Resguardo inicial

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/Makefile

@@ -0,0 +1,26 @@
+# SPDX-License-Identifier: GPL-2.0
+#
+# Cavium Liquidio ethernet device driver
+#
+
+obj-$(CONFIG_LIQUIDIO_CORE) += liquidio-core.o
+liquidio-core-y := \
+		lio_ethtool.o		\
+		lio_core.o		\
+		request_manager.o	\
+		response_manager.o	\
+		octeon_device.o		\
+		cn66xx_device.o		\
+		cn68xx_device.o		\
+		cn23xx_pf_device.o	\
+		cn23xx_vf_device.o	\
+		octeon_mailbox.o	\
+		octeon_mem_ops.o	\
+		octeon_droq.o		\
+		octeon_nic.o
+
+obj-$(CONFIG_LIQUIDIO) += liquidio.o
+liquidio-y := lio_main.o octeon_console.o lio_vf_rep.o
+
+obj-$(CONFIG_LIQUIDIO_VF) += liquidio_vf.o
+liquidio_vf-y := lio_vf_main.o

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn23xx_pf_device.h

@@ -0,0 +1,68 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file  cn23xx_device.h
+ * \brief Host Driver: Routines that perform CN23XX specific operations.
+ */
+
+#ifndef __CN23XX_PF_DEVICE_H__
+#define __CN23XX_PF_DEVICE_H__
+
+#include "cn23xx_pf_regs.h"
+
+/* Register address and configuration for a CN23XX devices.
+ * If device specific changes need to be made then add a struct to include
+ * device specific fields as shown in the commented section
+ */
+struct octeon_cn23xx_pf {
+	/** PCI interrupt summary register */
+	u8 __iomem *intr_sum_reg64;
+
+	/** PCI interrupt enable register */
+	u8 __iomem *intr_enb_reg64;
+
+	/** The PCI interrupt mask used by interrupt handler */
+	u64 intr_mask64;
+
+	struct octeon_config *conf;
+};
+
+#define CN23XX_SLI_DEF_BP			0x40
+
+struct oct_vf_stats {
+	u64 rx_packets;
+	u64 tx_packets;
+	u64 rx_bytes;
+	u64 tx_bytes;
+	u64 broadcast;
+	u64 multicast;
+};
+
+int setup_cn23xx_octeon_pf_device(struct octeon_device *oct);
+
+u32 cn23xx_pf_get_oq_ticks(struct octeon_device *oct, u32 time_intr_in_us);
+
+int cn23xx_sriov_config(struct octeon_device *oct);
+
+int cn23xx_fw_loaded(struct octeon_device *oct);
+
+void cn23xx_tell_vf_its_macaddr_changed(struct octeon_device *oct, int vfidx,
+					u8 *mac);
+
+int cn23xx_get_vf_stats(struct octeon_device *oct, int ifidx,
+			struct oct_vf_stats *stats);
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn23xx_pf_regs.h

@@ -0,0 +1,599 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file cn23xx_regs.h
+ * \brief Host Driver: Register Address and Register Mask values for
+ * Octeon CN23XX devices.
+ */
+
+#ifndef __CN23XX_PF_REGS_H__
+#define __CN23XX_PF_REGS_H__
+
+#define     CN23XX_CONFIG_VENDOR_ID	0x00
+#define     CN23XX_CONFIG_DEVICE_ID	0x02
+
+#define     CN23XX_CONFIG_XPANSION_BAR             0x38
+
+#define     CN23XX_CONFIG_MSIX_CAP		   0x50
+#define     CN23XX_CONFIG_MSIX_LMSI		   0x54
+#define     CN23XX_CONFIG_MSIX_UMSI		   0x58
+#define     CN23XX_CONFIG_MSIX_MSIMD		   0x5C
+#define     CN23XX_CONFIG_MSIX_MSIMM		   0x60
+#define     CN23XX_CONFIG_MSIX_MSIMP		   0x64
+
+#define     CN23XX_CONFIG_PCIE_CAP                 0x70
+#define     CN23XX_CONFIG_PCIE_DEVCAP              0x74
+#define     CN23XX_CONFIG_PCIE_DEVCTL              0x78
+#define     CN23XX_CONFIG_PCIE_LINKCAP             0x7C
+#define     CN23XX_CONFIG_PCIE_LINKCTL             0x80
+#define     CN23XX_CONFIG_PCIE_SLOTCAP             0x84
+#define     CN23XX_CONFIG_PCIE_SLOTCTL             0x88
+#define     CN23XX_CONFIG_PCIE_DEVCTL2             0x98
+#define     CN23XX_CONFIG_PCIE_LINKCTL2            0xA0
+#define     CN23XX_CONFIG_PCIE_UNCORRECT_ERR_MASK  0x108
+#define     CN23XX_CONFIG_PCIE_CORRECT_ERR_STATUS  0x110
+#define     CN23XX_CONFIG_PCIE_DEVCTL_MASK         0x00040000
+
+#define     CN23XX_PCIE_SRIOV_FDL		   0x188
+#define     CN23XX_PCIE_SRIOV_FDL_BIT_POS	   0x10
+#define     CN23XX_PCIE_SRIOV_FDL_MASK		   0xFF
+
+#define     CN23XX_CONFIG_PCIE_FLTMSK              0x720
+
+#define     CN23XX_CONFIG_SRIOV_VFDEVID            0x190
+
+#define     CN23XX_CONFIG_SRIOV_BAR_START	   0x19C
+#define     CN23XX_CONFIG_SRIOV_BARX(i)		\
+		(CN23XX_CONFIG_SRIOV_BAR_START + ((i) * 4))
+#define     CN23XX_CONFIG_SRIOV_BAR_PF		   0x08
+#define     CN23XX_CONFIG_SRIOV_BAR_64BIT	   0x04
+#define     CN23XX_CONFIG_SRIOV_BAR_IO		   0x01
+
+/* ##############  BAR0 Registers ################ */
+
+#define    CN23XX_SLI_CTL_PORT_START               0x286E0
+#define    CN23XX_PORT_OFFSET                      0x10
+
+#define    CN23XX_SLI_CTL_PORT(p)                  \
+		(CN23XX_SLI_CTL_PORT_START + ((p) * CN23XX_PORT_OFFSET))
+
+/* 2 scatch registers (64-bit)  */
+#define    CN23XX_SLI_WINDOW_CTL                   0x282E0
+#define    CN23XX_SLI_SCRATCH1                     0x283C0
+#define    CN23XX_SLI_SCRATCH2                     0x283D0
+#define    CN23XX_SLI_WINDOW_CTL_DEFAULT           0x200000ULL
+
+/* 1 registers (64-bit)  - SLI_CTL_STATUS */
+#define    CN23XX_SLI_CTL_STATUS                   0x28570
+
+/* SLI Packet Input Jabber Register (64 bit register)
+ * <31:0> for Byte count for limiting sizes of packet sizes
+ * that are allowed for sli packet inbound packets.
+ * the default value is 0xFA00(=64000).
+ */
+#define    CN23XX_SLI_PKT_IN_JABBER                0x29170
+/* The input jabber is used to determine the TSO max size.
+ * Due to H/W limitation, this needs to be reduced to 60000
+ * in order to use H/W TSO and avoid the WQE malformation
+ * PKO_BUG_24989_WQE_LEN
+ */
+#define    CN23XX_DEFAULT_INPUT_JABBER             0xEA60 /*60000*/
+
+#define    CN23XX_WIN_WR_ADDR_LO                   0x20000
+#define    CN23XX_WIN_WR_ADDR_HI                   0x20004
+#define    CN23XX_WIN_WR_ADDR64                    CN23XX_WIN_WR_ADDR_LO
+
+#define    CN23XX_WIN_RD_ADDR_LO                   0x20010
+#define    CN23XX_WIN_RD_ADDR_HI                   0x20014
+#define    CN23XX_WIN_RD_ADDR64                    CN23XX_WIN_RD_ADDR_LO
+
+#define    CN23XX_WIN_WR_DATA_LO                   0x20020
+#define    CN23XX_WIN_WR_DATA_HI                   0x20024
+#define    CN23XX_WIN_WR_DATA64                    CN23XX_WIN_WR_DATA_LO
+
+#define    CN23XX_WIN_RD_DATA_LO                   0x20040
+#define    CN23XX_WIN_RD_DATA_HI                   0x20044
+#define    CN23XX_WIN_RD_DATA64                    CN23XX_WIN_RD_DATA_LO
+
+#define    CN23XX_WIN_WR_MASK_LO                   0x20030
+#define    CN23XX_WIN_WR_MASK_HI                   0x20034
+#define    CN23XX_WIN_WR_MASK_REG                  CN23XX_WIN_WR_MASK_LO
+#define    CN23XX_SLI_MAC_CREDIT_CNT               0x23D70
+
+/* 4 registers (64-bit) for mapping IOQs to MACs(PEMs)-
+ * SLI_PKT_MAC(0..3)_PF(0..1)_RINFO
+ */
+#define    CN23XX_SLI_PKT_MAC_RINFO_START64       0x29030
+
+/*1 register (64-bit) to determine whether IOQs are in reset. */
+#define    CN23XX_SLI_PKT_IOQ_RING_RST            0x291E0
+
+/* Each Input Queue register is at a 16-byte Offset in BAR0 */
+#define    CN23XX_IQ_OFFSET                       0x20000
+
+#define    CN23XX_MAC_RINFO_OFFSET                0x20
+#define    CN23XX_PF_RINFO_OFFSET                 0x10
+
+#define CN23XX_SLI_PKT_MAC_RINFO64(mac, pf)		\
+		(CN23XX_SLI_PKT_MAC_RINFO_START64 +     \
+		 ((mac) * CN23XX_MAC_RINFO_OFFSET) +	\
+		 ((pf) * CN23XX_PF_RINFO_OFFSET))
+
+/** mask for total rings, setting TRS to base */
+#define    CN23XX_PKT_MAC_CTL_RINFO_TRS               BIT_ULL(16)
+/** mask for starting ring number: setting SRN <6:0> = 0x7F */
+#define    CN23XX_PKT_MAC_CTL_RINFO_SRN               (0x7F)
+
+/* Starting bit of the TRS field in CN23XX_SLI_PKT_MAC_RINFO64 register */
+#define    CN23XX_PKT_MAC_CTL_RINFO_TRS_BIT_POS     16
+/* Starting bit of SRN field in CN23XX_SLI_PKT_MAC_RINFO64 register */
+#define    CN23XX_PKT_MAC_CTL_RINFO_SRN_BIT_POS     0
+/* Starting bit of RPVF field in CN23XX_SLI_PKT_MAC_RINFO64 register */
+#define    CN23XX_PKT_MAC_CTL_RINFO_RPVF_BIT_POS     32
+/* Starting bit of NVFS field in CN23XX_SLI_PKT_MAC_RINFO64 register */
+#define    CN23XX_PKT_MAC_CTL_RINFO_NVFS_BIT_POS     48
+
+/*###################### REQUEST QUEUE #########################*/
+
+/* 64 registers for Input Queue Instr Count - SLI_PKT_IN_DONE0_CNTS */
+#define    CN23XX_SLI_IQ_INSTR_COUNT_START64     0x10040
+
+/* 64 registers for Input Queues Start Addr - SLI_PKT0_INSTR_BADDR */
+#define    CN23XX_SLI_IQ_BASE_ADDR_START64       0x10010
+
+/* 64 registers for Input Doorbell - SLI_PKT0_INSTR_BAOFF_DBELL */
+#define    CN23XX_SLI_IQ_DOORBELL_START          0x10020
+
+/* 64 registers for Input Queue size - SLI_PKT0_INSTR_FIFO_RSIZE */
+#define    CN23XX_SLI_IQ_SIZE_START              0x10030
+
+/* 64 registers (64-bit) - ES, RO, NS, Arbitration for Input Queue Data &
+ * gather list fetches. SLI_PKT(0..63)_INPUT_CONTROL.
+ */
+#define    CN23XX_SLI_IQ_PKT_CONTROL_START64    0x10000
+
+/*------- Request Queue Macros ---------*/
+#define    CN23XX_SLI_IQ_PKT_CONTROL64(iq)          \
+		(CN23XX_SLI_IQ_PKT_CONTROL_START64 + ((iq) * CN23XX_IQ_OFFSET))
+
+#define    CN23XX_SLI_IQ_BASE_ADDR64(iq)          \
+		(CN23XX_SLI_IQ_BASE_ADDR_START64 + ((iq) * CN23XX_IQ_OFFSET))
+
+#define    CN23XX_SLI_IQ_SIZE(iq)                 \
+		(CN23XX_SLI_IQ_SIZE_START + ((iq) * CN23XX_IQ_OFFSET))
+
+#define    CN23XX_SLI_IQ_DOORBELL(iq)             \
+		(CN23XX_SLI_IQ_DOORBELL_START + ((iq) * CN23XX_IQ_OFFSET))
+
+#define    CN23XX_SLI_IQ_INSTR_COUNT64(iq)          \
+		(CN23XX_SLI_IQ_INSTR_COUNT_START64 + ((iq) * CN23XX_IQ_OFFSET))
+
+/*------------------ Masks ----------------*/
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM                  BIT_ULL(32)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM                 BIT(29)
+/* Number of instructions to be read in one MAC read request.
+ * setting to Max value(4)
+ */
+#define    CN23XX_PKT_INPUT_CTL_RDSIZE                  (3 << 25)
+#define    CN23XX_PKT_INPUT_CTL_IS_64B                  BIT(24)
+#define    CN23XX_PKT_INPUT_CTL_RST                     BIT(23)
+#define    CN23XX_PKT_INPUT_CTL_QUIET                   BIT(28)
+#define    CN23XX_PKT_INPUT_CTL_RING_ENB                BIT(22)
+#define    CN23XX_PKT_INPUT_CTL_DATA_NS                 BIT(8)
+#define    CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP        BIT(6)
+#define    CN23XX_PKT_INPUT_CTL_DATA_RO                 BIT(5)
+#define    CN23XX_PKT_INPUT_CTL_USE_CSR                 BIT(4)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_NS               BIT(3)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_ES_64B_SWAP      (2)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_RO               (1)
+
+/** Rings per Virtual Function **/
+#define    CN23XX_PKT_INPUT_CTL_RPVF_MASK               (0x3F)
+#define    CN23XX_PKT_INPUT_CTL_RPVF_POS                (48)
+/** These bits[47:44] select the Physical function number within the MAC */
+#define    CN23XX_PKT_INPUT_CTL_PF_NUM_MASK             (0x7)
+#define    CN23XX_PKT_INPUT_CTL_PF_NUM_POS              (45)
+/** These bits[43:32] select the function number within the PF */
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM_MASK             (0x1FFF)
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM_POS              (32)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM_MASK            (0x3)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM_POS             (29)
+#define    CN23XX_PKT_IN_DONE_WMARK_MASK                (0xFFFFULL)
+#define    CN23XX_PKT_IN_DONE_WMARK_BIT_POS             (32)
+#define    CN23XX_PKT_IN_DONE_CNT_MASK                  (0x00000000FFFFFFFFULL)
+
+#ifdef __LITTLE_ENDIAN_BITFIELD
+#define    CN23XX_PKT_INPUT_CTL_MASK				\
+		(CN23XX_PKT_INPUT_CTL_RDSIZE		|	\
+		 CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP	|	\
+		 CN23XX_PKT_INPUT_CTL_USE_CSR)
+#else
+#define    CN23XX_PKT_INPUT_CTL_MASK				\
+		(CN23XX_PKT_INPUT_CTL_RDSIZE		|	\
+		 CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP	|	\
+		 CN23XX_PKT_INPUT_CTL_USE_CSR		|	\
+		 CN23XX_PKT_INPUT_CTL_GATHER_ES_64B_SWAP)
+#endif
+
+/** Masks for SLI_PKT_IN_DONE(0..63)_CNTS Register */
+#define    CN23XX_IN_DONE_CNTS_PI_INT               BIT_ULL(62)
+#define    CN23XX_IN_DONE_CNTS_CINT_ENB             BIT_ULL(48)
+
+/*############################ OUTPUT QUEUE #########################*/
+
+/* 64 registers for Output queue control - SLI_PKT(0..63)_OUTPUT_CONTROL */
+#define    CN23XX_SLI_OQ_PKT_CONTROL_START       0x10050
+
+/* 64 registers for Output queue buffer and info size - SLI_PKT0_OUT_SIZE */
+#define    CN23XX_SLI_OQ0_BUFF_INFO_SIZE         0x10060
+
+/* 64 registers for Output Queue Start Addr - SLI_PKT0_SLIST_BADDR */
+#define    CN23XX_SLI_OQ_BASE_ADDR_START64       0x10070
+
+/* 64 registers for Output Queue Packet Credits - SLI_PKT0_SLIST_BAOFF_DBELL */
+#define    CN23XX_SLI_OQ_PKT_CREDITS_START       0x10080
+
+/* 64 registers for Output Queue size - SLI_PKT0_SLIST_FIFO_RSIZE */
+#define    CN23XX_SLI_OQ_SIZE_START              0x10090
+
+/* 64 registers for Output Queue Packet Count - SLI_PKT0_CNTS */
+#define    CN23XX_SLI_OQ_PKT_SENT_START          0x100B0
+
+/* 64 registers for Output Queue INT Levels - SLI_PKT0_INT_LEVELS */
+#define    CN23XX_SLI_OQ_PKT_INT_LEVELS_START64   0x100A0
+
+/* Each Output Queue register is at a 16-byte Offset in BAR0 */
+#define    CN23XX_OQ_OFFSET                      0x20000
+
+/* 1 (64-bit register) for Output Queue backpressure across all rings. */
+#define    CN23XX_SLI_OQ_WMARK                   0x29180
+
+/* Global pkt control register */
+#define    CN23XX_SLI_GBL_CONTROL                0x29210
+
