From nobody Thu Sep 24 20:41:59 2026 Received: from smtpbgbr2.qq.com (smtpbgbr2.qq.com [54.207.22.56]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 72E4A37AA7A; Sun, 20 Sep 2026 09:27:21 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=54.207.22.56 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896451; cv=none; b=d/On/kKLRj0Wk1l2veQP5pB4uN+fhQ0pmp1M0NECQ+hhE/khtykyocZu/lxU55Y4K/Z2wBUKKR+9xebDXzZsAs4bNdzV3J9Pi8JUtQY6SboOAFsDvPIuJmNIFpiEGbHjEHrmZQQYrCwVekCcwE4hM84TAYF4XfAFl70awZc0m3k= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896451; c=relaxed/simple; bh=MlWmU13XI05zP/Mzpcx1XNDz9QtqjfCfb+dAEjOUfHY=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=np/f7DIs/B1BTeTkF7vYsyyKu586qlzxftjHRGm3W7eZWNG9ikJVHvGl4GkJkTOXLv++gVFVVMZGZFmginRtDoR9HEwWWluttTQrZlqP+VSGpLG+ClPkLKYcT5csjEYWiSsn55eB9c0omuRURzrTorFUzKUnktnqTKZ0wWWhPUQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com; spf=pass smtp.mailfrom=mucse.com; arc=none smtp.client-ip=54.207.22.56 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mucse.com X-QQ-mid: esmtpgz10t1789896375t656137e7 X-QQ-Originating-IP: Jv2/ProIa+HADsVqGJPenK8KFQ0kuiX0Z7jDI2Yx5+E= Received: from nic-Precision-5820-Tower ( [203.174.112.180]) by bizesmtp.qq.com (ESMTP) with id ; Sun, 20 Sep 2026 17:26:12 +0800 (CST) X-QQ-SSF: 0000000000000000000000000000000 X-QQ-GoodBg: 0 X-BIZMAIL-ID: 1750633442563770851 EX-QQ-RecipientCnt: 11 From: Dong Yibo To: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, vadim.fedorenko@linux.dev, u.kleine-koenig@baylibre.com Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, dong100@mucse.com, yaojun@mucse.com Subject: [PATCH net-next v13 1/5] net: rnpgbe: Add interrupt handling Date: Sun, 20 Sep 2026 17:26:11 +0800 Message-ID: X-Mailer: git-send-email 2.50.1 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-QQ-SENDSIZE: 520 Feedback-ID: esmtpgz:mucse.com:qybglogicsvrgz:qybglogicsvrgz3a-1 X-QQ-XMAILINFO: MO65cnFISSdTLkHJKugoiNVUUq5KGhrGMJNNodB3zqQEaWE2mxQrIA66 0Ohd1MrtQxWMdsKmwQKwHAg7tBFsvewXplfKSb6szbzVfyG4Ok+FaQ1nOPPWq2m0NC+OhnB kkyk1Mrk2j9QQCq4rWAXY89jueNmovw6Qwi7sgNA7ORfI6NKBN9FeOGN7AhhGQlcPBHCMOK oWi0nr2Gf8cV5AfjJ+J0kuhRwPJrOCU3V/ioca5Oij3qSKK4ttc/U/56A1I4JPcILBJsdX5 Qok3ia0vKplh/RYzSZJ6UFwzbQ5wGo+kmYQWY5L/RAklJ+drRU1apoQER1X3zmFqmsCu0GQ GzPjjZ7+OqGai2kG7shvsIrwTz5uXMXBdzah6FIBQmfj+t7lrOtfcVaskbqkxrLXzSbwaOV mF+cV1P/S4A9wG7XHPTlB3PXDn0Vob9lKzgnbGRY1hNuGuqoVgH3YR/pXJRDd/t33yWxmdW +Z5q5gFYz7QQwBtphUPeItPLHF1cxQPzEurADANW8nVt0Tx4vD3uc/jfxiYxfGulbZRu4xJ Sic8YHYUUm9/vtOgpKW/ZTf119zCVn8f2ls3zFApwLnzVs3RTu7+73S0JOTfSFHa0QAUsLR 1Y1/DlPl4/IdQQX+zEqlxpHmCb7obK2v/WW243BCCcU/0vq0ogYskVJTCO1rZGBsh/6xVnd JflEpSYe+vpJ4+uvZRyzqPZYaM+IPgF3Klg0afvY6xazqlV5NrDXujNzT6IiIkDPjslS3p+ SJDtAYOWmIk4AhaFeii2fNcUmj6rBWIAGuhuAxebR1S7qsoix0nVYcqWEG1ozHl3z8q7m+2 HDC4NNNsovu71WLqsIkPOmT0JAZ2ZPskzgKpVMUcrzgRNy5U5ivwi4PoSRS4G+UhK8ceqCg pGyo+OpIOIMvV55Fdl8ftD0wG0rExNjP7i3dhQlfIRpSBzLWPA/GEw23pSF4Esv1PdVqOjq R8mvksQpMozfk1Laj0kWvmZ5vqwXjN4UPokniXYGP866yLdKGryZxwXOFDV3cjlTGDAiLsW uaE3TAFe1kQKW91rYNnL6YvbZLucV7MqZQDI/EjC97XG4ZQZBaZHlTtpWLh0oO110J/kaVi g== X-QQ-XMRINFO: Nq+8W0+stu50tPAe92KXseR0ZZmBTk3gLg== X-QQ-RECHKSPAM: 0 Content-Type: text/plain; charset="utf-8" Add interrupt setup and teardown for the RNPGBE driver. Require at least two MSI-X vectors. Reserve vector 0 for the mailbox and assign the remaining vectors to the data path. Set up mailbox workqueue plumbing for later asynchronous firmware event handling, while data-path interrupts schedule NAPI. Signed-off-by: Dong Yibo --- drivers/net/ethernet/mucse/rnpgbe/Makefile | 3 +- drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h | 47 ++ .../net/ethernet/mucse/rnpgbe/rnpgbe_chip.c | 4 + drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h | 2 + .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.c | 571 ++++++++++++++++++ .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.h | 34 ++ .../net/ethernet/mucse/rnpgbe/rnpgbe_main.c | 56 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c | 12 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c | 11 + .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h | 1 + 10 files changed, 731 insertions(+), 10 deletions(-) create mode 100644 drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c create mode 100644 drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h diff --git a/drivers/net/ethernet/mucse/rnpgbe/Makefile b/drivers/net/ether= net/mucse/rnpgbe/Makefile index de8bcb7772ab..17574cad392a 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/Makefile +++ b/drivers/net/ethernet/mucse/rnpgbe/Makefile @@ -8,4 +8,5 @@ obj-$(CONFIG_MGBE) +=3D rnpgbe.o rnpgbe-objs :=3D rnpgbe_main.o\ rnpgbe_chip.o\ rnpgbe_mbx.o\ - rnpgbe_mbx_fw.o + rnpgbe_mbx_fw.o\ + rnpgbe_lib.o diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h b/drivers/net/ether= net/mucse/rnpgbe/rnpgbe.h index 5b024f9f7e17..40e41521a279 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h @@ -6,6 +6,11 @@ =20 #include #include +#include +#include +#include + +#include "rnpgbe_hw.h" =20 enum rnpgbe_boards { board_n500, @@ -35,21 +40,61 @@ enum { =20 struct mucse_hw { void __iomem *hw_addr; + void __iomem *ring_msix_base; struct pci_dev *pdev; struct mucse_mbx_info mbx; int port; u8 pfvfnum; }; =20 +struct mucse_ring { + struct mucse_ring *next; + struct mucse_q_vector *q_vector; + void __iomem *ring_addr; + void __iomem *irq_mask; + void __iomem *trig; + u8 queue_index; + /* hw ring idx */ + u8 rnpgbe_queue_idx; +} ____cacheline_internodealigned_in_smp; + +struct mucse_ring_container { + struct mucse_ring *ring; + u16 count; +}; + +struct mucse_q_vector { + struct mucse *mucse; + /* hardware interrupt vector number */ + int hw_vector; + struct mucse_ring_container rx, tx; + struct napi_struct napi; + char name[IFNAMSIZ + 18]; + /* for dynamic allocation of rings associated with this q_vector */ + struct mucse_ring ring[] ____cacheline_internodealigned_in_smp; +}; + struct mucse_stats { u64 tx_dropped; }; =20 +#define MAX_Q_VECTORS 8 + struct mucse { struct net_device *netdev; struct pci_dev *pdev; struct mucse_hw hw; struct mucse_stats stats; + struct mucse_ring *tx_ring[RNPGBE_MAX_QUEUES] + ____cacheline_aligned_in_smp; + struct mucse_ring *rx_ring[RNPGBE_MAX_QUEUES] + ____cacheline_aligned_in_smp; + struct mucse_q_vector *q_vector[MAX_Q_VECTORS]; + int num_tx_queues; + int num_q_vectors; + int num_rx_queues; + char mbx_name[32]; + struct work_struct mbx_work; }; =20 int rnpgbe_get_permanent_mac(struct mucse_hw *hw, u8 *perm_addr); @@ -68,4 +113,6 @@ int rnpgbe_init_hw(struct mucse_hw *hw, int board_type); =20 #define mucse_hw_wr32(hw, reg, val) \ writel((val), (hw)->hw_addr + (reg)) +#define mucse_hw_rd32(hw, reg) \ + readl((hw)->hw_addr + (reg)) #endif /* _RNPGBE_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_chip.c index ebc7b3750157..921cc325a991 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c @@ -89,6 +89,8 @@ static void rnpgbe_init_n500(struct mucse_hw *hw) { struct mucse_mbx_info *mbx =3D &hw->mbx; =20 + hw->ring_msix_base =3D hw->hw_addr + MUCSE_N500_RING_MSIX_BASE; + mbx->fwpf_ctrl_base =3D MUCSE_N500_FWPF_CTRL_BASE; mbx->fwpf_shm_base =3D MUCSE_N500_FWPF_SHM_BASE; } @@ -104,6 +106,8 @@ static void rnpgbe_init_n210(struct mucse_hw *hw) { struct mucse_mbx_info *mbx =3D &hw->mbx; =20 + hw->ring_msix_base =3D hw->hw_addr + MUCSE_N210_RING_MSIX_BASE; + mbx->fwpf_ctrl_base =3D MUCSE_N210_FWPF_CTRL_BASE; mbx->fwpf_shm_base =3D MUCSE_N210_FWPF_SHM_BASE; } diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h b/drivers/net/et= hernet/mucse/rnpgbe/rnpgbe_hw.h index e77e6bc3d3e3..0dce78e4a91b 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h @@ -6,10 +6,12 @@ =20 #define MUCSE_N500_FWPF_CTRL_BASE 0x28b00 #define MUCSE_N500_FWPF_SHM_BASE 0x2d000 +#define MUCSE_N500_RING_MSIX_BASE 0x28700 #define MUCSE_GBE_PFFW_MBX_CTRL_OFFSET 0x5500 #define MUCSE_GBE_FWPF_MBX_MASK_OFFSET 0x5700 #define MUCSE_N210_FWPF_CTRL_BASE 0x29400 #define MUCSE_N210_FWPF_SHM_BASE 0x2d900 +#define MUCSE_N210_RING_MSIX_BASE 0x29000 =20 #define RNPGBE_DMA_AXI_EN 0x0010 =20 diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.c new file mode 100644 index 000000000000..9196fa130ca9 --- /dev/null +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c @@ -0,0 +1,571 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright(c) 2020 - 2025 Mucse Corporation. */ + +#include +#include + +#include "rnpgbe_lib.h" +#include "rnpgbe.h" +#include "rnpgbe_mbx_fw.h" + +static void rnpgbe_mbx_work(struct work_struct *work) +{ + struct mucse *mucse =3D container_of(work, struct mucse, mbx_work); + + mucse_fw_irq_handler(&mucse->hw); +} + +/** + * rnpgbe_msix_other - Other irq handler + * @irq: interrupt number + * @data: private data + * + * Return: IRQ_HANDLED + **/ +static irqreturn_t rnpgbe_msix_other(int irq, void *data) +{ + struct mucse *mucse =3D (struct mucse *)data; + + queue_work(system_percpu_wq, &mucse->mbx_work); + + return IRQ_HANDLED; +} + +static void rnpgbe_irq_disable_queues(struct mucse_q_vector *q_vector) +{ + struct mucse_ring *ring; + + /* + * TX/RX pairs share interrupt control registers; update them through + * the TX ring list. + */ + mucse_for_each_ring(ring, q_vector->tx) { + writel(INT_VALID, ring->trig); + writel((RX_INT_MASK | TX_INT_MASK), ring->irq_mask); + } + + /* flush posted writes to ensure hardware sees the mask */ + if (q_vector->tx.ring) + readl(q_vector->tx.ring->irq_mask); +} + +static void rnpgbe_irq_enable_queues(struct mucse_q_vector *q_vector) +{ + struct mucse_ring *ring; + + /* + * TX/RX pairs share interrupt control registers; update them through + * the TX ring list. + */ + mucse_for_each_ring(ring, q_vector->tx) { + writel(0, ring->irq_mask); + + /* Re-trigger hw to re-check events lost while masked. */ + writel(INT_VALID | TX_INT_MASK | RX_INT_MASK, ring->trig); + } +} + +/** + * rnpgbe_poll - NAPI polling callback + * @napi: structure for representing this polling device + * @budget: polling budget + * + * Complete NAPI polling and re-enable queue interrupts. Ring cleaning is + * added when TX and RX support is enabled. + * + * Return: 0 + **/ +static int rnpgbe_poll(struct napi_struct *napi, int budget) +{ + struct mucse_q_vector *q_vector =3D + container_of(napi, struct mucse_q_vector, napi); + int work_done =3D 0; + + /* Exit if we are called by netpoll */ + if (unlikely(!budget)) + return 0; + + if (likely(napi_complete_done(napi, work_done))) + rnpgbe_irq_enable_queues(q_vector); + + return work_done; +} + +/** + * rnpgbe_request_mbx_irq - Register mbx routine + * @mucse: pointer to private structure + * + * Register the dedicated mailbox handler on vector 0. + * + * Return: 0 on success, negative on failure + **/ +int rnpgbe_request_mbx_irq(struct mucse *mucse) +{ + struct pci_dev *pdev =3D mucse->pdev; + int err =3D 0; + + snprintf(mucse->mbx_name, sizeof(mucse->mbx_name), + "rnpgbe-mbx:%s", pci_name(pdev)); + INIT_WORK(&mucse->mbx_work, rnpgbe_mbx_work); + + err =3D request_irq(pci_irq_vector(pdev, 0), rnpgbe_msix_other, 0, + mucse->mbx_name, mucse); + + return err; +} + +/** + * rnpgbe_free_mbx_irq - Remove mbx routine + * @mucse: pointer to private structure + **/ +void rnpgbe_free_mbx_irq(struct mucse *mucse) +{ + struct pci_dev *pdev =3D mucse->pdev; + + free_irq(pci_irq_vector(pdev, 0), mucse); + cancel_work_sync(&mucse->mbx_work); +} + +/** + * rnpgbe_set_num_queues - Set initial TX/RX queue counts + * @mucse: pointer to private structure + * + * Start with one TX queue and one RX queue. + **/ +static void rnpgbe_set_num_queues(struct mucse *mucse) +{ + mucse->num_tx_queues =3D 1; + mucse->num_rx_queues =3D 1; +} + +/** + * rnpgbe_set_interrupt_capability - Set MSI-X interrupt capability + * @mucse: pointer to private structure + * + * Attempt to configure the interrupts using the best available + * capabilities of the hardware. + * + * Return: 0 on success, negative on failure + **/ +static int rnpgbe_set_interrupt_capability(struct mucse *mucse) +{ + int v_budget; + + v_budget =3D min3(mucse->num_tx_queues, mucse->num_rx_queues, + MAX_Q_VECTORS); + v_budget =3D min_t(int, v_budget, num_online_cpus()); + /* add one vector for mbx */ + v_budget +=3D 1; + + /* Vector 0 is reserved for mailbox events and must not share NAPI. */ + v_budget =3D pci_alloc_irq_vectors(mucse->pdev, 2, v_budget, + PCI_IRQ_MSIX); + if (v_budget < 0) + return v_budget; + + /* q_vectors do not include the mailbox vector. */ + mucse->num_q_vectors =3D v_budget - 1; + + return 0; +} + +/** + * mucse_add_ring - Add ring to ring container + * @ring: ring to be added + * @head: ring container + **/ +static void mucse_add_ring(struct mucse_ring *ring, + struct mucse_ring_container *head) +{ + ring->next =3D head->ring; + head->ring =3D ring; + head->count++; +} + +/** + * rnpgbe_alloc_q_vector - Allocate memory for a single interrupt vector + * @mucse: pointer to private structure + * @eth_queue_idx: queue_index idx for this q_vector + * @vector_idx: q_vector array index + * @r_idx: starting hardware ring index + * @r_count: number of TX/RX ring pairs + * @step: ring step + * + * Return: 0 on success. If allocation fails we return -ENOMEM. + **/ +static int rnpgbe_alloc_q_vector(struct mucse *mucse, + int eth_queue_idx, int vector_idx, int r_idx, + int r_count, int step) +{ + int rxr_idx =3D r_idx, txr_idx =3D r_idx; + struct mucse_hw *hw =3D &mucse->hw; + struct mucse_q_vector *q_vector; + int txr_count, rxr_count, idx; + struct mucse_ring *ring; + int ring_count; + + /* + * TX and RX rings are always allocated as pairs with the same hardware + * ring index. Interrupt control is shared by the pair, so the TX ring + * list is used to update the common registers. + */ + txr_count =3D r_count; + rxr_count =3D r_count; + ring_count =3D txr_count + rxr_count; + + q_vector =3D kzalloc_flex(*q_vector, ring, ring_count); + if (!q_vector) + return -ENOMEM; + + netif_napi_add(mucse->netdev, &q_vector->napi, rnpgbe_poll); + /* tie q_vector and mucse together */ + mucse->q_vector[vector_idx] =3D q_vector; + q_vector->mucse =3D mucse; + q_vector->hw_vector =3D vector_idx; + /* Vector 0 is reserved for the mailbox. */ + q_vector->hw_vector++; + + ring =3D q_vector->ring; + + for (idx =3D 0; idx < txr_count; idx++) { + mucse_add_ring(ring, &q_vector->tx); + ring->queue_index =3D eth_queue_idx + idx; + ring->rnpgbe_queue_idx =3D txr_idx; + ring->ring_addr =3D hw->hw_addr + RING_OFFSET(txr_idx); + ring->irq_mask =3D ring->ring_addr + RNPGBE_DMA_INT_MASK; + ring->trig =3D ring->ring_addr + RNPGBE_DMA_INT_TRIG; + ring->q_vector =3D q_vector; + mucse->tx_ring[ring->queue_index] =3D ring; + txr_idx +=3D step; + ring++; + } + + for (idx =3D 0; idx < rxr_count; idx++) { + mucse_add_ring(ring, &q_vector->rx); + ring->queue_index =3D eth_queue_idx + idx; + ring->rnpgbe_queue_idx =3D rxr_idx; + ring->ring_addr =3D hw->hw_addr + RING_OFFSET(rxr_idx); + ring->irq_mask =3D ring->ring_addr + RNPGBE_DMA_INT_MASK; + ring->trig =3D ring->ring_addr + RNPGBE_DMA_INT_TRIG; + ring->q_vector =3D q_vector; + mucse->rx_ring[ring->queue_index] =3D ring; + rxr_idx +=3D step; + ring++; + } + + return 0; +} + +/** + * rnpgbe_free_q_vector - Free memory allocated for specific interrupt vec= tor + * @mucse: pointer to private structure + * @vector_idx: q_vector array index + * + * 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 rnpgbe_free_q_vector(struct mucse *mucse, int vector_idx) +{ + struct mucse_q_vector *q_vector =3D mucse->q_vector[vector_idx]; + struct mucse_ring *ring; + + mucse_for_each_ring(ring, q_vector->tx) + mucse->tx_ring[ring->queue_index] =3D NULL; + mucse_for_each_ring(ring, q_vector->rx) + mucse->rx_ring[ring->queue_index] =3D NULL; + mucse->q_vector[vector_idx] =3D NULL; + netif_napi_del(&q_vector->napi); + kfree(q_vector); +} + +/** + * rnpgbe_alloc_q_vectors - Allocate memory for interrupt vectors + * @mucse: pointer to private structure + * + * Return: 0 on success, or -ENOMEM on allocation failure. + **/ +static int rnpgbe_alloc_q_vectors(struct mucse *mucse) +{ + int err, ring_cnt, v_remaining =3D mucse->num_q_vectors; + int r_remaining =3D min_t(int, mucse->num_tx_queues, + mucse->num_rx_queues); + int num_q_vectors =3D 0; + int eth_queue_idx =3D 0; + int vector_idx =3D 0; + int ring_step =3D 1; + int ring_idx =3D 0; + + for (; r_remaining > 0 && v_remaining > 0; v_remaining--) { + ring_cnt =3D DIV_ROUND_UP(r_remaining, v_remaining); + err =3D rnpgbe_alloc_q_vector(mucse, eth_queue_idx, + vector_idx, ring_idx, ring_cnt, + ring_step); + if (err) + goto err_free_q_vector; + ring_idx +=3D ring_step * ring_cnt; + eth_queue_idx +=3D ring_cnt; + r_remaining -=3D ring_cnt; + num_q_vectors++; + vector_idx++; + } + /* Record the number of queue vectors actually allocated */ + mucse->num_q_vectors =3D num_q_vectors; + mucse->num_tx_queues =3D eth_queue_idx; + mucse->num_rx_queues =3D eth_queue_idx; + + return 0; + +err_free_q_vector: + mucse->num_tx_queues =3D 0; + mucse->num_rx_queues =3D 0; + mucse->num_q_vectors =3D 0; + + while (vector_idx--) + rnpgbe_free_q_vector(mucse, vector_idx); + + return err; +} + +/** + * rnpgbe_reset_interrupt_capability - Reset irq capability setup + * @mucse: pointer to private structure + **/ +static void rnpgbe_reset_interrupt_capability(struct mucse *mucse) +{ + pci_free_irq_vectors(mucse->pdev); +} + +/** + * rnpgbe_init_interrupt_scheme - Initialize the interrupt scheme + * @mucse: pointer to private structure + * + * Set up one TX/RX queue pair with a dedicated mailbox vector. + * + * Return: 0 on success, negative on failure + **/ +int rnpgbe_init_interrupt_scheme(struct mucse *mucse) +{ + int err; + + rnpgbe_set_num_queues(mucse); + + err =3D rnpgbe_set_interrupt_capability(mucse); + if (err) + return err; + + err =3D rnpgbe_alloc_q_vectors(mucse); + if (err) { + rnpgbe_reset_interrupt_capability(mucse); + return err; + } + + return 0; +} + +/** + * rnpgbe_free_q_vectors - Free memory allocated for interrupt vectors + * @mucse: pointer to private structure + * + * 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 rnpgbe_free_q_vectors(struct mucse *mucse) +{ + int vector_idx =3D mucse->num_q_vectors; + + mucse->num_q_vectors =3D 0; + + while (vector_idx--) + rnpgbe_free_q_vector(mucse, vector_idx); +} + +/** + * rnpgbe_clear_interrupt_scheme - Clear the current interrupt scheme sett= ings + * @mucse: pointer to private structure + * + * Clear interrupt specific resources and reset the structure + **/ +void rnpgbe_clear_interrupt_scheme(struct mucse *mucse) +{ + rnpgbe_free_q_vectors(mucse); + rnpgbe_reset_interrupt_capability(mucse); +} + +/** + * rnpgbe_msix_clean_rings - MSI-x interrupt handler for ring irq + * @irq: interrupt number + * @data: private data + * + * rnpgbe_msix_clean_rings handle irq from ring, start napi + * Return: IRQ_HANDLED + **/ +static irqreturn_t rnpgbe_msix_clean_rings(int irq, void *data) +{ + struct mucse_q_vector *q_vector =3D (struct mucse_q_vector *)data; + + rnpgbe_irq_disable_queues(q_vector); + napi_schedule_irqoff(&q_vector->napi); + + return IRQ_HANDLED; +} + +/** + * rnpgbe_request_irq - Initialize interrupts + * @mucse: pointer to private structure + * + * Attempts to configure interrupts using the best available + * capabilities of the hardware and kernel. + * + * Return: 0 on success, negative value on failure + **/ +int rnpgbe_request_irq(struct mucse *mucse) +{ + struct net_device *netdev =3D mucse->netdev; + struct pci_dev *pdev =3D mucse->pdev; + struct mucse_q_vector *q_vector; + int err, i; + + for (i =3D 0; i < mucse->num_q_vectors; i++) { + q_vector =3D mucse->q_vector[i]; + + snprintf(q_vector->name, sizeof(q_vector->name), + "%s-%s-%d", netdev->name, "TxRx", i); + + err =3D request_irq(pci_irq_vector(pdev, i + 1), + rnpgbe_msix_clean_rings, 0, + q_vector->name, q_vector); + if (err) { + dev_err(&pdev->dev, "MSI-X req err %d: %d\n", + i + 1, err); + goto err_free_irqs; + } + } + + return 0; +err_free_irqs: + while (i--) { + q_vector =3D mucse->q_vector[i]; + synchronize_irq(pci_irq_vector(pdev, i + 1)); + free_irq(pci_irq_vector(pdev, i + 1), q_vector); + } + + return err; +} + +/** + * rnpgbe_free_irq - Free interrupts + * @mucse: pointer to private structure + * + * Attempts to free interrupts according initialized type. + **/ +void rnpgbe_free_irq(struct mucse *mucse) +{ + struct pci_dev *pdev =3D mucse->pdev; + struct mucse_q_vector *q_vector; + + for (int i =3D 0; i < mucse->num_q_vectors; i++) { + q_vector =3D mucse->q_vector[i]; + if (!q_vector) + continue; + + free_irq(pci_irq_vector(pdev, i + 1), q_vector); + } +} + +/** + * rnpgbe_set_ring_vector - Set the ring_vector registers, + * mapping interrupt causes to vectors + * @mucse: pointer to private structure + * @queue: queue to map the corresponding interrupt to + * @vector: the vector num to map to the corresponding queue + * + */ +static void rnpgbe_set_ring_vector(struct mucse *mucse, + u8 queue, u8 vector) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 data; + + data =3D hw->pfvfnum << 24; + data |=3D (vector << 8); + data |=3D vector; + writel(data, hw->ring_msix_base + RING_VECTOR(queue)); +} + +/** + * rnpgbe_configure_msix - Configure MSI-X hardware + * @mucse: pointer to private structure + * + * rnpgbe_configure_msix sets up the hardware to properly generate MSI-X + * interrupts. + **/ +static void rnpgbe_configure_msix(struct mucse *mucse) +{ + struct mucse_q_vector *q_vector; + + for (int i =3D 0; i < mucse->num_q_vectors; i++) { + struct mucse_ring *ring; + + q_vector =3D mucse->q_vector[i]; + /* TX/RX pairs share vector registers. + * Update them through TX. + */ + mucse_for_each_ring(ring, q_vector->tx) { + rnpgbe_set_ring_vector(mucse, ring->rnpgbe_queue_idx, + q_vector->hw_vector); + } + } +} + +static void rnpgbe_irq_enable(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_q_vectors; i++) + rnpgbe_irq_enable_queues(mucse->q_vector[i]); +} + +/** + * rnpgbe_irq_disable - Mask queue interrupt generation on the NIC + * @mucse: board private structure + **/ +void rnpgbe_irq_disable(struct mucse *mucse) +{ + struct pci_dev *pdev =3D mucse->pdev; + + for (int i =3D 0; i < mucse->num_q_vectors; i++) { + rnpgbe_irq_disable_queues(mucse->q_vector[i]); + synchronize_irq(pci_irq_vector(pdev, i + 1)); + } +} + +static void rnpgbe_napi_enable_all(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_q_vectors; i++) + napi_enable(&mucse->q_vector[i]->napi); +} + +static void rnpgbe_napi_disable_all(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_q_vectors; i++) + napi_disable(&mucse->q_vector[i]->napi); +} + +void rnpgbe_down(struct mucse *mucse) +{ + rnpgbe_irq_disable(mucse); + rnpgbe_napi_disable_all(mucse); + synchronize_net(); + rnpgbe_irq_disable(mucse); +} + +/** + * rnpgbe_up_complete - Final step for port up + * @mucse: pointer to private structure + **/ +void rnpgbe_up_complete(struct mucse *mucse) +{ + rnpgbe_configure_msix(mucse); + rnpgbe_napi_enable_all(mucse); + rnpgbe_irq_enable(mucse); +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.h new file mode 100644 index 000000000000..d38c08df6b7c --- /dev/null +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h @@ -0,0 +1,34 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright(c) 2020 - 2025 Mucse Corporation. */ + +#ifndef _RNPGBE_LIB_H +#define _RNPGBE_LIB_H + +struct mucse; +struct mucse_hw; + +#define RING_OFFSET(n) (0x1000 + 0x100 * (n)) +#define RNPGBE_DMA_INT_MASK 0x24 +#define TX_INT_MASK BIT(1) +#define RX_INT_MASK BIT(0) +#define INT_VALID (BIT(16) | BIT(17)) +#define RNPGBE_DMA_INT_TRIG 0x2c /* lost-interrupt recovery trigger = */ +/* | 31:24 | .... | 15:8 | 7:0 | */ +/* | pfvfnum | | tx vector | rx vector | */ +#define RING_VECTOR(n) (0x04 * (n)) + +#define mucse_for_each_ring(pos, head)\ + for (typeof((head).ring) __pos =3D (head).ring;\ + __pos ? ({ pos =3D __pos; 1; }) : 0;\ + __pos =3D __pos->next) + +int rnpgbe_init_interrupt_scheme(struct mucse *mucse); +void rnpgbe_clear_interrupt_scheme(struct mucse *mucse); +int rnpgbe_request_mbx_irq(struct mucse *mucse); +void rnpgbe_free_mbx_irq(struct mucse *mucse); +int rnpgbe_request_irq(struct mucse *mucse); +void rnpgbe_free_irq(struct mucse *mucse); +void rnpgbe_irq_disable(struct mucse *mucse); +void rnpgbe_down(struct mucse *mucse); +void rnpgbe_up_complete(struct mucse *mucse); +#endif diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_main.c index 70a2b0082ba8..debed2f3fe64 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c @@ -7,6 +7,7 @@ =20 #include "rnpgbe.h" #include "rnpgbe_hw.h" +#include "rnpgbe_lib.h" #include "rnpgbe_mbx_fw.h" =20 static const char rnpgbe_driver_name[] =3D "rnpgbe"; @@ -32,11 +33,28 @@ static struct pci_device_id rnpgbe_pci_tbl[] =3D { * The open entry point is called when a network interface is made * active by the system (IFF_UP). * - * Return: 0 + * Return: 0 on success, negative value on failure **/ static int rnpgbe_open(struct net_device *netdev) { + struct mucse *mucse =3D netdev_priv(netdev); + int err; + + err =3D rnpgbe_request_irq(mucse); + if (err) + return err; + + err =3D netif_set_real_num_queues(netdev, mucse->num_tx_queues, + mucse->num_rx_queues); + if (err) + goto err_free_irqs; + + rnpgbe_up_complete(mucse); + return 0; +err_free_irqs: + rnpgbe_free_irq(mucse); + return err; } =20 /** @@ -50,6 +68,11 @@ static int rnpgbe_open(struct net_device *netdev) **/ static int rnpgbe_close(struct net_device *netdev) { + struct mucse *mucse =3D netdev_priv(netdev); + + rnpgbe_down(mucse); + rnpgbe_free_irq(mucse); + return 0; } =20 @@ -166,11 +189,33 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, goto err_powerdown; } =20 + err =3D rnpgbe_init_interrupt_scheme(mucse); + if (err) { + dev_err(&pdev->dev, "init interrupt failed %d\n", err); + goto err_powerdown; + } + + err =3D netif_set_real_num_queues(netdev, mucse->num_tx_queues, + mucse->num_rx_queues); + if (err) + goto err_clear_interrupt; + + err =3D rnpgbe_request_mbx_irq(mucse); + if (err) { + dev_err(&pdev->dev, "register mbx irq failed %d\n", err); + goto err_clear_interrupt; + } + err =3D register_netdev(netdev); if (err) - goto err_powerdown; + goto err_remove_mbx; =20 return 0; + +err_remove_mbx: + rnpgbe_free_mbx_irq(mucse); +err_clear_interrupt: + rnpgbe_clear_interrupt_scheme(mucse); err_powerdown: /* notify powerdown only powerup ok */ if (!err_notify) { @@ -253,9 +298,11 @@ static void rnpgbe_rm_adapter(struct pci_dev *pdev) return; netdev =3D mucse->netdev; unregister_netdev(netdev); + rnpgbe_free_mbx_irq(mucse); err =3D rnpgbe_send_notify(hw, false, mucse_fw_powerup); if (err) dev_warn(&pdev->dev, "Send powerdown to hw failed %d\n", err); + rnpgbe_clear_interrupt_scheme(mucse); free_netdev(netdev); } =20 @@ -287,8 +334,11 @@ static void rnpgbe_dev_shutdown(struct pci_dev *pdev) rtnl_lock(); netif_device_detach(netdev); if (netif_running(netdev)) - rnpgbe_close(netdev); + dev_close(netdev); rtnl_unlock(); + + rnpgbe_free_mbx_irq(mucse); + rnpgbe_clear_interrupt_scheme(mucse); pci_disable_device(pdev); } =20 diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_mbx.c index c46408698263..2310f01df1f5 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c @@ -84,7 +84,7 @@ static u32 mucse_mbx_get_lock_pf(struct mucse_hw *hw) * @hw: pointer to the HW structure * * Pair with mucse_release_mbx_lock_pf() - * This function maybe used in an irq handler. + * All mailbox access runs in process context. * * Return: 0 on success, negative errno on failure **/ @@ -93,11 +93,11 @@ static int mucse_obtain_mbx_lock_pf(struct mucse_hw *hw) struct mucse_mbx_info *mbx =3D &hw->mbx; u32 val; =20 - return read_poll_timeout_atomic(mucse_mbx_get_lock_pf, - val, val & MUCSE_MBX_PFU, - mbx->delay_us, - mbx->timeout_us, - false, hw); + return read_poll_timeout(mucse_mbx_get_lock_pf, + val, val & MUCSE_MBX_PFU, + mbx->delay_us, + mbx->timeout_us, + false, hw); } =20 /** diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c b/drivers/ne= t/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c index 5ba74997beac..e28f3193aa85 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c @@ -197,3 +197,14 @@ int mucse_mbx_get_macaddr(struct mucse_hw *hw, int pfv= fnum, =20 return 0; } + +/** + * mucse_fw_irq_handler - Deferred firmware mailbox notification hook + * @hw: pointer to the HW structure + * + * No asynchronous firmware events are enabled until link status support is + * added. + **/ +void mucse_fw_irq_handler(struct mucse_hw *hw) +{ +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h b/drivers/ne= t/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h index fe996aeffc4d..a6bf5de55aa3 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h @@ -99,4 +99,5 @@ int mucse_mbx_powerup(struct mucse_hw *hw, bool is_poweru= p); int mucse_mbx_reset_hw(struct mucse_hw *hw); int mucse_mbx_get_macaddr(struct mucse_hw *hw, int pfvfnum, u8 *mac_addr, int port); +void mucse_fw_irq_handler(struct mucse_hw *hw); #endif /* _RNPGBE_MBX_FW_H */ --=20 2.50.1 From nobody Thu Sep 24 20:41:59 2026 Received: from smtpbgau1.qq.com (smtpbgau1.qq.com [54.206.16.166]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id D98A834A3BF; Sun, 20 Sep 2026 09:27:39 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=54.206.16.166 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896464; cv=none; b=MeDvkMbvqc1h8w/vfE8+1iKRBDNWZKodimc0/syXT2qtkM1yiCq3U7QoWtV1Vp2FDtnQMFXbyVh3bc/s2fJ6cunwBXtcllsgtdokEhFqnfgvo7dpfR+m4Zz42ClwEaV9cYEEDE/1px9w6LkSS8hug3fpdeV74u8BPQvlekRZ+8I= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896464; c=relaxed/simple; bh=HGLEaLNxkAi/zUauLB0WR+eycFdbZwuIxbkKH4VpU+k=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=AIuwi2MYyisbjNnb2R2Iwxk42w5ZnKHsmpgMr/qkz/xxXVd+uRF188zlFNAhR9iTMd1/otSMXrci9ND8T618qLFut2/8UQVi1jWDSq5oOjmmnUYy/v/0Lqe5PqqbfiVthiD56pQbJ67eJdIt9ABVljoFuLfytQBaNvSEfqN4zYE= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com; spf=pass smtp.mailfrom=mucse.com; arc=none smtp.client-ip=54.206.16.166 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mucse.com X-QQ-mid: esmtpgz13t1789896386tceb340d0 X-QQ-Originating-IP: bhLS54ZidWpnn1SlWEFYL7zp0mIAqp/iavWA0udlzAQ= Received: from nic-Precision-5820-Tower ( [203.174.112.180]) by bizesmtp.qq.com (ESMTP) with id ; Sun, 20 Sep 2026 17:26:24 +0800 (CST) X-QQ-SSF: 0000000000000000000000000000000 X-QQ-GoodBg: 0 X-BIZMAIL-ID: 3566157444938278617 EX-QQ-RecipientCnt: 11 From: Dong Yibo To: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, vadim.fedorenko@linux.dev, u.kleine-koenig@baylibre.com Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, dong100@mucse.com, yaojun@mucse.com Subject: [PATCH net-next v13 2/5] net: rnpgbe: Add basic TX packet transmission support Date: Sun, 20 Sep 2026 17:26:22 +0800 Message-ID: <5CF5340DA3D4CE3C+20260920092622.345496-1-dong100@mucse.com> X-Mailer: git-send-email 2.50.1 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-QQ-SENDSIZE: 520 Feedback-ID: esmtpgz:mucse.com:qybglogicsvrgz:qybglogicsvrgz3a-1 X-QQ-XMAILINFO: MwqrwaLzgdebJwAypfET9qx5hrZa1J1kotnLJB5V5lXiApJBO6lSLXFT NQO8rprq8bJTPDzBKQ8SHbmHpzAPzRlNJxJzKnVfkuwZLAEfU+ZzMN0zfdGJHcD5WAQlmmc zBPZ8v4W/l2VWirWe5dlksL20yQWeotFQJIqw034ctrQLKzNtYgOy22/J+GsRWkwqcCQIeg +0BXeY1wu4nkRm+Iwl5KUYS4um9WynUUzjmNpJxy1EhtDq+zdeebMaKRsCW0+TGxnoh80iy kTcTY/wIxe06BRBSCA7IzRi9ZSyYM+7VidhbJX0zETi9nGIn9kZdGQMRaxMCZAElOPCV9Fy uU17J4vZM7qPbAeIS98HPQb6G/nhlniJdM9v7V8znMJJzLRATsVy1w147d1vrTv9aX8yVLs pMUM5kbI3AIIUaUsZBIhF88eO1AnYS0pR7oWuWUzEUjweTw7y48KSH5f20E8Y+hgUVcOmJe Ni7iDaVRAZIk/8kX1ajH/Ma+pW963fpetbiUaefgVJaVGtqOaHeTDF3FuovXKTI1grKjZ/9 CwYW0ONOzpHn61xRnm2IbkvEe0JqIoSvUEpzD0/vWZdkDzGZfhoxpuN+rjDjyTI/gvUacR+ yDlTBqc8md+JV2M6UauKlY4OI9hajuosL3iSA40MFBDIdmi+mtTRch4F3D4bOdAxMrQZsqH urCNE5e28+3dBUrLjKo9UBIjCtZHM2D5sq/ga/dRl0tZhd/EfBWgu+WsqPOPdt1h/ZSvZoI aarkNN39kOPvCpypVYJyDMMsot7u5ilX676DPSNuySSjaUwqN0xx+gUbOkXSrqwdd2v8ZEa dzYPquM+JqB927Fj6Ohgy9ZDZiZJITwu+ubuobexalMBPPO2Y8CbQDTW61fxi+p/si2LMSK UB/6SspPWoVihkR/aBFKaLVb5vyX08V5W8r6Us7r00tOwvxy3Z7g29RthrKx8B31cZaDv5b N61EbaU1KCmAqQPLTzMRdNf71wKTulltlhmutPm02BXFyDVmetWhBSwKnfbaq7nsLdrDyC5 TFXoeIZQT9Fem6axHd3wnTdcbkf6uCjfHPwsgJmEVsGqF0d3HWtkPghjYM2P8= X-QQ-XMRINFO: NI4Ajvh11aEjEMj13RCX7UuhPEoou2bs1g== X-QQ-RECHKSPAM: 0 Content-Type: text/plain; charset="utf-8" Add the basic TX path for the RNPGBE driver. Introduce the TX descriptor structure and buffer management, the rnpgbe_xmit_frame_ring() transmit function, TX ring allocation and teardown, and TX completion handling in rnpgbe_clean_tx_irq(). Track TX packet and byte statistics, and report dropped and tx_dropped counts through ndo_get_stats64(). Configure streaming and coherent DMA masks for the hardware's 56-bit DMA addressing capability. Add cycles_per_us to struct mucse_hw to scale the TX interrupt timer, and enable NETIF_F_HIGHDMA when the DMA mask allows it. A TX DMA idle timeout is terminal for the current driver instance. The hardware stops PCIe DMA requests, allowing cleanup to release DMA resources. Recovery requires a chip-level reset and driver rebind. Signed-off-by: Dong Yibo --- drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h | 86 ++- .../net/ethernet/mucse/rnpgbe/rnpgbe_chip.c | 4 + drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h | 7 + .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.c | 709 +++++++++++++++++- .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.h | 30 + .../net/ethernet/mucse/rnpgbe/rnpgbe_main.c | 63 +- 6 files changed, 875 insertions(+), 24 deletions(-) diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h b/drivers/net/ether= net/mucse/rnpgbe/rnpgbe.h index 40e41521a279..864cbf0caf44 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h @@ -7,6 +7,7 @@ #include #include #include +#include #include #include =20 @@ -44,20 +45,85 @@ struct mucse_hw { struct pci_dev *pdev; struct mucse_mbx_info mbx; int port; + u16 cycles_per_us; u8 pfvfnum; }; =20 +struct rnpgbe_tx_desc { + __le64 pkt_addr; /* Packet buffer address */ + union { + __le64 vlan_cmd_bsz; + struct { + __le32 blen_mac_ip_len; + __le32 vlan_cmd; /* vlan & cmd status */ + }; + }; +#define M_TXD_CMD_RS 0x040000 /* Report Status */ +#define M_TXD_STAT_DD 0x020000 /* Descriptor Done */ +#define M_TXD_CMD_EOP 0x010000 /* End of Packet */ +}; + +#define M_TX_DESC(R, i) (&(((struct rnpgbe_tx_desc *)((R)->desc))[i])) + +struct mucse_tx_buffer { + struct rnpgbe_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); + bool mapped_as_page; /* true if dma was mapped with dma_map_page */ +}; + +struct mucse_queue_stats { + u64 packets; + u64 bytes; + atomic64_t dropped; +}; + struct mucse_ring { struct mucse_ring *next; struct mucse_q_vector *q_vector; + struct net_device *netdev; + struct device *dev; + void *desc; + struct mucse_tx_buffer *tx_buffer_info; void __iomem *ring_addr; + void __iomem *tail; void __iomem *irq_mask; void __iomem *trig; u8 queue_index; /* hw ring idx */ u8 rnpgbe_queue_idx; + u8 pfvfnum; + u16 count; + u16 next_to_use; + u16 next_to_clean; + dma_addr_t dma; + unsigned int size; + struct mucse_queue_stats stats; + struct u64_stats_sync syncp; } ____cacheline_internodealigned_in_smp; =20 +static inline u16 mucse_desc_unused(struct mucse_ring *ring) +{ + u16 ntc =3D ring->next_to_clean; + u16 ntu =3D ring->next_to_use; + + return ((ntc > ntu) ? 