximum rx prefix used by any architecture. */ #define EFX_MAX_RX_PREFIX_SIZE 16 -/* Number of RX buffers to recycle pages for. When creating the RX page recycle - * ring, this number is divided by the number of buffers per page to calculate - * the number of pages to store in the RX page recycle ring. - */ -#define EFX_RECYCLE_RING_SIZE_IOMMU 4096 -#define EFX_RECYCLE_RING_SIZE_NOIOMMU (2 * EFX_RX_PREFERRED_BATCH) - /* Size of buffer allocated for skb header area. */ #define EFX_SKB_HEADERS 128u -/* This is the percentage fill level below which new RX descriptors - * will be added to the RX descriptor ring. - */ -static unsigned int rx_refill_threshold; - /* Each packet can consume up to ceil(max_frame_len / buffer_size) buffers */ #define EFX_RX_MAX_FRAGS DIV_ROUND_UP(EFX_MAX_FRAME_LEN(EFX_MAX_MTU), \ EFX_RX_USR_BUF_SIZE) -/* - * RX maximum head room required. - * - * This must be at least 1 to prevent overflow, plus one packet-worth - * to allow pipelined receives. - */ -#define EFX_RXD_HEAD_ROOM (1 + EFX_RX_MAX_FRAGS) - -static inline u8 *efx_rx_buf_va(struct efx_rx_buffer *buf) -{ - return page_address(buf->page) + buf->page_offset; -} - -static inline u32 efx_rx_buf_hash(struct efx_nic *efx, const u8 *eh) -{ -#if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) - return __le32_to_cpup((const __le32 *)(eh + efx->rx_packet_hash_offset)); -#else - const u8 *data = eh + efx->rx_packet_hash_offset; - return (u32)data[0] | - (u32)data[1] << 8 | - (u32)data[2] << 16 | - (u32)data[3] << 24; -#endif -} - -static inline struct efx_rx_buffer * -efx_rx_buf_next(struct efx_rx_queue *rx_queue, struct efx_rx_buffer *rx_buf) -{ - if (unlikely(rx_buf == efx_rx_buffer(rx_queue, rx_queue->ptr_mask))) - return efx_rx_buffer(rx_queue, 0); - else - return rx_buf + 1; -} - static inline void efx_sync_rx_buffer(struct efx_nic *efx, struct efx_rx_buffer *rx_buf, unsigned int len) @@ -94,301 +48,6 @@ static inline void efx_sync_rx_buffer(struct efx_nic *efx, DMA_FROM_DEVICE); } -void efx_rx_config_page_split(struct efx_nic *efx) -{ - efx->rx_page_buf_step = ALIGN(efx->rx_dma_len + efx->rx_ip_align + - XDP_PACKET_HEADROOM, - EFX_RX_BUF_ALIGNMENT); - efx->rx_bufs_per_page = efx->rx_buffer_order ? 1 : - ((PAGE_SIZE - sizeof(struct efx_rx_page_state)) / - efx->rx_page_buf_step); - efx->rx_buffer_truesize = (PAGE_SIZE << efx->rx_buffer_order) / - efx->rx_bufs_per_page; - efx->rx_pages_per_batch = DIV_ROUND_UP(EFX_RX_PREFERRED_BATCH, - efx->rx_bufs_per_page); -} - -/* Check the RX page recycle ring for a page that can be reused. */ -static struct page *efx_reuse_page(struct efx_rx_queue *rx_queue) -{ - struct efx_nic *efx = rx_queue->efx; - struct page *page; - struct efx_rx_page_state *state; - unsigned index; - - index = rx_queue->page_remove & rx_queue->page_ptr_mask; - page = rx_queue->page_ring[index]; - if (page == NULL) - return NULL; - - rx_queue->page_ring[index] = NULL; - /* page_remove cannot exceed page_add. */ - if (rx_queue->page_remove != rx_queue->page_add) - ++rx_queue->page_remove; - - /* If page_count is 1 then we