+/* Backpressure enable register for PF0  */
+#define    CN23XX_SLI_OUT_BP_EN_W1S              0x29260
+
+/* Backpressure enable register for PF1  */
+#define    CN23XX_SLI_OUT_BP_EN2_W1S             0x29270
+
+/* Backpressure disable register for PF0  */
+#define    CN23XX_SLI_OUT_BP_EN_W1C              0x29280
+
+/* Backpressure disable register for PF1  */
+#define    CN23XX_SLI_OUT_BP_EN2_W1C             0x29290
+
+/*------- Output Queue Macros ---------*/
+
+#define    CN23XX_SLI_OQ_PKT_CONTROL(oq)          \
+		(CN23XX_SLI_OQ_PKT_CONTROL_START + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_BASE_ADDR64(oq)          \
+		(CN23XX_SLI_OQ_BASE_ADDR_START64 + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_SIZE(oq)                 \
+		(CN23XX_SLI_OQ_SIZE_START + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_BUFF_INFO_SIZE(oq)                 \
+		(CN23XX_SLI_OQ0_BUFF_INFO_SIZE + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_PKTS_SENT(oq)            \
+		(CN23XX_SLI_OQ_PKT_SENT_START + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_PKTS_CREDIT(oq)          \
+		(CN23XX_SLI_OQ_PKT_CREDITS_START + ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_PKT_INT_LEVELS(oq)		\
+		(CN23XX_SLI_OQ_PKT_INT_LEVELS_START64 +	\
+		 ((oq) * CN23XX_OQ_OFFSET))
+
+/*Macro's for accessing CNT and TIME separately from INT_LEVELS*/
+#define    CN23XX_SLI_OQ_PKT_INT_LEVELS_CNT(oq)		\
+		(CN23XX_SLI_OQ_PKT_INT_LEVELS_START64 + \
+		 ((oq) * CN23XX_OQ_OFFSET))
+
+#define    CN23XX_SLI_OQ_PKT_INT_LEVELS_TIME(oq)	\
+		(CN23XX_SLI_OQ_PKT_INT_LEVELS_START64 +	\
+		 ((oq) * CN23XX_OQ_OFFSET) + 4)
+
+/*------------------ Masks ----------------*/
+#define    CN23XX_PKT_OUTPUT_CTL_TENB                  BIT(13)
+#define    CN23XX_PKT_OUTPUT_CTL_CENB                  BIT(12)
+#define    CN23XX_PKT_OUTPUT_CTL_IPTR                  BIT(11)
+#define    CN23XX_PKT_OUTPUT_CTL_ES                    BIT(9)
+#define    CN23XX_PKT_OUTPUT_CTL_NSR                   BIT(8)
+#define    CN23XX_PKT_OUTPUT_CTL_ROR                   BIT(7)
+#define    CN23XX_PKT_OUTPUT_CTL_DPTR                  BIT(6)
+#define    CN23XX_PKT_OUTPUT_CTL_BMODE                 BIT(5)
+#define    CN23XX_PKT_OUTPUT_CTL_ES_P                  BIT(3)
+#define    CN23XX_PKT_OUTPUT_CTL_NSR_P                 BIT(2)
+#define    CN23XX_PKT_OUTPUT_CTL_ROR_P                 BIT(1)
+#define    CN23XX_PKT_OUTPUT_CTL_RING_ENB              BIT(0)
+
+/*######################### Mailbox Reg Macros ########################*/
+#define    CN23XX_SLI_PKT_MBOX_INT_START             0x10210
+#define    CN23XX_SLI_PKT_PF_VF_MBOX_SIG_START       0x10200
+#define    CN23XX_SLI_MAC_PF_MBOX_INT_START          0x27380
+
+#define    CN23XX_SLI_MBOX_OFFSET		     0x20000
+#define    CN23XX_SLI_MBOX_SIG_IDX_OFFSET	     0x8
+
+#define    CN23XX_SLI_PKT_MBOX_INT(q)          \
+		(CN23XX_SLI_PKT_MBOX_INT_START + ((q) * CN23XX_SLI_MBOX_OFFSET))
+
+#define    CN23XX_SLI_PKT_PF_VF_MBOX_SIG(q, idx)		\
+		(CN23XX_SLI_PKT_PF_VF_MBOX_SIG_START +		\
+		 ((q) * CN23XX_SLI_MBOX_OFFSET +		\
+		  (idx) * CN23XX_SLI_MBOX_SIG_IDX_OFFSET))
+
+#define    CN23XX_SLI_MAC_PF_MBOX_INT(mac, pf)		\
+		(CN23XX_SLI_MAC_PF_MBOX_INT_START +	\
+		 ((mac) * CN23XX_MAC_INT_OFFSET +	\
+		  (pf) * CN23XX_PF_INT_OFFSET))
+
+/*######################### DMA Counters #########################*/
+
+/* 2 registers (64-bit) - DMA Count - 1 for each DMA counter 0/1. */
+#define    CN23XX_DMA_CNT_START                   0x28400
+
+/* 2 registers (64-bit) - DMA Timer 0/1, contains DMA timer values */
+/* SLI_DMA_0_TIM */
+#define    CN23XX_DMA_TIM_START                   0x28420
+
+/* 2 registers (64-bit) - DMA count & Time Interrupt threshold -
+ * SLI_DMA_0_INT_LEVEL
+ */
+#define    CN23XX_DMA_INT_LEVEL_START             0x283E0
+
+/* Each DMA register is at a 16-byte Offset in BAR0 */
+#define    CN23XX_DMA_OFFSET                      0x10
+
+/*---------- DMA Counter Macros ---------*/
+#define    CN23XX_DMA_CNT(dq)                      \
+		(CN23XX_DMA_CNT_START + ((dq) * CN23XX_DMA_OFFSET))
+
+#define    CN23XX_DMA_INT_LEVEL(dq)                \
+		(CN23XX_DMA_INT_LEVEL_START + ((dq) * CN23XX_DMA_OFFSET))
+
+#define    CN23XX_DMA_PKT_INT_LEVEL(dq)            \
+		(CN23XX_DMA_INT_LEVEL_START + ((dq) * CN23XX_DMA_OFFSET))
+
+#define    CN23XX_DMA_TIME_INT_LEVEL(dq)           \
+		(CN23XX_DMA_INT_LEVEL_START + 4 + ((dq) * CN23XX_DMA_OFFSET))
+
+#define    CN23XX_DMA_TIM(dq)                     \
+		(CN23XX_DMA_TIM_START + ((dq) * CN23XX_DMA_OFFSET))
+
+/*######################## MSIX TABLE #########################*/
+
+#define	CN23XX_MSIX_TABLE_ADDR_START		0x0
+#define	CN23XX_MSIX_TABLE_DATA_START		0x8
+
+#define	CN23XX_MSIX_TABLE_SIZE			0x10
+#define	CN23XX_MSIX_TABLE_ENTRIES		0x41
+
+#define CN23XX_MSIX_ENTRY_VECTOR_CTL	BIT_ULL(32)
+
+#define	CN23XX_MSIX_TABLE_ADDR(idx)		\
+	(CN23XX_MSIX_TABLE_ADDR_START + ((idx) * CN23XX_MSIX_TABLE_SIZE))
+
+#define	CN23XX_MSIX_TABLE_DATA(idx)		\
+	(CN23XX_MSIX_TABLE_DATA_START + ((idx) * CN23XX_MSIX_TABLE_SIZE))
+
+/*######################## INTERRUPTS #########################*/
+#define CN23XX_MAC_INT_OFFSET   0x20
+#define CN23XX_PF_INT_OFFSET    0x10
+
+/* 1 register (64-bit) for Interrupt Summary */
+#define    CN23XX_SLI_INT_SUM64            0x27000
+
+/* 4 registers (64-bit) for Interrupt Enable for each Port */
+#define    CN23XX_SLI_INT_ENB64            0x27080
+
+#define    CN23XX_SLI_MAC_PF_INT_SUM64(mac, pf)			\
+		(CN23XX_SLI_INT_SUM64 +				\
+		 ((mac) * CN23XX_MAC_INT_OFFSET) +		\
+		 ((pf) * CN23XX_PF_INT_OFFSET))
+
+#define    CN23XX_SLI_MAC_PF_INT_ENB64(mac, pf)		\
+		(CN23XX_SLI_INT_ENB64 +			\
+		 ((mac) * CN23XX_MAC_INT_OFFSET) +	\
+		 ((pf) * CN23XX_PF_INT_OFFSET))
+
+/* 1 register (64-bit) to indicate which Output Queue reached pkt threshold */
+#define    CN23XX_SLI_PKT_CNT_INT                0x29130
+
+/* 1 register (64-bit) to indicate which Output Queue reached time threshold */
+#define    CN23XX_SLI_PKT_TIME_INT               0x29140
+
+/*------------------ Interrupt Masks ----------------*/
+
+#define    CN23XX_INTR_PO_INT			BIT_ULL(63)
+#define    CN23XX_INTR_PI_INT			BIT_ULL(62)
+#define    CN23XX_INTR_MBOX_INT			BIT_ULL(61)
+#define    CN23XX_INTR_RESEND			BIT_ULL(60)
+
+#define    CN23XX_INTR_CINT_ENB                 BIT_ULL(48)
+#define    CN23XX_INTR_MBOX_ENB                 BIT(0)
+
+#define    CN23XX_INTR_RML_TIMEOUT_ERR           (1)
+
+#define    CN23XX_INTR_MIO_INT                   BIT(1)
+
+#define    CN23XX_INTR_RESERVED1                 (3 << 2)
+
+#define    CN23XX_INTR_PKT_COUNT                 BIT(4)
+#define    CN23XX_INTR_PKT_TIME                  BIT(5)
+
+#define    CN23XX_INTR_RESERVED2                 (3 << 6)
+
+#define    CN23XX_INTR_M0UPB0_ERR                BIT(8)
+#define    CN23XX_INTR_M0UPWI_ERR                BIT(9)
+#define    CN23XX_INTR_M0UNB0_ERR                BIT(10)
+#define    CN23XX_INTR_M0UNWI_ERR                BIT(11)
+
+#define    CN23XX_INTR_RESERVED3                 (0xFFFFFULL << 12)
+
+#define    CN23XX_INTR_DMA0_FORCE                BIT_ULL(32)
+#define    CN23XX_INTR_DMA1_FORCE                BIT_ULL(33)
+
+#define    CN23XX_INTR_DMA0_COUNT                BIT_ULL(34)
+#define    CN23XX_INTR_DMA1_COUNT                BIT_ULL(35)
+
+#define    CN23XX_INTR_DMA0_TIME                 BIT_ULL(36)
+#define    CN23XX_INTR_DMA1_TIME                 BIT_ULL(37)
+
+#define    CN23XX_INTR_RESERVED4                 (0x7FFFFULL << 38)
+
+#define    CN23XX_INTR_VF_MBOX                   BIT_ULL(57)
+#define    CN23XX_INTR_DMAVF_ERR                 BIT_ULL(58)
+#define    CN23XX_INTR_DMAPF_ERR                 BIT_ULL(59)
+
+#define    CN23XX_INTR_PKTVF_ERR                 BIT_ULL(60)
+#define    CN23XX_INTR_PKTPF_ERR                 BIT_ULL(61)
+#define    CN23XX_INTR_PPVF_ERR                  BIT_ULL(62)
+#define    CN23XX_INTR_PPPF_ERR                  BIT_ULL(63)
+
+#define    CN23XX_INTR_DMA0_DATA                 (CN23XX_INTR_DMA0_TIME)
+#define    CN23XX_INTR_DMA1_DATA                 (CN23XX_INTR_DMA1_TIME)
+
+#define    CN23XX_INTR_DMA_DATA                  \
+		(CN23XX_INTR_DMA0_DATA | CN23XX_INTR_DMA1_DATA)
+
+/* By fault only TIME based */
+#define    CN23XX_INTR_PKT_DATA                  (CN23XX_INTR_PKT_TIME)
+/* For both COUNT and TIME based */
+/* #define    CN23XX_INTR_PKT_DATA                  \
+ * (CN23XX_INTR_PKT_COUNT | CN23XX_INTR_PKT_TIME)
+ */
+
+/* Sum of interrupts for all PCI-Express Data Interrupts */
+#define    CN23XX_INTR_PCIE_DATA                 \
+		(CN23XX_INTR_DMA_DATA | CN23XX_INTR_PKT_DAT)
+
+/* Sum of interrupts for error events */
+#define    CN23XX_INTR_ERR			\
+		(CN23XX_INTR_M0UPB0_ERR	|	\
+		 CN23XX_INTR_M0UPWI_ERR	|	\
+		 CN23XX_INTR_M0UNB0_ERR	|	\
+		 CN23XX_INTR_M0UNWI_ERR	|	\
+		 CN23XX_INTR_DMAVF_ERR	|	\
+		 CN23XX_INTR_DMAPF_ERR	|	\
+		 CN23XX_INTR_PKTPF_ERR	|	\
+		 CN23XX_INTR_PPPF_ERR	|	\
+		 CN23XX_INTR_PPVF_ERR)
+
+/* Programmed Mask for Interrupt Sum */
+#define    CN23XX_INTR_MASK			\
+		(CN23XX_INTR_DMA_DATA	|	\
+		 CN23XX_INTR_DMA0_FORCE	|	\
+		 CN23XX_INTR_DMA1_FORCE	|	\
+		 CN23XX_INTR_MIO_INT	|	\
+		 CN23XX_INTR_ERR)
+
+/* 4 Registers (64 - bit) */
+#define    CN23XX_SLI_S2M_PORT_CTL_START         0x23D80
+#define    CN23XX_SLI_S2M_PORTX_CTL(port)	\
+		(CN23XX_SLI_S2M_PORT_CTL_START + ((port) * 0x10))
+
+#define    CN23XX_SLI_MAC_NUMBER                 0x20050
+
+/** PEM(0..3)_BAR1_INDEX(0..15)address is defined as
+ *  addr = (0x00011800C0000100  |port <<24 |idx <<3 )
+ *  Here, port is PEM(0..3) & idx is INDEX(0..15)
+ */
+#define    CN23XX_PEM_BAR1_INDEX_START             0x00011800C0000100ULL
+#define    CN23XX_PEM_OFFSET                       24
+#define    CN23XX_BAR1_INDEX_OFFSET                3
+
+#define    CN23XX_PEM_BAR1_INDEX_REG(port, idx)		\
+		(CN23XX_PEM_BAR1_INDEX_START + (((u64)port) << CN23XX_PEM_OFFSET) + \
+		 ((idx) << CN23XX_BAR1_INDEX_OFFSET))
+
+/*############################ DPI #########################*/
+
+/* 1 register (64-bit) - provides DMA Enable */
+#define    CN23XX_DPI_CTL                 0x0001df0000000040ULL
+
+/* 1 register (64-bit) - Controls the DMA IO Operation */
+#define    CN23XX_DPI_DMA_CONTROL         0x0001df0000000048ULL
+
+/* 1 register (64-bit) - Provides DMA Instr'n Queue Enable  */
+#define    CN23XX_DPI_REQ_GBL_ENB         0x0001df0000000050ULL
+
+/* 1 register (64-bit) - DPI_REQ_ERR_RSP
+ * Indicates which Instr'n Queue received error response from the IO sub-system
+ */
+#define    CN23XX_DPI_REQ_ERR_RSP         0x0001df0000000058ULL
+
+/* 1 register (64-bit) - DPI_REQ_ERR_RST
+ * Indicates which Instr'n Queue dropped an Instr'n
+ */
+#define    CN23XX_DPI_REQ_ERR_RST         0x0001df0000000060ULL
+
+/* 6 register (64-bit) - DPI_DMA_ENG(0..5)_EN
+ * Provides DMA Engine Queue Enable
+ */
+#define    CN23XX_DPI_DMA_ENG0_ENB        0x0001df0000000080ULL
+#define    CN23XX_DPI_DMA_ENG_ENB(eng) (CN23XX_DPI_DMA_ENG0_ENB + ((eng) * 8))
+
+/* 8 register (64-bit) - DPI_DMA(0..7)_REQQ_CTL
+ * Provides control bits for transaction on 8 Queues
+ */
+#define    CN23XX_DPI_DMA_REQQ0_CTL       0x0001df0000000180ULL
+#define    CN23XX_DPI_DMA_REQQ_CTL(q_no)	\
+		(CN23XX_DPI_DMA_REQQ0_CTL + ((q_no) * 8))
+
+/* 6 register (64-bit) - DPI_ENG(0..5)_BUF
+ * Provides DMA Engine FIFO (Queue) Size
+ */
+#define    CN23XX_DPI_DMA_ENG0_BUF        0x0001df0000000880ULL
+#define    CN23XX_DPI_DMA_ENG_BUF(eng)   \
+		(CN23XX_DPI_DMA_ENG0_BUF + ((eng) * 8))
+
+/* 4 Registers (64-bit) */
+#define    CN23XX_DPI_SLI_PRT_CFG_START   0x0001df0000000900ULL
+#define    CN23XX_DPI_SLI_PRTX_CFG(port)        \
+		(CN23XX_DPI_SLI_PRT_CFG_START + ((port) * 0x8))
+
+/* Masks for DPI_DMA_CONTROL Register */
+#define    CN23XX_DPI_DMA_COMMIT_MODE     BIT_ULL(58)
+#define    CN23XX_DPI_DMA_PKT_EN          BIT_ULL(56)
+#define    CN23XX_DPI_DMA_ENB             (0x0FULL << 48)
+/* Set the DMA Control, to update packet count not byte count sent by DMA,
+ * when we use Interrupt Coalescing (CA mode)
+ */
+#define    CN23XX_DPI_DMA_O_ADD1          BIT(19)
+/*selecting 64-bit Byte Swap Mode */
+#define    CN23XX_DPI_DMA_O_ES            BIT(15)
+#define    CN23XX_DPI_DMA_O_MODE          BIT(14)
+
+#define    CN23XX_DPI_DMA_CTL_MASK			\
+		(CN23XX_DPI_DMA_COMMIT_MODE	|	\
+		 CN23XX_DPI_DMA_PKT_EN		|	\
+		 CN23XX_DPI_DMA_O_ES		|	\
+		 CN23XX_DPI_DMA_O_MODE)
+
+/*############################ RST #########################*/
+
+#define    CN23XX_RST_BOOT            0x0001180006001600ULL
+#define    CN23XX_RST_SOFT_RST        0x0001180006001680ULL
+
+#define    CN23XX_LMC0_RESET_CTL               0x0001180088000180ULL
+#define    CN23XX_LMC0_RESET_CTL_DDR3RST_MASK  0x0000000000000001ULL
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn23xx_vf_device.c

@@ -0,0 +1,683 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/vmalloc.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "cn23xx_vf_device.h"
+#include "octeon_main.h"
+#include "octeon_mailbox.h"
+
+u32 cn23xx_vf_get_oq_ticks(struct octeon_device *oct, u32 time_intr_in_us)
+{
+	/* This gives the SLI clock per microsec */
+	u32 oqticks_per_us = (u32)oct->pfvf_hsword.coproc_tics_per_us;
+
+	/* This gives the clock cycles per millisecond */
+	oqticks_per_us *= 1000;
+
+	/* This gives the oq ticks (1024 core clock cycles) per millisecond */
+	oqticks_per_us /= 1024;
+
+	/* time_intr is in microseconds. The next 2 steps gives the oq ticks
+	 * corressponding to time_intr.