0 : ring->count) + ntc - ntu - 1; +} + +static inline __le64 build_ctob(u32 vlan_cmd, u32 mac_ip_len, u32 size) +{ + return cpu_to_le64(((u64)vlan_cmd << 32) | ((u64)mac_ip_len << 16) | + ((u64)size)); +} + +static inline struct netdev_queue *txring_txq(const struct mucse_ring *rin= g) +{ + return netdev_get_tx_queue(ring->netdev, ring->queue_index); +} + struct mucse_ring_container { struct mucse_ring *ring; u16 count; @@ -69,31 +135,37 @@ struct mucse_q_vector { int hw_vector; struct mucse_ring_container rx, tx; struct napi_struct napi; + struct rcu_head rcu; char name[IFNAMSIZ + 18]; /* for dynamic allocation of rings associated with this q_vector */ struct mucse_ring ring[] ____cacheline_internodealigned_in_smp; }; =20 -struct mucse_stats { - u64 tx_dropped; -}; - #define MAX_Q_VECTORS 8 =20 +#define M_DEFAULT_TXD 512 +#define M_DEFAULT_TX_WORK 256 + +enum mucse_state_t { + __MUCSE_AXI_FAULT, +}; + struct mucse { struct net_device *netdev; struct pci_dev *pdev; struct mucse_hw hw; - struct mucse_stats stats; struct mucse_ring *tx_ring[RNPGBE_MAX_QUEUES] ____cacheline_aligned_in_smp; struct mucse_ring *rx_ring[RNPGBE_MAX_QUEUES] ____cacheline_aligned_in_smp; struct mucse_q_vector *q_vector[MAX_Q_VECTORS]; + int tx_ring_item_count; + int tx_work_limit; int num_tx_queues; int num_q_vectors; int num_rx_queues; char mbx_name[32]; + unsigned long state; struct work_struct mbx_work; }; =20 @@ -115,4 +187,8 @@ int rnpgbe_init_hw(struct mucse_hw *hw, int board_type); writel((val), (hw)->hw_addr + (reg)) #define mucse_hw_rd32(hw, reg) \ readl((hw)->hw_addr + (reg)) +#define mucse_ring_wr32(ring, reg, val) \ + writel((val), (ring)->ring_addr + (reg)) +#define mucse_ring_rd32(ring, reg) \ + readl((ring)->ring_addr + (reg)) #endif /* _RNPGBE_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_chip.c index 921cc325a991..291e77d573fe 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c @@ -93,6 +93,8 @@ static void rnpgbe_init_n500(struct mucse_hw *hw) =20 mbx->fwpf_ctrl_base =3D MUCSE_N500_FWPF_CTRL_BASE; mbx->fwpf_shm_base =3D MUCSE_N500_FWPF_SHM_BASE; + + hw->cycles_per_us =3D M_DEFAULT_N500_MHZ; } =20 /** @@ -110,6 +112,8 @@ static void rnpgbe_init_n210(struct mucse_hw *hw) =20 mbx->fwpf_ctrl_base =3D MUCSE_N210_FWPF_CTRL_BASE; mbx->fwpf_shm_base =3D MUCSE_N210_FWPF_SHM_BASE; + + hw->cycles_per_us =3D M_DEFAULT_N210_MHZ; } =20 /** diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h b/drivers/net/et= hernet/mucse/rnpgbe/rnpgbe_hw.h index 0dce78e4a91b..6dc29ebe6fa7 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h @@ -7,13 +7,20 @@ #define MUCSE_N500_FWPF_CTRL_BASE 0x28b00 #define MUCSE_N500_FWPF_SHM_BASE 0x2d000 #define MUCSE_N500_RING_MSIX_BASE 0x28700 +#define M_DEFAULT_N500_MHZ 125 #define MUCSE_GBE_PFFW_MBX_CTRL_OFFSET 0x5500 #define MUCSE_GBE_FWPF_MBX_MASK_OFFSET 0x5700 #define MUCSE_N210_FWPF_CTRL_BASE 0x29400 #define MUCSE_N210_FWPF_SHM_BASE 0x2d900 #define MUCSE_N210_RING_MSIX_BASE 0x29000 +#define M_DEFAULT_N210_MHZ 62 =20 +#define RNPGBE_DMA_STATUS 0x0008 +#define TX_AXI_RW_EN 0xc +/* DMA_STATUS_REG[23:20]: tx_wr, tx_rd, rx_wr, rx_rd done status. */ +#define RNPGBE_DMA_TX_STATUS GENMASK_U32(23, 22) #define RNPGBE_DMA_AXI_EN 0x0010 +#define RNPGBE_TX_MIN_PKT_LEN 33 =20 #define RNPGBE_MAX_QUEUES 8 #endif /* _RNPGBE_HW_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.c index 9196fa130ca9..af5cef987051 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c @@ -3,6 +3,9 @@ =20 #include #include +#include +#include +#include =20 #include "rnpgbe_lib.h" #include "rnpgbe.h" @@ -43,7 +46,6 @@ static void rnpgbe_irq_disable_queues(struct mucse_q_vect= or *q_vector) writel(INT_VALID, ring->trig); writel((RX_INT_MASK | TX_INT_MASK), ring->irq_mask); } - /* flush posted writes to ensure hardware sees the mask */ if (q_vector->tx.ring) readl(q_vector->tx.ring->irq_mask); @@ -65,26 +67,143 @@ static void rnpgbe_irq_enable_queues(struct mucse_q_ve= ctor *q_vector) } } =20 +/** + * rnpgbe_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 + * + * Return: true if cleaning completed within the budget, otherwise false + **/ +static bool rnpgbe_clean_tx_irq(struct mucse_q_vector *q_vector, + struct mucse_ring *tx_ring, + int napi_budget) +{ + int budget =3D q_vector->mucse->tx_work_limit; + u64 total_bytes =3D 0, total_packets =3D 0; + struct mucse *mucse =3D q_vector->mucse; + struct mucse_tx_buffer *tx_buffer; + struct rnpgbe_tx_desc *tx_desc; + int i =3D tx_ring->next_to_clean; + + tx_buffer =3D &tx_ring->tx_buffer_info[i]; + tx_desc =3D M_TX_DESC(tx_ring, i); + i -=3D tx_ring->count; + + do { + struct rnpgbe_tx_desc *eop_desc =3D 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 */ + dma_rmb(); + + /* if eop DD is not set pending work has not been completed */ + if (!(eop_desc->vlan_cmd & cpu_to_le32(M_TXD_STAT_DD))) + break; + /* clear next_to_watch to prevent false hangs */ + tx_buffer->next_to_watch =3D NULL; + total_bytes +=3D tx_buffer->bytecount; + total_packets +=3D tx_buffer->gso_segs; + napi_consume_skb(tx_buffer->skb, napi_budget); + if (tx_buffer->mapped_as_page) { + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } else { + dma_unmap_single(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } + tx_buffer->skb =3D NULL; + dma_unmap_len_set(tx_buffer, len, 0); + + /* unmap remaining buffers */ + while (tx_desc !=3D eop_desc) { + tx_buffer++; + tx_desc++; + i++; + if (unlikely(!i)) { + i -=3D tx_ring->count; + tx_buffer =3D tx_ring->tx_buffer_info; + tx_desc =3D M_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 -=3D tx_ring->count; + tx_buffer =3D tx_ring->tx_buffer_info; + tx_desc =3D M_TX_DESC(tx_ring, 0); + } + + prefetch(tx_desc); + budget--; + } while (likely(budget > 0)); + + i +=3D tx_ring->count; + tx_ring->next_to_clean =3D i; + u64_stats_update_begin(&tx_ring->syncp); + tx_ring->stats.bytes +=3D total_bytes; + tx_ring->stats.packets +=3D total_packets; + u64_stats_update_end(&tx_ring->syncp); + +#define TX_WAKE_THRESHOLD (DESC_NEEDED * 2) + __netif_txq_completed_wake(txring_txq(tx_ring), + total_packets, total_bytes, + mucse_desc_unused(tx_ring), + TX_WAKE_THRESHOLD, + !netif_running(mucse->netdev)); + + return !!budget; +} + /** * rnpgbe_poll - NAPI polling callback * @napi: structure for representing this polling device * @budget: polling budget * - * Complete NAPI polling and re-enable queue interrupts. Ring cleaning is - * added when TX and RX support is enabled. + * Clean completed TX packets and complete NAPI polling when all queues are + * clean. * - * Return: 0 + * Return: 0 if all TX queues are clean, otherwise @budget **/ static int rnpgbe_poll(struct napi_struct *napi, int budget) { struct mucse_q_vector *q_vector =3D container_of(napi, struct mucse_q_vector, napi); + bool clean_complete =3D true; + struct mucse_ring *ring; int work_done =3D 0; =20 + mucse_for_each_ring(ring, q_vector->tx) { + if (!rnpgbe_clean_tx_irq(q_vector, ring, budget)) + clean_complete =3D false; + } + /* Exit if we are called by netpoll */ if (unlikely(!budget)) return 0; =20 + if (!clean_complete) + return budget; + if (likely(napi_complete_done(napi, work_done))) rnpgbe_irq_enable_queues(q_vector); =20 @@ -228,14 +347,19 @@ static int rnpgbe_alloc_q_vector(struct mucse *mucse, ring =3D q_vector->ring; =20 for (idx =3D 0; idx < txr_count; idx++) { + ring->dev =3D &mucse->pdev->dev; mucse_add_ring(ring, &q_vector->tx); + ring->count =3D mucse->tx_ring_item_count; + ring->netdev =3D mucse->netdev; ring->queue_index =3D eth_queue_idx + idx; ring->rnpgbe_queue_idx =3D txr_idx; ring->ring_addr =3D hw->hw_addr + RING_OFFSET(txr_idx); ring->irq_mask =3D ring->ring_addr + RNPGBE_DMA_INT_MASK; ring->trig =3D ring->ring_addr + RNPGBE_DMA_INT_TRIG; ring->q_vector =3D q_vector; - mucse->tx_ring[ring->queue_index] =3D ring; + ring->pfvfnum =3D hw->pfvfnum; + u64_stats_init(&ring->syncp); + WRITE_ONCE(mucse->tx_ring[ring->queue_index], ring); txr_idx +=3D step; ring++; } @@ -248,7 +372,7 @@ static int rnpgbe_alloc_q_vector(struct mucse *mucse, ring->irq_mask =3D ring->ring_addr + RNPGBE_DMA_INT_MASK; ring->trig =3D ring->ring_addr + RNPGBE_DMA_INT_TRIG; ring->q_vector =3D q_vector; - mucse->rx_ring[ring->queue_index] =3D ring; + WRITE_ONCE(mucse->rx_ring[ring->queue_index], ring); rxr_idx +=3D step; ring++; } @@ -271,12 +395,12 @@ static void rnpgbe_free_q_vector(struct mucse *mucse,= int vector_idx) struct mucse_ring *ring; =20 mucse_for_each_ring(ring, q_vector->tx) - mucse->tx_ring[ring->queue_index] =3D NULL; + WRITE_ONCE(mucse->tx_ring[ring->queue_index], NULL); mucse_for_each_ring(ring, q_vector->rx) - mucse->rx_ring[ring->queue_index] =3D NULL; + WRITE_ONCE(mucse->rx_ring[ring->queue_index], NULL); mucse->q_vector[vector_idx] =3D NULL; netif_napi_del(&q_vector->napi); - kfree(q_vector); + kfree_rcu(q_vector, rcu); } =20 /** @@ -551,12 +675,173 @@ static void rnpgbe_napi_disable_all(struct mucse *mu= cse) napi_disable(&mucse->q_vector[i]->napi); } =20 +static void rnpgbe_stop_tx_ring(struct mucse_ring *tx_ring) +{ + if (!tx_ring->tx_buffer_info) + return; + + /* Stop hw. No new descriptors are fetched after TX_START=3D0. + * DMA for descriptors fetched before the stop may still be in flight. + */ + mucse_ring_wr32(tx_ring, RNPGBE_TX_START, 0); + /* Flush posted write to ensure hardware sees TX_START=3D0 */ + (void)mucse_ring_rd32(tx_ring, RNPGBE_TX_START); +} + +static int rnpgbe_wait_tx_dma_idle(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 dma_status; + int err; + + /* A timeout indicates a terminal AXI fault. Hardware stops all PCIe + * DMA requests in this state, including requests accepted before the + * fault. Teardown may therefore release mappings and descriptor memory. + * Recovery requires a chip-level reset. + */ + err =3D readl_poll_timeout(hw->hw_addr + RNPGBE_DMA_STATUS, + dma_status, + (dma_status & RNPGBE_DMA_TX_STATUS) =3D=3D + RNPGBE_DMA_TX_STATUS, + 10, 100000); + if (err) { + set_bit(__MUCSE_AXI_FAULT, &mucse->state); + dev_err(&mucse->pdev->dev, + "TX DMA failed to quiesce, status %#x\n", dma_status); + } + + return err; +} + +/** + * rnpgbe_stop_all_tx_rings - Stop TX DMA on all queues + * @mucse: board private structure + * + * Return: 0 when TX DMA was quiesced, negative errno otherwise + **/ +static int rnpgbe_stop_all_tx_rings(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 dma_axi_ctl; + int err =3D 0; + + for (int i =3D 0; i < mucse->num_tx_queues; i++) + rnpgbe_stop_tx_ring(mucse->tx_ring[i]); + + if (mucse->num_tx_queues && + !test_bit(__MUCSE_AXI_FAULT, &mucse->state)) + err =3D rnpgbe_wait_tx_dma_idle(mucse); + + dma_axi_ctl =3D mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + dma_axi_ctl &=3D ~TX_AXI_RW_EN; + mucse_hw_wr32(hw, RNPGBE_DMA_AXI_EN, dma_axi_ctl); + /* Flush the posted write before continuing. */ + (void)mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + + if (!test_bit(__MUCSE_AXI_FAULT, &mucse->state)) + return 0; + + return err ? err : -EIO; +} + +/** + * rnpgbe_clean_tx_ring - Free Tx Buffers + * @tx_ring: ring to be cleaned + **/ +static void rnpgbe_clean_tx_ring(struct mucse_ring *tx_ring) +{ + struct mucse_tx_buffer *tx_buffer; + u16 i =3D tx_ring->next_to_clean; + unsigned long size; + + /* ring already cleared, nothing to do */ + if (!tx_ring->tx_buffer_info) + return; + + tx_buffer =3D &tx_ring->tx_buffer_info[i]; + + while (i !=3D tx_ring->next_to_use) { + struct rnpgbe_tx_desc *eop_desc, *tx_desc; + + dev_kfree_skb_any(tx_buffer->skb); + /* unmap skb header data */ + if (dma_unmap_len(tx_buffer, len)) { + if (tx_buffer->mapped_as_page) { + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } else { + dma_unmap_single(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } + } + eop_desc =3D tx_buffer->next_to_watch; + tx_desc =3D M_TX_DESC(tx_ring, i); + /* unmap remaining buffers */ + while (tx_desc !=3D eop_desc) { + tx_buffer++; + tx_desc++; + i++; + if (unlikely(i =3D=3D tx_ring->count)) { + i =3D 0; + tx_buffer =3D tx_ring->tx_buffer_info; + tx_desc =3D M_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 =3D=3D tx_ring->count)) { + i =3D 0; + tx_buffer =3D tx_ring->tx_buffer_info; + } + } + + netdev_tx_reset_queue(txring_txq(tx_ring)); + size =3D sizeof(struct mucse_tx_buffer) * tx_ring->count; + memset(tx_ring->tx_buffer_info, 0, size); + /* Zero out the descriptor ring */ + memset(tx_ring->desc, 0, tx_ring->size); + tx_ring->next_to_use =3D 0; + tx_ring->next_to_clean =3D 0; +} + +/** + * rnpgbe_clean_all_tx_rings - Stop TX DMA and free Tx buffers for all que= ues + * @mucse: board private structure + **/ +void rnpgbe_clean_all_tx_rings(struct mucse *mucse) +{ + /* rnpgbe_stop_all_tx_rings() logs and latches any AXI fault. Hardware + * stops DMA requests in this state, so cleanup intentionally ignores + * an error and releases the mappings and descriptor memory. + */ + rnpgbe_stop_all_tx_rings(mucse); + + for (int i =3D 0; i < mucse->num_tx_queues; i++) + rnpgbe_clean_tx_ring(mucse->tx_ring[i]); +} + void rnpgbe_down(struct mucse *mucse) { + struct net_device *netdev =3D mucse->netdev; + rnpgbe_irq_disable(mucse); rnpgbe_napi_disable_all(mucse); synchronize_net(); rnpgbe_irq_disable(mucse); + netif_tx_disable(netdev); + rnpgbe_clean_all_tx_rings(mucse); } =20 /** @@ -565,7 +850,413 @@ void rnpgbe_down(struct mucse *mucse) **/ void rnpgbe_up_complete(struct mucse *mucse) { + struct net_device *netdev =3D mucse->netdev; + rnpgbe_configure_msix(mucse); rnpgbe_napi_enable_all(mucse); rnpgbe_irq_enable(mucse); + netif_tx_start_all_queues(netdev); +} + +/** + * rnpgbe_free_tx_resources - Free Tx Resources per Queue + * @tx_ring: tx descriptor ring for a specific queue + * + * Free all transmit software resources + **/ +static void rnpgbe_free_tx_resources(struct mucse_ring *tx_ring) +{ + vfree(tx_ring->tx_buffer_info); + tx_ring->tx_buffer_info =3D 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 =3D NULL; +} + +/** + * rnpgbe_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 rnpgbe_setup_tx_resources(struct mucse_ring *tx_ring) +{ + struct device *dev =3D tx_ring->dev; + int size; + + size =3D sizeof(struct mucse_tx_buffer) * tx_ring->count; + + tx_ring->tx_buffer_info =3D vzalloc(size); + if (!tx_ring->tx_buffer_info) + goto err_return; + /* round up to nearest 4K */ + tx_ring->size =3D tx_ring->count * sizeof(struct rnpgbe_tx_desc); + tx_ring->size =3D ALIGN(tx_ring->size, 4096); + tx_ring->desc =3D dma_alloc_coherent(dev, tx_ring->size, &tx_ring->dma, + GFP_KERNEL); + if (!tx_ring->desc) + goto err_free_buffer; + + tx_ring->next_to_use =3D 0; + tx_ring->next_to_clean =3D 0; + + return 0; + +err_free_buffer: + vfree(tx_ring->tx_buffer_info); +err_return: + tx_ring->tx_buffer_info =3D NULL; + return -ENOMEM; +} + +/** + * rnpgbe_configure_tx_ring - Configure Tx ring after Reset + * @mucse: pointer to private structure + * @ring: structure containing ring specific data + * + * Configure the Tx descriptor ring after a reset. + **/ +static void rnpgbe_configure_tx_ring(struct mucse *mucse, + struct mucse_ring *ring) +{ + struct mucse_hw *hw =3D &mucse->hw; + + mucse_ring_wr32(ring, RNPGBE_TX_BASE_ADDR_LO, (u32)ring->dma); + mucse_ring_wr32(ring, RNPGBE_TX_BASE_ADDR_HI, + (u32)(((u64)ring->dma) >> 32) | + ((u32)hw->pfvfnum << 24)); + mucse_ring_wr32(ring, RNPGBE_TX_LEN, ring->count); + ring->next_to_clean =3D mucse_ring_rd32(ring, RNPGBE_TX_HEAD) % + ring->count; + ring->next_to_use =3D ring->next_to_clean; + ring->tail =3D ring->ring_addr + RNPGBE_TX_TAIL; + writel(ring->next_to_use, ring->tail); + mucse_ring_wr32(ring, RNPGBE_TX_FETCH_CTRL, M_DEFAULT_TX_FETCH); + mucse_ring_wr32(ring, RNPGBE_TX_INT_TIMER, + M_DEFAULT_INT_TIMER * hw->cycles_per_us); + mucse_ring_wr32(ring, RNPGBE_TX_INT_PKTCNT, M_DEFAULT_INT_PKTCNT); +} + +/** + * rnpgbe_configure_tx - Configure Transmit Unit after Reset + * @mucse: pointer to private structure + * + * Configure the Tx DMA after a reset. + * + * Return: 0 on success, negative errno if TX DMA did not quiesce + **/ +int rnpgbe_configure_tx(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 i, dma_axi_ctl; + int err; + + err =3D rnpgbe_stop_all_tx_rings(mucse); + if (err) + return err; + + for (i =3D 0; i < mucse->num_tx_queues; i++) + rnpgbe_configure_tx_ring(mucse, mucse->tx_ring[i]); + + /* Ensure all ring configuration is visible before enabling TX DMA. */ + wmb(); + dma_axi_ctl =3D mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + dma_axi_ctl |=3D TX_AXI_RW_EN; + mucse_hw_wr32(hw, RNPGBE_DMA_AXI_EN, dma_axi_ctl); + /* Ensure TX_AXI_RW_EN is visible before starting the rings. */ + (void)mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + + for (i =3D 0; i < mucse->num_tx_queues; i++) + mucse_ring_wr32(mucse->tx_ring[i], RNPGBE_TX_START, 1); + + return 0; +} + +/** + * rnpgbe_setup_all_tx_resources - allocate all queues Tx resources + * @mucse: pointer to private structure + * + * Allocate memory for tx_ring. + * + * Return: 0 on success, negative on failure + **/ +int rnpgbe_setup_all_tx_resources(struct mucse *mucse) +{ + int i, err =3D 0; + + for (i =3D 0; i < mucse->num_tx_queues; i++) { + err =3D rnpgbe_setup_tx_resources(mucse->tx_ring[i]); + if (!err) + continue; + + goto err_free_res; + } + + return 0; +err_free_res: + while (i--) + rnpgbe_free_tx_resources(mucse->tx_ring[i]); + return err; +} + +/** + * rnpgbe_free_all_tx_resources - Free Tx Resources for All Queues + * @mucse: pointer to private structure + * + * Free all transmit software resources + **/ +void rnpgbe_free_all_tx_resources(struct mucse *mucse) +{ + for (int i =3D 0; i < (mucse->num_tx_queues); i++) + rnpgbe_free_tx_resources(mucse->tx_ring[i]); +} + +static int rnpgbe_tx_map(struct mucse_ring *tx_ring, + struct mucse_tx_buffer *first, u32 mac_ip_len, + u32 tx_flags) +{ + struct mucse_tx_buffer *tx_buffer; + struct rnpgbe_tx_desc *tx_desc; + struct skb_shared_info *shinfo; + unsigned int data_len, size; + struct sk_buff *skb; + skb_frag_t *frag; + dma_addr_t dma; + int nr_frags; + int frag_idx; + /* Hardware requires this in the upper 8 bits of + * the descriptor address + */ + u64 fun_id; + u16 i; + + skb =3D first->skb; + shinfo =3D skb_shinfo(skb); + fun_id =3D (u64)tx_ring->pfvfnum << 56; + nr_frags =3D shinfo->nr_frags; + i =3D tx_ring->next_to_use; + frag_idx =3D 0; + tx_desc =3D M_TX_DESC(tx_ring, i); + size =3D skb_headlen(skb); + data_len =3D skb->data_len; + + if (size) { + dma =3D dma_map_single(tx_ring->dev, skb->data, size, + DMA_TO_DEVICE); + first->mapped_as_page =3D false; + } else if (data_len) { + while (frag_idx < nr_frags && + !skb_frag_size(&shinfo->frags[frag_idx])) + frag_idx++; + + if (frag_idx =3D=3D nr_frags) + goto err_unmap; + + frag =3D &shinfo->frags[frag_idx]; + size =3D skb_frag_size(frag); + + dma =3D skb_frag_dma_map(tx_ring->dev, frag, 0, + size, DMA_TO_DEVICE); + first->mapped_as_page =3D true; + data_len -=3D size; + frag_idx++; + } else { + goto err_unmap; + } + + tx_buffer =3D first; + + dma_unmap_len_set(tx_buffer, len, 0); + dma_unmap_addr_set(tx_buffer, dma, 0); + + for (;; frag_idx++) { + if (dma_mapping_error(tx_ring->dev, dma)) + goto err_unmap; + + /* record length, and DMA address */ + dma_unmap_len_set(tx_buffer, len, size); + dma_unmap_addr_set(tx_buffer, dma, dma); + + tx_desc->pkt_addr =3D cpu_to_le64(dma | fun_id); + + while (unlikely(size > M_MAX_DATA_PER_TXD)) { + tx_desc->vlan_cmd_bsz =3D build_ctob(tx_flags, + mac_ip_len, + M_MAX_DATA_PER_TXD); + i++; + tx_desc++; + if (i =3D=3D tx_ring->count) { + tx_desc =3D M_TX_DESC(tx_ring, 0); + i =3D 0; + } + + tx_buffer =3D &tx_ring->tx_buffer_info[i]; + dma_unmap_len_set(tx_buffer, len, 0); + dma +=3D M_MAX_DATA_PER_TXD; + size -=3D M_MAX_DATA_PER_TXD; + tx_desc->pkt_addr =3D cpu_to_le64(dma | fun_id); + } + + if (likely(!data_len)) + break; + tx_desc->vlan_cmd_bsz =3D build_ctob(tx_flags, mac_ip_len, size); + i++; + tx_desc++; + if (i =3D=3D tx_ring->count) { + tx_desc =3D M_TX_DESC(tx_ring, 0); + i =3D 0; + } + + while (frag_idx < nr_frags && + !skb_frag_size(&shinfo->frags[frag_idx])) + frag_idx++; + + if (frag_idx =3D=3D nr_frags) + goto err_unmap; + + frag =3D &shinfo->frags[frag_idx]; + size =3D skb_frag_size(frag); + data_len -=3D size; + dma =3D skb_frag_dma_map(tx_ring->dev, frag, 0, size, + DMA_TO_DEVICE); + tx_buffer =3D &tx_ring->tx_buffer_info[i]; + tx_buffer->mapped_as_page =3D true; + } + + /* write last descriptor with RS and EOP bits */ + tx_desc->vlan_cmd_bsz =3D build_ctob(tx_flags | M_TXD_CMD_EOP | + M_TXD_CMD_RS, + mac_ip_len, size); + + /* 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. + */ + dma_wmb(); + /* set next_to_watch value indicating a packet is present */ + first->next_to_watch =3D tx_desc; + i++; + if (i =3D=3D tx_ring->count) + i =3D 0; + tx_ring->next_to_use =3D i; + skb_tx_timestamp(skb); + netdev_tx_sent_queue(txring_txq(tx_ring), first->bytecount); + /* notify HW of packet */ + writel(i, tx_ring->tail); + + return 0; +err_unmap: + for (;;) { + tx_buffer =3D &tx_ring->tx_buffer_info[i]; + if (dma_unmap_len(tx_buffer, len)) { + if (tx_buffer->mapped_as_page) { + dma_unmap_page(tx_ring->dev, + dma_unmap_addr(tx_buffer, dma), + dma_unmap_len(tx_buffer, len), + DMA_TO_DEVICE); + } else { + dma_unmap_single(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); + dma_unmap_addr_set(tx_buffer, dma, 0); + if (tx_buffer =3D=3D first) + break; + if (i =3D=3D 0) + i +=3D tx_ring->count; + i--; + } + dev_kfree_skb_any(first->skb); + first->skb =3D NULL; + tx_ring->next_to_use =3D i; + + return -ENOMEM; +} + +netdev_tx_t rnpgbe_xmit_frame_ring(struct sk_buff *skb, + struct mucse_ring *tx_ring) +{ + /* hw requires non-zero mac_ip_len even without offload; + * Will be updated per-packet when offload is added + */ + u32 mac_ip_len =3D M_DEFAULT_MAC_IP_LEN; + struct mucse_tx_buffer *first; + u32 tx_flags =3D 0; + unsigned short f; + u16 count; + + count =3D TXD_USE_COUNT(skb_headlen(skb)); + for (f =3D 0; f < skb_shinfo(skb)->nr_frags; f++) { + skb_frag_t *frag_temp =3D &skb_shinfo(skb)->frags[f]; + + count +=3D TXD_USE_COUNT(skb_frag_size(frag_temp)); + } + + if (!netif_txq_maybe_stop(txring_txq(tx_ring), + mucse_desc_unused(tx_ring), + count + RESV_DESC_NEEDED, + count + RESV_DESC_NEEDED)) + return NETDEV_TX_BUSY; + + /* record the location of the first descriptor for this packet */ + first =3D &tx_ring->tx_buffer_info[tx_ring->next_to_use]; + first->skb =3D skb; + first->bytecount =3D skb->len; + first->gso_segs =3D 1; + + if (rnpgbe_tx_map(tx_ring, first, mac_ip_len, tx_flags)) { + atomic64_inc(&tx_ring->stats.dropped); + + goto out; + } + + /* NETDEV_TX_BUSY is expensive. So stop advancing the TX queue. */ + netif_txq_maybe_stop(txring_txq(tx_ring), + mucse_desc_unused(tx_ring), + DESC_NEEDED, DESC_NEEDED); +out: + return NETDEV_TX_OK; +} + +/** + * rnpgbe_get_stats64 - Get stats for this netdev + * @netdev: network interface device structure + * @stats: stats data + **/ +void rnpgbe_get_stats64(struct net_device *netdev, + struct rtnl_link_stats64 *stats) +{ + struct mucse *mucse =3D netdev_priv(netdev); + int i; + + rcu_read_lock(); + for (i =3D 0; i < mucse->num_tx_queues; i++) { + struct mucse_ring *ring =3D READ_ONCE(mucse->tx_ring[i]); + u64 bytes, packets, dropped; + unsigned int start; + + if (ring) { + do { + start =3D u64_stats_fetch_begin(&ring->syncp); + packets =3D ring->stats.packets; + bytes =3D ring->stats.bytes; + } while (u64_stats_fetch_retry(&ring->syncp, start)); + dropped =3D atomic64_read(&ring->stats.dropped); + + stats->tx_packets +=3D packets; + stats->tx_dropped +=3D dropped; + stats->tx_bytes +=3D bytes; + } + } + rcu_read_unlock(); } diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.h index d38c08df6b7c..1fe290ed224d 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h @@ -6,17 +6,39 @@ =20 struct mucse; struct mucse_hw; +struct mucse_ring; =20 #define RING_OFFSET(n) (0x1000 + 0x100 * (n)) +#define RNPGBE_TX_START 0x18 #define RNPGBE_DMA_INT_MASK 0x24 #define TX_INT_MASK BIT(1) #define RX_INT_MASK BIT(0) #define INT_VALID (BIT(16) | BIT(17)) #define RNPGBE_DMA_INT_TRIG 0x2c /* lost-interrupt recovery trigger = */ +#define RNPGBE_TX_BASE_ADDR_HI 0x60 +#define RNPGBE_TX_BASE_ADDR_LO 0x64 +#define RNPGBE_TX_LEN 0x68 +#define RNPGBE_TX_HEAD 0x6c +#define RNPGBE_TX_TAIL 0x70 +#define M_DEFAULT_TX_FETCH 0x80008 +#define RNPGBE_TX_FETCH_CTRL 0x74 +#define M_DEFAULT_INT_TIMER 100 +#define RNPGBE_TX_INT_TIMER 0x78 +#define M_DEFAULT_INT_PKTCNT 48 +#define RNPGBE_TX_INT_PKTCNT 0x7c /* | 31:24 | .... | 15:8 | 7:0 | */ /* | pfvfnum | | tx vector | rx vector | */ #define RING_VECTOR(n) (0x04 * (n)) =20 +#define M_MAX_TXD_PWR 12 +#define M_MAX_DATA_PER_TXD (0x1 << M_MAX_TXD_PWR) +#define TXD_USE_COUNT(S) DIV_ROUND_UP((S), M_MAX_DATA_PER_TXD) +#define DESC_NEEDED (MAX_SKB_FRAGS + 4) +/* 2 desc gap to keep tail from touching head */ +/* 1 desc for context descriptor */ +#define RESV_DESC_NEEDED 3 +/* Hardware requires this to be nonzero */ +#define M_DEFAULT_MAC_IP_LEN 20 #define mucse_for_each_ring(pos, head)\ for (typeof((head).ring) __pos =3D (head).ring;\ __pos ? ({ pos =3D __pos; 1; }) : 0;\ @@ -31,4 +53,12 @@ void rnpgbe_free_irq(struct mucse *mucse); void rnpgbe_irq_disable(struct mucse *mucse); void rnpgbe_down(struct mucse *mucse); void rnpgbe_up_complete(struct mucse *mucse); +int rnpgbe_configure_tx(struct mucse *mucse); +void rnpgbe_clean_all_tx_rings(struct mucse *mucse); +int rnpgbe_setup_all_tx_resources(struct mucse *mucse); +void rnpgbe_free_all_tx_resources(struct mucse *mucse); +netdev_tx_t rnpgbe_xmit_frame_ring(struct sk_buff *skb, + struct mucse_ring *tx_ring); +void rnpgbe_get_stats64(struct net_device *netdev, + struct rtnl_link_stats64 *stats); #endif diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_main.c index debed2f3fe64..cdda5f3d09b2 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c @@ -2,6 +2,7 @@ /* Copyright(c) 2020 - 2025 Mucse Corporation. */ =20 #include +#include #include #include =20 @@ -26,6 +27,19 @@ static struct pci_device_id rnpgbe_pci_tbl[] =3D { { }, }; =20 +/** + * rnpgbe_configure - Configure the hardware + * @mucse: pointer to private structure + * + * Configure Tx registers in hardware. + * + * Return: 0 on success, negative errno if hardware configuration fails + **/ +static int rnpgbe_configure(struct mucse *mucse) +{ + return rnpgbe_configure_tx(mucse); +} + /** * rnpgbe_open - Called when a network interface is made active * @netdev: network interface device structure @@ -40,6 +54,9 @@ static int rnpgbe_open(struct net_device *netdev) struct mucse *mucse =3D netdev_priv(netdev); int err; =20 + if (test_bit(__MUCSE_AXI_FAULT, &mucse->state)) + return -EIO; + err =3D rnpgbe_request_irq(mucse); if (err) return err; @@ -49,9 +66,19 @@ static int rnpgbe_open(struct net_device *netdev) if (err) goto err_free_irqs; =20 + err =3D rnpgbe_setup_all_tx_resources(mucse); + if (err) + goto err_free_irqs; + + err =3D rnpgbe_configure(mucse); + if (err) + goto err_free_tx; rnpgbe_up_complete(mucse); =20 return 0; +err_free_tx: + rnpgbe_clean_all_tx_rings(mucse); + rnpgbe_free_all_tx_resources(mucse); err_free_irqs: rnpgbe_free_irq(mucse); return err; @@ -72,6 +99,7 @@ static int rnpgbe_close(struct net_device *netdev) =20 rnpgbe_down(mucse); rnpgbe_free_irq(mucse); + rnpgbe_free_all_tx_resources(mucse); =20 return 0; } @@ -81,25 +109,37 @@ static int rnpgbe_close(struct net_device *netdev) * @skb: skb structure to be sent * @netdev: network interface device structure * - * Return: NETDEV_TX_OK + * Return: NETDEV_TX_OK or NETDEV_TX_BUSY when insufficient descriptors **/ static netdev_tx_t rnpgbe_xmit_frame(struct sk_buff *skb, struct net_device *netdev) { struct mucse *mucse =3D netdev_priv(netdev); + struct mucse_ring *tx_ring; =20 - dev_kfree_skb_any(skb); - mucse->stats.tx_dropped++; + tx_ring =3D mucse->tx_ring[skb_get_queue_mapping(skb)]; =20 - return NETDEV_TX_OK; + if (unlikely(skb_put_padto(skb, RNPGBE_TX_MIN_PKT_LEN))) { + atomic64_inc(&tx_ring->stats.dropped); + return NETDEV_TX_OK; + } + + return rnpgbe_xmit_frame_ring(skb, tx_ring); } =20 static const struct net_device_ops rnpgbe_netdev_ops =3D { .ndo_open =3D rnpgbe_open, .ndo_stop =3D rnpgbe_close, .ndo_start_xmit =3D rnpgbe_xmit_frame, + .ndo_get_stats64 =3D rnpgbe_get_stats64, }; =20 +static void rnpgbe_sw_init(struct mucse *mucse) +{ + mucse->tx_ring_item_count =3D M_DEFAULT_TXD; + mucse->tx_work_limit =3D M_DEFAULT_TX_WORK; +} + /** * rnpgbe_add_adapter - Add netdev for this pci_dev * @pdev: PCI device information structure @@ -172,6 +212,7 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, } =20 netdev->netdev_ops =3D &rnpgbe_netdev_ops; + rnpgbe_sw_init(mucse); err =3D rnpgbe_reset_hw(hw); if (err) { dev_err(&pdev->dev, "Hw reset failed %d\n", err); @@ -206,6 +247,13 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, goto err_clear_interrupt; } =20 + netdev->features |=3D NETIF_F_SG; + netdev->hw_features |=3D NETIF_F_SG; + if (dma_get_mask(&pdev->dev) > DMA_BIT_MASK(32)) { + netdev->features |=3D NETIF_F_HIGHDMA; + netdev->hw_features |=3D NETIF_F_HIGHDMA; + } + err =3D register_netdev(netdev); if (err) goto err_remove_mbx; @@ -248,12 +296,7 @@ static int rnpgbe_probe(struct pci_dev *pdev, const st= ruct pci_device_id *id) if (err) return err; =20 - err =3D dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(56)); - if (err) { - dev_err(&pdev->dev, - "No usable DMA configuration, aborting %d\n", err); - goto err_disable_dev; - } + dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(56)); =20 err =3D pci_request_mem_regions(pdev, rnpgbe_driver_name); if (err) { --=20 2.50.1 From nobody Thu Sep 24 20:41:59 2026 Received: from smtpbgjp3.qq.com (smtpbgjp3.qq.com [54.92.39.34]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id BA2292FD69D; Sun, 20 Sep 2026 09:27:39 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=54.92.39.34 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896464; cv=none; b=Ko2UhPLNxdR8X3HOZhHuc10S4AZSkRoAyGcJt4z7QHrlsdIEDsHeBxfkoHugQZdot9LesKQ2WCTTSaFHhAmEXd0A5WN2gNkt7dQcFeKBkOo9BA/JoE0+P/xY+JFI5QZR7j55WU0hEaBfT9c/LYhllf5+tXLaYAJIWLjf7kPn5oA= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896464; c=relaxed/simple; bh=BQqS6deHv3rHT6xSKQReI2F/IAcRwJiHp36Ajcl4laQ=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=kSUDMdWyuvE2e1lbff3DvqAD9n2muNTARFvLdDP8qQDlZpmmjNG3d6ccMoFoDUB3HhRP7XfRIXJxENFO1QIeBiA/wrN+opk3CQFzjchtEs/g6W3jvysX/I1kuNr50aJKq18zj1OmXThGudIye9Uq2bDMintOzzq4QwDOrZrAjyc= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com; spf=pass smtp.mailfrom=mucse.com; arc=none smtp.client-ip=54.92.39.34 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mucse.com X-QQ-mid: esmtpgz13t1789896402t1d075298 X-QQ-Originating-IP: i9r92vmuipLU9FOdi7p3PwhimVfrG5nbwkGUyUzwC4k= Received: from nic-Precision-5820-Tower ( [203.174.112.180]) by bizesmtp.qq.com (ESMTP) with id ; Sun, 20 Sep 2026 17:26:40 +0800 (CST) X-QQ-SSF: 0000000000000000000000000000000 X-QQ-GoodBg: 0 X-BIZMAIL-ID: 6976190405568439834 EX-QQ-RecipientCnt: 11 From: Dong Yibo To: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, vadim.fedorenko@linux.dev, u.kleine-koenig@baylibre.com Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, dong100@mucse.com, yaojun@mucse.com Subject: [PATCH net-next v13 3/5] net: rnpgbe: Add basic RX data path support Date: Sun, 20 Sep 2026 17:26:38 +0800 Message-ID: X-Mailer: git-send-email 2.50.1 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-QQ-SENDSIZE: 520 Feedback-ID: esmtpgz:mucse.com:qybglogicsvrgz:qybglogicsvrgz3a-1 X-QQ-XMAILINFO: MWxZUaxBV+u35yRnHB27m64uSVl4zEyohWdcZixJTZ2devKQhvuQoCGZ OusMSpQsZbYfuaeNinyd17jPrvmzLIqG6Ku6gm5OyqYFRNzGeDIoavMxbH5X7KMVXiCQ+Ug pEC7DgmAh6tNfdWnu3ZLHwvysR4PXyda/mMzNnanR8kXJvTvPCtu/HbnoG1yM14bM5c4hf1 g7MupKpzQ71lzt9bjf//mX+6tguvN1yvo1jon6W++nWg8x9Ff3Vl5ALMIl9Ujmpi88HtCjs OHB254mS7qbdt3PbMkQMYEIneVt7u0KhJuxNH0wJh9l4XkaUQF9sjWjX0tKlaHAlPICx7pd 1cluOtRptpF3rHOQU3LeKvlUOvcoB6+p2qlS/QDIgZuQ0yHaUMNIlGBlQcIM9aeeZFnE+bd VjyjvaOvjCbT1rDPZHp36upk+ATyadbnomJhbt+qFAZl86iJVKtmp1eC6dSQWcdkeBG0IdF E7Z4dJ4AMNGueQ3ZxJRm1FrYasdKKjRTN7l2kqNXZ7X75J3xlfl3qiidAXXI6DR0IPjgV/q U3ssZ6kVqBfT/fmn+LWHw1oXNu6Shz3mZJHjccokaZjq6hi4KdgXanyg1S1H3duLTdEePOV BuMbNNoK2nBI5LTKddM3Mm/lis+hT0ujU/t5/PhpjOKCGbq4EqpDZ5km+Z71T6NTEX6oZ3+ 6I+EkPUyBS43AvYl+FmHHBhtAojliOEGo7SGZlD/T+qa+ZRN9DvUJcFmW+Xj3LxQ/BEhfoF elmfdMZn3nXU++YfUl4SkPINvUgRPZ2SWoTMfL0qWb9AKkVgbERHH3j6K/JcceD9xHIa8Ic BP9PCszDLSi6MNAmmOdKOramcvFDWpQB7wMZ6jv/Sq0fdhM+HgDKUPXxM2MLPl/Qoo6ZvXL YamB6E402u3HjHIdQfngj+EiSuTdkjyOkPZlD6XiDPsXKLBpj/olxdrtJHDYvxMkPXWRskE 7qtzHWO/NLD5tcXE+XmqkCYJtEFBNBYmHK0bGtItB4JrPfCa93IzJ8rPYba5w6LVy/2CTbX 0oKaCce/t7Gp35Bvy7wrwfg8aCxt/sjEgnDzOK/pVPDotC5+r4xaA3yBq+rzo9HjIznYxIl 7SUdeTCX89I X-QQ-XMRINFO: NyFYKkN4Ny6FuXrnB5Ye7Aabb3ujjtK+gg== X-QQ-RECHKSPAM: 0 Content-Type: text/plain; charset="utf-8" Add basic RX DMA data path for the RNPGBE driver. Introduce the RX descriptor structure (union rnpgbe_rx_desc) with the hardware write-back format, and manage RX buffers with page_pool using a single page per descriptor without page splitting. Implement the NAPI poll callback, RX ring setup and teardown, and packet assembly from page buffers, along with RX statistics. Keep a 16-descriptor gap in the fixed 512-entry RX ring so at most 496 descriptors are posted to hardware. Retry failed page allocations with one timer per q_vector that schedules NAPI and is cancelled before RX cleanup and q_vector teardown. An RX DMA quiesce timeout is a terminal AXI fault: keep DMA disabled, release existing mappings safely, and reject subsequent opens. Recovery requires a chip-level reset and driver rebind. Signed-off-by: Dong Yibo --- drivers/net/ethernet/mucse/Kconfig | 1 + drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h | 58 +- drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h | 2 + .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.c | 794 +++++++++++++++++- .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.h | 21 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_main.c | 19 +- 6 files changed, 887 insertions(+), 8 deletions(-) diff --git a/drivers/net/ethernet/mucse/Kconfig b/drivers/net/ethernet/mucs= e/Kconfig index 0b3e853d625f..be0fdf268484 100644 --- a/drivers/net/ethernet/mucse/Kconfig +++ b/drivers/net/ethernet/mucse/Kconfig @@ -19,6 +19,7 @@ if NET_VENDOR_MUCSE config MGBE tristate "Mucse(R) 1GbE PCI Express adapters support" depends on PCI + select PAGE_POOL help This driver supports Mucse(R) 1GbE PCI Express family of adapters. diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h b/drivers/net/ether= net/mucse/rnpgbe/rnpgbe.h index 864cbf0caf44..c1a72e8e2770 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h @@ -8,6 +8,7 @@ #include #include #include +#include #include #include =20 @@ -63,7 +64,32 @@ struct rnpgbe_tx_desc { #define M_TXD_CMD_EOP 0x010000 /* End of Packet */ }; =20 +union rnpgbe_rx_desc { + struct { + __le64 pkt_addr; /* Packet buffer address */ + __le64 resv_cmd; /* cmd status */ + }; + struct { + __le32 rss_hash; /* RSS HASH */ + __le16 mark; /* mark info */ + __le16 rev1; + __le16 len; /* Packet length */ + __le16 padding_len; + __le16 vlan; /* VLAN tag */ + __le16 cmd; /* cmd status */ +#define M_RXD_STAT_DD BIT(1) /* Descriptor Done */ +#define M_RXD_STAT_EOP BIT(0) /* End of Packet */ + } wb; +}; + #define M_TX_DESC(R, i) (&(((struct rnpgbe_tx_desc *)((R)->desc))[i])) +#define M_RX_DESC(R, i) (&(((union rnpgbe_rx_desc *)((R)->desc))[i])) + +static inline __le16 rnpgbe_test_staterr(union rnpgbe_rx_desc *rx_desc, + const u16 stat_err_bits) +{ + return rx_desc->wb.cmd & cpu_to_le16(stat_err_bits); +} =20 struct mucse_tx_buffer { struct rnpgbe_tx_desc *next_to_watch; @@ -81,13 +107,24 @@ struct mucse_queue_stats { atomic64_t dropped; }; =20 +struct mucse_rx_buffer { + struct sk_buff *skb; + dma_addr_t dma; + struct page *page; + u32 page_offset; +}; + struct mucse_ring { struct mucse_ring *next; struct mucse_q_vector *q_vector; struct net_device *netdev; struct device *dev; + struct page_pool *page_pool; void *desc; - struct mucse_tx_buffer *tx_buffer_info; + union { + struct mucse_tx_buffer *tx_buffer_info; + struct mucse_rx_buffer *rx_buffer_info; + }; void __iomem *ring_addr; void __iomem *tail; void __iomem *irq_mask; @@ -103,8 +140,12 @@ struct mucse_ring { unsigned int size; struct mucse_queue_stats stats; struct u64_stats_sync syncp; + bool drop_status; } ____cacheline_internodealigned_in_smp; =20 +/* Refill RX descriptors in batches of this size. */ +#define M_RX_BUFFER_WRITE 16 + static inline u16 mucse_desc_unused(struct mucse_ring *ring) { u16 ntc =3D ring->next_to_clean; @@ -113,6 +154,18 @@ static inline u16 mucse_desc_unused(struct mucse_ring = *ring) return ((ntc > ntu) ? 0 : ring->count) + ntc - ntu - 1; } =20 +static inline u16 mucse_desc_unused_rx(struct mucse_ring *ring) +{ + u16 ntc =3D ring->next_to_clean; + u16 ntu =3D ring->next_to_use; + + /* Keep M_RX_BUFFER_WRITE descriptors unused so the ring is not filled + * completely. Refill is attempted once at least this many descriptors + * are available. + */ + return ((ntc > ntu) ? 