From e0244903d4a6a27f9f1e8b46b89afa4130aa9164 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:56:56 -0700 Subject: [PATCH 01/15] fm10k: set netdev features in one location Don't change netdev hw_features later in fm10k_probe, instead set all values inside fm10k_alloc_netdev. To do so, we need to know the MAC type (whether it is PF or VF) in order to determine what to do. This helps ensure that all logic regarding features is co-located. Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k.h | 2 +- .../net/ethernet/intel/fm10k/fm10k_netdev.c | 31 ++++++++++++------- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 9 +----- 3 files changed, 21 insertions(+), 21 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k.h b/drivers/net/ethernet/intel/fm10k/fm10k.h index 48809e5d3f79..b178905fa43e 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k.h @@ -484,7 +484,7 @@ void fm10k_netpoll(struct net_device *netdev); #endif /* Netdev */ -struct net_device *fm10k_alloc_netdev(void); +struct net_device *fm10k_alloc_netdev(const struct fm10k_info *info); int fm10k_setup_rx_resources(struct fm10k_ring *); int fm10k_setup_tx_resources(struct fm10k_ring *); void fm10k_free_rx_resources(struct fm10k_ring *); diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c index 7781e80896a6..79f6b7dd2362 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_netdev.c @@ -1388,8 +1388,9 @@ static const struct net_device_ops fm10k_netdev_ops = { #define DEFAULT_DEBUG_LEVEL_SHIFT 3 -struct net_device *fm10k_alloc_netdev(void) +struct net_device *fm10k_alloc_netdev(const struct fm10k_info *info) { + netdev_features_t hw_features; struct fm10k_intfc *interface; struct net_device *dev; @@ -1412,27 +1413,31 @@ struct net_device *fm10k_alloc_netdev(void) NETIF_F_TSO | NETIF_F_TSO6 | NETIF_F_TSO_ECN | - NETIF_F_GSO_UDP_TUNNEL | NETIF_F_RXHASH | NETIF_F_RXCSUM; + /* Only the PF can support VXLAN and NVGRE tunnel offloads */ + if (info->mac == fm10k_mac_pf) { + dev->hw_enc_features = NETIF_F_IP_CSUM | + NETIF_F_TSO | + NETIF_F_TSO6 | + NETIF_F_TSO_ECN | + NETIF_F_GSO_UDP_TUNNEL | + NETIF_F_IPV6_CSUM | + NETIF_F_SG; + + dev->features |= NETIF_F_GSO_UDP_TUNNEL; + } + /* all features defined to this point should be changeable */ - dev->hw_features |= dev->features; + hw_features = dev->features; /* allow user to enable L2 forwarding acceleration */ - dev->hw_features |= NETIF_F_HW_L2FW_DOFFLOAD; + hw_features |= NETIF_F_HW_L2FW_DOFFLOAD; /* configure VLAN features */ dev->vlan_features |= dev->features; - /* configure tunnel offloads */ - dev->hw_enc_features |= NETIF_F_IP_CSUM | - NETIF_F_TSO | - NETIF_F_TSO6 | - NETIF_F_TSO_ECN | - NETIF_F_GSO_UDP_TUNNEL | - NETIF_F_IPV6_CSUM; - /* we want to leave these both on as we cannot disable VLAN tag * insertion or stripping on the hardware since it is contained * in the FTAG and not in the frame itself. @@ -1443,5 +1448,7 @@ struct net_device *fm10k_alloc_netdev(void) dev->priv_flags |= IFF_UNICAST_FLT; + dev->hw_features |= hw_features; + return dev; } diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index 5fbffbaefe32..1af4b222d003 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -1722,13 +1722,6 @@ static int fm10k_sw_init(struct fm10k_intfc *interface, pci_resource_len(pdev, 4)); hw->sw_addr = interface->sw_addr; - /* Only the PF can support VXLAN and NVGRE offloads */ - if (hw->mac.type != fm10k_mac_pf) { - netdev->hw_enc_features = 0; - netdev->features &= ~NETIF_F_GSO_UDP_TUNNEL; - netdev->hw_features &= ~NETIF_F_GSO_UDP_TUNNEL; - } - /* initialize DCBNL interface */ fm10k_dcbnl_set_ops(netdev); @@ -1894,7 +1887,7 @@ static int fm10k_probe(struct pci_dev *pdev, pci_set_master(pdev); pci_save_state(pdev); - netdev = fm10k_alloc_netdev(); + netdev = fm10k_alloc_netdev(fm10k_info_tbl[ent->driver_data]); if (!netdev) { err = -ENOMEM; goto err_alloc_netdev; From 0e8d5b5975401c83641efd5d4595e6cdbe9e9e2f Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:56:57 -0700 Subject: [PATCH 02/15] fm10k: reset max_queues on init_hw_vf failure VF drivers must detect how many queues are available. Previously, the driver assumed that each VF has at minimum 1 queue. This assumption is incorrect, since it is possible that the PF has not yet assigned the queues to the VF by the time the VF checks. To resolve this, we added a check first to ensure that the first queue is infact owned by the VF at init_hw_vf time. However, the code flow did not reset hw->mac.max_queues to 0. In some cases, such as during reinit flows, we call init_hw_vf without clearing the previous value of hw->mac.max_queues. Due to this, when init_hw_vf errors out, if its error code is not properly handled the VF driver may still believe it has queues which no longer belong to it. Fix this by clearing the hw->mac.max_queues on exit due to errors. Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_vf.