From patchwork Fri Feb 3 09:11:26 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127226 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 61839C636CC for ; Fri, 3 Feb 2023 09:15:51 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232570AbjBCJPt (ORCPT ); Fri, 3 Feb 2023 04:15:49 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36174 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232445AbjBCJPr (ORCPT ); Fri, 3 Feb 2023 04:15:47 -0500 Received: from smtpbgbr2.qq.com (smtpbgbr2.qq.com [54.207.22.56]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 0DFE88E48F for ; Fri, 3 Feb 2023 01:15:42 -0800 (PST) X-QQ-mid: bizesmtp68t1675415734tfemy0dx Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:34 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: l3NhZMCtXe0eW2mBt2WCW2+r/L/+HUdQwSUnoVKq5kQpSU8wq87Q+w7bPwT18 spgie5bsQw8lUioOfVBz+fGJ1CNG/sjCNpmj1Me9/6oJYbLhBZd2NmPa122giSG7Ilu1snp vF551EWC3NGFkU65A7Db8h2MmcLtpIuGYGS2AYxyXdy8uZErTzbthqH4uV7pzrZy1VprzJm GLeMzcBfYHZtthc2cYPA+741IDG1lLRqbBIk7wwz+8aKGejLrttnroeaNEGqzL68nSxXEvI 7es/O0wVQp7M7NQcUqntbEFkjivFsAN5yYSFhNzQq0Vl6KUDx743+B7Ts99kuF+ZPTHJf7u G6uLcEnE1smi8KBdvF5UDBI8OfC6GAUYoqvAKTJySFkbxiYr+tmRCJBGxJQz8CgU4OoiHel zlP/MyXmOUaCxTfyEhdrcg== X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 01/10] net: libwx: Add irq flow functions Date: Fri, 3 Feb 2023 17:11:26 +0800 Message-Id: <20230203091135.3294377-2-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org From: Mengyuan Lou Add irq flow functions for ngbe and txgbe. Alloc pcie msix irqs for drivers, otherwise fall back to msi/legacy. Signed-off-by: Mengyuan Lou Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/libwx/Makefile | 2 +- drivers/net/ethernet/wangxun/libwx/wx_hw.c | 55 +- drivers/net/ethernet/wangxun/libwx/wx_hw.h | 3 + drivers/net/ethernet/wangxun/libwx/wx_lib.c | 609 +++++++++++++++++++ drivers/net/ethernet/wangxun/libwx/wx_lib.h | 20 + drivers/net/ethernet/wangxun/libwx/wx_type.h | 101 +++ 6 files changed, 788 insertions(+), 2 deletions(-) create mode 100644 drivers/net/ethernet/wangxun/libwx/wx_lib.c create mode 100644 drivers/net/ethernet/wangxun/libwx/wx_lib.h diff --git a/drivers/net/ethernet/wangxun/libwx/Makefile b/drivers/net/ethernet/wangxun/libwx/Makefile index 1ed5e23af944..850d1615cd18 100644 --- a/drivers/net/ethernet/wangxun/libwx/Makefile +++ b/drivers/net/ethernet/wangxun/libwx/Makefile @@ -4,4 +4,4 @@ obj-$(CONFIG_LIBWX) += libwx.o -libwx-objs := wx_hw.o +libwx-objs := wx_hw.o wx_lib.o diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c index 3d7ba0c0df38..b39fdef5a5b0 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c @@ -14,7 +14,7 @@ static void wx_intr_disable(struct wx *wx, u64 qmask) { u32 mask; - mask = (qmask & 0xFFFFFFFF); + mask = (qmask & U32_MAX); if (mask) wr32(wx, WX_PX_IMS(0), mask); @@ -25,6 +25,45 @@ static void wx_intr_disable(struct wx *wx, u64 qmask) } } +void wx_intr_enable(struct wx *wx, u64 qmask) +{ + u32 mask; + + mask = (qmask & U32_MAX); + if (mask) + wr32(wx, WX_PX_IMC(0), mask); + if (wx->mac.type == wx_mac_sp) { + mask = (qmask >> 32); + if (mask) + wr32(wx, WX_PX_IMC(1), mask); + } +} +EXPORT_SYMBOL(wx_intr_enable); + +/** + * wx_irq_disable - Mask off interrupt generation on the NIC + * @wx: board private structure + **/ +void wx_irq_disable(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + + wr32(wx, WX_PX_MISC_IEN, 0); + wx_intr_disable(wx, WX_INTR_ALL); + + if (pdev->msix_enabled) { + int vector; + + for (vector = 0; vector < wx->num_q_vectors; vector++) + synchronize_irq(wx->msix_entries[vector].vector); + + synchronize_irq(wx->msix_entries[vector].vector); + } else { + synchronize_irq(pdev->irq); + } +} +EXPORT_SYMBOL(wx_irq_disable); + /* cmd_addr is used for some special command: * 1. to be sector address, when implemented erase sector command * 2. to be flash address when implemented read, write flash address @@ -844,6 +883,20 @@ void wx_disable_rx(struct wx *wx) } EXPORT_SYMBOL(wx_disable_rx); +static void wx_configure_isb(struct wx *wx) +{ + /* set ISB Address */ + wr32(wx, WX_PX_ISB_ADDR_L, wx->isb_dma & DMA_BIT_MASK(32)); + if (IS_ENABLED(CONFIG_ARCH_DMA_ADDR_T_64BIT)) + wr32(wx, WX_PX_ISB_ADDR_H, upper_32_bits(wx->isb_dma)); +} + +void wx_configure(struct wx *wx) +{ + wx_configure_isb(wx); +} +EXPORT_SYMBOL(wx_configure); + /** * wx_disable_pcie_master - Disable PCI-express master access * @wx: pointer to hardware structure diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.h b/drivers/net/ethernet/wangxun/libwx/wx_hw.h index 803983546f3a..60bda129cfa6 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.h @@ -4,6 +4,8 @@ #ifndef _WX_HW_H_ #define _WX_HW_H_ +void wx_intr_enable(struct wx *wx, u64 qmask); +void wx_irq_disable(struct wx *wx); int wx_check_flash_load(struct wx *wx, u32 check_bit); void wx_control_hw(struct wx *wx, bool drv); int wx_mng_present(struct wx *wx); @@ -20,6 +22,7 @@ void wx_mac_set_default_filter(struct wx *wx, u8 *addr); void wx_flush_sw_mac_table(struct wx *wx); int wx_set_mac(struct net_device *netdev, void *p); void wx_disable_rx(struct wx *wx); +void wx_configure(struct wx *wx); int wx_disable_pcie_master(struct wx *wx); int wx_stop_adapter(struct wx *wx); void wx_reset_misc(struct wx *wx); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c new file mode 100644 index 000000000000..8f8a9e3151c2 --- /dev/null +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c @@ -0,0 +1,609 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2019 - 2022 Beijing WangXun Technology Co., Ltd. */ + +#include +#include +#include + +#include "wx_type.h" +#include "wx_lib.h" + +/** + * wx_poll - NAPI polling RX/TX cleanup routine + * @napi: napi struct with our devices info in it + * @budget: amount of work driver is allowed to do this pass, in packets + * + * This function will clean all queues associated with a q_vector. + **/ +static int wx_poll(struct napi_struct *napi, int budget) +{ + return 0; +} + +/** + * wx_set_rss_queues: Allocate queues for RSS + * @wx: board private structure to initialize + * + * This is our "base" multiqueue mode. RSS (Receive Side Scaling) will try + * to allocate one Rx queue per CPU, and if available, one Tx queue per CPU. + * + **/ +static void wx_set_rss_queues(struct wx *wx) +{ + wx->num_rx_queues = wx->mac.max_rx_queues; + wx->num_tx_queues = wx->mac.max_tx_queues; +} + +static void wx_set_num_queues(struct wx *wx) +{ + /* Start with base case */ + wx->num_rx_queues = 1; + wx->num_tx_queues = 1; + wx->queues_per_pool = 1; + + wx_set_rss_queues(wx); +} + +/** + * wx_acquire_msix_vectors - acquire MSI-X vectors + * @wx: board private structure + * + * Attempts to acquire a suitable range of MSI-X vector interrupts. Will + * return a negative error code if unable to acquire MSI-X vectors for any + * reason. + */ +static int wx_acquire_msix_vectors(struct wx *wx) +{ + struct irq_affinity affd = {0, }; + int nvecs, i; + + nvecs = min_t(int, num_online_cpus(), wx->mac.max_msix_vectors); + + wx->msix_entries = kcalloc(nvecs, + sizeof(struct msix_entry), + GFP_KERNEL); + if (!wx->msix_entries) + return -ENOMEM; + + nvecs = pci_alloc_irq_vectors_affinity(wx->pdev, nvecs, + nvecs, + PCI_IRQ_MSIX | PCI_IRQ_AFFINITY, + &affd); + if (nvecs < 0) { + wx_err(wx, "Failed to allocate MSI-X interrupts. Err: %d\n", nvecs); + kfree(wx->msix_entries); + wx->msix_entries = NULL; + return nvecs; + } + + for (i = 0; i < nvecs; i++) { + wx->msix_entries[i].entry = i; + wx->msix_entries[i].vector = pci_irq_vector(wx->pdev, i); + } + + /* one for msix_other */ + nvecs -= 1; + wx->num_q_vectors = nvecs; + wx->num_rx_queues = nvecs; + wx->num_tx_queues = nvecs; + + return 0; +} + +/** + * wx_set_interrupt_capability - set MSI-X or MSI if supported + * @wx: board private structure to initialize + * + * Attempt to configure the interrupts using the best available + * capabilities of the hardware and the kernel. + **/ +static int wx_set_interrupt_capability(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + int nvecs, ret; + + /* We will try to get MSI-X interrupts first */ + ret = wx_acquire_msix_vectors(wx); + if (ret == 0 || (ret == -ENOMEM)) + return ret; + + wx->num_rx_queues = 1; + wx->num_tx_queues = 1; + wx->num_q_vectors = 1; + + /* minmum one for queue, one for misc*/ + nvecs = 1; + nvecs = pci_alloc_irq_vectors(pdev, nvecs, + nvecs, PCI_IRQ_MSI | PCI_IRQ_LEGACY); + if (nvecs == 1) { + if (pdev->msi_enabled) + wx_err(wx, "Fallback to MSI.\n"); + else + wx_err(wx, "Fallback to LEGACY.\n"); + } else { + wx_err(wx, "Failed to allocate MSI/LEGACY interrupts. Error: %d\n", nvecs); + return nvecs; + } + + pdev->irq = pci_irq_vector(pdev, 0); + + return 0; +} + +/** + * wx_cache_ring_rss - Descriptor ring to register mapping for RSS + * @wx: board private structure to initialize + * + * Cache the descriptor ring offsets for RSS, ATR, FCoE, and SR-IOV. + * + **/ +static void wx_cache_ring_rss(struct wx *wx) +{ + u16 i; + + for (i = 0; i < wx->num_rx_queues; i++) + wx->rx_ring[i]->reg_idx = i; + + for (i = 0; i < wx->num_tx_queues; i++) + wx->tx_ring[i]->reg_idx = i; +} + +static void wx_add_ring(struct wx_ring *ring, struct wx_ring_container *head) +{ + ring->next = head->ring; + head->ring = ring; + head->count++; +} + +/** + * wx_alloc_q_vector - Allocate memory for a single interrupt vector + * @wx: board private structure to initialize + * @v_count: q_vectors allocated on wx, used for ring interleaving + * @v_idx: index of vector in wx struct + * @txr_count: total number of Tx rings to allocate + * @txr_idx: index of first Tx ring to allocate + * @rxr_count: total number of Rx rings to allocate + * @rxr_idx: index of first Rx ring to allocate + * + * We allocate one q_vector. If allocation fails we return -ENOMEM. + **/ +static int wx_alloc_q_vector(struct wx *wx, + unsigned int v_count, unsigned int v_idx, + unsigned int txr_count, unsigned int txr_idx, + unsigned int rxr_count, unsigned int rxr_idx) +{ + struct wx_q_vector *q_vector; + int ring_count, default_itr; + struct wx_ring *ring; + + /* note this will allocate space for the ring structure as well! */ + ring_count = txr_count + rxr_count; + + q_vector = kzalloc(struct_size(q_vector, ring, ring_count), + GFP_KERNEL); + if (!q_vector) + return -ENOMEM; + + /* initialize NAPI */ + netif_napi_add(wx->netdev, &q_vector->napi, + wx_poll); + + /* tie q_vector and wx together */ + wx->q_vector[v_idx] = q_vector; + q_vector->wx = wx; + q_vector->v_idx = v_idx; + + /* initialize pointer to rings */ + ring = q_vector->ring; + + if (wx->mac.type == wx_mac_sp) + default_itr = WX_12K_ITR; + else + default_itr = WX_7K_ITR; + /* initialize ITR */ + if (txr_count && !rxr_count) + /* tx only vector */ + q_vector->itr = wx->tx_itr_setting ? + default_itr : wx->tx_itr_setting; + else + /* rx or rx/tx vector */ + q_vector->itr = wx->rx_itr_setting ? + default_itr : wx->rx_itr_setting; + + while (txr_count) { + /* assign generic ring traits */ + ring->dev = &wx->pdev->dev; + ring->netdev = wx->netdev; + + /* configure backlink on ring */ + ring->q_vector = q_vector; + + /* update q_vector Tx values */ + wx_add_ring(ring, &q_vector->tx); + + /* apply Tx specific ring traits */ + ring->count = wx->tx_ring_count; + + ring->queue_index = txr_idx; + + /* assign ring to wx */ + wx->tx_ring[txr_idx] = ring; + + /* update count and index */ + txr_count--; + txr_idx += v_count; + + /* push pointer to next ring */ + ring++; + } + + while (rxr_count) { + /* assign generic ring traits */ + ring->dev = &wx->pdev->dev; + ring->netdev = wx->netdev; + + /* configure backlink on ring */ + ring->q_vector = q_vector; + + /* update q_vector Rx values */ + wx_add_ring(ring, &q_vector->rx); + + /* apply Rx specific ring traits */ + ring->count = wx->rx_ring_count; + ring->queue_index = rxr_idx; + + /* assign ring to wx */ + wx->rx_ring[rxr_idx] = ring; + + /* update count and index */ + rxr_count--; + rxr_idx += v_count; + + /* push pointer to next ring */ + ring++; + } + + return 0; +} + +/** + * wx_free_q_vector - Free memory allocated for specific interrupt vector + * @wx: board private structure to initialize + * @v_idx: Index of vector to be freed + * + * This function frees the memory allocated to the q_vector. In addition if + * NAPI is enabled it will delete any references to the NAPI struct prior + * to freeing the q_vector. + **/ +static void wx_free_q_vector(struct wx *wx, int v_idx) +{ + struct wx_q_vector *q_vector = wx->q_vector[v_idx]; + struct wx_ring *ring; + + wx_for_each_ring(ring, q_vector->tx) + wx->tx_ring[ring->queue_index] = NULL; + + wx_for_each_ring(ring, q_vector->rx) + wx->rx_ring[ring->queue_index] = NULL; + + wx->q_vector[v_idx] = NULL; + netif_napi_del(&q_vector->napi); + kfree_rcu(q_vector, rcu); +} + +/** + * wx_alloc_q_vectors - Allocate memory for interrupt vectors + * @wx: board private structure to initialize + * + * We allocate one q_vector per queue interrupt. If allocation fails we + * return -ENOMEM. + **/ +static int wx_alloc_q_vectors(struct wx *wx) +{ + unsigned int rxr_idx = 0, txr_idx = 0, v_idx = 0; + unsigned int rxr_remaining = wx->num_rx_queues; + unsigned int txr_remaining = wx->num_tx_queues; + unsigned int q_vectors = wx->num_q_vectors; + int rqpv, tqpv; + int err; + + for (; v_idx < q_vectors; v_idx++) { + rqpv = DIV_ROUND_UP(rxr_remaining, q_vectors - v_idx); + tqpv = DIV_ROUND_UP(txr_remaining, q_vectors - v_idx); + err = wx_alloc_q_vector(wx, q_vectors, v_idx, + tqpv, txr_idx, + rqpv, rxr_idx); + + if (err) + goto err_out; + + /* update counts and index */ + rxr_remaining -= rqpv; + txr_remaining -= tqpv; + rxr_idx++; + txr_idx++; + } + + return 0; + +err_out: + wx->num_tx_queues = 0; + wx->num_rx_queues = 0; + wx->num_q_vectors = 0; + + while (v_idx--) + wx_free_q_vector(wx, v_idx); + + return -ENOMEM; +} + +/** + * wx_free_q_vectors - Free memory allocated for interrupt vectors + * @wx: board private structure to initialize + * + * This function frees the memory allocated to the q_vectors. In addition if + * NAPI is enabled it will delete any references to the NAPI struct prior + * to freeing the q_vector. + **/ +static void wx_free_q_vectors(struct wx *wx) +{ + int v_idx = wx->num_q_vectors; + + wx->num_tx_queues = 0; + wx->num_rx_queues = 0; + wx->num_q_vectors = 0; + + while (v_idx--) + wx_free_q_vector(wx, v_idx); +} + +void wx_reset_interrupt_capability(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + + if (!pdev->msi_enabled && !pdev->msix_enabled) + return; + + pci_free_irq_vectors(wx->pdev); + if (pdev->msix_enabled) { + kfree(wx->msix_entries); + wx->msix_entries = NULL; + } +} +EXPORT_SYMBOL(wx_reset_interrupt_capability); + +/** + * wx_clear_interrupt_scheme - Clear the current interrupt scheme settings + * @wx: board private structure to clear interrupt scheme on + * + * We go through and clear interrupt specific resources and reset the structure + * to pre-load conditions + **/ +void wx_clear_interrupt_scheme(struct wx *wx) +{ + wx_free_q_vectors(wx); + wx_reset_interrupt_capability(wx); +} +EXPORT_SYMBOL(wx_clear_interrupt_scheme); + +int wx_init_interrupt_scheme(struct wx *wx) +{ + int ret; + + /* Number of supported queues */ + wx_set_num_queues(wx); + + /* Set interrupt mode */ + ret = wx_set_interrupt_capability(wx); + if (ret) { + wx_err(wx, "Allocate irq vectors for failed.\n"); + return ret; + } + + /* Allocate memory for queues */ + ret = wx_alloc_q_vectors(wx); + if (ret) { + wx_err(wx, "Unable to allocate memory for queue vectors.\n"); + wx_reset_interrupt_capability(wx); + return ret; + } + + wx_cache_ring_rss(wx); + + return 0; +} +EXPORT_SYMBOL(wx_init_interrupt_scheme); + +irqreturn_t wx_msix_clean_rings(int __always_unused irq, void *data) +{ + struct wx_q_vector *q_vector = data; + + /* EIAM disabled interrupts (on this vector) for us */ + if (q_vector->rx.ring || q_vector->tx.ring) + napi_schedule_irqoff(&q_vector->napi); + + return IRQ_HANDLED; +} +EXPORT_SYMBOL(wx_msix_clean_rings); + +void wx_free_irq(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + int vector; + + if (!