0 : ring->count) + ntc - ntu - M_RX_BUFFER_WRITE; +} + static inline __le64 build_ctob(u32 vlan_cmd, u32 mac_ip_len, u32 size) { return cpu_to_le64(((u64)vlan_cmd << 32) | ((u64)mac_ip_len << 16) | @@ -136,6 +189,7 @@ struct mucse_q_vector { struct mucse_ring_container rx, tx; struct napi_struct napi; struct rcu_head rcu; + struct timer_list rx_alloc_timer; char name[IFNAMSIZ + 18]; /* for dynamic allocation of rings associated with this q_vector */ struct mucse_ring ring[] ____cacheline_internodealigned_in_smp; @@ -144,6 +198,7 @@ struct mucse_q_vector { #define MAX_Q_VECTORS 8 =20 #define M_DEFAULT_TXD 512 +#define M_DEFAULT_RXD 512 #define M_DEFAULT_TX_WORK 256 =20 enum mucse_state_t { @@ -163,6 +218,7 @@ struct mucse { int tx_work_limit; int num_tx_queues; int num_q_vectors; + int rx_ring_item_count; int num_rx_queues; char mbx_name[32]; unsigned long state; diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h b/drivers/net/et= hernet/mucse/rnpgbe/rnpgbe_hw.h index 6dc29ebe6fa7..1d87edfba3d7 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h @@ -19,6 +19,8 @@ #define TX_AXI_RW_EN 0xc /* DMA_STATUS_REG[23:20]: tx_wr, tx_rd, rx_wr, rx_rd done status. */ #define RNPGBE_DMA_TX_STATUS GENMASK_U32(23, 22) +#define RNPGBE_DMA_RX_STATUS GENMASK_U32(21, 20) +#define RX_AXI_RW_EN 0x03 #define RNPGBE_DMA_AXI_EN 0x0010 #define RNPGBE_TX_MIN_PKT_LEN 33 =20 diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.c index af5cef987051..9afd10273a62 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c @@ -4,8 +4,10 @@ #include #include #include +#include #include #include +#include =20 #include "rnpgbe_lib.h" #include "rnpgbe.h" @@ -174,15 +176,436 @@ static bool rnpgbe_clean_tx_irq(struct mucse_q_vecto= r *q_vector, return !!budget; } =20 +static bool mucse_alloc_mapped_page(struct mucse_ring *rx_ring, + struct mucse_rx_buffer *bi) +{ + struct page *page =3D bi->page; + dma_addr_t dma; + + if (page) { + /* Buffer is being reused without going back through the + * page_pool. Do dma_sync for hw use. + */ + dma_sync_single_range_for_device(rx_ring->dev, bi->dma, + bi->page_offset, + PAGE_SIZE - bi->page_offset, + DMA_FROM_DEVICE); + return true; + } + + page =3D page_pool_dev_alloc_pages(rx_ring->page_pool); + if (unlikely(!page)) + return false; + dma =3D page_pool_get_dma_addr(page); + + bi->dma =3D dma; + bi->page =3D page; + bi->page_offset =3D RNPGBE_SKB_PAD; + + return true; +} + +static void mucse_update_rx_tail(struct mucse_ring *rx_ring, + u32 val) +{ + rx_ring->next_to_use =3D val; + writel(val, rx_ring->tail); +} + +/** + * rnpgbe_alloc_rx_buffers - Replace used receive buffers + * @rx_ring: ring to place buffers on + * @cleaned_count: number of buffers to replace + * + * Return: true if alloc failed + **/ +static bool rnpgbe_alloc_rx_buffers(struct mucse_ring *rx_ring, + u16 cleaned_count) +{ + u64 fun_id =3D ((u64)(rx_ring->pfvfnum) << 56); + union rnpgbe_rx_desc *rx_desc; + u16 i =3D rx_ring->next_to_use; + struct mucse_rx_buffer *bi; + bool err =3D false; + u64 addr; + /* nothing to do */ + if (!cleaned_count) + return err; + + rx_desc =3D M_RX_DESC(rx_ring, i); + bi =3D &rx_ring->rx_buffer_info[i]; + i -=3D rx_ring->count; + + do { + if (!mucse_alloc_mapped_page(rx_ring, bi)) { + err =3D true; + break; + } + + addr =3D (u64)(bi->dma + bi->page_offset); + rx_desc->pkt_addr =3D cpu_to_le64(addr | fun_id); + /* clean dd */ + rx_desc->resv_cmd =3D 0; + rx_desc++; + bi++; + i++; + if (unlikely(!i)) { + rx_desc =3D M_RX_DESC(rx_ring, 0); + bi =3D rx_ring->rx_buffer_info; + i -=3D rx_ring->count; + } + cleaned_count--; + } while (cleaned_count); + + i +=3D rx_ring->count; + + /* Publish the tail even when allocation stopped early, so a stale + * tail from a previous session cannot point hardware at a + * reallocated ring. + */ + dma_wmb(); + mucse_update_rx_tail(rx_ring, i); + + return err; +} + +/** + * rnpgbe_get_buffer - Get the rx_buffer to be used + * @rx_ring: pointer to rx ring + * @skb: pointer skb for this packet + * @size: data size in this desc + * + * Return: rx_buffer. + **/ +static struct mucse_rx_buffer *rnpgbe_get_buffer(struct mucse_ring *rx_rin= g, + struct sk_buff **skb, + const unsigned int size) +{ + struct mucse_rx_buffer *rx_buffer; + + rx_buffer =3D &rx_ring->rx_buffer_info[rx_ring->next_to_clean]; + *skb =3D rx_buffer->skb; + prefetchw(page_address(rx_buffer->page) + rx_buffer->page_offset); + /* 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); + + return rx_buffer; +} + +/** + * rnpgbe_add_rx_frag - Add non-linear data to the skb + * @rx_buffer: pointer to rx_buffer + * @skb: pointer skb for this packet + * @size: data size in this desc + **/ +static void rnpgbe_add_rx_frag(struct mucse_rx_buffer *rx_buffer, + struct sk_buff *skb, + unsigned int size) +{ + unsigned int truesize =3D PAGE_SIZE; + + skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_buffer->page, + rx_buffer->page_offset, size, truesize); +} + +/** + * rnpgbe_build_skb - Try to build a skb based on rx_buffer + * @rx_buffer: pointer to rx_buffer + * @size: data size in this desc + * + * Return: skb for this rx_buffer + **/ +static struct sk_buff *rnpgbe_build_skb(struct mucse_rx_buffer *rx_buffer, + unsigned int size) +{ + void *va =3D page_address(rx_buffer->page) + rx_buffer->page_offset; + unsigned int truesize =3D PAGE_SIZE; + struct sk_buff *skb; + + net_prefetch(va); + /* build an skb around the page buffer */ + skb =3D napi_build_skb(va - RNPGBE_SKB_PAD, truesize); + if (unlikely(!skb)) + return NULL; + /* update pointers within the skb to store the data */ + skb_reserve(skb, RNPGBE_SKB_PAD); + __skb_put(skb, size); + skb_mark_for_recycle(skb); + + return skb; +} + +/** + * rnpgbe_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. + * + * Return: true for not end of packet + **/ +static bool rnpgbe_is_non_eop(struct mucse_ring *rx_ring, + union rnpgbe_rx_desc *rx_desc, + struct sk_buff *skb) +{ + u32 ntc =3D rx_ring->next_to_clean + 1; + + /* fetch, update, and store next to clean */ + ntc =3D (ntc < rx_ring->count) ? ntc : 0; + rx_ring->next_to_clean =3D ntc; + prefetch(M_RX_DESC(rx_ring, ntc)); + /* if we are the last buffer then there is nothing else to do */ + if (likely(rnpgbe_test_staterr(rx_desc, M_RXD_STAT_EOP))) + return false; + if (skb_shinfo(skb)->nr_frags < MAX_SKB_FRAGS) { + /* place skb in next buffer to be received */ + rx_ring->rx_buffer_info[ntc].skb =3D skb; + } else { + atomic64_inc(&rx_ring->stats.dropped); + /* too much frags, force free */ + dev_kfree_skb_any(skb); + rx_ring->drop_status =3D true; + } + /* we should clean it since we used all info in it */ + rx_desc->wb.cmd =3D 0; + + return true; +} + +/** + * rnpgbe_cleanup_headers - Correct corrupted or empty headers + * @skb: current socket buffer containing buffer in progress + * + * Return: true if an error was encountered and skb was freed. + **/ +static bool rnpgbe_cleanup_headers(struct sk_buff *skb) +{ + /* if eth_skb_pad returns an error the skb was freed */ + if (eth_skb_pad(skb)) + return true; + + return false; +} + +/** + * rnpgbe_process_skb_fields - Set the RX queue and protocol fields + * @rx_ring: RX descriptor ring containing the queue information + * @skb: skb currently being received + * + * Records the RX queue that received the skb and sets its protocol from + * the Ethernet header. + **/ +static void rnpgbe_process_skb_fields(struct mucse_ring *rx_ring, + struct sk_buff *skb) +{ + struct net_device *dev =3D rx_ring->netdev; + + skb_record_rx_queue(skb, rx_ring->queue_index); + skb->protocol =3D eth_type_trans(skb, dev); +} + +/** + * rnpgbe_rx_alloc_retry - Retry RX buffer allocation + * @timer: RX allocation retry timer + * + * Schedules NAPI after RX buffer allocation fails. + **/ +static void rnpgbe_rx_alloc_retry(struct timer_list *timer) +{ + struct mucse_q_vector *q_vector =3D + timer_container_of(q_vector, timer, rx_alloc_timer); + + napi_schedule(&q_vector->napi); +} + +static void rnpgbe_schedule_rx_retry(struct mucse *mucse) +{ + struct mucse_q_vector *q_vector; + struct mucse_ring *ring; + + for (int i =3D 0; i < mucse->num_q_vectors; i++) { + q_vector =3D mucse->q_vector[i]; + mucse_for_each_ring(ring, q_vector->rx) { + if (mucse_desc_unused_rx(ring)) { + mod_timer(&q_vector->rx_alloc_timer, + jiffies + msecs_to_jiffies(500)); + break; + } + } + } +} + +/** + * rnpgbe_clean_rx_irq - Clean completed descriptors from Rx ring + * @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 + * + * rnpgbe_clean_rx_irq tries to check dd in desc, handle this desc + * if dd is set which means data is write-back by hw + * + * Return: amount of work completed. + **/ +static int rnpgbe_clean_rx_irq(struct mucse_q_vector *q_vector, + struct mucse_ring *rx_ring, + int budget) +{ + unsigned int max_size =3D SKB_WITH_OVERHEAD(PAGE_SIZE) - RNPGBE_SKB_PAD; + unsigned int total_rx_bytes =3D 0, total_rx_packets =3D 0; + u16 cleaned_count =3D mucse_desc_unused_rx(rx_ring); + unsigned int work_done =3D 0; + + while (likely(work_done < budget)) { + struct mucse_rx_buffer *rx_buffer; + union rnpgbe_rx_desc *rx_desc; + struct sk_buff *skb; + unsigned int size; + + if (cleaned_count >=3D M_RX_BUFFER_WRITE) { + if (rnpgbe_alloc_rx_buffers(rx_ring, cleaned_count)) { + mod_timer(&q_vector->rx_alloc_timer, + jiffies + msecs_to_jiffies(500)); + cleaned_count =3D mucse_desc_unused_rx(rx_ring); + } else { + cleaned_count =3D 0; + } + } + rx_desc =3D M_RX_DESC(rx_ring, rx_ring->next_to_clean); + + if (!rnpgbe_test_staterr(rx_desc, M_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(); + /* Hardware guarantees the first descriptor of each packet is at + * least 33 bytes, including multi-descriptor packets. + * Multi-descriptor packets are only used for jumbo frames over + * 1536 bytes (M_DEFAULT_SG =3D 96). Each descriptor is at most + * 1536 bytes. Small packets use a single descriptor. + */ + size =3D le16_to_cpu(rx_desc->wb.len); + + if (unlikely(rx_ring->drop_status)) { + cleaned_count++; + /* drop data until eop */ + if (rnpgbe_test_staterr(rx_desc, M_RXD_STAT_EOP)) { + rx_ring->drop_status =3D false; + work_done++; + } + + rx_desc->wb.cmd =3D 0; + rx_ring->next_to_clean++; + if (rx_ring->next_to_clean >=3D rx_ring->count) + rx_ring->next_to_clean =3D 0; + continue; + } + + if (unlikely(!size || size > max_size)) { + struct mucse_rx_buffer *err_rx_buffer; + u16 idx =3D rx_ring->next_to_clean; + + cleaned_count++; + atomic64_inc(&rx_ring->stats.dropped); + + /* Free the partial skb from a previous non-EOP + * descriptor before advancing next_to_clean. + */ + err_rx_buffer =3D &rx_ring->rx_buffer_info[idx]; + if (unlikely(err_rx_buffer->skb)) { + dev_kfree_skb_any(err_rx_buffer->skb); + err_rx_buffer->skb =3D NULL; + } + + /* drop data until eop */ + if (rnpgbe_test_staterr(rx_desc, M_RXD_STAT_EOP)) + work_done++; + else + rx_ring->drop_status =3D true; + + rx_desc->wb.cmd =3D 0; + rx_ring->next_to_clean++; + if (rx_ring->next_to_clean >=3D rx_ring->count) + rx_ring->next_to_clean =3D 0; + continue; + } + + rx_buffer =3D rnpgbe_get_buffer(rx_ring, &skb, size); + + if (skb) + rnpgbe_add_rx_frag(rx_buffer, skb, size); + else + skb =3D rnpgbe_build_skb(rx_buffer, size); + + if (!skb) { + cleaned_count++; + + /* drop until eop if multiple descriptors */ + if (rnpgbe_test_staterr(rx_desc, M_RXD_STAT_EOP)) + work_done++; + else + rx_ring->drop_status =3D true; + + rx_desc->wb.cmd =3D 0; + rx_ring->next_to_clean++; + atomic64_inc(&rx_ring->stats.dropped); + if (rx_ring->next_to_clean >=3D rx_ring->count) + rx_ring->next_to_clean =3D 0; + + continue; + } + + rx_buffer->page =3D NULL; + rx_buffer->skb =3D NULL; + cleaned_count++; + + if (rnpgbe_is_non_eop(rx_ring, rx_desc, skb)) + continue; + + /* verify the packet layout is correct */ + if (rnpgbe_cleanup_headers(skb)) { + /* we should clean it since we used all info in it */ + atomic64_inc(&rx_ring->stats.dropped); + rx_desc->wb.cmd =3D 0; + work_done++; + continue; + } + + /* probably a little skewed due to removing CRC */ + total_rx_bytes +=3D skb->len; + rnpgbe_process_skb_fields(rx_ring, skb); + rx_desc->wb.cmd =3D 0; + napi_gro_receive(&q_vector->napi, skb); + /* Update packet statistics and NAPI work accounting. */ + total_rx_packets++; + work_done++; + } + + u64_stats_update_begin(&rx_ring->syncp); + rx_ring->stats.packets +=3D total_rx_packets; + rx_ring->stats.bytes +=3D total_rx_bytes; + u64_stats_update_end(&rx_ring->syncp); + + return work_done; +} + /** * rnpgbe_poll - NAPI polling callback * @napi: structure for representing this polling device * @budget: polling budget * - * Clean completed TX packets and complete NAPI polling when all queues are - * clean. + * Clean completed TX and RX packets and complete NAPI polling when all + * queues are clean. * - * Return: 0 if all TX queues are clean, otherwise @budget + * Return: number of RX packets processed, or @budget if polling continues **/ static int rnpgbe_poll(struct napi_struct *napi, int budget) { @@ -190,6 +613,7 @@ static int rnpgbe_poll(struct napi_struct *napi, int bu= dget) container_of(napi, struct mucse_q_vector, napi); bool clean_complete =3D true; struct mucse_ring *ring; + int per_ring_budget; int work_done =3D 0; =20 mucse_for_each_ring(ring, q_vector->tx) { @@ -201,6 +625,20 @@ static int rnpgbe_poll(struct napi_struct *napi, int b= udget) if (unlikely(!budget)) return 0; =20 + if (q_vector->rx.count > 1) + per_ring_budget =3D max(budget / q_vector->rx.count, 1); + else + per_ring_budget =3D budget; + + mucse_for_each_ring(ring, q_vector->rx) { + int cleaned =3D 0; + + cleaned =3D rnpgbe_clean_rx_irq(q_vector, ring, per_ring_budget); + work_done +=3D cleaned; + if (cleaned >=3D per_ring_budget) + clean_complete =3D false; + } + if (!clean_complete) return budget; =20 @@ -340,6 +778,7 @@ static int rnpgbe_alloc_q_vector(struct mucse *mucse, /* tie q_vector and mucse together */ mucse->q_vector[vector_idx] =3D q_vector; q_vector->mucse =3D mucse; + timer_setup(&q_vector->rx_alloc_timer, rnpgbe_rx_alloc_retry, 0); q_vector->hw_vector =3D vector_idx; /* Vector 0 is reserved for the mailbox. */ q_vector->hw_vector++; @@ -365,13 +804,18 @@ static int rnpgbe_alloc_q_vector(struct mucse *mucse, } =20 for (idx =3D 0; idx < rxr_count; idx++) { + ring->dev =3D &mucse->pdev->dev; mucse_add_ring(ring, &q_vector->rx); + ring->count =3D mucse->rx_ring_item_count; + ring->netdev =3D mucse->netdev; ring->queue_index =3D eth_queue_idx + idx; ring->rnpgbe_queue_idx =3D rxr_idx; ring->ring_addr =3D hw->hw_addr + RING_OFFSET(rxr_idx); ring->irq_mask =3D ring->ring_addr + RNPGBE_DMA_INT_MASK; ring->trig =3D ring->ring_addr + RNPGBE_DMA_INT_TRIG; ring->q_vector =3D q_vector; + ring->pfvfnum =3D hw->pfvfnum; + u64_stats_init(&ring->syncp); WRITE_ONCE(mucse->rx_ring[ring->queue_index], ring); rxr_idx +=3D step; ring++; @@ -394,6 +838,8 @@ static void rnpgbe_free_q_vector(struct mucse *mucse, i= nt vector_idx) struct mucse_q_vector *q_vector =3D mucse->q_vector[vector_idx]; struct mucse_ring *ring; =20 + timer_delete_sync(&q_vector->rx_alloc_timer); + mucse_for_each_ring(ring, q_vector->tx) WRITE_ONCE(mucse->tx_ring[ring->queue_index], NULL); mucse_for_each_ring(ring, q_vector->rx) @@ -713,6 +1159,31 @@ static int rnpgbe_wait_tx_dma_idle(struct mucse *mucs= e) return err; } =20 +static int rnpgbe_wait_rx_dma_idle(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 dma_status; + int err; + + /* A timeout indicates a terminal AXI fault. Hardware stops all PCIe + * DMA requests in this state, including requests accepted before the + * fault. Teardown may therefore release mappings and descriptor memory. + * Recovery requires a chip-level reset. + */ + err =3D readl_poll_timeout(hw->hw_addr + RNPGBE_DMA_STATUS, + dma_status, + (dma_status & RNPGBE_DMA_RX_STATUS) =3D=3D + RNPGBE_DMA_RX_STATUS, + 10, 100000); + if (err) { + set_bit(__MUCSE_AXI_FAULT, &mucse->state); + dev_err(&mucse->pdev->dev, + "RX DMA failed to quiesce, status %#x\n", dma_status); + } + + return err; +} + /** * rnpgbe_stop_all_tx_rings - Stop TX DMA on all queues * @mucse: board private structure @@ -832,6 +1303,99 @@ void rnpgbe_clean_all_tx_rings(struct mucse *mucse) rnpgbe_clean_tx_ring(mucse->tx_ring[i]); } =20 +static void rnpgbe_stop_rx_ring(struct mucse_ring *rx_ring) +{ + /* Stop hw. hardware design guarantees: + * - No new descriptors will be fetched after RX_START=3D0 + * - No DMA will be initiated for already-fetched descriptors + */ + mucse_ring_wr32(rx_ring, RNPGBE_RX_START, 0); + /* Flush posted write to ensure hardware sees RX_START=3D0 */ + (void)mucse_ring_rd32(rx_ring, RNPGBE_RX_START); +} + +/** + * rnpgbe_stop_all_rx_rings - Stop RX DMA for all queues + * @mucse: board private structure + * + * Return: 0 if RX DMA is quiesced and no AXI fault is latched, + * negative errno otherwise + **/ +static int rnpgbe_stop_all_rx_rings(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 dma_axi_ctl; + int err =3D 0; + + for (int i =3D 0; i < mucse->num_rx_queues; i++) + rnpgbe_stop_rx_ring(mucse->rx_ring[i]); + + if (mucse->num_rx_queues && + !test_bit(__MUCSE_AXI_FAULT, &mucse->state)) + err =3D rnpgbe_wait_rx_dma_idle(mucse); + + dma_axi_ctl =3D mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + dma_axi_ctl &=3D ~RX_AXI_RW_EN; + mucse_hw_wr32(hw, RNPGBE_DMA_AXI_EN, dma_axi_ctl); + /* Flush the posted write before continuing. */ + (void)mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + + if (!test_bit(__MUCSE_AXI_FAULT, &mucse->state)) + return 0; + + return err ? err : -EIO; +} + +/** + * rnpgbe_clean_rx_ring - Free Rx Buffers per Queue + * @rx_ring: ring to free buffers from + **/ +static void rnpgbe_clean_rx_ring(struct mucse_ring *rx_ring) +{ + struct mucse_rx_buffer *rx_buffer; + u16 i; + + /* ring already cleared, nothing to do */ + if (!rx_ring->rx_buffer_info) + return; + /* Free all the Rx ring sk_buffs */ + for (i =3D 0; i < rx_ring->count; i++) { + rx_buffer =3D &rx_ring->rx_buffer_info[i]; + + if (rx_buffer->skb) { + struct sk_buff *skb =3D rx_buffer->skb; + + dev_kfree_skb(skb); + rx_buffer->skb =3D NULL; + } + + if (rx_buffer->page) { + page_pool_put_full_page(rx_ring->page_pool, + rx_buffer->page, false); + rx_buffer->page =3D NULL; + } + } + + rx_ring->next_to_clean =3D 0; + rx_ring->next_to_use =3D 0; +} + +/** + * rnpgbe_clean_all_rx_rings - Free Rx buffers for all queues + * @mucse: board private structure + **/ +static void rnpgbe_clean_all_rx_rings(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_rx_queues; i++) + rnpgbe_clean_rx_ring(mucse->rx_ring[i]); +} + +static void rnpgbe_cancel_rx_retry_timers(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_q_vectors; i++) + timer_delete_sync(&mucse->q_vector[i]->rx_alloc_timer); +} + void rnpgbe_down(struct mucse *mucse) { struct net_device *netdev =3D mucse->netdev; @@ -841,7 +1405,10 @@ void rnpgbe_down(struct mucse *mucse) synchronize_net(); rnpgbe_irq_disable(mucse); netif_tx_disable(netdev); + rnpgbe_stop_all_rx_rings(mucse); + rnpgbe_cancel_rx_retry_timers(mucse); rnpgbe_clean_all_tx_rings(mucse); + rnpgbe_clean_all_rx_rings(mucse); } =20 /** @@ -854,6 +1421,7 @@ void rnpgbe_up_complete(struct mucse *mucse) =20 rnpgbe_configure_msix(mucse); rnpgbe_napi_enable_all(mucse); + rnpgbe_schedule_rx_retry(mucse); rnpgbe_irq_enable(mucse); netif_tx_start_all_queues(netdev); } @@ -1258,5 +1826,225 @@ void rnpgbe_get_stats64(struct net_device *netdev, stats->tx_bytes +=3D bytes; } } + + for (i =3D 0; i < mucse->num_rx_queues; i++) { + struct mucse_ring *ring =3D READ_ONCE(mucse->rx_ring[i]); + u64 bytes, packets, dropped; + unsigned int start; + + if (ring) { + do { + start =3D u64_stats_fetch_begin(&ring->syncp); + packets =3D ring->stats.packets; + bytes =3D ring->stats.bytes; + } while (u64_stats_fetch_retry(&ring->syncp, start)); + dropped =3D atomic64_read(&ring->stats.dropped); + + stats->rx_packets +=3D packets; + stats->rx_dropped +=3D dropped; + stats->rx_bytes +=3D bytes; + } + } rcu_read_unlock(); } + +static int mucse_alloc_page_pool(struct mucse_ring *rx_ring) +{ + struct page_pool_params pp_params =3D { + .flags =3D PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV, + .order =3D 0, + .pool_size =3D rx_ring->count, + .nid =3D dev_to_node(rx_ring->dev), + .dev =3D rx_ring->dev, + .napi =3D &rx_ring->q_vector->napi, + .dma_dir =3D DMA_FROM_DEVICE, + .offset =3D 0, + .max_len =3D PAGE_SIZE, + .netdev =3D rx_ring->netdev, + .queue_idx =3D rx_ring->queue_index, + }; + int ret =3D 0; + + rx_ring->page_pool =3D page_pool_create(&pp_params); + if (IS_ERR(rx_ring->page_pool)) { + ret =3D PTR_ERR(rx_ring->page_pool); + rx_ring->page_pool =3D NULL; + } + + return ret; +} + +/** + * rnpgbe_setup_rx_resources - allocate