hold the only reference to this page. */ - if (page_count(page) == 1) { - ++rx_queue->page_recycle_count; - return page; - } else { - state = page_address(page); - dma_unmap_page(&efx->pci_dev->dev, state->dma_addr, - PAGE_SIZE << efx->rx_buffer_order, - DMA_FROM_DEVICE); - put_page(page); - ++rx_queue->page_recycle_failed; - } - - return NULL; -} - -/** - * efx_init_rx_buffers - create EFX_RX_BATCH page-based RX buffers - * - * @rx_queue: Efx RX queue - * - * This allocates a batch of pages, maps them for DMA, and populates - * struct efx_rx_buffers for each one. Return a negative error code or - * 0 on success. If a single page can be used for multiple buffers, - * then the page will either be inserted fully, or not at all. - */ -static int efx_init_rx_buffers(struct efx_rx_queue *rx_queue, bool atomic) -{ - struct efx_nic *efx = rx_queue->efx; - struct efx_rx_buffer *rx_buf; - struct page *page; - unsigned int page_offset; - struct efx_rx_page_state *state; - dma_addr_t dma_addr; - unsigned index, count; - - count = 0; - do { - page = efx_reuse_page(rx_queue); - if (page == NULL) { - page = alloc_pages(__GFP_COMP | - (atomic ? GFP_ATOMIC : GFP_KERNEL), - efx->rx_buffer_order); - if (unlikely(page == NULL)) - return -ENOMEM; - dma_addr = - dma_map_page(&efx->pci_dev->dev, page, 0, - PAGE_SIZE << efx->rx_buffer_order, - DMA_FROM_DEVICE); - if (unlikely(dma_mapping_error(&efx->pci_dev->dev, - dma_addr))) { - __free_pages(page, efx->rx_buffer_order); - return -EIO; - } - state = page_address(page); - state->dma_addr = dma_addr; - } else { - state = page_address(page); - dma_addr = state->dma_addr; - } - - dma_addr += sizeof(struct efx_rx_page_state); - page_offset = sizeof(struct efx_rx_page_state); - - do { - index = rx_queue->added_count & rx_queue->ptr_mask; - rx_buf = efx_rx_buffer(rx_queue, index); - rx_buf->dma_addr = dma_addr + efx->rx_ip_align + - XDP_PACKET_HEADROOM; - rx_buf->page = page; - rx_buf->page_offset = page_offset + efx->rx_ip_align + - XDP_PACKET_HEADROOM; - rx_buf->len = efx->rx_dma_len; - rx_buf->flags = 0; - ++rx_queue->added_count; - get_page(page); - dma_addr += efx->rx_page_buf_step; - page_offset += efx->rx_page_buf_step; - } while (page_offset + efx->rx_page_buf_step <= PAGE_SIZE); - - rx_buf->flags = EFX_RX_BUF_LAST_IN_PAGE; - } while (++count < efx->rx_pages_per_batch); - - return 0; -} - -/* Unmap a DMA-mapped page. This function is only called for the final RX - * buffer in a page. - */ -static void efx_unmap_rx_buffer(struct efx_nic *efx, - struct efx_rx_buffer *rx_buf) -{ - struct page *page = rx_buf->page; - - if (page) { - struct efx_rx_page_state *state = page_address(page); - dma_unmap_page(&efx->pci_dev->dev, - state->dma_addr, - PAGE_SIZE << efx->rx_buffer_order, - DMA_FROM_DEVICE); - } -} - -static void efx_free_rx_buffers(struct efx_rx_queue *rx_queue, - struct efx_rx_buffer *rx_buf, - unsigned int num_bufs) -{ - do { - if (rx_buf->page) { - put_page(rx_buf->page); - rx_buf->page = NULL; - } - rx_buf = efx_rx_buf_next(rx_queue, rx_buf); - } while (--num_bufs); -} - -/* Attempt