From 40735e7543f94fc88f58d94fd5f88daa4a2f2c6e Mon Sep 17 00:00:00 2001 From: Jijie Shao Date: Tue, 15 Oct 2024 20:35:12 +0800 Subject: [PATCH 01/16] net: hibmcge: Implement .ndo_start_xmit function Implement .ndo_start_xmit function to fill the information of the packet to be transmitted into the tx descriptor, and then the hardware will transmit the packet using the information in the tx descriptor. In addition, we also implemented the tx_handler function to enable the tx descriptor to be reused, and .ndo_tx_timeout function to print some information when the hardware is busy. Signed-off-by: Jijie Shao Signed-off-by: Paolo Abeni --- .../ethernet/hisilicon/hibmcge/hbg_common.h | 48 ++++ .../net/ethernet/hisilicon/hibmcge/hbg_hw.c | 18 ++ .../net/ethernet/hisilicon/hibmcge/hbg_hw.h | 2 + .../net/ethernet/hisilicon/hibmcge/hbg_irq.c | 8 +- .../net/ethernet/hisilicon/hibmcge/hbg_main.c | 54 +++- .../net/ethernet/hisilicon/hibmcge/hbg_reg.h | 19 ++ .../net/ethernet/hisilicon/hibmcge/hbg_txrx.c | 270 ++++++++++++++++++ .../net/ethernet/hisilicon/hibmcge/hbg_txrx.h | 39 +++ 8 files changed, 455 insertions(+), 3 deletions(-) create mode 100644 drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c create mode 100644 drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h index f60c2cc828fb..cceb674bd1c8 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h @@ -13,14 +13,61 @@ #define HBG_RX_SKIP1 0x00 #define HBG_RX_SKIP2 0x01 #define HBG_VECTOR_NUM 4 +#define HBG_PCU_CACHE_LINE_SIZE 32 +#define HBG_TX_TIMEOUT_BUF_LEN 1024 + +enum hbg_dir { + HBG_DIR_TX = 1 << 0, + HBG_DIR_RX = 1 << 1, + HBG_DIR_TX_RX = HBG_DIR_TX | HBG_DIR_RX, +}; + +enum hbg_tx_state { + HBG_TX_STATE_COMPLETE = 0, /* clear state, must fix to 0 */ + HBG_TX_STATE_START, +}; enum hbg_nic_state { HBG_NIC_STATE_EVENT_HANDLING = 0, }; +struct hbg_buffer { + u32 state; + dma_addr_t state_dma; + + struct sk_buff *skb; + dma_addr_t skb_dma; + u32 skb_len; + + enum hbg_dir dir; + struct hbg_ring *ring; + struct hbg_priv *priv; +}; + +struct hbg_ring { + struct hbg_buffer *queue; + dma_addr_t queue_dma; + + union { + u32 head; + u32 ntc; + }; + union { + u32 tail; + u32 ntu; + }; + u32 len; + + enum hbg_dir dir; + struct hbg_priv *priv; + struct napi_struct napi; + char *tout_log_buf; /* tx timeout log buffer */ +}; + enum hbg_hw_event_type { HBG_HW_EVENT_NONE = 0, HBG_HW_EVENT_INIT, /* driver is loading */ + HBG_HW_EVENT_RESET, }; struct hbg_dev_specs { @@ -73,6 +120,7 @@ struct hbg_priv { unsigned long state; struct hbg_mac mac; struct hbg_vector vectors; + struct hbg_ring tx_ring; }; #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c index 5258c5376987..05c71887b25a 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c @@ -73,6 +73,7 @@ static int hbg_hw_dev_specs_init(struct hbg_priv *priv) if (!is_valid_ether_addr((u8 *)specs->mac_addr.sa_data)) return -EADDRNOTAVAIL; + specs->max_frame_len = HBG_PCU_CACHE_LINE_SIZE + specs->max_mtu; return 0; } @@ -171,6 +172,23 @@ void hbg_hw_mac_enable(struct hbg_priv *priv, u32 enable) HBG_REG_PORT_ENABLE_RX_B, enable); } +u32 hbg_hw_get_fifo_used_num(struct hbg_priv *priv, enum hbg_dir dir) +{ + if (dir & HBG_DIR_TX) + return hbg_reg_read_field(priv, HBG_REG_CF_CFF_DATA_NUM_ADDR, + HBG_REG_CF_CFF_DATA_NUM_ADDR_TX_M); + + return 0; +} + +void hbg_hw_set_tx_desc(struct hbg_priv *priv, struct hbg_tx_desc *tx_desc) +{ + hbg_reg_write(priv, HBG_REG_TX_CFF_ADDR_0_ADDR, tx_desc->word0); + hbg_reg_write(priv, HBG_REG_TX_CFF_ADDR_1_ADDR, tx_desc->word1); + hbg_reg_write(priv, HBG_REG_TX_CFF_ADDR_2_ADDR, tx_desc->word2); + hbg_reg_write(priv, HBG_REG_TX_CFF_ADDR_3_ADDR, tx_desc->word3); +} + void hbg_hw_adjust_link(struct hbg_priv *priv, u32 speed, u32 duplex) { hbg_reg_write_field(priv, HBG_REG_PORT_MODE_ADDR, diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h index 0ce500e907b3..508e41cce41e 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h @@ -52,5 +52,7 @@ void hbg_hw_irq_enable(struct hbg_priv *priv, u32 mask, bool enable); void hbg_hw_set_mtu(struct hbg_priv *priv, u16 mtu); void hbg_hw_mac_enable(struct hbg_priv *priv, u32 enable); void hbg_hw_set_uc_addr(struct hbg_priv *priv, u64 mac_addr); +u32 hbg_hw_get_fifo_used_num(struct hbg_priv *priv, enum hbg_dir dir); +void hbg_hw_set_tx_desc(struct hbg_priv *priv, struct hbg_tx_desc *tx_desc); #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c index 66e047fb2d79..fed5502e51c8 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c @@ -13,6 +13,12 @@ static void hbg_irq_handle_err(struct hbg_priv *priv, "receive error interrupt: %s\n", irq_info->name); } +static void hbg_irq_handle_tx(struct hbg_priv *priv, + struct hbg_irq_info *irq_info) +{ + napi_schedule(&priv->tx_ring.napi); +} + #define HBG_TXRX_IRQ_I(name, handle) \ {#name, HBG_INT_MSK_##name##_B, false, false, 0, handle} #define HBG_ERR_IRQ_I(name, need_print) \ @@ -20,7 +26,7 @@ static void hbg_irq_handle_err(struct hbg_priv *priv, static struct hbg_irq_info hbg_irqs[] = { HBG_TXRX_IRQ_I(RX, NULL), - HBG_TXRX_IRQ_I(TX, NULL), + HBG_TXRX_IRQ_I(TX, hbg_irq_handle_tx), HBG_ERR_IRQ_I(MAC_MII_FIFO_ERR, true), HBG_ERR_IRQ_I(MAC_PCS_RX_FIFO_ERR, true), HBG_ERR_IRQ_I(MAC_PCS_TX_FIFO_ERR, true), diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c index 3f6f3b381c50..9bea5e21066f 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c @@ -9,6 +9,9 @@ #include "hbg_hw.h" #include "hbg_irq.h" #include "hbg_mdio.h" +#include "hbg_txrx.h" + +static void hbg_change_mtu(struct hbg_priv *priv, int new_mtu); static void hbg_all_irq_enable(struct hbg_priv *priv, bool enabled) { @@ -24,6 +27,11 @@ static void hbg_all_irq_enable(struct hbg_priv *priv, bool enabled) static int hbg_net_open(struct net_device *netdev) { struct hbg_priv *priv = netdev_priv(netdev); + int ret; + + ret = hbg_txrx_init(priv); + if (ret) + return ret; hbg_all_irq_enable(priv, true); hbg_hw_mac_enable(priv, HBG_STATUS_ENABLE); @@ -33,6 +41,26 @@ static int hbg_net_open(struct net_device *netdev) return 0; } +/* This function only can be called after hbg_txrx_uninit() */ +static int hbg_hw_txrx_clear(struct hbg_priv *priv) +{ + int ret; + + /* After ring buffers have been released, + * do a reset to release hw fifo rx ring buffer + */ + ret = hbg_hw_event_notify(priv, HBG_HW_EVENT_RESET); + if (ret) + return ret; + + /* After reset, regs need to be reconfigured */ + hbg_hw_init(priv); + hbg_hw_set_uc_addr(priv, ether_addr_to_u64(priv->netdev->dev_addr)); + hbg_change_mtu(priv, priv->netdev->mtu); + + return 0; +} + static int hbg_net_stop(struct net_device *netdev) { struct hbg_priv *priv = netdev_priv(netdev); @@ -41,8 +69,8 @@ static int hbg_net_stop(struct net_device *netdev) netif_stop_queue(netdev); hbg_hw_mac_enable(priv, HBG_STATUS_DISABLE); hbg_all_irq_enable(priv, false); - - return 0; + hbg_txrx_uninit(priv); + return hbg_hw_txrx_clear(priv); } static int hbg_net_set_mac_address(struct net_device *netdev, void *addr) @@ -86,12 +114,33 @@ static int hbg_net_change_mtu(struct net_device *netdev, int new_mtu) return 0; } +static void hbg_net_tx_timeout(struct net_device *netdev, unsigned int txqueue) +{ + struct hbg_priv *priv = netdev_priv(netdev); + struct hbg_ring *ring = &priv->tx_ring; + char *buf = ring->tout_log_buf; + u32 pos = 0; + + pos += scnprintf(buf + pos, HBG_TX_TIMEOUT_BUF_LEN - pos, + "ring used num: %u, fifo used num: %u\n", + hbg_get_queue_used_num(ring), + hbg_hw_get_fifo_used_num(priv, HBG_DIR_TX)); + pos += scnprintf(buf + pos, HBG_TX_TIMEOUT_BUF_LEN - pos, + "ntc: %u, ntu: %u, irq enabled: %u\n", + ring->ntc, ring->ntu, + hbg_hw_irq_is_enabled(priv, HBG_INT_MSK_TX_B)); + + netdev_info(netdev, "%s", buf); +} + static const struct net_device_ops hbg_netdev_ops = { .ndo_open = hbg_net_open, .ndo_stop = hbg_net_stop, + .ndo_start_xmit = hbg_net_start_xmit, .ndo_validate_addr = eth_validate_addr, .ndo_set_mac_address = hbg_net_set_mac_address, .ndo_change_mtu = hbg_net_change_mtu, + .ndo_tx_timeout = hbg_net_tx_timeout, }; static int hbg_init(struct hbg_priv *priv) @@ -170,6 +219,7 @@ static int hbg_probe(struct pci_dev *pdev, const struct pci_device_id *ent) netdev->max_mtu = priv->dev_specs.max_mtu; netdev->min_mtu = priv->dev_specs.min_mtu; netdev->netdev_ops = &hbg_netdev_ops; + netdev->watchdog_timeo = 5 * HZ; hbg_change_mtu(priv, ETH_DATA_LEN); hbg_net_set_mac_address(priv->netdev, &priv->dev_specs.mac_addr); diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h index fbc17ca5564f..0b7926a2a4fa 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h @@ -80,6 +80,12 @@ #define HBG_REG_CF_INTRPT_CLR_ADDR (HBG_REG_SGMII_BASE + 0x0438) #define HBG_REG_MAX_FRAME_LEN_ADDR (HBG_REG_SGMII_BASE + 0x0444) #define HBG_REG_MAX_FRAME_LEN_M GENMASK(15, 0) +#define HBG_REG_CF_CFF_DATA_NUM_ADDR (HBG_REG_SGMII_BASE + 0x045C) +#define HBG_REG_CF_CFF_DATA_NUM_ADDR_TX_M GENMASK(8, 0) +#define HBG_REG_TX_CFF_ADDR_0_ADDR (HBG_REG_SGMII_BASE + 0x0488) +#define HBG_REG_TX_CFF_ADDR_1_ADDR (HBG_REG_SGMII_BASE + 0x048C) +#define HBG_REG_TX_CFF_ADDR_2_ADDR (HBG_REG_SGMII_BASE + 0x0490) +#define HBG_REG_TX_CFF_ADDR_3_ADDR (HBG_REG_SGMII_BASE + 0x0494) #define HBG_REG_RX_BUF_SIZE_ADDR (HBG_REG_SGMII_BASE + 0x04E4) #define HBG_REG_RX_BUF_SIZE_M GENMASK(15, 0) #define HBG_REG_BUS_CTRL_ADDR (HBG_REG_SGMII_BASE + 0x04E8) @@ -108,4 +114,17 @@ enum hbg_port_mode { HBG_PORT_MODE_SGMII_1000M = 0x8, }; +struct hbg_tx_desc { + u32 word0; + u32 word1; + u32 word2; /* pkt_addr */ + u32 word3; /* clear_addr */ +}; + +#define HBG_TX_DESC_W0_IP_OFF_M GENMASK(30, 26) +#define HBG_TX_DESC_W0_l3_CS_B BIT(2) +#define HBG_TX_DESC_W0_WB_B BIT(1) +#define HBG_TX_DESC_W0_l4_CS_B BIT(0) +#define HBG_TX_DESC_W1_SEND_LEN_M GENMASK(19, 4) + #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c new file mode 100644 index 000000000000..6f828d7e77b3 --- /dev/null +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c @@ -0,0 +1,270 @@ +// SPDX-License-Identifier: GPL-2.0+ +// Copyright (c) 2024 Hisilicon Limited. + +#include +#include "hbg_common.h" +#include "hbg_irq.h" +#include "hbg_reg.h" +#include "hbg_txrx.h" + +#define netdev_get_tx_ring(netdev) \ + (&(((struct hbg_priv *)netdev_priv(netdev))->tx_ring)) + +#define buffer_to_dma_dir(buffer) (((buffer)->dir == HBG_DIR_RX) ? \ + DMA_FROM_DEVICE : DMA_TO_DEVICE) + +#define hbg_queue_used_num(head, tail, ring) ({ \ + typeof(ring) _ring = (ring); \ + ((tail) + _ring->len - (head)) % _ring->len; }) +#define hbg_queue_left_num(head, tail, ring) ({ \ + typeof(ring) _r = (ring); \ + _r->len - hbg_queue_used_num((head), (tail), _r) - 1; }) +#define hbg_queue_is_empty(head, tail, ring) \ + (hbg_queue_used_num((head), (tail), (ring)) == 0) +#define hbg_queue_next_prt(p, ring) (((p) + 1) % (ring)->len) + +#define HBG_TX_STOP_THRS 2 +#define HBG_TX_START_THRS (2 * HBG_TX_STOP_THRS) + +static int hbg_dma_map(struct hbg_buffer *buffer) +{ + struct hbg_priv *priv = buffer->priv; + + buffer->skb_dma = dma_map_single(&priv->pdev->dev, + buffer->skb->data, buffer->skb_len, + buffer_to_dma_dir(buffer)); + if (unlikely(dma_mapping_error(&priv->pdev->dev, buffer->skb_dma))) + return -ENOMEM; + + return 0; +} + +static void hbg_dma_unmap(struct hbg_buffer *buffer) +{ + struct hbg_priv *priv = buffer->priv; + + if (unlikely(!buffer->skb_dma)) + return; + + dma_unmap_single(&priv->pdev->dev, buffer->skb_dma, buffer->skb_len, + buffer_to_dma_dir(buffer)); + buffer->skb_dma = 0; +} + +static void hbg_init_tx_desc(struct hbg_buffer *buffer, + struct hbg_tx_desc *tx_desc) +{ + u32 ip_offset = buffer->skb->network_header - buffer->skb->mac_header; + u32 word0 = 0; + + word0 |= FIELD_PREP(HBG_TX_DESC_W0_WB_B, HBG_STATUS_ENABLE); + word0 |= FIELD_PREP(HBG_TX_DESC_W0_IP_OFF_M, ip_offset); + if (likely(buffer->skb->ip_summed == CHECKSUM_PARTIAL)) { + word0 |= FIELD_PREP(HBG_TX_DESC_W0_l3_CS_B, HBG_STATUS_ENABLE); + word0 |= FIELD_PREP(HBG_TX_DESC_W0_l4_CS_B, HBG_STATUS_ENABLE); + } + + tx_desc->word0 = word0; + tx_desc->word1 = FIELD_PREP(HBG_TX_DESC_W1_SEND_LEN_M, + buffer->skb->len); + tx_desc->word2 = buffer->skb_dma; + tx_desc->word3 = buffer->state_dma; +} + +netdev_tx_t hbg_net_start_xmit(struct sk_buff *skb, struct net_device *netdev) +{ + struct hbg_ring *ring = netdev_get_tx_ring(netdev); + struct hbg_priv *priv = netdev_priv(netdev); + /* This smp_load_acquire() pairs with smp_store_release() in + * hbg_napi_tx_recycle() called in tx interrupt handle process. + */ + u32 ntc = smp_load_acquire(&ring->ntc); + struct hbg_buffer *buffer; + struct hbg_tx_desc tx_desc; + u32 ntu = ring->ntu; + + if (unlikely(!skb->len || + skb->len > hbg_spec_max_frame_len(priv, HBG_DIR_TX))) { + dev_kfree_skb_any(skb); + netdev->stats.tx_errors++; + return NETDEV_TX_OK; + } + + if (!netif_subqueue_maybe_stop(netdev, 0, + hbg_queue_left_num(ntc, ntu, ring), + HBG_TX_STOP_THRS, HBG_TX_START_THRS)) + return NETDEV_TX_BUSY; + + buffer = &ring->queue[ntu]; + buffer->skb = skb; + buffer->skb_len = skb->len; + if (unlikely(hbg_dma_map(buffer))) { + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; + } + + buffer->state = HBG_TX_STATE_START; + hbg_init_tx_desc(buffer, &tx_desc); + hbg_hw_set_tx_desc(priv, &tx_desc); + + /* This smp_store_release() pairs with smp_load_acquire() in + * hbg_napi_tx_recycle() called in tx interrupt handle process. + */ + smp_store_release(&ring->ntu, hbg_queue_next_prt(ntu, ring)); + dev_sw_netstats_tx_add(netdev, 1, skb->len); + return NETDEV_TX_OK; +} + +static void hbg_buffer_free_skb(struct hbg_buffer *buffer) +{ + if (unlikely(!buffer->skb)) + return; + + dev_kfree_skb_any(buffer->skb); + buffer->skb = NULL; +} + +static void hbg_buffer_free(struct hbg_buffer *buffer) +{ + hbg_dma_unmap(buffer); + hbg_buffer_free_skb(buffer); +} + +static int hbg_napi_tx_recycle(struct napi_struct *napi, int budget) +{ + struct hbg_ring *ring = container_of(napi, struct hbg_ring, napi); + /* This smp_load_acquire() pairs with smp_store_release() in + * hbg_net_start_xmit() called in xmit process. + */ + u32 ntu = smp_load_acquire(&ring->ntu); + struct hbg_priv *priv = ring->priv; + struct hbg_buffer *buffer; + u32 ntc = ring->ntc; + int packet_done = 0; + + /* We need do cleanup even if budget is 0. + * Per NAPI documentation budget is for Rx. + * So We hardcode the amount of work Tx NAPI does to 128. + */ + budget = 128; + while (packet_done < budget) { + if (unlikely(hbg_queue_is_empty(ntc, ntu, ring))) + break; + + /* make sure HW write desc complete */ + dma_rmb(); + + buffer = &ring->queue[ntc]; + if (buffer->state != HBG_TX_STATE_COMPLETE) + break; + + hbg_buffer_free(buffer); + ntc = hbg_queue_next_prt(ntc, ring); + packet_done++; + } + + /* This smp_store_release() pairs with smp_load_acquire() in + * hbg_net_start_xmit() called in xmit process. + */ + smp_store_release(&ring->ntc, ntc); + netif_wake_queue(priv->netdev); + + if (likely(packet_done < budget && + napi_complete_done(napi, packet_done))) + hbg_hw_irq_enable(priv, HBG_INT_MSK_TX_B, true); + + return packet_done; +} + +static void hbg_ring_uninit(struct hbg_ring *ring) +{ + struct hbg_buffer *buffer; + u32 i; + + if (!ring->queue) + return; + + napi_disable(&ring->napi); + netif_napi_del(&ring->napi); + + for (i = 0; i < ring->len; i++) { + buffer = &ring->queue[i]; + hbg_buffer_free(buffer); + buffer->ring = NULL; + buffer->priv = NULL; + } + + dma_free_coherent(&ring->priv->pdev->dev, + ring->len * sizeof(*ring->queue), + ring->queue, ring->queue_dma); + ring->queue = NULL; + ring->queue_dma = 0; + ring->len = 0; + ring->priv = NULL; +} + +static int hbg_ring_init(struct hbg_priv *priv, struct hbg_ring *ring, + int (*napi_poll)(struct napi_struct *, int), + enum hbg_dir dir) +{ + struct hbg_buffer *buffer; + u32 i, len; + + len = hbg_get_spec_fifo_max_num(priv, dir) + 1; + ring->queue = dma_alloc_coherent(&priv->pdev->dev, + len * sizeof(*ring->queue), + &ring->queue_dma, GFP_KERNEL); + if (!ring->queue) + return -ENOMEM; + + for (i = 0; i < len; i++) { + buffer = &ring->queue[i]; + buffer->skb_len = 0; + buffer->dir = dir; + buffer->ring = ring; + buffer->priv = priv; + buffer->state_dma = ring->queue_dma + (i * sizeof(*buffer)); + } + + ring->dir = dir; + ring->priv = priv; + ring->ntc = 0; + ring->ntu = 0; + ring->len = len; + + netif_napi_add_tx(priv->netdev, &ring->napi, napi_poll); + napi_enable(&ring->napi); + return 0; +} + +static int hbg_tx_ring_init(struct hbg_priv *priv) +{ + struct hbg_ring *tx_ring = &priv->tx_ring; + + if (!tx_ring->tout_log_buf) + tx_ring->tout_log_buf = devm_kmalloc(&priv->pdev->dev, + HBG_TX_TIMEOUT_BUF_LEN, + GFP_KERNEL); + + if (!tx_ring->tout_log_buf) + return -ENOMEM; + + return hbg_ring_init(priv, tx_ring, hbg_napi_tx_recycle, HBG_DIR_TX); +} + +int hbg_txrx_init(struct hbg_priv *priv) +{ + int ret; + + ret = hbg_tx_ring_init(priv); + if (ret) + dev_err(&priv->pdev->dev, + "failed to init tx ring, ret = %d\n", ret); + + return ret; +} + +void hbg_txrx_uninit(struct hbg_priv *priv) +{ + hbg_ring_uninit(&priv->tx_ring); +} diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h new file mode 100644 index 000000000000..2883a5899ae2 --- /dev/null +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h @@ -0,0 +1,39 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ +/* Copyright (c) 2024 Hisilicon Limited. */ + +#ifndef __HBG_TXRX_H +#define __HBG_TXRX_H + +#include +#include "hbg_hw.h" + +static inline u32 hbg_spec_max_frame_len(struct hbg_priv *priv, + enum hbg_dir dir) +{ + return (dir == HBG_DIR_TX) ? priv->dev_specs.max_frame_len : + priv->dev_specs.rx_buf_size; +} + +static inline u32 hbg_get_spec_fifo_max_num(struct hbg_priv *priv, + enum hbg_dir dir) +{ + return (dir == HBG_DIR_TX) ? priv->dev_specs.tx_fifo_num : + priv->dev_specs.rx_fifo_num; +} + +static inline bool hbg_fifo_is_full(struct hbg_priv *priv, enum hbg_dir dir) +{ + return hbg_hw_get_fifo_used_num(priv, dir) >= + hbg_get_spec_fifo_max_num(priv, dir); +} + +static inline u32 hbg_get_queue_used_num(struct hbg_ring *ring) +{ + return (ring->ntu + ring->len - ring->ntc) % ring->len; +} + +netdev_tx_t hbg_net_start_xmit(struct sk_buff *skb, struct net_device *netdev); +int hbg_txrx_init(struct hbg_priv *priv); +void hbg_txrx_uninit(struct hbg_priv *priv); + +#endif -- 2.51.0 From f72e2559406191f2c7e2d15b78df6066594b0a9e Mon Sep 17 00:00:00 2001 From: Jijie Shao Date: Tue, 15 Oct 2024 20:35:13 +0800 Subject: [PATCH 02/16] net: hibmcge: Implement rx_poll function to receive packets Implement rx_poll function to read the rx descriptor after receiving the rx interrupt. Adjust the skb based on the descriptor to complete the reception of the packet. Signed-off-by: Jijie Shao Signed-off-by: Paolo Abeni --- .../ethernet/hisilicon/hibmcge/hbg_common.h | 5 + .../net/ethernet/hisilicon/hibmcge/hbg_hw.c | 10 ++ .../net/ethernet/hisilicon/hibmcge/hbg_hw.h | 1 + .../net/ethernet/hisilicon/hibmcge/hbg_irq.c | 8 +- .../net/ethernet/hisilicon/hibmcge/hbg_reg.h | 13 ++ .../net/ethernet/hisilicon/hibmcge/hbg_txrx.c | 143 +++++++++++++++++- 6 files changed, 177 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h index cceb674bd1c8..96daf058d387 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_common.h @@ -15,6 +15,10 @@ #define HBG_VECTOR_NUM 4 #define HBG_PCU_CACHE_LINE_SIZE 32 #define HBG_TX_TIMEOUT_BUF_LEN 1024 +#define HBG_RX_DESCR 0x01 + +#define HBG_PACKET_HEAD_SIZE ((HBG_RX_SKIP1 + HBG_RX_SKIP2 + \ + HBG_RX_DESCR) * HBG_PCU_CACHE_LINE_SIZE) enum hbg_dir { HBG_DIR_TX = 1 << 0, @@ -121,6 +125,7 @@ struct hbg_priv { struct hbg_mac mac; struct hbg_vector vectors; struct hbg_ring tx_ring; + struct hbg_ring rx_ring; }; #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c index 05c71887b25a..05295c2ad439 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c @@ -74,6 +74,7 @@ static int hbg_hw_dev_specs_init(struct hbg_priv *priv) return -EADDRNOTAVAIL; specs->max_frame_len = HBG_PCU_CACHE_LINE_SIZE + specs->max_mtu; + specs->rx_buf_size = HBG_PACKET_HEAD_SIZE + specs->max_frame_len; return 0; } @@ -178,6 +179,10 @@ u32 hbg_hw_get_fifo_used_num(struct hbg_priv *priv, enum hbg_dir dir) return hbg_reg_read_field(priv, HBG_REG_CF_CFF_DATA_NUM_ADDR, HBG_REG_CF_CFF_DATA_NUM_ADDR_TX_M); + if (dir & HBG_DIR_RX) + return hbg_reg_read_field(priv, HBG_REG_CF_CFF_DATA_NUM_ADDR, + HBG_REG_CF_CFF_DATA_NUM_ADDR_RX_M); + return 0; } @@ -189,6 +194,11 @@ void hbg_hw_set_tx_desc(struct hbg_priv *priv, struct hbg_tx_desc *tx_desc) hbg_reg_write(priv, HBG_REG_TX_CFF_ADDR_3_ADDR, tx_desc->word3); } +void hbg_hw_fill_buffer(struct hbg_priv *priv, u32 buffer_dma_addr) +{ + hbg_reg_write(priv, HBG_REG_RX_CFF_ADDR_ADDR, buffer_dma_addr); +} + void hbg_hw_adjust_link(struct hbg_priv *priv, u32 speed, u32 duplex) { hbg_reg_write_field(priv, HBG_REG_PORT_MODE_ADDR, diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h index 508e41cce41e..14fb39241c93 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.h @@ -54,5 +54,6 @@ void hbg_hw_mac_enable(struct hbg_priv *priv, u32 enable); void hbg_hw_set_uc_addr(struct hbg_priv *priv, u64 mac_addr); u32 hbg_hw_get_fifo_used_num(struct hbg_priv *priv, enum hbg_dir dir); void hbg_hw_set_tx_desc(struct hbg_priv *priv, struct hbg_tx_desc *tx_desc); +void hbg_hw_fill_buffer(struct hbg_priv *priv, u32 buffer_dma_addr); #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c index fed5502e51c8..25dd25f096fe 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_irq.c @@ -19,13 +19,19 @@ static void hbg_irq_handle_tx(struct hbg_priv *priv, napi_schedule(&priv->tx_ring.napi); } +static void hbg_irq_handle_rx(struct hbg_priv *priv, + struct hbg_irq_info *irq_info) +{ + napi_schedule(&priv->rx_ring.napi); +} + #define HBG_TXRX_IRQ_I(name, handle) \ {#name, HBG_INT_MSK_##name##_B, false, false, 0, handle} #define HBG_ERR_IRQ_I(name, need_print) \ {#name, HBG_INT_MSK_##name##_B, true, need_print, 0, hbg_irq_handle_err} static struct hbg_irq_info hbg_irqs[] = { - HBG_TXRX_IRQ_I(RX, NULL), + HBG_TXRX_IRQ_I(RX, hbg_irq_handle_rx), HBG_TXRX_IRQ_I(TX, hbg_irq_handle_tx), HBG_ERR_IRQ_I(MAC_MII_FIFO_ERR, true), HBG_ERR_IRQ_I(MAC_PCS_RX_FIFO_ERR, true), diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h index 0b7926a2a4fa..57d81c6d7633 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_reg.h @@ -82,10 +82,12 @@ #define HBG_REG_MAX_FRAME_LEN_M GENMASK(15, 0) #define HBG_REG_CF_CFF_DATA_NUM_ADDR (HBG_REG_SGMII_BASE + 0x045C) #define HBG_REG_CF_CFF_DATA_NUM_ADDR_TX_M GENMASK(8, 0) +#define HBG_REG_CF_CFF_DATA_NUM_ADDR_RX_M GENMASK(24, 16) #define HBG_REG_TX_CFF_ADDR_0_ADDR (HBG_REG_SGMII_BASE + 0x0488) #define HBG_REG_TX_CFF_ADDR_1_ADDR (HBG_REG_SGMII_BASE + 0x048C) #define HBG_REG_TX_CFF_ADDR_2_ADDR (HBG_REG_SGMII_BASE + 0x0490) #define HBG_REG_TX_CFF_ADDR_3_ADDR (HBG_REG_SGMII_BASE + 0x0494) +#define HBG_REG_RX_CFF_ADDR_ADDR (HBG_REG_SGMII_BASE + 0x04A0) #define HBG_REG_RX_BUF_SIZE_ADDR (HBG_REG_SGMII_BASE + 0x04E4) #define HBG_REG_RX_BUF_SIZE_M GENMASK(15, 0) #define HBG_REG_BUS_CTRL_ADDR (HBG_REG_SGMII_BASE + 0x04E8) @@ -127,4 +129,15 @@ struct hbg_tx_desc { #define HBG_TX_DESC_W0_l4_CS_B BIT(0) #define HBG_TX_DESC_W1_SEND_LEN_M GENMASK(19, 4) +struct hbg_rx_desc { + u32 word0; + u32 word1; /* tag */ + u32 word2; + u32 word3; + u32 word4; + u32 word5; +}; + +#define HBG_RX_DESC_W2_PKT_LEN_M GENMASK(31, 16) + #endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c index 6f828d7e77b3..f4f256a0dfea 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.c @@ -21,7 +21,12 @@ _r->len - hbg_queue_used_num((head), (tail), _r) - 1; }) #define hbg_queue_is_empty(head, tail, ring) \ (hbg_queue_used_num((head), (tail), (ring)) == 0) +#define hbg_queue_is_full(head, tail, ring) \ + (hbg_queue_left_num((head), (tail), (ring)) == 0) #define hbg_queue_next_prt(p, ring) (((p) + 1) % (ring)->len) +#define hbg_queue_move_next(p, ring) ({ \ + typeof(ring) _ring = (ring); \ + _ring->p = hbg_queue_next_prt(_ring->p, _ring); }) #define HBG_TX_STOP_THRS 2 #define HBG_TX_START_THRS (2 * HBG_TX_STOP_THRS) @@ -124,6 +129,20 @@ static void hbg_buffer_free_skb(struct hbg_buffer *buffer) buffer->skb = NULL; } +static int hbg_buffer_alloc_skb(struct hbg_buffer *buffer) +{ + u32 len = hbg_spec_max_frame_len(buffer->priv, buffer->dir); + struct hbg_priv *priv = buffer->priv; + + buffer->skb = netdev_alloc_skb(priv->netdev, len); + if (unlikely(!buffer->skb)) + return -ENOMEM; + + buffer->skb_len = len; + memset(buffer->skb->data, 0, HBG_PACKET_HEAD_SIZE); + return 0; +} + static void hbg_buffer_free(struct hbg_buffer *buffer) { hbg_dma_unmap(buffer); @@ -176,6 +195,92 @@ static int hbg_napi_tx_recycle(struct napi_struct *napi, int budget) return packet_done; } +static int hbg_rx_fill_one_buffer(struct hbg_priv *priv) +{ + struct hbg_ring *ring = &priv->rx_ring; + struct hbg_buffer *buffer; + int ret; + + if (hbg_queue_is_full(ring->ntc, ring->ntu, ring)) + return 0; + + buffer = &ring->queue[ring->ntu]; + ret = hbg_buffer_alloc_skb(buffer); + if (unlikely(ret)) + return ret; + + ret = hbg_dma_map(buffer); + if (unlikely(ret)) { + hbg_buffer_free_skb(buffer); + return ret; + } + + hbg_hw_fill_buffer(priv, buffer->skb_dma); + hbg_queue_move_next(ntu, ring); + return 0; +} + +static bool hbg_sync_data_from_hw(struct hbg_priv *priv, + struct hbg_buffer *buffer) +{ + struct hbg_rx_desc *rx_desc; + + /* make sure HW write desc complete */ + dma_rmb(); + + dma_sync_single_for_cpu(&priv->pdev->dev, buffer->skb_dma, + buffer->skb_len, DMA_FROM_DEVICE); + + rx_desc = (struct hbg_rx_desc *)buffer->skb->data; + return FIELD_GET(HBG_RX_DESC_W2_PKT_LEN_M, rx_desc->word2) != 0; +} + +static int hbg_napi_rx_poll(struct napi_struct *napi, int budget) +{ + struct hbg_ring *ring = container_of(napi, struct hbg_ring, napi); + struct hbg_priv *priv = ring->priv; + struct hbg_rx_desc *rx_desc; + struct hbg_buffer *buffer; + u32 packet_done = 0; + u32 pkt_len; + + while (packet_done < budget) { + if (unlikely(hbg_queue_is_empty(ring->ntc, ring->ntu, ring))) + break; + + buffer = &ring->queue[ring->ntc]; + if (unlikely(!buffer->skb)) + goto next_buffer; + + if (unlikely(!hbg_sync_data_from_hw(priv, buffer))) + break; + rx_desc = (struct hbg_rx_desc *)buffer->skb->data; + pkt_len = FIELD_GET(HBG_RX_DESC_W2_PKT_LEN_M, rx_desc->word2); + + hbg_dma_unmap(buffer); + + skb_reserve(buffer->skb, HBG_PACKET_HEAD_SIZE + NET_IP_ALIGN); + skb_put(buffer->skb, pkt_len); + buffer->skb->protocol = eth_type_trans(buffer->skb, + priv->netdev); + + dev_sw_netstats_rx_add(priv->netdev, pkt_len); + napi_gro_receive(napi, buffer->skb); + buffer->skb = NULL; + +next_buffer: + hbg_rx_fill_one_buffer(priv); + hbg_queue_move_next(ntc, ring); + packet_done++; + } + + if (likely(packet_done < budget && + napi_complete_done(napi, packet_done))) + hbg_hw_irq_enable(priv, HBG_INT_MSK_RX_B, true); + + return packet_done; +} + static void hbg_ring_uninit(struct hbg_ring *ring) { struct hbg_buffer *buffer; @@ -232,7 +337,11 @@ static int hbg_ring_init(struct hbg_priv *priv, struct hbg_ring *ring, ring->ntu = 0; ring->len = len; - netif_napi_add_tx(priv->netdev, &ring->napi, napi_poll); + if (dir == HBG_DIR_TX) + netif_napi_add_tx(priv->netdev, &ring->napi, napi_poll); + else + netif_napi_add(priv->netdev, &ring->napi, napi_poll); + napi_enable(&ring->napi); return 0; } @@ -252,14 +361,43 @@ static int hbg_tx_ring_init(struct hbg_priv *priv) return hbg_ring_init(priv, tx_ring, hbg_napi_tx_recycle, HBG_DIR_TX); } +static int hbg_rx_ring_init(struct hbg_priv *priv) +{ + int ret; + u32 i; + + ret = hbg_ring_init(priv, &priv->rx_ring, hbg_napi_rx_poll, HBG_DIR_RX); + if (ret) + return ret; + + for (i = 0; i < priv->rx_ring.len - 1; i++) { + ret = hbg_rx_fill_one_buffer(priv); + if (ret) { + hbg_ring_uninit(&priv->rx_ring); + return ret; + } + } + + return 0; +} + int hbg_txrx_init(struct hbg_priv *priv) { int ret; ret = hbg_tx_ring_init(priv); - if (ret) + if (ret) { dev_err(&priv->pdev->dev, "failed to init tx ring, ret = %d\n", ret); + return ret; + } + + ret = hbg_rx_ring_init(priv); + if (ret) { + dev_err(&priv->pdev->dev, + "failed to init rx ring, ret = %d\n", ret); + hbg_ring_uninit(&priv->tx_ring); + } return ret; } @@ -267,4 +405,5 @@ int hbg_txrx_init(struct hbg_priv *priv) void hbg_txrx_uninit(struct hbg_priv *priv) { hbg_ring_uninit(&priv->tx_ring); + hbg_ring_uninit(&priv->rx_ring); } -- 2.51.0 From e8d13548bd0831c8b3cb53ff9ec20d34ef2523ef Mon Sep 17 00:00:00 2001 From: Jijie Shao Date: Tue, 15 Oct 2024 20:35:14 +0800 Subject: [PATCH 03/16] net: hibmcge: Implement some ethtool_ops functions Implement the .get_drvinfo .get_link .get_link_ksettings to get the basic information and working status of the driver. Implement the .set_link_ksettings to modify the rate, duplex, and auto-negotiation status. Signed-off-by: Jijie Shao Reviewed-by: Andrew Lunn Reviewed-by: Kalesh AP Signed-off-by: Paolo Abeni --- .../ethernet/hisilicon/hibmcge/hbg_ethtool.c | 17 +++++++++++++++++ .../ethernet/hisilicon/hibmcge/hbg_ethtool.h | 11 +++++++++++ .../net/ethernet/hisilicon/hibmcge/hbg_main.c | 2 ++ 3 files changed, 30 insertions(+) create mode 100644 drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.c create mode 100644 drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.h diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.c new file mode 100644 index 000000000000..c3370114aef3 --- /dev/null +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.c @@ -0,0 +1,17 @@ +// SPDX-License-Identifier: GPL-2.0+ +// Copyright (c) 2024 Hisilicon Limited. + +#include +#include +#include "hbg_ethtool.h" + +static const struct ethtool_ops hbg_ethtool_ops = { + .get_link = ethtool_op_get_link, + .get_link_ksettings = phy_ethtool_get_link_ksettings, + .set_link_ksettings = phy_ethtool_set_link_ksettings, +}; + +void hbg_ethtool_set_ops(struct net_device *netdev) +{ + netdev->ethtool_ops = &hbg_ethtool_ops; +} diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.h new file mode 100644 index 000000000000..628707ec2686 --- /dev/null +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_ethtool.h @@ -0,0 +1,11 @@ +/* SPDX-License-Identifier: GPL-2.0+ */ +/* Copyright (c) 2024 Hisilicon Limited. */ + +#ifndef __HBG_ETHTOOL_H +#define __HBG_ETHTOOL_H + +#include + +void hbg_ethtool_set_ops(struct net_device *netdev); + +#endif diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c index 9bea5e21066f..75505fb5cc4a 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c @@ -6,6 +6,7 @@ #include #include #include "hbg_common.h" +#include "hbg_ethtool.h" #include "hbg_hw.h" #include "hbg_irq.h" #include "hbg_mdio.h" @@ -223,6 +224,7 @@ static int hbg_probe(struct pci_dev *pdev, const struct pci_device_id *ent) hbg_change_mtu(priv, ETH_DATA_LEN); hbg_net_set_mac_address(priv->netdev, &priv->dev_specs.mac_addr); + hbg_ethtool_set_ops(netdev); ret = devm_register_netdev(dev, netdev); if (ret) -- 2.51.0 From 81e176de6ad4a1c4345bce61b2876b01c6c24c28 Mon Sep 17 00:00:00 2001 From: Jijie Shao Date: Tue, 15 Oct 2024 20:35:15 +0800 Subject: [PATCH 04/16] net: hibmcge: Add a Makefile and update Kconfig for hibmcge Add a Makefile and update Kconfig to build hibmcge driver. Signed-off-by: Jijie Shao Signed-off-by: Paolo Abeni --- drivers/net/ethernet/hisilicon/Kconfig | 18 +++++++++++++++++- drivers/net/ethernet/hisilicon/Makefile | 1 + .../net/ethernet/hisilicon/hibmcge/Makefile | 8 ++++++++ 3 files changed, 26 insertions(+), 1 deletion(-) create mode 100644 drivers/net/ethernet/hisilicon/hibmcge/Makefile diff --git a/drivers/net/ethernet/hisilicon/Kconfig b/drivers/net/ethernet/hisilicon/Kconfig index 3312e1d93c3b..65302c41bfb1 100644 --- a/drivers/net/ethernet/hisilicon/Kconfig +++ b/drivers/net/ethernet/hisilicon/Kconfig @@ -7,7 +7,6 @@ config NET_VENDOR_HISILICON bool "Hisilicon devices" default y depends on OF || ACPI - depends on ARM || ARM64 || COMPILE_TEST help If you have a network (Ethernet) card belonging to this class, say Y. @@ -18,6 +17,8 @@ config NET_VENDOR_HISILICON if NET_VENDOR_HISILICON +if ARM || ARM64 || COMPILE_TEST + config HIX5HD2_GMAC tristate "Hisilicon HIX5HD2 Family Network Device Support" select PHYLIB @@ -141,4 +142,19 @@ config HNS3_ENET endif #HNS3 +endif # ARM || ARM64 || COMPILE_TEST + +config HIBMCGE + tristate "Hisilicon BMC Gigabit Ethernet Device Support" + depends on PCI && PCI_MSI + select PHYLIB + select MOTORCOMM_PHY + select REALTEK_PHY + help + If you wish to compile a kernel for a BMC with HIBMC-xx_gmac + then you should answer Y to this. This makes this driver suitable for use + on certain boards such as the HIBMC-210. + + If you are unsure, say N. + endif # NET_VENDOR_HISILICON diff --git a/drivers/net/ethernet/hisilicon/Makefile b/drivers/net/ethernet/hisilicon/Makefile index 7f76d412047a..0e2cadfea8ff 100644 --- a/drivers/net/ethernet/hisilicon/Makefile +++ b/drivers/net/ethernet/hisilicon/Makefile @@ -9,3 +9,4 @@ obj-$(CONFIG_HNS_MDIO) += hns_mdio.o obj-$(CONFIG_HNS) += hns/ obj-$(CONFIG_HNS3) += hns3/ obj-$(CONFIG_HISI_FEMAC) += hisi_femac.o +obj-$(CONFIG_HIBMCGE) += hibmcge/ diff --git a/drivers/net/ethernet/hisilicon/hibmcge/Makefile b/drivers/net/ethernet/hisilicon/hibmcge/Makefile new file mode 100644 index 000000000000..ae58ac38c206 --- /dev/null +++ b/drivers/net/ethernet/hisilicon/hibmcge/Makefile @@ -0,0 +1,8 @@ +# SPDX-License-Identifier: GPL-2.0+ +# +# Makefile for the HISILICON BMC GE network device drivers. +# + +obj-$(CONFIG_HIBMCGE) += hibmcge.o + +hibmcge-objs = hbg_main.o hbg_hw.o hbg_mdio.o hbg_irq.o hbg_txrx.o hbg_ethtool.o -- 2.51.0 From f9a002a13054b23ecba0904510121f61632797c8 Mon Sep 17 00:00:00 2001 From: Jijie Shao Date: Tue, 15 Oct 2024 20:35:16 +0800 Subject: [PATCH 05/16] net: hibmcge: Add maintainer for hibmcge Add myself as the maintainer for the hibmcge ethernet driver. Signed-off-by: Jijie Shao Signed-off-by: Paolo Abeni --- MAINTAINERS | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/MAINTAINERS b/MAINTAINERS index 16933f9faa0f..aed1fa42cfd2 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -10278,6 +10278,12 @@ S: Maintained W: http://www.hisilicon.com F: drivers/net/ethernet/hisilicon/hns3/ +HISILICON NETWORK HIBMCGE DRIVER +M: Jijie Shao +L: netdev@vger.kernel.org +S: Maintained +F: drivers/net/ethernet/hisilicon/hibmcge/ + HISILICON NETWORK SUBSYSTEM DRIVER M: Jian Shen M: Salil Mehta -- 2.51.0 From 700814fa41cecbd62a415e77524f494cafe5bc42 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:03 +0300 Subject: [PATCH 06/16] net/mlx5: Refactor QoS group scheduling element creation Introduce `esw_qos_create_group_sched_elem` to handle the creation of group scheduling elements for E-Switch QoS, Transmit Scheduling Arbiter (TSAR). This reduces duplication and simplifies code for TSAR setup. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 64 +++++++++---------- 1 file changed, 31 insertions(+), 33 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index ee6f76a6f0b5..7732f948e9c6 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -371,6 +371,33 @@ static int esw_qos_set_group_max_rate(struct mlx5_esw_rate_group *group, return err; } +static int esw_qos_create_group_sched_elem(struct mlx5_core_dev *dev, u32 parent_element_id, + u32 *tsar_ix) +{ + u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; + void *attr; + + if (!mlx5_qos_element_type_supported(dev, + SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR, + SCHEDULING_HIERARCHY_E_SWITCH) || + !mlx5_qos_tsar_type_supported(dev, + TSAR_ELEMENT_TSAR_TYPE_DWRR, + SCHEDULING_HIERARCHY_E_SWITCH)) + return -EOPNOTSUPP; + + MLX5_SET(scheduling_context, tsar_ctx, element_type, + SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR); + MLX5_SET(scheduling_context, tsar_ctx, parent_element_id, + parent_element_id); + attr = MLX5_ADDR_OF(scheduling_context, tsar_ctx, element_attributes); + MLX5_SET(tsar_element, attr, tsar_type, TSAR_ELEMENT_TSAR_TYPE_DWRR); + + return mlx5_create_scheduling_element_cmd(dev, + SCHEDULING_HIERARCHY_E_SWITCH, + tsar_ctx, + tsar_ix); +} + static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, u32 max_rate, u32 bw_share) { @@ -496,21 +523,11 @@ static void __esw_qos_free_rate_group(struct mlx5_esw_rate_group *group) static struct mlx5_esw_rate_group * __esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) { - u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; struct mlx5_esw_rate_group *group; - int tsar_ix, err; - void *attr; + u32 tsar_ix; + int err; - MLX5_SET(scheduling_context, tsar_ctx, element_type, - SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR); - MLX5_SET(scheduling_context, tsar_ctx, parent_element_id, - esw->qos.root_tsar_ix); - attr = MLX5_ADDR_OF(scheduling_context, tsar_ctx, element_attributes); - MLX5_SET(tsar_element, attr, tsar_type, TSAR_ELEMENT_TSAR_TYPE_DWRR); - err = mlx5_create_scheduling_element_cmd(esw->dev, - SCHEDULING_HIERARCHY_E_SWITCH, - tsar_ctx, - &tsar_ix); + err = esw_qos_create_group_sched_elem(esw->dev, esw->qos.root_tsar_ix, &tsar_ix); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch create TSAR for group failed"); return ERR_PTR(err); @@ -591,32 +608,13 @@ static int __esw_qos_destroy_rate_group(struct mlx5_esw_rate_group *group, static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) { - u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; struct mlx5_core_dev *dev = esw->dev; - void *attr; int err; if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling)) return -EOPNOTSUPP; - if (!mlx5_qos_element_type_supported(dev, - SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR, - SCHEDULING_HIERARCHY_E_SWITCH) || - !mlx5_qos_tsar_type_supported(dev, - TSAR_ELEMENT_TSAR_TYPE_DWRR, - SCHEDULING_HIERARCHY_E_SWITCH)) - return -EOPNOTSUPP; - - MLX5_SET(scheduling_context, tsar_ctx, element_type, - SCHEDULING_CONTEXT_ELEMENT_TYPE_TSAR); - - attr = MLX5_ADDR_OF(scheduling_context, tsar_ctx, element_attributes); - MLX5_SET(tsar_element, attr, tsar_type, TSAR_ELEMENT_TSAR_TYPE_DWRR); - - err = mlx5_create_scheduling_element_cmd(dev, - SCHEDULING_HIERARCHY_E_SWITCH, - tsar_ctx, - &esw->qos.root_tsar_ix); + err = esw_qos_create_group_sched_elem(esw->dev, 0, &esw->qos.root_tsar_ix); if (err) { esw_warn(dev, "E-Switch create root TSAR failed (%d)\n", err); return err; -- 2.51.0 From 4235fe2cb8e98dec6f49a86c93704b6a4762c4a2 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:04 +0300 Subject: [PATCH 07/16] net/mlx5: Introduce node type to rate group structure Introduce the `sched_node_type` enum to represent both the group and its members as scheduling nodes in the rate hierarchy. Add the `type` field to the rate group structure to specify the type of the node membership in the rate hierarchy. Generalize comments to reflect this flexibility within the rate group structure. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 28 ++++++++++++------- 1 file changed, 18 insertions(+), 10 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 7732f948e9c6..b324a6b1b9ff 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -61,6 +61,10 @@ static void esw_qos_domain_release(struct mlx5_eswitch *esw) esw->qos.domain = NULL; } +enum sched_node_type { + SCHED_NODE_TYPE_VPORTS_TSAR, +}; + struct mlx5_esw_rate_group { u32 tsar_ix; /* Bandwidth parameters. */ @@ -68,11 +72,13 @@ struct mlx5_esw_rate_group { u32 min_rate; /* A computed value indicating relative min_rate between group members. */ u32 bw_share; - /* Membership in the qos domain 'groups' list. */ + /* Membership in the parent list. */ struct list_head parent_entry; + /* The type of this group node in the rate hierarchy. */ + enum sched_node_type type; /* The eswitch this group belongs to. */ struct mlx5_eswitch *esw; - /* Vport members of this group.*/ + /* Members of this group.*/ struct list_head members; }; @@ -499,7 +505,7 @@ static int esw_qos_vport_update_group(struct mlx5_vport *vport, } static struct mlx5_esw_rate_group * -__esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix) +__esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_node_type type) { struct mlx5_esw_rate_group *group; @@ -509,6 +515,7 @@ __esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix) group->esw = esw; group->tsar_ix = tsar_ix; + group->type = type; INIT_LIST_HEAD(&group->members); list_add_tail(&group->parent_entry, &esw->qos.domain->groups); return group; @@ -521,7 +528,7 @@ static void __esw_qos_free_rate_group(struct mlx5_esw_rate_group *group) } static struct mlx5_esw_rate_group * -__esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) +__esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) { struct mlx5_esw_rate_group *group; u32 tsar_ix; @@ -533,7 +540,7 @@ __esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *ex return ERR_PTR(err); } - group = __esw_qos_alloc_rate_group(esw, tsar_ix); + group = __esw_qos_alloc_rate_group(esw, tsar_ix, SCHED_NODE_TYPE_VPORTS_TSAR); if (!group) { NL_SET_ERR_MSG_MOD(extack, "E-Switch alloc group failed"); err = -ENOMEM; @@ -563,7 +570,7 @@ static int esw_qos_get(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) static void esw_qos_put(struct mlx5_eswitch *esw); static struct mlx5_esw_rate_group * -esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) +esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) { struct mlx5_esw_rate_group *group; int err; @@ -576,7 +583,7 @@ esw_qos_create_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *exta if (err) return ERR_PTR(err); - group = __esw_qos_create_rate_group(esw, extack); + group = __esw_qos_create_vports_rate_group(esw, extack); if (IS_ERR(group)) esw_qos_put(esw); @@ -621,12 +628,13 @@ static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *exta } if (MLX5_CAP_QOS(dev, log_esw_max_sched_depth)) { - esw->qos.group0 = __esw_qos_create_rate_group(esw, extack); + esw->qos.group0 = __esw_qos_create_vports_rate_group(esw, extack); } else { /* The eswitch doesn't support scheduling groups. * Create a software-only group0 using the root TSAR to attach vport QoS to. */ - if (!__esw_qos_alloc_rate_group(esw, esw->qos.root_tsar_ix)) + if (!__esw_qos_alloc_rate_group(esw, esw->qos.root_tsar_ix, + SCHED_NODE_TYPE_VPORTS_TSAR)) esw->qos.group0 = ERR_PTR(-ENOMEM); } if (IS_ERR(esw->qos.group0)) { @@ -1038,7 +1046,7 @@ int mlx5_esw_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv, goto unlock; } - group = esw_qos_create_rate_group(esw, extack); + group = esw_qos_create_vports_rate_group(esw, extack); if (IS_ERR(group)) { err = PTR_ERR(group); goto unlock; -- 2.51.0 From 54200dbc685c6fcfcd59b132ff5d9979563bc01d Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:05 +0300 Subject: [PATCH 08/16] net/mlx5: Add parent group support in rate group structure Introduce a `parent` field in the `mlx5_esw_rate_group` structure to support hierarchical group relationships. The `parent` can reference another group or be set to `NULL`, indicating the group is connected to the root TSAR. This change enables the ability to manage groups in a hierarchical structure for future enhancements. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index b324a6b1b9ff..f2a0d59fa5bb 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -72,6 +72,8 @@ struct mlx5_esw_rate_group { u32 min_rate; /* A computed value indicating relative min_rate between group members. */ u32 bw_share; + /* The parent group of this group. */ + struct mlx5_esw_rate_group *parent; /* Membership in the parent list. */ struct list_head parent_entry; /* The type of this group node in the rate hierarchy. */ @@ -505,7 +507,8 @@ static int esw_qos_vport_update_group(struct mlx5_vport *vport, } static struct mlx5_esw_rate_group * -__esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_node_type type) +__esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_node_type type, + struct mlx5_esw_rate_group *parent) { struct mlx5_esw_rate_group *group; @@ -516,6 +519,7 @@ __esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_nod group->esw = esw; group->tsar_ix = tsar_ix; group->type = type; + group->parent = parent; INIT_LIST_HEAD(&group->members); list_add_tail(&group->parent_entry, &esw->qos.domain->groups); return group; @@ -528,7 +532,8 @@ static void __esw_qos_free_rate_group(struct mlx5_esw_rate_group *group) } static struct mlx5_esw_rate_group * -__esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) +__esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct mlx5_esw_rate_group *parent, + struct netlink_ext_ack *extack) { struct mlx5_esw_rate_group *group; u32 tsar_ix; @@ -540,7 +545,7 @@ __esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ return ERR_PTR(err); } - group = __esw_qos_alloc_rate_group(esw, tsar_ix, SCHED_NODE_TYPE_VPORTS_TSAR); + group = __esw_qos_alloc_rate_group(esw, tsar_ix, SCHED_NODE_TYPE_VPORTS_TSAR, parent); if (!group) { NL_SET_ERR_MSG_MOD(extack, "E-Switch alloc group failed"); err = -ENOMEM; @@ -583,7 +588,7 @@ esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ac if (err) return ERR_PTR(err); - group = __esw_qos_create_vports_rate_group(esw, extack); + group = __esw_qos_create_vports_rate_group(esw, NULL, extack); if (IS_ERR(group)) esw_qos_put(esw); @@ -628,13 +633,13 @@ static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *exta } if (MLX5_CAP_QOS(dev, log_esw_max_sched_depth)) { - esw->qos.group0 = __esw_qos_create_vports_rate_group(esw, extack); + esw->qos.group0 = __esw_qos_create_vports_rate_group(esw, NULL, extack); } else { /* The eswitch doesn't support scheduling groups. * Create a software-only group0 using the root TSAR to attach vport QoS to. */ if (!__esw_qos_alloc_rate_group(esw, esw->qos.root_tsar_ix, - SCHED_NODE_TYPE_VPORTS_TSAR)) + SCHED_NODE_TYPE_VPORTS_TSAR, NULL)) esw->qos.group0 = ERR_PTR(-ENOMEM); } if (IS_ERR(esw->qos.group0)) { -- 2.51.0 From 24e54e870d11db3b5a52f8bb5729c55989e44ef5 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:06 +0300 Subject: [PATCH 09/16] net/mlx5: Restrict domain list insertion to root TSAR ancestors Update the logic for adding rate groups to the E-Switch domain list, ensuring only groups with the root Transmit Scheduling Arbiter as their parent are included. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index f2a0d59fa5bb..dd6fe729f456 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -511,6 +511,7 @@ __esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_nod struct mlx5_esw_rate_group *parent) { struct mlx5_esw_rate_group *group; + struct list_head *parent_list; group = kzalloc(sizeof(*group), GFP_KERNEL); if (!group) @@ -521,7 +522,9 @@ __esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_nod group->type = type; group->parent = parent; INIT_LIST_HEAD(&group->members); - list_add_tail(&group->parent_entry, &esw->qos.domain->groups); + parent_list = parent ? &parent->members : &esw->qos.domain->groups; + list_add_tail(&group->parent_entry, parent_list); + return group; } -- 2.51.0 From 72a1d121fa6bf079bddb1df9fd19b394eb3ff5ee Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:07 +0300 Subject: [PATCH 10/16] net/mlx5: Rename vport QoS group reference to parent Rename the `group` field in the `mlx5_vport` structure to `parent` to clarify the vport's role as a member of a parent group and distinguish it from the concept of a general group. Additionally, rename `group_entry` to `parent_entry` to reflect this update. This distinction will be important for handling more complex group structures and scheduling elements. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- .../mlx5/core/esw/diag/qos_tracepoint.h | 8 ++-- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 42 +++++++++---------- .../net/ethernet/mellanox/mlx5/core/eswitch.h | 6 ++- 3 files changed, 29 insertions(+), 27 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h index 645bad0d625f..2aea01959073 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h @@ -35,18 +35,18 @@ DECLARE_EVENT_CLASS(mlx5_esw_vport_qos_template, __field(unsigned int, sched_elem_ix) __field(unsigned int, bw_share) __field(unsigned int, max_rate) - __field(void *, group) + __field(void *, parent) ), TP_fast_assign(__assign_str(devname); __entry->vport_id = vport->vport; __entry->sched_elem_ix = vport->qos.esw_sched_elem_ix; __entry->bw_share = bw_share; __entry->max_rate = max_rate; - __entry->group = vport->qos.group; + __entry->parent = vport->qos.parent; ), - TP_printk("(%s) vport=%hu sched_elem_ix=%u bw_share=%u, max_rate=%u group=%p\n", + TP_printk("(%s) vport=%hu sched_elem_ix=%u bw_share=%u, max_rate=%u parent=%p\n", __get_str(devname), __entry->vport_id, __entry->sched_elem_ix, - __entry->bw_share, __entry->max_rate, __entry->group + __entry->bw_share, __entry->max_rate, __entry->parent ) ); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index dd6fe729f456..837c4dda814d 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -84,11 +84,11 @@ struct mlx5_esw_rate_group { struct list_head members; }; -static void esw_qos_vport_set_group(struct mlx5_vport *vport, struct mlx5_esw_rate_group *group) +static void esw_qos_vport_set_parent(struct mlx5_vport *vport, struct mlx5_esw_rate_group *parent) { - list_del_init(&vport->qos.group_entry); - vport->qos.group = group; - list_add_tail(&vport->qos.group_entry, &group->members); + list_del_init(&vport->qos.parent_entry); + vport->qos.parent = parent; + list_add_tail(&vport->qos.parent_entry, &parent->members); } static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_ix, @@ -131,7 +131,7 @@ static int esw_qos_vport_config(struct mlx5_vport *vport, u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) { - struct mlx5_core_dev *dev = vport->qos.group->esw->dev; + struct mlx5_core_dev *dev = vport->qos.parent->esw->dev; int err; err = esw_qos_sched_elem_config(dev, vport->qos.esw_sched_elem_ix, max_rate, bw_share); @@ -157,7 +157,7 @@ static u32 esw_qos_calculate_group_min_rate_divider(struct mlx5_esw_rate_group * /* Find max min_rate across all vports in this group. * This will correspond to fw_max_bw_share in the final bw_share calculation. */ - list_for_each_entry(vport, &group->members, qos.group_entry) { + list_for_each_entry(vport, &group->members, qos.parent_entry) { if (vport->qos.min_rate > max_guarantee) max_guarantee = vport->qos.min_rate; } @@ -217,7 +217,7 @@ static int esw_qos_normalize_group_min_rate(struct mlx5_esw_rate_group *group, u32 bw_share; int err; - list_for_each_entry(vport, &group->members, qos.group_entry) { + list_for_each_entry(vport, &group->members, qos.parent_entry) { bw_share = esw_qos_calc_bw_share(vport->qos.min_rate, divider, fw_max_bw_share); if (bw_share == vport->qos.bw_share) @@ -286,7 +286,7 @@ static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport, previous_min_rate = vport->qos.min_rate; vport->qos.min_rate = min_rate; - err = esw_qos_normalize_group_min_rate(vport->qos.group, extack); + err = esw_qos_normalize_group_min_rate(vport->qos.parent, extack); if (err) vport->qos.min_rate = previous_min_rate; @@ -311,7 +311,7 @@ static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport, /* Use parent group limit if new max rate is 0. */ if (!max_rate) - act_max_rate = vport->qos.group->max_rate; + act_max_rate = vport->qos.parent->max_rate; err = esw_qos_vport_config(vport, act_max_rate, vport->qos.bw_share, extack); @@ -366,7 +366,7 @@ static int esw_qos_set_group_max_rate(struct mlx5_esw_rate_group *group, group->max_rate = max_rate; /* Any unlimited vports in the group should be set with the value of the group. */ - list_for_each_entry(vport, &group->members, qos.group_entry) { + list_for_each_entry(vport, &group->members, qos.parent_entry) { if (vport->qos.max_rate) continue; @@ -409,9 +409,9 @@ static int esw_qos_create_group_sched_elem(struct mlx5_core_dev *dev, u32 parent static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, u32 max_rate, u32 bw_share) { + struct mlx5_esw_rate_group *parent = vport->qos.parent; u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; - struct mlx5_esw_rate_group *group = vport->qos.group; - struct mlx5_core_dev *dev = group->esw->dev; + struct mlx5_core_dev *dev = parent->esw->dev; void *attr; int err; @@ -424,7 +424,7 @@ static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT); attr = MLX5_ADDR_OF(scheduling_context, sched_ctx, element_attributes); MLX5_SET(vport_element, attr, vport_number, vport->vport); - MLX5_SET(scheduling_context, sched_ctx, parent_element_id, group->tsar_ix); + MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent->tsar_ix); MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate); MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share); @@ -458,7 +458,7 @@ static int esw_qos_update_group_scheduling_element(struct mlx5_vport *vport, return err; } - esw_qos_vport_set_group(vport, new_group); + esw_qos_vport_set_parent(vport, new_group); /* Use new group max rate if vport max rate is unlimited. */ max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_group->max_rate; err = esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share); @@ -470,7 +470,7 @@ static int esw_qos_update_group_scheduling_element(struct mlx5_vport *vport, return 0; err_sched: - esw_qos_vport_set_group(vport, curr_group); + esw_qos_vport_set_parent(vport, curr_group); max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_group->max_rate; if (esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share)) esw_warn(curr_group->esw->dev, "E-Switch vport group restore failed (vport=%d)\n", @@ -488,7 +488,7 @@ static int esw_qos_vport_update_group(struct mlx5_vport *vport, int err; esw_assert_qos_lock_held(esw); - curr_group = vport->qos.group; + curr_group = vport->qos.parent; new_group = group ?: esw->qos.group0; if (curr_group == new_group) return 0; @@ -715,8 +715,8 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, if (err) return err; - INIT_LIST_HEAD(&vport->qos.group_entry); - esw_qos_vport_set_group(vport, esw->qos.group0); + INIT_LIST_HEAD(&vport->qos.parent_entry); + esw_qos_vport_set_parent(vport, esw->qos.group0); err = esw_qos_vport_create_sched_element(vport, max_rate, bw_share); if (err) @@ -743,10 +743,10 @@ void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport) esw_qos_lock(esw); if (!vport->qos.enabled) goto unlock; - WARN(vport->qos.group != esw->qos.group0, + WARN(vport->qos.parent != esw->qos.group0, "Disabling QoS on port before detaching it from group"); - dev = vport->qos.group->esw->dev; + dev = vport->qos.parent->esw->dev; err = mlx5_destroy_scheduling_element_cmd(dev, SCHEDULING_HIERARCHY_E_SWITCH, vport->qos.esw_sched_elem_ix); @@ -888,7 +888,7 @@ int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 /* Eswitch QoS wasn't enabled yet. Enable it and vport QoS. */ err = esw_qos_vport_enable(vport, rate_mbps, vport->qos.bw_share, NULL); } else { - struct mlx5_core_dev *dev = vport->qos.group->esw->dev; + struct mlx5_core_dev *dev = vport->qos.parent->esw->dev; MLX5_SET(scheduling_context, ctx, max_average_bw, rate_mbps); bitmask = MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h index 3b901bd36d4b..e789fb14989b 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h @@ -221,8 +221,10 @@ struct mlx5_vport { u32 max_rate; /* A computed value indicating relative min_rate between vports in a group. */ u32 bw_share; - struct mlx5_esw_rate_group *group; - struct list_head group_entry; + /* The parent group of this vport scheduling element. */ + struct mlx5_esw_rate_group *parent; + /* Membership in the parent 'members' list. */ + struct list_head parent_entry; } qos; u16 vport; -- 2.51.0 From 1c25d4388ba69bd67a29b20278f9da46d092dd72 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:08 +0300 Subject: [PATCH 11/16] net/mlx5: Introduce node struct and rename group terminology to node Introduce the `mlx5_esw_sched_node` struct, consolidating all rate hierarchy related details, including membership and scheduling parameters. Since the group concept aligns with the `mlx5_esw_sched_node`, replace the `mlx5_esw_rate_group` struct with it and rename the "group" terminology to "node" throughout the rate hierarchy. All relevant code paths and structures have been updated to use the "node" terminology accordingly, laying the groundwork for future patches that will unify the handling of different types of members within the rate hierarchy. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Reviewed-by: Daniel Machon Signed-off-by: Paolo Abeni --- .../mellanox/mlx5/core/esw/devlink_port.c | 2 +- .../mlx5/core/esw/diag/qos_tracepoint.h | 40 +- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 374 +++++++++--------- .../net/ethernet/mellanox/mlx5/core/eswitch.h | 16 +- 4 files changed, 217 insertions(+), 215 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c index 86af1891395f..d0f38818363f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c @@ -195,7 +195,7 @@ void mlx5_esw_offloads_devlink_port_unregister(struct mlx5_vport *vport) return; dl_port = vport->dl_port; - mlx5_esw_qos_vport_update_group(vport, NULL, NULL); + mlx5_esw_qos_vport_update_node(vport, NULL, NULL); devl_rate_leaf_destroy(&dl_port->dl_port); devl_port_unregister(&dl_port->dl_port); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h index 2aea01959073..0b50ef0871f2 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h @@ -62,57 +62,57 @@ DEFINE_EVENT(mlx5_esw_vport_qos_template, mlx5_esw_vport_qos_config, TP_ARGS(dev, vport, bw_share, max_rate) ); -DECLARE_EVENT_CLASS(mlx5_esw_group_qos_template, +DECLARE_EVENT_CLASS(mlx5_esw_node_qos_template, TP_PROTO(const struct mlx5_core_dev *dev, - const struct mlx5_esw_rate_group *group, + const struct mlx5_esw_sched_node *node, unsigned int tsar_ix), - TP_ARGS(dev, group, tsar_ix), + TP_ARGS(dev, node, tsar_ix), TP_STRUCT__entry(__string(devname, dev_name(dev->device)) - __field(const void *, group) + __field(const void *, node) __field(unsigned int, tsar_ix) ), TP_fast_assign(__assign_str(devname); - __entry->group = group; + __entry->node = node; __entry->tsar_ix = tsar_ix; ), - TP_printk("(%s) group=%p tsar_ix=%u\n", - __get_str(devname), __entry->group, __entry->tsar_ix + TP_printk("(%s) node=%p tsar_ix=%u\n", + __get_str(devname), __entry->node, __entry->tsar_ix ) ); -DEFINE_EVENT(mlx5_esw_group_qos_template, mlx5_esw_group_qos_create, +DEFINE_EVENT(mlx5_esw_node_qos_template, mlx5_esw_node_qos_create, TP_PROTO(const struct mlx5_core_dev *dev, - const struct mlx5_esw_rate_group *group, + const struct mlx5_esw_sched_node *node, unsigned int tsar_ix), - TP_ARGS(dev, group, tsar_ix) + TP_ARGS(dev, node, tsar_ix) ); -DEFINE_EVENT(mlx5_esw_group_qos_template, mlx5_esw_group_qos_destroy, +DEFINE_EVENT(mlx5_esw_node_qos_template, mlx5_esw_node_qos_destroy, TP_PROTO(const struct mlx5_core_dev *dev, - const struct mlx5_esw_rate_group *group, + const struct mlx5_esw_sched_node *node, unsigned int tsar_ix), - TP_ARGS(dev, group, tsar_ix) + TP_ARGS(dev, node, tsar_ix) ); -TRACE_EVENT(mlx5_esw_group_qos_config, +TRACE_EVENT(mlx5_esw_node_qos_config, TP_PROTO(const struct mlx5_core_dev *dev, - const struct mlx5_esw_rate_group *group, + const struct mlx5_esw_sched_node *node, unsigned int tsar_ix, u32 bw_share, u32 max_rate), - TP_ARGS(dev, group, tsar_ix, bw_share, max_rate), + TP_ARGS(dev, node, tsar_ix, bw_share, max_rate), TP_STRUCT__entry(__string(devname, dev_name(dev->device)) - __field(const void *, group) + __field(const void *, node) __field(unsigned int, tsar_ix) __field(unsigned int, bw_share) __field(unsigned int, max_rate) ), TP_fast_assign(__assign_str(devname); - __entry->group = group; + __entry->node = node; __entry->tsar_ix = tsar_ix; __entry->bw_share = bw_share; __entry->max_rate = max_rate; ), - TP_printk("(%s) group=%p tsar_ix=%u bw_share=%u max_rate=%u\n", - __get_str(devname), __entry->group, __entry->tsar_ix, + TP_printk("(%s) node=%p tsar_ix=%u bw_share=%u max_rate=%u\n", + __get_str(devname), __entry->node, __entry->tsar_ix, __entry->bw_share, __entry->max_rate ) ); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 837c4dda814d..840568c66a1a 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -11,12 +11,12 @@ /* Minimum supported BW share value by the HW is 1 Mbit/sec */ #define MLX5_MIN_BW_SHARE 1 -/* Holds rate groups associated with an E-Switch. */ +/* Holds rate nodes associated with an E-Switch. */ struct mlx5_qos_domain { /* Serializes access to all qos changes in the qos domain. */ struct mutex lock; - /* List of all mlx5_esw_rate_groups. */ - struct list_head groups; + /* List of all mlx5_esw_sched_nodes. */ + struct list_head nodes; }; static void esw_qos_lock(struct mlx5_eswitch *esw) @@ -43,7 +43,7 @@ static struct mlx5_qos_domain *esw_qos_domain_alloc(void) return NULL; mutex_init(&qos_domain->lock); - INIT_LIST_HEAD(&qos_domain->groups); + INIT_LIST_HEAD(&qos_domain->nodes); return qos_domain; } @@ -65,30 +65,30 @@ enum sched_node_type { SCHED_NODE_TYPE_VPORTS_TSAR, }; -struct mlx5_esw_rate_group { - u32 tsar_ix; +struct mlx5_esw_sched_node { + u32 ix; /* Bandwidth parameters. */ u32 max_rate; u32 min_rate; - /* A computed value indicating relative min_rate between group members. */ + /* A computed value indicating relative min_rate between node's children. */ u32 bw_share; - /* The parent group of this group. */ - struct mlx5_esw_rate_group *parent; - /* Membership in the parent list. */ - struct list_head parent_entry; - /* The type of this group node in the rate hierarchy. */ + /* The parent node in the rate hierarchy. */ + struct mlx5_esw_sched_node *parent; + /* Entry in the parent node's children list. */ + struct list_head entry; + /* The type of this node in the rate hierarchy. */ enum sched_node_type type; - /* The eswitch this group belongs to. */ + /* The eswitch this node belongs to. */ struct mlx5_eswitch *esw; - /* Members of this group.