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_vf.c b/drivers/net/ethernet/intel/fm10k/fm10k_vf.c index 3a18ef1cc017..d512575c33f3 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_vf.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_vf.c @@ -105,8 +105,10 @@ static s32 fm10k_init_hw_vf(struct fm10k_hw *hw) /* verify we have at least 1 queue */ if (!~fm10k_read_reg(hw, FM10K_TXQCTL(0)) || - !~fm10k_read_reg(hw, FM10K_RXQCTL(0))) - return FM10K_ERR_NO_RESOURCES; + !~fm10k_read_reg(hw, FM10K_RXQCTL(0))) { + err = FM10K_ERR_NO_RESOURCES; + goto reset_max_queues; + } /* determine how many queues we have */ for (i = 1; tqdloc0 && (i < FM10K_MAX_QUEUES_POOL); i++) { @@ -124,7 +126,7 @@ static s32 fm10k_init_hw_vf(struct fm10k_hw *hw) /* shut down queues we own and reset DMA configuration */ err = fm10k_disable_queues_generic(hw, i); if (err) - return err; + goto reset_max_queues; /* record maximum queue count */ hw->mac.max_queues = i; @@ -134,6 +136,11 @@ static s32 fm10k_init_hw_vf(struct fm10k_hw *hw) FM10K_TXQCTL_VID_MASK) >> FM10K_TXQCTL_VID_SHIFT; return 0; + +reset_max_queues: + hw->mac.max_queues = 0; + + return err; } /* This structure defines the attibutes to be parsed below */ From 1343c65f70ee1b1f968a08b30e1836a4e37116cd Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:56:58 -0700 Subject: [PATCH 03/15] fm10k: always check init_hw for errors A recent change modified init_hw in some flows the function may fail on VF devices. For example, if a VF doesn't yet own its own queues. However, many callers of init_hw didn't bother to check the error code. Other callers checked but only displayed diagnostic messages without actually handling the consequences. Fix this by (a) always returning and preventing the netdevice from going up, and (b) printing the diagnostic in every flow for consistency. This should resolve an issue where VF drivers would attempt to come up before the PF has finished assigning queues. In addition, change the dmesg output to explicitly show the actual function that failed, instead of combining reset_hw and init_hw into a single check, to help for future debugging. Fixes: 1d568b0f6424 ("fm10k: do not assume VF always has 1 queue") Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 34 +++++++++++++++++--- 1 file changed, 29 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index 1af4b222d003..9c21d1e45543 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -163,9 +163,17 @@ static void fm10k_reinit(struct fm10k_intfc *interface) interface->last_reset = jiffies + (10 * HZ); /* reset and initialize the hardware so it is in a known state */ - err = hw->mac.ops.reset_hw(hw) ? : hw->mac.ops.init_hw(hw); - if (err) + err = hw->mac.ops.reset_hw(hw); + if (err) { + dev_err(&interface->pdev->dev, "reset_hw failed: %d\n", err); + goto reinit_err; + } + + err = hw->mac.ops.init_hw(hw); + if (err) { dev_err(&interface->pdev->dev, "init_hw failed: %d\n", err); + goto reinit_err; + } /* reassociate interrupts */ fm10k_mbx_request_irq(interface); @@ -193,6 +201,10 @@ static void fm10k_reinit(struct fm10k_intfc *interface) fm10k_iov_resume(interface->pdev); +reinit_err: + if (err) + netif_device_detach(netdev); + rtnl_unlock(); clear_bit(__FM10K_RESETTING, &interface->state); @@ -1684,7 +1696,13 @@ static int fm10k_sw_init(struct fm10k_intfc *interface, interface->last_reset = jiffies + (10 * HZ); /* reset and initialize the hardware so it is in a known state */ - err = hw->mac.ops.reset_hw(hw) ? : hw->mac.ops.init_hw(hw); + err = hw->mac.ops.reset_hw(hw); + if (err) { + dev_err(&pdev->dev, "reset_hw failed: %d\n", err); + return err; + } + + err = hw->mac.ops.init_hw(hw); if (err) { dev_err(&pdev->dev, "init_hw failed: %d\n", err); return err; @@ -2064,8 +2082,10 @@ static int fm10k_resume(struct pci_dev *pdev) /* reset hardware to known state */ err = hw->mac.ops.init_hw(&interface->hw); - if (err) + if (err) { + dev_err(&pdev->dev, "init_hw failed: %d\n", err); return err; + } /* reset statistics starting values */ hw->mac.ops.rebind_hw_stats(hw, &interface->stats); @@ -2241,7 +2261,11 @@ static void fm10k_io_resume(struct pci_dev *pdev) int err = 0; /* reset hardware to known state */ - hw->mac.ops.init_hw(&interface->hw); + err = hw->mac.ops.init_hw(&interface->hw); + if (err) { + dev_err(&pdev->dev, "init_hw failed: %d\n", err); + return; + } /* reset statistics starting values */ hw->mac.ops.rebind_hw_stats(hw, &interface->stats); From 875328e4bce696e85edcda3c4b0ec80fd525e3a3 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:56:59 -0700 Subject: [PATCH 04/15] fm10k: reinitialize queuing scheme after calling init_hw The init_hw function may fail, and in the case of VFs, it might change the number of maximum queues available. Thus, for every flow which checks init_hw, we need to ensure that we clear the queue scheme before, and initialize it after. The fm10k_io_slot_reset path will end up triggering a reset so fm10k_reinit needs this change. The fm10k_io_error_detected and fm10k_io_resume also need to properly clear and reinitialize the queue scheme. Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index 9c21d1e45543..faf81492d009 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -159,6 +159,9 @@ static void fm10k_reinit(struct fm10k_intfc *interface) fm10k_mbx_free_irq(interface); + /* free interrupts */ + fm10k_clear_queueing_scheme(interface); + /* delay any future reset requests */ interface->last_reset = jiffies + (10 * HZ); @@ -175,6 +178,12 @@ static void fm10k_reinit(struct fm10k_intfc *interface) goto reinit_err; } + err = fm10k_init_queueing_scheme(interface); + if (err) { + dev_err(&interface->pdev->dev, "init_queueing_scheme failed: %d\n", err); + goto reinit_err; + } + /* reassociate interrupts */ fm10k_mbx_request_irq(interface); @@ -2198,6 +2207,9 @@ static pci_ers_result_t fm10k_io_error_detected(struct pci_dev *pdev, if (netif_running(netdev)) fm10k_close(netdev); + /* free interrupts */ + fm10k_clear_queueing_scheme(interface); + fm10k_mbx_free_irq(interface); pci_disable_device(pdev); @@ -2270,6 +2282,12 @@ static void fm10k_io_resume(struct pci_dev *pdev) /* reset statistics starting values */ hw->mac.ops.rebind_hw_stats(hw, &interface->stats); + err = fm10k_init_queueing_scheme(interface); + if (err) { + dev_err(&interface->pdev->dev, "init_queueing_scheme failed: %d\n", err); + return; + } + /* reassociate interrupts */ fm10k_mbx_request_irq(interface); From 9d4955b45888b1b7c9f2a954cf6aa1269904bb98 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:00 -0700 Subject: [PATCH 05/15] fm10k: Correct typecast in fm10k_update_xc_addr_pf Since the resultant data type of the mac_update.mac_upper field is u16, it does not make sense to typecast u8 variables to u32 first. Since we're modifying fm10k_pf.c, also update the copyright year. Reported-by: Matthew Vick Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pf.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c index 8c0bdc4e4edd..b1e09756a6ef 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c @@ -1,5 +1,5 @@ /* Intel Ethernet Switch Host Interface Driver - * Copyright(c) 2013 - 2014 Intel Corporation. + * Copyright(c) 2013 - 2015 Intel Corporation. * * This program is free software; you can redistribute it and/or modify it * under the terms and conditions of the GNU General Public License, @@ -334,8 +334,8 @@ static s32 fm10k_update_xc_addr_pf(struct fm10k_hw *hw, u16 glort, ((u32)mac[3] << 16) | ((u32)mac[4] << 8) | ((u32)mac[5])); - mac_update.mac_upper = cpu_to_le16(((u32)mac[0] << 8) | - ((u32)mac[1])); + mac_update.mac_upper = cpu_to_le16(((u16)mac[0] << 8) | + ((u16)mac[1])); mac_update.vlan = cpu_to_le16(vid); mac_update.glort = cpu_to_le16(glort); mac_update.action = add ? 0 : 1; From cdf32c94bd3569d4e46b8f993e0fd8e45d438d18 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:01 -0700 Subject: [PATCH 06/15] fm10k: explicitly typecast vlan values to u16 Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pf.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c index b1e09756a6ef..00f7a29e734f 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c @@ -1212,7 +1212,7 @@ s32 fm10k_iov_msg_mac_vlan_pf(struct fm10k_hw *hw, u32 **results, set = !(vid & FM10K_VLAN_CLEAR); vid &= ~FM10K_VLAN_CLEAR; - err = fm10k_iov_select_vid(vf_info, vid); + err = fm10k_iov_select_vid(vf_info, (u16)vid); if (err < 0) return err; else @@ -1242,7 +1242,7 @@ s32 fm10k_iov_msg_mac_vlan_pf(struct fm10k_hw *hw, u32 **results, if (err < 0) return err; else - vlan = err; + vlan = (u16)err; /* notify switch of request for new unicast address */ err = hw->mac.ops.update_uc_addr(hw, vf_info->glort, @@ -1268,7 +1268,7 @@ s32 fm10k_iov_msg_mac_vlan_pf(struct fm10k_hw *hw, u32 **results, if (err < 0) return err; else - vlan = err; + vlan = (u16)err; /* notify switch of request for new multicast address */ err = hw->mac.ops.update_mc_addr(hw, vf_info->glort, From 17d39fac0888bfd624f61f758c8cce60632a3394 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:02 -0700 Subject: [PATCH 07/15] fm10k: add statistics for actual DWORD count of mbmem mailbox A previous bug was uncovered by addition of a debug stat to indicate the actual number of DWORDS we pulled from the mbmem. It turned out this was not the same as the tx_dwords counter. While the previous bug fix should have corrected this in all cases, add some debug stats that count the number of DWORDs pushed or pulled from the mbmem. A future debugger may take advantage of this statistic for debugging purposes. Since we're modifying fm10k_mbx.h, update the copyright year as well. Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c | 2 ++ drivers/net/ethernet/intel/fm10k/fm10k_mbx.c | 4 ++++ drivers/net/ethernet/intel/fm10k/fm10k_mbx.h | 4 +++- 3 files changed, 9 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c index 2ce0eba5e040..7c0156112b5a 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c @@ -114,9 +114,11 @@ static const struct fm10k_stats fm10k_gstrings_mbx_stats[] = { FM10K_MBX_STAT("mbx_tx_oversized", tx_dropped), FM10K_MBX_STAT("mbx_tx_messages", tx_messages), FM10K_MBX_STAT("mbx_tx_dwords", tx_dwords), + FM10K_MBX_STAT("mbx_tx_mbmem_pulled", tx_mbmem_pulled), FM10K_MBX_STAT("mbx_rx_messages", rx_messages), FM10K_MBX_STAT("mbx_rx_dwords", rx_dwords), FM10K_MBX_STAT("mbx_rx_parse_err", rx_parse_err), + FM10K_MBX_STAT("mbx_rx_mbmem_pushed", rx_mbmem_pushed), }; #define FM10K_GLOBAL_STATS_LEN ARRAY_SIZE(fm10k_gstrings_global_stats) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_mbx.c b/drivers/net/ethernet/intel/fm10k/fm10k_mbx.c index af09a1b272e6..2bce47490723 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_mbx.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_mbx.c @@ -375,6 +375,8 @@ static void fm10k_mbx_write_copy(struct fm10k_hw *hw, if (!tail) tail++; + mbx->tx_mbmem_pulled++; + /* write message to hardware FIFO */ fm10k_write_reg(hw, mbmem + tail++, *(head++)); } while (--len && --end); @@ -459,6 +461,8 @@ static void fm10k_mbx_read_copy(struct fm10k_hw *hw, if (!head) head++; + mbx->rx_mbmem_pushed++; + /* read message from hardware FIFO */ *(tail++) = fm10k_read_reg(hw, mbmem + head++); } while (--len && --end); diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_mbx.h b/drivers/net/ethernet/intel/fm10k/fm10k_mbx.h index 0419a7f0035e..c4f18a8f176c 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_mbx.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k_mbx.h @@ -1,5 +1,5 @@ /* Intel Ethernet Switch Host Interface Driver - * Copyright(c) 2013 - 2014 Intel Corporation. + * Copyright(c) 2013 - 2015 Intel Corporation. * * This program is free software; you can redistribute it and/or modify it * under the terms and conditions of the GNU General Public License, @@ -291,8 +291,10 @@ struct fm10k_mbx_info { u64 tx_dropped; u64 tx_messages; u64 tx_dwords; + u64 tx_mbmem_pulled; u64 rx_messages; u64 rx_dwords; + u64 rx_mbmem_pushed; u64 rx_parse_err; /* Buffer to store messages */ From 5680ea692407ce5113f3391c2e6b5dbe6eec84d8 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:03 -0700 Subject: [PATCH 08/15] fm10k: rename mbx_tx_oversized statistic to mbx_tx_dropped Originally this statistic was renamed because the method of dropping was called "drop_oversized_messages", but this logic has changed much, and this counter does actually represent messages which we failed to transmit for a number of reasons. Rename the counter back to tx_dropped since this is when it will increment, and it is less confusing. Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c index 7c0156112b5a..fd29145a52e4 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c @@ -111,7 +111,7 @@ static const struct fm10k_stats fm10k_gstrings_pf_stats[] = { static const struct fm10k_stats fm10k_gstrings_mbx_stats[] = { FM10K_MBX_STAT("mbx_tx_busy", tx_busy), - FM10K_MBX_STAT("mbx_tx_oversized", tx_dropped), + FM10K_MBX_STAT("mbx_tx_dropped", tx_dropped), FM10K_MBX_STAT("mbx_tx_messages", tx_messages), FM10K_MBX_STAT("mbx_tx_dwords", tx_dwords), FM10K_MBX_STAT("mbx_tx_mbmem_pulled", tx_mbmem_pulled), From 20076fa18571e2e274f51a2012ff5ef5c82e93a5 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:05 -0700 Subject: [PATCH 09/15] fm10k: Add support for ITR scaling based on PCIe link speed The Intel Ethernet Switch FM10000 Host Interface interrupt throttle timers are based on the PCIe link speed. Because of this, the value being programmed into the ITR registers must be scaled accordingly. For the PF, this is as simple as reading the PCIe link speed and storing the result. However, in the case of SR-IOV, the VF's interrupt throttle timers are based on the link speed of the PF. However, the VF is unable to get the link speed information from its configuration space, so the PF must inform it of what scale to use. Rather than pass this scale via mailbox message, take advantage of unused bits in the TDLEN register to pass the scale. It is the responsibility of the PF to program this for the VF while setting up the VF queues and the responsibility of the VF to get the information accordingly. This is preferable because it allows the VF to set up the interrupts properly during initialization and matches how the MAC address is passed in the TDBAL/TDBAH registers. Since we're modifying fm10k_type.h, we may as well also update the copyright year. Reported-by: Matthew Vick Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pf.c | 22 ++++++++++++++++++- drivers/net/ethernet/intel/fm10k/fm10k_type.h | 17 +++++++++++++- drivers/net/ethernet/intel/fm10k/fm10k_vf.c | 20 +++++++++++++++-- 3 files changed, 55 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c index 00f7a29e734f..8b9b6ba5b92b 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pf.