to recycle the page if there is an RX recycle ring; the page can - * only be added if this is the final RX buffer, to prevent pages being used in - * the descriptor ring and appearing in the recycle ring simultaneously. - */ -static void efx_recycle_rx_page(struct efx_channel *channel, - struct efx_rx_buffer *rx_buf) -{ - struct page *page = rx_buf->page; - struct efx_rx_queue *rx_queue = efx_channel_get_rx_queue(channel); - struct efx_nic *efx = rx_queue->efx; - unsigned index; - - /* Only recycle the page after processing the final buffer. */ - if (!(rx_buf->flags & EFX_RX_BUF_LAST_IN_PAGE)) - return; - - index = rx_queue->page_add & rx_queue->page_ptr_mask; - if (rx_queue->page_ring[index] == NULL) { - unsigned read_index = rx_queue->page_remove & - rx_queue->page_ptr_mask; - - /* The next slot in the recycle ring is available, but - * increment page_remove if the read pointer currently - * points here. - */ - if (read_index == index) - ++rx_queue->page_remove; - rx_queue->page_ring[index] = page; - ++rx_queue->page_add; - return; - } - ++rx_queue->page_recycle_full; - efx_unmap_rx_buffer(efx, rx_buf); - put_page(rx_buf->page); -} - -static void efx_fini_rx_buffer(struct efx_rx_queue *rx_queue, - struct efx_rx_buffer *rx_buf) -{ - /* Release the page reference we hold for the buffer. */ - if (rx_buf->page) - put_page(rx_buf->page); - - /* If this is the last buffer in a page, unmap and free it. */ - if (rx_buf->flags & EFX_RX_BUF_LAST_IN_PAGE) { - efx_unmap_rx_buffer(rx_queue->efx, rx_buf); - efx_free_rx_buffers(rx_queue, rx_buf, 1); - } - rx_buf->page = NULL; -} - -/* Recycle the pages that are used by buffers that have just been received. */ -static void efx_recycle_rx_pages(struct efx_channel *channel, - struct efx_rx_buffer *rx_buf, - unsigned int n_frags) -{ - struct efx_rx_queue *rx_queue = efx_channel_get_rx_queue(channel); - - do { - efx_recycle_rx_page(channel, rx_buf); - rx_buf = efx_rx_buf_next(rx_queue, rx_buf); - } while (--n_frags); -} - -static void efx_discard_rx_packet(struct efx_channel *channel, - struct efx_rx_buffer *rx_buf, - unsigned int n_frags) -{ - struct efx_rx_queue *rx_queue = efx_channel_get_rx_queue(channel); - - efx_recycle_rx_pages(channel, rx_buf, n_frags); - - efx_free_rx_buffers(rx_queue, rx_buf, n_frags); -} - -/** - * efx_fast_push_rx_descriptors - push new RX descriptors quickly - * @rx_queue: RX descriptor queue - * - * This will aim to fill the RX descriptor queue up to - * @rx_queue->@max_fill. If there is insufficient atomic - * memory to do so, a slow fill will be scheduled. - * - * The caller must provide serialisation (none is used here). In practise, - * this means this function must run from the NAPI handler, or be called - * when NAPI is disabled. - */ -void efx_fast_push_rx_descriptors(struct efx_rx_queue *rx_queue, bool atomic) -{ - struct efx_nic *efx = rx_queue->efx; - unsigned int fill_level, batch_size; - int space, rc = 0; - - if (!rx_queue->refill_enabled) - return; - - /* Calculate current fill level, and exit if we don't need to fill */ - fill_level = (rx_queue->added_count - rx_queue->removed_count); - EFX_WARN_ON_ONCE_PARANOID(fill_level > rx_queue->efx->rxq_entries); - if (fill_level >= rx_queue->fast_fill_trigger) - goto out; - - /* Record minimum fill level */ - if (unlikely(fill_level < rx_queue->min_fill)) { - if (fill_level) - rx_queue->min_fill = fill_level; - } - - batch_size = efx->rx_pages_per_batch * efx->rx_bufs_per_page; - space = rx_queue->max_fill - fill_level; - EFX_WARN_ON_ONCE_PARANOID(space < batch_size); - - netif_vdbg(rx_queue->efx, rx_status, rx_queue->efx->net_dev, - "RX queue %d fast-filling descriptor ring from" - " level %d to level %d\n", - efx_rx_queue_index(rx_queue), fill_level, - rx_queue->max_fill); - - - do { - rc = efx_init_rx_buffers(rx_queue, atomic); - if (unlikely(rc)) { - /* Ensure that we don't leave the rx queue empty */ - efx_schedule_slow_fill(rx_queue); - goto out; - } - } while ((space -= batch_size) >= batch_size); - - netif_vdbg(rx_queue->efx, rx_status, rx_queue->efx->net_dev, - "RX queue %d fast-filled descriptor ring " - "to level %d\n", efx_rx_queue_index(rx_queue), - rx_queue->added_count - rx_queue->removed_count); - - out: - if (rx_queue->notified_count != rx_queue->added_count) - efx_nic_notify_rx_desc(rx_queue); -} - -void efx_rx_slow_fill(struct timer_list *t) -{ - struct efx_rx_queue *rx_queue = from_timer(rx_queue, t, slow_fill); - - /* Post an event to cause NAPI to run and refill the queue */ - efx_nic_generate_fill_event(rx_queue); - ++rx_queue->slow_fill_count; -} - static void efx_rx_packet__check_len(struct efx_rx_queue *rx_queue, struct efx_rx_buffer *rx_buf, int len) @@ -412,53 +71,6 @@ static void efx_rx_packet__check_len(struct efx_rx_queue *rx_queue, efx_rx_queue_channel(rx_queue)->n_rx_overlength++; } -/* Pass a received packet up through GRO. GRO can handle pages - * regardless of checksum state and skbs with a good checksum. - */ -static void -efx_rx_packet_gro(struct efx_channel *channel, struct efx_rx_buffer *rx_buf, - unsigned int n_frags, u8 *eh) -{ - struct napi_struct *napi = &channel->napi_str; - struct efx_nic *efx = channel->efx; - struct sk_buff *skb; - - skb = napi_get_frags(napi); - if (unlikely(!skb)) { - struct efx_rx_queue *rx_queue; - - rx_queue = efx_channel_get_rx_queue(channel); - efx_free_rx_buffers(rx_queue, rx_buf, n_frags); - return; - } - - if (efx->net_dev->features & NETIF_F_RXHASH) - skb_set_hash(skb, efx_rx_buf_hash(efx, eh), - PKT_HASH_TYPE_L3); - skb->ip_summed = ((rx_buf->flags & EFX_RX_PKT_CSUMMED) ? - CHECKSUM_UNNECESSARY : CHECKSUM_NONE); - skb->csum_level = !!(rx_buf->flags & EFX_RX_PKT_CSUM_LEVEL); - - for (;;) { - skb_fill_page_desc(skb, skb_shinfo(skb)->nr_frags, - rx_buf->page, rx_buf->page_offset, - rx_buf->len); - rx_buf->page = NULL; - skb->len += rx_buf->len; - if (skb_shinfo(skb)->nr_frags == n_frags) - break; - - rx_buf = efx_rx_buf_next(&channel->rx_queue, rx_buf); - } - - skb->data_len = skb->len; - skb->truesize += n_frags * efx->rx_buffer_truesize; - - skb_record_rx_queue(skb, channel->rx_queue.core_index); - - napi_gro_frags(napi); -} - /* Allocate