+	 */
+	oqticks_per_us *= time_intr_in_us;
+	oqticks_per_us /= 1000;
+
+	return oqticks_per_us;
+}
+
+static int cn23xx_vf_reset_io_queues(struct octeon_device *oct, u32 num_queues)
+{
+	u32 loop = BUSY_READING_REG_VF_LOOP_COUNT;
+	int ret_val = 0;
+	u32 q_no;
+	u64 d64;
+
+	for (q_no = 0; q_no < num_queues; q_no++) {
+		/* set RST bit to 1. This bit applies to both IQ and OQ */
+		d64 = octeon_read_csr64(oct,
+					CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no));
+		d64 |= CN23XX_PKT_INPUT_CTL_RST;
+		octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no),
+				   d64);
+	}
+
+	/* wait until the RST bit is clear or the RST and QUIET bits are set */
+	for (q_no = 0; q_no < num_queues; q_no++) {
+		u64 reg_val = octeon_read_csr64(oct,
+					CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no));
+		while ((READ_ONCE(reg_val) & CN23XX_PKT_INPUT_CTL_RST) &&
+		       !(READ_ONCE(reg_val) & CN23XX_PKT_INPUT_CTL_QUIET) &&
+		       loop) {
+			WRITE_ONCE(reg_val, octeon_read_csr64(
+			    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no)));
+			loop--;
+		}
+		if (!loop) {
+			dev_err(&oct->pci_dev->dev,
+				"clearing the reset reg failed or setting the quiet reg failed for qno: %u\n",
+				q_no);
+			return -1;
+		}
+		WRITE_ONCE(reg_val, READ_ONCE(reg_val) &
+			   ~CN23XX_PKT_INPUT_CTL_RST);
+		octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no),
+				   READ_ONCE(reg_val));
+
+		WRITE_ONCE(reg_val, octeon_read_csr64(
+		    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no)));
+		if (READ_ONCE(reg_val) & CN23XX_PKT_INPUT_CTL_RST) {
+			dev_err(&oct->pci_dev->dev,
+				"clearing the reset failed for qno: %u\n",
+				q_no);
+			ret_val = -1;
+		}
+	}
+
+	return ret_val;
+}
+
+static int cn23xx_vf_setup_global_input_regs(struct octeon_device *oct)
+{
+	struct octeon_cn23xx_vf *cn23xx = (struct octeon_cn23xx_vf *)oct->chip;
+	struct octeon_instr_queue *iq;
+	u64 q_no, intr_threshold;
+	u64 d64;
+
+	if (cn23xx_vf_reset_io_queues(oct, oct->sriov_info.rings_per_vf))
+		return -1;
+
+	for (q_no = 0; q_no < (oct->sriov_info.rings_per_vf); q_no++) {
+		void __iomem *inst_cnt_reg;
+
+		octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_DOORBELL(q_no),
+				   0xFFFFFFFF);
+		iq = oct->instr_queue[q_no];
+
+		if (iq)
+			inst_cnt_reg = iq->inst_cnt_reg;
+		else
+			inst_cnt_reg = (u8 *)oct->mmio[0].hw_addr +
+				       CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no);
+
+		d64 = octeon_read_csr64(oct,
+					CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no));
+
+		d64 &= 0xEFFFFFFFFFFFFFFFL;
+
+		octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no),
+				   d64);
+
+		/* Select ES, RO, NS, RDSIZE,DPTR Fomat#0 for
+		 * the Input Queues
+		 */
+		octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no),
+				   CN23XX_PKT_INPUT_CTL_MASK);
+
+		/* set the wmark level to trigger PI_INT */
+		intr_threshold = CFG_GET_IQ_INTR_PKT(cn23xx->conf) &
+				 CN23XX_PKT_IN_DONE_WMARK_MASK;
+
+		writeq((readq(inst_cnt_reg) &
+			~(CN23XX_PKT_IN_DONE_WMARK_MASK <<
+			  CN23XX_PKT_IN_DONE_WMARK_BIT_POS)) |
+		       (intr_threshold << CN23XX_PKT_IN_DONE_WMARK_BIT_POS),
+		       inst_cnt_reg);
+	}
+	return 0;
+}
+
+static void cn23xx_vf_setup_global_output_regs(struct octeon_device *oct)
+{
+	u32 reg_val;
+	u32 q_no;
+
+	for (q_no = 0; q_no < (oct->sriov_info.rings_per_vf); q_no++) {
+		octeon_write_csr(oct, CN23XX_VF_SLI_OQ_PKTS_CREDIT(q_no),
+				 0xFFFFFFFF);
+
+		reg_val =
+		    octeon_read_csr(oct, CN23XX_VF_SLI_OQ_PKTS_SENT(q_no));
+
+		reg_val &= 0xEFFFFFFFFFFFFFFFL;
+
+		reg_val =
+		    octeon_read_csr(oct, CN23XX_VF_SLI_OQ_PKT_CONTROL(q_no));
+
+		/* clear IPTR */
+		reg_val &= ~CN23XX_PKT_OUTPUT_CTL_IPTR;
+
+		/* set DPTR */
+		reg_val |= CN23XX_PKT_OUTPUT_CTL_DPTR;
+
+		/* reset BMODE */
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_BMODE);
+
+		/* No Relaxed Ordering, No Snoop, 64-bit Byte swap
+		 * for Output Queue ScatterList reset ROR_P, NSR_P
+		 */
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_ROR_P);
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_NSR_P);
+
+#ifdef __LITTLE_ENDIAN_BITFIELD
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_ES_P);
+#else
+		reg_val |= (CN23XX_PKT_OUTPUT_CTL_ES_P);
+#endif
+		/* No Relaxed Ordering, No Snoop, 64-bit Byte swap
+		 * for Output Queue Data reset ROR, NSR
+		 */
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_ROR);
+		reg_val &= ~(CN23XX_PKT_OUTPUT_CTL_NSR);
+		/* set the ES bit */
+		reg_val |= (CN23XX_PKT_OUTPUT_CTL_ES);
+
+		/* write all the selected settings */
+		octeon_write_csr(oct, CN23XX_VF_SLI_OQ_PKT_CONTROL(q_no),
+				 reg_val);
+	}
+}
+
+static int cn23xx_setup_vf_device_regs(struct octeon_device *oct)
+{
+	if (cn23xx_vf_setup_global_input_regs(oct))
+		return -1;
+
+	cn23xx_vf_setup_global_output_regs(oct);
+
+	return 0;
+}
+
+static void cn23xx_setup_vf_iq_regs(struct octeon_device *oct, u32 iq_no)
+{
+	struct octeon_instr_queue *iq = oct->instr_queue[iq_no];
+	u64 pkt_in_done;
+
+	/* Write the start of the input queue's ring and its size */
+	octeon_write_csr64(oct, CN23XX_VF_SLI_IQ_BASE_ADDR64(iq_no),
+			   iq->base_addr_dma);
+	octeon_write_csr(oct, CN23XX_VF_SLI_IQ_SIZE(iq_no), iq->max_count);
+
+	/* Remember the doorbell & instruction count register addr
+	 * for this queue
+	 */
+	iq->doorbell_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_VF_SLI_IQ_DOORBELL(iq_no);
+	iq->inst_cnt_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_VF_SLI_IQ_INSTR_COUNT64(iq_no);
+	dev_dbg(&oct->pci_dev->dev, "InstQ[%d]:dbell reg @ 0x%p instcnt_reg @ 0x%p\n",
+		iq_no, iq->doorbell_reg, iq->inst_cnt_reg);
+
+	/* Store the current instruction counter (used in flush_iq
+	 * calculation)
+	 */
+	pkt_in_done = readq(iq->inst_cnt_reg);
+
+	if (oct->msix_on) {
+		/* Set CINT_ENB to enable IQ interrupt */
+		writeq((pkt_in_done | CN23XX_INTR_CINT_ENB),
+		       iq->inst_cnt_reg);
+	}
+	iq->reset_instr_cnt = 0;
+}
+
+static void cn23xx_setup_vf_oq_regs(struct octeon_device *oct, u32 oq_no)
+{
+	struct octeon_droq *droq = oct->droq[oq_no];
+
+	octeon_write_csr64(oct, CN23XX_VF_SLI_OQ_BASE_ADDR64(oq_no),
+			   droq->desc_ring_dma);
+	octeon_write_csr(oct, CN23XX_VF_SLI_OQ_SIZE(oq_no), droq->max_count);
+
+	octeon_write_csr(oct, CN23XX_VF_SLI_OQ_BUFF_INFO_SIZE(oq_no),
+			 droq->buffer_size);
+
+	/* Get the mapped address of the pkt_sent and pkts_credit regs */
+	droq->pkts_sent_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_VF_SLI_OQ_PKTS_SENT(oq_no);
+	droq->pkts_credit_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_VF_SLI_OQ_PKTS_CREDIT(oq_no);
+}
+
+static void cn23xx_vf_mbox_thread(struct work_struct *work)
+{
+	struct cavium_wk *wk = (struct cavium_wk *)work;
+	struct octeon_mbox *mbox = (struct octeon_mbox *)wk->ctxptr;
+
+	octeon_mbox_process_message(mbox);
+}
+
+static int cn23xx_free_vf_mbox(struct octeon_device *oct)
+{
+	cancel_delayed_work_sync(&oct->mbox[0]->mbox_poll_wk.work);
+	vfree(oct->mbox[0]);
+	return 0;
+}
+
+static int cn23xx_setup_vf_mbox(struct octeon_device *oct)
+{
+	struct octeon_mbox *mbox = NULL;
+
+	mbox = vzalloc(sizeof(*mbox));
+	if (!mbox)
+		return 1;
+
+	spin_lock_init(&mbox->lock);
+
+	mbox->oct_dev = oct;
+
+	mbox->q_no = 0;
+
+	mbox->state = OCTEON_MBOX_STATE_IDLE;
+
+	/* VF mbox interrupt reg */
+	mbox->mbox_int_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_VF_SLI_PKT_MBOX_INT(0);
+	/* VF reads from SIG0 reg */
+	mbox->mbox_read_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_SLI_PKT_PF_VF_MBOX_SIG(0, 0);
+	/* VF writes into SIG1 reg */
+	mbox->mbox_write_reg =
+	    (u8 *)oct->mmio[0].hw_addr + CN23XX_SLI_PKT_PF_VF_MBOX_SIG(0, 1);
+
+	INIT_DELAYED_WORK(&mbox->mbox_poll_wk.work,
+			  cn23xx_vf_mbox_thread);
+
+	mbox->mbox_poll_wk.ctxptr = mbox;
+
+	oct->mbox[0] = mbox;
+
+	writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+
+	return 0;
+}
+
+static int cn23xx_enable_vf_io_queues(struct octeon_device *oct)
+{
+	u32 q_no;
+
+	for (q_no = 0; q_no < oct->num_iqs; q_no++) {
+		u64 reg_val;
+
+		/* set the corresponding IQ IS_64B bit */
+		if (oct->io_qmask.iq64B & BIT_ULL(q_no)) {
+			reg_val = octeon_read_csr64(
+			    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no));
+			reg_val |= CN23XX_PKT_INPUT_CTL_IS_64B;
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no), reg_val);
+		}
+
+		/* set the corresponding IQ ENB bit */
+		if (oct->io_qmask.iq & BIT_ULL(q_no)) {
+			reg_val = octeon_read_csr64(
+			    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no));
+			reg_val |= CN23XX_PKT_INPUT_CTL_RING_ENB;
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(q_no), reg_val);
+		}
+	}
+	for (q_no = 0; q_no < oct->num_oqs; q_no++) {
+		u32 reg_val;
+
+		/* set the corresponding OQ ENB bit */
+		if (oct->io_qmask.oq & BIT_ULL(q_no)) {
+			reg_val = octeon_read_csr(
+			    oct, CN23XX_VF_SLI_OQ_PKT_CONTROL(q_no));
+			reg_val |= CN23XX_PKT_OUTPUT_CTL_RING_ENB;
+			octeon_write_csr(
+			    oct, CN23XX_VF_SLI_OQ_PKT_CONTROL(q_no), reg_val);
+		}
+	}
+
+	return 0;
+}
+
+static void cn23xx_disable_vf_io_queues(struct octeon_device *oct)
+{
+	u32 num_queues = oct->num_iqs;
+
+	/* per HRM, rings can only be disabled via reset operation,
+	 * NOT via SLI_PKT()_INPUT/OUTPUT_CONTROL[ENB]
+	 */
+	if (num_queues < oct->num_oqs)
+		num_queues = oct->num_oqs;
+
+	cn23xx_vf_reset_io_queues(oct, num_queues);
+}
+
+void cn23xx_vf_ask_pf_to_do_flr(struct octeon_device *oct)
+{
+	struct octeon_mbox_cmd mbox_cmd;
+
+	mbox_cmd.msg.u64 = 0;
+	mbox_cmd.msg.s.type = OCTEON_MBOX_REQUEST;
+	mbox_cmd.msg.s.resp_needed = 0;
+	mbox_cmd.msg.s.cmd = OCTEON_VF_FLR_REQUEST;
+	mbox_cmd.msg.s.len = 1;
+	mbox_cmd.q_no = 0;
+	mbox_cmd.recv_len = 0;
+	mbox_cmd.recv_status = 0;
+	mbox_cmd.fn = NULL;
+	mbox_cmd.fn_arg = NULL;
+
+	octeon_mbox_write(oct, &mbox_cmd);
+}
+EXPORT_SYMBOL_GPL(cn23xx_vf_ask_pf_to_do_flr);
+
+static void octeon_pfvf_hs_callback(struct octeon_device *oct,
+				    struct octeon_mbox_cmd *cmd,
+				    void *arg)
+{
+	u32 major = 0;
+
+	memcpy((uint8_t *)&oct->pfvf_hsword, cmd->msg.s.params,
+	       CN23XX_MAILBOX_MSGPARAM_SIZE);
+	if (cmd->recv_len > 1)  {
+		major = ((struct lio_version *)(cmd->data))->major;
+		major = major << 16;
+	}
+
+	atomic_set((atomic_t *)arg, major | 1);
+}
+
+int cn23xx_octeon_pfvf_handshake(struct octeon_device *oct)
+{
+	struct octeon_mbox_cmd mbox_cmd;
+	u32 q_no, count = 0;
+	atomic_t status;
+	u32 pfmajor;
+	u32 vfmajor;
+	u32 ret;
+
+	/* Sending VF_ACTIVE indication to the PF driver */
+	dev_dbg(&oct->pci_dev->dev, "requesting info from pf\n");
+
+	mbox_cmd.msg.u64 = 0;
+	mbox_cmd.msg.s.type = OCTEON_MBOX_REQUEST;
+	mbox_cmd.msg.s.resp_needed = 1;
+	mbox_cmd.msg.s.cmd = OCTEON_VF_ACTIVE;
+	mbox_cmd.msg.s.len = 2;
+	mbox_cmd.data[0] = 0;
+	((struct lio_version *)&mbox_cmd.data[0])->major =
+						LIQUIDIO_BASE_MAJOR_VERSION;
+	((struct lio_version *)&mbox_cmd.data[0])->minor =
+						LIQUIDIO_BASE_MINOR_VERSION;
+	((struct lio_version *)&mbox_cmd.data[0])->micro =
+						LIQUIDIO_BASE_MICRO_VERSION;
+	mbox_cmd.q_no = 0;
+	mbox_cmd.recv_len = 0;
+	mbox_cmd.recv_status = 0;
+	mbox_cmd.fn = octeon_pfvf_hs_callback;
+	mbox_cmd.fn_arg = &status;
+
+	octeon_mbox_write(oct, &mbox_cmd);
+
+	atomic_set(&status, 0);
+
+	do {
+		schedule_timeout_uninterruptible(1);
+	} while ((!atomic_read(&status)) && (count++ < 100000));
+
+	ret = atomic_read(&status);
+	if (!ret) {
+		dev_err(&oct->pci_dev->dev, "octeon_pfvf_handshake timeout\n");
+		return 1;
+	}
+
+	for (q_no = 0 ; q_no < oct->num_iqs ; q_no++)
+		oct->instr_queue[q_no]->txpciq.s.pkind = oct->pfvf_hsword.pkind;
+
+	vfmajor = LIQUIDIO_BASE_MAJOR_VERSION;
+	pfmajor = ret >> 16;
+	if (pfmajor != vfmajor) {
+		dev_err(&oct->pci_dev->dev,
+			"VF Liquidio driver (major version %d) is not compatible with Liquidio PF driver (major version %d)\n",
+			vfmajor, pfmajor);
+		return 1;
+	}
+
+	dev_dbg(&oct->pci_dev->dev,
+		"VF Liquidio driver (major version %d), Liquidio PF driver (major version %d)\n",
+		vfmajor, pfmajor);
+
+	dev_dbg(&oct->pci_dev->dev, "got data from pf pkind is %d\n",
+		oct->pfvf_hsword.pkind);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(cn23xx_octeon_pfvf_handshake);
+
+static void cn23xx_handle_vf_mbox_intr(struct octeon_ioq_vector *ioq_vector)
+{
+	struct octeon_device *oct = ioq_vector->oct_dev;
+	u64 mbox_int_val;
+
+	if (!ioq_vector->droq_index) {
+		/* read and clear by writing 1 */
+		mbox_int_val = readq(oct->mbox[0]->mbox_int_reg);
+		writeq(mbox_int_val, oct->mbox[0]->mbox_int_reg);
+		if (octeon_mbox_read(oct->mbox[0]))
+			schedule_delayed_work(&oct->mbox[0]->mbox_poll_wk.work,
+					      msecs_to_jiffies(0));
+	}
+}
+
+static u64 cn23xx_vf_msix_interrupt_handler(void *dev)
+{
+	struct octeon_ioq_vector *ioq_vector = (struct octeon_ioq_vector *)dev;
+	struct octeon_device *oct = ioq_vector->oct_dev;
+	struct octeon_droq *droq = oct->droq[ioq_vector->droq_index];
+	u64 pkts_sent;
+	u64 ret = 0;
+
+	dev_dbg(&oct->pci_dev->dev, "In %s octeon_dev @ %p\n", __func__, oct);
+	pkts_sent = readq(droq->pkts_sent_reg);
+
+	/* If our device has interrupted, then proceed. Also check
+	 * for all f's if interrupt was triggered on an error
+	 * and the PCI read fails.
+	 */
+	if (!pkts_sent || (pkts_sent == 0xFFFFFFFFFFFFFFFFULL))
+		return ret;
+
+	/* Write count reg in sli_pkt_cnts to clear these int. */
+	if ((pkts_sent & CN23XX_INTR_PO_INT) ||
+	    (pkts_sent & CN23XX_INTR_PI_INT)) {
+		if (pkts_sent & CN23XX_INTR_PO_INT)
+			ret |= MSIX_PO_INT;
+	}
+
+	if (pkts_sent & CN23XX_INTR_PI_INT)
+		/* We will clear the count when we update the read_index. */
+		ret |= MSIX_PI_INT;
+
+	if (pkts_sent & CN23XX_INTR_MBOX_INT) {
+		cn23xx_handle_vf_mbox_intr(ioq_vector);
+		ret |= MSIX_MBOX_INT;
+	}
+
+	return ret;
+}
+
+static u32 cn23xx_update_read_index(struct octeon_instr_queue *iq)
+{
+	u32 pkt_in_done = readl(iq->inst_cnt_reg);
+	u32 last_done;
+	u32 new_idx;
+
+	last_done = pkt_in_done - iq->pkt_in_done;
+	iq->pkt_in_done = pkt_in_done;
+
+	/* Modulo of the new index with the IQ size will give us
+	 * the new index.  The iq->reset_instr_cnt is always zero for
+	 * cn23xx, so no extra adjustments are needed.