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pf.c @@ -150,19 +150,26 @@ static s32 fm10k_init_hw_pf(struct fm10k_hw *hw) FM10K_TPH_RXCTRL_HDR_WROEN); } - /* set max hold interval to align with 1.024 usec in all modes */ + /* set max hold interval to align with 1.024 usec in all modes and + * store ITR scale + */ switch (hw->bus.speed) { case fm10k_bus_speed_2500: dma_ctrl = FM10K_DMA_CTRL_MAX_HOLD_1US_GEN1; + hw->mac.itr_scale = FM10K_TDLEN_ITR_SCALE_GEN1; break; case fm10k_bus_speed_5000: dma_ctrl = FM10K_DMA_CTRL_MAX_HOLD_1US_GEN2; + hw->mac.itr_scale = FM10K_TDLEN_ITR_SCALE_GEN2; break; case fm10k_bus_speed_8000: dma_ctrl = FM10K_DMA_CTRL_MAX_HOLD_1US_GEN3; + hw->mac.itr_scale = FM10K_TDLEN_ITR_SCALE_GEN3; break; default: dma_ctrl = 0; + /* just in case, assume Gen3 ITR scale */ + hw->mac.itr_scale = FM10K_TDLEN_ITR_SCALE_GEN3; break; } @@ -903,6 +910,13 @@ static s32 fm10k_iov_assign_default_mac_vlan_pf(struct fm10k_hw *hw, fm10k_write_reg(hw, FM10K_TDBAL(vf_q_idx), tdbal); fm10k_write_reg(hw, FM10K_TDBAH(vf_q_idx), tdbah); + /* Provide the VF the ITR scale, using software-defined fields in TDLEN + * to pass the information during VF initialization. See definition of + * FM10K_TDLEN_ITR_SCALE_SHIFT for more details. + */ + fm10k_write_reg(hw, FM10K_TDLEN(vf_q_idx), hw->mac.itr_scale << + FM10K_TDLEN_ITR_SCALE_SHIFT); + err_out: /* configure Queue control register */ txqctl = ((u32)vf_vid << FM10K_TXQCTL_VID_SHIFT) & @@ -1035,6 +1049,12 @@ static s32 fm10k_iov_reset_resources_pf(struct fm10k_hw *hw, for (i = queues_per_pool; i--;) { fm10k_write_reg(hw, FM10K_TDBAL(vf_q_idx + i), tdbal); fm10k_write_reg(hw, FM10K_TDBAH(vf_q_idx + i), tdbah); + /* See definition of FM10K_TDLEN_ITR_SCALE_SHIFT for an + * explanation of how TDLEN is used. + */ + fm10k_write_reg(hw, FM10K_TDLEN(vf_q_idx + i), + hw->mac.itr_scale << + FM10K_TDLEN_ITR_SCALE_SHIFT); fm10k_write_reg(hw, FM10K_TQMAP(qmap_idx + i), vf_q_idx + i); fm10k_write_reg(hw, FM10K_RQMAP(qmap_idx + i), vf_q_idx + i); } diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_type.h b/drivers/net/ethernet/intel/fm10k/fm10k_type.h index 35afd711d144..02727250ce1f 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_type.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k_type.h @@ -1,5 +1,5 @@ /* Intel Ethernet Switch Host Interface Driver - * Copyright(c) 2013 - 2014 Intel Corporation. + * Copyright(c) 2013 - 2015 Intel Corporation. * * This program is free software; you can redistribute it and/or modify it * under the terms and conditions of the GNU General Public License, @@ -272,6 +272,20 @@ struct fm10k_hw; #define FM10K_TDBAL(_n) ((0x40 * (_n)) + 0x8000) #define FM10K_TDBAH(_n) ((0x40 * (_n)) + 0x8001) #define FM10K_TDLEN(_n) ((0x40 * (_n)) + 0x8002) +/* When fist initialized, VFs need to know the Interrupt Throttle Rate (ITR) + * scale which is based on the PCIe speed but the speed information in the PCI + * configuration space may not be accurate. The PF already knows the ITR scale + * but there is no defined method to pass that information from the PF to the + * VF. This is accomplished during VF initialization by temporarily co-opting + * the yet-to-be-used TDLEN register to have the PF store the ITR shift for + * the VF to retrieve before the VF needs to use the TDLEN register for its + * intended purpose, i.e. before the Tx resources are allocated. + */ +#define FM10K_TDLEN_ITR_SCALE_SHIFT 9 +#define FM10K_TDLEN_ITR_SCALE_MASK 0x00000E00 +#define FM10K_TDLEN_ITR_SCALE_GEN1 2 +#define FM10K_TDLEN_ITR_SCALE_GEN2 1 +#define FM10K_TDLEN_ITR_SCALE_GEN3 0 #define FM10K_TPH_TXCTRL(_n) ((0x40 * (_n)) + 0x8003) #define FM10K_TPH_TXCTRL_DESC_TPHEN 0x00000020 #define FM10K_TPH_TXCTRL_DESC_RROEN 0x00000200 @@ -560,6 +574,7 @@ struct fm10k_mac_info { bool get_host_state; bool tx_ready; u32 dglort_map; + u8 itr_scale; }; struct fm10k_swapi_table_info { diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_vf.c b/drivers/net/ethernet/intel/fm10k/fm10k_vf.c index d512575c33f3..2af697df5abc 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_vf.