and construct an SKB around page fragments */ static struct sk_buff *efx_rx_mk_skb(struct efx_channel *channel, struct efx_rx_buffer *rx_buf, @@ -805,174 +417,6 @@ out: channel->rx_pkt_n_frags = 0; } -int efx_probe_rx_queue(struct efx_rx_queue *rx_queue) -{ - struct efx_nic *efx = rx_queue->efx; - unsigned int entries; - int rc; - - /* Create the smallest power-of-two aligned ring */ - entries = max(roundup_pow_of_two(efx->rxq_entries), EFX_MIN_DMAQ_SIZE); - EFX_WARN_ON_PARANOID(entries > EFX_MAX_DMAQ_SIZE); - rx_queue->ptr_mask = entries - 1; - - netif_dbg(efx, probe, efx->net_dev, - "creating RX queue %d size %#x mask %#x\n", - efx_rx_queue_index(rx_queue), efx->rxq_entries, - rx_queue->ptr_mask); - - /* Allocate RX buffers */ - rx_queue->buffer = kcalloc(entries, sizeof(*rx_queue->buffer), - GFP_KERNEL); - if (!rx_queue->buffer) - return -ENOMEM; - - rc = efx_nic_probe_rx(rx_queue); - if (rc) { - kfree(rx_queue->buffer); - rx_queue->buffer = NULL; - } - - return rc; -} - -static void efx_init_rx_recycle_ring(struct efx_nic *efx, - struct efx_rx_queue *rx_queue) -{ - unsigned int bufs_in_recycle_ring, page_ring_size; - - /* Set the RX recycle ring size */ -#ifdef CONFIG_PPC64 - bufs_in_recycle_ring = EFX_RECYCLE_RING_SIZE_IOMMU; -#else - if (iommu_present(&pci_bus_type)) - bufs_in_recycle_ring = EFX_RECYCLE_RING_SIZE_IOMMU; - else - bufs_in_recycle_ring = EFX_RECYCLE_RING_SIZE_NOIOMMU; -#endif /* CONFIG_PPC64 */ - - page_ring_size = roundup_pow_of_two(bufs_in_recycle_ring / - efx->rx_bufs_per_page); - rx_queue->page_ring = kcalloc(page_ring_size, - sizeof(*rx_queue->page_ring), GFP_KERNEL); - rx_queue->page_ptr_mask = page_ring_size - 1; -} - -void efx_init_rx_queue(struct efx_rx_queue *rx_queue) -{ - struct efx_nic *efx = rx_queue->efx; - unsigned int max_fill, trigger, max_trigger; - int rc = 0; - - netif_dbg(rx_queue->efx, drv, rx_queue->efx->net_dev, - "initialising RX queue %d\n", efx_rx_queue_index(rx_queue)); - - /* Initialise ptr fields */ - rx_queue->added_count = 0; - rx_queue->notified_count = 0; - rx_queue->removed_count = 0; - rx_queue->min_fill = -1U; - efx_init_rx_recycle_ring(efx, rx_queue); - - rx_queue->page_remove = 0; - rx_queue->page_add = rx_queue->page_ptr_mask + 1; - rx_queue->page_recycle_count = 0; - rx_queue->page_recycle_failed = 0; - rx_queue->page_recycle_full = 0; - - /* Initialise limit fields */ - max_fill = efx->rxq_entries - EFX_RXD_HEAD_ROOM; - max_trigger = - max_fill - efx->rx_pages_per_batch * efx->rx_bufs_per_page; - if (rx_refill_threshold != 0) { - trigger = max_fill * min(rx_refill_threshold, 100U) / 100U; - if (trigger > max_trigger) - trigger = max_trigger; - } else { - trigger = max_trigger; - } - - rx_queue->max_fill = max_fill; - rx_queue->fast_fill_trigger = trigger; - rx_queue->refill_enabled = true; - - /* Initialise XDP queue information */ - rc = xdp_rxq_info_reg(&rx_queue->xdp_rxq_info, efx->net_dev, - rx_queue->core_index); - - if (rc) { - netif_err(efx, rx_err, efx->net_dev, - "Failure to initialise XDP queue information