*/ - struct list_head members; + /* The children nodes of this node, empty list for leaf nodes. */ + struct list_head children; }; -static void esw_qos_vport_set_parent(struct mlx5_vport *vport, struct mlx5_esw_rate_group *parent) +static void esw_qos_vport_set_parent(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent) { list_del_init(&vport->qos.parent_entry); vport->qos.parent = parent; - list_add_tail(&vport->qos.parent_entry, &parent->members); + list_add_tail(&vport->qos.parent_entry, &parent->children); } static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_ix, @@ -112,17 +112,17 @@ static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_i bitmask); } -static int esw_qos_group_config(struct mlx5_esw_rate_group *group, - u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) +static int esw_qos_node_config(struct mlx5_esw_sched_node *node, + u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) { - struct mlx5_core_dev *dev = group->esw->dev; + struct mlx5_core_dev *dev = node->esw->dev; int err; - err = esw_qos_sched_elem_config(dev, group->tsar_ix, max_rate, bw_share); + err = esw_qos_sched_elem_config(dev, node->ix, max_rate, bw_share); if (err) - NL_SET_ERR_MSG_MOD(extack, "E-Switch modify group TSAR element failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch modify node TSAR element failed"); - trace_mlx5_esw_group_qos_config(dev, group, group->tsar_ix, bw_share, max_rate); + trace_mlx5_esw_node_qos_config(dev, node, node->ix, bw_share, max_rate); return err; } @@ -148,16 +148,16 @@ static int esw_qos_vport_config(struct mlx5_vport *vport, return 0; } -static u32 esw_qos_calculate_group_min_rate_divider(struct mlx5_esw_rate_group *group) +static u32 esw_qos_calculate_node_min_rate_divider(struct mlx5_esw_sched_node *node) { - u32 fw_max_bw_share = MLX5_CAP_QOS(group->esw->dev, max_tsar_bw_share); + u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); struct mlx5_vport *vport; u32 max_guarantee = 0; - /* Find max min_rate across all vports in this group. + /* Find max min_rate across all vports in this node. * This will correspond to fw_max_bw_share in the final bw_share calculation. */ - list_for_each_entry(vport, &group->members, qos.parent_entry) { + list_for_each_entry(vport, &node->children, qos.parent_entry) { if (vport->qos.min_rate > max_guarantee) max_guarantee = vport->qos.min_rate; } @@ -165,13 +165,13 @@ static u32 esw_qos_calculate_group_min_rate_divider(struct mlx5_esw_rate_group * if (max_guarantee) return max_t(u32, max_guarantee / fw_max_bw_share, 1); - /* If vports max min_rate divider is 0 but their group has bw_share + /* If vports max min_rate divider is 0 but their node has bw_share * configured, then set bw_share for vports to minimal value. */ - if (group->bw_share) + if (node->bw_share) return 1; - /* A divider of 0 sets bw_share for all group vports to 0, + /* A divider of 0 sets bw_share for all node vports to 0, * effectively disabling min guarantees. */ return 0; @@ -180,23 +180,23 @@ static u32 esw_qos_calculate_group_min_rate_divider(struct mlx5_esw_rate_group * static u32 esw_qos_calculate_min_rate_divider(struct mlx5_eswitch *esw) { u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share); - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; u32 max_guarantee = 0; - /* Find max min_rate across all esw groups. + /* Find max min_rate across all esw nodes. * This will correspond to fw_max_bw_share in the final bw_share calculation. */ - list_for_each_entry(group, &esw->qos.domain->groups, parent_entry) { - if (group->esw == esw && group->tsar_ix != esw->qos.root_tsar_ix && - group->min_rate > max_guarantee) - max_guarantee = group->min_rate; + list_for_each_entry(node, &esw->qos.domain->nodes, entry) { + if (node->esw == esw && node->ix != esw->qos.root_tsar_ix && + node->min_rate > max_guarantee) + max_guarantee = node->min_rate; } if (max_guarantee) return max_t(u32, max_guarantee / fw_max_bw_share, 1); - /* If no group has min_rate configured, a divider of 0 sets all - * groups' bw_share to 0, effectively disabling min guarantees. + /* If no node has min_rate configured, a divider of 0 sets all + * nodes' bw_share to 0, effectively disabling min guarantees. */ return 0; } @@ -208,16 +208,16 @@ static u32 esw_qos_calc_bw_share(u32 min_rate, u32 divider, u32 fw_max) return min_t(u32, max_t(u32, DIV_ROUND_UP(min_rate, divider), MLX5_MIN_BW_SHARE), fw_max); } -static int esw_qos_normalize_group_min_rate(struct mlx5_esw_rate_group *group, - struct netlink_ext_ack *extack) +static int esw_qos_normalize_node_min_rate(struct mlx5_esw_sched_node *node, + struct netlink_ext_ack *extack) { - u32 fw_max_bw_share = MLX5_CAP_QOS(group->esw->dev, max_tsar_bw_share); - u32 divider = esw_qos_calculate_group_min_rate_divider(group); + u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); + u32 divider = esw_qos_calculate_node_min_rate_divider(node); struct mlx5_vport *vport; u32 bw_share; int err; - list_for_each_entry(vport, &group->members, qos.parent_entry) { + list_for_each_entry(vport, &node->children, qos.parent_entry) { bw_share = esw_qos_calc_bw_share(vport->qos.min_rate, divider, fw_max_bw_share); if (bw_share == vport->qos.bw_share) @@ -237,28 +237,29 @@ static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw, struct netlink_e { u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share); u32 divider = esw_qos_calculate_min_rate_divider(esw); - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; u32 bw_share; int err; - list_for_each_entry(group, &esw->qos.domain->groups, parent_entry) { - if (group->esw != esw || group->tsar_ix == esw->qos.root_tsar_ix) + list_for_each_entry(node, &esw->qos.domain->nodes, entry) { + if (node->esw != esw || node->ix == esw->qos.root_tsar_ix) continue; - bw_share = esw_qos_calc_bw_share(group->min_rate, divider, fw_max_bw_share); + bw_share = esw_qos_calc_bw_share(node->min_rate, divider, + fw_max_bw_share); - if (bw_share == group->bw_share) + if (bw_share == node->bw_share) continue; - err = esw_qos_group_config(group, group->max_rate, bw_share, extack); + err = esw_qos_node_config(node, node->max_rate, bw_share, extack); if (err) return err; - group->bw_share = bw_share; + node->bw_share = bw_share; - /* All the group's vports need to be set with default bw_share + /* All the node's vports need to be set with default bw_share * to enable them with QOS */ - err = esw_qos_normalize_group_min_rate(group, extack); + err = esw_qos_normalize_node_min_rate(node, extack); if (err) return err; @@ -286,7 +287,7 @@ static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport, previous_min_rate = vport->qos.min_rate; vport->qos.min_rate = min_rate; - err = esw_qos_normalize_group_min_rate(vport->qos.parent, extack); + err = esw_qos_normalize_node_min_rate(vport->qos.parent, extack); if (err) vport->qos.min_rate = previous_min_rate; @@ -309,7 +310,7 @@ static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport, if (max_rate == vport->qos.max_rate) return 0; - /* Use parent group limit if new max rate is 0. */ + /* Use parent node limit if new max rate is 0. */ if (!max_rate) act_max_rate = vport->qos.parent->max_rate; @@ -321,10 +322,10 @@ static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport, return err; } -static int esw_qos_set_group_min_rate(struct mlx5_esw_rate_group *group, - u32 min_rate, struct netlink_ext_ack *extack) +static int esw_qos_set_node_min_rate(struct mlx5_esw_sched_node *node, + u32 min_rate, struct netlink_ext_ack *extack) { - struct mlx5_eswitch *esw = group->esw; + struct mlx5_eswitch *esw = node->esw; u32 previous_min_rate; int err; @@ -332,17 +333,17 @@ static int esw_qos_set_group_min_rate(struct mlx5_esw_rate_group *group, MLX5_CAP_QOS(esw->dev, max_tsar_bw_share) < MLX5_MIN_BW_SHARE) return -EOPNOTSUPP; - if (min_rate == group->min_rate) + if (min_rate == node->min_rate) return 0; - previous_min_rate = group->min_rate; - group->min_rate = min_rate; + previous_min_rate = node->min_rate; + node->min_rate = min_rate; err = esw_qos_normalize_min_rate(esw, extack); if (err) { - NL_SET_ERR_MSG_MOD(extack, "E-Switch group min rate setting failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch node min rate setting failed"); /* Attempt restoring previous configuration */ - group->min_rate = previous_min_rate; + node->min_rate = previous_min_rate; if (esw_qos_normalize_min_rate(esw, extack)) NL_SET_ERR_MSG_MOD(extack, "E-Switch BW share restore failed"); } @@ -350,23 +351,23 @@ static int esw_qos_set_group_min_rate(struct mlx5_esw_rate_group *group, return err; } -static int esw_qos_set_group_max_rate(struct mlx5_esw_rate_group *group, - u32 max_rate, struct netlink_ext_ack *extack) +static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node, + u32 max_rate, struct netlink_ext_ack *extack) { struct mlx5_vport *vport; int err; - if (group->max_rate == max_rate) + if (node->max_rate == max_rate) return 0; - err = esw_qos_group_config(group, max_rate, group->bw_share, extack); + err = esw_qos_node_config(node, max_rate, node->bw_share, extack); if (err) return err; - group->max_rate = max_rate; + node->max_rate = max_rate; - /* Any unlimited vports in the group should be set with the value of the group. */ - list_for_each_entry(vport, &group->members, qos.parent_entry) { + /* Any unlimited vports in the node should be set with the value of the node. */ + list_for_each_entry(vport, &node->children, qos.parent_entry) { if (vport->qos.max_rate) continue; @@ -379,8 +380,8 @@ static int esw_qos_set_group_max_rate(struct mlx5_esw_rate_group *group, return err; } -static int esw_qos_create_group_sched_elem(struct mlx5_core_dev *dev, u32 parent_element_id, - u32 *tsar_ix) +static int esw_qos_create_node_sched_elem(struct mlx5_core_dev *dev, u32 parent_element_id, + u32 *tsar_ix) { u32 tsar_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; void *attr; @@ -409,7 +410,7 @@ static int esw_qos_create_group_sched_elem(struct mlx5_core_dev *dev, u32 parent static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, u32 max_rate, u32 bw_share) { - struct mlx5_esw_rate_group *parent = vport->qos.parent; + struct mlx5_esw_sched_node *parent = vport->qos.parent; u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; struct mlx5_core_dev *dev = parent->esw->dev; void *attr; @@ -424,7 +425,7 @@ static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT); attr = MLX5_ADDR_OF(scheduling_context, sched_ctx, element_attributes); MLX5_SET(vport_element, attr, vport_number, vport->vport); - MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent->tsar_ix); + MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent->ix); MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate); MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share); @@ -442,15 +443,15 @@ static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, return 0; } -static int esw_qos_update_group_scheduling_element(struct mlx5_vport *vport, - struct mlx5_esw_rate_group *curr_group, - struct mlx5_esw_rate_group *new_group, - struct netlink_ext_ack *extack) +static int esw_qos_update_node_scheduling_element(struct mlx5_vport *vport, + struct mlx5_esw_sched_node *curr_node, + struct mlx5_esw_sched_node *new_node, + struct netlink_ext_ack *extack) { u32 max_rate; int err; - err = mlx5_destroy_scheduling_element_cmd(curr_group->esw->dev, + err = mlx5_destroy_scheduling_element_cmd(curr_node->esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, vport->qos.esw_sched_elem_ix); if (err) { @@ -458,129 +459,129 @@ static int esw_qos_update_group_scheduling_element(struct mlx5_vport *vport, return err; } - esw_qos_vport_set_parent(vport, new_group); - /* Use new group max rate if vport max rate is unlimited. */ - max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_group->max_rate; + esw_qos_vport_set_parent(vport, new_node); + /* Use new node max rate if vport max rate is unlimited. */ + max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_node->max_rate; err = esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share); if (err) { - NL_SET_ERR_MSG_MOD(extack, "E-Switch vport group set failed."); + NL_SET_ERR_MSG_MOD(extack, "E-Switch vport node set failed."); goto err_sched; } return 0; err_sched: - esw_qos_vport_set_parent(vport, curr_group); - max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_group->max_rate; + esw_qos_vport_set_parent(vport, curr_node); + max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_node->max_rate; if (esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share)) - esw_warn(curr_group->esw->dev, "E-Switch vport group restore failed (vport=%d)\n", + esw_warn(curr_node->esw->dev, "E-Switch vport node restore failed (vport=%d)\n", vport->vport); return err; } -static int esw_qos_vport_update_group(struct mlx5_vport *vport, - struct mlx5_esw_rate_group *group, - struct netlink_ext_ack *extack) +static int esw_qos_vport_update_node(struct mlx5_vport *vport, + struct mlx5_esw_sched_node *node, + struct netlink_ext_ack *extack) { struct mlx5_eswitch *esw = vport->dev->priv.eswitch; - struct mlx5_esw_rate_group *new_group, *curr_group; + struct mlx5_esw_sched_node *new_node, *curr_node; int err; esw_assert_qos_lock_held(esw); - curr_group = vport->qos.parent; - new_group = group ?: esw->qos.group0; - if (curr_group == new_group) + curr_node = vport->qos.parent; + new_node = node ?: esw->qos.node0; + if (curr_node == new_node) return 0; - err = esw_qos_update_group_scheduling_element(vport, curr_group, new_group, extack); + err = esw_qos_update_node_scheduling_element(vport, curr_node, new_node, extack); if (err) return err; - /* Recalculate bw share weights of old and new groups */ - if (vport->qos.bw_share || new_group->bw_share) { - esw_qos_normalize_group_min_rate(curr_group, extack); - esw_qos_normalize_group_min_rate(new_group, extack); + /* Recalculate bw share weights of old and new nodes */ + if (vport->qos.bw_share || new_node->bw_share) { + esw_qos_normalize_node_min_rate(curr_node, extack); + esw_qos_normalize_node_min_rate(new_node, extack); } return 0; } -static struct mlx5_esw_rate_group * -__esw_qos_alloc_rate_group(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_node_type type, - struct mlx5_esw_rate_group *parent) +static struct mlx5_esw_sched_node * +__esw_qos_alloc_node(struct mlx5_eswitch *esw, u32 tsar_ix, enum sched_node_type type, + struct mlx5_esw_sched_node *parent) { - struct mlx5_esw_rate_group *group; - struct list_head *parent_list; + struct list_head *parent_children; + struct mlx5_esw_sched_node *node; - group = kzalloc(sizeof(*group), GFP_KERNEL); - if (!group) + node = kzalloc(sizeof(*node), GFP_KERNEL); + if (!node) return NULL; - group->esw = esw; - group->tsar_ix = tsar_ix; - group->type = type; - group->parent = parent; - INIT_LIST_HEAD(&group->members); - parent_list = parent ? &parent->members : &esw->qos.domain->groups; - list_add_tail(&group->parent_entry, parent_list); + node->esw = esw; + node->ix = tsar_ix; + node->type = type; + node->parent = parent; + INIT_LIST_HEAD(&node->children); + parent_children = parent ? &parent->children : &esw->qos.domain->nodes; + list_add_tail(&node->entry, parent_children); - return group; + return node; } -static void __esw_qos_free_rate_group(struct mlx5_esw_rate_group *group) +static void __esw_qos_free_node(struct mlx5_esw_sched_node *node) { - list_del(&group->parent_entry); - kfree(group); + list_del(&node->entry); + kfree(node); } -static struct mlx5_esw_rate_group * -__esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct mlx5_esw_rate_group *parent, +static struct mlx5_esw_sched_node * +__esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct mlx5_esw_sched_node *parent, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; u32 tsar_ix; int err; - err = esw_qos_create_group_sched_elem(esw->dev, esw->qos.root_tsar_ix, &tsar_ix); + err = esw_qos_create_node_sched_elem(esw->dev, esw->qos.root_tsar_ix, &tsar_ix); if (err) { - NL_SET_ERR_MSG_MOD(extack, "E-Switch create TSAR for group failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch create TSAR for node failed"); return ERR_PTR(err); } - group = __esw_qos_alloc_rate_group(esw, tsar_ix, SCHED_NODE_TYPE_VPORTS_TSAR, parent); - if (!group) { - NL_SET_ERR_MSG_MOD(extack, "E-Switch alloc group failed"); + node = __esw_qos_alloc_node(esw, tsar_ix, SCHED_NODE_TYPE_VPORTS_TSAR, parent); + if (!node) { + NL_SET_ERR_MSG_MOD(extack, "E-Switch alloc node failed"); err = -ENOMEM; - goto err_alloc_group; + goto err_alloc_node; } err = esw_qos_normalize_min_rate(esw, extack); if (err) { - NL_SET_ERR_MSG_MOD(extack, "E-Switch groups normalization failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed"); goto err_min_rate; } - trace_mlx5_esw_group_qos_create(esw->dev, group, group->tsar_ix); + trace_mlx5_esw_node_qos_create(esw->dev, node, node->ix); - return group; + return node; err_min_rate: - __esw_qos_free_rate_group(group); -err_alloc_group: + __esw_qos_free_node(node); +err_alloc_node: if (mlx5_destroy_scheduling_element_cmd(esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, tsar_ix)) - NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR for group failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR for node failed"); return ERR_PTR(err); } static int