From: Alexander Duyck Date: Tue, 27 Oct 2009 15:49:27 +0000 (+0000) Subject: igb: add new data structure for handling interrupts and NAPI X-Git-Tag: v2.6.33-rc1~388^2~698 X-Git-Url: https://openfabrics.org/gitweb/?a=commitdiff_plain;h=047e0030f1e601233ae5b03910602ec93c620bce;p=~emulex%2Finfiniband.git igb: add new data structure for handling interrupts and NAPI Add a new igb_q_vector data structure to handle interrupts and NAPI. This helps to abstract the rings away from the adapter struct. In addition it allows for a bit of consolidation since a tx and rx ring can share a q_vector. Signed-off-by: Alexander Duyck Signed-off-by: Jeff Kirsher Signed-off-by: David S. Miller --- diff --git a/drivers/net/igb/igb.h b/drivers/net/igb/igb.h index b805b1c63f8..86492c8957e 100644 --- a/drivers/net/igb/igb.h +++ b/drivers/net/igb/igb.h @@ -55,6 +55,8 @@ struct igb_adapter; #define IGB_DEFAULT_ITR 3 /* dynamic */ #define IGB_MAX_ITR_USECS 10000 #define IGB_MIN_ITR_USECS 10 +#define NON_Q_VECTORS 1 +#define MAX_Q_VECTORS 8 /* Transmit and receive queues */ #define IGB_MAX_RX_QUEUES (adapter->vfs_allocated_count ? \ @@ -149,25 +151,38 @@ struct igb_rx_queue_stats { u64 drops; }; -struct igb_ring { +struct igb_q_vector { struct igb_adapter *adapter; /* backlink */ - void *desc; /* descriptor ring memory */ - dma_addr_t dma; /* phys address of the ring */ - unsigned int size; /* length of desc. ring in bytes */ - unsigned int count; /* number of desc. in the ring */ + struct igb_ring *rx_ring; + struct igb_ring *tx_ring; + struct napi_struct napi; + + u32 eims_value; + u16 cpu; + + u16 itr_val; + u8 set_itr; + u8 itr_shift; + void __iomem *itr_register; + + char name[IFNAMSIZ + 9]; +}; + +struct igb_ring { + struct igb_q_vector *q_vector; /* backlink to q_vector */ + void *desc; /* descriptor ring memory */ + dma_addr_t dma; /* phys address of the ring */ + unsigned int size; /* length of desc. ring in bytes */ + unsigned int count; /* number of desc. in the ring */ u16 next_to_use; u16 next_to_clean; u16 head; u16 tail; struct igb_buffer *buffer_info; /* array of buffer info structs */ - u32 eims_value; - u32 itr_val; - u16 itr_register; - u16 cpu; + u8 queue_index; + u8 reg_idx; - u16 queue_index; - u16 reg_idx; unsigned int total_bytes; unsigned int total_packets; @@ -181,13 +196,8 @@ struct igb_ring { struct { struct igb_rx_queue_stats rx_stats; u64 rx_queue_drops; - struct napi_struct napi; - int set_itr; - struct igb_ring *buddy; }; }; - - char name[IFNAMSIZ + 5]; }; #define E1000_RX_DESC_ADV(R, i) \ @@ -254,7 +264,6 @@ struct igb_adapter { /* OS defined structs */ struct net_device *netdev; - struct napi_struct napi; struct pci_dev *pdev; struct cyclecounter cycles; struct timecounter clock; @@ -272,6 +281,9 @@ struct igb_adapter { struct igb_ring test_rx_ring; int msg_enable; + + unsigned int num_q_vectors; + struct igb_q_vector *q_vector[MAX_Q_VECTORS]; struct msix_entry *msix_entries; u32 eims_enable_mask; u32 eims_other; diff --git a/drivers/net/igb/igb_ethtool.c b/drivers/net/igb/igb_ethtool.c index dafb25bfd9e..f71276fec3f 100644 --- a/drivers/net/igb/igb_ethtool.c +++ b/drivers/net/igb/igb_ethtool.c @@ -1907,7 +1907,6 @@ static int igb_set_coalesce(struct net_device *netdev, struct ethtool_coalesce *ec) { struct igb_adapter *adapter = netdev_priv(netdev); - struct e1000_hw *hw = &adapter->hw; int i; if ((ec->rx_coalesce_usecs > IGB_MAX_ITR_USECS) || @@ -1925,8 +1924,11 @@ static int igb_set_coalesce(struct net_device *netdev, adapter->itr = adapter->itr_setting; } - for (i = 0; i < adapter->num_rx_queues; i++) - wr32(adapter->rx_ring[i].itr_register, adapter->itr); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + q_vector->itr_val = adapter->itr; + q_vector->set_itr = 1; + } return 0; } diff --git a/drivers/net/igb/igb_main.c b/drivers/net/igb/igb_main.c index 2ffe0997b83..c15eb4c3916 100644 --- a/drivers/net/igb/igb_main.c +++ b/drivers/net/igb/igb_main.c @@ -111,16 +111,14 @@ static void igb_set_uta(struct igb_adapter *adapter); static irqreturn_t igb_intr(int irq, void *); static irqreturn_t igb_intr_msi(int irq, void *); static irqreturn_t igb_msix_other(int irq, void *); -static irqreturn_t igb_msix_rx(int irq, void *); -static irqreturn_t igb_msix_tx(int irq, void *); +static irqreturn_t igb_msix_ring(int irq, void *); #ifdef CONFIG_IGB_DCA -static void igb_update_rx_dca(struct igb_ring *); -static void igb_update_tx_dca(struct igb_ring *); +static void igb_update_dca(struct igb_q_vector *); static void igb_setup_dca(struct igb_adapter *); #endif /* CONFIG_IGB_DCA */ -static bool igb_clean_tx_irq(struct igb_ring *); +static bool igb_clean_tx_irq(struct igb_q_vector *); static int igb_poll(struct napi_struct *, int); -static bool igb_clean_rx_irq_adv(struct igb_ring *, int *, int); +static bool igb_clean_rx_irq_adv(struct igb_q_vector *, int *, int); static void igb_alloc_rx_buffers_adv(struct igb_ring *, int); static int igb_ioctl(struct net_device *, struct ifreq *, int cmd); static void igb_tx_timeout(struct net_device *); @@ -374,7 +372,7 @@ module_exit(igb_exit_module); static void igb_cache_ring_register(struct igb_adapter *adapter) { int i; - unsigned int rbase_offset = adapter->vfs_allocated_count; + u32 rbase_offset = adapter->vfs_allocated_count; switch (adapter->hw.mac.type) { case e1000_82576: @@ -400,6 +398,18 @@ static void igb_cache_ring_register(struct igb_adapter *adapter) } } +static void igb_free_queues(struct igb_adapter *adapter) +{ + kfree(adapter->tx_ring); + kfree(adapter->rx_ring); + + adapter->tx_ring = NULL; + adapter->rx_ring = NULL; + + adapter->num_rx_queues = 0; + adapter->num_tx_queues = 0; +} + /** * igb_alloc_queues - Allocate memory for all rings * @adapter: board private structure to initialize @@ -414,59 +424,48 @@ static int igb_alloc_queues(struct igb_adapter *adapter) adapter->tx_ring = kcalloc(adapter->num_tx_queues, sizeof(struct igb_ring), GFP_KERNEL); if (!adapter->tx_ring) - return -ENOMEM; + goto err; adapter->rx_ring = kcalloc(adapter->num_rx_queues, sizeof(struct igb_ring), GFP_KERNEL); - if (!adapter->rx_ring) { - kfree(adapter->tx_ring); - return -ENOMEM; - } - - adapter->rx_ring->buddy = adapter->tx_ring; + if (!adapter->rx_ring) + goto err; for (i = 0; i < adapter->num_tx_queues; i++) { struct igb_ring *ring = &(adapter->tx_ring[i]); ring->count = adapter->tx_ring_count; - ring->adapter = adapter; ring->queue_index = i; } for (i = 0; i < adapter->num_rx_queues; i++) { struct igb_ring *ring = &(adapter->rx_ring[i]); ring->count = adapter->rx_ring_count; - ring->adapter = adapter; ring->queue_index = i; - ring->itr_register = E1000_ITR; - - /* set a default napi handler for each rx_ring */ - netif_napi_add(adapter->netdev, &ring->napi, igb_poll, 64); } igb_cache_ring_register(adapter); - return 0; -} -static void igb_free_queues(struct igb_adapter *adapter) -{ - int i; - - for (i = 0; i < adapter->num_rx_queues; i++) - netif_napi_del(&adapter->rx_ring[i].napi); + return 0; - adapter->num_rx_queues = 0; - adapter->num_tx_queues = 0; +err: + igb_free_queues(adapter); - kfree(adapter->tx_ring); - kfree(adapter->rx_ring); + return -ENOMEM; } #define IGB_N0_QUEUE -1 -static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue, - int tx_queue, int msix_vector) +static void igb_assign_vector(struct igb_q_vector *q_vector, int msix_vector) { u32 msixbm = 0; + struct igb_adapter *adapter = q_vector->adapter; struct e1000_hw *hw = &adapter->hw; u32 ivar, index; + int rx_queue = IGB_N0_QUEUE; + int tx_queue = IGB_N0_QUEUE; + + if (q_vector->rx_ring) + rx_queue = q_vector->rx_ring->reg_idx; + if (q_vector->tx_ring) + tx_queue = q_vector->tx_ring->reg_idx; switch (hw->mac.type) { case e1000_82575: @@ -474,16 +473,12 @@ static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue, bitmask for the EICR/EIMS/EIMC registers. To assign one or more queues to a vector, we write the appropriate bits into the MSIXBM register for that vector. */ - if (rx_queue > IGB_N0_QUEUE) { + if (rx_queue > IGB_N0_QUEUE) msixbm = E1000_EICR_RX_QUEUE0 << rx_queue; - adapter->rx_ring[rx_queue].eims_value = msixbm; - } - if (tx_queue > IGB_N0_QUEUE) { + if (tx_queue > IGB_N0_QUEUE) msixbm |= E1000_EICR_TX_QUEUE0 << tx_queue; - adapter->tx_ring[tx_queue].eims_value = - E1000_EICR_TX_QUEUE0 << tx_queue; - } array_wr32(E1000_MSIXBM(0), msix_vector, msixbm); + q_vector->eims_value = msixbm; break; case e1000_82576: /* 82576 uses a table-based method for assigning vectors. @@ -491,35 +486,34 @@ static void igb_assign_vector(struct igb_adapter *adapter, int rx_queue, a vector number along with a "valid" bit. Sadly, the layout of the table is somewhat counterintuitive. */ if (rx_queue > IGB_N0_QUEUE) { - index = (rx_queue >> 1) + adapter->vfs_allocated_count; + index = (rx_queue & 0x7); ivar = array_rd32(E1000_IVAR0, index); - if (rx_queue & 0x1) { - /* vector goes into third byte of register */ - ivar = ivar & 0xFF00FFFF; - ivar |= (msix_vector | E1000_IVAR_VALID) << 16; - } else { + if (rx_queue < 8) { /* vector goes into low byte of register */ ivar = ivar & 0xFFFFFF00; ivar |= msix_vector | E1000_IVAR_VALID; + } else { + /* vector goes into third byte of register */ + ivar = ivar & 0xFF00FFFF; + ivar |= (msix_vector | E1000_IVAR_VALID) << 16; } - adapter->rx_ring[rx_queue].eims_value= 1 << msix_vector; array_wr32(E1000_IVAR0, index, ivar); } if (tx_queue > IGB_N0_QUEUE) { - index = (tx_queue >> 1) + adapter->vfs_allocated_count; + index = (tx_queue & 0x7); ivar = array_rd32(E1000_IVAR0, index); - if (tx_queue & 0x1) { - /* vector goes into high byte of register */ - ivar = ivar & 0x00FFFFFF; - ivar |= (msix_vector | E1000_IVAR_VALID) << 24; - } else { + if (tx_queue < 8) { /* vector goes into second byte of register */ ivar = ivar & 0xFFFF00FF; ivar |= (msix_vector | E1000_IVAR_VALID) << 8; + } else { + /* vector goes into high byte of register */ + ivar = ivar & 0x00FFFFFF; + ivar |= (msix_vector | E1000_IVAR_VALID) << 24; } - adapter->tx_ring[tx_queue].eims_value= 1 << msix_vector; array_wr32(E1000_IVAR0, index, ivar); } + q_vector->eims_value = 1 << msix_vector; break; default: BUG(); @@ -540,43 +534,10 @@ static void igb_configure_msix(struct igb_adapter *adapter) struct e1000_hw *hw = &adapter->hw; adapter->eims_enable_mask = 0; - if (hw->mac.type == e1000_82576) - /* Turn on MSI-X capability first, or our settings - * won't stick. And it will take days to debug. */ - wr32(E1000_GPIE, E1000_GPIE_MSIX_MODE | - E1000_GPIE_PBA | E1000_GPIE_EIAME | - E1000_GPIE_NSICR); - - for (i = 0; i < adapter->num_tx_queues; i++) { - struct igb_ring *tx_ring = &adapter->tx_ring[i]; - igb_assign_vector(adapter, IGB_N0_QUEUE, i, vector++); - adapter->eims_enable_mask |= tx_ring->eims_value; - if (tx_ring->itr_val) - writel(tx_ring->itr_val, - hw->hw_addr + tx_ring->itr_register); - else - writel(1, hw->hw_addr + tx_ring->itr_register); - } - - for (i = 0; i < adapter->num_rx_queues; i++) { - struct igb_ring *rx_ring = &adapter->rx_ring[i]; - rx_ring->buddy = NULL; - igb_assign_vector(adapter, i, IGB_N0_QUEUE, vector++); - adapter->eims_enable_mask |= rx_ring->eims_value; - if (rx_ring->itr_val) - writel(rx_ring->itr_val, - hw->hw_addr + rx_ring->itr_register); - else - writel(1, hw->hw_addr + rx_ring->itr_register); - } - /* set vector for other causes, i.e. link changes */ switch (hw->mac.type) { case e1000_82575: - array_wr32(E1000_MSIXBM(0), vector++, - E1000_EIMS_OTHER); - tmp = rd32(E1000_CTRL_EXT); /* enable MSI-X PBA support*/ tmp |= E1000_CTRL_EXT_PBA_CLR; @@ -586,22 +547,40 @@ static void igb_configure_msix(struct igb_adapter *adapter) tmp |= E1000_CTRL_EXT_IRCA; wr32(E1000_CTRL_EXT, tmp); - adapter->eims_enable_mask |= E1000_EIMS_OTHER; + + /* enable msix_other interrupt */ + array_wr32(E1000_MSIXBM(0), vector++, + E1000_EIMS_OTHER); adapter->eims_other = E1000_EIMS_OTHER; break; case e1000_82576: + /* Turn on MSI-X capability first, or our settings + * won't stick. And it will take days to debug. */ + wr32(E1000_GPIE, E1000_GPIE_MSIX_MODE | + E1000_GPIE_PBA | E1000_GPIE_EIAME | + E1000_GPIE_NSICR); + + /* enable msix_other interrupt */ + adapter->eims_other = 1 << vector; tmp = (vector++ | E1000_IVAR_VALID) << 8; - wr32(E1000_IVAR_MISC, tmp); - adapter->eims_enable_mask = (1 << (vector)) - 1; - adapter->eims_other = 1 << (vector - 1); + wr32(E1000_IVAR_MISC, tmp); break; default: /* do nothing, since nothing else supports MSI-X */ break; } /* switch (hw->mac.type) */ + + adapter->eims_enable_mask |= adapter->eims_other; + + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + igb_assign_vector(q_vector, vector++); + adapter->eims_enable_mask |= q_vector->eims_value; + } + wrfl(); } @@ -614,43 +593,40 @@ static void igb_configure_msix(struct igb_adapter *adapter) static int igb_request_msix(struct igb_adapter *adapter) { struct net_device *netdev = adapter->netdev; + struct e1000_hw *hw = &adapter->hw; int i, err = 0, vector = 0; - vector = 0; - - for (i = 0; i < adapter->num_tx_queues; i++) { - struct igb_ring *ring = &(adapter->tx_ring[i]); - sprintf(ring->name, "%s-tx-%d", netdev->name, i); - err = request_irq(adapter->msix_entries[vector].vector, - &igb_msix_tx, 0, ring->name, - &(adapter->tx_ring[i])); - if (err) - goto out; - ring->itr_register = E1000_EITR(0) + (vector << 2); - ring->itr_val = 976; /* ~4000 ints/sec */ - vector++; - } - for (i = 0; i < adapter->num_rx_queues; i++) { - struct igb_ring *ring = &(adapter->rx_ring[i]); - if (strlen(netdev->name) < (IFNAMSIZ - 5)) - sprintf(ring->name, "%s-rx-%d", netdev->name, i); + err = request_irq(adapter->msix_entries[vector].vector, + &igb_msix_other, 0, netdev->name, adapter); + if (err) + goto out; + vector++; + + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + + q_vector->itr_register = hw->hw_addr + E1000_EITR(vector); + + if (q_vector->rx_ring && q_vector->tx_ring) + sprintf(q_vector->name, "%s-TxRx-%u", netdev->name, + q_vector->rx_ring->queue_index); + else if (q_vector->tx_ring) + sprintf(q_vector->name, "%s-tx-%u", netdev->name, + q_vector->tx_ring->queue_index); + else if (q_vector->rx_ring) + sprintf(q_vector->name, "%s-rx-%u", netdev->name, + q_vector->rx_ring->queue_index); else - memcpy(ring->name, netdev->name, IFNAMSIZ); + sprintf(q_vector->name, "%s-unused", netdev->name); + err = request_irq(adapter->msix_entries[vector].vector, - &igb_msix_rx, 0, ring->name, - &(adapter->rx_ring[i])); + &igb_msix_ring, 0, q_vector->name, + q_vector); if (err) goto out; - ring->itr_register = E1000_EITR(0) + (vector << 2); - ring->itr_val = adapter->itr; vector++; } - err = request_irq(adapter->msix_entries[vector].vector, - &igb_msix_other, 0, netdev->name, netdev); - if (err) - goto out; - igb_configure_msix(adapter); return 0; out: @@ -663,11 +639,44 @@ static void igb_reset_interrupt_capability(struct igb_adapter *adapter) pci_disable_msix(adapter->pdev); kfree(adapter->msix_entries); adapter->msix_entries = NULL; - } else if (adapter->flags & IGB_FLAG_HAS_MSI) + } else if (adapter->flags & IGB_FLAG_HAS_MSI) { pci_disable_msi(adapter->pdev); - return; + } } +/** + * igb_free_q_vectors - Free memory allocated for interrupt vectors + * @adapter: board private structure to initialize + * + * This function frees the memory allocated to the q_vectors. In addition if + * NAPI is enabled it will delete any references to the NAPI struct prior + * to freeing the q_vector. + **/ +static void igb_free_q_vectors(struct igb_adapter *adapter) +{ + int v_idx; + + for (v_idx = 0; v_idx < adapter->num_q_vectors; v_idx++) { + struct igb_q_vector *q_vector = adapter->q_vector[v_idx]; + adapter->q_vector[v_idx] = NULL; + netif_napi_del(&q_vector->napi); + kfree(q_vector); + } + adapter->num_q_vectors = 0; +} + +/** + * igb_clear_interrupt_scheme - reset the device to a state of no interrupts + * + * This function resets the device so that it has 0 rx queues, tx queues, and + * MSI-X interrupts allocated. + */ +static void igb_clear_interrupt_scheme(struct igb_adapter *adapter) +{ + igb_free_queues(adapter); + igb_free_q_vectors(adapter); + igb_reset_interrupt_capability(adapter); +} /** * igb_set_interrupt_capability - set MSI or MSI-X if supported @@ -681,11 +690,20 @@ static void igb_set_interrupt_capability(struct igb_adapter *adapter) int numvecs, i; /* Number of supported queues. */ - /* Having more queues than CPUs doesn't make sense. */ adapter->num_rx_queues = min_t(u32, IGB_MAX_RX_QUEUES, num_online_cpus()); adapter->num_tx_queues = min_t(u32, IGB_MAX_TX_QUEUES, num_online_cpus()); - numvecs = adapter->num_tx_queues + adapter->num_rx_queues + 1; + /* start with one vector for every rx queue */ + numvecs = adapter->num_rx_queues; + + /* if tx handler is seperate add 1 for every tx queue */ + numvecs += adapter->num_tx_queues; + + /* store the number of vectors reserved for queues */ + adapter->num_q_vectors = numvecs; + + /* add 1 vector for link status interrupts */ + numvecs++; adapter->msix_entries = kcalloc(numvecs, sizeof(struct msix_entry), GFP_KERNEL); if (!adapter->msix_entries) @@ -721,6 +739,7 @@ msi_only: #endif adapter->num_rx_queues = 1; adapter->num_tx_queues = 1; + adapter->num_q_vectors = 1; if (!pci_enable_msi(adapter->pdev)) adapter->flags |= IGB_FLAG_HAS_MSI; out: @@ -729,6 +748,139 @@ out: return; } +/** + * igb_alloc_q_vectors - Allocate memory for interrupt vectors + * @adapter: board private structure to initialize + * + * We allocate one q_vector per queue interrupt. If allocation fails we + * return -ENOMEM. + **/ +static int igb_alloc_q_vectors(struct igb_adapter *adapter) +{ + struct igb_q_vector *q_vector; + struct e1000_hw *hw = &adapter->hw; + int v_idx; + + for (v_idx = 0; v_idx < adapter->num_q_vectors; v_idx++) { + q_vector = kzalloc(sizeof(struct igb_q_vector), GFP_KERNEL); + if (!q_vector) + goto err_out; + q_vector->adapter = adapter; + q_vector->itr_shift = (hw->mac.type == e1000_82575) ? 