+	 */
+	new_idx = (iq->octeon_read_index +
+		   (u32)(last_done & CN23XX_PKT_IN_DONE_CNT_MASK)) %
+		  iq->max_count;
+
+	return new_idx;
+}
+
+static void cn23xx_enable_vf_interrupt(struct octeon_device *oct, u8 intr_flag)
+{
+	struct octeon_cn23xx_vf *cn23xx = (struct octeon_cn23xx_vf *)oct->chip;
+	u32 q_no, time_threshold;
+
+	if (intr_flag & OCTEON_OUTPUT_INTR) {
+		for (q_no = 0; q_no < oct->num_oqs; q_no++) {
+			/* Set up interrupt packet and time thresholds
+			 * for all the OQs
+			 */
+			time_threshold = cn23xx_vf_get_oq_ticks(
+				oct, (u32)CFG_GET_OQ_INTR_TIME(cn23xx->conf));
+
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_OQ_PKT_INT_LEVELS(q_no),
+			    (CFG_GET_OQ_INTR_PKT(cn23xx->conf) |
+			     ((u64)time_threshold << 32)));
+		}
+	}
+
+	if (intr_flag & OCTEON_INPUT_INTR) {
+		for (q_no = 0; q_no < oct->num_oqs; q_no++) {
+			/* Set CINT_ENB to enable IQ interrupt */
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no),
+			    ((octeon_read_csr64(
+				  oct, CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no)) &
+			      ~CN23XX_PKT_IN_DONE_CNT_MASK) |
+			     CN23XX_INTR_CINT_ENB));
+		}
+	}
+
+	/* Set queue-0 MBOX_ENB to enable VF mailbox interrupt */
+	if (intr_flag & OCTEON_MBOX_INTR) {
+		octeon_write_csr64(
+		    oct, CN23XX_VF_SLI_PKT_MBOX_INT(0),
+		    (octeon_read_csr64(oct, CN23XX_VF_SLI_PKT_MBOX_INT(0)) |
+		     CN23XX_INTR_MBOX_ENB));
+	}
+}
+
+static void cn23xx_disable_vf_interrupt(struct octeon_device *oct, u8 intr_flag)
+{
+	u32 q_no;
+
+	if (intr_flag & OCTEON_OUTPUT_INTR) {
+		for (q_no = 0; q_no < oct->num_oqs; q_no++) {
+			/* Write all 1's in INT_LEVEL reg to disable PO_INT */
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_OQ_PKT_INT_LEVELS(q_no),
+			    0x3fffffffffffff);
+		}
+	}
+	if (intr_flag & OCTEON_INPUT_INTR) {
+		for (q_no = 0; q_no < oct->num_oqs; q_no++) {
+			octeon_write_csr64(
+			    oct, CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no),
+			    (octeon_read_csr64(
+				 oct, CN23XX_VF_SLI_IQ_INSTR_COUNT64(q_no)) &
+			     ~(CN23XX_INTR_CINT_ENB |
+			       CN23XX_PKT_IN_DONE_CNT_MASK)));
+		}
+	}
+
+	if (intr_flag & OCTEON_MBOX_INTR) {
+		octeon_write_csr64(
+		    oct, CN23XX_VF_SLI_PKT_MBOX_INT(0),
+		    (octeon_read_csr64(oct, CN23XX_VF_SLI_PKT_MBOX_INT(0)) &
+		     ~CN23XX_INTR_MBOX_ENB));
+	}
+}
+
+int cn23xx_setup_octeon_vf_device(struct octeon_device *oct)
+{
+	struct octeon_cn23xx_vf *cn23xx = (struct octeon_cn23xx_vf *)oct->chip;
+	u32 rings_per_vf;
+	u64 reg_val;
+
+	if (octeon_map_pci_barx(oct, 0, 0))
+		return 1;
+
+	/* INPUT_CONTROL[RPVF] gives the VF IOq count */
+	reg_val = octeon_read_csr64(oct, CN23XX_VF_SLI_IQ_PKT_CONTROL64(0));
+
+	oct->pf_num = (reg_val >> CN23XX_PKT_INPUT_CTL_PF_NUM_POS) &
+		      CN23XX_PKT_INPUT_CTL_PF_NUM_MASK;
+	oct->vf_num = (reg_val >> CN23XX_PKT_INPUT_CTL_VF_NUM_POS) &
+		      CN23XX_PKT_INPUT_CTL_VF_NUM_MASK;
+
+	reg_val = reg_val >> CN23XX_PKT_INPUT_CTL_RPVF_POS;
+
+	rings_per_vf = reg_val & CN23XX_PKT_INPUT_CTL_RPVF_MASK;
+
+	cn23xx->conf  = oct_get_config_info(oct, LIO_23XX);
+	if (!cn23xx->conf) {
+		dev_err(&oct->pci_dev->dev, "%s No Config found for CN23XX\n",
+			__func__);
+		octeon_unmap_pci_barx(oct, 0);
+		return 1;
+	}
+
+	if (oct->sriov_info.rings_per_vf > rings_per_vf) {
+		dev_warn(&oct->pci_dev->dev,
+			 "num_queues:%d greater than PF configured rings_per_vf:%d. Reducing to %d.\n",
+			 oct->sriov_info.rings_per_vf, rings_per_vf,
+			 rings_per_vf);
+		oct->sriov_info.rings_per_vf = rings_per_vf;
+	} else {
+		if (rings_per_vf > num_present_cpus()) {
+			dev_warn(&oct->pci_dev->dev,
+				 "PF configured rings_per_vf:%d greater than num_cpu:%d. Using rings_per_vf:%d equal to num cpus\n",
+				 rings_per_vf,
+				 num_present_cpus(),
+				 num_present_cpus());
+			oct->sriov_info.rings_per_vf =
+				num_present_cpus();
+		} else {
+			oct->sriov_info.rings_per_vf = rings_per_vf;
+		}
+	}
+
+	oct->fn_list.setup_iq_regs = cn23xx_setup_vf_iq_regs;
+	oct->fn_list.setup_oq_regs = cn23xx_setup_vf_oq_regs;
+	oct->fn_list.setup_mbox = cn23xx_setup_vf_mbox;
+	oct->fn_list.free_mbox = cn23xx_free_vf_mbox;
+
+	oct->fn_list.msix_interrupt_handler = cn23xx_vf_msix_interrupt_handler;
+
+	oct->fn_list.setup_device_regs = cn23xx_setup_vf_device_regs;
+	oct->fn_list.update_iq_read_idx = cn23xx_update_read_index;
+
+	oct->fn_list.enable_interrupt = cn23xx_enable_vf_interrupt;
+	oct->fn_list.disable_interrupt = cn23xx_disable_vf_interrupt;
+
+	oct->fn_list.enable_io_queues = cn23xx_enable_vf_io_queues;
+	oct->fn_list.disable_io_queues = cn23xx_disable_vf_io_queues;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(cn23xx_setup_octeon_vf_device);

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn23xx_vf_device.h

@@ -0,0 +1,46 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file  cn23xx_device.h
+ * \brief Host Driver: Routines that perform CN23XX specific operations.
+ */
+
+#ifndef __CN23XX_VF_DEVICE_H__
+#define __CN23XX_VF_DEVICE_H__
+
+#include "cn23xx_vf_regs.h"
+
+/* Register address and configuration for a CN23XX devices.
+ * If device specific changes need to be made then add a struct to include
+ * device specific fields as shown in the commented section
+ */
+struct octeon_cn23xx_vf {
+	struct octeon_config *conf;
+};
+
+#define BUSY_READING_REG_VF_LOOP_COUNT		10000
+
+#define CN23XX_MAILBOX_MSGPARAM_SIZE		6
+
+void cn23xx_vf_ask_pf_to_do_flr(struct octeon_device *oct);
+
+int cn23xx_octeon_pfvf_handshake(struct octeon_device *oct);
+
+int cn23xx_setup_octeon_vf_device(struct octeon_device *oct);
+
+u32 cn23xx_vf_get_oq_ticks(struct octeon_device *oct, u32 time_intr_in_us);
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn23xx_vf_regs.h

@@ -0,0 +1,274 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file cn23xx_vf_regs.h
+ * \brief Host Driver: Register Address and Register Mask values for
+ * Octeon CN23XX vf functions.
+ */
+
+#ifndef __CN23XX_VF_REGS_H__
+#define __CN23XX_VF_REGS_H__
+
+#define     CN23XX_CONFIG_XPANSION_BAR             0x38
+
+#define     CN23XX_CONFIG_PCIE_CAP                 0x70
+#define     CN23XX_CONFIG_PCIE_DEVCAP              0x74
+#define     CN23XX_CONFIG_PCIE_DEVCTL              0x78
+#define     CN23XX_CONFIG_PCIE_LINKCAP             0x7C
+#define     CN23XX_CONFIG_PCIE_LINKCTL             0x80
+#define     CN23XX_CONFIG_PCIE_SLOTCAP             0x84
+#define     CN23XX_CONFIG_PCIE_SLOTCTL             0x88
+
+#define     CN23XX_CONFIG_PCIE_FLTMSK              0x720
+
+/* The input jabber is used to determine the TSO max size.
+ * Due to H/W limitation, this needs to be reduced to 60000
+ * in order to use H/W TSO and avoid the WQE malformation
+ * PKO_BUG_24989_WQE_LEN
+ */
+#define    CN23XX_DEFAULT_INPUT_JABBER             0xEA60 /*60000*/
+
+/* ##############  BAR0 Registers ################ */
+
+/* Each Input Queue register is at a 16-byte Offset in BAR0 */
+#define    CN23XX_VF_IQ_OFFSET                     0x20000
+
+/*###################### REQUEST QUEUE #########################*/
+
+/* 64 registers for Input Queue Instr Count - SLI_PKT_IN_DONE0_CNTS */
+#define    CN23XX_VF_SLI_IQ_INSTR_COUNT_START64     0x10040
+
+/* 64 registers for Input Queues Start Addr - SLI_PKT0_INSTR_BADDR */
+#define    CN23XX_VF_SLI_IQ_BASE_ADDR_START64       0x10010
+
+/* 64 registers for Input Doorbell - SLI_PKT0_INSTR_BAOFF_DBELL */
+#define    CN23XX_VF_SLI_IQ_DOORBELL_START          0x10020
+
+/* 64 registers for Input Queue size - SLI_PKT0_INSTR_FIFO_RSIZE */
+#define    CN23XX_VF_SLI_IQ_SIZE_START              0x10030
+
+/* 64 registers (64-bit) - ES, RO, NS, Arbitration for Input Queue Data &
+ * gather list fetches. SLI_PKT(0..63)_INPUT_CONTROL.
+ */
+#define    CN23XX_VF_SLI_IQ_PKT_CONTROL_START64     0x10000
+
+/*------- Request Queue Macros ---------*/
+#define CN23XX_VF_SLI_IQ_PKT_CONTROL64(iq)		\
+	(CN23XX_VF_SLI_IQ_PKT_CONTROL_START64 + ((iq) * CN23XX_VF_IQ_OFFSET))
+
+#define CN23XX_VF_SLI_IQ_BASE_ADDR64(iq)		\
+	(CN23XX_VF_SLI_IQ_BASE_ADDR_START64 + ((iq) * CN23XX_VF_IQ_OFFSET))
+
+#define CN23XX_VF_SLI_IQ_SIZE(iq)			\
+	(CN23XX_VF_SLI_IQ_SIZE_START + ((iq) * CN23XX_VF_IQ_OFFSET))
+
+#define CN23XX_VF_SLI_IQ_DOORBELL(iq)			\
+	(CN23XX_VF_SLI_IQ_DOORBELL_START + ((iq) * CN23XX_VF_IQ_OFFSET))
+
+#define CN23XX_VF_SLI_IQ_INSTR_COUNT64(iq)		\
+	(CN23XX_VF_SLI_IQ_INSTR_COUNT_START64 + ((iq) * CN23XX_VF_IQ_OFFSET))
+
+/*------------------ Masks ----------------*/
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM                  BIT_ULL(32)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM                 BIT(29)
+/* Number of instructions to be read in one MAC read request.
+ * setting to Max value(4)
+ */
+#define    CN23XX_PKT_INPUT_CTL_RDSIZE                  (3 << 25)
+#define    CN23XX_PKT_INPUT_CTL_IS_64B                  BIT(24)
+#define    CN23XX_PKT_INPUT_CTL_RST                     BIT(23)
+#define    CN23XX_PKT_INPUT_CTL_QUIET                   BIT(28)
+#define    CN23XX_PKT_INPUT_CTL_RING_ENB                BIT(22)
+#define    CN23XX_PKT_INPUT_CTL_DATA_NS                 BIT(8)
+#define    CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP        BIT(6)
+#define    CN23XX_PKT_INPUT_CTL_DATA_RO                 BIT(5)
+#define    CN23XX_PKT_INPUT_CTL_USE_CSR                 BIT(4)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_NS               BIT(3)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_ES_64B_SWAP      (2)
+#define    CN23XX_PKT_INPUT_CTL_GATHER_RO               (1)
+
+/** Rings per Virtual Function [RO] **/
+#define    CN23XX_PKT_INPUT_CTL_RPVF_MASK               (0x3F)
+#define    CN23XX_PKT_INPUT_CTL_RPVF_POS                (48)
+/* These bits[47:44][RO] give the Physical function number info within the MAC*/
+#define    CN23XX_PKT_INPUT_CTL_PF_NUM_MASK             (0x7)
+#define    CN23XX_PKT_INPUT_CTL_PF_NUM_POS              (45)
+/** These bits[43:32][RO] give the virtual function number info within the PF*/
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM_MASK             (0x1FFF)
+#define    CN23XX_PKT_INPUT_CTL_VF_NUM_POS              (32)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM_MASK            (0x3)
+#define    CN23XX_PKT_INPUT_CTL_MAC_NUM_POS             (29)
+#define    CN23XX_PKT_IN_DONE_WMARK_MASK                (0xFFFFULL)
+#define    CN23XX_PKT_IN_DONE_WMARK_BIT_POS             (32)
+#define    CN23XX_PKT_IN_DONE_CNT_MASK                  (0x00000000FFFFFFFFULL)
+
+#ifdef __LITTLE_ENDIAN_BITFIELD
+#define CN23XX_PKT_INPUT_CTL_MASK			\
+	(CN23XX_PKT_INPUT_CTL_RDSIZE			\
+	 | CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP	\
+	 | CN23XX_PKT_INPUT_CTL_USE_CSR)
+#else
+#define CN23XX_PKT_INPUT_CTL_MASK			\
+	(CN23XX_PKT_INPUT_CTL_RDSIZE			\
+	 | CN23XX_PKT_INPUT_CTL_DATA_ES_64B_SWAP	\
+	 | CN23XX_PKT_INPUT_CTL_USE_CSR			\
+	 | CN23XX_PKT_INPUT_CTL_GATHER_ES_64B_SWAP)
+#endif
+
+/** Masks for SLI_PKT_IN_DONE(0..63)_CNTS Register */
+#define    CN23XX_IN_DONE_CNTS_PI_INT               BIT_ULL(62)
+#define    CN23XX_IN_DONE_CNTS_CINT_ENB             BIT_ULL(48)
+
+/*############################ OUTPUT QUEUE #########################*/
+
+/* 64 registers for Output queue control - SLI_PKT(0..63)_OUTPUT_CONTROL */
+#define    CN23XX_VF_SLI_OQ_PKT_CONTROL_START       0x10050
+
+/* 64 registers for Output queue buffer and info size - SLI_PKT0_OUT_SIZE */
+#define    CN23XX_VF_SLI_OQ0_BUFF_INFO_SIZE         0x10060
+
+/* 64 registers for Output Queue Start Addr - SLI_PKT0_SLIST_BADDR */
+#define    CN23XX_VF_SLI_OQ_BASE_ADDR_START64       0x10070
+
+/* 64 registers for Output Queue Packet Credits - SLI_PKT0_SLIST_BAOFF_DBELL */
+#define    CN23XX_VF_SLI_OQ_PKT_CREDITS_START       0x10080
+
+/* 64 registers for Output Queue size - SLI_PKT0_SLIST_FIFO_RSIZE */
+#define    CN23XX_VF_SLI_OQ_SIZE_START              0x10090
+
+/* 64 registers for Output Queue Packet Count - SLI_PKT0_CNTS */
+#define    CN23XX_VF_SLI_OQ_PKT_SENT_START          0x100B0
+
+/* 64 registers for Output Queue INT Levels - SLI_PKT0_INT_LEVELS */
+#define    CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_START64  0x100A0
+
+/* Each Output Queue register is at a 16-byte Offset in BAR0 */
+#define    CN23XX_VF_OQ_OFFSET                      0x20000
+
+/*------- Output Queue Macros ---------*/
+
+#define CN23XX_VF_SLI_OQ_PKT_CONTROL(oq)		\
+	(CN23XX_VF_SLI_OQ_PKT_CONTROL_START + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_BASE_ADDR64(oq)		\
+	(CN23XX_VF_SLI_OQ_BASE_ADDR_START64 + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_SIZE(oq)			\
+	(CN23XX_VF_SLI_OQ_SIZE_START + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_BUFF_INFO_SIZE(oq)		\
+	(CN23XX_VF_SLI_OQ0_BUFF_INFO_SIZE + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_PKTS_SENT(oq)		\
+	(CN23XX_VF_SLI_OQ_PKT_SENT_START + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_PKTS_CREDIT(oq)		\
+	(CN23XX_VF_SLI_OQ_PKT_CREDITS_START + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_PKT_INT_LEVELS(oq)		\