(pdev->msix_enabled)) { + free_irq(pdev->irq, wx); + return; + } + + for (vector = 0; vector < wx->num_q_vectors; vector++) { + struct wx_q_vector *q_vector = wx->q_vector[vector]; + struct msix_entry *entry = &wx->msix_entries[vector]; + + /* free only the irqs that were actually requested */ + if (!q_vector->rx.ring && !q_vector->tx.ring) + continue; + + free_irq(entry->vector, q_vector); + } + + free_irq(wx->msix_entries[vector].vector, wx); +} +EXPORT_SYMBOL(wx_free_irq); + +/** + * wx_setup_isb_resources - allocate interrupt status resources + * @wx: board private structure + * + * Return 0 on success, negative on failure + **/ +int wx_setup_isb_resources(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + + wx->isb_mem = dma_alloc_coherent(&pdev->dev, + sizeof(u32) * 4, + &wx->isb_dma, + GFP_KERNEL); + if (!wx->isb_mem) { + wx_err(wx, "Alloc isb_mem failed\n"); + return -ENOMEM; + } + + return 0; +} +EXPORT_SYMBOL(wx_setup_isb_resources); + +/** + * wx_free_isb_resources - allocate all queues Rx resources + * @wx: board private structure + * + * Return 0 on success, negative on failure + **/ +void wx_free_isb_resources(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + + dma_free_coherent(&pdev->dev, sizeof(u32) * 4, + wx->isb_mem, wx->isb_dma); + wx->isb_mem = NULL; +} +EXPORT_SYMBOL(wx_free_isb_resources); + +u32 wx_misc_isb(struct wx *wx, enum wx_isb_idx idx) +{ + u32 cur_tag = 0; + + cur_tag = wx->isb_mem[WX_ISB_HEADER]; + wx->isb_tag[idx] = cur_tag; + + return (__force u32)cpu_to_le32(wx->isb_mem[idx]); +} +EXPORT_SYMBOL(wx_misc_isb); + +/** + * wx_set_ivar - set the IVAR registers, mapping interrupt causes to vectors + * @wx: pointer to wx struct + * @direction: 0 for Rx, 1 for Tx, -1 for other causes + * @queue: queue to map the corresponding interrupt to + * @msix_vector: the vector to map to the corresponding queue + * + **/ +static void wx_set_ivar(struct wx *wx, s8 direction, + u16 queue, u16 msix_vector) +{ + u32 ivar, index; + + if (direction == -1) { + /* other causes */ + msix_vector |= WX_PX_IVAR_ALLOC_VAL; + index = 0; + ivar = rd32(wx, WX_PX_MISC_IVAR); + ivar &= ~(0xFF << index); + ivar |= (msix_vector << index); + wr32(wx, WX_PX_MISC_IVAR, ivar); + } else { + /* tx or rx causes */ + msix_vector |= WX_PX_IVAR_ALLOC_VAL; + index = ((16 * (queue & 1)) + (8 * direction)); + ivar = rd32(wx, WX_PX_IVAR(queue >> 1)); + ivar &= ~(0xFF << index); + ivar |= (msix_vector << index); + wr32(wx, WX_PX_IVAR(queue >> 1), ivar); + } +} + +/** + * wx_write_eitr - write EITR register in hardware specific way + * @q_vector: structure containing interrupt and ring information + * + * This function is made to be called by ethtool and by the driver + * when it needs to update EITR registers at runtime. Hardware + * specific quirks/differences are taken care of here. + */ +static void wx_write_eitr(struct wx_q_vector *q_vector) +{ + struct wx *wx = q_vector->wx; + int v_idx = q_vector->v_idx; + u32 itr_reg; + + if (wx->mac.type == wx_mac_sp) + itr_reg = q_vector->itr & WX_SP_MAX_EITR; + else + itr_reg = q_vector->itr & WX_EM_MAX_EITR; + + itr_reg |= WX_PX_ITR_CNT_WDIS; + + wr32(wx, WX_PX_ITR(v_idx), itr_reg); +} + +/** + * wx_configure_vectors - Configure vectors for hardware + * @wx: board private structure + * + * wx_configure_vectors sets up the hardware to properly generate MSI-X/MSI/LEGACY + * interrupts. + **/ +void wx_configure_vectors(struct wx *wx) +{ + struct pci_dev *pdev = wx->pdev; + u32 eitrsel = 0; + u16 v_idx; + + if (pdev->msix_enabled) { + /* Populate MSIX to EITR Select */ + wr32(wx, WX_PX_ITRSEL, eitrsel); + /* use EIAM to auto-mask when MSI-X interrupt is asserted + * this saves a register write for every interrupt + */ + wr32(wx, WX_PX_GPIE, WX_PX_GPIE_MODEL); + } else { + /* legacy interrupts, use EIAM to auto-mask when reading EICR, + * specifically only auto mask tx and rx interrupts. + */ + wr32(wx, WX_PX_GPIE, 0); + } + + /* Populate the IVAR table and set the ITR values to the + * corresponding register. + */ + for (v_idx = 0; v_idx < wx->num_q_vectors; v_idx++) { + struct wx_q_vector *q_vector = wx->q_vector[v_idx]; + struct wx_ring *ring; + + wx_for_each_ring(ring, q_vector->rx) + wx_set_ivar(wx, 0, ring->reg_idx, v_idx); + + wx_for_each_ring(ring, q_vector->tx) + wx_set_ivar(wx, 1, ring->reg_idx, v_idx); + + wx_write_eitr(q_vector); + } + + wx_set_ivar(wx, -1, 0, v_idx); + if (pdev->msix_enabled) + wr32(wx, WX_PX_ITR(v_idx), 1950); +} +EXPORT_SYMBOL(wx_configure_vectors); + +MODULE_LICENSE("GPL"); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.h b/drivers/net/ethernet/wangxun/libwx/wx_lib.h new file mode 100644 index 000000000000..8ae657155f34 --- /dev/null +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.h @@ -0,0 +1,20 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* + * WangXun Gigabit PCI Express Linux driver + * Copyright (c) 2019 - 2022 Beijing WangXun Technology Co., Ltd. + */ + +#ifndef _WX_LIB_H_ +#define _WX_LIB_H_ + +void wx_reset_interrupt_capability(struct wx *wx); +void wx_clear_interrupt_scheme(struct wx *wx); +int wx_init_interrupt_scheme(struct wx *wx); +irqreturn_t wx_msix_clean_rings(int __always_unused irq, void *data); +void wx_free_irq(struct wx *wx); +int wx_setup_isb_resources(struct wx *wx); +void wx_free_isb_resources(struct wx *wx); +u32 wx_misc_isb(struct wx *wx, enum wx_isb_idx idx); +void wx_configure_vectors(struct wx *wx); + +#endif /* _NGBE_LIB_H_ */ diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index c86a37914d43..eb2a1918fb12 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -5,6 +5,7 @@ #define _WX_TYPE_H_ #include +#include /* Vendor ID */ #ifndef PCI_VENDOR_ID_WANGXUN @@ -66,6 +67,22 @@ #define WX_CFG_PORT_CTL 0x14400 #define WX_CFG_PORT_CTL_DRV_LOAD BIT(3) +/* GPIO Registers */ +#define WX_GPIO_DR 0x14800 +#define WX_GPIO_DR_0 BIT(0) /* SDP0 Data Value */ +#define WX_GPIO_DR_1 BIT(1) /* SDP1 Data Value */ +#define WX_GPIO_DDR 0x14804 +#define WX_GPIO_DDR_0 BIT(0) /* SDP0 IO direction */ +#define WX_GPIO_DDR_1 BIT(1) /* SDP1 IO direction */ +#define WX_GPIO_CTL 0x14808 +#define WX_GPIO_INTEN 0x14830 +#define WX_GPIO_INTEN_0 BIT(0) +#define WX_GPIO_INTEN_1 BIT(1) +#define WX_GPIO_INTMASK 0x14834 +#define WX_GPIO_INTTYPE_LEVEL 0x14838 +#define WX_GPIO_POLARITY 0x1483C +#define WX_GPIO_EOI 0x1484C + /*********************** Transmit DMA registers **************************/ /* transmit global control */ #define WX_TDM_CTL 0x18000 @@ -151,8 +168,28 @@ /* Interrupt Registers */ #define WX_BME_CTL 0x12020 #define WX_PX_MISC_IC 0x100 +#define WX_PX_MISC_ICS 0x104 +#define WX_PX_MISC_IEN 0x108 +#define WX_PX_INTA 0x110 +#define WX_PX_GPIE 0x118 +#define WX_PX_GPIE_MODEL BIT(0) +#define WX_PX_IC 0x120 #define WX_PX_IMS(_i) (0x140 + (_i) * 4) +#define WX_PX_IMC(_i) (0x150 + (_i) * 4) +#define WX_PX_ISB_ADDR_L 0x160 +#define WX_PX_ISB_ADDR_H 0x164 #define WX_PX_TRANSACTION_PENDING 0x168 +#define WX_PX_ITRSEL 0x180 +#define WX_PX_ITR(_i) (0x200 + (_i) * 4) +#define WX_PX_ITR_CNT_WDIS BIT(31) +#define WX_PX_MISC_IVAR 0x4FC +#define WX_PX_IVAR(_i) (0x500 + (_i) * 4) + +#define WX_PX_IVAR_ALLOC_VAL 0x80 /* Interrupt Allocation valid */ +#define WX_7K_ITR 595 +#define WX_12K_ITR 336 +#define WX_SP_MAX_EITR 0x00000FF8U +#define WX_EM_MAX_EITR 0x00007FFCU /* transmit DMA Registers */ #define WX_PX_TR_CFG(_i) (0x03010 + ((_i) * 0x40)) @@ -312,6 +349,58 @@ enum wx_reset_type { WX_GLOBAL_RESET }; +/* iterator for handling rings in ring container */ +#define wx_for_each_ring(posm, headm) \ + for (posm = (headm).ring; posm; posm = posm->next) + +struct wx_ring_container { + struct wx_ring *ring; /* pointer to linked list of rings */ + u8 count; /* total number of rings in vector */ + u8 itr; /* current ITR setting for ring */ +}; + +struct wx_ring { + struct wx_ring *next; /* pointer to next ring in q_vector */ + struct wx_q_vector *q_vector; /* backpointer to host q_vector */ + struct net_device *netdev; /* netdev ring belongs to */ + struct device *dev; /* device for DMA mapping */ + + u16 count; /* amount of descriptors */ + + u8 queue_index; /* needed for multiqueue queue management */ + u8 reg_idx; /* holds the special value that gets + * the hardware register offset + * associated with this ring, which is + * different for DCB and RSS modes + */ +} ____cacheline_internodealigned_in_smp; + +struct wx_q_vector { + struct wx *wx; + int cpu; /* CPU for DCA */ + u16 v_idx; /* index of q_vector within array, also used for + * finding the bit in EICR and friends that + * represents the vector for this ring + */ + u16 itr; /* Interrupt throttle rate written to EITR */ + struct wx_ring_container rx, tx; + struct napi_struct napi; + struct rcu_head rcu; /* to avoid race with update stats on free */ + + char name[IFNAMSIZ + 17]; + + /* for dynamic allocation of rings associated with this q_vector */ + struct wx_ring ring[0] ____cacheline_internodealigned_in_smp; +}; + +enum wx_isb_idx { + WX_ISB_HEADER, + WX_ISB_MISC, + WX_ISB_VEC0, + WX_ISB_VEC1, + WX_ISB_MAX +}; + struct wx { u8 __iomem *hw_addr; struct pci_dev *pdev; @@ -360,6 +449,18 @@ struct wx { u32 tx_ring_count; u32 rx_ring_count; + struct wx_ring *tx_ring[64] ____cacheline_aligned_in_smp; + struct wx_ring *rx_ring[64]; + struct wx_q_vector *q_vector[64]; + + unsigned int queues_per_pool; + struct msix_entry *msix_entries; + + /* misc interrupt status block */ + dma_addr_t isb_dma; + u32 *isb_mem; + u32 isb_tag[WX_ISB_MAX]; + #define WX_MAX_RETA_ENTRIES 128 u8 rss_indir_tbl[WX_MAX_RETA_ENTRIES]; From patchwork Fri Feb 3 09:11:27 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127225 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id ABEE7C636D6 for ; Fri, 3 Feb 2023 09:15:47 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232234AbjBCJPq (ORCPT ); Fri, 3 Feb 2023 04:15:46 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36174 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230435AbjBCJPp (ORCPT ); Fri, 3 Feb 2023 04:15:45 -0500 Received: from smtpbguseast1.qq.com (smtpbguseast1.qq.com [54.204.34.129]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 269528E49E for ; Fri, 3 Feb 2023 01:15:42 -0800 (PST) X-QQ-mid: bizesmtp68t1675415737tqdhx0qq Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:37 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: qxAR8rpszOtLNSsuvS/aQLTvZSXhXtvZOnA6/9pA0RYX+1rmNo5MBdw33Z6YM JNGyDEY/6A9x0txxb/t5d6EDg+48zuAJArPElySjXp+xBMP3Nl3yrKzrsyVLSlNn9RmrK35 1Ds5V5SmNGH2ZSQ1P2O1i8DBeIeNRG3aCGNG8fS6jIn2D236UeOJ1YNujDF/fy9+dnFdp+h ak+yWl47/p5sPeY+hWKk9L+mAAYf4L7XSDS9wOJb3Bnti83XGPk7o70zc77kXa/iv/Zltbn aKsY80Gx7TP3/wStkALU1Y68iJwJ0Jb/FEGcnH3Cy77ZdLf6w/IwvJGbt+6mR34l03yZsgH Rbc5kdnG3lWF/cgs4ivqF3Nib+TDAwvEcXoGw4XWI0pFp6763w4tFOGwBBllM1JNdmRdWKF pDJqhKXkWR79w9+Db6C2iQ== X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Subject: [PATCH net-next v3 02/10] net: ngbe: Add irqs request flow Date: Fri, 3 Feb 2023 17:11:27 +0800 Message-Id: <20230203091135.3294377-3-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org From: Mengyuan Lou Add request_irq for tx/rx rings and misc other events. If the application is successful, config vertors for interrupts. Enable some base interrupts mask in ngbe_irq_enable. Signed-off-by: Mengyuan Lou --- drivers/net/ethernet/wangxun/ngbe/ngbe_main.c | 189 +++++++++++++++++- drivers/net/ethernet/wangxun/ngbe/ngbe_type.h | 14 ++ 2 files changed, 202 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c index ed52f80b5475..a9772f929274 100644 --- a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c +++ b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c @@ -13,6 +13,7 @@ #include "../libwx/wx_type.h" #include "../libwx/wx_hw.h" +#include "../libwx/wx_lib.h" #include "ngbe_type.h" #include "ngbe_mdio.h" #include "ngbe_hw.h" @@ -148,6 +149,161 @@ static int ngbe_sw_init(struct wx *wx) return 0; } +/** + * ngbe_irq_enable - Enable default interrupt generation settings + * @wx: board private structure + * @queues: enable all queues interrupts + **/ +static void ngbe_irq_enable(struct wx *wx, bool queues) +{ + u32 mask; + + /* enable misc interrupt */ + mask = NGBE_PX_MISC_IEN_MASK; + + wr32(wx, WX_GPIO_DDR, WX_GPIO_DDR_0); + wr32(wx, WX_GPIO_INTEN, WX_GPIO_INTEN_0 | WX_GPIO_INTEN_1); + wr32(wx, WX_GPIO_INTTYPE_LEVEL, 0x0); + wr32(wx, WX_GPIO_POLARITY, wx->gpio_ctrl ? 0 : 0x3); + + wr32(wx, WX_PX_MISC_IEN, mask); + + /* mask interrupt */ + if (queues) + wx_intr_enable(wx, NGBE_INTR_ALL); + else + wx_intr_enable(wx, NGBE_INTR_MISC(wx)); +} + +/** + * ngbe_intr - msi/legacy mode Interrupt Handler + * @irq: interrupt number + * @data: pointer to a network interface device structure + **/ +static irqreturn_t ngbe_intr(int __always_unused irq, void *data) +{ + struct wx_q_vector *q_vector; + struct wx *wx = data; + struct pci_dev *pdev; + u32 eicr; + + q_vector = wx->q_vector[0]; + pdev = wx->pdev; + + eicr = wx_misc_isb(wx, WX_ISB_VEC0); + if (!eicr) { + /* shared interrupt alert! + * the interrupt that we masked before the EICR read. + */ + if (netif_running(wx->netdev)) + ngbe_irq_enable(wx, true); + return IRQ_NONE; /* Not our interrupt */ + } + wx->isb_mem[WX_ISB_VEC0] = 0; + if (!(pdev->msi_enabled)) + wr32(wx, WX_PX_INTA, 1); + + wx->isb_mem[WX_ISB_MISC] = 0; + /* would disable interrupts here but it is auto disabled */ + napi_schedule_irqoff(&q_vector->napi); + + if (netif_running(wx->netdev)) + ngbe_irq_enable(wx, false); + + return IRQ_HANDLED; +} + +static irqreturn_t ngbe_msix_other(int __always_unused irq, void *data) +{ + struct wx *wx = data; + + /* re-enable the original interrupt state, no lsc, no queues */ + if (netif_running(wx->netdev)) + ngbe_irq_enable(wx, false); + + return IRQ_HANDLED; +} + +/** + * ngbe_request_msix_irqs - Initialize MSI-X interrupts + * @wx: board private structure + * + * ngbe_request_msix_irqs allocates MSI-X vectors and requests + * interrupts from the kernel. + **/ +static int ngbe_request_msix_irqs(struct wx *wx) +{ + struct net_device *netdev = wx->netdev; + int vector, err; + + for (vector = 0; vector < wx->num_q_vectors; vector++) { + struct wx_q_vector *q_vector = wx->q_vector[vector]; + struct msix_entry *entry = &wx->msix_entries[vector]; + + if (q_vector->tx.ring && q_vector->rx.ring) + snprintf(q_vector->name, sizeof(q_vector->name) - 1, + "%s-TxRx-%d", netdev->name, entry->entry); + else + /* skip this unused q_vector */ + continue; + + err = request_irq(entry->vector, wx_msix_clean_rings, 0, + q_vector->name, q_vector); + if (err) { + wx_err(wx, "request_irq failed for MSIX interrupt %s Error: %d\n", + q_vector->name, err); + goto free_queue_irqs; + } + } + + err = request_irq(wx->msix_entries[vector].vector, + ngbe_msix_other, 0, netdev->name, wx); + + if (err) { + wx_err(wx, "request_irq for msix_other failed: %d\n", err); + goto free_queue_irqs; + } + + return 0; + +free_queue_irqs: + while (vector) { + vector--; + free_irq(wx->msix_entries[vector].vector, + wx->q_vector[vector]); + } + wx_reset_interrupt_capability(wx); + return err; +} + +/** + * ngbe_request_irq - initialize interrupts + * @wx: board private structure + * + * Attempts to configure interrupts using the best available + * capabilities of the hardware and kernel. + **/ +static int ngbe_request_irq(struct wx *wx) +{ + struct net_device *netdev = wx->netdev; + struct pci_dev *pdev = wx->pdev; + int err; + + if (pdev->msix_enabled) + err = ngbe_request_msix_irqs(wx); + else if (pdev->msi_enabled) + err = request_irq(pdev->irq, ngbe_intr, 0, + netdev->name, wx); + else + err = request_irq(pdev->irq, ngbe_intr, IRQF_SHARED, + netdev->name, wx); + + if (err) + wx_err(wx, "request_irq failed, Error %d\n", err); + + return err; +} + static void ngbe_disable_device(struct wx *wx) { struct net_device *netdev = wx->netdev; @@ -157,6 +313,7 @@ static void ngbe_disable_device(struct wx *wx) netif_tx_disable(netdev); if (wx->gpio_ctrl) ngbe_sfp_modules_txrx_powerctl(wx, false); + wx_irq_disable(wx); } static void ngbe_down(struct wx *wx) @@ -167,8 +324,15 @@ static void ngbe_down(struct wx *wx) static void ngbe_up(struct wx *wx) { + wx_configure_vectors(wx); + + /* clear any pending interrupts, may auto mask */ + rd32(wx, WX_PX_IC); + rd32(wx, WX_PX_MISC_IC); + ngbe_irq_enable(wx, true); if (wx->gpio_ctrl) ngbe_sfp_modules_txrx_powerctl(wx, true); + phy_start(wx->phydev); } @@ -187,12 +351,26 @@ static int ngbe_open(struct net_device *netdev) int err; wx_control_hw(wx, true); + + err = wx_setup_isb_resources(wx); + if (err) + return err; + + wx_configure(wx); + + err = ngbe_request_irq(wx); + if (err) + goto err_req_irq; + err = ngbe_phy_connect(wx); if (err) return err; ngbe_up(wx); return 0; +err_req_irq: + wx_free_isb_resources(wx); + return err; } /** @@ -211,6 +389,8 @@ static int ngbe_close(struct net_device *netdev) struct wx *wx = netdev_priv(netdev); ngbe_down(wx); + wx_free_irq(wx); + wx_free_isb_resources(wx); phy_disconnect(wx->phydev); wx_control_hw(wx, false); @@ -411,10 +591,14 @@ static int ngbe_probe(struct pci_dev *pdev, eth_hw_addr_set(netdev, wx->mac.perm_addr); wx_mac_set_default_filter(wx, wx->mac.perm_addr); + err = wx_init_interrupt_scheme(wx); + if (err) + goto err_free_mac_table; + /* phy Interface Configuration */ err = ngbe_mdio_init(wx); if (err) - goto err_free_mac_table; + goto err_clear_interrupt_scheme; err = register_netdev(netdev); if (err) @@ -431,6 +615,8 @@ static int ngbe_probe(struct pci_dev *pdev, err_register: wx_control_hw(wx, false); +err_clear_interrupt_scheme: + wx_clear_interrupt_scheme(wx); err_free_mac_table: kfree(wx->mac_table); err_pci_release_regions: @@ -462,6 +648,7 @@ static void ngbe_remove(struct pci_dev *pdev) pci_select_bars(pdev, IORESOURCE_MEM)); kfree(wx->mac_table); + wx_clear_interrupt_scheme(wx); pci_disable_pcie_error_reporting(pdev); pci_disable_device(pdev); diff --git a/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h b/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h index fd71260f73de..b60f5f0f64fa 100644 --- a/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h +++ b/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h @@ -90,6 +90,20 @@ enum NGBE_MSCA_CMD_value { #define NGBE_GPIO_DDR_0 BIT(0) /* SDP0 IO direction */ #define NGBE_GPIO_DDR_1 BIT(1) /* SDP1 IO direction */ +/* Extended Interrupt Enable Set */ +#define NGBE_PX_MISC_IEN_DEV_RST BIT(10) +#define NGBE_PX_MISC_IEN_ETH_LK BIT(18) +#define NGBE_PX_MISC_IEN_INT_ERR BIT(20) +#define NGBE_PX_MISC_IEN_GPIO BIT(26) +#define NGBE_PX_MISC_IEN_MASK ( \ + NGBE_PX_MISC_IEN_DEV_RST | \ + NGBE_PX_MISC_IEN_ETH_LK | \ + NGBE_PX_MISC_IEN_INT_ERR | \ + NGBE_PX_MISC_IEN_GPIO) + +#define NGBE_INTR_ALL 0x1FF +#define NGBE_INTR_MISC(A) BIT((A)->num_q_vectors) + #define NGBE_PHY_CONFIG(reg_offset) (0x14000 + ((reg_offset) * 4)) #define NGBE_CFG_LAN_SPEED 0x14440 #define NGBE_CFG_PORT_ST 0x14404 From patchwork Fri Feb 3 09:11:28 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127227 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id CCDDCC636CC for ; Fri, 3 Feb 2023 09:16:01 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232594AbjBCJQB (ORCPT ); Fri, 3 Feb 2023 04:16:01 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36288 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231307AbjBCJPw (ORCPT ); Fri, 3 Feb 2023 04:15:52 -0500 Received: from smtpbg151.qq.com (smtpbg151.qq.com [18.169.211.239]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 07E4995D2D for ; Fri, 3 Feb 2023 01:15:49 -0800 (PST) X-QQ-mid: bizesmtp68t1675415740tdk7m7v5 Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:39 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: lkL5M32tl2AxbT0+ypf5VY/MI4CquVqQcNBgd2c0fUkN7hlYGSmuXuapbPMhg xnT7cgf5W9kwRH/QSreJjEwPoPlyo2Yyzz1vBfYagKUweaoaI30OjljsVxlfnDF2Gx+wO2u 55Pay7lVbx3EhsneffuHc7Ig3sFOHQVZVDLbEKoqvf0y8JNhdK5XLwMs4mZc1Lge9adPOKQ hU4UFTikUCyV+9beLctTdtpet5I1ANN590riwTteSa4X7KkSOGcQ0nkqHH9Rd0Fw6VbdVVj etkoU+YxCg7rJT/toGXJMVxdsy2kxLc0ypvqlev42ggSIzTz7yjh8oZ6Mu4SynegdkC0B9F JCYIFCKdt6RRUrqeZHxQJ7tFmCVXKPZAjKPcn85l2KG1H+bhU3oArtrNTe32VPKkif6qA64 b5kAKz8OMkg= X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 03/10] net: txgbe: Add interrupt support Date: Fri, 3 Feb 2023 17:11:28 +0800 Message-Id: <20230203091135.3294377-4-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Determine proper interrupt scheme to enable and handle interrupt. Signed-off-by: Jiawen Wu --- .../net/ethernet/wangxun/txgbe/txgbe_main.c | 201 ++++++++++++++++++ .../net/ethernet/wangxun/txgbe/txgbe_type.h | 18 ++ 2 files changed, 219 insertions(+) diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c index aa4d09df3b01..48cca2fb54c7 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c @@ -11,6 +11,7 @@ #include #include "../libwx/wx_type.h" +#include "../libwx/wx_lib.h" #include "../libwx/wx_hw.h" #include "txgbe_type.h" #include "txgbe_hw.h" @@ -72,9 +73,158 @@ static int txgbe_enumerate_functions(struct wx *wx) return physfns; } +/** + * txgbe_irq_enable - Enable default interrupt generation settings + * @wx: pointer to private structure + * @queues: enable irqs for queues + **/ +static void txgbe_irq_enable(struct wx *wx, bool queues) +{ + /* unmask interrupt */ + wx_intr_enable(wx, TXGBE_INTR_MISC(wx)); + if (queues) + wx_intr_enable(wx, TXGBE_INTR_QALL(wx)); +} + +/** + * txgbe_intr - msi/legacy mode Interrupt Handler + * @irq: interrupt number + * @data: pointer to a network interface device structure + **/ +static irqreturn_t txgbe_intr(int __always_unused irq, void *data) +{ + struct wx_q_vector *q_vector; + struct wx *wx = data; + struct pci_dev *pdev; + u32 eicr; + + q_vector = wx->q_vector[0]; + pdev = wx->pdev; + + eicr = wx_misc_isb(wx, WX_ISB_VEC0); + if (!eicr) { + /* shared interrupt alert! + * the interrupt that we masked before the ICR read. + */ + if (netif_running(wx->netdev)) + txgbe_irq_enable(wx, true); + return IRQ_NONE; /* Not our interrupt */ + } + wx->isb_mem[WX_ISB_VEC0] = 0; + if (!