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_vf.c @@ -28,7 +28,7 @@ static s32 fm10k_stop_hw_vf(struct fm10k_hw *hw) { u8 *perm_addr = hw->mac.perm_addr; - u32 bal = 0, bah = 0; + u32 bal = 0, bah = 0, tdlen; s32 err; u16 i; @@ -48,6 +48,9 @@ static s32 fm10k_stop_hw_vf(struct fm10k_hw *hw) ((u32)perm_addr[2]); } + /* restore default itr_scale for next VF initialization */ + tdlen = hw->mac.itr_scale << FM10K_TDLEN_ITR_SCALE_SHIFT; + /* The queues have already been disabled so we just need to * update their base address registers */ @@ -56,6 +59,12 @@ static s32 fm10k_stop_hw_vf(struct fm10k_hw *hw) fm10k_write_reg(hw, FM10K_TDBAH(i), bah); fm10k_write_reg(hw, FM10K_RDBAL(i), bal); fm10k_write_reg(hw, FM10K_RDBAH(i), bah); + /* Restore ITR scale in software-defined mechanism in TDLEN + * for next VF initialization. See definition of + * FM10K_TDLEN_ITR_SCALE_SHIFT for more details on the use of + * TDLEN here. + */ + fm10k_write_reg(hw, FM10K_TDLEN(i), tdlen); } return 0; @@ -131,9 +140,16 @@ static s32 fm10k_init_hw_vf(struct fm10k_hw *hw) /* record maximum queue count */ hw->mac.max_queues = i; - /* fetch default VLAN */ + /* fetch default VLAN and ITR scale */ hw->mac.default_vid = (fm10k_read_reg(hw, FM10K_TXQCTL(0)) & FM10K_TXQCTL_VID_MASK) >> FM10K_TXQCTL_VID_SHIFT; + /* Read the ITR scale from TDLEN. See the definition of + * FM10K_TDLEN_ITR_SCALE_SHIFT for more information about how TDLEN is + * used here. + */ + hw->mac.itr_scale = (fm10k_read_reg(hw, FM10K_TDLEN(0)) & + FM10K_TDLEN_ITR_SCALE_MASK) >> + FM10K_TDLEN_ITR_SCALE_SHIFT; return 0; From 584373f5b98aed81ff5a432d91b6e16d7554a5c9 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:06 -0700 Subject: [PATCH 10/15] fm10k: introduce ITR_IS_ADAPTIVE macro Define a macro for identifying when the itr value is dynamic or adaptive. The concept was taken from i40e. This helps make clear what the check is, and reduces the line length to something more reasonable in a few places. Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k.h | 2 ++ drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c | 6 ++---- drivers/net/ethernet/intel/fm10k/fm10k_main.c | 2 +- 3 files changed, 5 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k.h b/drivers/net/ethernet/intel/fm10k/fm10k.h index b178905fa43e..cea0d94d8f5f 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k.h @@ -172,6 +172,8 @@ struct fm10k_ring_container { #define FM10K_ITR_20K 50 /* 50us */ #define FM10K_ITR_ADAPTIVE 0x8000 /* adaptive interrupt moderation flag */ +#define ITR_IS_ADAPTIVE(itr) (!!(itr & FM10K_ITR_ADAPTIVE)) + #define FM10K_ITR_ENABLE (FM10K_ITR_AUTOMASK | FM10K_ITR_MASK_CLEAR) static inline struct netdev_queue *txring_txq(const struct fm10k_ring *ring) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c index fd29145a52e4..a505a502bb8e 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c @@ -701,12 +701,10 @@ static int fm10k_get_coalesce(struct net_device *dev, { struct fm10k_intfc *interface = netdev_priv(dev); - ec->use_adaptive_tx_coalesce = - !!(interface->tx_itr & FM10K_ITR_ADAPTIVE); + ec->use_adaptive_tx_coalesce = ITR_IS_ADAPTIVE(interface->tx_itr); ec->tx_coalesce_usecs = interface->tx_itr & ~FM10K_ITR_ADAPTIVE; - ec->use_adaptive_rx_coalesce = - !!(interface->rx_itr & FM10K_ITR_ADAPTIVE); + ec->use_adaptive_rx_coalesce = ITR_IS_ADAPTIVE(interface->rx_itr); ec->rx_coalesce_usecs = interface->rx_itr & ~FM10K_ITR_ADAPTIVE; return 0; diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_main.c b/drivers/net/ethernet/intel/fm10k/fm10k_main.c index 746a1986690b..21d49550935e 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_main.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_main.c @@ -1366,7 +1366,7 @@ static void fm10k_update_itr(struct fm10k_ring_container *ring_container) unsigned int avg_wire_size, packets; /* Only update ITR if we are using adaptive setting */ - if (!(ring_container->itr & FM10K_ITR_ADAPTIVE)) + if (!ITR_IS_ADAPTIVE(ring_container->itr)) goto clear_counts; packets = ring_container->total_packets; From 242722dd3d0af32703d4ebb4af63c92a2c85b835 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:07 -0700 Subject: [PATCH 11/15] fm10k: Update adaptive ITR algorithm The existing adaptive ITR algorithm is overly restrictive. It throttles incorrectly for various traffic rates, and does not produce good performance. The algorithm now allows for more interrupts per second, and does some calculation to help improve for smaller packet loads. In addition, take into account the new itr_scale from the hardware which indicates how much to scale due to PCIe link speed. Reported-by: Matthew Vick Reported-by: Alex Duyck Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k.h | 1 + drivers/net/ethernet/intel/fm10k/fm10k_main.c | 52 ++++++++++++++----- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 6 ++- 3 files changed, 45 insertions(+), 14 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k.h b/drivers/net/ethernet/intel/fm10k/fm10k.h index cea0d94d8f5f..bb799b368996 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k.h @@ -164,6 +164,7 @@ struct fm10k_ring_container { unsigned int total_packets; /* total packets processed this int */ u16 work_limit; /* total work allowed per interrupt */ u16 itr; /* interrupt throttle rate value */ + u8 itr_scale; /* ITR adjustment scaler based on PCI speed */ u8 count; /* total number of rings in vector */ }; diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_main.c b/drivers/net/ethernet/intel/fm10k/fm10k_main.c index 21d49550935e..d422cd187101 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_main.