+	(CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_START64 + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+/* Macro's for accessing CNT and TIME separately from INT_LEVELS */
+#define CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_CNT(oq)	\
+	(CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_START64 + ((oq) * CN23XX_VF_OQ_OFFSET))
+
+#define CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_TIME(oq)	\
+	(CN23XX_VF_SLI_OQ_PKT_INT_LEVELS_START64 +	\
+	 ((oq) * CN23XX_VF_OQ_OFFSET) + 4)
+
+/*------------------ Masks ----------------*/
+#define    CN23XX_PKT_OUTPUT_CTL_TENB                  BIT(13)
+#define    CN23XX_PKT_OUTPUT_CTL_CENB                  BIT(12)
+#define    CN23XX_PKT_OUTPUT_CTL_IPTR                  BIT(11)
+#define    CN23XX_PKT_OUTPUT_CTL_ES                    BIT(9)
+#define    CN23XX_PKT_OUTPUT_CTL_NSR                   BIT(8)
+#define    CN23XX_PKT_OUTPUT_CTL_ROR                   BIT(7)
+#define    CN23XX_PKT_OUTPUT_CTL_DPTR                  BIT(6)
+#define    CN23XX_PKT_OUTPUT_CTL_BMODE                 BIT(5)
+#define    CN23XX_PKT_OUTPUT_CTL_ES_P                  BIT(3)
+#define    CN23XX_PKT_OUTPUT_CTL_NSR_P                 BIT(2)
+#define    CN23XX_PKT_OUTPUT_CTL_ROR_P                 BIT(1)
+#define    CN23XX_PKT_OUTPUT_CTL_RING_ENB              BIT(0)
+
+/*######################### Mailbox Reg Macros ########################*/
+#define    CN23XX_VF_SLI_PKT_MBOX_INT_START            0x10210
+#define    CN23XX_SLI_PKT_PF_VF_MBOX_SIG_START         0x10200
+
+#define    CN23XX_SLI_MBOX_OFFSET                      0x20000
+#define    CN23XX_SLI_MBOX_SIG_IDX_OFFSET              0x8
+
+#define CN23XX_VF_SLI_PKT_MBOX_INT(q)	\
+	(CN23XX_VF_SLI_PKT_MBOX_INT_START + ((q) * CN23XX_SLI_MBOX_OFFSET))
+
+#define CN23XX_SLI_PKT_PF_VF_MBOX_SIG(q, idx)		\
+	(CN23XX_SLI_PKT_PF_VF_MBOX_SIG_START +		\
+	 ((q) * CN23XX_SLI_MBOX_OFFSET +		\
+	  (idx) * CN23XX_SLI_MBOX_SIG_IDX_OFFSET))
+
+/*######################## INTERRUPTS #########################*/
+
+#define    CN23XX_VF_SLI_INT_SUM_START		  0x100D0
+
+#define CN23XX_VF_SLI_INT_SUM(q)			\
+	(CN23XX_VF_SLI_INT_SUM_START + ((q) * CN23XX_VF_IQ_OFFSET))
+
+/*------------------ Interrupt Masks ----------------*/
+
+#define    CN23XX_INTR_PO_INT                   BIT_ULL(63)
+#define    CN23XX_INTR_PI_INT                   BIT_ULL(62)
+#define    CN23XX_INTR_MBOX_INT                 BIT_ULL(61)
+#define    CN23XX_INTR_RESEND                   BIT_ULL(60)
+
+#define    CN23XX_INTR_CINT_ENB                 BIT_ULL(48)
+#define    CN23XX_INTR_MBOX_ENB                 BIT(0)
+
+/*############################ MIO #########################*/
+#define    CN23XX_MIO_PTP_CLOCK_CFG       0x0001070000000f00ULL
+#define    CN23XX_MIO_PTP_CLOCK_LO        0x0001070000000f08ULL
+#define    CN23XX_MIO_PTP_CLOCK_HI        0x0001070000000f10ULL
+#define    CN23XX_MIO_PTP_CLOCK_COMP      0x0001070000000f18ULL
+#define    CN23XX_MIO_PTP_TIMESTAMP       0x0001070000000f20ULL
+#define    CN23XX_MIO_PTP_EVT_CNT         0x0001070000000f28ULL
+#define    CN23XX_MIO_PTP_CKOUT_THRESH_LO 0x0001070000000f30ULL
+#define    CN23XX_MIO_PTP_CKOUT_THRESH_HI 0x0001070000000f38ULL
+#define    CN23XX_MIO_PTP_CKOUT_HI_INCR   0x0001070000000f40ULL
+#define    CN23XX_MIO_PTP_CKOUT_LO_INCR   0x0001070000000f48ULL
+#define    CN23XX_MIO_PTP_PPS_THRESH_LO   0x0001070000000f50ULL
+#define    CN23XX_MIO_PTP_PPS_THRESH_HI   0x0001070000000f58ULL
+#define    CN23XX_MIO_PTP_PPS_HI_INCR     0x0001070000000f60ULL
+#define    CN23XX_MIO_PTP_PPS_LO_INCR     0x0001070000000f68ULL
+
+/*############################ RST #########################*/
+#define    CN23XX_RST_BOOT                0x0001180006001600ULL
+
+/*######################## MSIX TABLE #########################*/
+
+#define    CN23XX_MSIX_TABLE_ADDR_START    0x0
+#define    CN23XX_MSIX_TABLE_DATA_START    0x8
+
+#define    CN23XX_MSIX_TABLE_SIZE          0x10
+#define    CN23XX_MSIX_TABLE_ENTRIES       0x41
+
+#define    CN23XX_MSIX_ENTRY_VECTOR_CTL    BIT_ULL(32)
+
+#define CN23XX_MSIX_TABLE_ADDR(idx)		\
+	(CN23XX_MSIX_TABLE_ADDR_START + ((idx) * CN23XX_MSIX_TABLE_SIZE))
+
+#define CN23XX_MSIX_TABLE_DATA(idx)		\
+	(CN23XX_MSIX_TABLE_DATA_START + ((idx) * CN23XX_MSIX_TABLE_SIZE))
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn66xx_device.c

@@ -0,0 +1,738 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+#include "cn66xx_regs.h"
+#include "cn66xx_device.h"
+
+int lio_cn6xxx_soft_reset(struct octeon_device *oct)
+{
+	octeon_write_csr64(oct, CN6XXX_WIN_WR_MASK_REG, 0xFF);
+
+	dev_dbg(&oct->pci_dev->dev, "BIST enabled for soft reset\n");
+
+	lio_pci_writeq(oct, 1, CN6XXX_CIU_SOFT_BIST);
+	octeon_write_csr64(oct, CN6XXX_SLI_SCRATCH1, 0x1234ULL);
+
+	lio_pci_readq(oct, CN6XXX_CIU_SOFT_RST);
+	lio_pci_writeq(oct, 1, CN6XXX_CIU_SOFT_RST);
+
+	/* Wait for 10ms as Octeon resets. */
+	mdelay(100);
+
+	if (octeon_read_csr64(oct, CN6XXX_SLI_SCRATCH1)) {
+		dev_err(&oct->pci_dev->dev, "Soft reset failed\n");
+		return 1;
+	}
+
+	dev_dbg(&oct->pci_dev->dev, "Reset completed\n");
+	octeon_write_csr64(oct, CN6XXX_WIN_WR_MASK_REG, 0xFF);
+
+	return 0;
+}
+
+void lio_cn6xxx_enable_error_reporting(struct octeon_device *oct)
+{
+	u32 val;
+
+	pci_read_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, &val);
+	if (val & 0x000c0000) {
+		dev_err(&oct->pci_dev->dev, "PCI-E Link error detected: 0x%08x\n",
+			val & 0x000c0000);
+	}
+
+	val |= 0xf;          /* Enable Link error reporting */
+
+	dev_dbg(&oct->pci_dev->dev, "Enabling PCI-E error reporting..\n");
+	pci_write_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, val);
+}
+
+void lio_cn6xxx_setup_pcie_mps(struct octeon_device *oct,
+			       enum octeon_pcie_mps mps)
+{
+	u32 val;
+	u64 r64;
+
+	/* Read config register for MPS */
+	pci_read_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, &val);
+
+	if (mps == PCIE_MPS_DEFAULT) {
+		mps = ((val & (0x7 << 5)) >> 5);
+	} else {
+		val &= ~(0x7 << 5);  /* Turn off any MPS bits */
+		val |= (mps << 5);   /* Set MPS */
+		pci_write_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, val);
+	}
+
+	/* Set MPS in DPI_SLI_PRT0_CFG to the same value. */
+	r64 = lio_pci_readq(oct, CN6XXX_DPI_SLI_PRTX_CFG(oct->pcie_port));
+	r64 |= (mps << 4);
+	lio_pci_writeq(oct, r64, CN6XXX_DPI_SLI_PRTX_CFG(oct->pcie_port));
+}
+
+void lio_cn6xxx_setup_pcie_mrrs(struct octeon_device *oct,
+				enum octeon_pcie_mrrs mrrs)
+{
+	u32 val;
+	u64 r64;
+
+	/* Read config register for MRRS */
+	pci_read_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, &val);
+
+	if (mrrs == PCIE_MRRS_DEFAULT) {
+		mrrs = ((val & (0x7 << 12)) >> 12);
+	} else {
+		val &= ~(0x7 << 12); /* Turn off any MRRS bits */
+		val |= (mrrs << 12); /* Set MRRS */
+		pci_write_config_dword(oct->pci_dev, CN6XXX_PCIE_DEVCTL, val);
+	}
+
+	/* Set MRRS in SLI_S2M_PORT0_CTL to the same value. */
+	r64 = octeon_read_csr64(oct, CN6XXX_SLI_S2M_PORTX_CTL(oct->pcie_port));
+	r64 |= mrrs;
+	octeon_write_csr64(oct, CN6XXX_SLI_S2M_PORTX_CTL(oct->pcie_port), r64);
+
+	/* Set MRRS in DPI_SLI_PRT0_CFG to the same value. */
+	r64 = lio_pci_readq(oct, CN6XXX_DPI_SLI_PRTX_CFG(oct->pcie_port));
+	r64 |= mrrs;
+	lio_pci_writeq(oct, r64, CN6XXX_DPI_SLI_PRTX_CFG(oct->pcie_port));
+}
+
+u32 lio_cn6xxx_coprocessor_clock(struct octeon_device *oct)
+{
+	/* Bits 29:24 of MIO_RST_BOOT holds the ref. clock multiplier
+	 * for SLI.
+	 */
+	return ((lio_pci_readq(oct, CN6XXX_MIO_RST_BOOT) >> 24) & 0x3f) * 50;
+}
+
+u32 lio_cn6xxx_get_oq_ticks(struct octeon_device *oct,
+			    u32 time_intr_in_us)
+{
+	/* This gives the SLI clock per microsec */
+	u32 oqticks_per_us = lio_cn6xxx_coprocessor_clock(oct);
+
+	/* core clock per us / oq ticks will be fractional. TO avoid that
+	 * we use the method below.
+	 */
+
+	/* This gives the clock cycles per millisecond */
+	oqticks_per_us *= 1000;
+
+	/* This gives the oq ticks (1024 core clock cycles) per millisecond */
+	oqticks_per_us /= 1024;
+
+	/* time_intr is in microseconds. The next 2 steps gives the oq ticks
+	 * corressponding to time_intr.
+	 */
+	oqticks_per_us *= time_intr_in_us;
+	oqticks_per_us /= 1000;
+
+	return oqticks_per_us;
+}
+
+void lio_cn6xxx_setup_global_input_regs(struct octeon_device *oct)
+{
+	/* Select Round-Robin Arb, ES, RO, NS for Input Queues */
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_INPUT_CONTROL,
+			 CN6XXX_INPUT_CTL_MASK);
+
+	/* Instruction Read Size - Max 4 instructions per PCIE Read */
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_INSTR_RD_SIZE,
+			   0xFFFFFFFFFFFFFFFFULL);
+
+	/* Select PCIE Port for all Input rings. */
+	octeon_write_csr64(oct, CN6XXX_SLI_IN_PCIE_PORT,
+			   (oct->pcie_port * 0x5555555555555555ULL));
+}
+
+static void lio_cn66xx_setup_pkt_ctl_regs(struct octeon_device *oct)
+{
+	u64 pktctl;
+
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+
+	pktctl = octeon_read_csr64(oct, CN6XXX_SLI_PKT_CTL);
+
+	/* 66XX SPECIFIC */
+	if (CFG_GET_OQ_MAX_Q(cn6xxx->conf) <= 4)
+		/* Disable RING_EN if only upto 4 rings are used. */
+		pktctl &= ~(1 << 4);
+	else
+		pktctl |= (1 << 4);
+
+	if (CFG_GET_IS_SLI_BP_ON(cn6xxx->conf))
+		pktctl |= 0xF;
+	else
+		/* Disable per-port backpressure. */
+		pktctl &= ~0xF;
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_CTL, pktctl);
+}
+
+void lio_cn6xxx_setup_global_output_regs(struct octeon_device *oct)
+{
+	u32 time_threshold;
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+
+	/* / Select PCI-E Port for all Output queues */
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_PCIE_PORT64,
+			   (oct->pcie_port * 0x5555555555555555ULL));
+
+	if (CFG_GET_IS_SLI_BP_ON(cn6xxx->conf)) {
+		octeon_write_csr64(oct, CN6XXX_SLI_OQ_WMARK, 32);
+	} else {
+		/* / Set Output queue watermark to 0 to disable backpressure */
+		octeon_write_csr64(oct, CN6XXX_SLI_OQ_WMARK, 0);
+	}
+
+	/* / Select Packet count instead of bytes for SLI_PKTi_CNTS[CNT] */
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_OUT_BMODE, 0);
+
+	/* Select ES, RO, NS setting from register for Output Queue Packet
+	 * Address
+	 */
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_DPADDR, 0xFFFFFFFF);
+
+	/* No Relaxed Ordering, No Snoop, 64-bit swap for Output
+	 * Queue ScatterList
+	 */
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_SLIST_ROR, 0);
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_SLIST_NS, 0);
+
+	/* / ENDIAN_SPECIFIC CHANGES - 0 works for LE. */
+#ifdef __BIG_ENDIAN_BITFIELD
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_SLIST_ES64,
+			   0x5555555555555555ULL);
+#else
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_SLIST_ES64, 0ULL);
+#endif
+
+	/* / No Relaxed Ordering, No Snoop, 64-bit swap for Output Queue Data */
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_DATA_OUT_ROR, 0);
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_DATA_OUT_NS, 0);
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_DATA_OUT_ES64,
+			   0x5555555555555555ULL);
+
+	/* / Set up interrupt packet and time threshold */
+	octeon_write_csr(oct, CN6XXX_SLI_OQ_INT_LEVEL_PKTS,
+			 (u32)CFG_GET_OQ_INTR_PKT(cn6xxx->conf));
+	time_threshold =
+		lio_cn6xxx_get_oq_ticks(oct, (u32)
+					CFG_GET_OQ_INTR_TIME(cn6xxx->conf));
+
+	octeon_write_csr(oct, CN6XXX_SLI_OQ_INT_LEVEL_TIME, time_threshold);
+}
+
+static int lio_cn6xxx_setup_device_regs(struct octeon_device *oct)
+{
+	lio_cn6xxx_setup_pcie_mps(oct, PCIE_MPS_DEFAULT);
+	lio_cn6xxx_setup_pcie_mrrs(oct, PCIE_MRRS_512B);
+	lio_cn6xxx_enable_error_reporting(oct);
+
+	lio_cn6xxx_setup_global_input_regs(oct);
+	lio_cn66xx_setup_pkt_ctl_regs(oct);
+	lio_cn6xxx_setup_global_output_regs(oct);
+
+	/* Default error timeout value should be 0x200000 to avoid host hang
+	 * when reads invalid register
+	 */
+	octeon_write_csr64(oct, CN6XXX_SLI_WINDOW_CTL, 0x200000ULL);
+	return 0;
+}
+
+void lio_cn6xxx_setup_iq_regs(struct octeon_device *oct, u32 iq_no)
+{
+	struct octeon_instr_queue *iq = oct->instr_queue[iq_no];
+
+	octeon_write_csr64(oct, CN6XXX_SLI_IQ_PKT_INSTR_HDR64(iq_no), 0);
+
+	/* Write the start of the input queue's ring and its size  */
+	octeon_write_csr64(oct, CN6XXX_SLI_IQ_BASE_ADDR64(iq_no),
+			   iq->base_addr_dma);
+	octeon_write_csr(oct, CN6XXX_SLI_IQ_SIZE(iq_no), iq->max_count);
+
+	/* Remember the doorbell & instruction count register addr for this
+	 * queue
+	 */
+	iq->doorbell_reg = oct->mmio[0].hw_addr + CN6XXX_SLI_IQ_DOORBELL(iq_no);
+	iq->inst_cnt_reg = oct->mmio[0].hw_addr
+			   + CN6XXX_SLI_IQ_INSTR_COUNT(iq_no);
+	dev_dbg(&oct->pci_dev->dev, "InstQ[%d]:dbell reg @ 0x%p instcnt_reg @ 0x%p\n",
+		iq_no, iq->doorbell_reg, iq->inst_cnt_reg);
+
+	/* Store the current instruction counter
+	 * (used in flush_iq calculation)
+	 */
+	iq->reset_instr_cnt = readl(iq->inst_cnt_reg);
+}
+
+static void lio_cn66xx_setup_iq_regs(struct octeon_device *oct, u32 iq_no)
+{
+	lio_cn6xxx_setup_iq_regs(oct, iq_no);
+
+	/* Backpressure for this queue - WMARK set to all F's. This effectively
+	 * disables the backpressure mechanism.