Rx resources (Descriptors) + * @rx_ring: rx descriptor ring (for a specific queue) to setup + * + * Return: 0 on success, negative on failure + **/ +static int rnpgbe_setup_rx_resources(struct mucse_ring *rx_ring) +{ + struct device *dev =3D rx_ring->dev; + int size; + + size =3D sizeof(struct mucse_rx_buffer) * rx_ring->count; + + rx_ring->rx_buffer_info =3D vzalloc(size); + + if (!rx_ring->rx_buffer_info) + goto err_return; + /* Round up to nearest 4K */ + rx_ring->size =3D rx_ring->count * sizeof(union rnpgbe_rx_desc); + rx_ring->size =3D ALIGN(rx_ring->size, 4096); + rx_ring->desc =3D dma_alloc_coherent(dev, rx_ring->size, &rx_ring->dma, + GFP_KERNEL); + if (!rx_ring->desc) + goto err_free_buffer; + + rx_ring->next_to_clean =3D 0; + rx_ring->next_to_use =3D 0; + + if (mucse_alloc_page_pool(rx_ring)) + goto err_free_desc; + + return 0; +err_free_desc: + dma_free_coherent(dev, rx_ring->size, rx_ring->desc, + rx_ring->dma); + rx_ring->desc =3D NULL; +err_free_buffer: + vfree(rx_ring->rx_buffer_info); +err_return: + rx_ring->rx_buffer_info =3D NULL; + return -ENOMEM; +} + +/** + * rnpgbe_free_rx_resources - Free Rx Resources + * @rx_ring: ring to clean the resources from + * + * RX DMA must be quiesced and all ring-held buffers released before this + * function is called. + **/ +static void rnpgbe_free_rx_resources(struct mucse_ring *rx_ring) +{ + vfree(rx_ring->rx_buffer_info); + rx_ring->rx_buffer_info =3D 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 =3D NULL; + if (rx_ring->page_pool) { + page_pool_destroy(rx_ring->page_pool); + rx_ring->page_pool =3D NULL; + } +} + +/** + * rnpgbe_setup_all_rx_resources - allocate all queues Rx resources + * @mucse: pointer to private structure + * + * Return: 0 on success, negative on failure + **/ +int rnpgbe_setup_all_rx_resources(struct mucse *mucse) +{ + int i, err =3D 0; + + for (i =3D 0; i < mucse->num_rx_queues; i++) { + err =3D rnpgbe_setup_rx_resources(mucse->rx_ring[i]); + if (!err) + continue; + + goto err_setup_rx; + } + + return 0; +err_setup_rx: + while (i--) + rnpgbe_free_rx_resources(mucse->rx_ring[i]); + return err; +} + +/** + * rnpgbe_free_all_rx_resources - Free Rx Resources for All Queues + * @mucse: pointer to private structure + * + * Free all receive software resources + **/ +void rnpgbe_free_all_rx_resources(struct mucse *mucse) +{ + for (int i =3D 0; i < mucse->num_rx_queues; i++) + rnpgbe_free_rx_resources(mucse->rx_ring[i]); +} + +/** + * rnpgbe_configure_rx_ring - Configure Rx ring info to hw + * @mucse: pointer to private structure + * @ring: structure containing ring specific data + * + * Configure the Rx descriptor ring after a reset. + **/ +static void rnpgbe_configure_rx_ring(struct mucse *mucse, + struct mucse_ring *ring) +{ + struct mucse_hw *hw =3D &mucse->hw; + + /* Set the descriptor registers after RX DMA has quiesced. */ + mucse_ring_wr32(ring, RNPGBE_RX_BASE_ADDR_LO, (u32)ring->dma); + mucse_ring_wr32(ring, RNPGBE_RX_BASE_ADDR_HI, + (u32)((u64)ring->dma >> 32) | + ((u32)hw->pfvfnum << 24)); + mucse_ring_wr32(ring, RNPGBE_RX_LEN, ring->count); + ring->tail =3D ring->ring_addr + RNPGBE_RX_TAIL; + ring->next_to_clean =3D mucse_ring_rd32(ring, RNPGBE_RX_HEAD) % + ring->count; + ring->next_to_use =3D ring->next_to_clean; + ring->drop_status =3D false; + mucse_ring_wr32(ring, RNPGBE_RX_SG_LEN, M_DEFAULT_SG); + mucse_ring_wr32(ring, RNPGBE_RX_FETCH, M_DEFAULT_RX_FETCH); + mucse_ring_wr32(ring, RNPGBE_RX_TIMEOUT_TH, 0); + mucse_ring_wr32(ring, RNPGBE_RX_INT_TIMER, + M_DEFAULT_INT_TIMER_R * hw->cycles_per_us); + mucse_ring_wr32(ring, RNPGBE_RX_INT_PKTCNT, M_DEFAULT_RX_INT_PKTCNT); + /* Fill the ring before RX DMA is enabled. Any allocation shortfall is + * retried after NAPI is enabled by rnpgbe_up_complete(). + */ + rnpgbe_alloc_rx_buffers(ring, mucse_desc_unused_rx(ring)); +} + +/** + * rnpgbe_configure_rx - Configure Receive Unit after Reset + * @mucse: pointer to private structure + * + * Configure the Rx unit after a reset. + * + * Return: 0 on success, negative errno if RX DMA does not quiesce or + * an AXI fault is already latched + **/ +int rnpgbe_configure_rx(struct mucse *mucse) +{ + struct mucse_hw *hw =3D &mucse->hw; + u32 dma_axi_ctl; + int err; + + err =3D rnpgbe_stop_all_rx_rings(mucse); + if (err) + return err; + + for (int i =3D 0; i < mucse->num_rx_queues; i++) + rnpgbe_configure_rx_ring(mucse, mucse->rx_ring[i]); + + /* Ensure all ring configuration is visible before enabling RX DMA. */ + wmb(); + dma_axi_ctl =3D mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + dma_axi_ctl |=3D RX_AXI_RW_EN; + mucse_hw_wr32(hw, RNPGBE_DMA_AXI_EN, dma_axi_ctl); + /* Ensure RX_AXI_RW_EN is visible before starting the rings. */ + (void)mucse_hw_rd32(hw, RNPGBE_DMA_AXI_EN); + + for (int i =3D 0; i < mucse->num_rx_queues; i++) + mucse_ring_wr32(mucse->rx_ring[i], RNPGBE_RX_START, 1); + + return 0; +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.h index 1fe290ed224d..76b5b0c0ae4a 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h @@ -9,12 +9,27 @@ struct mucse_hw; struct mucse_ring; =20 #define RING_OFFSET(n) (0x1000 + 0x100 * (n)) +#define RNPGBE_RX_START 0x10 #define RNPGBE_TX_START 0x18 #define RNPGBE_DMA_INT_MASK 0x24 #define TX_INT_MASK BIT(1) #define RX_INT_MASK BIT(0) #define INT_VALID (BIT(16) | BIT(17)) #define RNPGBE_DMA_INT_TRIG 0x2c /* lost-interrupt recovery trigger = */ +#define RNPGBE_RX_BASE_ADDR_HI 0x30 +#define RNPGBE_RX_BASE_ADDR_LO 0x34 +#define RNPGBE_RX_LEN 0x38 +#define RNPGBE_RX_HEAD 0x3c +#define RNPGBE_RX_TAIL 0x40 +#define M_DEFAULT_RX_FETCH 0x100020 +#define RNPGBE_RX_FETCH 0x44 +#define M_DEFAULT_INT_TIMER_R 30 +#define RNPGBE_RX_INT_TIMER 0x48 +#define M_DEFAULT_RX_INT_PKTCNT 64 +#define RNPGBE_RX_INT_PKTCNT 0x4c +#define RNPGBE_RX_TIMEOUT_TH 0x54 +#define M_DEFAULT_SG 96 /* unit 16b, 1536 bytes */ +#define RNPGBE_RX_SG_LEN 0x58 #define RNPGBE_TX_BASE_ADDR_HI 0x60 #define RNPGBE_TX_BASE_ADDR_LO 0x64 #define RNPGBE_TX_LEN 0x68 @@ -37,13 +52,14 @@ struct mucse_ring; /* 2 desc gap to keep tail from touching head */ /* 1 desc for context descriptor */ #define RESV_DESC_NEEDED 3 +#define RNPGBE_SKB_PAD (NET_SKB_PAD + NET_IP_ALIGN) + /* Hardware requires this to be nonzero */ #define M_DEFAULT_MAC_IP_LEN 20 #define mucse_for_each_ring(pos, head)\ for (typeof((head).ring) __pos =3D (head).ring;\ __pos ? ({ pos =3D __pos; 1; }) : 0;\ __pos =3D __pos->next) - int rnpgbe_init_interrupt_scheme(struct mucse *mucse); void rnpgbe_clear_interrupt_scheme(struct mucse *mucse); int rnpgbe_request_mbx_irq(struct mucse *mucse); @@ -54,6 +70,7 @@ void rnpgbe_irq_disable(struct mucse *mucse); void rnpgbe_down(struct mucse *mucse); void rnpgbe_up_complete(struct mucse *mucse); int rnpgbe_configure_tx(struct mucse *mucse); +int rnpgbe_configure_rx(struct mucse *mucse); void rnpgbe_clean_all_tx_rings(struct mucse *mucse); int rnpgbe_setup_all_tx_resources(struct mucse *mucse); void rnpgbe_free_all_tx_resources(struct mucse *mucse); @@ -61,4 +78,6 @@ netdev_tx_t rnpgbe_xmit_frame_ring(struct sk_buff *skb, struct mucse_ring *tx_ring); void rnpgbe_get_stats64(struct net_device *netdev, struct rtnl_link_stats64 *stats); +int rnpgbe_setup_all_rx_resources(struct mucse *mucse); +void rnpgbe_free_all_rx_resources(struct mucse *mucse); #endif diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_main.c index cdda5f3d09b2..b0363509a3c5 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c @@ -31,13 +31,19 @@ static struct pci_device_id rnpgbe_pci_tbl[] =3D { * rnpgbe_configure - Configure the hardware * @mucse: pointer to private structure * - * Configure Tx registers in hardware. + * Configure Tx and Rx registers in hardware. * * Return: 0 on success, negative errno if hardware configuration fails **/ static int rnpgbe_configure(struct mucse *mucse) { - return rnpgbe_configure_tx(mucse); + int err; + + err =3D rnpgbe_configure_tx(mucse); + if (err) + return err; + + return rnpgbe_configure_rx(mucse); } =20 /** @@ -69,13 +75,18 @@ static int rnpgbe_open(struct net_device *netdev) err =3D rnpgbe_setup_all_tx_resources(mucse); if (err) goto err_free_irqs; + err =3D rnpgbe_setup_all_rx_resources(mucse); + if (err) + goto err_free_tx; =20 err =3D rnpgbe_configure(mucse); if (err) - goto err_free_tx; + goto err_free_rx; rnpgbe_up_complete(mucse); =20 return 0; +err_free_rx: + rnpgbe_free_all_rx_resources(mucse); err_free_tx: rnpgbe_clean_all_tx_rings(mucse); rnpgbe_free_all_tx_resources(mucse); @@ -100,6 +111,7 @@ static int rnpgbe_close(struct net_device *netdev) rnpgbe_down(mucse); rnpgbe_free_irq(mucse); rnpgbe_free_all_tx_resources(mucse); + rnpgbe_free_all_rx_resources(mucse); =20 return 0; } @@ -137,6 +149,7 @@ static const struct net_device_ops rnpgbe_netdev_ops = =3D { static void rnpgbe_sw_init(struct mucse *mucse) { mucse->tx_ring_item_count =3D M_DEFAULT_TXD; + mucse->rx_ring_item_count =3D M_DEFAULT_RXD; mucse->tx_work_limit =3D M_DEFAULT_TX_WORK; } =20 --=20 2.50.1 From nobody Thu Sep 24 20:41:59 2026 Received: from smtpbg150.qq.com (smtpbg150.qq.com [18.132.163.193]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id DC9433B8948; Sun, 20 Sep 2026 09:28:08 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=18.132.163.193 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896493; cv=none; b=clYis+H9CGdrzoUiSi1LBflK7cZITH3M+rngpHw3FgIcqrbaGZ/Mw/Fb1u3hnsVMbIDqr941r3z4FevRt2qyDlRuKI7tsgZokj58rSU342GjwT5/EtPxBuS0GZNLjYyBdDrzmfEsNoNDE75K6kmd0Tb2DxkuPmamB8QekYjkXeo= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896493; c=relaxed/simple; bh=5TUnrPmqwGV1Gsud7gNOYe1pyIJxfgNKUiEOMupFByY=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=Dw4YKUafj2L3vFe65ICuWTHDPLrieGAReXSW/YhWfwkJhvQusUSppyMR9LzfsyELNiMoJoany0RSqXBYeZntT1oT8WG9CountsJjdV3yABGic7R+PugErEqWJYHnqYb1eLuZm+1IRut+czFGjHby8ov8PUJcURVSZRa1XjfuN+g= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com; spf=pass smtp.mailfrom=mucse.com; arc=none smtp.client-ip=18.132.163.193 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mucse.com X-QQ-mid: esmtpgz14t1789896413t3c276fd0 X-QQ-Originating-IP: S3RiTUx+dK8uextlYLQz6TdtJVUe9ATjfZly5O6k10k= Received: from nic-Precision-5820-Tower ( [203.174.112.180]) by bizesmtp.qq.com (ESMTP) with id ; Sun, 20 Sep 2026 17:26:50 +0800 (CST) X-QQ-SSF: 0000000000000000000000000000000 X-QQ-GoodBg: 0 X-BIZMAIL-ID: 2193611732860423393 EX-QQ-RecipientCnt: 11 From: Dong Yibo To: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, vadim.fedorenko@linux.dev, u.kleine-koenig@baylibre.com Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, dong100@mucse.com, yaojun@mucse.com Subject: [PATCH net-next v13 4/5] net: rnpgbe: Add receive mode support Date: Sun, 20 Sep 2026 17:26:49 +0800 Message-ID: <057BAFCF77CDC8BE+20260920092649.345634-1-dong100@mucse.com> X-Mailer: git-send-email 2.50.1 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-QQ-SENDSIZE: 520 Feedback-ID: esmtpgz:mucse.com:qybglogicsvrgz:qybglogicsvrgz3a-1 X-QQ-XMAILINFO: MyIXMys/8kCtF8eXvJzNiGCC1xCQbvbQ/Ef5/qgautOdAy2AiAhxqCVM 0SdCCzkw181+LYg5nsrsbirtl4jfeuUIZpDRs8mu5cMrCbOuH6W+9KzraU2DgH2kbE1iJ+P yTmKh1JpZAIUqrKxb9LRD8iw8TPt8UdxuvHFfhVWUf4/EtdkEmQWPwJt7w+eXqGK/K1bUo+ TIUUFWvsW3iNrU3LfOiIrVjjjg2rODsoOe3MX6RoVYyjd8viT/McLCGtQKMT9c62s5VVij5 Dobu0QTAXReUdsKCKig+c+nixd191Q018cBSmr6RCQQKaYUPFGNIouVZQj91uq6O9W9d5en 3k2CcmUpn/fMRlAEJkCoP52Q5gogz3bLj1RQh1coqPshMo3WBqtWbioYIBwyeSXEGLWB8p2 s099juV0zkqSYrDqJLiO/C5HrVBUTyVbDKNz90nejDrlyj5C4AqhI7FvVGRWxDywi60R5Kg MShurd+Q1/EJ5ARyq3rvr2LBp4DZ2jDl8rLhkTNHTPOzBVR4oirbP7cF94YkhIgFXJzp7pT 4r4o5oupHvMoJY4FnaqNku8KuhIEPbzpj1qmBYjEi6f4jsjylUHuJ6Rxbphlt4SUSigHSwA CrcrvRKzinNFrazkqUxZEUeeUol7Wbmzpc3PUR2bZg/GPEUvSTA8W5IlYQqfsNHUB30c9IP ZXIyi2XcSA23Fl6EwG3t4whPKho+P2KLZMWUWCVizMfJkpm05xRSxe41dhzbPE8SwH3s58X 5uhI87AJYI3/ZAgidwXamkZBuPI6o80vNUSUsFBFk8s67+gj+egBg8C9L9PzoEoXYunFgxO fnaxwrIzq3vL0earPaeadruKBgtrmRNovp1bCZLCw8ilsFJpgPkZkKwfBwWKDDEltXAe3KU fYj4v+xIe+AEf02fL4zhyKWXRftnqRWgmfnydcJqv20ZvnLhtj7Td2J75vouktnGsAuS4F9 7rOTkU+u9sXyiU8Y01yZjm1Axj/M0FcvZUuj0fwjMJBm5P1fxQLI10FE2dIDQCbkOPCgFHB EmIls66KxK8/HU534mFnB5ADVsjc2pARmqjESMLhp727c5WAn1BNLKEwnanUYL+xgzZEOuT A== X-QQ-XMRINFO: Mp0Kj//9VHAxzExpfF+O8yhSrljjwrznVg== X-QQ-RECHKSPAM: 0 Content-Type: text/plain; charset="utf-8" Add chip-level receive filtering and the ndo_set_rx_mode callback. Program the primary and additional unicast addresses into the available RAR entries, and use the 4096-bit multicast hash table for multicast addresses. Advertise RAR-based secondary unicast filtering to the networking core, so it does not force promiscuous mode for a non-empty secondary unicast list. Fall back to accepting all unicast addresses if the RAR table is exhausted, and handle IFF_PROMISC and IFF_ALLMULTI through the chip-level filter control register. Signed-off-by: Dong Yibo --- drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h | 1 + .../net/ethernet/mucse/rnpgbe/rnpgbe_chip.c | 75 +++++++++++++++++++ drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h | 15 ++++ .../net/ethernet/mucse/rnpgbe/rnpgbe_main.c | 12 ++- 4 files changed, 100 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h b/drivers/net/ether= net/mucse/rnpgbe/rnpgbe.h index c1a72e8e2770..bd7491671e1e 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h @@ -231,6 +231,7 @@ int rnpgbe_send_notify(struct mucse_hw *hw, bool enable, int mode); int rnpgbe_init_hw(struct mucse_hw *hw, int board_type); +void rnpgbe_set_rx_mode(struct net_device *netdev); =20 /* Device IDs */ #define PCI_VENDOR_ID_MUCSE 0x8848 diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_chip.c index 291e77d573fe..4e2835b71f32 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c @@ -149,3 +149,78 @@ int rnpgbe_init_hw(struct mucse_hw *hw, int board_type) =20 return 0; } + +static void rnpgbe_set_rar(struct mucse_hw *hw, u32 index, const u8 *addr) +{ + u32 rar_low, rar_high; + + /* The RAR stores the Ethernet address in reverse byte order. */ + rar_low =3D (u32)addr[5] | ((u32)addr[4] << 8) | + ((u32)addr[3] << 16) | ((u32)addr[2] << 24); + rar_high =3D (u32)addr[1] | ((u32)addr[0] << 8) | + RNPGBE_RX_RAR_VALID; + mucse_hw_wr32(hw, RNPGBE_RX_RAR_HIGH(index), + rar_high & ~RNPGBE_RX_RAR_VALID); + mucse_hw_wr32(hw, RNPGBE_RX_RAR_LOW(index), rar_low); + mucse_hw_wr32(hw, RNPGBE_RX_RAR_HIGH(index), rar_high); +} + +static void rnpgbe_clear_rar(struct mucse_hw *hw, u32 index) +{ + mucse_hw_wr32(hw, RNPGBE_RX_RAR_HIGH(index), 0); + mucse_hw_wr32(hw, RNPGBE_RX_RAR_LOW(index), 0); +} + +void rnpgbe_set_rx_mode(struct net_device *netdev) +{ + u32 mcast_hash[RNPGBE_RX_MCAST_HASH_ENTRIES] =3D {}; + u32 mcast_ctrl =3D RNPGBE_RX_UCAST_TABLE_EN; + u32 filter_ctrl =3D RNPGBE_RX_FILTER_BCAST; + struct netdev_hw_addr *ha; + struct mucse_hw *hw; + struct mucse *mucse; + int rar =3D 1; + u16 hash; + int i; + + mucse =3D netdev_priv(netdev); + hw =3D &mucse->hw; + + /* RAR 0 always holds the interface's primary unicast address. */ + rnpgbe_set_rar(hw, 0, netdev->dev_addr); + + netdev_for_each_uc_addr(ha, netdev) { + if (rar =3D=3D RNPGBE_RX_RAR_ENTRIES) { + filter_ctrl |=3D RNPGBE_RX_FILTER_UCAST_ALL; + break; + } + + rnpgbe_set_rar(hw, rar, ha->addr); + rar++; + } + + for (; rar < RNPGBE_RX_RAR_ENTRIES; rar++) + rnpgbe_clear_rar(hw, rar); + + if (netdev->flags & IFF_PROMISC) { + filter_ctrl |=3D RNPGBE_RX_FILTER_UCAST_ALL | + RNPGBE_RX_FILTER_MCAST_ALL; + } else if (netdev->flags & IFF_ALLMULTI) { + filter_ctrl |=3D RNPGBE_RX_FILTER_MCAST_ALL; + } else { + /* MTA type 0 uses ((addr[4] << 8) | addr[5]) & 0xfff. */ + netdev_for_each_mc_addr(ha, netdev) { + hash =3D ((ha->addr[4] & 0xf) << 8) | ha->addr[5]; + mcast_hash[hash >> 5] |=3D BIT(hash & 0x1f); + } + + if (!netdev_mc_empty(netdev)) + mcast_ctrl |=3D RNPGBE_RX_MCAST_HASH_EN; + } + + for (i =3D 0; i < RNPGBE_RX_MCAST_HASH_ENTRIES; i++) + mucse_hw_wr32(hw, RNPGBE_RX_MCAST_HASH(i), mcast_hash[i]); + + mucse_hw_wr32(hw, RNPGBE_RX_MCAST_CTRL, mcast_ctrl); + mucse_hw_wr32(hw, RNPGBE_RX_FILTER_CTRL, filter_ctrl); +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h b/drivers/net/et= hernet/mucse/rnpgbe/rnpgbe_hw.h index 1d87edfba3d7..feb84704d602 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h @@ -24,5 +24,20 @@ #define RNPGBE_DMA_AXI_EN 0x0010 #define RNPGBE_TX_MIN_PKT_LEN 33 =20 +#define MUCSE_ETH_OFF(_n) (0x10000 + (_n)) +#define RNPGBE_RX_RAR_LOW(_n) MUCSE_ETH_OFF(0xa000 + 4 * (_n)) +#define RNPGBE_RX_RAR_HIGH(_n) MUCSE_ETH_OFF(0xa400 + 4 * (_n)) +#define RNPGBE_RX_MCAST_HASH(_n) MUCSE_ETH_OFF(0xac00 + 4 * (_n)) +#define RNPGBE_RX_FILTER_CTRL MUCSE_ETH_OFF(0x9110) +#define RNPGBE_RX_MCAST_CTRL MUCSE_ETH_OFF(0x9114) +#define RNPGBE_RX_RAR_VALID BIT(31) +#define RNPGBE_RX_FILTER_BCAST BIT(10) +#define RNPGBE_RX_FILTER_UCAST_ALL BIT(9) +#define RNPGBE_RX_FILTER_MCAST_ALL BIT(8) +#define RNPGBE_RX_MCAST_HASH_EN BIT(4) +#define RNPGBE_RX_UCAST_TABLE_EN BIT(3) +/* The final two of the 32 hardware RAR entries are reserved for NCSI. */ +#define RNPGBE_RX_RAR_ENTRIES 30 +#define RNPGBE_RX_MCAST_HASH_ENTRIES 128 #define RNPGBE_MAX_QUEUES 8 #endif /* _RNPGBE_HW_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_main.c index b0363509a3c5..35eb858eb2c1 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c @@ -43,6 +43,10 @@ static int rnpgbe_configure(struct mucse *mucse) if (err) return err; =20 + netif_addr_lock_bh(mucse->netdev); + rnpgbe_set_rx_mode(mucse->netdev); + netif_addr_unlock_bh(mucse->netdev); + return rnpgbe_configure_rx(mucse); } =20 @@ -140,9 +144,10 @@ static netdev_tx_t rnpgbe_xmit_frame(struct sk_buff *s= kb, } =20 static const struct net_device_ops rnpgbe_netdev_ops =3D { - .ndo_open =3D rnpgbe_open, - .ndo_stop =3D rnpgbe_close, - .ndo_start_xmit =3D rnpgbe_xmit_frame, + .ndo_open =3D rnpgbe_open, + .ndo_stop =3D rnpgbe_close, + .ndo_start_xmit =3D rnpgbe_xmit_frame, + .ndo_set_rx_mode =3D rnpgbe_set_rx_mode, .ndo_get_stats64 =3D rnpgbe_get_stats64, }; =20 @@ -225,6 +230,7 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, } =20 netdev->netdev_ops =3D &rnpgbe_netdev_ops; + netdev->priv_flags |=3D IFF_UNICAST_FLT; rnpgbe_sw_init(mucse); err =3D rnpgbe_reset_hw(hw); if (err) { --=20 2.50.1 From nobody Thu Sep 24 20:41:59 2026 Received: from smtpbgbr1.qq.com (smtpbgbr1.qq.com [54.207.19.206]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id D5DB43E51E4; Sun, 20 Sep 2026 09:27:50 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=54.207.19.206 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896475; cv=none; b=rl288J9LOGd2i+NQutFoNC1JA2FzpMMKfTdhsqP7fl4uh0TwEkL66++f53ONFgseJU+1BlaiLr4MlfPmTrsrqczahA0esdr0pe/BVNpZ4za4kyYW45OKEXHfRIE9imn6C1UZII1+MBsYZmTEb3w1lNmmZCvvng0v7ahho6+hNKM= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1789896475; c=relaxed/simple; bh=nQ6IzEryi3nz6wxgeDZUCbYsoZ+YAv77pLjg4htTm1U=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=UJxXNW0iNEd/WMRL0SWiP7PHTM5ijxU5N3TwowOyfJOCmZwMILQ3UnHxIKO8vSTCd01bH5Eu//MdrZRHBBLNCdPZAxES0tz3RvMxM6xgtjgED0dHcoJTaR7bSe6Zreiprb+rllBC8rkQjtZhIMJTf/oO6E7mOlar4/rK/EGA1cQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com; spf=pass smtp.mailfrom=mucse.com; arc=none smtp.client-ip=54.207.19.206 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=mucse.