rc=%d\n", - rc); - efx->xdp_rxq_info_failed = true; - } else { - rx_queue->xdp_rxq_info_valid = true; - } - - /* Set up RX descriptor ring */ - efx_nic_init_rx(rx_queue); -} - -void efx_fini_rx_queue(struct efx_rx_queue *rx_queue) -{ - int i; - struct efx_nic *efx = rx_queue->efx; - struct efx_rx_buffer *rx_buf; - - netif_dbg(rx_queue->efx, drv, rx_queue->efx->net_dev, - "shutting down RX queue %d\n", efx_rx_queue_index(rx_queue)); - - del_timer_sync(&rx_queue->slow_fill); - - /* Release RX buffers from the current read ptr to the write ptr */ - if (rx_queue->buffer) { - for (i = rx_queue->removed_count; i < rx_queue->added_count; - i++) { - unsigned index = i & rx_queue->ptr_mask; - rx_buf = efx_rx_buffer(rx_queue, index); - efx_fini_rx_buffer(rx_queue, rx_buf); - } - } - - /* Unmap and release the pages in the recycle ring. Remove the ring. */ - for (i = 0; i <= rx_queue->page_ptr_mask; i++) { - struct page *page = rx_queue->page_ring[i]; - struct efx_rx_page_state *state; - - if (page == NULL) - continue; - - state = page_address(page); - dma_unmap_page(&efx->pci_dev->dev, state->dma_addr, - PAGE_SIZE << efx->rx_buffer_order, - DMA_FROM_DEVICE); - put_page(page); - } - kfree(rx_queue->page_ring); - rx_queue->page_ring = NULL; - - if (rx_queue->xdp_rxq_info_valid) - xdp_rxq_info_unreg(&rx_queue->xdp_rxq_info); - - rx_queue->xdp_rxq_info_valid = false; -} - -void efx_remove_rx_queue(struct efx_rx_queue *rx_queue) -{ - netif_dbg(rx_queue->efx, drv, rx_queue->efx->net_dev, - "destroying RX queue %d\n", efx_rx_queue_index(rx_queue)); - - efx_nic_remove_rx(rx_queue); - - kfree(rx_queue->buffer); - rx_queue->buffer = NULL; -} - - -module_param(rx_refill_threshold, uint, 0444); -MODULE_PARM_DESC(rx_refill_threshold, - "RX descriptor ring refill threshold (%)"); - #ifdef CONFIG_RFS_ACCEL static void efx_filter_rfs_work(struct work_struct *data) @@ -1206,37 +650,3 @@ bool __efx_filter_rfs_expire(struct efx_channel *channel, unsigned int quota) } #endif /* CONFIG_RFS_ACCEL */ - -/** - * efx_filter_is_mc_recipient - test whether spec is a multicast recipient - * @spec: Specification to test - * - * Return: %true if the specification is a non-drop RX filter that - * matches a local MAC address I/G bit value of 1 or matches a local - * IPv4 or IPv6 address value in the respective multicast address - * range. Otherwise %false. - */ -bool efx_filter_is_mc_recipient(const struct efx_filter_spec *spec) -{ - if (!(spec->flags & EFX_FILTER_FLAG_RX) || - spec->dmaq_id == EFX_FILTER_RX_DMAQ_ID_DROP) - return false; - - if (spec->match_flags & - (EFX_FILTER_MATCH_LOC_MAC | EFX_FILTER_MATCH_LOC_MAC_IG) && - is_multicast_ether_addr(spec->loc_mac)) - return true; - - if ((spec->match_flags & - (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_LOC_HOST)) == - (EFX_FILTER_MATCH_ETHER_TYPE | EFX_FILTER_MATCH_LOC_HOST)) { - if (spec->ether_type == htons(ETH_P_IP) && - ipv4_is_multicast(spec->loc_host[0])) - return true; - if (spec->ether_type == htons(ETH_P_IPV6) && - ((const u8 *)spec->loc_host)[0] == 0xff) - return true; - } - - return false; -} |