+	 */
+	octeon_write_csr64(oct, CN66XX_SLI_IQ_BP64(iq_no),
+			   (0xFFFFFFFFULL << 32));
+}
+
+void lio_cn6xxx_setup_oq_regs(struct octeon_device *oct, u32 oq_no)
+{
+	u32 intr;
+	struct octeon_droq *droq = oct->droq[oq_no];
+
+	octeon_write_csr64(oct, CN6XXX_SLI_OQ_BASE_ADDR64(oq_no),
+			   droq->desc_ring_dma);
+	octeon_write_csr(oct, CN6XXX_SLI_OQ_SIZE(oq_no), droq->max_count);
+
+	octeon_write_csr(oct, CN6XXX_SLI_OQ_BUFF_INFO_SIZE(oq_no),
+			 droq->buffer_size);
+
+	/* Get the mapped address of the pkt_sent and pkts_credit regs */
+	droq->pkts_sent_reg =
+		oct->mmio[0].hw_addr + CN6XXX_SLI_OQ_PKTS_SENT(oq_no);
+	droq->pkts_credit_reg =
+		oct->mmio[0].hw_addr + CN6XXX_SLI_OQ_PKTS_CREDIT(oq_no);
+
+	/* Enable this output queue to generate Packet Timer Interrupt */
+	intr = octeon_read_csr(oct, CN6XXX_SLI_PKT_TIME_INT_ENB);
+	intr |= (1 << oq_no);
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_TIME_INT_ENB, intr);
+
+	/* Enable this output queue to generate Packet Timer Interrupt */
+	intr = octeon_read_csr(oct, CN6XXX_SLI_PKT_CNT_INT_ENB);
+	intr |= (1 << oq_no);
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_CNT_INT_ENB, intr);
+}
+
+int lio_cn6xxx_enable_io_queues(struct octeon_device *oct)
+{
+	u32 mask;
+
+	mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_INSTR_SIZE);
+	mask |= oct->io_qmask.iq64B;
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_INSTR_SIZE, mask);
+
+	mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_INSTR_ENB);
+	mask |= oct->io_qmask.iq;
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_INSTR_ENB, mask);
+
+	mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_OUT_ENB);
+	mask |= oct->io_qmask.oq;
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_OUT_ENB, mask);
+
+	return 0;
+}
+
+void lio_cn6xxx_disable_io_queues(struct octeon_device *oct)
+{
+	int i;
+	u32 mask, loop = HZ;
+	u32 d32;
+
+	/* Reset the Enable bits for Input Queues. */
+	mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_INSTR_ENB);
+	mask ^= oct->io_qmask.iq;
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_INSTR_ENB, mask);
+
+	/* Wait until hardware indicates that the queues are out of reset. */
+	mask = (u32)oct->io_qmask.iq;
+	d32 = octeon_read_csr(oct, CN6XXX_SLI_PORT_IN_RST_IQ);
+	while (((d32 & mask) != mask) && loop--) {
+		d32 = octeon_read_csr(oct, CN6XXX_SLI_PORT_IN_RST_IQ);
+		schedule_timeout_uninterruptible(1);
+	}
+
+	/* Reset the doorbell register for each Input queue. */
+	for (i = 0; i < MAX_OCTEON_INSTR_QUEUES(oct); i++) {
+		if (!(oct->io_qmask.iq & BIT_ULL(i)))
+			continue;
+		octeon_write_csr(oct, CN6XXX_SLI_IQ_DOORBELL(i), 0xFFFFFFFF);
+		d32 = octeon_read_csr(oct, CN6XXX_SLI_IQ_DOORBELL(i));
+	}
+
+	/* Reset the Enable bits for Output Queues. */
+	mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_OUT_ENB);
+	mask ^= oct->io_qmask.oq;
+	octeon_write_csr(oct, CN6XXX_SLI_PKT_OUT_ENB, mask);
+
+	/* Wait until hardware indicates that the queues are out of reset. */
+	loop = HZ;
+	mask = (u32)oct->io_qmask.oq;
+	d32 = octeon_read_csr(oct, CN6XXX_SLI_PORT_IN_RST_OQ);
+	while (((d32 & mask) != mask) && loop--) {
+		d32 = octeon_read_csr(oct, CN6XXX_SLI_PORT_IN_RST_OQ);
+		schedule_timeout_uninterruptible(1);
+	}
+	;
+
+	/* Reset the doorbell register for each Output queue. */
+	for (i = 0; i < MAX_OCTEON_OUTPUT_QUEUES(oct); i++) {
+		if (!(oct->io_qmask.oq & BIT_ULL(i)))
+			continue;
+		octeon_write_csr(oct, CN6XXX_SLI_OQ_PKTS_CREDIT(i), 0xFFFFFFFF);
+		d32 = octeon_read_csr(oct, CN6XXX_SLI_OQ_PKTS_CREDIT(i));
+
+		d32 = octeon_read_csr(oct, CN6XXX_SLI_OQ_PKTS_SENT(i));
+		octeon_write_csr(oct, CN6XXX_SLI_OQ_PKTS_SENT(i), d32);
+	}
+
+	d32 = octeon_read_csr(oct, CN6XXX_SLI_PKT_CNT_INT);
+	if (d32)
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_CNT_INT, d32);
+
+	d32 = octeon_read_csr(oct, CN6XXX_SLI_PKT_TIME_INT);
+	if (d32)
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_TIME_INT, d32);
+}
+
+void
+lio_cn6xxx_bar1_idx_setup(struct octeon_device *oct,
+			  u64 core_addr,
+			  u32 idx,
+			  int valid)
+{
+	u64 bar1;
+
+	if (valid == 0) {
+		bar1 = lio_pci_readq(oct, CN6XXX_BAR1_REG(idx, oct->pcie_port));
+		lio_pci_writeq(oct, (bar1 & 0xFFFFFFFEULL),
+			       CN6XXX_BAR1_REG(idx, oct->pcie_port));
+		bar1 = lio_pci_readq(oct, CN6XXX_BAR1_REG(idx, oct->pcie_port));
+		return;
+	}
+
+	/* Bits 17:4 of the PCI_BAR1_INDEXx stores bits 35:22 of
+	 * the Core Addr
+	 */
+	lio_pci_writeq(oct, (((core_addr >> 22) << 4) | PCI_BAR1_MASK),
+		       CN6XXX_BAR1_REG(idx, oct->pcie_port));
+
+	bar1 = lio_pci_readq(oct, CN6XXX_BAR1_REG(idx, oct->pcie_port));
+}
+
+void lio_cn6xxx_bar1_idx_write(struct octeon_device *oct,
+			       u32 idx,
+			       u32 mask)
+{
+	lio_pci_writeq(oct, mask, CN6XXX_BAR1_REG(idx, oct->pcie_port));
+}
+
+u32 lio_cn6xxx_bar1_idx_read(struct octeon_device *oct, u32 idx)
+{
+	return (u32)lio_pci_readq(oct, CN6XXX_BAR1_REG(idx, oct->pcie_port));
+}
+
+u32
+lio_cn6xxx_update_read_index(struct octeon_instr_queue *iq)
+{
+	u32 new_idx = readl(iq->inst_cnt_reg);
+
+	/* The new instr cnt reg is a 32-bit counter that can roll over. We have
+	 * noted the counter's initial value at init time into
+	 * reset_instr_cnt
+	 */
+	if (iq->reset_instr_cnt < new_idx)
+		new_idx -= iq->reset_instr_cnt;
+	else
+		new_idx += (0xffffffff - iq->reset_instr_cnt) + 1;
+
+	/* Modulo of the new index with the IQ size will give us
+	 * the new index.
+	 */
+	new_idx %= iq->max_count;
+
+	return new_idx;
+}
+
+void lio_cn6xxx_enable_interrupt(struct octeon_device *oct,
+				 u8 unused __attribute__((unused)))
+{
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+	u64 mask = cn6xxx->intr_mask64 | CN6XXX_INTR_DMA0_FORCE;
+
+	/* Enable Interrupt */
+	writeq(mask, cn6xxx->intr_enb_reg64);
+}
+
+void lio_cn6xxx_disable_interrupt(struct octeon_device *oct,
+				  u8 unused __attribute__((unused)))
+{
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+
+	/* Disable Interrupts */
+	writeq(0, cn6xxx->intr_enb_reg64);
+}
+
+static void lio_cn6xxx_get_pcie_qlmport(struct octeon_device *oct)
+{
+	/* CN63xx Pass2 and newer parts implements the SLI_MAC_NUMBER register
+	 * to determine the PCIE port #
+	 */
+	oct->pcie_port = octeon_read_csr(oct, CN6XXX_SLI_MAC_NUMBER) & 0xff;
+
+	dev_dbg(&oct->pci_dev->dev, "Using PCIE Port %d\n", oct->pcie_port);
+}
+
+static void
+lio_cn6xxx_process_pcie_error_intr(struct octeon_device *oct, u64 intr64)
+{
+	dev_err(&oct->pci_dev->dev, "Error Intr: 0x%016llx\n",
+		CVM_CAST64(intr64));
+}
+
+static int lio_cn6xxx_process_droq_intr_regs(struct octeon_device *oct)
+{
+	struct octeon_droq *droq;
+	int oq_no;
+	u32 pkt_count, droq_time_mask, droq_mask, droq_int_enb;
+	u32 droq_cnt_enb, droq_cnt_mask;
+
+	droq_cnt_enb = octeon_read_csr(oct, CN6XXX_SLI_PKT_CNT_INT_ENB);
+	droq_cnt_mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_CNT_INT);
+	droq_mask = droq_cnt_mask & droq_cnt_enb;
+
+	droq_time_mask = octeon_read_csr(oct, CN6XXX_SLI_PKT_TIME_INT);
+	droq_int_enb = octeon_read_csr(oct, CN6XXX_SLI_PKT_TIME_INT_ENB);
+	droq_mask |= (droq_time_mask & droq_int_enb);
+
+	droq_mask &= oct->io_qmask.oq;
+
+	oct->droq_intr = 0;
+
+	for (oq_no = 0; oq_no < MAX_OCTEON_OUTPUT_QUEUES(oct); oq_no++) {
+		if (!(droq_mask & BIT_ULL(oq_no)))
+			continue;
+
+		droq = oct->droq[oq_no];
+		pkt_count = octeon_droq_check_hw_for_pkts(droq);
+		if (pkt_count) {
+			oct->droq_intr |= BIT_ULL(oq_no);
+			if (droq->ops.poll_mode) {
+				u32 value;
+				u32 reg;
+
+				struct octeon_cn6xxx *cn6xxx =
+					(struct octeon_cn6xxx *)oct->chip;
+
+				/* disable interrupts for this droq */
+				spin_lock
+					(&cn6xxx->lock_for_droq_int_enb_reg);
+				reg = CN6XXX_SLI_PKT_TIME_INT_ENB;
+				value = octeon_read_csr(oct, reg);
+				value &= ~(1 << oq_no);
+				octeon_write_csr(oct, reg, value);
+				reg = CN6XXX_SLI_PKT_CNT_INT_ENB;
+				value = octeon_read_csr(oct, reg);
+				value &= ~(1 << oq_no);
+				octeon_write_csr(oct, reg, value);
+
+				spin_unlock(&cn6xxx->lock_for_droq_int_enb_reg);
+			}
+		}
+	}
+
+	droq_time_mask &= oct->io_qmask.oq;
+	droq_cnt_mask &= oct->io_qmask.oq;
+
+	/* Reset the PKT_CNT/TIME_INT registers. */
+	if (droq_time_mask)
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_TIME_INT, droq_time_mask);
+
+	if (droq_cnt_mask)      /* reset PKT_CNT register:66xx */
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_CNT_INT, droq_cnt_mask);
+
+	return 0;
+}
+
+irqreturn_t lio_cn6xxx_process_interrupt_regs(void *dev)
+{
+	struct octeon_device *oct = (struct octeon_device *)dev;
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+	u64 intr64;
+
+	intr64 = readq(cn6xxx->intr_sum_reg64);
+
+	/* If our device has interrupted, then proceed.
+	 * Also check for all f's if interrupt was triggered on an error
+	 * and the PCI read fails.
+	 */
+	if (!intr64 || (intr64 == 0xFFFFFFFFFFFFFFFFULL))
+		return IRQ_NONE;
+
+	oct->int_status = 0;
+
+	if (intr64 & CN6XXX_INTR_ERR)
+		lio_cn6xxx_process_pcie_error_intr(oct, intr64);
+
+	if (intr64 & CN6XXX_INTR_PKT_DATA) {
+		lio_cn6xxx_process_droq_intr_regs(oct);
+		oct->int_status |= OCT_DEV_INTR_PKT_DATA;
+	}
+
+	if (intr64 & CN6XXX_INTR_DMA0_FORCE)
+		oct->int_status |= OCT_DEV_INTR_DMA0_FORCE;
+
+	if (intr64 & CN6XXX_INTR_DMA1_FORCE)
+		oct->int_status |= OCT_DEV_INTR_DMA1_FORCE;
+
+	/* Clear the current interrupts */
+	writeq(intr64, cn6xxx->intr_sum_reg64);
+
+	return IRQ_HANDLED;
+}
+
+void lio_cn6xxx_setup_reg_address(struct octeon_device *oct,
+				  void *chip,
+				  struct octeon_reg_list *reg_list)
+{
+	u8 __iomem *bar0_pciaddr = oct->mmio[0].hw_addr;
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)chip;
+
+	reg_list->pci_win_wr_addr_hi =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_ADDR_HI);
+	reg_list->pci_win_wr_addr_lo =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_ADDR_LO);
+	reg_list->pci_win_wr_addr =
+		(u64 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_ADDR64);
+
+	reg_list->pci_win_rd_addr_hi =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_ADDR_HI);
+	reg_list->pci_win_rd_addr_lo =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_ADDR_LO);
+	reg_list->pci_win_rd_addr =
+		(u64 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_ADDR64);
+
+	reg_list->pci_win_wr_data_hi =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_DATA_HI);
+	reg_list->pci_win_wr_data_lo =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_DATA_LO);
+	reg_list->pci_win_wr_data =
+		(u64 __iomem *)(bar0_pciaddr + CN6XXX_WIN_WR_DATA64);
+
+	reg_list->pci_win_rd_data_hi =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_DATA_HI);
+	reg_list->pci_win_rd_data_lo =
+		(u32 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_DATA_LO);
+	reg_list->pci_win_rd_data =
+		(u64 __iomem *)(bar0_pciaddr + CN6XXX_WIN_RD_DATA64);
+
+	lio_cn6xxx_get_pcie_qlmport(oct);
+
+	cn6xxx->intr_sum_reg64 = bar0_pciaddr + CN6XXX_SLI_INT_SUM64;
+	cn6xxx->intr_mask64 = CN6XXX_INTR_MASK;
+	cn6xxx->intr_enb_reg64 =
+		bar0_pciaddr + CN6XXX_SLI_INT_ENB64(oct->pcie_port);
+}
+
+int lio_setup_cn66xx_octeon_device(struct octeon_device *oct)
+{
+	struct octeon_cn6xxx *cn6xxx = (struct octeon_cn6xxx *)oct->chip;
+
+	if (octeon_map_pci_barx(oct, 0, 0))
+		return 1;
+
+	if (octeon_map_pci_barx(oct, 1, MAX_BAR1_IOREMAP_SIZE)) {
+		dev_err(&oct->pci_dev->dev, "%s CN66XX BAR1 map failed\n",
+			__func__);
+		octeon_unmap_pci_barx(oct, 0);
+		return 1;
+	}
+
+	spin_lock_init(&cn6xxx->lock_for_droq_int_enb_reg);
+
+	oct->fn_list.setup_iq_regs = lio_cn66xx_setup_iq_regs;
+	oct->fn_list.setup_oq_regs = lio_cn6xxx_setup_oq_regs;
+
+	oct->fn_list.soft_reset = lio_cn6xxx_soft_reset;
+	oct->fn_list.setup_device_regs = lio_cn6xxx_setup_device_regs;
+	oct->fn_list.update_iq_read_idx = lio_cn6xxx_update_read_index;
+
+	oct->fn_list.bar1_idx_setup = lio_cn6xxx_bar1_idx_setup;
+	oct->fn_list.bar1_idx_write = lio_cn6xxx_bar1_idx_write;
+	oct->fn_list.bar1_idx_read = lio_cn6xxx_bar1_idx_read;
+
+	oct->fn_list.process_interrupt_regs = lio_cn6xxx_process_interrupt_regs;
+	oct->fn_list.enable_interrupt = lio_cn6xxx_enable_interrupt;
+	oct->fn_list.disable_interrupt = lio_cn6xxx_disable_interrupt;
+
+	oct->fn_list.enable_io_queues = lio_cn6xxx_enable_io_queues;
+	oct->fn_list.disable_io_queues = lio_cn6xxx_disable_io_queues;
+
+	lio_cn6xxx_setup_reg_address(oct, oct->chip, &oct->reg_list);
+
+	cn6xxx->conf = (struct octeon_config *)
+		       oct_get_config_info(oct, LIO_210SV);
+	if (!cn6xxx->conf) {
+		dev_err(&oct->pci_dev->dev, "%s No Config found for CN66XX\n",
+			__func__);
+		octeon_unmap_pci_barx(oct, 0);
+		octeon_unmap_pci_barx(oct, 1);
+		return 1;
+	}
+
+	oct->coproc_clock_rate = 1000000ULL * lio_cn6xxx_coprocessor_clock(oct);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(lio_setup_cn66xx_octeon_device);
+
+int lio_validate_cn6xxx_config_info(struct octeon_device *oct,
+				    struct octeon_config *conf6xxx)
+{
+	if (CFG_GET_IQ_MAX_Q(conf6xxx) > CN6XXX_MAX_INPUT_QUEUES) {
+		dev_err(&oct->pci_dev->dev, "%s: Num IQ (%d) exceeds Max (%d)\n",
+			__func__, CFG_GET_IQ_MAX_Q(conf6xxx),
+			CN6XXX_MAX_INPUT_QUEUES);
+		return 1;
+	}
+
+	if (CFG_GET_OQ_MAX_Q(conf6xxx) > CN6XXX_MAX_OUTPUT_QUEUES) {
+		dev_err(&oct->pci_dev->dev, "%s: Num OQ (%d) exceeds Max (%d)\n",
+			__func__, CFG_GET_OQ_MAX_Q(conf6xxx),
+			CN6XXX_MAX_OUTPUT_QUEUES);
+		return 1;
+	}
+
+	if (CFG_GET_IQ_INSTR_TYPE(conf6xxx) != OCTEON_32BYTE_INSTR &&
+	    CFG_GET_IQ_INSTR_TYPE(conf6xxx) != OCTEON_64BYTE_INSTR) {
+		dev_err(&oct->pci_dev->dev, "%s: Invalid instr type for IQ\n",
+			__func__);
+		return 1;
+	}
+	if (!CFG_GET_OQ_REFILL_THRESHOLD(conf6xxx)) {
+		dev_err(&oct->pci_dev->dev, "%s: Invalid parameter for OQ\n",
+			__func__);
+		return 1;
+	}
+
+	if (!(CFG_GET_OQ_INTR_TIME(conf6xxx))) {
+		dev_err(&oct->pci_dev->dev, "%s: No Time Interrupt for OQ\n",
+			__func__);
+		return 1;
+	}
+
+	return 0;
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn66xx_device.h

@@ -0,0 +1,97 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file  cn66xx_device.h
+ *  \brief Host Driver: Routines that perform CN66XX specific operations.
+ */
+
+#ifndef __CN66XX_DEVICE_H__
+#define  __CN66XX_DEVICE_H__
+
+/* Register address and configuration for a CN6XXX devices.
+ * If device specific changes need to be made then add a struct to include
+ * device specific fields as shown in the commented section
+ */
+struct octeon_cn6xxx {
+	/** PCI interrupt summary register */
+	u8 __iomem *intr_sum_reg64;
+
+	/** PCI interrupt enable register */
+	u8 __iomem *intr_enb_reg64;
+
+	/** The PCI interrupt mask used by interrupt handler */
+	u64 intr_mask64;
+
+	struct octeon_config *conf;
+
+	/* Example additional fields - not used currently
+	 *  struct {
+	 *  }cn6xyz;
+	 */
+
+	/* For the purpose of atomic access to interrupt enable reg */
+	spinlock_t lock_for_droq_int_enb_reg;
+
+};
+
+enum octeon_pcie_mps {
+	PCIE_MPS_DEFAULT = -1,	/* Use the default setup by BIOS */
+	PCIE_MPS_128B = 0,
+	PCIE_MPS_256B = 1
+};
+
+enum octeon_pcie_mrrs {
+	PCIE_MRRS_DEFAULT = -1,	/* Use the default setup by BIOS */
+	PCIE_MRRS_128B = 0,
+	PCIE_MRRS_256B = 1,
+	PCIE_MRRS_512B = 2,
+	PCIE_MRRS_1024B = 3,
+	PCIE_MRRS_2048B = 4,
+	PCIE_MRRS_4096B = 5
+};
+
+/* Common functions for 66xx and 68xx */
+int lio_cn6xxx_soft_reset(struct octeon_device *oct);
+void lio_cn6xxx_enable_error_reporting(struct octeon_device *oct);
+void lio_cn6xxx_setup_pcie_mps(struct octeon_device *oct,
+			       enum octeon_pcie_mps mps);
+void lio_cn6xxx_setup_pcie_mrrs(struct octeon_device *oct,
+				enum octeon_pcie_mrrs mrrs);
+void lio_cn6xxx_setup_global_input_regs(struct octeon_device *oct);
+void lio_cn6xxx_setup_global_output_regs(struct octeon_device *oct);
+void lio_cn6xxx_setup_iq_regs(struct octeon_device *oct, u32 iq_no);
+void lio_cn6xxx_setup_oq_regs(struct octeon_device *oct, u32 oq_no);
+int lio_cn6xxx_enable_io_queues(struct octeon_device *oct);
+void lio_cn6xxx_disable_io_queues(struct octeon_device *oct);
+irqreturn_t lio_cn6xxx_process_interrupt_regs(void *dev);
+void lio_cn6xxx_bar1_idx_setup(struct octeon_device *oct, u64 core_addr,
+			       u32 idx, int valid);
+void lio_cn6xxx_bar1_idx_write(struct octeon_device *oct, u32 idx, u32 mask);
+u32 lio_cn6xxx_bar1_idx_read(struct octeon_device *oct, u32 idx);
+u32
+lio_cn6xxx_update_read_index(struct octeon_instr_queue *iq);
+void lio_cn6xxx_enable_interrupt(struct octeon_device *oct, u8 unused);
+void lio_cn6xxx_disable_interrupt(struct octeon_device *oct, u8 unused);
+void lio_cn6xxx_setup_reg_address(struct octeon_device *oct, void *chip,
+				  struct octeon_reg_list *reg_list);
+u32 lio_cn6xxx_coprocessor_clock(struct octeon_device *oct);
+u32 lio_cn6xxx_get_oq_ticks(struct octeon_device *oct, u32 time_intr_in_us);
+int lio_setup_cn66xx_octeon_device(struct octeon_device *oct);
+int lio_validate_cn6xxx_config_info(struct octeon_device *oct,
+				    struct octeon_config *conf6xxx);
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn66xx_regs.h

@@ -0,0 +1,530 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file cn66xx_regs.h
+ *  \brief Host Driver: Register Address and Register Mask values for
+ *  Octeon CN66XX devices.