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mucse.com X-QQ-mid: zesmtpgz6t1789896424t4e4cbacb X-QQ-Originating-IP: b3kW9XhTQyVeiiuFQD/3VCtiOt6wA6o3uqVOqwFupl0= Received: from nic-Precision-5820-Tower ( [203.174.112.180]) by bizesmtp.qq.com (ESMTP) with id ; Sun, 20 Sep 2026 17:27:01 +0800 (CST) X-QQ-SSF: 0000000000000000000000000000000 X-QQ-GoodBg: 0 X-BIZMAIL-ID: 15872735073132448030 EX-QQ-RecipientCnt: 11 From: Dong Yibo To: andrew+netdev@lunn.ch, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, vadim.fedorenko@linux.dev, u.kleine-koenig@baylibre.com Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org, dong100@mucse.com, yaojun@mucse.com Subject: [PATCH net-next v13 5/5] net: rnpgbe: Add link status handling support Date: Sun, 20 Sep 2026 17:27:00 +0800 Message-ID: X-Mailer: git-send-email 2.50.1 In-Reply-To: References: Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable X-QQ-SENDSIZE: 520 Feedback-ID: zesmtpgz:mucse.com:qybglogicsvrgz:qybglogicsvrgz3a-1 X-QQ-XMAILINFO: N+ARrNpmMERciaTr23Q6VUpaTJ618Tj5u3LKvvriSvtvdZXg6pvKG36a YePYGLSJ1DolRUGiyJr5WjzBaM5LvTfy4FxfsWCEb9UWtfv9+udlaBTreBFhhR0PmAvt7lf slr/SXYxE7/2MoTOM+wBPeClqtCGR9FAbEdHf17vq96Td6QPQNYry+WD5gCKSnPbRHrApPN zR3y+lNYp/a8xZiJaCR8wz5PHregEvYur/HgJBcQywPNLtye40AfU4ToAgiTT7+P8HL/E7V VfVj2mJwb2H3GQAJpQ54JzyDaCISKr4FFqos20ZGpUepY4GKAkpHubTA/Gcd3tVI3Rk7sdC iW7nDZ4MwVX7UelsWQlqgzunIGXQ0pjyEXuFsiuwcXHjLOW5CQTkjrdq/hWRbVCDXG/vU82 Kis65Bg+obWzyhLNnxNcgSpNAMWh6c0CATbaiLzdgDw/9ZRV/bDQhWcstMGWpTld0VmyYmY A4S4Kdy7ouiJ/S+MXsiVUaHfGKSkb6Oq1fEooV3eLqHBdVM5pzqrweaoDq0i2W5W45lNriW wJCKOpySO6ipedFHl3i8e/zeNxu5qcH9jZk57YVyudG/cigwoxEH1DAuLqLhCCfUgPBHlce 8ZRFfxrti7Mihd16HpTIHyWB9iXtGIQcFUeTkYzIr0WZx3tmpDjbLmmqcAh9wwmEjvQ2vSP gqWiznMDXDk0nD46k1ATnCiaF1yXbuRynYiPayonshHwlR9TP/gbbBg2cdf8qFqTMEJO8cK mANzTyWwIKY9qn5BEw17o277AxKeakOk1zySaVW5lZvwoNovkXN5G5Oo79VYCBIfsXqorGr fy40sWDabEZx26Q2t0Kl8bf3WvScd+PbVJSEn9ezhegCE1KqOjBFGBwO+HTa5aX2g79dzXs dFMgo9kdg1UiyDzQ92kGwesh7Z8v01wdxy3VVNzGEhc8XkLGnKQXIx3pDgQj7nQ1zE9Cmrr QkomU2eRE7tvb4dL5fcKkqRoQumGwXc3KUHZtsiw0PAE6+SvUYoUii+OO20NkMi+10P2F2w 4z4MOG3SC3aeHjSOxhXtCSd3PD2ScbVuV8eo96Z7q3vSTcS54FJRZAMe5bns3HTNiRnxa1h qe2KoyFfxGyfO2+koysl7Q= X-QQ-XMRINFO: MPJ6Tf5t3I/ylTmHUqvI8+Wpn+Gzalws3A== X-QQ-RECHKSPAM: 0 Content-Type: text/plain; charset="utf-8" Add link status handling for the RNPGBE driver. Introduce link state data structures for speed, duplex and link state, and process firmware link events delivered over the mailbox, since firmware controls the link and phylink is not used. Add a service task to handle pending events and manage carrier status with netif_carrier_on()/netif_carrier_off(). Notify firmware when the interface is brought up and after the data path is quiesced during teardown, and fail interface open if the firmware port-up or link-report setup fails. Validate firmware link events before updating the cached link state. Enable the GMAC receiver in receive-all mode after the chip-level receive filter has been configured, and document the driver-to-firmware link-state snapshot and reset semantics. Configure firmware after reset to enable autonegotiation. Advertise 10/100Mb/s half- and full-duplex modes and 1000 Mb/s full-duplex mode. Fail probe if firmware does not acknowledge the link setup command. Signed-off-by: Dong Yibo --- drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h | 15 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_chip.c | 54 +++- drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h | 17 ++ .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.c | 173 ++++++++++- .../net/ethernet/mucse/rnpgbe/rnpgbe_lib.h | 3 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_main.c | 20 +- .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c | 169 ++++++++++- .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx.h | 4 + .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c | 278 +++++++++++++++++- .../net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h | 59 ++++ 10 files changed, 776 insertions(+), 16 deletions(-) diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h b/drivers/net/ether= net/mucse/rnpgbe/rnpgbe.h index bd7491671e1e..10aff417b647 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe.h @@ -5,6 +5,7 @@ #define _RNPGBE_H =20 #include +#include #include #include #include @@ -33,11 +34,10 @@ struct mucse_mbx_info { u32 fwpf_ctrl_base; }; =20 -/* Enum for firmware notification modes, - * more modes (e.g., portup, link_report) will be added in future - **/ enum { mucse_fw_powerup, + mucse_fw_portup, + mucse_fw_link_report_en, }; =20 struct mucse_hw { @@ -46,8 +46,11 @@ struct mucse_hw { struct pci_dev *pdev; struct mucse_mbx_info mbx; int port; + int speed; + bool link; u16 cycles_per_us; u8 pfvfnum; + u8 duplex; }; =20 struct rnpgbe_tx_desc { @@ -222,16 +225,22 @@ struct mucse { int num_rx_queues; char mbx_name[32]; unsigned long state; + atomic_t link_pending; + atomic_t mbx_irq_seq; struct work_struct mbx_work; + struct delayed_work serv_task; + spinlock_t link_lock; /* spinlock for link update */ }; =20 int rnpgbe_get_permanent_mac(struct mucse_hw *hw, u8 *perm_addr); int rnpgbe_reset_hw(struct mucse_hw *hw); +int rnpgbe_setup_default_link(struct mucse_hw *hw); int rnpgbe_send_notify(struct mucse_hw *hw, bool enable, int mode); int rnpgbe_init_hw(struct mucse_hw *hw, int board_type); void rnpgbe_set_rx_mode(struct net_device *netdev); +void rnpgbe_set_link(struct mucse_hw *hw, bool linkup); =20 /* Device IDs */ #define PCI_VENDOR_ID_MUCSE 0x8848 diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_chip.c index 4e2835b71f32..b56263de4a36 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_chip.c @@ -4,12 +4,19 @@ #include #include #include +#include =20 #include "rnpgbe.h" #include "rnpgbe_hw.h" #include "rnpgbe_mbx.h" #include "rnpgbe_mbx_fw.h" =20 +#define RNPGBE_DEFAULT_ADV (RNPGBE_FW_ADV_10_FULL | \ + RNPGBE_FW_ADV_100_FULL | \ + RNPGBE_FW_ADV_1000_FULL | \ + RNPGBE_FW_ADV_10_HALF | \ + RNPGBE_FW_ADV_100_HALF) + /** * rnpgbe_get_permanent_mac - Get permanent mac * @hw: hw information structure @@ -53,6 +60,18 @@ int rnpgbe_reset_hw(struct mucse_hw *hw) return mucse_mbx_reset_hw(hw); } =20 +/** + * rnpgbe_setup_default_link - Configure the default link settings + * @hw: hardware information structure + * + * Return: 0 on success, negative errno on failure + **/ +int rnpgbe_setup_default_link(struct mucse_hw *hw) +{ + return mucse_mbx_set_link(hw, RNPGBE_DEFAULT_ADV, true, 0, 0, + ETH_TP_MDI_AUTO); +} + /** * rnpgbe_send_notify - Echo fw status * @hw: hw information structure @@ -66,11 +85,17 @@ int rnpgbe_send_notify(struct mucse_hw *hw, int mode) { int err; - /* Keep switch struct to support more modes in the future */ + switch (mode) { case mucse_fw_powerup: err =3D mucse_mbx_powerup(hw, enable); break; + case mucse_fw_portup: + err =3D mucse_mbx_phyup(hw, enable); + break; + case mucse_fw_link_report_en: + err =3D mucse_mbx_link_report(hw, enable); + break; default: err =3D -EINVAL; } @@ -224,3 +249,30 @@ void rnpgbe_set_rx_mode(struct net_device *netdev) mucse_hw_wr32(hw, RNPGBE_RX_MCAST_CTRL, mcast_ctrl); mucse_hw_wr32(hw, RNPGBE_RX_FILTER_CTRL, filter_ctrl); } + +/** + * rnpgbe_set_link - Set the hardware link state + * @hw: hw information structure + * @linkup: link on or not + * + * rnpgbe_set_link setup link status + * + **/ +void rnpgbe_set_link(struct mucse_hw *hw, bool linkup) +{ + u32 value =3D mucse_hw_rd32(hw, GMAC_CONTROL); + + /* The chip-level filter is programmed by ndo_set_rx_mode(). Keep the + * GMAC in receive-all mode so it does not discard frames accepted by + * that filter. + */ + if (linkup) { + mucse_hw_wr32(hw, GMAC_FRAME_FILTER, GMAC_RX_ALL); + value |=3D GMAC_CONTROL_RE; + mucse_hw_wr32(hw, GMAC_CONTROL, value); + } else { + value &=3D ~GMAC_CONTROL_RE; + mucse_hw_wr32(hw, GMAC_CONTROL, value); + mucse_hw_wr32(hw, GMAC_FRAME_FILTER, 0); + } +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h b/drivers/net/et= hernet/mucse/rnpgbe/rnpgbe_hw.h index feb84704d602..5659e91f3497 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_hw.h @@ -21,6 +21,17 @@ #define RNPGBE_DMA_TX_STATUS GENMASK_U32(23, 22) #define RNPGBE_DMA_RX_STATUS GENMASK_U32(21, 20) #define RX_AXI_RW_EN 0x03 +/* RNPGBE_LINK_ST is a driver-owned link-state snapshot consumed by firmwa= re. + * Set the driver-state marker; all other driver status fields start clear. + */ +#define M_DEFAULT_ST 0xa0000000 +/* Firmware never reports 0xf in the speed field. */ +#define M_INVALID_ST (M_DEFAULT_ST | GENMASK_U32(11, 8)) +/* Driver-reported fields: 25:24 pause, 11:8 speed, 6 LLDP, 4 duplex, + * and 0 link up/down. M_DEFAULT_ST resets these fields so firmware reports + * the current hardware state again. + */ +#define RNPGBE_LINK_ST 0x000c #define RNPGBE_DMA_AXI_EN 0x0010 #define RNPGBE_TX_MIN_PKT_LEN 33 =20 @@ -39,5 +50,11 @@ /* The final two of the 32 hardware RAR entries are reserved for NCSI. */ #define RNPGBE_RX_RAR_ENTRIES 30 #define RNPGBE_RX_MCAST_HASH_ENTRIES 128 + +#define MUCSE_GMAC_OFF(_n) (0x20000 + (_n)) +#define GMAC_CONTROL_RE 0x00000004 +#define GMAC_CONTROL MUCSE_GMAC_OFF(0) +#define GMAC_RX_ALL (BIT(31) | BIT(0)) +#define GMAC_FRAME_FILTER MUCSE_GMAC_OFF(0x4) #define RNPGBE_MAX_QUEUES 8 #endif /* _RNPGBE_HW_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.c index 9afd10273a62..2a5349f03a02 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.c @@ -16,8 +16,14 @@ static void rnpgbe_mbx_work(struct work_struct *work) { struct mucse *mucse =3D container_of(work, struct mucse, mbx_work); + int irq_seq; =20 + irq_seq =3D atomic_read(&mucse->mbx_irq_seq); mucse_fw_irq_handler(&mucse->hw); + + /* A mailbox interrupt while this work runs needs another invocation. */ + if (irq_seq !=3D atomic_read(&mucse->mbx_irq_seq)) + queue_work(system_percpu_wq, &mucse->mbx_work); } =20 /** @@ -31,6 +37,7 @@ static irqreturn_t rnpgbe_msix_other(int irq, void *data) { struct mucse *mucse =3D (struct mucse *)data; =20 + atomic_inc(&mucse->mbx_irq_seq); queue_work(system_percpu_wq, &mucse->mbx_work); =20 return IRQ_HANDLED; @@ -664,6 +671,7 @@ int rnpgbe_request_mbx_irq(struct mucse *mucse) snprintf(mucse->mbx_name, sizeof(mucse->mbx_name), "rnpgbe-mbx:%s", pci_name(pdev)); INIT_WORK(&mucse->mbx_work, rnpgbe_mbx_work); + atomic_set(&mucse->mbx_irq_seq, 0); =20 err =3D request_irq(pci_irq_vector(pdev, 0), rnpgbe_msix_other, 0, mucse->mbx_name, mucse); @@ -1399,7 +1407,19 @@ static void rnpgbe_cancel_rx_retry_timers(struct muc= se *mucse) void rnpgbe_down(struct mucse *mucse) { struct net_device *netdev =3D mucse->netdev; + struct mucse_hw *hw =3D &mucse->hw; + unsigned long flags; + int err; =20 + spin_lock_irqsave(&mucse->link_lock, flags); + WRITE_ONCE(hw->link, false); + WRITE_ONCE(hw->speed, 0); + WRITE_ONCE(hw->duplex, 0); + atomic_set(&mucse->link_pending, 0); + spin_unlock_irqrestore(&mucse->link_lock, flags); + cancel_delayed_work_sync(&mucse->serv_task); + + netif_carrier_off(netdev); rnpgbe_irq_disable(mucse); rnpgbe_napi_disable_all(mucse); synchronize_net(); @@ -1408,22 +1428,77 @@ void rnpgbe_down(struct mucse *mucse) rnpgbe_stop_all_rx_rings(mucse); rnpgbe_cancel_rx_retry_timers(mucse); rnpgbe_clean_all_tx_rings(mucse); + + err =3D rnpgbe_send_notify(hw, false, mucse_fw_link_report_en); + if (err) { + dev_warn(&hw->pdev->dev, "Send link report to hw failed %d\n", + err); + dev_warn(&hw->pdev->dev, "Fw will still report link event\n"); + } + + err =3D rnpgbe_send_notify(hw, false, mucse_fw_portup); + if (err) { + dev_warn(&hw->pdev->dev, "Send port down to hw failed %d\n", + err); + dev_warn(&hw->pdev->dev, "Port is not truly down\n"); + } + rnpgbe_set_link(hw, false); rnpgbe_clean_all_rx_rings(mucse); } =20 /** * rnpgbe_up_complete - Final step for port up * @mucse: pointer to private structure + * + * Return: 0 on success, negative errno if firmware setup fails **/ -void rnpgbe_up_complete(struct mucse *mucse) +int rnpgbe_up_complete(struct mucse *mucse) { struct net_device *netdev =3D mucse->netdev; + struct mucse_hw *hw =3D &mucse->hw; + unsigned long flags; + int err; =20 rnpgbe_configure_msix(mucse); rnpgbe_napi_enable_all(mucse); + /* Clear stale link state and tell firmware the driver hasn't + * seen any link yet. Firmware will re-assert LINK_CHANGE_EVT + * when link_report_en is sent below since RNPGBE_LINK_ST + * no longer matches the actual (possibly up) link state. + */ + spin_lock_irqsave(&mucse->link_lock, flags); + WRITE_ONCE(hw->link, false); + WRITE_ONCE(hw->speed, 0); + WRITE_ONCE(hw->duplex, 0); + atomic_set(&mucse->link_pending, 0); + /* echo fw driver now in default state */ + mucse_hw_wr32(hw, RNPGBE_LINK_ST, M_DEFAULT_ST); + /* Keep the cached state and default snapshot atomic with mailbox + * link-event handling. + */ + spin_unlock_irqrestore(&mucse->link_lock, flags); rnpgbe_schedule_rx_retry(mucse); rnpgbe_irq_enable(mucse); netif_tx_start_all_queues(netdev); + err =3D rnpgbe_send_notify(hw, true, mucse_fw_portup); + if (err) { + dev_err(&hw->pdev->dev, + "Failed to notify firmware that port is up: %d\n", err); + return err; + } + /* Firmware checks RNPGBE_LINK_ST (driver's last-received link state) + * and only asserts LINK_CHANGE_EVT when it differs from the actual link + * state AND link_report_en is true. + */ + err =3D rnpgbe_send_notify(hw, true, mucse_fw_link_report_en); + if (err) { + dev_err(&hw->pdev->dev, + "Failed to enable firmware link reporting: %d\n", + err); + return err; + } + + return 0; } =20 /** @@ -2048,3 +2123,99 @@ int rnpgbe_configure_rx(struct mucse *mucse) =20 return 0; } + +/** + * rnpgbe_process_link_event - Consume a pending link event + * @mucse: pointer to the device private structure + * @link: link status snapshot + * @speed: link speed snapshot + * @duplex: link duplex snapshot + * + * Return: true if a link event was consumed, false otherwise + **/ +static bool rnpgbe_process_link_event(struct mucse *mucse, bool *link, + int *speed, u8 *duplex) +{ + struct mucse_hw *hw =3D &mucse->hw; + unsigned long flags; + + if (!atomic_xchg(&mucse->link_pending, 0)) + return false; + + spin_lock_irqsave(&mucse->link_lock, flags); + *link =3D hw->link; + *speed =3D hw->speed; + *duplex =3D hw->duplex; + spin_unlock_irqrestore(&mucse->link_lock, flags); + + return true; +} + +/** + * rnpgbe_link_is_up - Update netif_carrier status and + * print link up message + * @mucse: pointer to the device private structure + * @speed: link speed snapshot + * @duplex: link duplex snapshot + **/ +static void rnpgbe_link_is_up(struct mucse *mucse, int speed, u8 duplex) +{ + struct net_device *netdev =3D mucse->netdev; + + /* Only continue if link was previously down */ + if (netif_carrier_ok(netdev)) + return; + + netdev_info(netdev, "NIC Link is Up %d Mbps, %s Duplex\n", + speed, duplex ? "Full" : "Half"); + netif_carrier_on(netdev); +} + +/** + * rnpgbe_link_is_down - Update netif_carrier status and + * print link down message + * @mucse: pointer to the private structure + **/ +static void rnpgbe_link_is_down(struct mucse *mucse) +{ + struct net_device *netdev =3D mucse->netdev; + + /* Only continue if link was up previously */ + if (!netif_carrier_ok(netdev)) + return; + netdev_info(netdev, "NIC Link is Down\n"); + netif_carrier_off(netdev); +} + +/** + * rnpgbe_process_link_subtask - Process a link-state update + * @mucse: pointer to the device private structure + **/ +static void rnpgbe_process_link_subtask(struct mucse *mucse) +{ + bool link; + int speed; + u8 duplex; + + if (!rnpgbe_process_link_event(mucse, &link, &speed, &duplex)) + return; + + if (link) + rnpgbe_link_is_up(mucse, speed, duplex); + else + rnpgbe_link_is_down(mucse); +} + +/** + * rnpgbe_service_task - Manages and runs subtasks + * @work: pointer to work_struct containing our data + **/ +void rnpgbe_service_task(struct work_struct *work) +{ + struct mucse *mucse =3D container_of(work, struct mucse, serv_task.work); + + if (!netif_running(mucse->netdev)) + return; + + rnpgbe_process_link_subtask(mucse); +} diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_lib.h index 76b5b0c0ae4a..08b7d22f98ee 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_lib.h @@ -68,7 +68,7 @@ int rnpgbe_request_irq(struct mucse *mucse); void rnpgbe_free_irq(struct mucse *mucse); void rnpgbe_irq_disable(struct mucse *mucse); void rnpgbe_down(struct mucse *mucse); -void rnpgbe_up_complete(struct mucse *mucse); +int rnpgbe_up_complete(struct mucse *mucse); int rnpgbe_configure_tx(struct mucse *mucse); int rnpgbe_configure_rx(struct mucse *mucse); void rnpgbe_clean_all_tx_rings(struct mucse *mucse); @@ -80,4 +80,5 @@ void rnpgbe_get_stats64(struct net_device *netdev, struct rtnl_link_stats64 *stats); int rnpgbe_setup_all_rx_resources(struct mucse *mucse); void rnpgbe_free_all_rx_resources(struct mucse *mucse); +void rnpgbe_service_task(struct work_struct *work); #endif diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c b/drivers/net/= ethernet/mucse/rnpgbe/rnpgbe_main.c index 35eb858eb2c1..9ba53bb3629b 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_main.c @@ -67,6 +67,7 @@ static int rnpgbe_open(struct net_device *netdev) if (test_bit(__MUCSE_AXI_FAULT, &mucse->state)) return -EIO; =20 + netif_carrier_off(netdev); err =3D rnpgbe_request_irq(mucse); if (err) return err; @@ -86,9 +87,16 @@ static int rnpgbe_open(struct net_device *netdev) err =3D rnpgbe_configure(mucse); if (err) goto err_free_rx; - rnpgbe_up_complete(mucse); + err =3D rnpgbe_up_complete(mucse); + if (err) + goto err_down; =20 return 0; +err_down: + rnpgbe_down(mucse); + rnpgbe_free_all_rx_resources(mucse); + rnpgbe_free_all_tx_resources(mucse); + goto err_free_irqs; err_free_rx: rnpgbe_free_all_rx_resources(mucse); err_free_tx: @@ -237,6 +245,11 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, dev_err(&pdev->dev, "Hw reset failed %d\n", err); goto err_powerdown; } + err =3D rnpgbe_setup_default_link(hw); + if (err) { + dev_err(&pdev->dev, "Setup link failed %d\n", err); + goto err_powerdown; + } =20 err =3D rnpgbe_get_permanent_mac(hw, perm_addr); if (!err) { @@ -249,6 +262,10 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, goto err_powerdown; } =20 + INIT_DELAYED_WORK(&mucse->serv_task, rnpgbe_service_task); + spin_lock_init(&mucse->link_lock); + atomic_set(&mucse->link_pending, 0); + err =3D rnpgbe_init_interrupt_scheme(mucse); if (err) { dev_err(&pdev->dev, "init interrupt failed %d\n", err); @@ -273,6 +290,7 @@ static int rnpgbe_add_adapter(struct pci_dev *pdev, netdev->hw_features |=3D NETIF_F_HIGHDMA; } =20 + netif_carrier_off(netdev); err =3D register_netdev(netdev); if (err) goto err_remove_mbx; diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_mbx.c index 2310f01df1f5..74f42be975c2 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.c @@ -197,7 +197,8 @@ static int mucse_read_mbx_pf(struct mucse_hw *hw, __le3= 2 *msg, u16 size) * mucse_check_for_msg_pf - Check to see if the fw has sent mail * @hw: pointer to the HW structure * - * Return: 0 if the fw has set the Status bit or else -EIO + * Return: 0 if firmware has posted a new message, -ENOMSG if there is no + * new message, or -EIO if the mailbox has been reset **/ static int mucse_check_for_msg_pf(struct mucse_hw *hw) { @@ -205,12 +206,11 @@ static int mucse_check_for_msg_pf(struct mucse_hw *hw) u16 fw_req; =20 fw_req =3D mucse_mbx_get_fwreq(mbx); - /* chip's register is reset to 0 when rc send reset - * mbx command. Return -EIO if in this state, others - * fw =3D=3D hw->mbx.fw_req means no new msg. - **/ - if (fw_req =3D=3D 0 || fw_req =3D=3D hw->mbx.fw_req) + /* The request counter is reset to zero by a mailbox reset. */ + if (!fw_req) return -EIO; + if (fw_req =3D=3D hw->mbx.fw_req) + return -ENOMSG; =20 return 0; } @@ -252,6 +252,68 @@ int mucse_poll_and_read_mbx(struct mucse_hw *hw, __le3= 2 *msg, u16 size) return mucse_read_mbx_pf(hw, msg, size); } =20 +/** + * mucse_mbx_event_begin - Lock and read a pending firmware event + * @hw: pointer to the HW structure + * @msg: the message buffer + * @size: length of buffer + * + * On success the hardware mailbox remains locked. The caller must process + * the event and call mucse_mbx_event_end() to acknowledge it and release + * the hardware mailbox lock. + * + * Firmware releases the hardware mailbox lock before raising the event + * interrupt and does not access the shared window again until the PF + * acknowledges the event. Therefore, failure to acquire the lock indicates + * a mailbox fault rather than transient contention. + * + * Return: 0 on success, -ENOMSG if no event is pending, or another negati= ve + * errno on failure + **/ +int mucse_mbx_event_begin(struct mucse_hw *hw, __le32 *msg, u16 size) +{ + const int size_in_words =3D size / sizeof(__le32); + struct mucse_mbx_info *mbx =3D &hw->mbx; + int off =3D MUCSE_MBX_FWPF_SHM; + int err; + + err =3D mucse_check_for_msg_pf(hw); + if (err) + return err; + + err =3D mucse_obtain_mbx_lock_pf(hw); + if (err) + return err; + + /* Check again after taking ownership of the shared mailbox. */ + err =3D mucse_check_for_msg_pf(hw); + if (err) { + mucse_release_mbx_lock_pf(hw, false); + return err; + } + + for (int i =3D 0; i < size_in_words; i++) + msg[i] =3D cpu_to_le32(mbx_data_rd32(mbx, off + 4 * i)); + + return 0; +} + +/** + * mucse_mbx_event_end - Acknowledge a firmware event and unlock the mailb= ox + * @hw: pointer to the HW structure + * + * Pair with a successful mucse_mbx_event_begin(). + **/ +void mucse_mbx_event_end(struct mucse_hw *hw) +{ + struct mucse_mbx_info *mbx =3D &hw->mbx; + + mbx_data_wr32(mbx, MUCSE_MBX_FWPF_SHM, 0); + hw->mbx.fw_req =3D mucse_mbx_get_fwreq(mbx); + mucse_mbx_inc_pf_ack(hw); + mucse_release_mbx_lock_pf(hw, false); +} + /** * mucse_mbx_get_fwack - Read fw ack from reg * @mbx: pointer to the MBX structure @@ -286,6 +348,36 @@ static void mucse_mbx_inc_pf_req(struct mucse_hw *hw) mbx_data_wr32(mbx, MUCSE_MBX_PF2FW_CNT, val); } =20 +/** + * mucse_ack_pending_event - Acknowledge a level-like firmware event + * @hw: pointer to the HW structure + * + * The caller must own the hardware mailbox lock. + * + * Return: 0 on success, negative errno on failure + **/ +static int mucse_ack_pending_event(struct mucse_hw *hw) +{ + struct mucse_mbx_info *mbx =3D &hw->mbx; + u16 fw_req; + + fw_req =3D mucse_mbx_get_fwreq(mbx); + if (fw_req =3D=3D hw->mbx.fw_req) + return 0; + if (!fw_req) + return -EIO; + + /* Firmware only posts LINK_CHANGE_EVT asynchronously. It is level-like, + * so leaving the link snapshot unchanged lets firmware report the + * current state again if reporting stays enabled. + */ + mbx_data_wr32(mbx, MUCSE_MBX_FWPF_SHM, 0); + hw->mbx.fw_req =3D fw_req; + mucse_mbx_inc_pf_ack(hw); + + return 0; +} + /** * mucse_write_mbx_pf - Place a message in the mailbox * @hw: pointer to the HW structure @@ -320,6 +412,44 @@ static int mucse_write_mbx_pf(struct mucse_hw *hw, con= st __le32 *msg, u16 size) return 0; } =20 +/** + * mucse_write_mbx_pf_coalesce_event - Write around a level-like event + * @hw: pointer to the HW structure + * @msg: the message buffer + * @size: length of buffer + * + * Return: 0 on success, negative errno on failure + **/ +static int mucse_write_mbx_pf_coalesce_event(struct mucse_hw *hw, + const __le32 *msg, u16 size) +{ + const int size_in_words =3D size / sizeof(__le32); + struct mucse_mbx_info *mbx =3D &hw->mbx; + int err; + + err =3D mucse_obtain_mbx_lock_pf(hw); + if (err) + return err; + + err =3D mucse_ack_pending_event(hw); + if (err) + goto release; + + for (int i =3D 0; i < size_in_words; i++) + mbx_data_wr32(mbx, MUCSE_MBX_FWPF_SHM + i * 4, + le32_to_cpu(msg[i])); + + hw->mbx.fw_ack =3D mucse_mbx_get_fwack(mbx); + mucse_mbx_inc_pf_req(hw); + mucse_release_mbx_lock_pf(hw, true); + + return 0; + +release: + mucse_release_mbx_lock_pf(hw, false); + return err; +} + /** * mucse_check_for_ack_pf - Check to see if the fw has ACKed * @hw: pointer to the HW structure @@ -381,6 +511,33 @@ int mucse_write_and_wait_ack_mbx(struct mucse_hw *hw, = const __le32 *msg, return mucse_poll_for_ack(hw); } =20 +/** + * mucse_write_mbx_coalesce_event - Send a command while coalescing an eve= nt + * @hw: pointer to the HW structure + * @msg: the message buffer + * @size: length of buffer + * + * A pending link event is acknowledged without updating its snapshot, then + * the PF command is written while the hardware mailbox remains locked. If + * event reporting remains enabled, firmware reports the event again becau= se + * its snapshot still differs from the hardware state. This function waits + * for firmware to acknowledge the PF command before returning. + * + * Return: 0 after firmware acknowledges the command, negative errno on + * failure + **/ +int mucse_write_mbx_coalesce_event(struct mucse_hw *hw, + const __le32 *msg, u16 size) +{ + int err; + + err =3D mucse_write_mbx_pf_coalesce_event(hw, msg, size); + if (err) + return err; + + return mucse_poll_for_ack(hw); +} + /** * mucse_mbx_reset - Reset mbx info, sync info from regs * @hw: pointer to the HW structure diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.h b/drivers/net/e= thernet/mucse/rnpgbe/rnpgbe_mbx.h index 75b88b18b04d..50597dd9dde9 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx.h @@ -16,6 +16,10 @@ =20 int mucse_write_and_wait_ack_mbx(struct mucse_hw *hw, const __le32 *msg, u16 size); +int mucse_write_mbx_coalesce_event(struct mucse_hw *hw, + const __le32 *msg, u16 size); void mucse_init_mbx_params_pf(struct mucse_hw *hw); int mucse_poll_and_read_mbx(struct mucse_hw *hw, __le32 *msg, u16 size); +int mucse_mbx_event_begin(struct mucse_hw *hw, __le32 *msg, u16 size); +void mucse_mbx_event_end(struct mucse_hw *hw); #endif /* _RNPGBE_MBX_H */ diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c b/drivers/ne= t/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c index e28f3193aa85..6415b6430efc 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.c @@ -3,6 +3,7 @@ =20 #include #include +#include =20 #include "rnpgbe.h" #include "rnpgbe_mbx.h" @@ -199,12 +200,283 @@ int mucse_mbx_get_macaddr(struct mucse_hw *hw, int p= fvfnum, } =20 /** - * mucse_fw_irq_handler - Deferred firmware mailbox notification hook + * mucse_mbx_set_link - Configure firmware link settings * @hw: pointer to the HW structure + * @advertising: firmware link-mode advertisement mask + * @autoneg: whether to enable autonegotiation + * @speed: forced link speed when autonegotiation is disabled + * @duplex: forced duplex mode when autonegotiation is disabled + * @mdix_ctrl: MDI/MDI-X control mode * - * No asynchronous firmware events are enabled until link status support is - * added. + * Firmware ignores @speed, @duplex when @autoneg is true. + * + * Return: 0 on success, negative errno on failure + **/ +int mucse_mbx_set_link(struct mucse_hw *hw, u32 advertising, bool autoneg, + u32 speed, u32 duplex, u32 mdix_ctrl) +{ + union mbx_fw_cmd_req_u req =3D { + .r =3D { + .datalen =3D cpu_to_le16(sizeof(req.r.phy_link_set) + + MUCSE_MBX_REQ_HDR_LEN), + .opcode =3D cpu_to_le16(PHY_LINK_SET), + .phy_link_set =3D { + .adv_speed_mask =3D cpu_to_le32(advertising), + .autoneg =3D cpu_to_le32(autoneg), + .speed =3D cpu_to_le32(speed), + .duplex =3D cpu_to_le32(duplex), + .nr_lane =3D cpu_to_le32(hw->port), + .tp_mdix_ctrl =3D cpu_to_le32(mdix_ctrl), + }, + }, + }; + int len, err; + + len =3D le16_to_cpu(req.r.datalen); + mutex_lock(&hw->mbx.lock); + err =3D mucse_write_and_wait_ack_mbx(hw, req.dwords, len); + mutex_unlock(&hw->mbx.lock); + + return err; +} + +/** + * mucse_mbx_phyup - Request that firmware bring the PHY up or down + * @hw: pointer to the HW structure + * @is_phyup: true for up, false for down + * + * mucse_mbx_phyup echo fw to change phy status + * + * Return: 0 on success, negative errno on failure + **/ +int mucse_mbx_phyup(struct mucse_hw *hw, bool is_phyup) +{ + union mbx_fw_cmd_req_u req =3D { + .r =3D { + .datalen =3D cpu_to_le16(sizeof(req.r.phy_status) + + MUCSE_MBX_REQ_HDR_LEN), + .opcode =3D cpu_to_le16(SET_PHY_UP), + .phy_status =3D { + .port_mask =3D cpu_to_le32(BIT(hw->port)), + .status =3D cpu_to_le32(is_phyup ? 1 : 0), + }, + }, + }; + int len, err; + + len =3D le16_to_cpu(req.r.datalen); + mutex_lock(&hw->mbx.lock); + err =3D mucse_write_mbx_coalesce_event(hw, req.dwords, len); + mutex_unlock(&hw->mbx.lock); + + return err; +} + +/** + * mucse_mbx_link_report - Configure firmware link-change event reporting + * @hw: pointer to the HW structure + * @is_report: true for report, false for no + * + * mucse_mbx_link_report echo fw to change event report state + * + * Return: 0 on success, negative errno on failure + **/ +int mucse_mbx_link_report(struct mucse_hw *hw, bool is_report) +{ + union mbx_fw_cmd_req_u req =3D { + .r =3D { + .datalen =3D cpu_to_le16(sizeof(req.r.report_status) + + MUCSE_MBX_REQ_HDR_LEN), + .opcode =3D cpu_to_le16(LINK_REPORT_EN), + .report_status =3D { + .port_mask =3D cpu_to_le16(BIT(hw->port)), + .status =3D cpu_to_le16(is_report ? 1 : 0), + }, + }, + }; + int len, err; + + len =3D le16_to_cpu(req.r.datalen); + mutex_lock(&hw->mbx.lock); + err =3D mucse_write_mbx_coalesce_event(hw, req.dwords, len); + mutex_unlock(&hw->mbx.lock); + + return err; +} + +static bool mucse_link_speed_valid(const struct mbx_fw_cmd_req *req) +{ + switch (le16_to_cpu(req->link_stat.st.speed)) { + case 10: + case 100: + case 1000: + return true; + default: + return false; + } +} + +static bool mucse_link_is_up(const struct mucse_hw *hw, + const struct mbx_fw_cmd_req *req) +{ + return le16_to_cpu(req->link_stat.port_status) & BIT(hw->port); +} + +/** + * mucse_update_link_status_reg - update driver speed inf to reg + * @hw: pointer to the HW structure + * @req: pointer to req data + * + * Update the driver's link-state snapshot exported to firmware. Firmware + * sends a new event when this snapshot differs from the hardware state. + * The default snapshot clears the driver-reported fields; + * a valid event then repopulates them, including the LLDP status in bit 6. + * + **/ +static void mucse_update_link_status_reg(struct mucse_hw *hw, + struct mbx_fw_cmd_req *req) +{ + u16 status =3D le16_to_cpu(req->link_stat.st.status); + u16 speed =3D le16_to_cpu(req->link_stat.st.speed); + u32 value; + + value =3D M_DEFAULT_ST; + + if (mucse_link_is_up(hw, req)) { + value |=3D BIT(0); + switch (speed) { + case 10: + value |=3D (mucse_speed_10 << 8); + break; + case 100: + value |=3D (mucse_speed_100 << 8); + break; + case 1000: + value |=3D (mucse_speed_1000 << 8); + break; + default: + break; + } + + value |=3D FIELD_PREP(BIT(4), + !!(req->link_stat.st.flags & DUPLEX_BIT)); + value |=3D FIELD_PREP(GENMASK_U32(25, 24), + status & GENMASK(1, 0)); + } else { + value &=3D ~BIT(0); + } + + if (status & ST_STATUS_LLDP_STATUS_MASK) + value |=3D BIT(6); + else + value &=3D ~BIT(6); + + mucse_hw_wr32(hw, RNPGBE_LINK_ST, value); +} + +/** + * mucse_mbx_fw_req_handler - Handle fw req + * @hw: pointer to the HW structure + * @req: pointer to req data + * + * mucse_mbx_fw_req_handler handler fw req, such as a link event req. + **/ +static void mucse_mbx_fw_req_handler(struct mucse_hw *hw, + struct mbx_fw_cmd_req *req) +{ + struct mucse *mucse =3D container_of(hw, struct mucse, hw); + u32 magic =3D le32_to_cpu(req->link_stat.port_magic); + unsigned long flags; + + if (le16_to_cpu(req->opcode) =3D=3D LINK_CHANGE_EVT) { + u16 speed =3D le16_to_cpu(req->link_stat.st.speed); + + spin_lock_irqsave(&mucse->link_lock, flags); + if (magic !=3D ST_VALID_MAGIC) { + /* Do not change the cached state for an invalid event. + * Use an invalid speed encoding to make firmware report + * again. + */ + mucse_hw_wr32(hw, RNPGBE_LINK_ST, M_INVALID_ST); + spin_unlock_irqrestore(&mucse->link_lock, flags); + return; + } + + if (mucse_link_is_up(hw, req) && + !mucse_link_speed_valid(req)) { + /* Do not acknowledge an invalid speed as valid. + * Keep the snapshot mismatched so firmware retries it. + * Firmware limits link reports to one per 500 ms. + */ + mucse_hw_wr32(hw, RNPGBE_LINK_ST, M_DEFAULT_ST); + spin_unlock_irqrestore(&mucse->link_lock, flags); + dev_warn_ratelimited(&hw->pdev->dev, + "unsupported link speed %u Mbps\n", + speed); + return; + } + + if (!netif_running(mucse->netdev)) { + rnpgbe_set_link(hw, false); + mucse_update_link_status_reg(hw, req); + spin_unlock_irqrestore(&mucse->link_lock, flags); + return; + } + + if (mucse_link_is_up(hw, req)) + WRITE_ONCE(hw->link, true); + else + WRITE_ONCE(hw->link, false); + + WRITE_ONCE(hw->speed, le16_to_cpu(req->link_stat.st.speed)); + WRITE_ONCE(hw->duplex, req->link_stat.st.flags & DUPLEX_BIT); + rnpgbe_set_link(hw, mucse_link_is_up(hw, req)); + /* update regs to notify link info is received */ + mucse_update_link_status_reg(hw, req); + atomic_set_release(&mucse->link_pending, 1); + /* Run link handling immediately. */ + mod_delayed_work(system_percpu_wq, &mucse->serv_task, 0); + spin_unlock_irqrestore(&mucse->link_lock, flags); + } +} + +/** + * mucse_fw_handle_event - Handle one pending firmware event + * @hw: pointer to the hardware structure + **/ +static void mucse_fw_handle_event(struct mucse_hw *hw) +{ + union mbx_fw_cmd_req_u msg =3D {}; + int err; + + /* try to check and read fw req */ + mutex_lock(&hw->mbx.lock); + err =3D mucse_mbx_event_begin(hw, msg.dwords, sizeof(msg)); + /* A PF command may have coalesced the event before stale mailbox work + * runs, in which case there is no event left to handle. + */ + if (err) { + mutex_unlock(&hw->mbx.lock); + if (err !=3D -ENOMSG) + dev_warn_ratelimited(&hw->pdev->dev, + "failed to read firmware event: %d\n", + err); + return; + } + + mucse_mbx_fw_req_handler(hw, &msg.r); + mucse_mbx_event_end(hw); + mutex_unlock(&hw->mbx.lock); +} + +/** + * mucse_fw_irq_handler - Handle one pending firmware mailbox event + * @hw: pointer to the HW structure + * + * Process at most one event per work-item invocation. The caller requeues + * mailbox work when a dedicated mailbox interrupt arrives during handling. **/ void mucse_fw_irq_handler(struct mucse_hw *hw) { + mucse_fw_handle_event(hw); } diff --git a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h b/drivers/ne= t/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h index a6bf5de55aa3..93ce0e796442 100644 --- a/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h +++ b/drivers/net/ethernet/mucse/rnpgbe/rnpgbe_mbx_fw.h @@ -14,9 +14,19 @@ enum MUCSE_FW_CMD { GET_HW_INFO =3D 0x0601, GET_MAC_ADDRESS =3D 0x0602, RESET_HW =3D 0x0603, + LINK_CHANGE_EVT =3D 0x0608, + LINK_REPORT_EN =3D 0x0613, + PHY_LINK_SET =3D 0x0630, + SET_PHY_UP =3D 0x0800, POWER_UP =3D 0x0803, }; =20 +#define RNPGBE_FW_ADV_10_FULL BIT(2) +#define RNPGBE_FW_ADV_100_FULL BIT(3) +#define RNPGBE_FW_ADV_1000_FULL BIT(4) +#define RNPGBE_FW_ADV_10_HALF BIT(10) +#define RNPGBE_FW_ADV_100_HALF BIT(11) + struct mucse_hw_info { u8 link_stat; u8 port_mask; @@ -36,6 +46,16 @@ struct mucse_hw_info { __le32 ext_info; } __packed; =20 +#define ST_STATUS_LLDP_STATUS_MASK BIT(12) + +#define DUPLEX_BIT BIT(0) +struct st_status { + u8 phyid; + u8 flags; + __le16 speed; + __le16 status; +} __packed; + struct mbx_fw_cmd_req { __le16 flags; __le16 opcode; @@ -55,6 +75,32 @@ struct mbx_fw_cmd_req { __le32 port_mask; __le32 pfvf_num; } get_mac_addr; + struct { + __le32 port_mask; + __le32 status; + } phy_status; + struct { + __le32 adv_speed_mask; + __le32 autoneg; + __le32 speed; + __le32 duplex; + __le32 nr_lane; + __le32 tp_mdix_ctrl; + } phy_link_set; + struct { + /* LINK_REPORT_EN uses its own 16-bit payload layout: + * status precedes port_mask. + */ + __le16 status; + __le16 port_mask; + } report_status; + struct { + __le16 changed_lanes; + __le16 port_status; + __le32 port_magic; +#define ST_VALID_MAGIC 0xa4a6a8a9 + struct st_status st; + } link_stat; }; } __packed; =20 @@ -94,10 +140,23 @@ union mbx_fw_cmd_reply_u { __le32 dwords[sizeof(struct mbx_fw_cmd_reply) / sizeof(__le32)]; }; =20 +/* Firmware encoding for RNPGBE_LINK_ST[11:8]. Valid link events report + * only these three speeds. + */ +enum mucse_speed { + mucse_speed_10 =3D 0, + mucse_speed_100 =3D 1, + mucse_speed_1000 =3D 2, +}; + int mucse_mbx_sync_fw(struct mucse_hw *hw); int mucse_mbx_powerup(struct mucse_hw *hw, bool is_powerup); int mucse_mbx_reset_hw(struct mucse_hw *hw); int mucse_mbx_get_macaddr(struct mucse_hw *hw, int pfvfnum, u8 *mac_addr, int port); +int mucse_mbx_set_link(struct mucse_hw *hw, u32 advertising, bool autoneg, + u32 speed, u32 duplex, u32 mdix_ctrl); +int mucse_mbx_phyup(struct mucse_hw *hw, bool is_phyup); +int mucse_mbx_link_report(struct mucse_hw *hw, bool is_report); void mucse_fw_irq_handler(struct mucse_hw *hw); #endif /* _RNPGBE_MBX_FW_H */ --=20 2.50.1