(pdev->msi_enabled)) + wr32(wx, WX_PX_INTA, 1); + + wx->isb_mem[WX_ISB_MISC] = 0; + /* would disable interrupts here but it is auto disabled */ + napi_schedule_irqoff(&q_vector->napi); + + /* re-enable link(maybe) and non-queue interrupts, no flush. + * txgbe_poll will re-enable the queue interrupts + */ + if (netif_running(wx->netdev)) + txgbe_irq_enable(wx, false); + + return IRQ_HANDLED; +} + +static irqreturn_t txgbe_msix_other(int __always_unused irq, void *data) +{ + struct wx *wx = data; + + /* re-enable the original interrupt state */ + if (netif_running(wx->netdev)) + txgbe_irq_enable(wx, false); + + return IRQ_HANDLED; +} + +/** + * txgbe_request_msix_irqs - Initialize MSI-X interrupts + * @wx: board private structure + * + * Allocate MSI-X vectors and request interrupts from the kernel. + **/ +static int txgbe_request_msix_irqs(struct wx *wx) +{ + struct net_device *netdev = wx->netdev; + int vector, err; + + for (vector = 0; vector < wx->num_q_vectors; vector++) { + struct wx_q_vector *q_vector = wx->q_vector[vector]; + struct msix_entry *entry = &wx->msix_entries[vector]; + + if (q_vector->tx.ring && q_vector->rx.ring) + snprintf(q_vector->name, sizeof(q_vector->name) - 1, + "%s-TxRx-%d", netdev->name, entry->entry); + else + /* skip this unused q_vector */ + continue; + + err = request_irq(entry->vector, wx_msix_clean_rings, 0, + q_vector->name, q_vector); + if (err) { + wx_err(wx, "request_irq failed for MSIX interrupt %s Error: %d\n", + q_vector->name, err); + goto free_queue_irqs; + } + } + + err = request_irq(wx->msix_entries[vector].vector, + txgbe_msix_other, 0, netdev->name, wx); + if (err) { + wx_err(wx, "request_irq for msix_other failed: %d\n", err); + goto free_queue_irqs; + } + + return 0; + +free_queue_irqs: + while (vector) { + vector--; + free_irq(wx->msix_entries[vector].vector, + wx->q_vector[vector]); + } + wx_reset_interrupt_capability(wx); + return err; +} + +/** + * txgbe_request_irq - initialize interrupts + * @wx: board private structure + * + * Attempt to configure interrupts using the best available + * capabilities of the hardware and kernel. + **/ +static int txgbe_request_irq(struct wx *wx) +{ + struct net_device *netdev = wx->netdev; + struct pci_dev *pdev = wx->pdev; + int err; + + if (pdev->msix_enabled) + err = txgbe_request_msix_irqs(wx); + else if (pdev->msi_enabled) + err = request_irq(wx->pdev->irq, &txgbe_intr, 0, + netdev->name, wx); + else + err = request_irq(wx->pdev->irq, &txgbe_intr, IRQF_SHARED, + netdev->name, wx); + + if (err) + wx_err(wx, "request_irq failed, Error %d\n", err); + + return err; +} + static void txgbe_up_complete(struct wx *wx) { wx_control_hw(wx, true); + wx_configure_vectors(wx); + + /* clear any pending interrupts, may auto mask */ + rd32(wx, WX_PX_IC); + rd32(wx, WX_PX_MISC_IC); + txgbe_irq_enable(wx, true); } static void txgbe_reset(struct wx *wx) @@ -104,6 +254,8 @@ static void txgbe_disable_device(struct wx *wx) netif_carrier_off(netdev); netif_tx_disable(netdev); + wx_irq_disable(wx); + if (wx->bus.func < 2) wr32m(wx, TXGBE_MIS_PRB_CTL, TXGBE_MIS_PRB_CTL_LAN_UP(wx->bus.func), 0); else @@ -132,6 +284,7 @@ static void txgbe_down(struct wx *wx) **/ static int txgbe_sw_init(struct wx *wx) { + u16 msix_count = 0; int err; wx->mac.num_rar_entries = TXGBE_SP_RAR_ENTRIES; @@ -156,6 +309,25 @@ static int txgbe_sw_init(struct wx *wx) break; } + /* Set common capability flags and settings */ + wx->max_q_vectors = TXGBE_MAX_MSIX_VECTORS; + err = wx_get_pcie_msix_counts(wx, &msix_count, TXGBE_MAX_MSIX_VECTORS); + if (err) + wx_err(wx, "Do not support MSI-X\n"); + wx->mac.max_msix_vectors = msix_count; + + /* enable itr by default in dynamic mode */ + wx->rx_itr_setting = 1; + wx->tx_itr_setting = 1; + + /* set default ring sizes */ + wx->tx_ring_count = TXGBE_DEFAULT_TXD; + wx->rx_ring_count = TXGBE_DEFAULT_RXD; + + /* set default work limits */ + wx->tx_work_limit = TXGBE_DEFAULT_TX_WORK; + wx->rx_work_limit = TXGBE_DEFAULT_RX_WORK; + return 0; } @@ -171,10 +343,28 @@ static int txgbe_sw_init(struct wx *wx) static int txgbe_open(struct net_device *netdev) { struct wx *wx = netdev_priv(netdev); + int err; + + err = wx_setup_isb_resources(wx); + if (err) + goto err_reset; + + wx_configure(wx); + + err = txgbe_request_irq(wx); + if (err) + goto err_free_isb; txgbe_up_complete(wx); return 0; + +err_free_isb: + wx_free_isb_resources(wx); +err_reset: + txgbe_reset(wx); + + return err; } /** @@ -187,6 +377,9 @@ static int txgbe_open(struct net_device *netdev) static void txgbe_close_suspend(struct wx *wx) { txgbe_disable_device(wx); + + wx_free_irq(wx); + wx_free_isb_resources(wx); } /** @@ -205,6 +398,8 @@ static int txgbe_close(struct net_device *netdev) struct wx *wx = netdev_priv(netdev); txgbe_down(wx); + wx_free_irq(wx); + wx_free_isb_resources(wx); wx_control_hw(wx, false); return 0; @@ -367,6 +562,10 @@ static int txgbe_probe(struct pci_dev *pdev, eth_hw_addr_set(netdev, wx->mac.perm_addr); wx_mac_set_default_filter(wx, wx->mac.perm_addr); + err = wx_init_interrupt_scheme(wx); + if (err) + goto err_free_mac_table; + /* Save off EEPROM version number and Option Rom version which * together make a unique identify for the eeprom */ @@ -435,6 +634,7 @@ static int txgbe_probe(struct pci_dev *pdev, return 0; err_release_hw: + wx_clear_interrupt_scheme(wx); wx_control_hw(wx, false); err_free_mac_table: kfree(wx->mac_table); @@ -468,6 +668,7 @@ static void txgbe_remove(struct pci_dev *pdev) pci_select_bars(pdev, IORESOURCE_MEM)); kfree(wx->mac_table); + wx_clear_interrupt_scheme(wx); pci_disable_pcie_error_reporting(pdev); diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h b/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h index cbd705a9f4bd..ec28cf60c91d 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h @@ -67,6 +67,7 @@ #define TXGBE_PBANUM1_PTR 0x06 #define TXGBE_PBANUM_PTR_GUARD 0xFAFA +#define TXGBE_MAX_MSIX_VECTORS 64 #define TXGBE_MAX_FDIR_INDICES 63 #define TXGBE_MAX_RX_QUEUES (TXGBE_MAX_FDIR_INDICES + 1) @@ -77,6 +78,23 @@ #define TXGBE_SP_RAR_ENTRIES 128 #define TXGBE_SP_MC_TBL_SIZE 128 +/* TX/RX descriptor defines */ +#define TXGBE_DEFAULT_TXD 512 +#define TXGBE_DEFAULT_TX_WORK 256 + +#if (PAGE_SIZE < 8192) +#define TXGBE_DEFAULT_RXD 512 +#define TXGBE_DEFAULT_RX_WORK 256 +#else +#define TXGBE_DEFAULT_RXD 256 +#define TXGBE_DEFAULT_RX_WORK 128 +#endif + +#define TXGBE_INTR_MISC(A) BIT((A)->num_q_vectors) +#define TXGBE_INTR_QALL(A) (TXGBE_INTR_MISC(A) - 1) + +#define TXGBE_MAX_EITR GENMASK(11, 3) + extern char txgbe_driver_name[]; #endif /* _TXGBE_TYPE_H_ */ From patchwork Fri Feb 3 09:11:29 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127234 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 6EC80C636CC for ; Fri, 3 Feb 2023 09:18:08 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232817AbjBCJSG (ORCPT ); Fri, 3 Feb 2023 04:18:06 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:37870 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232576AbjBCJRZ (ORCPT ); Fri, 3 Feb 2023 04:17:25 -0500 X-Greylist: delayed 72 seconds by postgrey-1.37 at lindbergh.monkeyblade.net; Fri, 03 Feb 2023 01:17:04 PST Received: from smtpbgbr1.qq.com (smtpbgbr1.qq.com [54.207.19.206]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 30D9F991DC for ; Fri, 3 Feb 2023 01:17:03 -0800 (PST) X-QQ-mid: bizesmtp68t1675415743t7x84n9n Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:42 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: f9mzMPD0x2n/OJk5D1mJD4Z5egxv1Sflb4mHcZsD1Hb2zywlYMM8LW4gvEgfH YlY7QtKcZwDuMBYCa+hbSHsZkkVx+bONnAFnvEkOzcC8PdiE9057fV2BI4Q52LMzfvWPcQx B5yFiOFUeyLEd1TQRUyq3nz3PvLKb4Q7efWzAX5ZfK7CtEcGxG5dRfBY9SWDioB2g9AkrwV J1rVsK+t9cB6C56hzcvlKxKolqgY0hEmVXJj7APaYcrWih9ys5NF9SUNmZPa1U6OssUhdqs RAIqIHeuXHbgGIOj5u6Dik0lG+kfFJyxcyHkGq6TekG0HylpLjkR5nDe87M7rS1OY2QKIRb RPy52dOBK4R1RrXs6Cqri4tOZ1RD3yErjsVNhJ6tPXHrpMEcJHzu13zPuPbZWnHG4A5iZ+h 5BpgktXhTYg= X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 04/10] net: libwx: Configure Rx and Tx unit on hardware Date: Fri, 3 Feb 2023 17:11:29 +0800 Message-Id: <20230203091135.3294377-5-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Configure hardware for preparing to process packets. Including configure receive and transmit unit of the MAC layer, and setup the specific rings. Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/libwx/wx_hw.c | 597 +++++++++++++++++++ drivers/net/ethernet/wangxun/libwx/wx_hw.h | 2 + drivers/net/ethernet/wangxun/libwx/wx_type.h | 85 +++ 3 files changed, 684 insertions(+) diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c index b39fdef5a5b0..db313b53a6b6 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c @@ -804,6 +804,37 @@ void wx_flush_sw_mac_table(struct wx *wx) } EXPORT_SYMBOL(wx_flush_sw_mac_table); +static int wx_add_mac_filter(struct wx *wx, u8 *addr, u16 pool) +{ + u32 i; + + if (is_zero_ether_addr(addr)) + return -EINVAL; + + for (i = 0; i < wx->mac.num_rar_entries; i++) { + if (wx->mac_table[i].state & WX_MAC_STATE_IN_USE) { + if (ether_addr_equal(addr, wx->mac_table[i].addr)) { + if (wx->mac_table[i].pools != (1ULL << pool)) { + memcpy(wx->mac_table[i].addr, addr, ETH_ALEN); + wx->mac_table[i].pools |= (1ULL << pool); + wx_sync_mac_table(wx); + return i; + } + } + } + + if (wx->mac_table[i].state & WX_MAC_STATE_IN_USE) + continue; + wx->mac_table[i].state |= (WX_MAC_STATE_MODIFIED | + WX_MAC_STATE_IN_USE); + memcpy(wx->mac_table[i].addr, addr, ETH_ALEN); + wx->mac_table[i].pools |= (1ULL << pool); + wx_sync_mac_table(wx); + return i; + } + return -ENOMEM; +} + static int wx_del_mac_filter(struct wx *wx, u8 *addr, u16 pool) { u32 i; @@ -828,6 +859,184 @@ static int wx_del_mac_filter(struct wx *wx, u8 *addr, u16 pool) return -ENOMEM; } +static int wx_available_rars(struct wx *wx) +{ + u32 i, count = 0; + + for (i = 0; i < wx->mac.num_rar_entries; i++) { + if (wx->mac_table[i].state == 0) + count++; + } + + return count; +} + +/** + * wx_write_uc_addr_list - write unicast addresses to RAR table + * @netdev: network interface device structure + * @pool: index for mac table + * + * Writes unicast address list to the RAR table. + * Returns: -ENOMEM on failure/insufficient address space + * 0 on no addresses written + * X on writing X addresses to the RAR table + **/ +static int wx_write_uc_addr_list(struct net_device *netdev, int pool) +{ + struct wx *wx = netdev_priv(netdev); + int count = 0; + + /* return ENOMEM indicating insufficient memory for addresses */ + if (netdev_uc_count(netdev) > wx_available_rars(wx)) + return -ENOMEM; + + if (!netdev_uc_empty(netdev)) { + struct netdev_hw_addr *ha; + + netdev_for_each_uc_addr(ha, netdev) { + wx_del_mac_filter(wx, ha->addr, pool); + wx_add_mac_filter(wx, ha->addr, pool); + count++; + } + } + return count; +} + +/** + * wx_mta_vector - Determines bit-vector in multicast table to set + * @wx: pointer to private structure + * @mc_addr: the multicast address + * + * Extracts the 12 bits, from a multicast address, to determine which + * bit-vector to set in the multicast table. The hardware uses 12 bits, from + * incoming rx multicast addresses, to determine the bit-vector to check in + * the MTA. Which of the 4 combination, of 12-bits, the hardware uses is set + * by the MO field of the MCSTCTRL. The MO field is set during initialization + * to mc_filter_type. + **/ +static u32 wx_mta_vector(struct wx *wx, u8 *mc_addr) +{ + u32 vector = 0; + + switch (wx->mac.mc_filter_type) { + case 0: /* use bits [47:36] of the address */ + vector = ((mc_addr[4] >> 4) | (((u16)mc_addr[5]) << 4)); + break; + case 1: /* use bits [46:35] of the address */ + vector = ((mc_addr[4] >> 3) | (((u16)mc_addr[5]) << 5)); + break; + case 2: /* use bits [45:34] of the address */ + vector = ((mc_addr[4] >> 2) | (((u16)mc_addr[5]) << 6)); + break; + case 3: /* use bits [43:32] of the address */ + vector = ((mc_addr[4]) | (((u16)mc_addr[5]) << 8)); + break; + default: /* Invalid mc_filter_type */ + wx_err(wx, "MC filter type param set incorrectly\n"); + break; + } + + /* vector can only be 12-bits or boundary will be exceeded */ + vector &= 0xFFF; + return vector; +} + +/** + * wx_set_mta - Set bit-vector in multicast table + * @wx: pointer to private structure + * @mc_addr: Multicast address + * + * Sets the bit-vector in the multicast table. + **/ +static void wx_set_mta(struct wx *wx, u8 *mc_addr) +{ + u32 vector, vector_bit, vector_reg; + + wx->addr_ctrl.mta_in_use++; + + vector = wx_mta_vector(wx, mc_addr); + wx_dbg(wx, " bit-vector = 0x%03X\n", vector); + + /* The MTA is a register array of 128 32-bit registers. It is treated + * like an array of 4096 bits. We want to set bit + * BitArray[vector_value]. So we figure out what register the bit is + * in, read it, OR in the new bit, then write back the new value. The + * register is determined by the upper 7 bits of the vector value and + * the bit within that register are determined by the lower 5 bits of + * the value. + */ + vector_reg = (vector >> 5) & 0x7F; + vector_bit = vector & 0x1F; + wx->mac.mta_shadow[vector_reg] |= (1 << vector_bit); +} + +/** + * wx_update_mc_addr_list - Updates MAC list of multicast addresses + * @wx: pointer to private structure + * @netdev: pointer to net device structure + * + * The given list replaces any existing list. Clears the MC addrs from receive + * address registers and the multicast table. Uses unused receive address + * registers for the first multicast addresses, and hashes the rest into the + * multicast table. + **/ +static void wx_update_mc_addr_list(struct wx *wx, struct net_device *netdev) +{ + struct netdev_hw_addr *ha; + u32 i, psrctl; + + /* Set the new number of MC addresses that we are being requested to + * use. + */ + wx->addr_ctrl.num_mc_addrs = netdev_mc_count(netdev); + wx->addr_ctrl.mta_in_use = 0; + + /* Clear mta_shadow */ + wx_dbg(wx, " Clearing MTA\n"); + memset(&wx->mac.mta_shadow, 0, sizeof(wx->mac.mta_shadow)); + + /* Update mta_shadow */ + netdev_for_each_mc_addr(ha, netdev) { + wx_dbg(wx, " Adding the multicast addresses:\n"); + wx_set_mta(wx, ha->addr); + } + + /* Enable mta */ + for (i = 0; i < wx->mac.mcft_size; i++) + wr32a(wx, WX_PSR_MC_TBL(0), i, + wx->mac.mta_shadow[i]); + + if (wx->addr_ctrl.mta_in_use > 0) { + psrctl = rd32(wx, WX_PSR_CTL); + psrctl &= ~(WX_PSR_CTL_MO | WX_PSR_CTL_MFE); + psrctl |= WX_PSR_CTL_MFE | + (wx->mac.mc_filter_type << WX_PSR_CTL_MO_SHIFT); + wr32(wx, WX_PSR_CTL, psrctl); + } + + wx_dbg(wx, "Update mc addr list Complete\n"); +} + +/** + * wx_write_mc_addr_list - write multicast addresses to MTA + * @netdev: network interface device structure + * + * Writes multicast address list to the MTA hash table. + * Returns: 0 on no addresses written + * X on writing X addresses to MTA + **/ +static int wx_write_mc_addr_list(struct net_device *netdev) +{ + struct wx *wx = netdev_priv(netdev); + + if (!netif_running(netdev)) + return 0; + + wx_update_mc_addr_list(wx, netdev); + + return netdev_mc_count(netdev); +} + /** * wx_set_mac - Change the Ethernet Address of the NIC * @netdev: network interface device structure @@ -883,6 +1092,385 @@ void wx_disable_rx(struct wx *wx) } EXPORT_SYMBOL(wx_disable_rx); +static void wx_enable_rx(struct wx *wx) +{ + u32 psrctl; + + /* enable mac receiver */ + wr32m(wx, WX_MAC_RX_CFG, + WX_MAC_RX_CFG_RE, WX_MAC_RX_CFG_RE); + + wr32m(wx, WX_RDB_PB_CTL, + WX_RDB_PB_CTL_RXEN, WX_RDB_PB_CTL_RXEN); + + if (wx->mac.set_lben) { + psrctl = rd32(wx, WX_PSR_CTL); + psrctl |= WX_PSR_CTL_SW_EN; + wr32(wx, WX_PSR_CTL, psrctl); + wx->mac.set_lben = false; + } +} + +/** + * wx_set_rxpba - Initialize Rx packet buffer + * @wx: pointer to private structure + **/ +static void wx_set_rxpba(struct wx *wx) +{ + u32 rxpktsize, txpktsize, txpbthresh; + + rxpktsize = wx->mac.rx_pb_size << WX_RDB_PB_SZ_SHIFT; + wr32(wx, WX_RDB_PB_SZ(0), rxpktsize); + + /* Only support an equally distributed Tx packet buffer strategy. */ + txpktsize = wx->mac.tx_pb_size; + txpbthresh = (txpktsize / 1024) - WX_TXPKT_SIZE_MAX; + wr32(wx, WX_TDB_PB_SZ(0), txpktsize); + wr32(wx, WX_TDM_PB_THRE(0), txpbthresh); +} + +static void wx_configure_port(struct wx *wx) +{ + u32 value, i; + + value = WX_CFG_PORT_CTL_D_VLAN | WX_CFG_PORT_CTL_QINQ; + wr32m(wx, WX_CFG_PORT_CTL, + WX_CFG_PORT_CTL_D_VLAN | + WX_CFG_PORT_CTL_QINQ, + value); + + wr32(wx, WX_CFG_TAG_TPID(0), + ETH_P_8021Q | ETH_P_8021AD << 16); + wx->tpid[0] = ETH_P_8021Q; + wx->tpid[1] = ETH_P_8021AD; + for (i = 1; i < 4; i++) + wr32(wx, WX_CFG_TAG_TPID(i), + ETH_P_8021Q | ETH_P_8021Q << 16); + for (i = 2; i < 8; i++) + wx->tpid[i] = ETH_P_8021Q; +} + +/** + * wx_disable_sec_rx_path - Stops the receive data path + * @wx: pointer to private structure + * + * Stops the receive data path and waits for the HW to internally empty + * the Rx security block + **/ +static int wx_disable_sec_rx_path(struct wx *wx) +{ + u32 secrx; + + wr32m(wx, WX_RSC_CTL, + WX_RSC_CTL_RX_DIS, WX_RSC_CTL_RX_DIS); + + return read_poll_timeout(rd32, secrx, secrx & WX_RSC_ST_RSEC_RDY, + 1000, 40000, false, wx, WX_RSC_ST); +} + +/** + * wx_enable_sec_rx_path - Enables the receive data path + * @wx: pointer to private structure + * + * Enables the receive data path. + **/ +static void wx_enable_sec_rx_path(struct wx *wx) +{ + wr32m(wx, WX_RSC_CTL, WX_RSC_CTL_RX_DIS, 0); + WX_WRITE_FLUSH(wx); +} + +void wx_set_rx_mode(struct net_device *netdev) +{ + struct wx *wx = netdev_priv(netdev); + u32 fctrl, vmolr, vlnctrl; + int count; + + /* Check for Promiscuous and All Multicast modes */ + fctrl = rd32(wx, WX_PSR_CTL); + fctrl &= ~(WX_PSR_CTL_UPE | WX_PSR_CTL_MPE); + vmolr = rd32(wx, WX_PSR_VM_L2CTL(0)); + vmolr &= ~(WX_PSR_VM_L2CTL_UPE | + WX_PSR_VM_L2CTL_MPE | + WX_PSR_VM_L2CTL_ROPE | + WX_PSR_VM_L2CTL_ROMPE); + vlnctrl = rd32(wx, WX_PSR_VLAN_CTL); + vlnctrl &= ~(WX_PSR_VLAN_CTL_VFE | WX_PSR_VLAN_CTL_CFIEN); + + /* set all bits that we expect to always be set */ + fctrl |= WX_PSR_CTL_BAM | WX_PSR_CTL_MFE; + vmolr |= WX_PSR_VM_L2CTL_BAM | + WX_PSR_VM_L2CTL_AUPE | + WX_PSR_VM_L2CTL_VACC; + vlnctrl |= WX_PSR_VLAN_CTL_VFE; + + wx->addr_ctrl.user_set_promisc = false; + if (netdev->flags & IFF_PROMISC) { + wx->addr_ctrl.user_set_promisc = true; + fctrl |= WX_PSR_CTL_UPE | WX_PSR_CTL_MPE; + /* pf don't want packets routing to vf, so clear UPE */ + vmolr |= WX_PSR_VM_L2CTL_MPE; + vlnctrl &= ~WX_PSR_VLAN_CTL_VFE; + } + + if (netdev->flags & IFF_ALLMULTI) { + fctrl |= WX_PSR_CTL_MPE; + vmolr |= WX_PSR_VM_L2CTL_MPE; + } + + if (netdev->features & NETIF_F_RXALL) { + vmolr |= (WX_PSR_VM_L2CTL_UPE | WX_PSR_VM_L2CTL_MPE); + vlnctrl &= ~WX_PSR_VLAN_CTL_VFE; + /* receive bad packets */ + wr32m(wx, WX_RSC_CTL, + WX_RSC_CTL_SAVE_MAC_ERR, + WX_RSC_CTL_SAVE_MAC_ERR); + } else { + vmolr |= WX_PSR_VM_L2CTL_ROPE | WX_PSR_VM_L2CTL_ROMPE; + } + + /* Write addresses to available RAR registers, if there is not + * sufficient space to store all the addresses then enable + * unicast promiscuous mode + */ + count = wx_write_uc_addr_list(netdev, 0); + if (count < 0) { + vmolr &= ~WX_PSR_VM_L2CTL_ROPE; + vmolr |= WX_PSR_VM_L2CTL_UPE; + } + + /* Write addresses to the MTA, if the attempt fails + * then we should just turn on promiscuous mode so + * that we can at least receive multicast traffic + */ + count = wx_write_mc_addr_list(netdev); + if (count < 0) { + vmolr &= ~WX_PSR_VM_L2CTL_ROMPE; + vmolr |= WX_PSR_VM_L2CTL_MPE; + } + + wr32(wx, WX_PSR_VLAN_CTL, vlnctrl); + wr32(wx, WX_PSR_CTL, fctrl); + wr32(wx, WX_PSR_VM_L2CTL(0), vmolr); +} +EXPORT_SYMBOL(wx_set_rx_mode); + +static void wx_set_rx_buffer_len(struct wx *wx) +{ + struct net_device *netdev = wx->netdev; + u32 mhadd, max_frame; + + max_frame = netdev->mtu + ETH_HLEN + ETH_FCS_LEN; + /* adjust max frame to be at least the size of a standard frame */ + if (max_frame < (ETH_FRAME_LEN + ETH_FCS_LEN)) + max_frame = (ETH_FRAME_LEN + ETH_FCS_LEN); + + mhadd = rd32(wx, WX_PSR_MAX_SZ); + if (max_frame != mhadd) + wr32(wx, WX_PSR_MAX_SZ, max_frame); +} + +/* Disable the specified rx queue */ +void wx_disable_rx_queue(struct wx *wx, struct wx_ring *ring) +{ + u8 reg_idx = ring->reg_idx; + u32 rxdctl; + int ret; + + /* write value back with RRCFG.EN bit cleared */ + wr32m(wx, WX_PX_RR_CFG(reg_idx), + WX_PX_RR_CFG_RR_EN, 0); + + /* the hardware may take up to 100us to really disable the rx queue */ + ret = read_poll_timeout(rd32, rxdctl, !(rxdctl & WX_PX_RR_CFG_RR_EN), + 10, 100, true, wx, WX_PX_RR_CFG(reg_idx)); + + if (ret == -ETIMEDOUT) { + /* Just for information */ + wx_err(wx, + "RRCFG.EN on Rx queue %d not cleared within the polling period\n", + reg_idx); + } +} +EXPORT_SYMBOL(wx_disable_rx_queue); + +static void wx_enable_rx_queue(struct wx *wx, struct wx_ring *ring) +{ + u8 reg_idx = ring->reg_idx; + u32 rxdctl; + int ret; + + ret = read_poll_timeout(rd32, rxdctl, rxdctl & WX_PX_RR_CFG_RR_EN, + 1000, 10000, true, wx, WX_PX_RR_CFG(reg_idx)); + + if (ret == -ETIMEDOUT) { + /* Just for information */ + wx_err(wx, + "RRCFG.EN on Rx queue %d not set within the polling period\n", + reg_idx); + } +} + +static void wx_configure_srrctl(struct wx *wx, + struct wx_ring *rx_ring) +{ + u16 reg_idx = rx_ring->reg_idx; + u32 srrctl; + + srrctl = rd32(wx, WX_PX_RR_CFG(reg_idx)); + srrctl &= ~(WX_PX_RR_CFG_RR_HDR_SZ | + WX_PX_RR_CFG_RR_BUF_SZ | + WX_PX_RR_CFG_SPLIT_MODE); + /* configure header buffer length, needed for RSC */ + srrctl |= WX_RXBUFFER_256 << WX_PX_RR_CFG_BHDRSIZE_SHIFT; + + /* configure the packet buffer length */ + srrctl |= WX_RX_BUFSZ >> WX_PX_RR_CFG_BSIZEPKT_SHIFT; + + wr32(wx, WX_PX_RR_CFG(reg_idx), srrctl); +} + +static void wx_configure_tx_ring(struct wx *wx, + struct wx_ring *ring) +{ + u32 txdctl = WX_PX_TR_CFG_ENABLE; + u8 reg_idx = ring->reg_idx; + int ret; + + /* disable queue to avoid issues while updating state */ + wr32(wx, WX_PX_TR_CFG(reg_idx), WX_PX_TR_CFG_SWFLSH); + WX_WRITE_FLUSH(wx); + + /* reset head and tail pointers */ + wr32(wx, WX_PX_TR_RP(reg_idx), 0); + wr32(wx, WX_PX_TR_WP(reg_idx), 0); + ring->tail = wx->hw_addr + WX_PX_TR_WP(reg_idx); + + if (ring->count < WX_MAX_TXD) + txdctl |= ring->count / 128 << WX_PX_TR_CFG_TR_SIZE_SHIFT; + txdctl |= 0x20 << WX_PX_TR_CFG_WTHRESH_SHIFT; + + /* enable queue */ + wr32(wx, WX_PX_TR_CFG(reg_idx), txdctl); + + /* poll to verify queue is enabled */ + ret = read_poll_timeout(rd32, txdctl, txdctl & WX_PX_TR_CFG_ENABLE, + 1000, 10000, true, wx, WX_PX_TR_CFG(reg_idx)); + if (ret == -ETIMEDOUT) + wx_err(wx, "Could not enable Tx Queue %d\n", reg_idx); +} + +static void wx_configure_rx_ring(struct wx *wx, + struct wx_ring *ring) +{ + u16 reg_idx = ring->reg_idx; + u32 rxdctl; + + /* disable queue to avoid issues while updating state */ + rxdctl = rd32(wx, WX_PX_RR_CFG(reg_idx)); + wx_disable_rx_queue(wx, ring); + + if (ring->count == WX_MAX_RXD) + rxdctl |= 0 << WX_PX_RR_CFG_RR_SIZE_SHIFT; + else + rxdctl |= (ring->count / 128) << WX_PX_RR_CFG_RR_SIZE_SHIFT; + + rxdctl |= 0x1 << WX_PX_RR_CFG_RR_THER_SHIFT; + wr32(wx, WX_PX_RR_CFG(reg_idx), rxdctl); + + /* reset head and tail pointers */ + wr32(wx, WX_PX_RR_RP(reg_idx), 0); + wr32(wx, WX_PX_RR_WP(reg_idx), 0); + ring->tail = wx->hw_addr + WX_PX_RR_WP(reg_idx); + + wx_configure_srrctl(wx, ring); + + /* enable receive descriptor ring */ + wr32m(wx, WX_PX_RR_CFG(reg_idx), + WX_PX_RR_CFG_RR_EN, WX_PX_RR_CFG_RR_EN); + + wx_enable_rx_queue(wx, ring); +} + +/** + * wx_configure_tx - Configure Transmit Unit after Reset + * @wx: pointer to private structure + * + * Configure the Tx unit of the MAC after a reset. + **/ +static void wx_configure_tx(struct wx *wx) +{ + u32 i; + + /* TDM_CTL.TE must be before Tx queues are enabled */ + wr32m(wx, WX_TDM_CTL, + WX_TDM_CTL_TE, WX_TDM_CTL_TE); + + /* Setup the HW Tx Head and Tail descriptor pointers */ + for (i = 0; i < wx->num_tx_queues; i++) + wx_configure_tx_ring(wx, wx->tx_ring[i]); + + wr32m(wx, WX_TSC_BUF_AE, WX_TSC_BUF_AE_THR, 0x10); + + if (wx->mac.type == wx_mac_em) + wr32m(wx, WX_TSC_CTL, WX_TSC_CTL_TX_DIS | WX_TSC_CTL_TSEC_DIS, 0x1); + + /* enable mac transmitter */ + wr32m(wx, WX_MAC_TX_CFG, + WX_MAC_TX_CFG_TE, WX_MAC_TX_CFG_TE); +} + +/** + * wx_configure_rx - Configure Receive Unit after Reset + * @wx: pointer to private structure + * + * Configure the Rx unit of the MAC after a reset. + **/ +static void wx_configure_rx(struct wx *wx) +{ + u32 psrtype, i; + int ret; + + wx_disable_rx(wx); + + psrtype = WX_RDB_PL_CFG_L4HDR | + WX_RDB_PL_CFG_L3HDR | + WX_RDB_PL_CFG_L2HDR | + WX_RDB_PL_CFG_TUN_TUNHDR | + WX_RDB_PL_CFG_TUN_TUNHDR; + wr32(wx, WX_RDB_PL_CFG(0), psrtype); + + /* enable hw crc stripping */ + wr32m(wx, WX_RSC_CTL, WX_RSC_CTL_CRC_STRIP, WX_RSC_CTL_CRC_STRIP); + + if (wx->mac.type == wx_mac_sp) { + u32 psrctl; + + /* RSC Setup */ + psrctl = rd32(wx, WX_PSR_CTL); + psrctl |= WX_PSR_CTL_RSC_ACK; /* Disable RSC for ACK packets */ + psrctl |= WX_PSR_CTL_RSC_DIS; + wr32(wx, WX_PSR_CTL, psrctl); + } + + /* set_rx_buffer_len must be called before ring initialization */ + wx_set_rx_buffer_len(wx); + + /* Setup the HW Rx Head and Tail Descriptor Pointers and + * the Base and Length of the Rx Descriptor Ring + */ + for (i = 0; i < wx->num_rx_queues; i++) + wx_configure_rx_ring(wx, wx->rx_ring[i]); + + /* Enable all receives, disable security engine prior to block traffic */ + ret = wx_disable_sec_rx_path(wx); + if (ret < 0) + wx_err(wx, "The register status is abnormal, please check device."); + + wx_enable_rx(wx); + wx_enable_sec_rx_path(wx); +} + static void wx_configure_isb(struct wx *wx) { /* set ISB Address */ @@ -893,6 +1481,15 @@ static void wx_configure_isb(struct wx *wx) void wx_configure(struct wx *wx) { + wx_set_rxpba(wx); + wx_configure_port(wx); + + wx_set_rx_mode(wx->netdev); + + wx_enable_sec_rx_path(wx); + + wx_configure_tx(wx); + wx_configure_rx(wx); wx_configure_isb(wx); } EXPORT_SYMBOL(wx_configure); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.h b/drivers/net/ethernet/wangxun/libwx/wx_hw.h index 60bda129cfa6..44dfd6ea442a 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.h @@ -22,6 +22,8 @@ void wx_mac_set_default_filter(struct wx *wx, u8 *addr); void wx_flush_sw_mac_table(struct wx *wx); int wx_set_mac(struct net_device *netdev, void *p); void wx_disable_rx(struct wx *wx); +void wx_set_rx_mode(struct net_device *netdev); +void wx_disable_rx_queue(struct wx *wx, struct wx_ring *ring); void wx_configure(struct wx *wx); int wx_disable_pcie_master(struct wx *wx); int wx_stop_adapter(struct wx *wx); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index eb2a1918fb12..1f6625f151ac 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -66,6 +66,9 @@ /* port cfg Registers */ #define WX_CFG_PORT_CTL 0x14400 #define WX_CFG_PORT_CTL_DRV_LOAD BIT(3) +#define WX_CFG_PORT_CTL_QINQ BIT(2) +#define WX_CFG_PORT_CTL_D_VLAN BIT(0) /* double vlan*/ +#define WX_CFG_TAG_TPID(_i) (0x14430 + ((_i) * 4)) /* GPIO Registers */ #define WX_GPIO_DR 0x14800 @@ -88,15 +91,25 @@ #define WX_TDM_CTL 0x18000 /* TDM CTL BIT */ #define WX_TDM_CTL_TE BIT(0) /* Transmit Enable */ +#define WX_TDM_PB_THRE(_i) (0x18020 + ((_i) * 4)) /***************************** RDB registers *********************************/ /* receive packet buffer */ #define WX_RDB_PB_CTL 0x19000 #define WX_RDB_PB_CTL_RXEN BIT(31) /* Enable Receiver */ #define WX_RDB_PB_CTL_DISABLED BIT(0) +#define WX_RDB_PB_SZ(_i) (0x19020 + ((_i) * 4)) +#define WX_RDB_PB_SZ_SHIFT 10 /* statistic */ #define WX_RDB_PFCMACDAL 0x19210 #define WX_RDB_PFCMACDAH 0x19214 +/* ring assignment */ +#define WX_RDB_PL_CFG(_i) (0x19300 + ((_i) * 4)) +#define WX_RDB_PL_CFG_L4HDR BIT(1) +#define WX_RDB_PL_CFG_L3HDR BIT(2) +#define WX_RDB_PL_CFG_L2HDR BIT(3) +#define WX_RDB_PL_CFG_TUN_TUNHDR BIT(4) +#define WX_RDB_PL_CFG_TUN_OUTL2HDR BIT(5) /******************************* PSR Registers *******************************/ /* psr control */ @@ -114,10 +127,24 @@ #define WX_PSR_CTL_MO_SHIFT 5 #define WX_PSR_CTL_MO (0x3 << WX_PSR_CTL_MO_SHIFT) #define WX_PSR_CTL_TPE BIT(4) +#define WX_PSR_MAX_SZ 0x15020 +#define WX_PSR_VLAN_CTL 0x15088 +#define WX_PSR_VLAN_CTL_CFIEN BIT(29) /* bit 29 */ +#define WX_PSR_VLAN_CTL_VFE BIT(30) /* bit 30 */ /* mcasst/ucast overflow tbl */ #define WX_PSR_MC_TBL(_i) (0x15200 + ((_i) * 4)) #define WX_PSR_UC_TBL(_i) (0x15400 + ((_i) * 4)) +/* VM L2 contorl */ +#define WX_PSR_VM_L2CTL(_i) (0x15600 + ((_i) * 4)) +#define WX_PSR_VM_L2CTL_UPE BIT(4) /* unicast promiscuous */ +#define WX_PSR_VM_L2CTL_VACC BIT(6) /* accept nomatched vlan */ +#define WX_PSR_VM_L2CTL_AUPE BIT(8) /* accept untagged packets */ +#define WX_PSR_VM_L2CTL_ROMPE BIT(9) /* accept packets in MTA tbl */ +#define WX_PSR_VM_L2CTL_ROPE BIT(10) /* accept packets in UC tbl */ +#define WX_PSR_VM_L2CTL_BAM BIT(11) /* accept broadcast packets */ +#define WX_PSR_VM_L2CTL_MPE BIT(12) /* multicast promiscuous */ + /* Management */ #define WX_PSR_MNG_FLEX_SEL 0x1582C #define WX_PSR_MNG_FLEX_DW_L(_i) (0x15A00 + ((_i) * 16)) @@ -139,6 +166,27 @@ #define WX_PSR_MAC_SWC_IDX 0x16210 #define WX_CLEAR_VMDQ_ALL 0xFFFFFFFFU +/********************************* RSEC **************************************/ +/* general rsec */ +#define WX_RSC_CTL 0x17000 +#define WX_RSC_CTL_SAVE_MAC_ERR BIT(6) +#define WX_RSC_CTL_CRC_STRIP BIT(2) +#define WX_RSC_CTL_RX_DIS BIT(1) +#define WX_RSC_ST 0x17004 +#define WX_RSC_ST_RSEC_RDY BIT(0) + +/****************************** TDB ******************************************/ +#define WX_TDB_PB_SZ(_i) (0x1CC00 + ((_i) * 4)) +#define WX_TXPKT_SIZE_MAX 0xA /* Max Tx Packet size */ + +/****************************** TSEC *****************************************/ +/* Security Control Registers */ +#define WX_TSC_CTL 0x1D000 +#define WX_TSC_CTL_TX_DIS BIT(1) +#define WX_TSC_CTL_TSEC_DIS BIT(0) +#define WX_TSC_BUF_AE 0x1D00C +#define WX_TSC_BUF_AE_THR GENMASK(9, 0) + /************************************** MNG ********************************/ #define WX_MNG_SWFW_SYNC 0x1E008 #define WX_MNG_SWFW_SYNC_SW_MB BIT(2) @@ -192,6 +240,10 @@ #define WX_EM_MAX_EITR 0x00007FFCU /* transmit DMA Registers */ +#define WX_PX_TR_BAL(_i) (0x03000 + ((_i) * 0x40)) +#define WX_PX_TR_BAH(_i) (0x03004 + ((_i) * 0x40)) +#define WX_PX_TR_WP(_i) (0x03008 + ((_i) * 0x40)) +#define WX_PX_TR_RP(_i) (0x0300C + ((_i) * 0x40)) #define WX_PX_TR_CFG(_i) (0x03010 + ((_i) * 0x40)) /* Transmit Config masks */ #define WX_PX_TR_CFG_ENABLE BIT(0) /* Ena specific Tx Queue */ @@ -201,8 +253,22 @@ #define WX_PX_TR_CFG_THRE_SHIFT 8 /* Receive DMA Registers */ +#define WX_PX_RR_BAL(_i) (0x01000 + ((_i) * 0x40)) +#define WX_PX_RR_BAH(_i) (0x01004 + ((_i) * 0x40)) +#define WX_PX_RR_WP(_i) (0x01008 + ((_i) * 0x40)) +#define WX_PX_RR_RP(_i) (0x0100C + ((_i) * 0x40)) #define WX_PX_RR_CFG(_i) (0x01010 + ((_i) * 0x40)) /* PX_RR_CFG bit definitions */ +#define WX_PX_RR_CFG_SPLIT_MODE BIT(26) +#define WX_PX_RR_CFG_RR_THER_SHIFT 16 +#define WX_PX_RR_CFG_RR_HDR_SZ GENMASK(15, 12) +#define WX_PX_RR_CFG_RR_BUF_SZ GENMASK(11, 8) +#define WX_PX_RR_CFG_BHDRSIZE_SHIFT 6 /* 64byte resolution (>> 6) + * + at bit 8 offset (<< 12) + * = (<< 6) + */ +#define WX_PX_RR_CFG_BSIZEPKT_SHIFT 2 /* so many KBs */ +#define WX_PX_RR_CFG_RR_SIZE_SHIFT 1 #define WX_PX_RR_CFG_RR_EN BIT(0) /* Number of 80 microseconds we wait for PCI Express master disable */ @@ -230,6 +296,20 @@ #define WX_MAC_STATE_MODIFIED 0x2 #define WX_MAC_STATE_IN_USE 0x4 +#define WX_MAX_RXD 8192 +#define WX_MAX_TXD 8192 + +/* Supported Rx Buffer Sizes */ +#define WX_RXBUFFER_256 256 /* Used for skb receive header */ +#define WX_RXBUFFER_2K 2048 +#define WX_MAX_RXBUFFER 16384 /* largest size for single descriptor */ + +#if MAX_SKB_FRAGS < 8 +#define WX_RX_BUFSZ ALIGN(WX_MAX_RXBUFFER / MAX_SKB_FRAGS, 1024) +#else +#define WX_RX_BUFSZ WX_RXBUFFER_2K +#endif + #define WX_CFG_PORT_ST 0x14404 /* Host Interface Command Structures */ @@ -307,9 +387,12 @@ struct wx_mac_info { bool set_lben; u8 addr[ETH_ALEN]; u8 perm_addr[ETH_ALEN]; + u32 mta_shadow[128]; s32 mc_filter_type; u32 mcft_size; u32 num_rar_entries; + u32 rx_pb_size; + u32 tx_pb_size; u32 max_tx_queues; u32 max_rx_queues; @@ -364,6 +447,7 @@ struct wx_ring { struct wx_q_vector *q_vector; /* backpointer to host q_vector */ struct net_device *netdev; /* netdev ring belongs to */ struct device *dev; /* device for DMA mapping */ + u8 __iomem *tail; u16 count; /* amount of descriptors */ @@ -420,6 +504,7 @@ struct wx { u16 oem_svid; u16 msg_enable; bool adapter_stopped; + u16 tpid[8]; char eeprom_id[32]; enum wx_reset_type reset_type; From patchwork Fri Feb 3 09:11:30 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127228 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id DE2EDC636D6 for ; Fri, 3 Feb 2023 09:16:03 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S231426AbjBCJQC (ORCPT ); Fri, 3 Feb 2023 04:16:02 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36352 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232576AbjBCJPy (ORCPT ); Fri, 3 Feb 2023 04:15:54 -0500 Received: from smtpbgsg2.qq.com (smtpbgsg2.qq.com [54.254.200.128]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 8BBD3991C5 for ; Fri, 3 Feb 2023 01:15:51 -0800 (PST) X-QQ-mid: bizesmtp68t1675415746tsv7bhh3 Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:45 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: uW0fSitQVIu417VCf4tpoYVtHoWqvWZbiR8qidNqeSXCbAlHivc/MjNH6JWJE N5KA3tEHfkdISDFyIXOMyTb/7T2gXJqEjJILXntHDaVc63TYtHneDYR0URjWPhcdPbQf3wb dPDWRarGuXyLsRvVri+G2mBDsxvd5MkGnsRkZDnISZHjeJyzwhbF2CWpiMQZwaC32I41hBa AZglDZj5M54on++bR4uSqSdFuCc1btw0onP26VVDuZ8exW0z0KucMjSFylCVm+6kfjSmNSh LAa54BMglhxky0SumrMPp45gCjeLrOS3EyPYQR/FvMdkaiVj+3B+3FB9xyA3wnB+XqIknjp xncSP/I/9P+7FKwF5F8A72Jvbsx7Dj9m5G47XDXFDHJHaCwNi/UPIsdQVkeUXJ7K49Nf+2Y ozwgLM4YiMtP48NmZ+ooNQ== X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 05/10] net: libwx: Allocate Rx and Tx resources Date: Fri, 3 Feb 2023 17:11:30 +0800 Message-Id: <20230203091135.3294377-6-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Setup Rx and Tx descriptors for specefic rings. Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/Kconfig | 1 + drivers/net/ethernet/wangxun/libwx/wx_hw.c | 8 + drivers/net/ethernet/wangxun/libwx/wx_lib.c | 303 +++++++++++++++++++ drivers/net/ethernet/wangxun/libwx/wx_lib.h | 2 + drivers/net/ethernet/wangxun/libwx/wx_type.h | 74 +++++ 5 files changed, 388 insertions(+) diff --git a/drivers/net/ethernet/wangxun/Kconfig b/drivers/net/ethernet/wangxun/Kconfig index 0922beac3ec0..c9d88673d306 100644 --- a/drivers/net/ethernet/wangxun/Kconfig +++ b/drivers/net/ethernet/wangxun/Kconfig @@ -18,6 +18,7 @@ if NET_VENDOR_WANGXUN config LIBWX tristate + select PAGE_POOL help Common library for Wangxun(R) Ethernet drivers. diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c index db313b53a6b6..99ff183d23e5 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c @@ -1335,12 +1335,16 @@ static void wx_configure_tx_ring(struct wx *wx, { u32 txdctl = WX_PX_TR_CFG_ENABLE; u8 reg_idx = ring->reg_idx; + u64 tdba = ring->dma; int ret; /* disable queue to avoid issues while updating state */ wr32(wx, WX_PX_TR_CFG(reg_idx), WX_PX_TR_CFG_SWFLSH); WX_WRITE_FLUSH(wx); + wr32(wx, WX_PX_TR_BAL(reg_idx), tdba & DMA_BIT_MASK(32)); + wr32(wx, WX_PX_TR_BAH(reg_idx), upper_32_bits(tdba)); + /* reset head and tail pointers */ wr32(wx, WX_PX_TR_RP(reg_idx), 0); wr32(wx, WX_PX_TR_WP(reg_idx), 0); @@ -1364,12 +1368,16 @@ static void wx_configure_rx_ring(struct wx *wx, struct wx_ring *ring) { u16 reg_idx = ring->reg_idx; + u64 rdba = ring->dma; u32 rxdctl; /* disable queue to avoid issues while updating state */ rxdctl = rd32(wx, WX_PX_RR_CFG(reg_idx)); wx_disable_rx_queue(wx, ring); + wr32(wx, WX_PX_RR_BAL(reg_idx), rdba & DMA_BIT_MASK(32)); + wr32(wx, WX_PX_RR_BAH(reg_idx), upper_32_bits(rdba)); + if (ring->count == WX_MAX_RXD) rxdctl |= 0 << WX_PX_RR_CFG_RR_SIZE_SHIFT; else diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c index 8f8a9e3151c2..db468d936915 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c @@ -2,6 +2,7 @@ /* Copyright (c) 2019 - 2022 Beijing WangXun Technology Co., Ltd. */ #include +#include #include #include @@ -192,6 +193,8 @@ static int wx_alloc_q_vector(struct wx *wx, wx->q_vector[v_idx] = q_vector; q_vector->wx = wx; q_vector->v_idx = v_idx; + if (cpu_online(v_idx)) + q_vector->numa_node = cpu_to_node(v_idx); /* initialize pointer to rings */ ring = q_vector->ring; @@ -606,4 +609,304 @@ void wx_configure_vectors(struct wx *wx) } EXPORT_SYMBOL(wx_configure_vectors); +/** + * wx_free_rx_resources - Free Rx Resources + * @rx_ring: ring to clean the resources from + * + * Free all receive software resources + **/ +static void wx_free_rx_resources(struct wx_ring *rx_ring) +{ + kvfree(rx_ring->rx_buffer_info); + rx_ring->rx_buffer_info = NULL; + + /* if not set, then don't free */ + if (!rx_ring->desc) + return; + + dma_free_coherent(rx_ring->dev, rx_ring->size, + rx_ring->desc, rx_ring->dma); + + rx_ring->desc = NULL; + + if (rx_ring->page_pool) { + page_pool_destroy(rx_ring->page_pool); + rx_ring->page_pool = NULL; + } +} + +/** + * wx_free_all_rx_resources - Free Rx Resources for All Queues + * @wx: pointer to hardware structure + * + * Free all receive software resources + **/ +static void wx_free_all_rx_resources(struct wx *wx) +{ + int i; + + for (i = 0; i < wx->num_rx_queues; i++) + wx_free_rx_resources(wx->rx_ring[i]); +} + +/** + * wx_free_tx_resources - Free Tx Resources per Queue + * @tx_ring: Tx descriptor ring for a specific queue + * + * Free all transmit software resources + **/ +static void wx_free_tx_resources(struct wx_ring *tx_ring) +{ + kvfree(tx_ring->tx_buffer_info); + tx_ring->tx_buffer_info = NULL; + + /* if not set, then don't free */ + if (!tx_ring->desc) + return; + + dma_free_coherent(tx_ring->dev, tx_ring->size, + tx_ring->desc, tx_ring->dma); + tx_ring->desc = NULL; +} + +/** + * wx_free_all_tx_resources - Free Tx Resources for All Queues + * @wx: pointer to hardware structure + * + * Free all transmit software resources + **/ +static void wx_free_all_tx_resources(struct wx *wx) +{ + int i; + + for (i = 0; i < wx->num_tx_queues; i++) + wx_free_tx_resources(wx->tx_ring[i]); +} + +void wx_free_resources(struct wx *wx) +{ + wx_free_isb_resources(wx); + wx_free_all_rx_resources(wx); + wx_free_all_tx_resources(wx); +} +EXPORT_SYMBOL(wx_free_resources); + +static int wx_alloc_page_pool(struct wx_ring *rx_ring) +{ + int ret = 0; + + struct page_pool_params pp_params = { + .flags = PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV, + .order = 0, + .pool_size = rx_ring->size, + .nid = dev_to_node(rx_ring->dev), + .dev = rx_ring->dev, + .dma_dir = DMA_FROM_DEVICE, + .offset = 0, + .max_len = PAGE_SIZE, + }; + + rx_ring->page_pool = page_pool_create(&pp_params); + if (IS_ERR(rx_ring->page_pool)) { + rx_ring->page_pool = NULL; + ret = PTR_ERR(rx_ring->page_pool); + } + + return ret; +} + +/** + * wx_setup_rx_resources - allocate Rx resources (Descriptors) + * @rx_ring: rx descriptor ring (for a specific queue) to setup + * + * Returns 0 on success, negative on failure + **/ +static int wx_setup_rx_resources(struct wx_ring *rx_ring) +{ + struct device *dev = rx_ring->dev; + int orig_node = dev_to_node(dev); + int numa_node = NUMA_NO_NODE; + int size, ret; + + size = sizeof(struct wx_rx_buffer) * rx_ring->count; + + if (rx_ring->q_vector) + numa_node = rx_ring->q_vector->numa_node; + + rx_ring->rx_buffer_info = kvmalloc_node(size, GFP_KERNEL, numa_node); + if (!rx_ring->rx_buffer_info) + rx_ring->rx_buffer_info = kvmalloc(size, GFP_KERNEL); + if (!rx_ring->rx_buffer_info) + goto err; + + /* Round up to nearest 4K */ + rx_ring->size = rx_ring->count * sizeof(union wx_rx_desc); + rx_ring->size = ALIGN(rx_ring->size, 4096); + + set_dev_node(dev, numa_node); + rx_ring->desc = dma_alloc_coherent(dev, rx_ring->size, + &rx_ring->dma, GFP_KERNEL); + if (!rx_ring->desc) { + set_dev_node(dev, orig_node); + rx_ring->desc = dma_alloc_coherent(dev, rx_ring->size, + &rx_ring->dma, GFP_KERNEL); + } + + if (!rx_ring->desc) + goto err; + + ret = wx_alloc_page_pool(rx_ring); + if (ret < 0) { + dev_err(rx_ring->dev, "Page pool creation failed: %d\n", ret); + goto err; + } + + return 0; +err: + kvfree(rx_ring->rx_buffer_info); + rx_ring->rx_buffer_info = NULL; + dev_err(dev, "Unable to allocate memory for the Rx descriptor ring\n"); + return -ENOMEM; +} + +/** + * wx_setup_all_rx_resources - allocate all queues Rx resources + * @wx: pointer to hardware structure + * + * If this function returns with an error, then it's possible one or + * more of the rings is populated (while the rest are not). It is the + * callers duty to clean those orphaned rings. + * + * Return 0 on success, negative on failure + **/ +static int wx_setup_all_rx_resources(struct wx *wx) +{ + int i, err = 0; + + for (i = 0; i < wx->num_rx_queues; i++) { + err = wx_setup_rx_resources(wx->rx_ring[i]); + if (!err) + continue; + + wx_err(wx, "Allocation for Rx Queue %u failed\n", i); + goto err_setup_rx; + } + + return 0; +err_setup_rx: + /* rewind the index freeing the rings as we go */ + while (i--) + wx_free_rx_resources(wx->rx_ring[i]); + return err; +} + +/** + * wx_setup_tx_resources - allocate Tx resources (Descriptors) + * @tx_ring: tx descriptor ring (for a specific queue) to setup + * + * Return 0 on success, negative on failure + **/ +static int wx_setup_tx_resources(struct wx_ring *tx_ring) +{ + struct device *dev = tx_ring->dev; + int orig_node = dev_to_node(dev); + int numa_node = NUMA_NO_NODE; + int size; + + size = sizeof(struct wx_tx_buffer) * tx_ring->count; + + if (tx_ring->q_vector) + numa_node = tx_ring->q_vector->numa_node; + + tx_ring->tx_buffer_info = kvmalloc_node(size, GFP_KERNEL, numa_node); + if (!tx_ring->tx_buffer_info) + tx_ring->tx_buffer_info = kvmalloc(size, GFP_KERNEL); + if (!tx_ring->tx_buffer_info) + goto err; + + /* round up to nearest 4K */ + tx_ring->size = tx_ring->count * sizeof(union wx_tx_desc); + tx_ring->size = ALIGN(tx_ring->size, 4096); + + set_dev_node(dev, numa_node); + tx_ring->desc = dma_alloc_coherent(dev, tx_ring->size, + &tx_ring->dma, GFP_KERNEL); + if (!tx_ring->desc) { + set_dev_node(dev, orig_node); + tx_ring->desc = dma_alloc_coherent(dev, tx_ring->size, + &tx_ring->dma, GFP_KERNEL); + } + + if (!tx_ring->desc) + goto err; + + return 0; + +err: + kvfree(tx_ring->tx_buffer_info); + tx_ring->tx_buffer_info = NULL; + dev_err(dev, "Unable to allocate memory for the Tx descriptor ring\n"); + return -ENOMEM; +} + +/** + * wx_setup_all_tx_resources - allocate all queues Tx resources + * @wx: pointer to private structure + * + * If this function returns with an error, then it's possible one or + * more of the rings is populated (while the rest are not). It is the + * callers duty to clean those orphaned rings. + * + * Return 0 on success, negative on failure + **/ +static int wx_setup_all_tx_resources(struct wx *wx) +{ + int i, err = 0; + + for (i = 0; i < wx->num_tx_queues; i++) { + err = wx_setup_tx_resources(wx->tx_ring[i]); + if (!err) + continue; + + wx_err(wx, "Allocation for Tx Queue %u failed\n", i); + goto err_setup_tx; + } + + return 0; +err_setup_tx: + /* rewind the index freeing the rings as we go */ + while (i--) + wx_free_tx_resources(wx->tx_ring[i]); + return err; +} + +int wx_setup_resources(struct wx *wx) +{ + int err; + + /* allocate transmit descriptors */ + err = wx_setup_all_tx_resources(wx); + if (err) + return err; + + /* allocate receive descriptors */ + err = wx_setup_all_rx_resources(wx); + if (err) + goto err_free_tx; + + err = wx_setup_isb_resources(wx); + if (err) + goto err_free_rx; + + return 0; + +err_free_rx: + wx_free_all_rx_resources(wx); +err_free_tx: + wx_free_all_tx_resources(wx); + + return err; +} +EXPORT_SYMBOL(wx_setup_resources); + MODULE_LICENSE("GPL"); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.h b/drivers/net/ethernet/wangxun/libwx/wx_lib.h index 8ae657155f34..6fa95752fc42 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.h @@ -16,5 +16,7 @@ int wx_setup_isb_resources(struct wx *wx); void wx_free_isb_resources(struct wx *wx); u32 wx_misc_isb(struct wx *wx, enum wx_isb_idx idx); void wx_configure_vectors(struct wx *wx); +void wx_free_resources(struct wx *wx); +int wx_setup_resources(struct wx *wx); #endif /* _NGBE_LIB_H_ */ diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index 1f6625f151ac..6d7e36239e22 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -432,6 +432,71 @@ enum wx_reset_type { WX_GLOBAL_RESET }; +/* Transmit Descriptor */ +union wx_tx_desc { + struct { + __le64 buffer_addr; /* Address of descriptor's data buf */ + __le32 cmd_type_len; + __le32 olinfo_status; + } read; + struct { + __le64 rsvd; /* Reserved */ + __le32 nxtseq_seed; + __le32 status; + } wb; +}; + +/* Receive Descriptor */ +union wx_rx_desc { + struct { + __le64 pkt_addr; /* Packet buffer address */ + __le64 hdr_addr; /* Header buffer address */ + } read; + struct { + struct { + union { + __le32 data; + struct { + __le16 pkt_info; /* RSS, Pkt type */ + __le16 hdr_info; /* Splithdr, hdrlen */ + } hs_rss; + } lo_dword; + union { + __le32 rss; /* RSS Hash */ + struct { + __le16 ip_id; /* IP id */ + __le16 csum; /* Packet Checksum */ + } csum_ip; + } hi_dword; + } lower; + struct { + __le32 status_error; /* ext status/error */ + __le16 length; /* Packet length */ + __le16 vlan; /* VLAN tag */ + } upper; + } wb; /* writeback */ +}; + +/* wrapper around a pointer to a socket buffer, + * so a DMA handle can be stored along with the buffer + */ +struct wx_tx_buffer { + union wx_tx_desc *next_to_watch; + struct sk_buff *skb; + unsigned int bytecount; + unsigned short gso_segs; + DEFINE_DMA_UNMAP_ADDR(dma); + DEFINE_DMA_UNMAP_LEN(len); +}; + +struct wx_rx_buffer { + struct sk_buff *skb; + dma_addr_t dma; + dma_addr_t page_dma; + struct page *page; + unsigned int page_offset; +}; + /* iterator for handling rings in ring container */ #define wx_for_each_ring(posm, headm) \ for (posm = (headm).ring; posm; posm = posm->next) @@ -447,7 +512,15 @@ struct wx_ring { struct wx_q_vector *q_vector; /* backpointer to host q_vector */ struct net_device *netdev; /* netdev ring belongs to */ struct device *dev; /* device for DMA mapping */ + struct page_pool *page_pool; + void *desc; /* descriptor ring memory */ + union { + struct wx_tx_buffer *tx_buffer_info; + struct wx_rx_buffer *rx_buffer_info; + }; u8 __iomem *tail; + dma_addr_t dma; /* phys. address of descriptor ring */ + unsigned int size; /* length in bytes */ u16 count; /* amount of descriptors */ @@ -462,6 +535,7 @@ struct wx_ring { struct wx_q_vector { struct wx *wx; int cpu; /* CPU for DCA */ + int numa_node; u16 v_idx; /* index of q_vector within array, also used for * finding the bit in EICR and friends that * represents the vector for this ring From patchwork Fri Feb 3 09:11:31 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127229 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 56764C636CC for ; Fri, 3 Feb 2023 09:16:14 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232646AbjBCJQM (ORCPT ); Fri, 3 Feb 2023 04:16:12 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36570 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232601AbjBCJQB (ORCPT ); Fri, 3 Feb 2023 04:16:01 -0500 Received: from smtpbguseast1.qq.com (smtpbguseast1.qq.com [54.204.34.129]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 209C7991C0 for ; Fri, 3 Feb 2023 01:15:54 -0800 (PST) X-QQ-mid: bizesmtp68t1675415749tc15eikw Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:48 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: Lc4bMlOoZUHUOeAgq4eW7hFLJM50SJvLoW7iL+eX1tWiICL8F1YYxv+XV1k2B i/hAHtiIyZZn4XdH9H5BAwaY+PiUEe4mQjn94l1smp6YDejilfYDCgdhgXMBf73X4xoZ+5y 7SD3jP1pkHPHUh7jlwVtIu9QZXss7AGNbt3YUWEp44dd7uuwivlXrTPqJsURcHqfaylXGY/ GCDwtip5GWaA1etZ1V4UrzIRXF8oOKHxCKovueVBi70qYWSiARGTIY0Jrk5cSiZD/taTEWA lal20W10WRs2V5TS+JDNE6v4VwysBZBhk7tFQeJ5Vi5Zr9TGKgwI0huSpEXKq3CrK7S1kQ5 mHvpRO/f4XJi9aUYBrJKvyT5xNsD5466ooonZ1SM2iO8v7bNtLVYvpF0m74WbC9TpsGp19K 7fPOwl4wPkg= X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 06/10] net: txgbe: Setup Rx and Tx ring Date: Fri, 3 Feb 2023 17:11:31 +0800 Message-Id: <20230203091135.3294377-7-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Improve the configuration of Rx and Tx ring, set Rx flags and implement ndo_set_rx_mode ops. Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/libwx/wx_type.h | 1 + .../net/ethernet/wangxun/txgbe/txgbe_main.c | 43 +++++++++++++++++-- .../net/ethernet/wangxun/txgbe/txgbe_type.h | 3 ++ 3 files changed, 44 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index 6d7e36239e22..479bc4fd2bd4 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -200,6 +200,7 @@ #define WX_MAC_TX_CFG 0x11000 #define WX_MAC_TX_CFG_TE BIT(0) #define WX_MAC_TX_CFG_SPEED_MASK GENMASK(30, 29) +#define WX_MAC_TX_CFG_SPEED_10G FIELD_PREP(WX_MAC_TX_CFG_SPEED_MASK, 0) #define WX_MAC_TX_CFG_SPEED_1G FIELD_PREP(WX_MAC_TX_CFG_SPEED_MASK, 3) #define WX_MAC_RX_CFG 0x11004 #define WX_MAC_RX_CFG_RE BIT(0) diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c index 48cca2fb54c7..3b50acb09699 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c @@ -218,6 +218,8 @@ static int txgbe_request_irq(struct wx *wx) static void txgbe_up_complete(struct wx *wx) { + u32 reg; + wx_control_hw(wx, true); wx_configure_vectors(wx); @@ -225,6 +227,15 @@ static void txgbe_up_complete(struct wx *wx) rd32(wx, WX_PX_IC); rd32(wx, WX_PX_MISC_IC); txgbe_irq_enable(wx, true); + + /* Configure MAC Rx and Tx when link is up */ + reg = rd32(wx, WX_MAC_RX_CFG); + wr32(wx, WX_MAC_RX_CFG, reg); + wr32(wx, WX_MAC_PKT_FLT, WX_MAC_PKT_FLT_PR); + reg = rd32(wx, WX_MAC_WDG_TIMEOUT); + wr32(wx, WX_MAC_WDG_TIMEOUT, reg); + reg = rd32(wx, WX_MAC_TX_CFG); + wr32(wx, WX_MAC_TX_CFG, (reg & ~WX_MAC_TX_CFG_SPEED_MASK) | WX_MAC_TX_CFG_SPEED_10G); } static void txgbe_reset(struct wx *wx) @@ -246,11 +257,17 @@ static void txgbe_reset(struct wx *wx) static void txgbe_disable_device(struct wx *wx) { struct net_device *netdev = wx->netdev; + u32 i; wx_disable_pcie_master(wx); /* disable receives */ wx_disable_rx(wx); + /* disable all enabled rx queues */ + for (i = 0; i < wx->num_rx_queues; i++) + /* this call also flushes the previous write */ + wx_disable_rx_queue(wx, wx->rx_ring[i]); + netif_carrier_off(netdev); netif_tx_disable(netdev); @@ -268,6 +285,13 @@ static void txgbe_disable_device(struct wx *wx) wr32m(wx, WX_MAC_TX_CFG, WX_MAC_TX_CFG_TE, 0); } + /* disable transmits in the hardware now that interrupts are off */ + for (i = 0; i < wx->num_tx_queues; i++) { + u8 reg_idx = wx->tx_ring[i]->reg_idx; + + wr32(wx, WX_PX_TR_CFG(reg_idx), WX_PX_TR_CFG_SWFLSH); + } + /* Disable the Tx DMA engine */ wr32m(wx, WX_TDM_CTL, WX_TDM_CTL_TE, 0); } @@ -291,6 +315,8 @@ static int txgbe_sw_init(struct wx *wx) wx->mac.max_tx_queues = TXGBE_SP_MAX_TX_QUEUES; wx->mac.max_rx_queues = TXGBE_SP_MAX_RX_QUEUES; wx->mac.mcft_size = TXGBE_SP_MC_TBL_SIZE; + wx->mac.rx_pb_size = TXGBE_SP_RX_PB_SIZE; + wx->mac.tx_pb_size = TXGBE_SP_TDB_PB_SZ; /* PCI config space info */ err = wx_sw_init(wx); @@ -345,7 +371,7 @@ static int txgbe_open(struct net_device *netdev) struct wx *wx = netdev_priv(netdev); int err; - err = wx_setup_isb_resources(wx); + err = wx_setup_resources(wx); if (err) goto err_reset; @@ -379,7 +405,7 @@ static void txgbe_close_suspend(struct wx *wx) txgbe_disable_device(wx); wx_free_irq(wx); - wx_free_isb_resources(wx); + wx_free_resources(wx); } /** @@ -399,7 +425,7 @@ static int txgbe_close(struct net_device *netdev) txgbe_down(wx); wx_free_irq(wx); - wx_free_isb_resources(wx); + wx_free_resources(wx); wx_control_hw(wx, false); return 0; @@ -445,6 +471,7 @@ static const struct net_device_ops txgbe_netdev_ops = { .ndo_open = txgbe_open, .ndo_stop = txgbe_close, .ndo_start_xmit = txgbe_xmit_frame, + .ndo_set_rx_mode = wx_set_rx_mode, .ndo_validate_addr = eth_validate_addr, .ndo_set_mac_address = wx_set_mac, }; @@ -549,6 +576,16 @@ static int txgbe_probe(struct pci_dev *pdev, } netdev->features |= NETIF_F_HIGHDMA; + netdev->features = NETIF_F_SG; + + /* copy netdev features into list of user selectable features */ + netdev->hw_features |= netdev->features | NETIF_F_RXALL; + + netdev->priv_flags |= IFF_UNICAST_FLT; + netdev->priv_flags |= IFF_SUPP_NOFCS; + + netdev->min_mtu = ETH_MIN_MTU; + netdev->max_mtu = TXGBE_MAX_JUMBO_FRAME_SIZE - (ETH_HLEN + ETH_FCS_LEN); /* make sure the EEPROM is good */ err = txgbe_validate_eeprom_checksum(wx, NULL); diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h b/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h index ec28cf60c91d..563ea51deca6 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_type.h @@ -77,6 +77,9 @@ #define TXGBE_SP_MAX_RX_QUEUES 128 #define TXGBE_SP_RAR_ENTRIES 128 #define TXGBE_SP_MC_TBL_SIZE 128 +#define TXGBE_SP_RX_PB_SIZE 512 +#define TXGBE_SP_TDB_PB_SZ (160 * 1024) /* 160KB Packet Buffer */ +#define TXGBE_MAX_JUMBO_FRAME_SIZE 9432 /* max payload 9414 */ /* TX/RX descriptor defines */ #define TXGBE_DEFAULT_TXD 512 From patchwork Fri Feb 3 09:11:32 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127230 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id B317CC61DA4 for ; Fri, 3 Feb 2023 09:16:29 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232726AbjBCJQ2 (ORCPT ); Fri, 3 Feb 2023 04:16:28 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36860 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232603AbjBCJQN (ORCPT ); Fri, 3 Feb 2023 04:16:13 -0500 Received: from smtpbgeu2.qq.com (smtpbgeu2.qq.com [18.194.254.142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 96307991D8 for ; Fri, 3 Feb 2023 01:15:58 -0800 (PST) X-QQ-mid: bizesmtp68t1675415751tuyp8e0r Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:51 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: iDzLjIm7mlYS/rCPdhYXkJ64pUQh0ScLuSJnJepAKMWbXUoo8Tdp2/DCQ02sd qJEgz85P5+DyR8BIRzR3dL7TnX9zCPgmzVA12I8SMI2lnkKMW825Il81QKdSW0wpAkONLk2 vATJ60+D2qwLmW6eHVWw2K8cXzYg3sJD78Yr1OusQ9IKTqLFldjQ8+pVA8TSNIkxoq1qx9v BRACGDk+V9dxMzJoNZvasANyosy75yhSpY/aMtpmH49Tv7eLi2BGW9BDECeHvgTTfPsZZq8 zlSW6LD7XXorvx070m/COJ4Wt6NSlfIyJmb4qtF1pAOhi+s6UP0Wj+r6flVDLIimj6YbRGg VJg/eZStozflifBLMqS8Ug4aGdd2WkjYEZumEUUBKBUfFs8Wqwh6QOXJRuG+FqfsEUP9C4L tri5fodtuXQ7ylAxV6xQ9Q== X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 07/10] net: libwx: Support to receive packets in NAPI Date: Fri, 3 Feb 2023 17:11:32 +0800 Message-Id: <20230203091135.3294377-8-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Clean all queues associated with a q_vector, to simple receive packets without hardware features. Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/libwx/wx_hw.c | 11 + drivers/net/ethernet/wangxun/libwx/wx_lib.c | 655 ++++++++++++++++++- drivers/net/ethernet/wangxun/libwx/wx_lib.h | 7 + drivers/net/ethernet/wangxun/libwx/wx_type.h | 34 + 4 files changed, 706 insertions(+), 1 deletion(-) diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c index 99ff183d23e5..f5ad5f991ab7 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c @@ -8,6 +8,7 @@ #include #include "wx_type.h" +#include "wx_lib.h" #include "wx_hw.h" static void wx_intr_disable(struct wx *wx, u64 qmask) @@ -1368,6 +1369,7 @@ static void wx_configure_rx_ring(struct wx *wx, struct wx_ring *ring) { u16 reg_idx = ring->reg_idx; + union wx_rx_desc *rx_desc; u64 rdba = ring->dma; u32 rxdctl; @@ -1393,11 +1395,20 @@ static void wx_configure_rx_ring(struct wx *wx, wx_configure_srrctl(wx, ring); + /* initialize rx_buffer_info */ + memset(ring->rx_buffer_info, 0, + sizeof(struct wx_rx_buffer) * ring->count); + + /* initialize Rx descriptor 0 */ + rx_desc = WX_RX_DESC(ring, 0); + rx_desc->wb.upper.length = 0; + /* enable receive descriptor ring */ wr32m(wx, WX_PX_RR_CFG(reg_idx), WX_PX_RR_CFG_RR_EN, WX_PX_RR_CFG_RR_EN); wx_enable_rx_queue(wx, ring); + wx_alloc_rx_buffers(ring, wx_desc_unused(ring)); } /** diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c index db468d936915..ea6a4207f503 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c @@ -8,6 +8,501 @@ #include "wx_type.h" #include "wx_lib.h" +#include "wx_hw.h" + +/* wx_test_staterr - tests bits in Rx descriptor status and error fields */ +static __le32 wx_test_staterr(union wx_rx_desc *rx_desc, + const u32 stat_err_bits) +{ + return rx_desc->wb.upper.status_error & cpu_to_le32(stat_err_bits); +} + +static bool wx_can_reuse_rx_page(struct wx_rx_buffer *rx_buffer, + int rx_buffer_pgcnt) +{ + unsigned int pagecnt_bias = rx_buffer->pagecnt_bias; + struct page *page = rx_buffer->page; + + /* avoid re-using remote and pfmemalloc pages */ + if (!dev_page_is_reusable(page)) + return false; + +#if (PAGE_SIZE < 8192) + /* if we are only owner of page we can reuse it */ + if (unlikely((rx_buffer_pgcnt - pagecnt_bias) > 1)) + return false; +#endif + + /* If we have drained the page fragment pool we need to update + * the pagecnt_bias and page count so that we fully restock the + * number of references the driver holds. + */ + if (unlikely(pagecnt_bias == 1)) { + page_ref_add(page, USHRT_MAX - 1); + rx_buffer->pagecnt_bias = USHRT_MAX; + } + + return true; +} + +/** + * wx_reuse_rx_page - page flip buffer and store it back on the ring + * @rx_ring: rx descriptor ring to store buffers on + * @old_buff: donor buffer to have page reused + * + * Synchronizes page for reuse by the adapter + **/ +static void wx_reuse_rx_page(struct wx_ring *rx_ring, + struct wx_rx_buffer *old_buff) +{ + u16 nta = rx_ring->next_to_alloc; + struct wx_rx_buffer *new_buff; + + new_buff = &rx_ring->rx_buffer_info[nta]; + + /* update, and store next to alloc */ + nta++; + rx_ring->next_to_alloc = (nta < rx_ring->count) ? nta : 0; + + /* transfer page from old buffer to new buffer */ + new_buff->page = old_buff->page; + new_buff->page_dma = old_buff->page_dma; + new_buff->page_offset = old_buff->page_offset; + new_buff->pagecnt_bias = old_buff->pagecnt_bias; +} + +static void wx_dma_sync_frag(struct wx_ring *rx_ring, + struct wx_rx_buffer *rx_buffer) +{ + struct sk_buff *skb = rx_buffer->skb; + skb_frag_t *frag = &skb_shinfo(skb)->frags[0]; + + dma_sync_single_range_for_cpu(rx_ring->dev, + WX_CB(skb)->dma, + skb_frag_off(frag), + skb_frag_size(frag), + DMA_FROM_DEVICE); + + /* If the page was released, just unmap it. */ + if (unlikely(WX_CB(skb)->page_released)) + page_pool_put_full_page(rx_ring->page_pool, rx_buffer->page, false); +} + +static struct wx_rx_buffer *wx_get_rx_buffer(struct wx_ring *rx_ring, + union wx_rx_desc *rx_desc, + struct sk_buff **skb, + int *rx_buffer_pgcnt) +{ + struct wx_rx_buffer *rx_buffer; + unsigned int size; + + rx_buffer = &rx_ring->rx_buffer_info[rx_ring->next_to_clean]; + size = le16_to_cpu(rx_desc->wb.upper.length); + +#if (PAGE_SIZE < 8192) + *rx_buffer_pgcnt = page_count(rx_buffer->page); +#else + *rx_buffer_pgcnt = 0; +#endif + + prefetchw(rx_buffer->page); + *skb = rx_buffer->skb; + + /* Delay unmapping of the first packet. It carries the header + * information, HW may still access the header after the writeback. + * Only unmap it when EOP is reached + */ + if (!wx_test_staterr(rx_desc, WX_RXD_STAT_EOP)) { + if (!*skb) + goto skip_sync; + } else { + if (*skb) + wx_dma_sync_frag(rx_ring, rx_buffer); + } + + /* we are reusing so sync this buffer for CPU use */ + dma_sync_single_range_for_cpu(rx_ring->dev, + rx_buffer->dma, + rx_buffer->page_offset, + size, + DMA_FROM_DEVICE); +skip_sync: + rx_buffer->pagecnt_bias--; + + return rx_buffer; +} + +static void wx_put_rx_buffer(struct wx_ring *rx_ring, + struct wx_rx_buffer *rx_buffer, + struct sk_buff *skb, + int rx_buffer_pgcnt) +{ + if (wx_can_reuse_rx_page(rx_buffer, rx_buffer_pgcnt)) { + /* hand second half of page back to the ring */ + wx_reuse_rx_page(rx_ring, rx_buffer); + } else { + if (!IS_ERR(skb) && WX_CB(skb)->dma == rx_buffer->dma) + /* the page has been released from the ring */ + WX_CB(skb)->page_released = true; + else + page_pool_put_full_page(rx_ring->page_pool, rx_buffer->page, false); + + __page_frag_cache_drain(rx_buffer->page, + rx_buffer->pagecnt_bias); + } + + /* clear contents of rx_buffer */ + rx_buffer->page = NULL; + rx_buffer->skb = NULL; +} + +static struct sk_buff *wx_build_skb(struct wx_ring *rx_ring, + struct wx_rx_buffer *rx_buffer, + union wx_rx_desc *rx_desc) +{ + unsigned int size = le16_to_cpu(rx_desc->wb.upper.length); +#if (PAGE_SIZE < 8192) + unsigned int truesize = WX_RX_BUFSZ; +#else + unsigned int truesize = ALIGN(size, L1_CACHE_BYTES); +#endif + struct sk_buff *skb = rx_buffer->skb; + + if (!skb) { + void *page_addr = page_address(rx_buffer->page) + + rx_buffer->page_offset; + + /* prefetch first cache line of first page */ + prefetch(page_addr); +#if L1_CACHE_BYTES < 128 + prefetch(page_addr + L1_CACHE_BYTES); +#endif + + /* allocate a skb to store the frags */ + skb = napi_alloc_skb(&rx_ring->q_vector->napi, WX_RXBUFFER_256); + if (unlikely(!skb)) + return NULL; + + /* we will be copying header into skb->data in + * pskb_may_pull so it is in our interest to prefetch + * it now to avoid a possible cache miss + */ + prefetchw(skb->data); + + if (size <= WX_RXBUFFER_256) { + memcpy(__skb_put(skb, size), page_addr, + ALIGN(size, sizeof(long))); + rx_buffer->pagecnt_bias++; + + return skb; + } + + if (!wx_test_staterr(rx_desc, WX_RXD_STAT_EOP)) + WX_CB(skb)->dma = rx_buffer->dma; + + skb_add_rx_frag(skb, 0, rx_buffer->page, + rx_buffer->page_offset, + size, truesize); + goto out; + + } else { + skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_buffer->page, + rx_buffer->page_offset, size, truesize); + } + +out: +#if (PAGE_SIZE < 8192) + /* flip page offset to other buffer */ + rx_buffer->page_offset ^= truesize; +#else + /* move offset up to the next cache line */ + rx_buffer->page_offset += truesize; +#endif + + return skb; +} + +static bool wx_alloc_mapped_page(struct wx_ring *rx_ring, + struct wx_rx_buffer *bi) +{ + struct page *page = bi->page; + dma_addr_t dma; + + /* since we are recycling buffers we should seldom need to alloc */ + if (likely(page)) + return true; + + page = page_pool_dev_alloc_pages(rx_ring->page_pool); + WARN_ON(!page); + dma = page_pool_get_dma_addr(page); + + bi->page_dma = dma; + bi->page = page; + bi->page_offset = 0; + page_ref_add(page, USHRT_MAX - 1); + bi->pagecnt_bias = USHRT_MAX; + + return true; +} + +/** + * wx_alloc_rx_buffers - Replace used receive buffers + * @rx_ring: ring to place buffers on + * @cleaned_count: number of buffers to replace + **/ +void wx_alloc_rx_buffers(struct wx_ring *rx_ring, u16 cleaned_count) +{ + u16 i = rx_ring->next_to_use; + union wx_rx_desc *rx_desc; + struct wx_rx_buffer *bi; + + /* nothing to do */ + if (!cleaned_count) + return; + + rx_desc = WX_RX_DESC(rx_ring, i); + bi = &rx_ring->rx_buffer_info[i]; + i -= rx_ring->count; + + do { + if (!wx_alloc_mapped_page(rx_ring, bi)) + break; + + /* sync the buffer for use by the device */ + dma_sync_single_range_for_device(rx_ring->dev, bi->dma, + bi->page_offset, + WX_RX_BUFSZ, + DMA_FROM_DEVICE); + + rx_desc->read.pkt_addr = + cpu_to_le64(bi->page_dma + bi->page_offset); + + rx_desc++; + bi++; + i++; + if (unlikely(!i)) { + rx_desc = WX_RX_DESC(rx_ring, 0); + bi = rx_ring->rx_buffer_info; + i -= rx_ring->count; + } + + /* clear the status bits for the next_to_use descriptor */ + rx_desc->wb.upper.status_error = 0; + + cleaned_count--; + } while (cleaned_count); + + i += rx_ring->count; + + if (rx_ring->next_to_use != i) { + rx_ring->next_to_use = i; + /* update next to alloc since we have filled the ring */ + rx_ring->next_to_alloc = i; + + /* Force memory writes to complete before letting h/w + * know there are new descriptors to fetch. (Only + * applicable for weak-ordered memory model archs, + * such as IA-64). + */ + wmb(); + writel(i, rx_ring->tail); + } +} + +u16 wx_desc_unused(struct wx_ring *ring) +{ + u16 ntc = ring->next_to_clean; + u16 ntu = ring->next_to_use; + + return ((ntc > ntu) ? 0 : ring->count) + ntc - ntu - 1; +} + +/** + * wx_is_non_eop - process handling of non-EOP buffers + * @rx_ring: Rx ring being processed + * @rx_desc: Rx descriptor for current buffer + * @skb: Current socket buffer containing buffer in progress + * + * This function updates next to clean. If the buffer is an EOP buffer + * this function exits returning false, otherwise it will place the + * sk_buff in the next buffer to be chained and return true indicating + * that this is in fact a non-EOP buffer. + **/ +static bool wx_is_non_eop(struct wx_ring *rx_ring, + union wx_rx_desc *rx_desc, + struct sk_buff *skb) +{ + u32 ntc = rx_ring->next_to_clean + 1; + + /* fetch, update, and store next to clean */ + ntc = (ntc < rx_ring->count) ? ntc : 0; + rx_ring->next_to_clean = ntc; + + prefetch(WX_RX_DESC(rx_ring, ntc)); + + /* if we are the last buffer then there is nothing else to do */ + if (likely(wx_test_staterr(rx_desc, WX_RXD_STAT_EOP))) + return false; + + rx_ring->rx_buffer_info[ntc].skb = skb; + + return true; +} + +static void wx_pull_tail(struct sk_buff *skb) +{ + skb_frag_t *frag = &skb_shinfo(skb)->frags[0]; + unsigned int pull_len; + unsigned char *va; + + /* it is valid to use page_address instead of kmap since we are + * working with pages allocated out of the lomem pool per + * alloc_page(GFP_ATOMIC) + */ + va = skb_frag_address(frag); + + /* we need the header to contain the greater of either ETH_HLEN or + * 60 bytes if the skb->len is less than 60 for skb_pad. + */ + pull_len = eth_get_headlen(skb->dev, va, WX_RXBUFFER_256); + + /* align pull length to size of long to optimize memcpy performance */ + skb_copy_to_linear_data(skb, va, ALIGN(pull_len, sizeof(long))); + + /* update all of the pointers */ + skb_frag_size_sub(frag, pull_len); + skb_frag_off_add(frag, pull_len); + skb->data_len -= pull_len; + skb->tail += pull_len; +} + +/** + * wx_cleanup_headers - Correct corrupted or empty headers + * @rx_ring: rx descriptor ring packet is being transacted on + * @rx_desc: pointer to the EOP Rx descriptor + * @skb: pointer to current skb being fixed + * + * Check for corrupted packet headers caused by senders on the local L2 + * embedded NIC switch not setting up their Tx Descriptors right. These + * should be very rare. + * + * Also address the case where we are pulling data in on pages only + * and as such no data is present in the skb header. + * + * In addition if skb is not at least 60 bytes we need to pad it so that + * it is large enough to qualify as a valid Ethernet frame. + * + * Returns true if an error was encountered and skb was freed. + **/ +static bool wx_cleanup_headers(struct wx_ring *rx_ring, + union wx_rx_desc *rx_desc, + struct sk_buff *skb) +{ + struct net_device *netdev = rx_ring->netdev; + + /* verify that the packet does not have any known errors */ + if (!netdev || + unlikely(wx_test_staterr(rx_desc, WX_RXD_ERR_RXE) && + !(netdev->features & NETIF_F_RXALL))) { + dev_kfree_skb_any(skb); + return true; + } + + /* place header in linear portion of buffer */ + if (!skb_headlen(skb)) + wx_pull_tail(skb); + + /* if eth_skb_pad returns an error the skb was freed */ + if (eth_skb_pad(skb)) + return true; + + return false; +} + +/** + * wx_clean_rx_irq - Clean completed descriptors from Rx ring - bounce buf + * @q_vector: structure containing interrupt and ring information + * @rx_ring: rx descriptor ring to transact packets on + * @budget: Total limit on number of packets to process + * + * This function provides a "bounce buffer" approach to Rx interrupt + * processing. The advantage to this is that on systems that have + * expensive overhead for IOMMU access this provides a means of avoiding + * it by maintaining the mapping of the page to the system. + * + * Returns amount of work completed. + **/ +static int wx_clean_rx_irq(struct wx_q_vector *q_vector, + struct wx_ring *rx_ring, + int budget) +{ + unsigned int total_rx_bytes = 0, total_rx_packets = 0; + u16 cleaned_count = wx_desc_unused(rx_ring); + + do { + struct wx_rx_buffer *rx_buffer; + union wx_rx_desc *rx_desc; + struct sk_buff *skb; + int rx_buffer_pgcnt; + + /* return some buffers to hardware, one at a time is too slow */ + if (cleaned_count >= WX_RX_BUFFER_WRITE) { + wx_alloc_rx_buffers(rx_ring, cleaned_count); + cleaned_count = 0; + } + + rx_desc = WX_RX_DESC(rx_ring, rx_ring->next_to_clean); + if (!wx_test_staterr(rx_desc, WX_RXD_STAT_DD)) + break; + + /* This memory barrier is needed to keep us from reading + * any other fields out of the rx_desc until we know the + * descriptor has been written back + */ + dma_rmb(); + + rx_buffer = wx_get_rx_buffer(rx_ring, rx_desc, &skb, &rx_buffer_pgcnt); + + /* retrieve a buffer from the ring */ + skb = wx_build_skb(rx_ring, rx_buffer, rx_desc); + + /* exit if we failed to retrieve a buffer */ + if (!skb) { + rx_buffer->pagecnt_bias++; + break; + } + + wx_put_rx_buffer(rx_ring, rx_buffer, skb, rx_buffer_pgcnt); + cleaned_count++; + + /* place incomplete frames back on ring for completion */ + if (wx_is_non_eop(rx_ring, rx_desc, skb)) + continue; + + /* verify the packet layout is correct */ + if (wx_cleanup_headers(rx_ring, rx_desc, skb)) + continue; + + /* probably a little skewed due to removing CRC */ + total_rx_bytes += skb->len; + + skb_record_rx_queue(skb, rx_ring->queue_index); + skb->protocol = eth_type_trans(skb, rx_ring->netdev); + napi_gro_receive(&q_vector->napi, skb); + + /* update budget accounting */ + total_rx_packets++; + } while (likely(total_rx_packets < budget)); + + u64_stats_update_begin(&rx_ring->syncp); + rx_ring->stats.packets += total_rx_packets; + rx_ring->stats.bytes += total_rx_bytes; + u64_stats_update_end(&rx_ring->syncp); + q_vector->rx.total_packets += total_rx_packets; + q_vector->rx.total_bytes += total_rx_bytes; + + return total_rx_packets; +} /** * wx_poll - NAPI polling RX/TX cleanup routine @@ -18,8 +513,68 @@ **/ static int wx_poll(struct napi_struct *napi, int budget) { - return 0; + struct wx_q_vector *q_vector = container_of(napi, struct wx_q_vector, napi); + int per_ring_budget, work_done = 0; + struct wx *wx = q_vector->wx; + bool clean_complete = true; + struct wx_ring *ring; + + /* Exit if we are called by netpoll */ + if (budget <= 0) + return budget; + + /* attempt to distribute budget to each queue fairly, but don't allow + * the budget to go below 1 because we'll exit polling + */ + if (q_vector->rx.count > 1) + per_ring_budget = max(budget / q_vector->rx.count, 1); + else + per_ring_budget = budget; + + wx_for_each_ring(ring, q_vector->rx) { + int cleaned = wx_clean_rx_irq(q_vector, ring, per_ring_budget); + + work_done += cleaned; + if (cleaned >= per_ring_budget) + clean_complete = false; + } + + /* If all work not completed, return budget and keep polling */ + if (!clean_complete) + return budget; + + /* all work done, exit the polling mode */ + if (likely(napi_complete_done(napi, work_done))) { + if (netif_running(wx->netdev)) + wx_intr_enable(wx, WX_INTR_Q(q_vector->v_idx)); + }; + + return min(work_done, budget - 1); +} + +void wx_napi_enable_all(struct wx *wx) +{ + struct wx_q_vector *q_vector; + int q_idx; + + for (q_idx = 0; q_idx < wx->num_q_vectors; q_idx++) { + q_vector = wx->q_vector[q_idx]; + napi_enable(&q_vector->napi); + } +} +EXPORT_SYMBOL(wx_napi_enable_all); + +void wx_napi_disable_all(struct wx *wx) +{ + struct wx_q_vector *q_vector; + int q_idx; + + for (q_idx = 0; q_idx < wx->num_q_vectors; q_idx++) { + q_vector = wx->q_vector[q_idx]; + napi_disable(&q_vector->napi); + } } +EXPORT_SYMBOL(wx_napi_disable_all); /** * wx_set_rss_queues: Allocate queues for RSS @@ -609,6 +1164,68 @@ void wx_configure_vectors(struct wx *wx) } EXPORT_SYMBOL(wx_configure_vectors); +/** + * wx_clean_rx_ring - Free Rx Buffers per Queue + * @rx_ring: ring to free buffers from + **/ +static void wx_clean_rx_ring(struct wx_ring *rx_ring) +{ + struct wx_rx_buffer *rx_buffer; + u16 i = rx_ring->next_to_clean; + + rx_buffer = &rx_ring->rx_buffer_info[i]; + + /* Free all the Rx ring sk_buffs */ + while (i != rx_ring->next_to_alloc) { + if (rx_buffer->skb) { + struct sk_buff *skb = rx_buffer->skb; + + if (WX_CB(skb)->page_released) + page_pool_put_full_page(rx_ring->page_pool, rx_buffer->page, false); + + dev_kfree_skb(skb); + } + + /* Invalidate cache lines that may have been written to by + * device so that we avoid corrupting memory. + */ + dma_sync_single_range_for_cpu(rx_ring->dev, + rx_buffer->dma, + rx_buffer->page_offset, + WX_RX_BUFSZ, + DMA_FROM_DEVICE); + + /* free resources associated with mapping */ + page_pool_put_full_page(rx_ring->page_pool, rx_buffer->page, false); + __page_frag_cache_drain(rx_buffer->page, + rx_buffer->pagecnt_bias); + + i++; + rx_buffer++; + if (i == rx_ring->count) { + i = 0; + rx_buffer = rx_ring->rx_buffer_info; + } + } + + rx_ring->next_to_alloc = 0; + rx_ring->next_to_clean = 0; + rx_ring->next_to_use = 0; +} + +/** + * wx_clean_all_rx_rings - Free Rx Buffers for all queues + * @wx: board private structure + **/ +void wx_clean_all_rx_rings(struct wx *wx) +{ + int i; + + for (i = 0; i < wx->num_rx_queues; i++) + wx_clean_rx_ring(wx->rx_ring[i]); +} +EXPORT_SYMBOL(wx_clean_all_rx_rings); + /** * wx_free_rx_resources - Free Rx Resources * @rx_ring: ring to clean the resources from @@ -617,6 +1234,7 @@ EXPORT_SYMBOL(wx_configure_vectors); **/ static void wx_free_rx_resources(struct wx_ring *rx_ring) { + wx_clean_rx_ring(rx_ring); kvfree(rx_ring->rx_buffer_info); rx_ring->rx_buffer_info = NULL; @@ -755,6 +1373,9 @@ static int wx_setup_rx_resources(struct wx_ring *rx_ring) if (!rx_ring->desc) goto err; + rx_ring->next_to_clean = 0; + rx_ring->next_to_use = 0; + ret = wx_alloc_page_pool(rx_ring); if (ret < 0) { dev_err(rx_ring->dev, "Page pool creation failed: %d\n", ret); @@ -909,4 +1530,36 @@ int wx_setup_resources(struct wx *wx) } EXPORT_SYMBOL(wx_setup_resources); +/** + * wx_get_stats64 - Get System Network Statistics + * @netdev: network interface device structure + * @stats: storage space for 64bit statistics + */ +void wx_get_stats64(struct net_device *netdev, + struct rtnl_link_stats64 *stats) +{ + struct wx *wx = netdev_priv(netdev); + int i; + + rcu_read_lock(); + for (i = 0; i < wx->num_rx_queues; i++) { + struct wx_ring *ring = READ_ONCE(wx->rx_ring[i]); + u64 bytes, packets; + unsigned int start; + + if (ring) { + do { + start = u64_stats_fetch_begin(&ring->syncp); + packets = ring->stats.packets; + bytes = ring->stats.bytes; + } while (u64_stats_fetch_retry(&ring->syncp, start)); + stats->rx_packets += packets; + stats->rx_bytes += bytes; + } + } + + rcu_read_unlock(); +} +EXPORT_SYMBOL(wx_get_stats64); + MODULE_LICENSE("GPL"); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.h b/drivers/net/ethernet/wangxun/libwx/wx_lib.h index 6fa95752fc42..8fc7f6f3d40e 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.h @@ -7,6 +7,10 @@ #ifndef _WX_LIB_H_ #define _WX_LIB_H_ +void wx_alloc_rx_buffers(struct wx_ring *rx_ring, u16 cleaned_count); +u16 wx_desc_unused(struct wx_ring *ring); +void wx_napi_enable_all(struct wx *wx); +void wx_napi_disable_all(struct wx *wx); void wx_reset_interrupt_capability(struct wx *wx); void wx_clear_interrupt_scheme(struct wx *wx); int wx_init_interrupt_scheme(struct wx *wx); @@ -16,7 +20,10 @@ int wx_setup_isb_resources(struct wx *wx); void wx_free_isb_resources(struct wx *wx); u32 wx_misc_isb(struct wx *wx, enum wx_isb_idx idx); void wx_configure_vectors(struct wx *wx); +void wx_clean_all_rx_rings(struct wx *wx); void wx_free_resources(struct wx *wx); int wx_setup_resources(struct wx *wx); +void wx_get_stats64(struct net_device *netdev, + struct rtnl_link_stats64 *stats); #endif /* _NGBE_LIB_H_ */ diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index 479bc4fd2bd4..9c340bd34803 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -311,8 +311,15 @@ #define WX_RX_BUFSZ WX_RXBUFFER_2K #endif +#define WX_RX_BUFFER_WRITE 16 /* Must be power of 2 */ #define WX_CFG_PORT_ST 0x14404 +/******************* Receive Descriptor bit definitions **********************/ +#define WX_RXD_STAT_DD BIT(0) /* Done */ +#define WX_RXD_STAT_EOP BIT(1) /* End of Packet */ + +#define WX_RXD_ERR_RXE BIT(29) /* Any MAC Error */ + /* Host Interface Command Structures */ struct wx_hic_hdr { u8 cmd; @@ -433,6 +440,15 @@ enum wx_reset_type { WX_GLOBAL_RESET }; +struct wx_cb { + dma_addr_t dma; + u16 append_cnt; /* number of skb's appended */ + bool page_released; + bool dma_released; +}; + +#define WX_CB(skb) ((struct wx_cb *)(skb)->cb) + /* Transmit Descriptor */ union wx_tx_desc { struct { @@ -478,6 +494,9 @@ union wx_rx_desc { } wb; /* writeback */ }; +#define WX_RX_DESC(R, i) \ + (&(((union wx_rx_desc *)((R)->desc))[i])) + /* wrapper around a pointer to a socket buffer, * so a DMA handle can be stored along with the buffer */ @@ -496,6 +515,12 @@ struct wx_rx_buffer { dma_addr_t page_dma; struct page *page; unsigned int page_offset; + u16 pagecnt_bias; +}; + +struct wx_queue_stats { + u64 packets; + u64 bytes; }; /* iterator for handling rings in ring container */ @@ -504,6 +529,8 @@ struct wx_rx_buffer { struct wx_ring_container { struct wx_ring *ring; /* pointer to linked list of rings */ + unsigned int total_bytes; /* total bytes processed this int */ + unsigned int total_packets; /* total packets processed this int */ u8 count; /* total number of rings in vector */ u8 itr; /* current ITR setting for ring */ }; @@ -531,6 +558,12 @@ struct wx_ring { * associated with this ring, which is * different for DCB and RSS modes */ + u16 next_to_use; + u16 next_to_clean; + u16 next_to_alloc; + + struct wx_queue_stats stats; + struct u64_stats_sync syncp; } ____cacheline_internodealigned_in_smp; struct wx_q_vector { @@ -632,6 +665,7 @@ struct wx { }; #define WX_INTR_ALL (~0ULL) +#define WX_INTR_Q(i) BIT(i) /* register operations */ #define wr32(a, reg, value) writel((value), ((a)->hw_addr + (reg))) From patchwork Fri Feb 3 09:11:33 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127232 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 267F0C636CC for ; Fri, 3 Feb 2023 09:16:41 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232771AbjBCJQj (ORCPT ); Fri, 3 Feb 2023 04:16:39 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:37038 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232632AbjBCJQX (ORCPT ); Fri, 3 Feb 2023 04:16:23 -0500 Received: from smtpbg150.qq.com (smtpbg150.qq.com [18.132.163.193]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id CA349991DB for ; Fri, 3 Feb 2023 01:16:03 -0800 (PST) X-QQ-mid: bizesmtp68t1675415755tl1a0q14 Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:54 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: hvKw+PyJ8U4k0ooBJPPH3YuECmuEPfYR7FgcMY5HyTCWKfBKltKvOKdwYhIHY JJ7xT220i8Kw8pkx/YFDVGVNdL2VTzoTHiSz5Bh/gVDag2Qk6LmIDduu/u2RjeUHtGx7xXC RTRH6n1AI9Q6C825+Y1KtI9c5nlKoVZqXeqr8r8Hz140kzt66iQqyPFT4mjMU35NeYT0+kS cOfSPXBa0ly4FUz9UZfDs6N1yr/EWU46iClNOBFVafX9BAKiy1hKu2DA/bO8aY2a4Wwu1TW LDy5hSZ1rhpDqxoBRZf8z6LNAyuw9kOsBC0NyYTEA8Wp+7JbICK1qP66l9X2P2kdcIhm2nW UXeonVzVTGzG4bPsebrKhztyMfeNjatpt32TGvtmtxpyTYZ5mLOvSKTZga4efLdmHaMRIMo nW/eHUC8XzA= X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 08/10] net: libwx: Add tx path to process packets Date: Fri, 3 Feb 2023 17:11:33 +0800 Message-Id: <20230203091135.3294377-9-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org From: Mengyuan Lou Support to transmit packets without hardware features. Signed-off-by: Mengyuan Lou Signed-off-by: Jiawen Wu --- drivers/net/ethernet/wangxun/libwx/wx_hw.c | 4 + drivers/net/ethernet/wangxun/libwx/wx_lib.c | 439 +++++++++++++++++++ drivers/net/ethernet/wangxun/libwx/wx_lib.h | 3 + drivers/net/ethernet/wangxun/libwx/wx_type.h | 19 + 4 files changed, 465 insertions(+) diff --git a/drivers/net/ethernet/wangxun/libwx/wx_hw.c b/drivers/net/ethernet/wangxun/libwx/wx_hw.c index f5ad5f991ab7..7db57f934a91 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_hw.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_hw.c @@ -1355,6 +1355,10 @@ static void wx_configure_tx_ring(struct wx *wx, txdctl |= ring->count / 128 << WX_PX_TR_CFG_TR_SIZE_SHIFT; txdctl |= 0x20 << WX_PX_TR_CFG_WTHRESH_SHIFT; + /* reinitialize tx_buffer_info */ + memset(ring->tx_buffer_info, 0, + sizeof(struct wx_tx_buffer) * ring->count); + /* enable queue */ wr32(wx, WX_PX_TR_CFG(reg_idx), txdctl); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.c b/drivers/net/ethernet/wangxun/libwx/wx_lib.c index ea6a4207f503..57e1871ea0c6 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.c +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.c @@ -504,6 +504,134 @@ static int wx_clean_rx_irq(struct wx_q_vector *q_vector, return total_rx_packets; } +static struct netdev_queue *wx_txring_txq(const struct wx_ring *ring) +{ + return netdev_get_tx_queue(ring->netdev, ring->queue_index); +} + +/** + * wx_clean_tx_irq - Reclaim resources after transmit completes + * @q_vector: structure containing interrupt and ring information + * @tx_ring: tx ring to clean + * @napi_budget: Used to determine if we are in netpoll + **/ +static bool wx_clean_tx_irq(struct wx_q_vector *q_vector, + struct wx_ring *tx_ring, int napi_budget) +{ + unsigned int budget = q_vector->wx->tx_work_limit; + unsigned int total_bytes = 0, total_packets = 0; + unsigned int i = tx_ring->next_to_clean; + struct wx_tx_buffer *tx_buffer; + union wx_tx_desc *tx_desc; + + if (!netif_carrier_ok(tx_ring->netdev)) + return true; + + tx_buffer = &tx_ring->tx_buffer_info[i]; + tx_desc = WX_TX_DESC(tx_ring, i); + i -= tx_ring->count; + + do { + union wx_tx_desc *eop_desc = tx_buffer->next_to_watch; + + /* if next_to_watch is not set then there is no work pending */ + if (!eop_desc) + break; + + /* prevent any other reads prior to eop_desc */ + smp_rmb(); + + /* if DD is not set pending work has not been completed */ + if (!(eop_desc->wb.status & cpu_to_le32(WX_TXD_STAT_DD))) + break; + + /* clear next_to_watch to prevent false hangs */ + tx_buffer->next_to_watch = NULL; + + /* update the statistics for this packet */ + total_bytes += tx_buffer->bytecount; + total_packets += tx_buffer->gso_segs; + + /* free the skb */ + napi_consume_skb(tx_buffer->skb, napi_budget); + + /* unmap skb header data */ + dma_unmap_single(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + + /* clear tx_buffer data */ + dma_unmap_len_set(tx_buffer, len, 0); + + /* unmap remaining buffers */ + while (tx_desc != eop_desc) { + tx_buffer++; + tx_desc++; + i++; + if (unlikely(!i)) { + i -= tx_ring->count; + tx_buffer = tx_ring->tx_buffer_info; + tx_desc = WX_TX_DESC(tx_ring, 0); + } + + /* unmap any remaining paged data */ + if (dma_unmap_len(tx_buffer, len)) { + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + dma_unmap_len_set(tx_buffer, len, 0); + } + } + + /* move us one more past the eop_desc for start of next pkt */ + tx_buffer++; + tx_desc++; + i++; + if (unlikely(!i)) { + i -= tx_ring->count; + tx_buffer = tx_ring->tx_buffer_info; + tx_desc = WX_TX_DESC(tx_ring, 0); + } + + /* issue prefetch for next Tx descriptor */ + prefetch(tx_desc); + + /* update budget accounting */ + budget--; + } while (likely(budget)); + + i += tx_ring->count; + tx_ring->next_to_clean = i; + u64_stats_update_begin(&tx_ring->syncp); + tx_ring->stats.bytes += total_bytes; + tx_ring->stats.packets += total_packets; + u64_stats_update_end(&tx_ring->syncp); + q_vector->tx.total_bytes += total_bytes; + q_vector->tx.total_packets += total_packets; + + netdev_tx_completed_queue(wx_txring_txq(tx_ring), + total_packets, total_bytes); + +#define TX_WAKE_THRESHOLD (DESC_NEEDED * 2) + if (unlikely(total_packets && netif_carrier_ok(tx_ring->netdev) && + (wx_desc_unused(tx_ring) >= TX_WAKE_THRESHOLD))) { + /* Make sure that anybody stopping the queue after this + * sees the new next_to_clean. + */ + smp_mb(); + + if (__netif_subqueue_stopped(tx_ring->netdev, + tx_ring->queue_index) && + netif_running(tx_ring->netdev)) + netif_wake_subqueue(tx_ring->netdev, + tx_ring->queue_index); + } + + return !!budget; +} + /** * wx_poll - NAPI polling RX/TX cleanup routine * @napi: napi struct with our devices info in it @@ -519,6 +647,11 @@ static int wx_poll(struct napi_struct *napi, int budget) bool clean_complete = true; struct wx_ring *ring; + wx_for_each_ring(ring, q_vector->tx) { + if (!wx_clean_tx_irq(q_vector, ring, budget)) + clean_complete = false; + } + /* Exit if we are called by netpoll */ if (budget <= 0) return budget; @@ -552,6 +685,216 @@ static int wx_poll(struct napi_struct *napi, int budget) return min(work_done, budget - 1); } +static int wx_maybe_stop_tx(struct wx_ring *tx_ring, u16 size) +{ + if (likely(wx_desc_unused(tx_ring) >= size)) + return 0; + + netif_stop_subqueue(tx_ring->netdev, tx_ring->queue_index); + + /* For the next check */ + smp_mb(); + + /* We need to check again in a case another CPU has just + * made room available. + */ + if (likely(wx_desc_unused(tx_ring) < size)) + return -EBUSY; + + /* A reprieve! - use start_queue because it doesn't call schedule */ + netif_start_subqueue(tx_ring->netdev, tx_ring->queue_index); + + return 0; +} + +static void wx_tx_map(struct wx_ring *tx_ring, + struct wx_tx_buffer *first) +{ + struct sk_buff *skb = first->skb; + struct wx_tx_buffer *tx_buffer; + u16 i = tx_ring->next_to_use; + unsigned int data_len, size; + union wx_tx_desc *tx_desc; + skb_frag_t *frag; + dma_addr_t dma; + u32 cmd_type; + + cmd_type = WX_TXD_DTYP_DATA | WX_TXD_IFCS; + tx_desc = WX_TX_DESC(tx_ring, i); + + tx_desc->read.olinfo_status = cpu_to_le32(skb->len << WX_TXD_PAYLEN_SHIFT); + + size = skb_headlen(skb); + data_len = skb->data_len; + dma = dma_map_single(tx_ring->dev, skb->data, size, DMA_TO_DEVICE); + + tx_buffer = first; + + for (frag = &skb_shinfo(skb)->frags[0];; frag++) { + if (dma_mapping_error(tx_ring->dev, dma)) + goto dma_error; + + /* record length, and DMA address */ + dma_unmap_len_set(tx_buffer, len, size); + dma_unmap_addr_set(tx_buffer, dma, dma); + + tx_desc->read.buffer_addr = cpu_to_le64(dma); + + while (unlikely(size > WX_MAX_DATA_PER_TXD)) { + tx_desc->read.cmd_type_len = + cpu_to_le32(cmd_type ^ WX_MAX_DATA_PER_TXD); + + i++; + tx_desc++; + if (i == tx_ring->count) { + tx_desc = WX_TX_DESC(tx_ring, 0); + i = 0; + } + tx_desc->read.olinfo_status = 0; + + dma += WX_MAX_DATA_PER_TXD; + size -= WX_MAX_DATA_PER_TXD; + + tx_desc->read.buffer_addr = cpu_to_le64(dma); + } + + if (likely(!data_len)) + break; + + tx_desc->read.cmd_type_len = cpu_to_le32(cmd_type ^ size); + + i++; + tx_desc++; + if (i == tx_ring->count) { + tx_desc = WX_TX_DESC(tx_ring, 0); + i = 0; + } + tx_desc->read.olinfo_status = 0; + + size = skb_frag_size(frag); + + data_len -= size; + + dma = skb_frag_dma_map(tx_ring->dev, frag, 0, size, + DMA_TO_DEVICE); + + tx_buffer = &tx_ring->tx_buffer_info[i]; + } + + /* write last descriptor with RS and EOP bits */ + cmd_type |= size | WX_TXD_EOP | WX_TXD_RS; + tx_desc->read.cmd_type_len = cpu_to_le32(cmd_type); + + netdev_tx_sent_queue(wx_txring_txq(tx_ring), first->bytecount); + + skb_tx_timestamp(skb); + + /* Force memory writes to complete before letting h/w know there + * are new descriptors to fetch. (Only applicable for weak-ordered + * memory model archs, such as IA-64). + * + * We also need this memory barrier to make certain all of the + * status bits have been updated before next_to_watch is written. + */ + wmb(); + + /* set next_to_watch value indicating a packet is present */ + first->next_to_watch = tx_desc; + + i++; + if (i == tx_ring->count) + i = 0; + + tx_ring->next_to_use = i; + + wx_maybe_stop_tx(tx_ring, DESC_NEEDED); + + if (netif_xmit_stopped(wx_txring_txq(tx_ring)) || !netdev_xmit_more()) + writel(i, tx_ring->tail); + + return; +dma_error: + dev_err(tx_ring->dev, "TX DMA map failed\n"); + + /* clear dma mappings for failed tx_buffer_info map */ + for (;;) { + tx_buffer = &tx_ring->tx_buffer_info[i]; + if (dma_unmap_len(tx_buffer, len)) + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + dma_unmap_len_set(tx_buffer, len, 0); + if (tx_buffer == first) + break; + if (i == 0) + i += tx_ring->count; + i--; + } + + dev_kfree_skb_any(first->skb); + first->skb = NULL; + + tx_ring->next_to_use = i; +} + +static netdev_tx_t wx_xmit_frame_ring(struct sk_buff *skb, + struct wx_ring *tx_ring) +{ + u16 count = TXD_USE_COUNT(skb_headlen(skb)); + struct wx_tx_buffer *first; + unsigned short f; + + /* need: 1 descriptor per page * PAGE_SIZE/WX_MAX_DATA_PER_TXD, + * + 1 desc for skb_headlen/WX_MAX_DATA_PER_TXD, + * + 2 desc gap to keep tail from touching head, + * + 1 desc for context descriptor, + * otherwise try next time + */ + for (f = 0; f < skb_shinfo(skb)->nr_frags; f++) + count += TXD_USE_COUNT(skb_frag_size(&skb_shinfo(skb)-> + frags[f])); + + if (wx_maybe_stop_tx(tx_ring, count + 3)) + return NETDEV_TX_BUSY; + + /* record the location of the first descriptor for this packet */ + first = &tx_ring->tx_buffer_info[tx_ring->next_to_use]; + first->skb = skb; + first->bytecount = skb->len; + first->gso_segs = 1; + + wx_tx_map(tx_ring, first); + + return NETDEV_TX_OK; +} + +netdev_tx_t wx_xmit_frame(struct sk_buff *skb, + struct net_device *netdev) +{ + unsigned int r_idx = skb->queue_mapping; + struct wx *wx = netdev_priv(netdev); + struct wx_ring *tx_ring; + + if (!netif_carrier_ok(netdev)) { + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; + } + + /* The minimum packet size for olinfo paylen is 17 so pad the skb + * in order to meet this minimum size requirement. + */ + if (skb_put_padto(skb, 17)) + return NETDEV_TX_OK; + + if (r_idx >= wx->num_tx_queues) + r_idx = r_idx % wx->num_tx_queues; + tx_ring = wx->tx_ring[r_idx]; + + return wx_xmit_frame_ring(skb, tx_ring); +} +EXPORT_SYMBOL(wx_xmit_frame); + void wx_napi_enable_all(struct wx *wx) { struct wx_q_vector *q_vector; @@ -1267,6 +1610,81 @@ static void wx_free_all_rx_resources(struct wx *wx) wx_free_rx_resources(wx->rx_ring[i]); } +/** + * wx_clean_tx_ring - Free Tx Buffers + * @tx_ring: ring to be cleaned + **/ +static void wx_clean_tx_ring(struct wx_ring *tx_ring) +{ + struct wx_tx_buffer *tx_buffer; + u16 i = tx_ring->next_to_clean; + + tx_buffer = &tx_ring->tx_buffer_info[i]; + + while (i != tx_ring->next_to_use) { + union wx_tx_desc *eop_desc, *tx_desc; + + /* Free all the Tx ring sk_buffs */ + dev_kfree_skb_any(tx_buffer->skb); + + /* unmap skb header data */ + dma_unmap_single(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + + /* check for eop_desc to determine the end of the packet */ + eop_desc = tx_buffer->next_to_watch; + tx_desc = WX_TX_DESC(tx_ring, i); + + /* unmap remaining buffers */ + while (tx_desc != eop_desc) { + tx_buffer++; + tx_desc++; + i++; + if (unlikely(i == tx_ring->count)) { + i = 0; + tx_buffer = tx_ring->tx_buffer_info; + tx_desc = WX_TX_DESC(tx_ring, 0); + } + + /* unmap any remaining paged data */ + if (dma_unmap_len(tx_buffer, len)) + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } + + /* move us one more past the eop_desc for start of next pkt */ + tx_buffer++; + i++; + if (unlikely(i == tx_ring->count)) { + i = 0; + tx_buffer = tx_ring->tx_buffer_info; + } + } + + netdev_tx_reset_queue(wx_txring_txq(tx_ring)); + + /* reset next_to_use and next_to_clean */ + tx_ring->next_to_use = 0; + tx_ring->next_to_clean = 0; +} + +/** + * wx_clean_all_tx_rings - Free Tx Buffers for all queues + * @wx: board private structure + **/ +void wx_clean_all_tx_rings(struct wx *wx) +{ + int i; + + for (i = 0; i < wx->num_tx_queues; i++) + wx_clean_tx_ring(wx->tx_ring[i]); +} +EXPORT_SYMBOL(wx_clean_all_tx_rings); + /** * wx_free_tx_resources - Free Tx Resources per Queue * @tx_ring: Tx descriptor ring for a specific queue @@ -1275,6 +1693,7 @@ static void wx_free_all_rx_resources(struct wx *wx) **/ static void wx_free_tx_resources(struct wx_ring *tx_ring) { + wx_clean_tx_ring(tx_ring); kvfree(tx_ring->tx_buffer_info); tx_ring->tx_buffer_info = NULL; @@ -1461,6 +1880,9 @@ static int wx_setup_tx_resources(struct wx_ring *tx_ring) if (!tx_ring->desc) goto err; + tx_ring->next_to_use = 0; + tx_ring->next_to_clean = 0; + return 0; err: @@ -1558,6 +1980,23 @@ void wx_get_stats64(struct net_device *netdev, } } + for (i = 0; i < wx->num_tx_queues; i++) { + struct wx_ring *ring = READ_ONCE(wx->tx_ring[i]); + u64 bytes, packets; + unsigned int start; + + if (ring) { + do { + start = u64_stats_fetch_begin(&ring->syncp); + packets = ring->stats.packets; + bytes = ring->stats.bytes; + } while (u64_stats_fetch_retry(&ring->syncp, + start)); + stats->tx_packets += packets; + stats->tx_bytes += bytes; + } + } + rcu_read_unlock(); } EXPORT_SYMBOL(wx_get_stats64); diff --git a/drivers/net/ethernet/wangxun/libwx/wx_lib.h b/drivers/net/ethernet/wangxun/libwx/wx_lib.h index 8fc7f6f3d40e..50ee41f1fa10 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_lib.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_lib.h @@ -9,6 +9,8 @@ void wx_alloc_rx_buffers(struct wx_ring *rx_ring, u16 cleaned_count); u16 wx_desc_unused(struct wx_ring *ring); +netdev_tx_t wx_xmit_frame(struct sk_buff *skb, + struct net_device *netdev); void wx_napi_enable_all(struct wx *wx); void wx_napi_disable_all(struct wx *wx); void wx_reset_interrupt_capability(struct wx *wx); @@ -21,6 +23,7 @@ void wx_free_isb_resources(struct wx *wx); u32 wx_misc_isb(struct wx *wx, enum wx_isb_idx idx); void wx_configure_vectors(struct wx *wx); void wx_clean_all_rx_rings(struct wx *wx); +void wx_clean_all_tx_rings(struct wx *wx); void wx_free_resources(struct wx *wx); int wx_setup_resources(struct wx *wx); void wx_get_stats64(struct net_device *netdev, diff --git a/drivers/net/ethernet/wangxun/libwx/wx_type.h b/drivers/net/ethernet/wangxun/libwx/wx_type.h index 9c340bd34803..eede93d4120d 100644 --- a/drivers/net/ethernet/wangxun/libwx/wx_type.h +++ b/drivers/net/ethernet/wangxun/libwx/wx_type.h @@ -312,6 +312,15 @@ #endif #define WX_RX_BUFFER_WRITE 16 /* Must be power of 2 */ + +#define WX_MAX_DATA_PER_TXD BIT(14) +/* Tx Descriptors needed, worst case */ +#define TXD_USE_COUNT(S) DIV_ROUND_UP((S), WX_MAX_DATA_PER_TXD) +#define DESC_NEEDED (MAX_SKB_FRAGS + 4) + +/* Ether Types */ +#define WX_ETH_P_CNM 0x22E7 + #define WX_CFG_PORT_ST 0x14404 /******************* Receive Descriptor bit definitions **********************/ @@ -320,6 +329,14 @@ #define WX_RXD_ERR_RXE BIT(29) /* Any MAC Error */ +/*********************** Transmit Descriptor Config Masks ****************/ +#define WX_TXD_STAT_DD BIT(0) /* Descriptor Done */ +#define WX_TXD_DTYP_DATA 0 /* Adv Data Descriptor */ +#define WX_TXD_PAYLEN_SHIFT 13 /* Desc PAYLEN shift */ +#define WX_TXD_EOP BIT(24) /* End of Packet */ +#define WX_TXD_IFCS BIT(25) /* Insert FCS */ +#define WX_TXD_RS BIT(27) /* Report Status */ + /* Host Interface Command Structures */ struct wx_hic_hdr { u8 cmd; @@ -496,6 +513,8 @@ union wx_rx_desc { #define WX_RX_DESC(R, i) \ (&(((union wx_rx_desc *)((R)->desc))[i])) +#define WX_TX_DESC(R, i) \ + (&(((union wx_tx_desc *)((R)->desc))[i])) /* wrapper around a pointer to a socket buffer, * so a DMA handle can be stored along with the buffer From patchwork Fri Feb 3 09:11:34 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127231 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 0EB06C61DA4 for ; Fri, 3 Feb 2023 09:16:42 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232773AbjBCJQk (ORCPT ); Fri, 3 Feb 2023 04:16:40 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:37046 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232675AbjBCJQX (ORCPT ); Fri, 3 Feb 2023 04:16:23 -0500 Received: from smtpbguseast3.qq.com (smtpbguseast3.qq.com [54.243.244.52]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id A65B195D31 for ; Fri, 3 Feb 2023 01:16:04 -0800 (PST) X-QQ-mid: bizesmtp68t1675415758tos65246 Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:15:57 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: lkL5M32tl2Bh0YGrduW6ZbtZs+gCZwgnNRB45cAfYyXy8GxAhwA5kSMPuW9hr WJlNStwFvz/AyavYziHePt/vxly6PhEqfgfH8CggQ8cLepWNPuEzr8EmdGqveIsLxiVrSLT b56f5O4u6Pkum7+YLmb8GVNcLWlHkJD4E0WYp7VTPSYgD6vxvEFKRIVN/O71xOMW9lxJNzJ DJ9+kX2CEfNdVOjJNr1g0QLoM2kspdVzwevWVPfDb2fjtx2eyOw8lrBhKuIR9AFdRNHsF9z YeTSVYPRGD4jR0iK9/nRCjP+pdA4XgTP3OpD0vJNy03K0Z5gvboQ1W+oCfbQjYUJO6dbQOa nkOeetQXgnGbV35lXQMif2v7UulsuQCBjNvTuI9D7oUlTXu8nyifdlGJxPVvD5Ik/V4RNWa oSkeuTbbkiDBJ5gqluDJrg== X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Cc: Jiawen Wu Subject: [PATCH net-next v3 09/10] net: txgbe: Support Rx and Tx process path Date: Fri, 3 Feb 2023 17:11:34 +0800 Message-Id: <20230203091135.3294377-10-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Clean Rx and Tx ring interrupts, process packets in the data path. Signed-off-by: Jiawen Wu --- .../net/ethernet/wangxun/txgbe/txgbe_main.c | 37 ++++++++++++++----- 1 file changed, 28 insertions(+), 9 deletions(-) diff --git a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c index 3b50acb09699..094df377726b 100644 --- a/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c +++ b/drivers/net/ethernet/wangxun/txgbe/txgbe_main.c @@ -223,6 +223,10 @@ static void txgbe_up_complete(struct wx *wx) wx_control_hw(wx, true); wx_configure_vectors(wx); + /* make sure to complete pre-operations */ + smp_mb__before_atomic(); + wx_napi_enable_all(wx); + /* clear any pending interrupts, may auto mask */ rd32(wx, WX_PX_IC); rd32(wx, WX_PX_MISC_IC); @@ -236,6 +240,10 @@ static void txgbe_up_complete(struct wx *wx) wr32(wx, WX_MAC_WDG_TIMEOUT, reg); reg = rd32(wx, WX_MAC_TX_CFG); wr32(wx, WX_MAC_TX_CFG, (reg & ~WX_MAC_TX_CFG_SPEED_MASK) | WX_MAC_TX_CFG_SPEED_10G); + + /* enable transmits */ + netif_tx_start_all_queues(wx->netdev); + netif_carrier_on(wx->netdev); } static void txgbe_reset(struct wx *wx) @@ -268,10 +276,12 @@ static void txgbe_disable_device(struct wx *wx) /* this call also flushes the previous write */ wx_disable_rx_queue(wx, wx->rx_ring[i]); + netif_tx_stop_all_queues(netdev); netif_carrier_off(netdev); netif_tx_disable(netdev); wx_irq_disable(wx); + wx_napi_disable_all(wx); if (wx->bus.func < 2) wr32m(wx, TXGBE_MIS_PRB_CTL, TXGBE_MIS_PRB_CTL_LAN_UP(wx->bus.func), 0); @@ -300,6 +310,9 @@ static void txgbe_down(struct wx *wx) { txgbe_disable_device(wx); txgbe_reset(wx); + + wx_clean_all_tx_rings(wx); + wx_clean_all_rx_rings(wx); } /** @@ -381,10 +394,21 @@ static int txgbe_open(struct net_device *netdev) if (err) goto err_free_isb; + /* Notify the stack of the actual queue counts. */ + err = netif_set_real_num_tx_queues(netdev, wx->num_tx_queues); + if (err) + goto err_free_irq; + + err = netif_set_real_num_rx_queues(netdev, wx->num_rx_queues); + if (err) + goto err_free_irq; + txgbe_up_complete(wx); return 0; +err_free_irq: + wx_free_irq(wx); err_free_isb: wx_free_isb_resources(wx); err_reset: @@ -403,8 +427,6 @@ static int txgbe_open(struct net_device *netdev) static void txgbe_close_suspend(struct wx *wx) { txgbe_disable_device(wx); - - wx_free_irq(wx); wx_free_resources(wx); } @@ -461,19 +483,14 @@ static void txgbe_shutdown(struct pci_dev *pdev) } } -static netdev_tx_t txgbe_xmit_frame(struct sk_buff *skb, - struct net_device *netdev) -{ - return NETDEV_TX_OK; -} - static const struct net_device_ops txgbe_netdev_ops = { .ndo_open = txgbe_open, .ndo_stop = txgbe_close, - .ndo_start_xmit = txgbe_xmit_frame, + .ndo_start_xmit = wx_xmit_frame, .ndo_set_rx_mode = wx_set_rx_mode, .ndo_validate_addr = eth_validate_addr, .ndo_set_mac_address = wx_set_mac, + .ndo_get_stats64 = wx_get_stats64, }; /** @@ -647,6 +664,8 @@ static int txgbe_probe(struct pci_dev *pdev, pci_set_drvdata(pdev, wx); + netif_tx_stop_all_queues(netdev); + /* calculate the expected PCIe bandwidth required for optimal * performance. Note that some older parts will never have enough * bandwidth due to being older generation PCIe parts. We clamp these From patchwork Fri Feb 3 09:11:35 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jiawen Wu X-Patchwork-Id: 13127233 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 81F3EC636CC for ; Fri, 3 Feb 2023 09:16:49 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232785AbjBCJQs (ORCPT ); Fri, 3 Feb 2023 04:16:48 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:36552 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S231307AbjBCJQZ (ORCPT ); Fri, 3 Feb 2023 04:16:25 -0500 Received: from smtpbguseast1.qq.com (smtpbguseast1.qq.com [54.204.34.129]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id A6A799B6F7 for ; Fri, 3 Feb 2023 01:16:06 -0800 (PST) X-QQ-mid: bizesmtp68t1675415760tjfqxv7o Received: from wxdbg.localdomain.com ( [183.129.236.74]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 03 Feb 2023 17:16:00 +0800 (CST) X-QQ-SSF: 01400000000000H0Y000B00A0000000 X-QQ-FEAT: lkL5M32tl2Cyix4MfR380DdxBZhMnSMAkcGWoKeVH4zqjs9ZT+EAw8PS/iZug YLLrvY3K5UXzV8DA6BS8+UD/jf+elDhFz76TG/ppb5egkpLfgTKPe0Oj8wTH/lWuKMCs9oV 1O9tJbe62nzivTcLefEuO3h2J605CxZJibnzaPRmhpXGbmmFjZXy5LwDjZ3xz6988mjiJbM hpKdElNCK4ubJEJshaDu3YlQNOdUiOQZzh7OO4AmXCIuUHU+BrDblLGySP3JShSz8LQq9Th fUKII6mUgnY9TH0REg2e1PqTLXNuxcUwXvgM8e2IOBUYe/XmoadcjX8AYehYNQ4hLa8Dg5e wCD57+0NbBHJcBH8uD3MPTsh4Z6cDYzanipMf5NmWWyAcHjK3NOf5XWquTIB+zJvffPrxOi pp3eEFbUBFA= X-QQ-GoodBg: 2 From: Jiawen Wu To: netdev@vger.kernel.org, mengyuanlou@net-swift.com Subject: [PATCH net-next v3 10/10] net: ngbe: Support Rx and Tx process path Date: Fri, 3 Feb 2023 17:11:35 +0800 Message-Id: <20230203091135.3294377-11-jiawenwu@trustnetic.com> X-Mailer: git-send-email 2.27.0 In-Reply-To: <20230203091135.3294377-1-jiawenwu@trustnetic.com> References: <20230203091135.3294377-1-jiawenwu@trustnetic.com> MIME-Version: 1.0 X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:trustnetic.com:qybglogicsvr:qybglogicsvr5 Precedence: bulk List-ID: X-Mailing-List: netdev@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org From: Mengyuan Lou Add enable and disable operation process for ngbe open/close. Clean Rx and Tx ring interrupts, process packets in the data path. Signed-off-by: Mengyuan Lou --- drivers/net/ethernet/wangxun/ngbe/ngbe_main.c | 70 +++++++++++++++---- drivers/net/ethernet/wangxun/ngbe/ngbe_type.h | 4 ++ 2 files changed, 61 insertions(+), 13 deletions(-) diff --git a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c index a9772f929274..f94d415daf3c 100644 --- a/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c +++ b/drivers/net/ethernet/wangxun/ngbe/ngbe_main.c @@ -113,6 +113,9 @@ static int ngbe_sw_init(struct wx *wx) wx->mac.num_rar_entries = NGBE_RAR_ENTRIES; wx->mac.max_rx_queues = NGBE_MAX_RX_QUEUES; wx->mac.max_tx_queues = NGBE_MAX_TX_QUEUES; + wx->mac.mcft_size = NGBE_MC_TBL_SIZE; + wx->mac.rx_pb_size = NGBE_RX_PB_SIZE; + wx->mac.tx_pb_size = NGBE_TDB_PB_SZ; /* PCI config space info */ err = wx_sw_init(wx); @@ -307,25 +310,46 @@ static int ngbe_request_irq(struct wx *wx) static void ngbe_disable_device(struct wx *wx) { struct net_device *netdev = wx->netdev; + u32 i; + /* disable all enabled rx queues */ + for (i = 0; i < wx->num_rx_queues; i++) + /* this call also flushes the previous write */ + wx_disable_rx_queue(wx, wx->rx_ring[i]); /* disable receives */ wx_disable_rx(wx); + wx_napi_disable_all(wx); + netif_tx_stop_all_queues(netdev); netif_tx_disable(netdev); if (wx->gpio_ctrl) ngbe_sfp_modules_txrx_powerctl(wx, false); wx_irq_disable(wx); + /* disable transmits in the hardware now that interrupts are off */ + for (i = 0; i < wx->num_tx_queues; i++) { + u8 reg_idx = wx->tx_ring[i]->reg_idx; + + wr32(wx, WX_PX_TR_CFG(reg_idx), WX_PX_TR_CFG_SWFLSH); + } } static void ngbe_down(struct wx *wx) { phy_stop(wx->phydev); ngbe_disable_device(wx); + wx_clean_all_tx_rings(wx); + wx_clean_all_rx_rings(wx); } static void ngbe_up(struct wx *wx) { wx_configure_vectors(wx); + /* make sure to complete pre-operations */ + smp_mb__before_atomic(); + wx_napi_enable_all(wx); + /* enable transmits */ + netif_tx_start_all_queues(wx->netdev); + /* clear any pending interrupts, may auto mask */ rd32(wx, WX_PX_IC); rd32(wx, WX_PX_MISC_IC); @@ -352,7 +376,7 @@ static int ngbe_open(struct net_device *netdev) wx_control_hw(wx, true); - err = wx_setup_isb_resources(wx); + err = wx_setup_resources(wx); if (err) return err; @@ -360,16 +384,29 @@ static int ngbe_open(struct net_device *netdev) err = ngbe_request_irq(wx); if (err) - goto err_req_irq; + goto err_free_resources; err = ngbe_phy_connect(wx); if (err) - return err; + goto err_free_irq; + + err = netif_set_real_num_tx_queues(netdev, wx->num_tx_queues); + if (err) + goto err_dis_phy; + + err = netif_set_real_num_rx_queues(netdev, wx->num_rx_queues); + if (err) + goto err_dis_phy; + ngbe_up(wx); return 0; -err_req_irq: - wx_free_isb_resources(wx); +err_dis_phy: + phy_disconnect(wx->phydev); +err_free_irq: + wx_free_irq(wx); +err_free_resources: + wx_free_resources(wx); return err; } @@ -390,19 +427,13 @@ static int ngbe_close(struct net_device *netdev) ngbe_down(wx); wx_free_irq(wx); - wx_free_isb_resources(wx); + wx_free_resources(wx); phy_disconnect(wx->phydev); wx_control_hw(wx, false); return 0; } -static netdev_tx_t ngbe_xmit_frame(struct sk_buff *skb, - struct net_device *netdev) -{ - return NETDEV_TX_OK; -} - static void ngbe_dev_shutdown(struct pci_dev *pdev, bool *enable_wake) { struct wx *wx = pci_get_drvdata(pdev); @@ -438,9 +469,11 @@ static void ngbe_shutdown(struct pci_dev *pdev) static const struct net_device_ops ngbe_netdev_ops = { .ndo_open = ngbe_open, .ndo_stop = ngbe_close, - .ndo_start_xmit = ngbe_xmit_frame, + .ndo_start_xmit = wx_xmit_frame, + .ndo_set_rx_mode = wx_set_rx_mode, .ndo_validate_addr = eth_validate_addr, .ndo_set_mac_address = wx_set_mac, + .ndo_get_stats64 = wx_get_stats64, }; /** @@ -516,6 +549,17 @@ static int ngbe_probe(struct pci_dev *pdev, netdev->netdev_ops = &ngbe_netdev_ops; netdev->features |= NETIF_F_HIGHDMA; + netdev->features = NETIF_F_SG; + + /* copy netdev features into list of user selectable features */ + netdev->hw_features |= netdev->features | + NETIF_F_RXALL; + + netdev->priv_flags |= IFF_UNICAST_FLT; + netdev->priv_flags |= IFF_SUPP_NOFCS; + + netdev->min_mtu = ETH_MIN_MTU; + netdev->max_mtu = NGBE_MAX_JUMBO_FRAME_SIZE - (ETH_HLEN + ETH_FCS_LEN); wx->bd_number = func_nums; /* setup the private structure */ diff --git a/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h b/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h index b60f5f0f64fa..a2351349785e 100644 --- a/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h +++ b/drivers/net/ethernet/wangxun/ngbe/ngbe_type.h @@ -134,6 +134,10 @@ enum NGBE_MSCA_CMD_value { #define NGBE_ETH_LENGTH_OF_ADDRESS 6 #define NGBE_MAX_MSIX_VECTORS 0x09 #define NGBE_RAR_ENTRIES 32 +#define NGBE_RX_PB_SIZE 42 +#define NGBE_MC_TBL_SIZE 128 +#define NGBE_TDB_PB_SZ (20 * 1024) /* 160KB Packet Buffer */ +#define NGBE_MAX_JUMBO_FRAME_SIZE 9432 /* max payload 9414 */ /* TX/RX descriptor defines */ #define NGBE_DEFAULT_TXD 512 /* default ring size */