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_main.c @@ -1363,7 +1363,7 @@ static bool fm10k_clean_tx_irq(struct fm10k_q_vector *q_vector, **/ static void fm10k_update_itr(struct fm10k_ring_container *ring_container) { - unsigned int avg_wire_size, packets; + unsigned int avg_wire_size, packets, itr_round; /* Only update ITR if we are using adaptive setting */ if (!ITR_IS_ADAPTIVE(ring_container->itr)) @@ -1375,18 +1375,44 @@ static void fm10k_update_itr(struct fm10k_ring_container *ring_container) avg_wire_size = ring_container->total_bytes / packets; - /* Add 24 bytes to size to account for CRC, preamble, and gap */ - avg_wire_size += 24; + /* The following is a crude approximation of: + * wmem_default / (size + overhead) = desired_pkts_per_int + * rate / bits_per_byte / (size + ethernet overhead) = pkt_rate + * (desired_pkt_rate / pkt_rate) * usecs_per_sec = ITR value + * + * Assuming wmem_default is 212992 and overhead is 640 bytes per + * packet, (256 skb, 64 headroom, 320 shared info), we can reduce the + * formula down to + * + * (34 * (size + 24)) / (size + 640) = ITR + * + * We first do some math on the packet size and then finally bitshift + * by 8 after rounding up. We also have to account for PCIe link speed + * difference as ITR scales based on this. + */ + if (avg_wire_size <= 360) { + /* Start at 250K ints/sec and gradually drop to 77K ints/sec */ + avg_wire_size *= 8; + avg_wire_size += 376; + } else if (avg_wire_size <= 1152) { + /* 77K ints/sec to 45K ints/sec */ + avg_wire_size *= 3; + avg_wire_size += 2176; + } else if (avg_wire_size <= 1920) { + /* 45K ints/sec to 38K ints/sec */ + avg_wire_size += 4480; + } else { + /* plateau at a limit of 38K ints/sec */ + avg_wire_size = 6656; + } - /* Don't starve jumbo frames */ - if (avg_wire_size > 3000) - avg_wire_size = 3000; - - /* Give a little boost to mid-size frames */ - if ((avg_wire_size > 300) && (avg_wire_size < 1200)) - avg_wire_size /= 3; - else - avg_wire_size /= 2; + /* Perform final bitshift for division after rounding up to ensure + * that the calculation will never get below a 1. The bit shift + * accounts for changes in the ITR due to PCIe link speed. + */ + itr_round = ACCESS_ONCE(ring_container->itr_scale) + 8; + avg_wire_size += (1 << itr_round) - 1; + avg_wire_size >>= itr_round; /* write back value and retain adaptive flag */ ring_container->itr = avg_wire_size | FM10K_ITR_ADAPTIVE; @@ -1604,6 +1630,7 @@ static int fm10k_alloc_q_vector(struct fm10k_intfc *interface, q_vector->tx.ring = ring; q_vector->tx.work_limit = FM10K_DEFAULT_TX_WORK; q_vector->tx.itr = interface->tx_itr; + q_vector->tx.itr_scale = interface->hw.mac.itr_scale; q_vector->tx.count = txr_count; while (txr_count) { @@ -1632,6 +1659,7 @@ static int fm10k_alloc_q_vector(struct fm10k_intfc *interface, /* save Rx ring container info */ q_vector->rx.ring = ring; q_vector->rx.itr = interface->rx_itr; + q_vector->rx.itr_scale = interface->hw.mac.itr_scale; q_vector->rx.count = rxr_count; while (rxr_count) { diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index faf81492d009..1f0d9bc3bf7b 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -880,7 +880,8 @@ static irqreturn_t fm10k_msix_mbx_vf(int __always_unused irq, void *data) /* re-enable mailbox interrupt and indicate 20us delay */ fm10k_write_reg(hw, FM10K_VFITR(FM10K_MBX_VECTOR), - FM10K_ITR_ENABLE | FM10K_MBX_INT_DELAY); + FM10K_ITR_ENABLE | (FM10K_MBX_INT_DELAY >> + hw->mac.itr_scale)); /* service upstream mailbox */ if (fm10k_mbx_trylock(interface)) { @@ -1111,7 +1112,8 @@ static irqreturn_t fm10k_msix_mbx_pf(int __always_unused irq, void *data) /* re-enable mailbox interrupt and indicate 20us delay */ fm10k_write_reg(hw, FM10K_ITR(FM10K_MBX_VECTOR), - FM10K_ITR_ENABLE | FM10K_MBX_INT_DELAY); + FM10K_ITR_ENABLE | (FM10K_MBX_INT_DELAY >> + hw->mac.itr_scale)); return IRQ_HANDLED; } From 436ea956bffd9974bc41ae9bd99930e29d9c807d Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:08 -0700 Subject: [PATCH 12/15] fm10k: use macro for default Tx and Rx ITR values Signed-off-by: Jacob Keller Reviewed-by: Bruce Allan Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k.h | 2 ++ drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c | 4 ++-- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 4 ++-- 3 files changed, 6 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k.h b/drivers/net/ethernet/intel/fm10k/fm10k.h index bb799b368996..115033f6539e 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k.h @@ -175,6 +175,8 @@ struct fm10k_ring_container { #define ITR_IS_ADAPTIVE(itr) (!!