16 : 0; + q_vector->itr_register = hw->hw_addr + E1000_EITR(0); + q_vector->itr_val = IGB_START_ITR; + q_vector->set_itr = 1; + netif_napi_add(adapter->netdev, &q_vector->napi, igb_poll, 64); + adapter->q_vector[v_idx] = q_vector; + } + return 0; + +err_out: + while (v_idx) { + v_idx--; + q_vector = adapter->q_vector[v_idx]; + netif_napi_del(&q_vector->napi); + kfree(q_vector); + adapter->q_vector[v_idx] = NULL; + } + return -ENOMEM; +} + +static void igb_map_rx_ring_to_vector(struct igb_adapter *adapter, + int ring_idx, int v_idx) +{ + struct igb_q_vector *q_vector; + + q_vector = adapter->q_vector[v_idx]; + q_vector->rx_ring = &adapter->rx_ring[ring_idx]; + q_vector->rx_ring->q_vector = q_vector; + q_vector->itr_val = adapter->itr; +} + +static void igb_map_tx_ring_to_vector(struct igb_adapter *adapter, + int ring_idx, int v_idx) +{ + struct igb_q_vector *q_vector; + + q_vector = adapter->q_vector[v_idx]; + q_vector->tx_ring = &adapter->tx_ring[ring_idx]; + q_vector->tx_ring->q_vector = q_vector; + q_vector->itr_val = adapter->itr; +} + +/** + * igb_map_ring_to_vector - maps allocated queues to vectors + * + * This function maps the recently allocated queues to vectors. + **/ +static int igb_map_ring_to_vector(struct igb_adapter *adapter) +{ + int i; + int v_idx = 0; + + if ((adapter->num_q_vectors < adapter->num_rx_queues) || + (adapter->num_q_vectors < adapter->num_tx_queues)) + return -ENOMEM; + + if (adapter->num_q_vectors >= + (adapter->num_rx_queues + adapter->num_tx_queues)) { + for (i = 0; i < adapter->num_rx_queues; i++) + igb_map_rx_ring_to_vector(adapter, i, v_idx++); + for (i = 0; i < adapter->num_tx_queues; i++) + igb_map_tx_ring_to_vector(adapter, i, v_idx++); + } else { + for (i = 0; i < adapter->num_rx_queues; i++) { + if (i < adapter->num_tx_queues) + igb_map_tx_ring_to_vector(adapter, i, v_idx); + igb_map_rx_ring_to_vector(adapter, i, v_idx++); + } + for (; i < adapter->num_tx_queues; i++) + igb_map_tx_ring_to_vector(adapter, i, v_idx++); + } + return 0; +} + +/** + * igb_init_interrupt_scheme - initialize interrupts, allocate queues/vectors + * + * This function initializes the interrupts and allocates all of the queues. + **/ +static int igb_init_interrupt_scheme(struct igb_adapter *adapter) +{ + struct pci_dev *pdev = adapter->pdev; + int err; + + igb_set_interrupt_capability(adapter); + + err = igb_alloc_q_vectors(adapter); + if (err) { + dev_err(&pdev->dev, "Unable to allocate memory for vectors\n"); + goto err_alloc_q_vectors; + } + + err = igb_alloc_queues(adapter); + if (err) { + dev_err(&pdev->dev, "Unable to allocate memory for queues\n"); + goto err_alloc_queues; + } + + err = igb_map_ring_to_vector(adapter); + if (err) { + dev_err(&pdev->dev, "Invalid q_vector to ring mapping\n"); + goto err_map_queues; + } + + + return 0; +err_map_queues: + igb_free_queues(adapter); +err_alloc_queues: + igb_free_q_vectors(adapter); +err_alloc_q_vectors: + igb_reset_interrupt_capability(adapter); + return err; +} + /** * igb_request_irq - initialize interrupts * @@ -738,6 +890,7 @@ out: static int igb_request_irq(struct igb_adapter *adapter) { struct net_device *netdev = adapter->netdev; + struct pci_dev *pdev = adapter->pdev; struct e1000_hw *hw = &adapter->hw; int err = 0; @@ -746,18 +899,36 @@ static int igb_request_irq(struct igb_adapter *adapter) if (!err) goto request_done; /* fall back to MSI */ - igb_reset_interrupt_capability(adapter); + igb_clear_interrupt_scheme(adapter); if (!pci_enable_msi(adapter->pdev)) adapter->flags |= IGB_FLAG_HAS_MSI; igb_free_all_tx_resources(adapter); igb_free_all_rx_resources(adapter); + adapter->num_tx_queues = 1; adapter->num_rx_queues = 1; - igb_alloc_queues(adapter); + adapter->num_q_vectors = 1; + err = igb_alloc_q_vectors(adapter); + if (err) { + dev_err(&pdev->dev, + "Unable to allocate memory for vectors\n"); + goto request_done; + } + err = igb_alloc_queues(adapter); + if (err) { + dev_err(&pdev->dev, + "Unable to allocate memory for queues\n"); + igb_free_q_vectors(adapter); + goto request_done; + } + igb_setup_all_tx_resources(adapter); + igb_setup_all_rx_resources(adapter); } else { switch (hw->mac.type) { case e1000_82575: wr32(E1000_MSIXBM(0), - (E1000_EICR_RX_QUEUE0 | E1000_EIMS_OTHER)); + (E1000_EICR_RX_QUEUE0 | + E1000_EICR_TX_QUEUE0 | + E1000_EIMS_OTHER)); break; case e1000_82576: wr32(E1000_IVAR0, E1000_IVAR_VALID); @@ -769,16 +940,17 @@ static int igb_request_irq(struct igb_adapter *adapter) if (adapter->flags & IGB_FLAG_HAS_MSI) { err = request_irq(adapter->pdev->irq, &igb_intr_msi, 0, - netdev->name, netdev); + netdev->name, adapter); if (!err) goto request_done; + /* fall back to legacy interrupts */ igb_reset_interrupt_capability(adapter); adapter->flags &= ~IGB_FLAG_HAS_MSI; } err = request_irq(adapter->pdev->irq, &igb_intr, IRQF_SHARED, - netdev->name, netdev); + netdev->name, adapter); if (err) dev_err(&adapter->pdev->dev, "Error %d getting interrupt\n", @@ -790,23 +962,19 @@ request_done: static void igb_free_irq(struct igb_adapter *adapter) { - struct net_device *netdev = adapter->netdev; - if (adapter->msix_entries) { int vector = 0, i; - for (i = 0; i < adapter->num_tx_queues; i++) - free_irq(adapter->msix_entries[vector++].vector, - &(adapter->tx_ring[i])); - for (i = 0; i < adapter->num_rx_queues; i++) - free_irq(adapter->msix_entries[vector++].vector, - &(adapter->rx_ring[i])); + free_irq(adapter->msix_entries[vector++].vector, adapter); - free_irq(adapter->msix_entries[vector++].vector, netdev); - return; + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + free_irq(adapter->msix_entries[vector++].vector, + q_vector); + } + } else { + free_irq(adapter->pdev->irq, adapter); } - - free_irq(adapter->pdev->irq, netdev); } /** @@ -967,8 +1135,10 @@ int igb_up(struct igb_adapter *adapter) clear_bit(__IGB_DOWN, &adapter->state); - for (i = 0; i < adapter->num_rx_queues; i++) - napi_enable(&adapter->rx_ring[i].napi); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + napi_enable(&q_vector->napi); + } if (adapter->msix_entries) igb_configure_msix(adapter); @@ -1012,8 +1182,10 @@ void igb_down(struct igb_adapter *adapter) wrfl(); msleep(10); - for (i = 0; i < adapter->num_rx_queues; i++) - napi_disable(&adapter->rx_ring[i].napi); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + napi_disable(&q_vector->napi); + } igb_irq_disable(adapter); @@ -1584,9 +1756,8 @@ err_eeprom: if (hw->flash_address) iounmap(hw->flash_address); - - igb_free_queues(adapter); err_sw_init: + igb_clear_interrupt_scheme(adapter); iounmap(hw->hw_addr); err_ioremap: free_netdev(netdev); @@ -1640,9 +1811,7 @@ static void __devexit igb_remove(struct pci_dev *pdev) if (!igb_check_reset_block(&adapter->hw)) igb_reset_phy(&adapter->hw); - igb_reset_interrupt_capability(adapter); - - igb_free_queues(adapter); + igb_clear_interrupt_scheme(adapter); #ifdef CONFIG_PCI_IOV /* reclaim resources allocated to VFs */ @@ -1696,9 +1865,7 @@ static int __devinit igb_sw_init(struct igb_adapter *adapter) /* This call may decrease the number of queues depending on * interrupt mode. */ - igb_set_interrupt_capability(adapter); - - if (igb_alloc_queues(adapter)) { + if (igb_init_interrupt_scheme(adapter)) { dev_err(&pdev->dev, "Unable to allocate memory for queues\n"); return -ENOMEM; } @@ -1768,8 +1935,10 @@ static int igb_open(struct net_device *netdev) /* From here on the code is the same as igb_up() */ clear_bit(__IGB_DOWN, &adapter->state); - for (i = 0; i < adapter->num_rx_queues; i++) - napi_enable(&adapter->rx_ring[i].napi); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + napi_enable(&q_vector->napi); + } /* Clear any pending interrupts. */ rd32(E1000_ICR); @@ -1858,14 +2027,13 @@ int igb_setup_tx_resources(struct igb_adapter *adapter, if (!tx_ring->desc) goto err; - tx_ring->adapter = adapter; tx_ring->next_to_use = 0; tx_ring->next_to_clean = 0; return 0; err: vfree(tx_ring->buffer_info); - dev_err(&adapter->pdev->dev, + dev_err(&pdev->dev, "Unable to allocate memory for the transmit descriptor ring\n"); return -ENOMEM; } @@ -1996,8 +2164,6 @@ int igb_setup_rx_resources(struct igb_adapter *adapter, rx_ring->next_to_clean = 0; rx_ring->next_to_use = 0; - rx_ring->adapter = adapter; - return 0; err: @@ -2308,7 +2474,7 @@ static void igb_configure_rx(struct igb_adapter *adapter) **/ void igb_free_tx_resources(struct igb_ring *tx_ring) { - struct pci_dev *pdev = tx_ring->adapter->pdev; + struct pci_dev *pdev = tx_ring->q_vector->adapter->pdev; igb_clean_tx_ring(tx_ring); @@ -2354,7 +2520,7 @@ static void igb_unmap_and_free_tx_resource(struct igb_adapter *adapter, **/ static void igb_clean_tx_ring(struct igb_ring *tx_ring) { - struct igb_adapter *adapter = tx_ring->adapter; + struct igb_adapter *adapter = tx_ring->q_vector->adapter; struct igb_buffer *buffer_info; unsigned long size; unsigned int i; @@ -2402,7 +2568,7 @@ static void igb_clean_all_tx_rings(struct igb_adapter *adapter) **/ void igb_free_rx_resources(struct igb_ring *rx_ring) { - struct pci_dev *pdev = rx_ring->adapter->pdev; + struct pci_dev *pdev = rx_ring->q_vector->adapter->pdev; igb_clean_rx_ring(rx_ring); @@ -2434,7 +2600,7 @@ static void igb_free_all_rx_resources(struct igb_adapter *adapter) **/ static void igb_clean_rx_ring(struct igb_ring *rx_ring) { - struct igb_adapter *adapter = rx_ring->adapter; + struct igb_adapter *adapter = rx_ring->q_vector->adapter; struct igb_buffer *buffer_info; struct pci_dev *pdev = adapter->pdev; unsigned long size; @@ -2749,7 +2915,6 @@ static void igb_watchdog_task(struct work_struct *work) struct net_device *netdev = adapter->netdev; struct igb_ring *tx_ring = adapter->tx_ring; u32 link; - u32 eics = 0; int i; link = igb_has_link(adapter); @@ -2848,8 +3013,11 @@ link_up: /* Cause software interrupt to ensure rx ring is cleaned */ if (adapter->msix_entries) { - for (i = 0; i < adapter->num_rx_queues; i++) - eics |= adapter->rx_ring[i].eims_value; + u32 eics = 0; + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + eics |= q_vector->eims_value; + } wr32(E1000_EICS, eics); } else { wr32(E1000_ICS, E1000_ICS_RXDMT0); @@ -2886,25 +3054,37 @@ enum latency_range { * parameter (see igb_param.c) * NOTE: This function is called only when operating in a multiqueue * receive environment. - * @rx_ring: pointer to ring + * @q_vector: pointer to q_vector **/ -static void igb_update_ring_itr(struct igb_ring *rx_ring) +static void igb_update_ring_itr(struct igb_q_vector *q_vector) { - int new_val = rx_ring->itr_val; + int new_val = q_vector->itr_val; int avg_wire_size = 0; - struct igb_adapter *adapter = rx_ring->adapter; - - if (!rx_ring->total_packets) - goto clear_counts; /* no packets, so don't do anything */ + struct igb_adapter *adapter = q_vector->adapter; /* For non-gigabit speeds, just fix the interrupt rate at 4000 * ints/sec - ITR timer value of 120 ticks. */ if (adapter->link_speed != SPEED_1000) { - new_val = 120; + new_val = 976; goto set_itr_val; } - avg_wire_size = rx_ring->total_bytes / rx_ring->total_packets; + + if (q_vector->rx_ring && q_vector->rx_ring->total_packets) { + struct igb_ring *ring = q_vector->rx_ring; + avg_wire_size = ring->total_bytes / ring->total_packets; + } + + if (q_vector->tx_ring && q_vector->tx_ring->total_packets) { + struct igb_ring *ring = q_vector->tx_ring; + avg_wire_size = max_t(u32, avg_wire_size, + (ring->total_bytes / + ring->total_packets)); + } + + /* if avg_wire_size isn't set no work was done */ + if (!avg_wire_size) + goto clear_counts; /* Add 24 bytes to size to account for CRC, preamble, and gap */ avg_wire_size += 24; @@ -2919,13 +3099,19 @@ static void igb_update_ring_itr(struct igb_ring *rx_ring) new_val = avg_wire_size / 2; set_itr_val: - if (new_val != rx_ring->itr_val) { - rx_ring->itr_val = new_val; - rx_ring->set_itr = 1; + if (new_val != q_vector->itr_val) { + q_vector->itr_val = new_val; + q_vector->set_itr = 1; } clear_counts: - rx_ring->total_bytes = 0; - rx_ring->total_packets = 0; + if (q_vector->rx_ring) { + q_vector->rx_ring->total_bytes = 0; + q_vector->rx_ring->total_packets = 0; + } + if (q_vector->tx_ring) { + q_vector->tx_ring->total_bytes = 0; + q_vector->tx_ring->total_packets = 0; + } } /** @@ -2942,7 +3128,7 @@ clear_counts: * NOTE: These calculations are only valid when operating in a single- * queue environment. * @adapter: pointer to adapter - * @itr_setting: current adapter->itr + * @itr_setting: current q_vector->itr_val * @packets: the number of packets during this measurement interval * @bytes: the number of bytes during this measurement interval **/ @@ -2994,8 +3180,9 @@ update_itr_done: static void igb_set_itr(struct igb_adapter *adapter) { + struct igb_q_vector *q_vector = adapter->q_vector[0]; u16 current_itr; - u32 new_itr = adapter->itr; + u32 new_itr = q_vector->itr_val; /* for non-gigabit speeds, just fix the interrupt rate at 4000 */ if (adapter->link_speed != SPEED_1000) { @@ -3009,15 +3196,11 @@ static void igb_set_itr(struct igb_adapter *adapter) adapter->rx_ring->total_packets, adapter->rx_ring->total_bytes); - if (adapter->rx_ring->buddy) { - adapter->tx_itr = igb_update_itr(adapter, - adapter->tx_itr, - adapter->tx_ring->total_packets, - adapter->tx_ring->total_bytes); - current_itr = max(adapter->rx_itr, adapter->tx_itr); - } else { - current_itr = adapter->rx_itr; - } + adapter->tx_itr = igb_update_itr(adapter, + adapter->tx_itr, + adapter->tx_ring->total_packets, + adapter->tx_ring->total_bytes); + current_itr = max(adapter->rx_itr, adapter->tx_itr); /* conservative mode (itr 3) eliminates the lowest_latency setting */ if (adapter->itr_setting == 3 && current_itr == lowest_latency) @@ -3041,18 +3224,17 @@ static void igb_set_itr(struct igb_adapter *adapter) set_itr_now: adapter->rx_ring->total_bytes = 0; adapter->rx_ring->total_packets = 0; - if (adapter->rx_ring->buddy) { - adapter->rx_ring->buddy->total_bytes = 0; - adapter->rx_ring->buddy->total_packets = 0; - } + adapter->tx_ring->total_bytes = 0; + adapter->tx_ring->total_packets = 0; - if (new_itr != adapter->itr) { + if (new_itr != q_vector->itr_val) { /* this attempts to bias the interrupt rate towards Bulk * by adding intermediate steps when interrupt rate is * increasing */ - new_itr = new_itr > adapter->itr ? - max((new_itr * adapter->itr) / - (new_itr + (adapter->itr >> 2)), new_itr) : + new_itr = new_itr > q_vector->itr_val ? + max((new_itr * q_vector->itr_val) / + (new_itr + (q_vector->itr_val >> 2)), + new_itr) : new_itr; /* Don't write the value here; it resets the adapter's * internal timer, and causes us to delay far longer than @@ -3060,15 +3242,13 @@ set_itr_now: * value at the beginning of the next interrupt so the timing * ends up being correct. */ - adapter->itr = new_itr; - adapter->rx_ring->itr_val = new_itr; - adapter->rx_ring->set_itr = 1; + q_vector->itr_val = new_itr; + q_vector->set_itr = 1; } return; } - #define IGB_TX_FLAGS_CSUM 0x00000001 #define IGB_TX_FLAGS_VLAN 0x00000002 #define IGB_TX_FLAGS_TSO 0x00000004 @@ -3781,14 +3961,12 @@ void igb_update_stats(struct igb_adapter *adapter) static irqreturn_t igb_msix_other(int irq, void *data) { - struct net_device *netdev = data; - struct igb_adapter *adapter = netdev_priv(netdev); + struct igb_adapter *adapter = data; struct e1000_hw *hw = &adapter->hw; u32 icr = rd32(E1000_ICR); - /* reading ICR causes bit 31 of EICR to be cleared */ - if(icr & E1000_ICR_DOUTSYNC) { + if (icr & E1000_ICR_DOUTSYNC) { /* HW is reporting DMA is out of sync */ adapter->stats.doosync++; } @@ -3810,119 +3988,79 @@ static irqreturn_t igb_msix_other(int irq, void *data) return IRQ_HANDLED; } -static irqreturn_t igb_msix_tx(int irq, void *data) +static void igb_write_itr(struct igb_q_vector *q_vector) { - struct igb_ring *tx_ring = data; - struct igb_adapter *adapter = tx_ring->adapter; - struct e1000_hw *hw = &adapter->hw; + u32 itr_val = q_vector->itr_val & 0x7FFC; -#ifdef CONFIG_IGB_DCA - if (adapter->flags & IGB_FLAG_DCA_ENABLED) - igb_update_tx_dca(tx_ring); -#endif + if (!q_vector->set_itr) + return; - tx_ring->total_bytes = 0; - tx_ring->total_packets = 0; + if (!itr_val) + itr_val = 0x4; - /* auto mask will automatically reenable the interrupt when we write - * EICS */ - if (!igb_clean_tx_irq(tx_ring)) - /* Ring was not completely cleaned, so fire another interrupt */ - wr32(E1000_EICS, tx_ring->eims_value); + if (q_vector->itr_shift) + itr_val |= itr_val << q_vector->itr_shift; else - wr32(E1000_EIMS, tx_ring->eims_value); + itr_val |= 0x8000000; - return IRQ_HANDLED; -} - -static void igb_write_itr(struct igb_ring *ring) -{ - struct e1000_hw *hw = &ring->adapter->hw; - if ((ring->adapter->itr_setting & 3) && ring->set_itr) { - switch (hw->mac.type) { - case e1000_82576: - wr32(ring->itr_register, ring->itr_val | - 0x80000000); - break; - default: - wr32(ring->itr_register, ring->itr_val | - (ring->itr_val << 16)); - break; - } - ring->set_itr = 0; - } + writel(itr_val, q_vector->itr_register); + q_vector->set_itr = 0; } -static irqreturn_t igb_msix_rx(int irq, void *data) +static irqreturn_t igb_msix_ring(int irq, void *data) { - struct igb_ring *rx_ring = data; - - /* Write the ITR value calculated at the end of the - * previous interrupt. - */ + struct igb_q_vector *q_vector = data; - igb_write_itr(rx_ring); + /* Write the ITR value calculated from the previous interrupt. */ + igb_write_itr(q_vector); - if (napi_schedule_prep(&rx_ring->napi)) - __napi_schedule(&rx_ring->napi); + napi_schedule(&q_vector->napi); -#ifdef CONFIG_IGB_DCA - if (rx_ring->adapter->flags & IGB_FLAG_DCA_ENABLED) - igb_update_rx_dca(rx_ring); -#endif - return IRQ_HANDLED; + return IRQ_HANDLED; } #ifdef CONFIG_IGB_DCA -static void igb_update_rx_dca(struct igb_ring *rx_ring) +static void igb_update_dca(struct igb_q_vector *q_vector) { - u32 dca_rxctrl; - struct igb_adapter *adapter = rx_ring->adapter; + struct igb_adapter *adapter = q_vector->adapter; struct e1000_hw *hw = &adapter->hw; int cpu = get_cpu(); - int q = rx_ring->reg_idx; - if (rx_ring->cpu != cpu) { - dca_rxctrl = rd32(E1000_DCA_RXCTRL(q)); - if (hw->mac.type == e1000_82576) { - dca_rxctrl &= ~E1000_DCA_RXCTRL_CPUID_MASK_82576; - dca_rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) << - E1000_DCA_RXCTRL_CPUID_SHIFT; + if (q_vector->cpu == cpu) + goto out_no_update; + + if (q_vector->tx_ring) { + int q = q_vector->tx_ring->reg_idx; + u32 dca_txctrl = rd32(E1000_DCA_TXCTRL(q)); + if (hw->mac.type == e1000_82575) { + dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK; + dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu); } else { + dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK_82576; + dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) << + E1000_DCA_TXCTRL_CPUID_SHIFT; + } + dca_txctrl |= E1000_DCA_TXCTRL_DESC_DCA_EN; + wr32(E1000_DCA_TXCTRL(q), dca_txctrl); + } + if (q_vector->rx_ring) { + int q = q_vector->rx_ring->reg_idx; + u32 dca_rxctrl = rd32(E1000_DCA_RXCTRL(q)); + if (hw->mac.type == e1000_82575) { dca_rxctrl &= ~E1000_DCA_RXCTRL_CPUID_MASK; dca_rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu); + } else { + dca_rxctrl &= ~E1000_DCA_RXCTRL_CPUID_MASK_82576; + dca_rxctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) << + E1000_DCA_RXCTRL_CPUID_SHIFT; } dca_rxctrl |= E1000_DCA_RXCTRL_DESC_DCA_EN; dca_rxctrl |= E1000_DCA_RXCTRL_HEAD_DCA_EN; dca_rxctrl |= E1000_DCA_RXCTRL_DATA_DCA_EN; wr32(E1000_DCA_RXCTRL(q), dca_rxctrl); - rx_ring->cpu = cpu; - } - put_cpu(); -} - -static void igb_update_tx_dca(struct igb_ring *tx_ring) -{ - u32 dca_txctrl; - struct igb_adapter *adapter = tx_ring->adapter; - struct e1000_hw *hw = &adapter->hw; - int cpu = get_cpu(); - int q = tx_ring->reg_idx; - - if (tx_ring->cpu != cpu) { - dca_txctrl = rd32(E1000_DCA_TXCTRL(q)); - if (hw->mac.type == e1000_82576) { - dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK_82576; - dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu) << - E1000_DCA_TXCTRL_CPUID_SHIFT; - } else { - dca_txctrl &= ~E1000_DCA_TXCTRL_CPUID_MASK; - dca_txctrl |= dca3_get_tag(&adapter->pdev->dev, cpu); - } - dca_txctrl |= E1000_DCA_TXCTRL_DESC_DCA_EN; - wr32(E1000_DCA_TXCTRL(q), dca_txctrl); - tx_ring->cpu = cpu; } + q_vector->cpu = cpu; +out_no_update: put_cpu(); } @@ -3937,13 +4075,10 @@ static void igb_setup_dca(struct igb_adapter *adapter) /* Always use CB2 mode, difference is masked in the CB driver. */ wr32(E1000_DCA_CTRL, E1000_DCA_CTRL_DCA_MODE_CB2); - for (i = 0; i < adapter->num_tx_queues; i++) { - adapter->tx_ring[i].cpu = -1; - igb_update_tx_dca(&adapter->tx_ring[i]); - } - for (i = 0; i < adapter->num_rx_queues; i++) { - adapter->rx_ring[i].cpu = -1; - igb_update_rx_dca(&adapter->rx_ring[i]); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + q_vector->cpu = -1; + igb_update_dca(q_vector); } } @@ -3972,7 +4107,7 @@ static int __igb_notify_dca(struct device *dev, void *data) case DCA_PROVIDER_REMOVE: if (adapter->flags & IGB_FLAG_DCA_ENABLED) { /* without this a class_device is left - * hanging around in the sysfs model */ + * hanging around in the sysfs model */ dca_remove_requester(dev); dev_info(&adapter->pdev->dev, "DCA disabled\n"); adapter->flags &= ~IGB_FLAG_DCA_ENABLED; @@ -4379,15 +4514,15 @@ static void igb_set_uta(struct igb_adapter *adapter) **/ static irqreturn_t igb_intr_msi(int irq, void *data) { - struct net_device *netdev = data; - struct igb_adapter *adapter = netdev_priv(netdev); + struct igb_adapter *adapter = data; + struct igb_q_vector *q_vector = adapter->q_vector[0]; struct e1000_hw *hw = &adapter->hw; /* read ICR disables interrupts using IAM */ u32 icr = rd32(E1000_ICR); - igb_write_itr(adapter->rx_ring); + igb_write_itr(q_vector); - if(icr & E1000_ICR_DOUTSYNC) { + if (icr & E1000_ICR_DOUTSYNC) { /* HW is reporting DMA is out of sync */ adapter->stats.doosync++; } @@ -4398,7 +4533,7 @@ static irqreturn_t igb_intr_msi(int irq, void *data) mod_timer(&adapter->watchdog_timer, jiffies + 1); } - napi_schedule(&adapter->rx_ring[0].napi); + napi_schedule(&q_vector->napi); return IRQ_HANDLED; } @@ -4410,8 +4545,8 @@ static irqreturn_t igb_intr_msi(int irq, void *data) **/ static irqreturn_t igb_intr(int irq, void *data) { - struct net_device *netdev = data; - struct igb_adapter *adapter = netdev_priv(netdev); + struct igb_adapter *adapter = data; + struct igb_q_vector *q_vector = adapter->q_vector[0]; struct e1000_hw *hw = &adapter->hw; /* Interrupt Auto-Mask...upon reading ICR, interrupts are masked. No * need for the IMC write */ @@ -4419,14 +4554,14 @@ static irqreturn_t igb_intr(int irq, void *data) if (!icr) return IRQ_NONE; /* Not our interrupt */ - igb_write_itr(adapter->rx_ring); + igb_write_itr(q_vector); /* IMS will not auto-mask if INT_ASSERTED is not set, and if it is * not set, then the adapter didn't send an interrupt */ if (!(icr & E1000_ICR_INT_ASSERTED)) return IRQ_NONE; - if(icr & E1000_ICR_DOUTSYNC) { + if (icr & E1000_ICR_DOUTSYNC) { /* HW is reporting DMA is out of sync */ adapter->stats.doosync++; } @@ -4438,26 +4573,26 @@ static irqreturn_t igb_intr(int irq, void *data) mod_timer(&adapter->watchdog_timer, jiffies + 1); } - napi_schedule(&adapter->rx_ring[0].napi); + napi_schedule(&q_vector->napi); return IRQ_HANDLED; } -static inline void igb_rx_irq_enable(struct igb_ring *rx_ring) +static inline void igb_ring_irq_enable(struct igb_q_vector *q_vector) { - struct igb_adapter *adapter = rx_ring->adapter; + struct igb_adapter *adapter = q_vector->adapter; struct e1000_hw *hw = &adapter->hw; if (adapter->itr_setting & 3) { - if (adapter->num_rx_queues == 1) + if (!adapter->msix_entries) igb_set_itr(adapter); else - igb_update_ring_itr(rx_ring); + igb_update_ring_itr(q_vector); } if (!test_bit(__IGB_DOWN, &adapter->state)) { if (adapter->msix_entries) - wr32(E1000_EIMS, rx_ring->eims_value); + wr32(E1000_EIMS, q_vector->eims_value); else igb_irq_enable(adapter); } @@ -4470,28 +4605,28 @@ static inline void igb_rx_irq_enable(struct igb_ring *rx_ring) **/ static int igb_poll(struct napi_struct *napi, int budget) { - struct igb_ring *rx_ring = container_of(napi, struct igb_ring, napi); - int work_done = 0; + struct igb_q_vector *q_vector = container_of(napi, + struct igb_q_vector, + napi); + int tx_clean_complete = 1, work_done = 0; #ifdef CONFIG_IGB_DCA - if (rx_ring->adapter->flags & IGB_FLAG_DCA_ENABLED) - igb_update_rx_dca(rx_ring); + if (q_vector->adapter->flags & IGB_FLAG_DCA_ENABLED) + igb_update_dca(q_vector); #endif - igb_clean_rx_irq_adv(rx_ring, &work_done, budget); + if (q_vector->tx_ring) + tx_clean_complete = igb_clean_tx_irq(q_vector); - if (rx_ring->buddy) { -#ifdef CONFIG_IGB_DCA - if (rx_ring->adapter->flags & IGB_FLAG_DCA_ENABLED) - igb_update_tx_dca(rx_ring->buddy); -#endif - if (!igb_clean_tx_irq(rx_ring->buddy)) - work_done = budget; - } + if (q_vector->rx_ring) + igb_clean_rx_irq_adv(q_vector, &work_done, budget); + + if (!tx_clean_complete) + work_done = budget; /* If not enough Rx work done, exit the polling mode */ if (work_done < budget) { napi_complete(napi); - igb_rx_irq_enable(rx_ring); + igb_ring_irq_enable(q_vector); } return work_done; @@ -4533,12 +4668,13 @@ static void igb_tx_hwtstamp(struct igb_adapter *adapter, struct sk_buff *skb) /** * igb_clean_tx_irq - Reclaim resources after transmit completes - * @adapter: board private structure + * @q_vector: pointer to q_vector containing needed info * returns true if ring is completely cleaned **/ -static bool igb_clean_tx_irq(struct igb_ring *tx_ring) +static bool igb_clean_tx_irq(struct igb_q_vector *q_vector) { - struct igb_adapter *adapter = tx_ring->adapter; + struct igb_adapter *adapter = q_vector->adapter; + struct igb_ring *tx_ring = q_vector->tx_ring; struct net_device *netdev = adapter->netdev; struct e1000_hw *hw = &adapter->hw; struct igb_buffer *buffer_info; @@ -4646,25 +4782,21 @@ static bool igb_clean_tx_irq(struct igb_ring *tx_ring) /** * igb_receive_skb - helper function to handle rx indications - * @ring: pointer to receive ring receving this packet - * @status: descriptor status field as written by hardware - * @rx_desc: receive descriptor containing vlan and type information. - * @skb: pointer to sk_buff to be indicated to stack + * @q_vector: structure containing interrupt and ring information + * @skb: packet to send up + * @vlan_tag: vlan tag for packet **/ -static void igb_receive_skb(struct igb_ring *ring, u8 status, - union e1000_adv_rx_desc * rx_desc, - struct sk_buff *skb) -{ - struct igb_adapter * adapter = ring->adapter; - bool vlan_extracted = (adapter->vlgrp && (status & E1000_RXD_STAT_VP)); - - skb_record_rx_queue(skb, ring->queue_index); - if (vlan_extracted) - vlan_gro_receive(&ring->napi, adapter->vlgrp, - le16_to_cpu(rx_desc->wb.upper.vlan), - skb); +static void igb_receive_skb(struct igb_q_vector *q_vector, + struct sk_buff *skb, + u16 vlan_tag) +{ + struct igb_adapter *adapter = q_vector->adapter; + + if (vlan_tag) + vlan_gro_receive(&q_vector->napi, adapter->vlgrp, + vlan_tag, skb); else - napi_gro_receive(&ring->napi, skb); + napi_gro_receive(&q_vector->napi, skb); } static inline void igb_rx_checksum_adv(struct igb_adapter *adapter, @@ -4712,11 +4844,12 @@ static inline u16 igb_get_hlen(struct igb_adapter *adapter, return hlen; } -static bool igb_clean_rx_irq_adv(struct igb_ring *rx_ring, - int *work_done, int budget) +static bool igb_clean_rx_irq_adv(struct igb_q_vector *q_vector, + int *work_done, int budget) { - struct igb_adapter *adapter = rx_ring->adapter; + struct igb_adapter *adapter = q_vector->adapter; struct net_device *netdev = adapter->netdev; + struct igb_ring *rx_ring = q_vector->rx_ring; struct e1000_hw *hw = &adapter->hw; struct pci_dev *pdev = adapter->pdev; union e1000_adv_rx_desc *rx_desc , *next_rxd; @@ -4728,6 +4861,7 @@ static bool igb_clean_rx_irq_adv(struct igb_ring *rx_ring, unsigned int i; u32 staterr; u16 length; + u16 vlan_tag; i = rx_ring->next_to_clean; buffer_info = &rx_ring->buffer_info[i]; @@ -4855,8 +4989,12 @@ send_up: igb_rx_checksum_adv(adapter, staterr, skb); skb->protocol = eth_type_trans(skb, netdev); + skb_record_rx_queue(skb, rx_ring->queue_index); + + vlan_tag = ((staterr & E1000_RXD_STAT_VP) ? + le16_to_cpu(rx_desc->wb.upper.vlan) : 0); - igb_receive_skb(rx_ring, staterr, rx_desc, skb); + igb_receive_skb(q_vector, skb, vlan_tag); next_desc: rx_desc->wb.upper.status_error = 0; @@ -4895,7 +5033,7 @@ next_desc: static void igb_alloc_rx_buffers_adv(struct igb_ring *rx_ring, int cleaned_count) { - struct igb_adapter *adapter = rx_ring->adapter; + struct igb_adapter *adapter = rx_ring->q_vector->adapter; struct net_device *netdev = adapter->netdev; struct pci_dev *pdev = adapter->pdev; union e1000_adv_rx_desc *rx_desc; @@ -5360,9 +5498,7 @@ static int __igb_shutdown(struct pci_dev *pdev, bool *enable_wake) if (netif_running(netdev)) igb_close(netdev); - igb_reset_interrupt_capability(adapter); - - igb_free_queues(adapter); + igb_clear_interrupt_scheme(adapter); #ifdef CONFIG_PM retval = pci_save_state(pdev); @@ -5457,9 +5593,7 @@ static int igb_resume(struct pci_dev *pdev) pci_enable_wake(pdev, PCI_D3hot, 0); pci_enable_wake(pdev, PCI_D3cold, 0); - igb_set_interrupt_capability(adapter); - - if (igb_alloc_queues(adapter)) { + if (igb_init_interrupt_scheme(adapter)) { dev_err(&pdev->dev, "Unable to allocate memory for queues\n"); return -ENOMEM; } @@ -5511,22 +5645,16 @@ static void igb_netpoll(struct net_device *netdev) int i; if (!adapter->msix_entries) { + struct igb_q_vector *q_vector = adapter->q_vector[0]; igb_irq_disable(adapter); - napi_schedule(&adapter->rx_ring[0].napi); + napi_schedule(&q_vector->napi); return; } - for (i = 0; i < adapter->num_tx_queues; i++) { - struct igb_ring *tx_ring = &adapter->tx_ring[i]; - wr32(E1000_EIMC, tx_ring->eims_value); - igb_clean_tx_irq(tx_ring); - wr32(E1000_EIMS, tx_ring->eims_value); - } - - for (i = 0; i < adapter->num_rx_queues; i++) { - struct igb_ring *rx_ring = &adapter->rx_ring[i]; - wr32(E1000_EIMC, rx_ring->eims_value); - napi_schedule(&rx_ring->napi); + for (i = 0; i < adapter->num_q_vectors; i++) { + struct igb_q_vector *q_vector = adapter->q_vector[i]; + wr32(E1000_EIMC, q_vector->eims_value); + napi_schedule(&q_vector->napi); } } #endif /* CONFIG_NET_POLL_CONTROLLER */