From bdc8cbd34d5d9d82c3db4df6a182b0530040062d Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:39 -0400 Subject: qede: get_channels() need to populate max tx/rx coalesce values Recent changes in kernel ethtool implementation requires the driver callback for get_channels() has to populate the values for max tx/rx coalesce fields. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c index 25a9b293ee8f..16a1a478a2bd 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c @@ -756,6 +756,8 @@ static void qede_get_channels(struct net_device *dev, struct qede_dev *edev = netdev_priv(dev); channels->max_combined = QEDE_MAX_RSS_CNT(edev); + channels->max_rx = QEDE_MAX_RSS_CNT(edev); + channels->max_tx = QEDE_MAX_RSS_CNT(edev); channels->combined_count = QEDE_QUEUE_CNT(edev) - edev->fp_num_tx - edev->fp_num_rx; channels->tx_count = edev->fp_num_tx; -- cgit v1.2.3 From ba300ce35123a9ba8ea55318b0f579cc1ab3f51a Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:40 -0400 Subject: qede: Do not allow RSS config for 100G devices RSS configuration is not supported for 100G adapters. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c index 16a1a478a2bd..bf04fb9e1ae6 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c @@ -1055,6 +1055,12 @@ static int qede_set_rxfh(struct net_device *dev, const u32 *indir, struct qede_dev *edev = netdev_priv(dev); int i; + if (edev->dev_info.common.num_hwfns > 1) { + DP_INFO(edev, + "RSS configuration is not supported for 100G devices\n"); + return -EOPNOTSUPP; + } + if (hfunc != ETH_RSS_HASH_NO_CHANGE && hfunc != ETH_RSS_HASH_TOP) return -EOPNOTSUPP; -- cgit v1.2.3 From 837d4eb6ed7cb0341079fac97e3037df6bef7482 Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:41 -0400 Subject: qede: Loopback implementation should ignore the normal traffic During the execution of loopback test, driver may receive the packets which are not originated by this test, loopback implementation need to skip those packets. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qede/qede.h | 1 + drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 80 +++++++++++++++---------- drivers/net/ethernet/qlogic/qede/qede_main.c | 3 +- 3 files changed, 49 insertions(+), 35 deletions(-) diff --git a/drivers/net/ethernet/qlogic/qede/qede.h b/drivers/net/ethernet/qlogic/qede/qede.h index 28c0e9f42c9e..d6a4a9a5eb4d 100644 --- a/drivers/net/ethernet/qlogic/qede/qede.h +++ b/drivers/net/ethernet/qlogic/qede/qede.h @@ -348,6 +348,7 @@ bool qede_has_rx_work(struct qede_rx_queue *rxq); int qede_txq_has_work(struct qede_tx_queue *txq); void qede_recycle_rx_bd_ring(struct qede_rx_queue *rxq, struct qede_dev *edev, u8 count); +void qede_update_rx_prod(struct qede_dev *edev, struct qede_rx_queue *rxq); #define RX_RING_SIZE_POW 13 #define RX_RING_SIZE ((u16)BIT(RX_RING_SIZE_POW)) diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c index bf04fb9e1ae6..7da184cfe72e 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c @@ -1207,8 +1207,8 @@ static int qede_selftest_receive_traffic(struct qede_dev *edev) struct qede_rx_queue *rxq = NULL; struct sw_rx_data *sw_rx_data; union eth_rx_cqe *cqe; + int i, rc = 0; u8 *data_ptr; - int i; for_each_queue(i) { if (edev->fp_array[i].type & QEDE_FASTPATH_RX) { @@ -1227,46 +1227,60 @@ static int qede_selftest_receive_traffic(struct qede_dev *edev) * queue and that the loopback traffic is not IP. */ for (i = 0; i < QEDE_SELFTEST_POLL_COUNT; i++) { - if (qede_has_rx_work(rxq)) + if (!qede_has_rx_work(rxq)) { + usleep_range(100, 200); + continue; + } + + hw_comp_cons = le16_to_cpu(*rxq->hw_cons_ptr); + sw_comp_cons = qed_chain_get_cons_idx(&rxq->rx_comp_ring); + + /* Memory barrier to prevent the CPU from doing speculative + * reads of CQE/BD before reading hw_comp_cons. If the CQE is + * read before it is written by FW, then FW writes CQE and SB, + * and then the CPU reads the hw_comp_cons, it will use an old + * CQE. + */ + rmb(); + + /* Get the CQE from the completion ring */ + cqe = (union eth_rx_cqe *)qed_chain_consume(&rxq->rx_comp_ring); + + /* Get the data from the SW ring */ + sw_rx_index = rxq->sw_rx_cons & NUM_RX_BDS_MAX; + sw_rx_data = &rxq->sw_rx_ring[sw_rx_index]; + fp_cqe = &cqe->fast_path_regular; + len = le16_to_cpu(fp_cqe->len_on_first_bd); + data_ptr = (u8 *)(page_address(sw_rx_data->data) + + fp_cqe->placement_offset + + sw_rx_data->page_offset); + if (ether_addr_equal(data_ptr, edev->ndev->dev_addr) && + ether_addr_equal(data_ptr + ETH_ALEN, + edev->ndev->dev_addr)) { + for (i = ETH_HLEN; i < len; i++) + if (data_ptr[i] != (unsigned char)(i & 0xff)) { + rc = -1; + break; + } + + qede_recycle_rx_bd_ring(rxq, edev, 1); + qed_chain_recycle_consumed(&rxq->rx_comp_ring); break; - usleep_range(100, 200); + } + + DP_INFO(edev, "Not the transmitted packet\n"); + qede_recycle_rx_bd_ring(rxq, edev, 1); + qed_chain_recycle_consumed(&rxq->rx_comp_ring); } - if (!qede_has_rx_work(rxq)) { + if (i == QEDE_SELFTEST_POLL_COUNT) { DP_NOTICE(edev, "Failed to receive the traffic\n"); return -1; } - hw_comp_cons = le16_to_cpu(*rxq->hw_cons_ptr); - sw_comp_cons = qed_chain_get_cons_idx(&rxq->rx_comp_ring); - - /* Memory barrier to prevent the CPU from doing speculative reads of CQE - * / BD before reading hw_comp_cons. If the CQE is read before it is - * written by FW, then FW writes CQE and SB, and then the CPU reads the - * hw_comp_cons, it will use an old CQE. - */ - rmb(); - - /* Get the CQE from the completion ring */ - cqe = (union eth_rx_cqe *)qed_chain_consume(&rxq->rx_comp_ring); - - /* Get the data from the SW ring */ - sw_rx_index = rxq->sw_rx_cons & NUM_RX_BDS_MAX; - sw_rx_data = &rxq->sw_rx_ring[sw_rx_index]; - fp_cqe = &cqe->fast_path_regular; - len = le16_to_cpu(fp_cqe->len_on_first_bd); - data_ptr = (u8 *)(page_address(sw_rx_data->data) + - fp_cqe->placement_offset + sw_rx_data->page_offset); - for (i = ETH_HLEN; i < len; i++) - if (data_ptr[i] != (unsigned char)(i & 0xff)) { - DP_NOTICE(edev, "Loopback test failed\n"); - qede_recycle_rx_bd_ring(rxq, edev, 1); - return -1; - } - - qede_recycle_rx_bd_ring(rxq, edev, 1); + qede_update_rx_prod(edev, rxq); - return 0; + return rc; } static int qede_selftest_run_loopback(struct qede_dev *edev, u32 loopback_mode) diff --git a/drivers/net/ethernet/qlogic/qede/qede_main.c b/drivers/net/ethernet/qlogic/qede/qede_main.c index 343038ca047d..da8ef69cedd6 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_main.c +++ b/drivers/net/ethernet/qlogic/qede/qede_main.c @@ -943,8 +943,7 @@ static inline int qede_realloc_rx_buffer(struct qede_dev *edev, return 0; } -static inline void qede_update_rx_prod(struct qede_dev *edev, - struct qede_rx_queue *rxq) +void qede_update_rx_prod(struct qede_dev *edev, struct qede_rx_queue *rxq) { u16 bd_prod = qed_chain_get_prod_idx(&rxq->rx_bd_ring); u16 cqe_prod = qed_chain_get_prod_idx(&rxq->rx_comp_ring); -- cgit v1.2.3 From 