(itr & FM10K_ITR_ADAPTIVE)) +#define FM10K_TX_ITR_DEFAULT FM10K_ITR_10K +#define FM10K_RX_ITR_DEFAULT FM10K_ITR_20K #define FM10K_ITR_ENABLE (FM10K_ITR_AUTOMASK | FM10K_ITR_MASK_CLEAR) static inline struct netdev_queue *txring_txq(const struct fm10k_ring *ring) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c index a505a502bb8e..109e2111bdda 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_ethtool.c @@ -729,10 +729,10 @@ static int fm10k_set_coalesce(struct net_device *dev, /* set initial values for adaptive ITR */ if (ec->use_adaptive_tx_coalesce) - tx_itr = FM10K_ITR_ADAPTIVE | FM10K_ITR_10K; + tx_itr = FM10K_ITR_ADAPTIVE | FM10K_TX_ITR_DEFAULT; if (ec->use_adaptive_rx_coalesce) - rx_itr = FM10K_ITR_ADAPTIVE | FM10K_ITR_20K; + rx_itr = FM10K_ITR_ADAPTIVE | FM10K_RX_ITR_DEFAULT; /* update interface */ interface->tx_itr = tx_itr; diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index 1f0d9bc3bf7b..0cdb79e8b36e 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -1771,8 +1771,8 @@ static int fm10k_sw_init(struct fm10k_intfc *interface, interface->rx_ring_count = FM10K_DEFAULT_RXD; /* set default interrupt moderation */ - interface->tx_itr = FM10K_ITR_10K; - interface->rx_itr = FM10K_ITR_ADAPTIVE | FM10K_ITR_20K; + interface->tx_itr = FM10K_TX_ITR_DEFAULT; + interface->rx_itr = FM10K_ITR_ADAPTIVE | FM10K_RX_ITR_DEFAULT; /* initialize vxlan_port list */ INIT_LIST_HEAD(&interface->vxlan_port); From dbf4284886c59a17dc0081cee39039ac2a546078 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:09 -0700 Subject: [PATCH 13/15] fm10k: change default Tx ITR to 25usec The current default ITR for Tx is overly restrictive. Using a simple netperf TCP_STREAM test, we top out at about 10Gb/s for a single thread when running using 1500 byte frames. By reducing the ITR value to 25usec (up to 40K interrupts a second from 10K), we are able to achieve 36Gb/s for a single thread TCP stream test. Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k.h b/drivers/net/ethernet/intel/fm10k/fm10k.h index 115033f6539e..fa26e20445a5 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k.h +++ b/drivers/net/ethernet/intel/fm10k/fm10k.h @@ -171,11 +171,12 @@ struct fm10k_ring_container { #define FM10K_ITR_MAX 0x0FFF /* maximum value for ITR */ #define FM10K_ITR_10K 100 /* 100us */ #define FM10K_ITR_20K 50 /* 50us */ +#define FM10K_ITR_40K 25 /* 25us */ #define FM10K_ITR_ADAPTIVE 0x8000 /* adaptive interrupt moderation flag */ #define ITR_IS_ADAPTIVE(itr) (!!(itr & FM10K_ITR_ADAPTIVE)) -#define FM10K_TX_ITR_DEFAULT FM10K_ITR_10K +#define FM10K_TX_ITR_DEFAULT FM10K_ITR_40K #define FM10K_RX_ITR_DEFAULT FM10K_ITR_20K #define FM10K_ITR_ENABLE (FM10K_ITR_AUTOMASK | FM10K_ITR_MASK_CLEAR) From c7bc952349c39d8ec7b1a7f6ef403be2a08d5e86 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:10 -0700 Subject: [PATCH 14/15] fm10k: TRIVIAL fix typo of hardware Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_pci.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c index 0cdb79e8b36e..15d8e10c2504 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_pci.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_pci.c @@ -584,7 +584,7 @@ static void fm10k_configure_tx_ring(struct fm10k_intfc *interface, /* store tail pointer */ ring->tail = &interface->uc_addr[FM10K_TDT(reg_idx)]; - /* reset ntu and ntc to place SW in sync with hardwdare */ + /* reset ntu and ntc to place SW in sync with hardware */ ring->next_to_clean = 0; ring->next_to_use = 0; @@ -690,7 +690,7 @@ static void fm10k_configure_rx_ring(struct fm10k_intfc *interface, /* store tail pointer */ ring->tail = &interface->uc_addr[FM10K_RDT(reg_idx)]; - /* reset ntu and ntc to place SW in sync with hardwdare */ + /* reset ntu and ntc to place SW in sync with hardware */ ring->next_to_clean = 0; ring->next_to_use = 0; ring->next_to_alloc = 0; From 03d13a51fb4494f3bf47f65e2be00e56c36d2b63 Mon Sep 17 00:00:00 2001 From: Jacob Keller Date: Fri, 16 Oct 2015 10:57:11 -0700 Subject: [PATCH 15/15] fm10k: TRIVIAL cleanup order at top of fm10k_xmit_frame Signed-off-by: Jacob Keller Tested-by: Krishneil Singh Signed-off-by: Jeff Kirsher --- drivers/net/ethernet/intel/fm10k/fm10k_main.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_main.c b/drivers/net/ethernet/intel/fm10k/fm10k_main.c index d422cd187101..1c17b6284daa 100644 --- a/drivers/net/ethernet/intel/fm10k/fm10k_main.c +++ b/drivers/net/ethernet/intel/fm10k/fm10k_main.c @@ -1094,11 +1094,11 @@ dma_error: netdev_tx_t fm10k_xmit_frame_ring(struct sk_buff *skb, struct fm10k_ring *tx_ring) { - struct fm10k_tx_buffer *first; - int tso; - u32 tx_flags = 0; - unsigned short f; u16 count = TXD_USE_COUNT(skb_headlen(skb)); + struct fm10k_tx_buffer *first; + unsigned short f; + u32 tx_flags = 0; + int tso; /* need: 1 descriptor per page * PAGE_SIZE/FM10K_MAX_DATA_PER_TXD, * + 1 desc for skb_headlen/FM10K_MAX_DATA_PER_TXD,