+ */
+
+#ifndef __CN66XX_REGS_H__
+#define __CN66XX_REGS_H__
+
+#define     CN6XXX_XPANSION_BAR             0x30
+
+#define     CN6XXX_MSI_CAP                  0x50
+#define     CN6XXX_MSI_ADDR_LO              0x54
+#define     CN6XXX_MSI_ADDR_HI              0x58
+#define     CN6XXX_MSI_DATA                 0x5C
+
+#define     CN6XXX_PCIE_CAP                 0x70
+#define     CN6XXX_PCIE_DEVCAP              0x74
+#define     CN6XXX_PCIE_DEVCTL              0x78
+#define     CN6XXX_PCIE_LINKCAP             0x7C
+#define     CN6XXX_PCIE_LINKCTL             0x80
+#define     CN6XXX_PCIE_SLOTCAP             0x84
+#define     CN6XXX_PCIE_SLOTCTL             0x88
+
+#define     CN6XXX_PCIE_ENH_CAP             0x100
+#define     CN6XXX_PCIE_UNCORR_ERR_STATUS   0x104
+#define     CN6XXX_PCIE_UNCORR_ERR_MASK     0x108
+#define     CN6XXX_PCIE_UNCORR_ERR          0x10C
+#define     CN6XXX_PCIE_CORR_ERR_STATUS     0x110
+#define     CN6XXX_PCIE_CORR_ERR_MASK       0x114
+#define     CN6XXX_PCIE_ADV_ERR_CAP         0x118
+
+#define     CN6XXX_PCIE_ACK_REPLAY_TIMER    0x700
+#define     CN6XXX_PCIE_OTHER_MSG           0x704
+#define     CN6XXX_PCIE_PORT_FORCE_LINK     0x708
+#define     CN6XXX_PCIE_ACK_FREQ            0x70C
+#define     CN6XXX_PCIE_PORT_LINK_CTL       0x710
+#define     CN6XXX_PCIE_LANE_SKEW           0x714
+#define     CN6XXX_PCIE_SYM_NUM             0x718
+#define     CN6XXX_PCIE_FLTMSK              0x720
+
+/* ##############  BAR0 Registers ################  */
+
+#define    CN6XXX_SLI_CTL_PORT0                    0x0050
+#define    CN6XXX_SLI_CTL_PORT1                    0x0060
+
+#define    CN6XXX_SLI_WINDOW_CTL                   0x02E0
+#define    CN6XXX_SLI_DBG_DATA                     0x0310
+#define    CN6XXX_SLI_SCRATCH1                     0x03C0
+#define    CN6XXX_SLI_SCRATCH2                     0x03D0
+#define    CN6XXX_SLI_CTL_STATUS                   0x0570
+
+#define    CN6XXX_WIN_WR_ADDR_LO                   0x0000
+#define    CN6XXX_WIN_WR_ADDR_HI                   0x0004
+#define    CN6XXX_WIN_WR_ADDR64                    CN6XXX_WIN_WR_ADDR_LO
+
+#define    CN6XXX_WIN_RD_ADDR_LO                   0x0010
+#define    CN6XXX_WIN_RD_ADDR_HI                   0x0014
+#define    CN6XXX_WIN_RD_ADDR64                    CN6XXX_WIN_RD_ADDR_LO
+
+#define    CN6XXX_WIN_WR_DATA_LO                   0x0020
+#define    CN6XXX_WIN_WR_DATA_HI                   0x0024
+#define    CN6XXX_WIN_WR_DATA64                    CN6XXX_WIN_WR_DATA_LO
+
+#define    CN6XXX_WIN_RD_DATA_LO                   0x0040
+#define    CN6XXX_WIN_RD_DATA_HI                   0x0044
+#define    CN6XXX_WIN_RD_DATA64                    CN6XXX_WIN_RD_DATA_LO
+
+#define    CN6XXX_WIN_WR_MASK_LO                   0x0030
+#define    CN6XXX_WIN_WR_MASK_HI                   0x0034
+#define    CN6XXX_WIN_WR_MASK_REG                  CN6XXX_WIN_WR_MASK_LO
+
+/* 1 register (32-bit) to enable Input queues */
+#define    CN6XXX_SLI_PKT_INSTR_ENB               0x1000
+
+/* 1 register (32-bit) to enable Output queues */
+#define    CN6XXX_SLI_PKT_OUT_ENB                 0x1010
+
+/* 1 register (32-bit) to determine whether Output queues are in reset. */
+#define    CN6XXX_SLI_PORT_IN_RST_OQ              0x11F0
+
+/* 1 register (32-bit) to determine whether Input queues are in reset. */
+#define    CN6XXX_SLI_PORT_IN_RST_IQ              0x11F4
+
+/*###################### REQUEST QUEUE #########################*/
+
+/* 1 register (32-bit) - instr. size of each input queue. */
+#define    CN6XXX_SLI_PKT_INSTR_SIZE             0x1020
+
+/* 32 registers for Input Queue Instr Count - SLI_PKT_IN_DONE0_CNTS */
+#define    CN6XXX_SLI_IQ_INSTR_COUNT_START       0x2000
+
+/* 32 registers for Input Queue Start Addr - SLI_PKT0_INSTR_BADDR */
+#define    CN6XXX_SLI_IQ_BASE_ADDR_START64       0x2800
+
+/* 32 registers for Input Doorbell - SLI_PKT0_INSTR_BAOFF_DBELL */
+#define    CN6XXX_SLI_IQ_DOORBELL_START          0x2C00
+
+/* 32 registers for Input Queue size - SLI_PKT0_INSTR_FIFO_RSIZE */
+#define    CN6XXX_SLI_IQ_SIZE_START              0x3000
+
+/* 32 registers for Instruction Header Options - SLI_PKT0_INSTR_HEADER */
+#define    CN6XXX_SLI_IQ_PKT_INSTR_HDR_START64   0x3400
+
+/* 1 register (64-bit) - Back Pressure for each input queue - SLI_PKT0_IN_BP */
+#define    CN66XX_SLI_INPUT_BP_START64           0x3800
+
+/* Each Input Queue register is at a 16-byte Offset in BAR0 */
+#define    CN6XXX_IQ_OFFSET                      0x10
+
+/* 1 register (32-bit) - ES, RO, NS, Arbitration for Input Queue Data &
+ * gather list fetches. SLI_PKT_INPUT_CONTROL.
+ */
+#define    CN6XXX_SLI_PKT_INPUT_CONTROL          0x1170
+
+/* 1 register (64-bit) - Number of instructions to read at one time
+ * - 2 bits for each input ring. SLI_PKT_INSTR_RD_SIZE.
+ */
+#define    CN6XXX_SLI_PKT_INSTR_RD_SIZE          0x11A0
+
+/* 1 register (64-bit) - Assign Input ring to MAC port
+ * - 2 bits for each input ring. SLI_PKT_IN_PCIE_PORT.
+ */
+#define    CN6XXX_SLI_IN_PCIE_PORT               0x11B0
+
+/*------- Request Queue Macros ---------*/
+#define    CN6XXX_SLI_IQ_BASE_ADDR64(iq)          \
+	(CN6XXX_SLI_IQ_BASE_ADDR_START64 + ((iq) * CN6XXX_IQ_OFFSET))
+
+#define    CN6XXX_SLI_IQ_SIZE(iq)                 \
+	(CN6XXX_SLI_IQ_SIZE_START + ((iq) * CN6XXX_IQ_OFFSET))
+
+#define    CN6XXX_SLI_IQ_PKT_INSTR_HDR64(iq)      \
+	(CN6XXX_SLI_IQ_PKT_INSTR_HDR_START64 + ((iq) * CN6XXX_IQ_OFFSET))
+
+#define    CN6XXX_SLI_IQ_DOORBELL(iq)             \
+	(CN6XXX_SLI_IQ_DOORBELL_START + ((iq) * CN6XXX_IQ_OFFSET))
+
+#define    CN6XXX_SLI_IQ_INSTR_COUNT(iq)          \
+	(CN6XXX_SLI_IQ_INSTR_COUNT_START + ((iq) * CN6XXX_IQ_OFFSET))
+
+#define    CN66XX_SLI_IQ_BP64(iq)                 \
+	(CN66XX_SLI_INPUT_BP_START64 + ((iq) * CN6XXX_IQ_OFFSET))
+
+/*------------------ Masks ----------------*/
+#define    CN6XXX_INPUT_CTL_ROUND_ROBIN_ARB         BIT(22)
+#define    CN6XXX_INPUT_CTL_DATA_NS                 BIT(8)
+#define    CN6XXX_INPUT_CTL_DATA_ES_64B_SWAP        BIT(6)
+#define    CN6XXX_INPUT_CTL_DATA_RO                 BIT(5)
+#define    CN6XXX_INPUT_CTL_USE_CSR                 BIT(4)
+#define    CN6XXX_INPUT_CTL_GATHER_NS               BIT(3)
+#define    CN6XXX_INPUT_CTL_GATHER_ES_64B_SWAP      BIT(2)
+#define    CN6XXX_INPUT_CTL_GATHER_RO               BIT(1)
+
+#ifdef __BIG_ENDIAN_BITFIELD
+#define    CN6XXX_INPUT_CTL_MASK                    \
+	(CN6XXX_INPUT_CTL_DATA_ES_64B_SWAP      \
+	  | CN6XXX_INPUT_CTL_USE_CSR              \
+	  | CN6XXX_INPUT_CTL_GATHER_ES_64B_SWAP)
+#else
+#define    CN6XXX_INPUT_CTL_MASK                    \
+	(CN6XXX_INPUT_CTL_DATA_ES_64B_SWAP     \
+	  | CN6XXX_INPUT_CTL_USE_CSR)
+#endif
+
+/*############################ OUTPUT QUEUE #########################*/
+
+/* 32 registers for Output queue buffer and info size - SLI_PKT0_OUT_SIZE */
+#define    CN6XXX_SLI_OQ0_BUFF_INFO_SIZE         0x0C00
+
+/* 32 registers for Output Queue Start Addr - SLI_PKT0_SLIST_BADDR */
+#define    CN6XXX_SLI_OQ_BASE_ADDR_START64       0x1400
+
+/* 32 registers for Output Queue Packet Credits - SLI_PKT0_SLIST_BAOFF_DBELL */
+#define    CN6XXX_SLI_OQ_PKT_CREDITS_START       0x1800
+
+/* 32 registers for Output Queue size - SLI_PKT0_SLIST_FIFO_RSIZE */
+#define    CN6XXX_SLI_OQ_SIZE_START              0x1C00
+
+/* 32 registers for Output Queue Packet Count - SLI_PKT0_CNTS */
+#define    CN6XXX_SLI_OQ_PKT_SENT_START          0x2400
+
+/* Each Output Queue register is at a 16-byte Offset in BAR0 */
+#define    CN6XXX_OQ_OFFSET                      0x10
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - Relaxed Ordering setting for reading Output Queues descriptors
+ * - SLI_PKT_SLIST_ROR
+ */
+#define    CN6XXX_SLI_PKT_SLIST_ROR              0x1030
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - No Snoop mode for reading Output Queues descriptors
+ * - SLI_PKT_SLIST_NS
+ */
+#define    CN6XXX_SLI_PKT_SLIST_NS               0x1040
+
+/* 1 register (64-bit) - 2 bits for each output queue
+ * - Endian-Swap mode for reading Output Queue descriptors
+ * - SLI_PKT_SLIST_ES
+ */
+#define    CN6XXX_SLI_PKT_SLIST_ES64             0x1050
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - InfoPtr mode for Output Queues.
+ * - SLI_PKT_IPTR
+ */
+#define    CN6XXX_SLI_PKT_IPTR                   0x1070
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - DPTR format selector for Output queues.
+ * - SLI_PKT_DPADDR
+ */
+#define    CN6XXX_SLI_PKT_DPADDR                 0x1080
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - Relaxed Ordering setting for reading Output Queues data
+ * - SLI_PKT_DATA_OUT_ROR
+ */
+#define    CN6XXX_SLI_PKT_DATA_OUT_ROR           0x1090
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - No Snoop mode for reading Output Queues data
+ * - SLI_PKT_DATA_OUT_NS
+ */
+#define    CN6XXX_SLI_PKT_DATA_OUT_NS            0x10A0
+
+/* 1 register (64-bit)  - 2 bits for each output queue
+ * - Endian-Swap mode for reading Output Queue data
+ * - SLI_PKT_DATA_OUT_ES
+ */
+#define    CN6XXX_SLI_PKT_DATA_OUT_ES64          0x10B0
+
+/* 1 register (32-bit) - 1 bit for each output queue
+ * - Controls whether SLI_PKTn_CNTS is incremented for bytes or for packets.
+ * - SLI_PKT_OUT_BMODE
+ */
+#define    CN6XXX_SLI_PKT_OUT_BMODE              0x10D0
+
+/* 1 register (64-bit) - 2 bits for each output queue
+ * - Assign PCIE port for Output queues
+ * - SLI_PKT_PCIE_PORT.
+ */
+#define    CN6XXX_SLI_PKT_PCIE_PORT64            0x10E0
+
+/* 1 (64-bit) register for Output Queue Packet Count Interrupt Threshold
+ * & Time Threshold. The same setting applies to all 32 queues.
+ * The register is defined as a 64-bit registers, but we use the
+ * 32-bit offsets to define distinct addresses.