0e191827383a6503a3bc547e63c74ff093f450f5 Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:42 -0400 Subject: qed*: Reduce the memory footprint for Rx path With the current default values for Rx path i.e., 8 queues of 8Kb entries each with 4Kb size, interface will consume 256Mb for Rx. The default values causing the driver probe to fail when the system memory is low. Based on the perforamnce results, rx-ring count value of 1Kb gives the comparable performance with Rx coalesce timeout of 12 seconds. Updating the default values. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qed/qed_main.c | 1 + drivers/net/ethernet/qlogic/qede/qede.h | 2 +- include/linux/qed/qed_if.h | 1 + 3 files changed, 3 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c index 8dc3f4670f64..c418360ba02a 100644 --- a/drivers/net/ethernet/qlogic/qed/qed_main.c +++ b/drivers/net/ethernet/qlogic/qed/qed_main.c @@ -878,6 +878,7 @@ static int qed_slowpath_start(struct qed_dev *cdev, } } + cdev->rx_coalesce_usecs = QED_DEFAULT_RX_USECS; rc = qed_nic_setup(cdev); if (rc) goto err; diff --git a/drivers/net/ethernet/qlogic/qede/qede.h b/drivers/net/ethernet/qlogic/qede/qede.h index d6a4a9a5eb4d..974689a13337 100644 --- a/drivers/net/ethernet/qlogic/qede/qede.h +++ b/drivers/net/ethernet/qlogic/qede/qede.h @@ -354,7 +354,7 @@ void qede_update_rx_prod(struct qede_dev *edev, struct qede_rx_queue *rxq); #define RX_RING_SIZE ((u16)BIT(RX_RING_SIZE_POW)) #define NUM_RX_BDS_MAX (RX_RING_SIZE - 1) #define NUM_RX_BDS_MIN 128 -#define NUM_RX_BDS_DEF NUM_RX_BDS_MAX +#define NUM_RX_BDS_DEF ((u16)BIT(10) - 1) #define TX_RING_SIZE_POW 13 #define TX_RING_SIZE ((u16)BIT(TX_RING_SIZE_POW)) diff --git a/include/linux/qed/qed_if.h b/include/linux/qed/qed_if.h index f9ae903bbb84..8978a60371f4 100644 --- a/include/linux/qed/qed_if.h +++ b/include/linux/qed/qed_if.h @@ -146,6 +146,7 @@ enum qed_led_mode { #define DIRECT_REG_RD(reg_addr) readl((void __iomem *)(reg_addr)) #define QED_COALESCE_MAX 0xFF +#define QED_DEFAULT_RX_USECS 12 /* forward */ struct qed_dev; -- cgit v1.2.3 From ed0dd91515bbe19a19cfccca366cf163ed581cac Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:43 -0400 Subject: qede: Reconfigure rss indirection direction table when rss count is updated Rx indirection table entries are in the range [0, (rss_count - 1)]. If user reduces the rss count, the table entries may not be in the ccorrect range. Need to reconfigure the table with new rss_count as a basis. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c index 7da184cfe72e..d47bdaa1f9a8 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c @@ -822,6 +822,13 @@ static int qede_set_channels(struct net_device *dev, edev->req_queues = count; edev->req_num_tx = channels->tx_count; edev->req_num_rx = channels->rx_count; + /* Reset the indirection table if rx queue count is updated */ + if ((edev->req_queues - edev->req_num_tx) != QEDE_RSS_COUNT(edev)) { + edev->rss_params_inited &= ~QEDE_RSS_INDIR_INITED; + memset(&edev->rss_params.rss_ind_table, 0, + sizeof(edev->rss_params.rss_ind_table)); + } + if (netif_running(dev)) qede_reload(edev, NULL, NULL); -- cgit v1.2.3 From 15c6de2c65048d45484e396194ba2598618d4cb4 Mon Sep 17 00:00:00 2001 From: Sudarsana Reddy Kalluru Date: Fri, 21 Oct 2016 04:43:44 -0400 Subject: qed: Zero-out the buffer paased to dcbx_query() API qed_dcbx_query_params() implementation populate the values to input buffer based on the dcbx mode and, the current negotiated state/params, the caller of this API need to memset the buffer to zero. Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: Yuval Mintz Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qed/qed_dcbx.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/ethernet/qlogic/qed/qed_dcbx.c b/drivers/net/ethernet/qlogic/qed/qed_dcbx.c index 130da1c0490b..a4789a93b692 100644 --- a/drivers/net/ethernet/qlogic/qed/qed_dcbx.c +++ b/drivers/net/ethernet/qlogic/qed/qed_dcbx.c @@ -1190,6 +1190,7 @@ int qed_dcbx_get_config_params(struct qed_hwfn *p_hwfn, if (!dcbx_info) return -ENOMEM; + memset(dcbx_info, 0, sizeof(*dcbx_info)); rc = qed_dcbx_query_params(p_hwfn, dcbx_info, QED_DCBX_OPERATIONAL_MIB); if (rc) { kfree(dcbx_info); @@ -1225,6 +1226,7 @@ static struct qed_dcbx_get *qed_dcbnl_get_dcbx(struct qed_hwfn *hwfn, if (!dcbx_info) return NULL; + memset(dcbx_info, 0, sizeof(*dcbx_info)); if (qed_dcbx_query_params(hwfn, dcbx_info, type)) { kfree(dcbx_info); return NULL; -- cgit v1.2.3 From fabd545c6d27ac1977fe567c43cd4c72fad04172 Mon Sep 17 00:00:00 2001 From: Manish Chopra Date: Fri, 21 Oct 2016 04:43:45 -0400 Subject: qede: Fix incorrrect usage of APIs for un-mapping DMA memory Driver uses incorrect APIs to unmap DMA memory which were mapped using dma_map_single(). This patch fixes it to use appropriate APIs for un-mapping DMA memory. Signed-off-by: Manish Chopra Signed-off-by: Yuval Mintz Signed-off-by: Sudarsana Reddy Kalluru Signed-off-by: David S. Miller --- drivers/net/ethernet/qlogic/qede/qede_ethtool.c | 4 ++-- drivers/net/ethernet/qlogic/qede/qede_main.c | 8 ++++---- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c index d47bdaa1f9a8..12251a1032d1 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_ethtool.c +++ b/drivers/net/ethernet/qlogic/qede/qede_ethtool.c @@ -1199,8 +1199,8 @@ static int qede_selftest_transmit_traffic(struct qede_dev *edev, } first_bd = (struct eth_tx_1st_bd *)qed_chain_consume(&txq->tx_pbl); - dma_unmap_page(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), - BD_UNMAP_LEN(first_bd), DMA_TO_DEVICE); + dma_unmap_single(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), + BD_UNMAP_LEN(first_bd), DMA_TO_DEVICE); txq->sw_tx_cons++; txq->sw_tx_ring[idx].skb = NULL; diff --git a/drivers/net/ethernet/qlogic/qede/qede_main.c b/drivers/net/ethernet/qlogic/qede/qede_main.c index da8ef69cedd6..444b271059b2 100644 --- a/drivers/net/ethernet/qlogic/qede/qede_main.c +++ b/drivers/net/ethernet/qlogic/qede/qede_main.c @@ -313,8 +313,8 @@ static int qede_free_tx_pkt(struct qede_dev *edev, split_bd_len = BD_UNMAP_LEN(split); bds_consumed++; } - dma_unmap_page(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), - BD_UNMAP_LEN(first_bd) + split_bd_len, DMA_TO_DEVICE); + dma_unmap_single(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), + BD_UNMAP_LEN(first_bd) + split_bd_len, DMA_TO_DEVICE); /* Unmap the data of the skb frags */ for (i = 0; i < skb_shinfo(skb)->nr_frags; i++, bds_consumed++) { @@ -359,8 +359,8 @@ static void qede_free_failed_tx_pkt(struct qede_dev *edev, nbd--; } - dma_unmap_page(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), - BD_UNMAP_LEN(first_bd) + split_bd_len, DMA_TO_DEVICE); + dma_unmap_single(&edev->pdev->dev, BD_UNMAP_ADDR(first_bd), + BD_UNMAP_LEN(first_bd) + split_bd_len, DMA_TO_DEVICE); /* Unmap the data of the skb frags */ for (i = 0; i < nbd; i++) { -- cgit v1.2.3