esw_qos_get(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack); static void esw_qos_put(struct mlx5_eswitch *esw); -static struct mlx5_esw_rate_group * -esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) +static struct mlx5_esw_sched_node * +esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; int err; esw_assert_qos_lock_held(esw); @@ -591,31 +592,30 @@ esw_qos_create_vports_rate_group(struct mlx5_eswitch *esw, struct netlink_ext_ac if (err) return ERR_PTR(err); - group = __esw_qos_create_vports_rate_group(esw, NULL, extack); - if (IS_ERR(group)) + node = __esw_qos_create_vports_sched_node(esw, NULL, extack); + if (IS_ERR(node)) esw_qos_put(esw); - return group; + return node; } -static int __esw_qos_destroy_rate_group(struct mlx5_esw_rate_group *group, - struct netlink_ext_ack *extack) +static int __esw_qos_destroy_node(struct mlx5_esw_sched_node *node, struct netlink_ext_ack *extack) { - struct mlx5_eswitch *esw = group->esw; + struct mlx5_eswitch *esw = node->esw; int err; - trace_mlx5_esw_group_qos_destroy(esw->dev, group, group->tsar_ix); + trace_mlx5_esw_node_qos_destroy(esw->dev, node, node->ix); err = mlx5_destroy_scheduling_element_cmd(esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, - group->tsar_ix); + node->ix); if (err) NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR_ID failed"); - __esw_qos_free_rate_group(group); + __esw_qos_free_node(node); err = esw_qos_normalize_min_rate(esw, extack); if (err) - NL_SET_ERR_MSG_MOD(extack, "E-Switch groups normalization failed"); + NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed"); return err; @@ -629,32 +629,34 @@ static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *exta if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling)) return -EOPNOTSUPP; - err = esw_qos_create_group_sched_elem(esw->dev, 0, &esw->qos.root_tsar_ix); + err = esw_qos_create_node_sched_elem(esw->dev, 0, &esw->qos.root_tsar_ix); if (err) { esw_warn(dev, "E-Switch create root TSAR failed (%d)\n", err); return err; } if (MLX5_CAP_QOS(dev, log_esw_max_sched_depth)) { - esw->qos.group0 = __esw_qos_create_vports_rate_group(esw, NULL, extack); + esw->qos.node0 = __esw_qos_create_vports_sched_node(esw, NULL, extack); } else { - /* The eswitch doesn't support scheduling groups. - * Create a software-only group0 using the root TSAR to attach vport QoS to. + /* The eswitch doesn't support scheduling nodes. + * Create a software-only node0 using the root TSAR to attach vport QoS to. */ - if (!__esw_qos_alloc_rate_group(esw, esw->qos.root_tsar_ix, - SCHED_NODE_TYPE_VPORTS_TSAR, NULL)) - esw->qos.group0 = ERR_PTR(-ENOMEM); + if (!__esw_qos_alloc_node(esw, + esw->qos.root_tsar_ix, + SCHED_NODE_TYPE_VPORTS_TSAR, + NULL)) + esw->qos.node0 = ERR_PTR(-ENOMEM); } - if (IS_ERR(esw->qos.group0)) { - err = PTR_ERR(esw->qos.group0); - esw_warn(dev, "E-Switch create rate group 0 failed (%d)\n", err); - goto err_group0; + if (IS_ERR(esw->qos.node0)) { + err = PTR_ERR(esw->qos.node0); + esw_warn(dev, "E-Switch create rate node 0 failed (%d)\n", err); + goto err_node0; } refcount_set(&esw->qos.refcnt, 1); return 0; -err_group0: +err_node0: if (mlx5_destroy_scheduling_element_cmd(esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, esw->qos.root_tsar_ix)) esw_warn(esw->dev, "E-Switch destroy root TSAR failed.\n"); @@ -666,11 +668,11 @@ static void esw_qos_destroy(struct mlx5_eswitch *esw) { int err; - if (esw->qos.group0->tsar_ix != esw->qos.root_tsar_ix) - __esw_qos_destroy_rate_group(esw->qos.group0, NULL); + if (esw->qos.node0->ix != esw->qos.root_tsar_ix) + __esw_qos_destroy_node(esw->qos.node0, NULL); else - __esw_qos_free_rate_group(esw->qos.group0); - esw->qos.group0 = NULL; + __esw_qos_free_node(esw->qos.node0); + esw->qos.node0 = NULL; err = mlx5_destroy_scheduling_element_cmd(esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, @@ -716,7 +718,7 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, return err; INIT_LIST_HEAD(&vport->qos.parent_entry); - esw_qos_vport_set_parent(vport, esw->qos.group0); + esw_qos_vport_set_parent(vport, esw->qos.node0); err = esw_qos_vport_create_sched_element(vport, max_rate, bw_share); if (err) @@ -743,8 +745,8 @@ void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport) esw_qos_lock(esw); if (!vport->qos.enabled) goto unlock; - WARN(vport->qos.parent != esw->qos.group0, - "Disabling QoS on port before detaching it from group"); + WARN(vport->qos.parent != esw->qos.node0, + "Disabling QoS on port before detaching it from node"); dev = vport->qos.parent->esw->dev; err = mlx5_destroy_scheduling_element_cmd(dev, @@ -1004,8 +1006,8 @@ unlock: int mlx5_esw_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void *priv, u64 tx_share, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group = priv; - struct mlx5_eswitch *esw = group->esw; + struct mlx5_esw_sched_node *node = priv; + struct mlx5_eswitch *esw = node->esw; int err; err = esw_qos_devlink_rate_to_mbps(esw->dev, "tx_share", &tx_share, extack); @@ -1013,7 +1015,7 @@ int mlx5_esw_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void return err; esw_qos_lock(esw); - err = esw_qos_set_group_min_rate(group, tx_share, extack); + err = esw_qos_set_node_min_rate(node, tx_share, extack); esw_qos_unlock(esw); return err; } @@ -1021,8 +1023,8 @@ int mlx5_esw_devlink_rate_node_tx_share_set(struct devlink_rate *rate_node, void int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void *priv, u64 tx_max, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group = priv; - struct mlx5_eswitch *esw = group->esw; + struct mlx5_esw_sched_node *node = priv; + struct mlx5_eswitch *esw = node->esw; int err; err = esw_qos_devlink_rate_to_mbps(esw->dev, "tx_max", &tx_max, extack); @@ -1030,7 +1032,7 @@ int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void * return err; esw_qos_lock(esw); - err = esw_qos_set_group_max_rate(group, tx_max, extack); + err = esw_qos_set_node_max_rate(node, tx_max, extack); esw_qos_unlock(esw); return err; } @@ -1038,7 +1040,7 @@ int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void * int mlx5_esw_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; struct mlx5_eswitch *esw; int err = 0; @@ -1054,13 +1056,13 @@ int mlx5_esw_devlink_rate_node_new(struct devlink_rate *rate_node, void **priv, goto unlock; } - group = esw_qos_create_vports_rate_group(esw, extack); - if (IS_ERR(group)) { - err = PTR_ERR(group); + node = esw_qos_create_vports_sched_node(esw, extack); + if (IS_ERR(node)) { + err = PTR_ERR(node); goto unlock; } - *priv = group; + *priv = node; unlock: esw_qos_unlock(esw); return err; @@ -1069,36 +1071,36 @@ unlock: int mlx5_esw_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group = priv; - struct mlx5_eswitch *esw = group->esw; + struct mlx5_esw_sched_node *node = priv; + struct mlx5_eswitch *esw = node->esw; int err; esw_qos_lock(esw); - err = __esw_qos_destroy_rate_group(group, extack); + err = __esw_qos_destroy_node(node, extack); esw_qos_put(esw); esw_qos_unlock(esw); return err; } -int mlx5_esw_qos_vport_update_group(struct mlx5_vport *vport, - struct mlx5_esw_rate_group *group, - struct netlink_ext_ack *extack) +int mlx5_esw_qos_vport_update_node(struct mlx5_vport *vport, + struct mlx5_esw_sched_node *node, + struct netlink_ext_ack *extack) { struct mlx5_eswitch *esw = vport->dev->priv.eswitch; int err = 0; - if (group && group->esw != esw) { + if (node && node->esw != esw) { NL_SET_ERR_MSG_MOD(extack, "Cross E-Switch scheduling is not supported"); return -EOPNOTSUPP; } esw_qos_lock(esw); - if (!vport->qos.enabled && !group) + if (!vport->qos.enabled && !node) goto unlock; err = esw_qos_vport_enable(vport, 0, 0, extack); if (!err) - err = esw_qos_vport_update_group(vport, group, extack); + err = esw_qos_vport_update_node(vport, node, extack); unlock: esw_qos_unlock(esw); return err; @@ -1109,12 +1111,12 @@ int mlx5_esw_devlink_rate_parent_set(struct devlink_rate *devlink_rate, void *priv, void *parent_priv, struct netlink_ext_ack *extack) { - struct mlx5_esw_rate_group *group; + struct mlx5_esw_sched_node *node; struct mlx5_vport *vport = priv; if (!parent) - return mlx5_esw_qos_vport_update_group(vport, NULL, extack); + return mlx5_esw_qos_vport_update_node(vport, NULL, extack); - group = parent_priv; - return mlx5_esw_qos_vport_update_group(vport, group, extack); + node = parent_priv; + return mlx5_esw_qos_vport_update_node(vport, node, extack); } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h index e789fb14989b..38f912f5a707 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h @@ -222,7 +222,7 @@ struct mlx5_vport { /* A computed value indicating relative min_rate between vports in a group. */ u32 bw_share; /* The parent group of this vport scheduling element. */ - struct mlx5_esw_rate_group *parent; + struct mlx5_esw_sched_node *parent; /* Membership in the parent 'members' list. */ struct list_head parent_entry; } qos; @@ -372,11 +372,11 @@ struct mlx5_eswitch { refcount_t refcnt; u32 root_tsar_ix; struct mlx5_qos_domain *domain; - /* Contains all vports with QoS enabled but no explicit group. - * Cannot be NULL if QoS is enabled, but may be a fake group - * referencing the root TSAR if the esw doesn't support groups. + /* Contains all vports with QoS enabled but no explicit node. + * Cannot be NULL if QoS is enabled, but may be a fake node + * referencing the root TSAR if the esw doesn't support nodes. */ - struct mlx5_esw_rate_group *group0; + struct mlx5_esw_sched_node *node0; } qos; struct mlx5_esw_bridge_offloads *br_offloads; @@ -436,9 +436,9 @@ int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw, u16 vport_num, bool setting); int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, u16 vport, u32 max_rate, u32 min_rate); -int mlx5_esw_qos_vport_update_group(struct mlx5_vport *vport, - struct mlx5_esw_rate_group *group, - struct netlink_ext_ack *extack); +int mlx5_esw_qos_vport_update_node(struct mlx5_vport *vport, + struct mlx5_esw_sched_node *node, + struct netlink_ext_ack *extack); int mlx5_eswitch_set_vepa(struct mlx5_eswitch *esw, u8 setting); int mlx5_eswitch_get_vepa(struct mlx5_eswitch *esw, u8 *setting); int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw, -- 2.51.0 From 88d5fbcb7ba0b94aeb054dd576ff6e43d026c1fe Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:09 +0300 Subject: [PATCH 12/16] net/mlx5: Refactor vport scheduling element creation function Modify the vport scheduling element creation function to get the parent node directly, aligning it with the group creation function. This ensures a consistent flow for scheduling elements creation, as the parent nodes already contain the device and parent element index. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 27 ++++++++++--------- 1 file changed, 15 insertions(+), 12 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 840568c66a1a..bcdb745994d2 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -407,10 +407,10 @@ static int esw_qos_create_node_sched_elem(struct mlx5_core_dev *dev, u32 parent_ tsar_ix); } -static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, - u32 max_rate, u32 bw_share) +static int +esw_qos_vport_create_sched_element(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent, + u32 max_rate, u32 bw_share, u32 *sched_elem_ix) { - struct mlx5_esw_sched_node *parent = vport->qos.parent; u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; struct mlx5_core_dev *dev = parent->esw->dev; void *attr; @@ -432,7 +432,7 @@ static int esw_qos_vport_create_sched_element(struct mlx5_vport *vport, err = mlx5_create_scheduling_element_cmd(dev, SCHEDULING_HIERARCHY_E_SWITCH, sched_ctx, - &vport->qos.esw_sched_elem_ix); + sched_elem_ix); if (err) { esw_warn(dev, "E-Switch create vport scheduling element failed (vport=%d,err=%d)\n", @@ -459,21 +459,23 @@ static int esw_qos_update_node_scheduling_element(struct mlx5_vport *vport, return err; } - esw_qos_vport_set_parent(vport, new_node); /* Use new node max rate if vport max rate is unlimited. */ max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_node->max_rate; - err = esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share); + err = esw_qos_vport_create_sched_element(vport, new_node, max_rate, vport->qos.bw_share, + &vport->qos.esw_sched_elem_ix); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch vport node set failed."); goto err_sched; } + esw_qos_vport_set_parent(vport, new_node); + return 0; err_sched: - esw_qos_vport_set_parent(vport, curr_node); max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_node->max_rate; - if (esw_qos_vport_create_sched_element(vport, max_rate, vport->qos.bw_share)) + if (esw_qos_vport_create_sched_element(vport, curr_node, max_rate, vport->qos.bw_share, + &vport->qos.esw_sched_elem_ix)) esw_warn(curr_node->esw->dev, "E-Switch vport node restore failed (vport=%d)\n", vport->vport); @@ -717,13 +719,14 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, if (err) return err; - INIT_LIST_HEAD(&vport->qos.parent_entry); - esw_qos_vport_set_parent(vport, esw->qos.node0); - - err = esw_qos_vport_create_sched_element(vport, max_rate, bw_share); + err = esw_qos_vport_create_sched_element(vport, esw->qos.node0, max_rate, bw_share, + &vport->qos.esw_sched_elem_ix); if (err) goto err_out; + INIT_LIST_HEAD(&vport->qos.parent_entry); + esw_qos_vport_set_parent(vport, esw->qos.node0); + vport->qos.enabled = true; trace_mlx5_esw_vport_qos_create(vport->dev, vport, bw_share, max_rate); -- 2.51.0 From 045815fe329ab7e1084591c5f85af32922d54c9a Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:10 +0300 Subject: [PATCH 13/16] net/mlx5: Refactor vport QoS to use scheduling node structure Refactor the vport QoS structure by moving group membership and scheduling details into the `mlx5_esw_sched_node` structure. This change consolidates the vport into the rate hierarchy by unifying the handling of different types of scheduling element nodes. In addition, add a direct reference to the mlx5_vport within the mlx5_esw_sched_node structure, to ensure that the vport is easily accessible when a scheduling node is associated with a vport. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Signed-off-by: Paolo Abeni --- .../mlx5/core/esw/diag/qos_tracepoint.h | 7 +- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 154 +++++++++++------- .../net/ethernet/mellanox/mlx5/core/esw/qos.h | 3 + .../net/ethernet/mellanox/mlx5/core/eswitch.c | 2 + .../net/ethernet/mellanox/mlx5/core/eswitch.h | 11 +- 5 files changed, 110 insertions(+), 67 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h index 0b50ef0871f2..43550a416a6f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/diag/qos_tracepoint.h @@ -9,6 +9,7 @@ #include #include "eswitch.h" +#include "qos.h" TRACE_EVENT(mlx5_esw_vport_qos_destroy, TP_PROTO(const struct mlx5_core_dev *dev, const struct mlx5_vport *vport), @@ -19,7 +20,7 @@ TRACE_EVENT(mlx5_esw_vport_qos_destroy, ), TP_fast_assign(__assign_str(devname); __entry->vport_id = vport->vport; - __entry->sched_elem_ix = vport->qos.esw_sched_elem_ix; + __entry->sched_elem_ix = mlx5_esw_qos_vport_get_sched_elem_ix(vport); ), TP_printk("(%s) vport=%hu sched_elem_ix=%u\n", __get_str(devname), __entry->vport_id, __entry->sched_elem_ix @@ -39,10 +40,10 @@ DECLARE_EVENT_CLASS(mlx5_esw_vport_qos_template, ), TP_fast_assign(__assign_str(devname); __entry->vport_id = vport->vport; - __entry->sched_elem_ix = vport->qos.esw_sched_elem_ix; + __entry->sched_elem_ix = mlx5_esw_qos_vport_get_sched_elem_ix(vport); __entry->bw_share = bw_share; __entry->max_rate = max_rate; - __entry->parent = vport->qos.parent; + __entry->parent = mlx5_esw_qos_vport_get_parent(vport); ), TP_printk("(%s) vport=%hu sched_elem_ix=%u bw_share=%u, max_rate=%u parent=%p\n", __get_str(devname), __entry->vport_id, __entry->sched_elem_ix, diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index bcdb745994d2..2d10453afc8a 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -63,6 +63,7 @@ static void esw_qos_domain_release(struct mlx5_eswitch *esw) enum sched_node_type { SCHED_NODE_TYPE_VPORTS_TSAR, + SCHED_NODE_TYPE_VPORT, }; struct mlx5_esw_sched_node { @@ -82,13 +83,34 @@ struct mlx5_esw_sched_node { struct mlx5_eswitch *esw; /* The children nodes of this node, empty list for leaf nodes. */ struct list_head children; + /* Valid only if this node is associated with a vport. */ + struct mlx5_vport *vport; }; -static void esw_qos_vport_set_parent(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent) +static void +esw_qos_node_set_parent(struct mlx5_esw_sched_node *node, struct mlx5_esw_sched_node *parent) { - list_del_init(&vport->qos.parent_entry); - vport->qos.parent = parent; - list_add_tail(&vport->qos.parent_entry, &parent->children); + list_del_init(&node->entry); + node->parent = parent; + list_add_tail(&node->entry, &parent->children); + node->esw = parent->esw; +} + +u32 mlx5_esw_qos_vport_get_sched_elem_ix(const struct mlx5_vport *vport) +{ + if (!vport->qos.sched_node) + return 0; + + return vport->qos.sched_node->ix; +} + +struct mlx5_esw_sched_node * +mlx5_esw_qos_vport_get_parent(const struct mlx5_vport *vport) +{ + if (!vport->qos.sched_node) + return NULL; + + return vport->qos.sched_node->parent; } static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_ix, @@ -131,10 +153,11 @@ static int esw_qos_vport_config(struct mlx5_vport *vport, u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) { - struct mlx5_core_dev *dev = vport->qos.parent->esw->dev; + struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; + struct mlx5_core_dev *dev = vport_node->parent->esw->dev; int err; - err = esw_qos_sched_elem_config(dev, vport->qos.esw_sched_elem_ix, max_rate, bw_share); + err = esw_qos_sched_elem_config(dev, vport_node->ix, max_rate, bw_share); if (err) { esw_warn(dev, "E-Switch modify vport scheduling element failed (vport=%d,err=%d)\n", @@ -151,15 +174,15 @@ static int esw_qos_vport_config(struct mlx5_vport *vport, static u32 esw_qos_calculate_node_min_rate_divider(struct mlx5_esw_sched_node *node) { u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); - struct mlx5_vport *vport; + struct mlx5_esw_sched_node *vport_node; u32 max_guarantee = 0; /* Find max min_rate across all vports in this node. * This will correspond to fw_max_bw_share in the final bw_share calculation. */ - list_for_each_entry(vport, &node->children, qos.parent_entry) { - if (vport->qos.min_rate > max_guarantee) - max_guarantee = vport->qos.min_rate; + list_for_each_entry(vport_node, &node->children, entry) { + if (vport_node->min_rate > max_guarantee) + max_guarantee = vport_node->min_rate; } if (max_guarantee) @@ -213,21 +236,22 @@ static int esw_qos_normalize_node_min_rate(struct mlx5_esw_sched_node *node, { u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); u32 divider = esw_qos_calculate_node_min_rate_divider(node); - struct mlx5_vport *vport; + struct mlx5_esw_sched_node *vport_node; u32 bw_share; int err; - list_for_each_entry(vport, &node->children, qos.parent_entry) { - bw_share = esw_qos_calc_bw_share(vport->qos.min_rate, divider, fw_max_bw_share); + list_for_each_entry(vport_node, &node->children, entry) { + bw_share = esw_qos_calc_bw_share(vport_node->min_rate, divider, fw_max_bw_share); - if (bw_share == vport->qos.bw_share) + if (bw_share == vport_node->bw_share) continue; - err = esw_qos_vport_config(vport, vport->qos.max_rate, bw_share, extack); + err = esw_qos_vport_config(vport_node->vport, vport_node->max_rate, bw_share, + extack); if (err) return err; - vport->qos.bw_share = bw_share; + vport_node->bw_share = bw_share; } return 0; @@ -271,25 +295,25 @@ static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw, struct netlink_e static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport, u32 min_rate, struct netlink_ext_ack *extack) { - struct mlx5_eswitch *esw = vport->dev->priv.eswitch; + struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; u32 fw_max_bw_share, previous_min_rate; bool min_rate_supported; int err; - esw_assert_qos_lock_held(esw); + esw_assert_qos_lock_held(vport_node->esw); fw_max_bw_share = MLX5_CAP_QOS(vport->dev, max_tsar_bw_share); min_rate_supported = MLX5_CAP_QOS(vport->dev, esw_bw_share) && fw_max_bw_share >= MLX5_MIN_BW_SHARE; if (min_rate && !min_rate_supported) return -EOPNOTSUPP; - if (min_rate == vport->qos.min_rate) + if (min_rate == vport_node->min_rate) return 0; - previous_min_rate = vport->qos.min_rate; - vport->qos.min_rate = min_rate; - err = esw_qos_normalize_node_min_rate(vport->qos.parent, extack); + previous_min_rate = vport_node->min_rate; + vport_node->min_rate = min_rate; + err = esw_qos_normalize_node_min_rate(vport_node->parent, extack); if (err) - vport->qos.min_rate = previous_min_rate; + vport_node->min_rate = previous_min_rate; return err; } @@ -297,27 +321,27 @@ static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport, static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport, u32 max_rate, struct netlink_ext_ack *extack) { - struct mlx5_eswitch *esw = vport->dev->priv.eswitch; + struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; u32 act_max_rate = max_rate; bool max_rate_supported; int err; - esw_assert_qos_lock_held(esw); + esw_assert_qos_lock_held(vport_node->esw); max_rate_supported = MLX5_CAP_QOS(vport->dev, esw_rate_limit); if (max_rate && !max_rate_supported) return -EOPNOTSUPP; - if (max_rate == vport->qos.max_rate) + if (max_rate == vport_node->max_rate) return 0; /* Use parent node limit if new max rate is 0. */ if (!max_rate) - act_max_rate = vport->qos.parent->max_rate; + act_max_rate = vport_node->parent->max_rate; - err = esw_qos_vport_config(vport, act_max_rate, vport->qos.bw_share, extack); + err = esw_qos_vport_config(vport, act_max_rate, vport_node->bw_share, extack); if (!err) - vport->qos.max_rate = max_rate; + vport_node->max_rate = max_rate; return err; } @@ -354,7 +378,7 @@ static int esw_qos_set_node_min_rate(struct mlx5_esw_sched_node *node, static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node, u32 max_rate, struct netlink_ext_ack *extack) { - struct mlx5_vport *vport; + struct mlx5_esw_sched_node *vport_node; int err; if (node->max_rate == max_rate) @@ -367,11 +391,12 @@ static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node, node->max_rate = max_rate; /* Any unlimited vports in the node should be set with the value of the node. */ - list_for_each_entry(vport, &node->children, qos.parent_entry) { - if (vport->qos.max_rate) + list_for_each_entry(vport_node, &node->children, entry) { + if (vport_node->max_rate) continue; - err = esw_qos_vport_config(vport, max_rate, vport->qos.bw_share, extack); + err = esw_qos_vport_config(vport_node->vport, max_rate, vport_node->bw_share, + extack); if (err) NL_SET_ERR_MSG_MOD(extack, "E-Switch vport implicit rate limit setting failed"); @@ -448,34 +473,37 @@ static int esw_qos_update_node_scheduling_element(struct mlx5_vport *vport, struct mlx5_esw_sched_node *new_node, struct netlink_ext_ack *extack) { + struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; u32 max_rate; int err; err = mlx5_destroy_scheduling_element_cmd(curr_node->esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, - vport->qos.esw_sched_elem_ix); + vport_node->ix); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy vport scheduling element failed"); return err; } /* Use new node max rate if vport max rate is unlimited. */ - max_rate = vport->qos.max_rate ? vport->qos.max_rate : new_node->max_rate; - err = esw_qos_vport_create_sched_element(vport, new_node, max_rate, vport->qos.bw_share, - &vport->qos.esw_sched_elem_ix); + max_rate = vport_node->max_rate ? vport_node->max_rate : new_node->max_rate; + err = esw_qos_vport_create_sched_element(vport, new_node, max_rate, + vport_node->bw_share, + &vport_node->ix); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch vport node set failed."); goto err_sched; } - esw_qos_vport_set_parent(vport, new_node); + esw_qos_node_set_parent(vport->qos.sched_node, new_node); return 0; err_sched: - max_rate = vport->qos.max_rate ? vport->qos.max_rate : curr_node->max_rate; - if (esw_qos_vport_create_sched_element(vport, curr_node, max_rate, vport->qos.bw_share, - &vport->qos.esw_sched_elem_ix)) + max_rate = vport_node->max_rate ? vport_node->max_rate : curr_node->max_rate; + if (esw_qos_vport_create_sched_element(vport, curr_node, max_rate, + vport_node->bw_share, + &vport_node->ix)) esw_warn(curr_node->esw->dev, "E-Switch vport node restore failed (vport=%d)\n", vport->vport); @@ -486,12 +514,13 @@ static int esw_qos_vport_update_node(struct mlx5_vport *vport, struct mlx5_esw_sched_node *node, struct netlink_ext_ack *extack) { + struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; struct mlx5_eswitch *esw = vport->dev->priv.eswitch; struct mlx5_esw_sched_node *new_node, *curr_node; int err; esw_assert_qos_lock_held(esw); - curr_node = vport->qos.parent; + curr_node = vport_node->parent; new_node = node ?: esw->qos.node0; if (curr_node == new_node) return 0; @@ -501,7 +530,7 @@ static int esw_qos_vport_update_node(struct mlx5_vport *vport, return err; /* Recalculate bw share weights of old and new nodes */ - if (vport->qos.bw_share || new_node->bw_share) { + if (vport_node->bw_share || new_node->bw_share) { esw_qos_normalize_node_min_rate(curr_node, extack); esw_qos_normalize_node_min_rate(new_node, extack); } @@ -709,6 +738,7 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) { struct mlx5_eswitch *esw = vport->dev->priv.eswitch; + u32 sched_elem_ix; int err; esw_assert_qos_lock_held(esw); @@ -720,18 +750,28 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, return err; err = esw_qos_vport_create_sched_element(vport, esw->qos.node0, max_rate, bw_share, - &vport->qos.esw_sched_elem_ix); + &sched_elem_ix); if (err) goto err_out; - INIT_LIST_HEAD(&vport->qos.parent_entry); - esw_qos_vport_set_parent(vport, esw->qos.node0); + vport->qos.sched_node = __esw_qos_alloc_node(esw, sched_elem_ix, SCHED_NODE_TYPE_VPORT, + esw->qos.node0); + if (!vport->qos.sched_node) { + err = -ENOMEM; + goto err_alloc; + } vport->qos.enabled = true; + vport->qos.sched_node->vport = vport; + trace_mlx5_esw_vport_qos_create(vport->dev, vport, bw_share, max_rate); return 0; +err_alloc: + if (mlx5_destroy_scheduling_element_cmd(esw->dev, + SCHEDULING_HIERARCHY_E_SWITCH, sched_elem_ix)) + esw_warn(esw->dev, "E-Switch destroy vport scheduling element failed.\n"); err_out: esw_qos_put(esw); @@ -741,6 +781,7 @@ err_out: void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport) { struct mlx5_eswitch *esw = vport->dev->priv.eswitch; + struct mlx5_esw_sched_node *vport_node; struct mlx5_core_dev *dev; int err; @@ -748,20 +789,23 @@ void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport) esw_qos_lock(esw); if (!vport->qos.enabled) goto unlock; - WARN(vport->qos.parent != esw->qos.node0, + vport_node = vport->qos.sched_node; + WARN(vport_node->parent != esw->qos.node0, "Disabling QoS on port before detaching it from node"); - dev = vport->qos.parent->esw->dev; + dev = vport_node->esw->dev; + trace_mlx5_esw_vport_qos_destroy(dev, vport); + err = mlx5_destroy_scheduling_element_cmd(dev, SCHEDULING_HIERARCHY_E_SWITCH, - vport->qos.esw_sched_elem_ix); + vport_node->ix); if (err) esw_warn(dev, "E-Switch destroy vport scheduling element failed (vport=%d,err=%d)\n", vport->vport, err); + __esw_qos_free_node(vport_node); memset(&vport->qos, 0, sizeof(vport->qos)); - trace_mlx5_esw_vport_qos_destroy(dev, vport); esw_qos_put(esw); unlock: @@ -794,8 +838,8 @@ bool mlx5_esw_qos_get_vport_rate(struct mlx5_vport *vport, u32 *max_rate, u32 *m esw_qos_lock(esw); enabled = vport->qos.enabled; if (enabled) { - *max_rate = vport->qos.max_rate; - *min_rate = vport->qos.min_rate; + *max_rate = vport->qos.sched_node->max_rate; + *min_rate = vport->qos.sched_node->min_rate; } esw_qos_unlock(esw); return enabled; @@ -891,16 +935,16 @@ int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 esw_qos_lock(esw); if (!vport->qos.enabled) { /* Eswitch QoS wasn't enabled yet. Enable it and vport QoS. */ - err = esw_qos_vport_enable(vport, rate_mbps, vport->qos.bw_share, NULL); + err = esw_qos_vport_enable(vport, rate_mbps, 0, NULL); } else { - struct mlx5_core_dev *dev = vport->qos.parent->esw->dev; + struct mlx5_core_dev *dev = vport->qos.sched_node->parent->esw->dev; MLX5_SET(scheduling_context, ctx, max_average_bw, rate_mbps); bitmask = MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW; err = mlx5_modify_scheduling_element_cmd(dev, SCHEDULING_HIERARCHY_E_SWITCH, ctx, - vport->qos.esw_sched_elem_ix, + vport->qos.sched_node->ix, bitmask); } esw_qos_unlock(esw); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h index b4045efbaf9e..61a6fdd5c267 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h @@ -13,6 +13,9 @@ int mlx5_esw_qos_set_vport_rate(struct mlx5_vport *evport, u32 max_rate, u32 min bool mlx5_esw_qos_get_vport_rate(struct mlx5_vport *vport, u32 *max_rate, u32 *min_rate); void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport); +u32 mlx5_esw_qos_vport_get_sched_elem_ix(const struct mlx5_vport *vport); +struct mlx5_esw_sched_node *mlx5_esw_qos_vport_get_parent(const struct mlx5_vport *vport); + int mlx5_esw_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, void *priv, u64 tx_share, struct netlink_ext_ack *extack); int mlx5_esw_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void *priv, diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c index 2bcd42305f46..09719e9b8611 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c @@ -1061,6 +1061,7 @@ static void mlx5_eswitch_clear_vf_vports_info(struct mlx5_eswitch *esw) unsigned long i; mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs) { + kfree(vport->qos.sched_node); memset(&vport->qos, 0, sizeof(vport->qos)); memset(&vport->info, 0, sizeof(vport->info)); vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO; @@ -1073,6 +1074,7 @@ static void mlx5_eswitch_clear_ec_vf_vports_info(struct mlx5_eswitch *esw) unsigned long i; mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs) { + kfree(vport->qos.sched_node); memset(&vport->qos, 0, sizeof(vport->qos)); memset(&vport->info, 0, sizeof(vport->info)); vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h index 38f912f5a707..e77ec82787de 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h @@ -216,15 +216,8 @@ struct mlx5_vport { struct { /* Initially false, set to true whenever any QoS features are used. */ bool enabled; - u32 esw_sched_elem_ix; - u32 min_rate; - u32 max_rate; - /* A computed value indicating relative min_rate between vports in a group. */ - u32 bw_share; - /* The parent group of this vport scheduling element. */ - struct mlx5_esw_sched_node *parent; - /* Membership in the parent 'members' list. */ - struct list_head parent_entry; + /* Vport scheduling element node. */ + struct mlx5_esw_sched_node *sched_node; } qos; u16 vport; -- 2.51.0 From ebecc37befb184c0a4316006df0eef83039a2475 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:11 +0300 Subject: [PATCH 14/16] net/mlx5: Remove vport QoS enabled flag Remove the `enabled` flag from the `vport->qos` struct, as QoS now relies solely on the `sched_node` pointer to determine whether QoS features are in use. Currently, the vport `qos` struct consists only of the `sched_node`, introducing an unnecessary two-level reference. However, the qos struct is retained as it will be extended in future patches to support new QoS features. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Signed-off-by: Paolo Abeni --- drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c | 13 ++++++------- drivers/net/ethernet/mellanox/mlx5/core/eswitch.h | 2 -- 2 files changed, 6 insertions(+), 9 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 2d10453afc8a..77e835fd099d 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -742,7 +742,7 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, int err; esw_assert_qos_lock_held(esw); - if (vport->qos.enabled) + if (vport->qos.sched_node) return 0; err = esw_qos_get(esw, extack); @@ -761,7 +761,6 @@ static int esw_qos_vport_enable(struct mlx5_vport *vport, goto err_alloc; } - vport->qos.enabled = true; vport->qos.sched_node->vport = vport; trace_mlx5_esw_vport_qos_create(vport->dev, vport, bw_share, max_rate); @@ -787,9 +786,9 @@ void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport) lockdep_assert_held(&esw->state_lock); esw_qos_lock(esw); - if (!vport->qos.enabled) - goto unlock; vport_node = vport->qos.sched_node; + if (!vport_node) + goto unlock; WARN(vport_node->parent != esw->qos.node0, "Disabling QoS on port before detaching it from node"); @@ -836,7 +835,7 @@ bool mlx5_esw_qos_get_vport_rate(struct mlx5_vport *vport, u32 *max_rate, u32 *m bool enabled; esw_qos_lock(esw); - enabled = vport->qos.enabled; + enabled = !!vport->qos.sched_node; if (enabled) { *max_rate = vport->qos.sched_node->max_rate; *min_rate = vport->qos.sched_node->min_rate; @@ -933,7 +932,7 @@ int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 } esw_qos_lock(esw); - if (!vport->qos.enabled) { + if (!vport->qos.sched_node) { /* Eswitch QoS wasn't enabled yet. Enable it and vport QoS. */ err = esw_qos_vport_enable(vport, rate_mbps, 0, NULL); } else { @@ -1142,7 +1141,7 @@ int mlx5_esw_qos_vport_update_node(struct mlx5_vport *vport, } esw_qos_lock(esw); - if (!vport->qos.enabled && !node) + if (!vport->qos.sched_node && !node) goto unlock; err = esw_qos_vport_enable(vport, 0, 0, extack); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h index e77ec82787de..14dd42d44e6f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h @@ -214,8 +214,6 @@ struct mlx5_vport { /* Protected with the E-Switch qos domain lock. */ struct { - /* Initially false, set to true whenever any QoS features are used. */ - bool enabled; /* Vport scheduling element node. */ struct mlx5_esw_sched_node *sched_node; } qos; -- 2.51.0 From 70744a46aabfaea925cd0ebf172e1bd21ce58148 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:12 +0300 Subject: [PATCH 15/16] net/mlx5: Simplify QoS scheduling element configuration Simplify the configuration of QoS scheduling elements by removing the separate functions `esw_qos_node_config` and `esw_qos_vport_config`. Instead, directly use the existing `esw_qos_sched_elem_config` function for both nodes and vports. This unification helps in generalizing operations on scheduling elements nodes. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 86 +++++++++---------- 1 file changed, 40 insertions(+), 46 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 77e835fd099d..7b243ba5558c 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -66,6 +66,11 @@ enum sched_node_type { SCHED_NODE_TYPE_VPORT, }; +static const char * const sched_node_type_str[] = { + [SCHED_NODE_TYPE_VPORTS_TSAR] = "vports TSAR", + [SCHED_NODE_TYPE_VPORT] = "vport", +}; + struct mlx5_esw_sched_node { u32 ix; /* Bandwidth parameters. */ @@ -113,11 +118,27 @@ mlx5_esw_qos_vport_get_parent(const struct mlx5_vport *vport) return vport->qos.sched_node->parent; } -static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_ix, - u32 max_rate, u32 bw_share) +static void esw_qos_sched_elem_config_warn(struct mlx5_esw_sched_node *node, int err) +{ + if (node->vport) { + esw_warn(node->esw->dev, + "E-Switch modify %s scheduling element failed (vport=%d,err=%d)\n", + sched_node_type_str[node->type], node->vport->vport, err); + return; + } + + esw_warn(node->esw->dev, + "E-Switch modify %s scheduling element failed (err=%d)\n", + sched_node_type_str[node->type], err); +} + +static int esw_qos_sched_elem_config(struct mlx5_esw_sched_node *node, u32 max_rate, u32 bw_share, + struct netlink_ext_ack *extack) { u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {}; + struct mlx5_core_dev *dev = node->esw->dev; u32 bitmask = 0; + int err; if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling)) return -EOPNOTSUPP; @@ -127,46 +148,22 @@ static int esw_qos_sched_elem_config(struct mlx5_core_dev *dev, u32 sched_elem_i bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW; bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE; - return mlx5_modify_scheduling_element_cmd(dev, - SCHEDULING_HIERARCHY_E_SWITCH, - sched_ctx, - sched_elem_ix, - bitmask); -} - -static int esw_qos_node_config(struct mlx5_esw_sched_node *node, - u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack) -{ - struct mlx5_core_dev *dev = node->esw->dev; - int err; - - err = esw_qos_sched_elem_config(dev, node->ix, max_rate, bw_share); - if (err) - NL_SET_ERR_MSG_MOD(extack, "E-Switch modify node TSAR element failed"); - - trace_mlx5_esw_node_qos_config(dev, node, node->ix, bw_share, max_rate); - - return err; -} - -static int esw_qos_vport_config(struct mlx5_vport *vport, - u32 max_rate, u32 bw_share, - struct netlink_ext_ack *extack) -{ - struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node; - struct mlx5_core_dev *dev = vport_node->parent->esw->dev; - int err; - - err = esw_qos_sched_elem_config(dev, vport_node->ix, max_rate, bw_share); + err = mlx5_modify_scheduling_element_cmd(dev, + SCHEDULING_HIERARCHY_E_SWITCH, + sched_ctx, + node->ix, + bitmask); if (err) { - esw_warn(dev, - "E-Switch modify vport scheduling element failed (vport=%d,err=%d)\n", - vport->vport, err); - NL_SET_ERR_MSG_MOD(extack, "E-Switch modify vport scheduling element failed"); + esw_qos_sched_elem_config_warn(node, err); + NL_SET_ERR_MSG_MOD(extack, "E-Switch modify scheduling element failed"); + return err; } - trace_mlx5_esw_vport_qos_config(dev, vport, bw_share, max_rate); + if (node->type == SCHED_NODE_TYPE_VPORTS_TSAR) + trace_mlx5_esw_node_qos_config(dev, node, node->ix, bw_share, max_rate); + else if (node->type == SCHED_NODE_TYPE_VPORT) + trace_mlx5_esw_vport_qos_config(dev, node->vport, bw_share, max_rate); return 0; } @@ -246,8 +243,7 @@ static int esw_qos_normalize_node_min_rate(struct mlx5_esw_sched_node *node, if (bw_share == vport_node->bw_share) continue; - err = esw_qos_vport_config(vport_node->vport, vport_node->max_rate, bw_share, - extack); + err = esw_qos_sched_elem_config(vport_node, vport_node->max_rate, bw_share, extack); if (err) return err; @@ -274,7 +270,7 @@ static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw, struct netlink_e if (bw_share == node->bw_share) continue; - err = esw_qos_node_config(node, node->max_rate, bw_share, extack); + err = esw_qos_sched_elem_config(node, node->max_rate, bw_share, extack); if (err) return err; @@ -338,8 +334,7 @@ static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport, if (!max_rate) act_max_rate = vport_node->parent->max_rate; - err = esw_qos_vport_config(vport, act_max_rate, vport_node->bw_share, extack); - + err = esw_qos_sched_elem_config(vport_node, act_max_rate, vport_node->bw_share, extack); if (!err) vport_node->max_rate = max_rate; @@ -384,7 +379,7 @@ static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node, if (node->max_rate == max_rate) return 0; - err = esw_qos_node_config(node, max_rate, node->bw_share, extack); + err = esw_qos_sched_elem_config(node, max_rate, node->bw_share, extack); if (err) return err; @@ -395,8 +390,7 @@ static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node, if (vport_node->max_rate) continue; - err = esw_qos_vport_config(vport_node->vport, max_rate, vport_node->bw_share, - extack); + err = esw_qos_sched_elem_config(vport_node, max_rate, vport_node->bw_share, extack); if (err) NL_SET_ERR_MSG_MOD(extack, "E-Switch vport implicit rate limit setting failed"); -- 2.51.0 From a1903bf50f2e67f8f5c67743d57c76ab19e795a5 Mon Sep 17 00:00:00 2001 From: Carolina Jubran Date: Wed, 16 Oct 2024 20:36:13 +0300 Subject: [PATCH 16/16] net/mlx5: Generalize QoS operations for nodes and vports Refactor QoS normalization and rate calculation functions to operate on mlx5_esw_sched_node, allowing for generalized handling of both vports and nodes. Signed-off-by: Carolina Jubran Reviewed-by: Cosmin Ratiu Signed-off-by: Tariq Toukan Signed-off-by: Paolo Abeni --- .../net/ethernet/mellanox/mlx5/core/esw/qos.c | 115 +++++++----------- 1 file changed, 43 insertions(+), 72 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c index 7b243ba5558c..7e7f99b38a37 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c @@ -168,45 +168,18 @@ static int esw_qos_sched_elem_config(struct mlx5_esw_sched_node *node, u32 max_r return 0; } -static u32 esw_qos_calculate_node_min_rate_divider(struct mlx5_esw_sched_node *node) -{ - u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); - struct mlx5_esw_sched_node *vport_node; - u32 max_guarantee = 0; - - /* Find max min_rate across all vports in this node. - * This will correspond to fw_max_bw_share in the final bw_share calculation. - */ - list_for_each_entry(vport_node, &node->children, entry) { - if (vport_node->min_rate > max_guarantee) - max_guarantee = vport_node->min_rate; - } - - if (max_guarantee) - return max_t(u32, max_guarantee / fw_max_bw_share, 1); - - /* If vports max min_rate divider is 0 but their node has bw_share - * configured, then set bw_share for vports to minimal value. - */ - if (node->bw_share) - return 1; - - /* A divider of 0 sets bw_share for all node vports to 0, - * effectively disabling min guarantees. - */ - return 0; -} - -static u32 esw_qos_calculate_min_rate_divider(struct mlx5_eswitch *esw) +static u32 esw_qos_calculate_min_rate_divider(struct mlx5_eswitch *esw, + struct mlx5_esw_sched_node *parent) { + struct list_head *nodes = parent ? &parent->children : &esw->qos.domain->nodes; u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share); struct mlx5_esw_sched_node *node; u32 max_guarantee = 0; - /* Find max min_rate across all esw nodes. + /* Find max min_rate across all nodes. * This will correspond to fw_max_bw_share in the final bw_share calculation. */ - list_for_each_entry(node, &esw->qos.domain->nodes, entry) { + list_for_each_entry(node, nodes, entry) { if (node->esw == esw && node->ix != esw->qos.root_tsar_ix && node->min_rate > max_guarantee) max_guarantee = node->min_rate; @@ -215,7 +188,14 @@ static u32 esw_qos_calculate_min_rate_divider(struct mlx5_eswitch *esw) if (max_guarantee) return max_t(u32, max_guarantee / fw_max_bw_share, 1); - /* If no node has min_rate configured, a divider of 0 sets all + /* If nodes max min_rate divider is 0 but their parent has bw_share + * configured, then set bw_share for nodes to minimal value. + */ + + if (parent && parent->bw_share) + return 1; + + /* If the node nodes has min_rate configured, a divider of 0 sets all * nodes' bw_share to 0, effectively disabling min guarantees. */ return 0; @@ -228,59 +208,50 @@ static u32 esw_qos_calc_bw_share(u32 min_rate, u32 divider, u32 fw_max) return min_t(u32, max_t(u32, DIV_ROUND_UP(min_rate, divider), MLX5_MIN_BW_SHARE), fw_max); } -static int esw_qos_normalize_node_min_rate(struct mlx5_esw_sched_node *node, - struct netlink_ext_ack *extack) +static int esw_qos_update_sched_node_bw_share(struct mlx5_esw_sched_node *node, + u32 divider, + struct netlink_ext_ack *extack) { u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share); - u32 divider = esw_qos_calculate_node_min_rate_divider(node); - struct mlx5_esw_sched_node *vport_node; u32 bw_share; int err; - list_for_each_entry(vport_node, &node->children, entry) { - bw_share = esw_qos_calc_bw_share(vport_node->min_rate, divider, fw_max_bw_share); + bw_share = esw_qos_calc_bw_share(node->min_rate, divider, fw_max_bw_share); - if (bw_share == vport_node->bw_share) - continue; + if (bw_share == node->bw_share) + return 0; - err = esw_qos_sched_elem_config(vport_node, vport_node->max_rate, bw_share, extack); - if (err) - return err; + err = esw_qos_sched_elem_config(node, node->max_rate, bw_share, extack); + if (err) + return err; - vport_node->bw_share = bw_share; - } + node->bw_share = bw_share; - return 0; + return err; } -static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack) +static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw, + struct mlx5_esw_sched_node *parent, + struct netlink_ext_ack *extack) { - u32 fw_max_bw_share = MLX5_CAP_QOS(esw->dev, max_tsar_bw_share); - u32 divider = esw_qos_calculate_min_rate_divider(esw); + struct list_head *nodes = parent ? &parent->children : &esw->qos.domain->nodes; + u32 divider = esw_qos_calculate_min_rate_divider(esw, parent); struct mlx5_esw_sched_node *node; - u32 bw_share; - int err; - list_for_each_entry(node, &esw->qos.domain->nodes, entry) { - if (node->esw != esw || node->ix == esw->qos.root_tsar_ix) - continue; - bw_share = esw_qos_calc_bw_share(node->min_rate, divider, - fw_max_bw_share); + list_for_each_entry(node, nodes, entry) { + int err; - if (bw_share == node->bw_share) + if (node->esw != esw || node->ix == esw->qos.root_tsar_ix) continue; - err = esw_qos_sched_elem_config(node, node->max_rate, bw_share, extack); + err = esw_qos_update_sched_node_bw_share(node, divider, extack); if (err) return err; - node->bw_share = bw_share; - - /* All the node's vports need to be set with default bw_share - * to enable them with QOS - */ - err = esw_qos_normalize_node_min_rate(node, extack); + if (list_empty(&node->children)) + continue; + err = esw_qos_normalize_min_rate(node->esw, node, extack); if (err) return err; } @@ -307,7 +278,7 @@ static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport, previous_min_rate = vport_node->min_rate; vport_node->min_rate = min_rate; - err = esw_qos_normalize_node_min_rate(vport_node->parent, extack); + err = esw_qos_normalize_min_rate(vport_node->parent->esw, vport_node->parent, extack); if (err) vport_node->min_rate = previous_min_rate; @@ -357,13 +328,13 @@ static int esw_qos_set_node_min_rate(struct mlx5_esw_sched_node *node, previous_min_rate = node->min_rate; node->min_rate = min_rate; - err = esw_qos_normalize_min_rate(esw, extack); + err = esw_qos_normalize_min_rate(esw, NULL, extack); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch node min rate setting failed"); /* Attempt restoring previous configuration */ node->min_rate = previous_min_rate; - if (esw_qos_normalize_min_rate(esw, extack)) + if (esw_qos_normalize_min_rate(esw, NULL, extack)) NL_SET_ERR_MSG_MOD(extack, "E-Switch BW share restore failed"); } @@ -525,8 +496,8 @@ static int esw_qos_vport_update_node(struct mlx5_vport *vport, /* Recalculate bw share weights of old and new nodes */ if (vport_node->bw_share || new_node->bw_share) { - esw_qos_normalize_node_min_rate(curr_node, extack); - esw_qos_normalize_node_min_rate(new_node, extack); + esw_qos_normalize_min_rate(curr_node->esw, curr_node, extack); + esw_qos_normalize_min_rate(new_node->esw, new_node, extack); } return 0; @@ -581,7 +552,7 @@ __esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct mlx5_esw_sch goto err_alloc_node; } - err = esw_qos_normalize_min_rate(esw, extack); + err = esw_qos_normalize_min_rate(esw, NULL, extack); if (err) { NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed"); goto err_min_rate; @@ -638,7 +609,7 @@ static int __esw_qos_destroy_node(struct mlx5_esw_sched_node *node, struct netli NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR_ID failed"); __esw_qos_free_node(node); - err = esw_qos_normalize_min_rate(esw, extack); + err = esw_qos_normalize_min_rate(esw, NULL, extack); if (err) NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed"); -- 2.51.0