+ */
+#define    CN6XXX_SLI_OQ_INT_LEVEL_PKTS          0x1120
+#define    CN6XXX_SLI_OQ_INT_LEVEL_TIME          0x1124
+
+/* 1 (64-bit register) for Output Queue backpressure across all rings. */
+#define    CN6XXX_SLI_OQ_WMARK                   0x1180
+
+/* 1 register to control output queue global backpressure & ring enable. */
+#define    CN6XXX_SLI_PKT_CTL                    0x1220
+
+/*------- Output Queue Macros ---------*/
+#define    CN6XXX_SLI_OQ_BASE_ADDR64(oq)          \
+	(CN6XXX_SLI_OQ_BASE_ADDR_START64 + ((oq) * CN6XXX_OQ_OFFSET))
+
+#define    CN6XXX_SLI_OQ_SIZE(oq)                 \
+	(CN6XXX_SLI_OQ_SIZE_START + ((oq) * CN6XXX_OQ_OFFSET))
+
+#define    CN6XXX_SLI_OQ_BUFF_INFO_SIZE(oq)                 \
+	(CN6XXX_SLI_OQ0_BUFF_INFO_SIZE + ((oq) * CN6XXX_OQ_OFFSET))
+
+#define    CN6XXX_SLI_OQ_PKTS_SENT(oq)            \
+	(CN6XXX_SLI_OQ_PKT_SENT_START + ((oq) * CN6XXX_OQ_OFFSET))
+
+#define    CN6XXX_SLI_OQ_PKTS_CREDIT(oq)          \
+	(CN6XXX_SLI_OQ_PKT_CREDITS_START + ((oq) * CN6XXX_OQ_OFFSET))
+
+/*######################### DMA Counters #########################*/
+
+/* 2 registers (64-bit) - DMA Count - 1 for each DMA counter 0/1. */
+#define    CN6XXX_DMA_CNT_START                   0x0400
+
+/* 2 registers (64-bit) - DMA Timer 0/1, contains DMA timer values
+ * SLI_DMA_0_TIM
+ */
+#define    CN6XXX_DMA_TIM_START                   0x0420
+
+/* 2 registers (64-bit) - DMA count & Time Interrupt threshold -
+ * SLI_DMA_0_INT_LEVEL
+ */
+#define    CN6XXX_DMA_INT_LEVEL_START             0x03E0
+
+/* Each DMA register is at a 16-byte Offset in BAR0 */
+#define    CN6XXX_DMA_OFFSET                      0x10
+
+/*---------- DMA Counter Macros ---------*/
+#define    CN6XXX_DMA_CNT(dq)                      \
+	(CN6XXX_DMA_CNT_START + ((dq) * CN6XXX_DMA_OFFSET))
+
+#define    CN6XXX_DMA_INT_LEVEL(dq)                \
+	(CN6XXX_DMA_INT_LEVEL_START + ((dq) * CN6XXX_DMA_OFFSET))
+
+#define    CN6XXX_DMA_PKT_INT_LEVEL(dq)            \
+	(CN6XXX_DMA_INT_LEVEL_START + ((dq) * CN6XXX_DMA_OFFSET))
+
+#define    CN6XXX_DMA_TIME_INT_LEVEL(dq)           \
+	(CN6XXX_DMA_INT_LEVEL_START + 4 + ((dq) * CN6XXX_DMA_OFFSET))
+
+#define    CN6XXX_DMA_TIM(dq)                      \
+	(CN6XXX_DMA_TIM_START + ((dq) * CN6XXX_DMA_OFFSET))
+
+/*######################## INTERRUPTS #########################*/
+
+/* 1 register (64-bit) for Interrupt Summary */
+#define    CN6XXX_SLI_INT_SUM64                  0x0330
+
+/* 1 register (64-bit) for Interrupt Enable */
+#define    CN6XXX_SLI_INT_ENB64_PORT0            0x0340
+#define    CN6XXX_SLI_INT_ENB64_PORT1            0x0350
+
+/* 1 register (32-bit) to enable Output Queue Packet/Byte Count Interrupt */
+#define    CN6XXX_SLI_PKT_CNT_INT_ENB            0x1150
+
+/* 1 register (32-bit) to enable Output Queue Packet Timer Interrupt */
+#define    CN6XXX_SLI_PKT_TIME_INT_ENB           0x1160
+
+/* 1 register (32-bit) to indicate which Output Queue reached pkt threshold */
+#define    CN6XXX_SLI_PKT_CNT_INT                0x1130
+
+/* 1 register (32-bit) to indicate which Output Queue reached time threshold */
+#define    CN6XXX_SLI_PKT_TIME_INT               0x1140
+
+/*------------------ Interrupt Masks ----------------*/
+
+#define    CN6XXX_INTR_RML_TIMEOUT_ERR           BIT(1)
+#define    CN6XXX_INTR_BAR0_RW_TIMEOUT_ERR       BIT(2)
+#define    CN6XXX_INTR_IO2BIG_ERR                BIT(3)
+#define    CN6XXX_INTR_PKT_COUNT                 BIT(4)
+#define    CN6XXX_INTR_PKT_TIME                  BIT(5)
+#define    CN6XXX_INTR_M0UPB0_ERR                BIT(8)
+#define    CN6XXX_INTR_M0UPWI_ERR                BIT(9)
+#define    CN6XXX_INTR_M0UNB0_ERR                BIT(10)
+#define    CN6XXX_INTR_M0UNWI_ERR                BIT(11)
+#define    CN6XXX_INTR_M1UPB0_ERR                BIT(12)
+#define    CN6XXX_INTR_M1UPWI_ERR                BIT(13)
+#define    CN6XXX_INTR_M1UNB0_ERR                BIT(14)
+#define    CN6XXX_INTR_M1UNWI_ERR                BIT(15)
+#define    CN6XXX_INTR_MIO_INT0                  BIT(16)
+#define    CN6XXX_INTR_MIO_INT1                  BIT(17)
+#define    CN6XXX_INTR_MAC_INT0                  BIT(18)
+#define    CN6XXX_INTR_MAC_INT1                  BIT(19)
+
+#define    CN6XXX_INTR_DMA0_FORCE                BIT_ULL(32)
+#define    CN6XXX_INTR_DMA1_FORCE                BIT_ULL(33)
+#define    CN6XXX_INTR_DMA0_COUNT                BIT_ULL(34)
+#define    CN6XXX_INTR_DMA1_COUNT                BIT_ULL(35)
+#define    CN6XXX_INTR_DMA0_TIME                 BIT_ULL(36)
+#define    CN6XXX_INTR_DMA1_TIME                 BIT_ULL(37)
+#define    CN6XXX_INTR_INSTR_DB_OF_ERR           BIT_ULL(48)
+#define    CN6XXX_INTR_SLIST_DB_OF_ERR           BIT_ULL(49)
+#define    CN6XXX_INTR_POUT_ERR                  BIT_ULL(50)
+#define    CN6XXX_INTR_PIN_BP_ERR                BIT_ULL(51)
+#define    CN6XXX_INTR_PGL_ERR                   BIT_ULL(52)
+#define    CN6XXX_INTR_PDI_ERR                   BIT_ULL(53)
+#define    CN6XXX_INTR_POP_ERR                   BIT_ULL(54)
+#define    CN6XXX_INTR_PINS_ERR                  BIT_ULL(55)
+#define    CN6XXX_INTR_SPRT0_ERR                 BIT_ULL(56)
+#define    CN6XXX_INTR_SPRT1_ERR                 BIT_ULL(57)
+#define    CN6XXX_INTR_ILL_PAD_ERR               BIT_ULL(60)
+
+#define    CN6XXX_INTR_DMA0_DATA                 (CN6XXX_INTR_DMA0_TIME)
+
+#define    CN6XXX_INTR_DMA1_DATA                 (CN6XXX_INTR_DMA1_TIME)
+
+#define    CN6XXX_INTR_DMA_DATA                  \
+	(CN6XXX_INTR_DMA0_DATA | CN6XXX_INTR_DMA1_DATA)
+
+#define    CN6XXX_INTR_PKT_DATA                  (CN6XXX_INTR_PKT_TIME | \
+						  CN6XXX_INTR_PKT_COUNT)
+
+/* Sum of interrupts for all PCI-Express Data Interrupts */
+#define    CN6XXX_INTR_PCIE_DATA                 \
+	(CN6XXX_INTR_DMA_DATA | CN6XXX_INTR_PKT_DATA)
+
+#define    CN6XXX_INTR_MIO                       \
+	(CN6XXX_INTR_MIO_INT0 | CN6XXX_INTR_MIO_INT1)
+
+#define    CN6XXX_INTR_MAC                       \
+	(CN6XXX_INTR_MAC_INT0 | CN6XXX_INTR_MAC_INT1)
+
+/* Sum of interrupts for error events */
+#define    CN6XXX_INTR_ERR                       \
+	(CN6XXX_INTR_BAR0_RW_TIMEOUT_ERR    \
+	   | CN6XXX_INTR_IO2BIG_ERR             \
+	   | CN6XXX_INTR_M0UPB0_ERR             \
+	   | CN6XXX_INTR_M0UPWI_ERR             \
+	   | CN6XXX_INTR_M0UNB0_ERR             \
+	   | CN6XXX_INTR_M0UNWI_ERR             \
+	   | CN6XXX_INTR_M1UPB0_ERR             \
+	   | CN6XXX_INTR_M1UPWI_ERR             \
+	   | CN6XXX_INTR_M1UNB0_ERR             \
+	   | CN6XXX_INTR_M1UNWI_ERR             \
+	   | CN6XXX_INTR_INSTR_DB_OF_ERR        \
+	   | CN6XXX_INTR_SLIST_DB_OF_ERR        \
+	   | CN6XXX_INTR_POUT_ERR               \
+	   | CN6XXX_INTR_PIN_BP_ERR             \
+	   | CN6XXX_INTR_PGL_ERR                \
+	   | CN6XXX_INTR_PDI_ERR                \
+	   | CN6XXX_INTR_POP_ERR                \
+	   | CN6XXX_INTR_PINS_ERR               \
+	   | CN6XXX_INTR_SPRT0_ERR              \
+	   | CN6XXX_INTR_SPRT1_ERR              \
+	   | CN6XXX_INTR_ILL_PAD_ERR)
+
+/* Programmed Mask for Interrupt Sum */
+#define    CN6XXX_INTR_MASK                      \
+	(CN6XXX_INTR_PCIE_DATA              \
+	   | CN6XXX_INTR_DMA0_FORCE             \
+	   | CN6XXX_INTR_DMA1_FORCE             \
+	   | CN6XXX_INTR_MIO                    \
+	   | CN6XXX_INTR_MAC                    \
+	   | CN6XXX_INTR_ERR)
+
+#define    CN6XXX_SLI_S2M_PORT0_CTL              0x3D80
+#define    CN6XXX_SLI_S2M_PORT1_CTL              0x3D90
+#define    CN6XXX_SLI_S2M_PORTX_CTL(port)        \
+	(CN6XXX_SLI_S2M_PORT0_CTL + ((port) * 0x10))
+
+#define    CN6XXX_SLI_INT_ENB64(port)            \
+	(CN6XXX_SLI_INT_ENB64_PORT0 + ((port) * 0x10))
+
+#define    CN6XXX_SLI_MAC_NUMBER                 0x3E00
+
+/* CN6XXX BAR1 Index registers. */
+#define    CN6XXX_PEM_BAR1_INDEX000                0x00011800C00000A8ULL
+#define    CN6XXX_PEM_OFFSET                       0x0000000001000000ULL
+
+#define    CN6XXX_BAR1_INDEX_START                 CN6XXX_PEM_BAR1_INDEX000
+#define    CN6XXX_PCI_BAR1_OFFSET                  0x8
+
+#define    CN6XXX_BAR1_REG(idx, port) \
+		(CN6XXX_BAR1_INDEX_START + ((port) * CN6XXX_PEM_OFFSET) + \
+		(CN6XXX_PCI_BAR1_OFFSET * (idx)))
+
+/*############################ DPI #########################*/
+
+#define    CN6XXX_DPI_CTL                 0x0001df0000000040ULL
+
+#define    CN6XXX_DPI_DMA_CONTROL         0x0001df0000000048ULL
+
+#define    CN6XXX_DPI_REQ_GBL_ENB         0x0001df0000000050ULL
+
+#define    CN6XXX_DPI_REQ_ERR_RSP         0x0001df0000000058ULL
+
+#define    CN6XXX_DPI_REQ_ERR_RST         0x0001df0000000060ULL
+
+#define    CN6XXX_DPI_DMA_ENG0_ENB        0x0001df0000000080ULL
+
+#define    CN6XXX_DPI_DMA_ENG_ENB(q_no)   \
+	(CN6XXX_DPI_DMA_ENG0_ENB + ((q_no) * 8))
+
+#define    CN6XXX_DPI_DMA_ENG0_BUF        0x0001df0000000880ULL
+
+#define    CN6XXX_DPI_DMA_ENG_BUF(q_no)   \
+	(CN6XXX_DPI_DMA_ENG0_BUF + ((q_no) * 8))
+
+#define    CN6XXX_DPI_SLI_PRT0_CFG        0x0001df0000000900ULL
+#define    CN6XXX_DPI_SLI_PRT1_CFG        0x0001df0000000908ULL
+#define    CN6XXX_DPI_SLI_PRTX_CFG(port)        \
+	(CN6XXX_DPI_SLI_PRT0_CFG + ((port) * 0x10))
+
+#define    CN6XXX_DPI_DMA_COMMIT_MODE     BIT_ULL(58)
+#define    CN6XXX_DPI_DMA_PKT_HP          BIT_ULL(57)
+#define    CN6XXX_DPI_DMA_PKT_EN          BIT_ULL(56)
+#define    CN6XXX_DPI_DMA_O_ES            BIT_ULL(15)
+#define    CN6XXX_DPI_DMA_O_MODE          BIT_ULL(14)
+
+#define    CN6XXX_DPI_DMA_CTL_MASK             \
+	(CN6XXX_DPI_DMA_COMMIT_MODE    |    \
+	 CN6XXX_DPI_DMA_PKT_HP         |    \
+	 CN6XXX_DPI_DMA_PKT_EN         |    \
+	 CN6XXX_DPI_DMA_O_ES           |    \
+	 CN6XXX_DPI_DMA_O_MODE)
+
+/*############################ CIU #########################*/
+
+#define    CN6XXX_CIU_SOFT_BIST           0x0001070000000738ULL
+#define    CN6XXX_CIU_SOFT_RST            0x0001070000000740ULL
+
+/*############################ MIO #########################*/
+#define    CN6XXX_MIO_PTP_CLOCK_CFG       0x0001070000000f00ULL
+#define    CN6XXX_MIO_PTP_CLOCK_LO        0x0001070000000f08ULL
+#define    CN6XXX_MIO_PTP_CLOCK_HI        0x0001070000000f10ULL
+#define    CN6XXX_MIO_PTP_CLOCK_COMP      0x0001070000000f18ULL
+#define    CN6XXX_MIO_PTP_TIMESTAMP       0x0001070000000f20ULL
+#define    CN6XXX_MIO_PTP_EVT_CNT         0x0001070000000f28ULL
+#define    CN6XXX_MIO_PTP_CKOUT_THRESH_LO 0x0001070000000f30ULL
+#define    CN6XXX_MIO_PTP_CKOUT_THRESH_HI 0x0001070000000f38ULL
+#define    CN6XXX_MIO_PTP_CKOUT_HI_INCR   0x0001070000000f40ULL
+#define    CN6XXX_MIO_PTP_CKOUT_LO_INCR   0x0001070000000f48ULL
+#define    CN6XXX_MIO_PTP_PPS_THRESH_LO   0x0001070000000f50ULL
+#define    CN6XXX_MIO_PTP_PPS_THRESH_HI   0x0001070000000f58ULL
+#define    CN6XXX_MIO_PTP_PPS_HI_INCR     0x0001070000000f60ULL
+#define    CN6XXX_MIO_PTP_PPS_LO_INCR     0x0001070000000f68ULL
+
+#define    CN6XXX_MIO_QLM4_CFG            0x00011800000015B0ULL
+#define    CN6XXX_MIO_RST_BOOT            0x0001180000001600ULL
+
+#define    CN6XXX_MIO_QLM_CFG_MASK        0x7
+
+/*############################ LMC #########################*/
+
+#define    CN6XXX_LMC0_RESET_CTL               0x0001180088000180ULL
+#define    CN6XXX_LMC0_RESET_CTL_DDR3RST_MASK  0x0000000000000001ULL
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn68xx_device.c

@@ -0,0 +1,184 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+#include "cn66xx_regs.h"
+#include "cn66xx_device.h"
+#include "cn68xx_device.h"
+#include "cn68xx_regs.h"
+
+static void lio_cn68xx_set_dpi_regs(struct octeon_device *oct)
+{
+	u32 i;
+	u32 fifo_sizes[6] = { 3, 3, 1, 1, 1, 8 };
+
+	lio_pci_writeq(oct, CN6XXX_DPI_DMA_CTL_MASK, CN6XXX_DPI_DMA_CONTROL);
+	dev_dbg(&oct->pci_dev->dev, "DPI_DMA_CONTROL: 0x%016llx\n",
+		lio_pci_readq(oct, CN6XXX_DPI_DMA_CONTROL));
+
+	for (i = 0; i < 6; i++) {
+		/* Prevent service of instruction queue for all DMA engines
+		 * Engine 5 will remain 0. Engines 0 - 4 will be setup by
+		 * core.
+		 */
+		lio_pci_writeq(oct, 0, CN6XXX_DPI_DMA_ENG_ENB(i));
+		lio_pci_writeq(oct, fifo_sizes[i], CN6XXX_DPI_DMA_ENG_BUF(i));
+		dev_dbg(&oct->pci_dev->dev, "DPI_ENG_BUF%d: 0x%016llx\n", i,
+			lio_pci_readq(oct, CN6XXX_DPI_DMA_ENG_BUF(i)));
+	}
+
+	/* DPI_SLI_PRT_CFG has MPS and MRRS settings that will be set
+	 * separately.
+	 */
+
+	lio_pci_writeq(oct, 1, CN6XXX_DPI_CTL);
+	dev_dbg(&oct->pci_dev->dev, "DPI_CTL: 0x%016llx\n",
+		lio_pci_readq(oct, CN6XXX_DPI_CTL));
+}
+
+static int lio_cn68xx_soft_reset(struct octeon_device *oct)
+{
+	lio_cn6xxx_soft_reset(oct);
+	lio_cn68xx_set_dpi_regs(oct);
+
+	return 0;
+}
+
+static void lio_cn68xx_setup_pkt_ctl_regs(struct octeon_device *oct)
+{
+	struct octeon_cn6xxx *cn68xx = (struct octeon_cn6xxx *)oct->chip;
+	u64 pktctl, tx_pipe, max_oqs;
+
+	pktctl = octeon_read_csr64(oct, CN6XXX_SLI_PKT_CTL);
+
+	/* 68XX specific */
+	max_oqs = CFG_GET_OQ_MAX_Q(CHIP_CONF(oct, cn6xxx));
+	tx_pipe  = octeon_read_csr64(oct, CN68XX_SLI_TX_PIPE);
+	tx_pipe &= 0xffffffffff00ffffULL; /* clear out NUMP field */
+	tx_pipe |= max_oqs << 16; /* put max_oqs in NUMP field */
+	octeon_write_csr64(oct, CN68XX_SLI_TX_PIPE, tx_pipe);
+
+	if (CFG_GET_IS_SLI_BP_ON(cn68xx->conf))
+		pktctl |= 0xF;
+	else
+		/* Disable per-port backpressure. */
+		pktctl &= ~0xF;
+	octeon_write_csr64(oct, CN6XXX_SLI_PKT_CTL, pktctl);
+}
+
+static int lio_cn68xx_setup_device_regs(struct octeon_device *oct)
+{
+	lio_cn6xxx_setup_pcie_mps(oct, PCIE_MPS_DEFAULT);
+	lio_cn6xxx_setup_pcie_mrrs(oct, PCIE_MRRS_256B);
+	lio_cn6xxx_enable_error_reporting(oct);
+
+	lio_cn6xxx_setup_global_input_regs(oct);
+	lio_cn68xx_setup_pkt_ctl_regs(oct);
+	lio_cn6xxx_setup_global_output_regs(oct);
+
+	/* Default error timeout value should be 0x200000 to avoid host hang
+	 * when reads invalid register
+	 */
+	octeon_write_csr64(oct, CN6XXX_SLI_WINDOW_CTL, 0x200000ULL);
+
+	return 0;
+}
+
+static inline void lio_cn68xx_vendor_message_fix(struct octeon_device *oct)
+{
+	u32 val = 0;
+
+	/* Set M_VEND1_DRP and M_VEND0_DRP bits */
+	pci_read_config_dword(oct->pci_dev, CN6XXX_PCIE_FLTMSK, &val);
+	val |= 0x3;
+	pci_write_config_dword(oct->pci_dev, CN6XXX_PCIE_FLTMSK, val);
+}
+
+static int lio_is_210nv(struct octeon_device *oct)
+{
+	u64 mio_qlm4_cfg = lio_pci_readq(oct, CN6XXX_MIO_QLM4_CFG);
+
+	return ((mio_qlm4_cfg & CN6XXX_MIO_QLM_CFG_MASK) == 0);
+}
+
+int lio_setup_cn68xx_octeon_device(struct octeon_device *oct)
+{
+	struct octeon_cn6xxx *cn68xx = (struct octeon_cn6xxx *)oct->chip;
+	u16 card_type = LIO_410NV;
+
+	if (octeon_map_pci_barx(oct, 0, 0))
+		return 1;
+
+	if (octeon_map_pci_barx(oct, 1, MAX_BAR1_IOREMAP_SIZE)) {
+		dev_err(&oct->pci_dev->dev, "%s CN68XX BAR1 map failed\n",
+			__func__);
+		octeon_unmap_pci_barx(oct, 0);
+		return 1;
+	}
+
+	spin_lock_init(&cn68xx->lock_for_droq_int_enb_reg);
+
+	oct->fn_list.setup_iq_regs = lio_cn6xxx_setup_iq_regs;
+	oct->fn_list.setup_oq_regs = lio_cn6xxx_setup_oq_regs;
+
+	oct->fn_list.process_interrupt_regs = lio_cn6xxx_process_interrupt_regs;
+	oct->fn_list.soft_reset = lio_cn68xx_soft_reset;
+	oct->fn_list.setup_device_regs = lio_cn68xx_setup_device_regs;
+	oct->fn_list.update_iq_read_idx = lio_cn6xxx_update_read_index;
+
+	oct->fn_list.bar1_idx_setup = lio_cn6xxx_bar1_idx_setup;
+	oct->fn_list.bar1_idx_write = lio_cn6xxx_bar1_idx_write;
+	oct->fn_list.bar1_idx_read = lio_cn6xxx_bar1_idx_read;
+
+	oct->fn_list.enable_interrupt = lio_cn6xxx_enable_interrupt;
+	oct->fn_list.disable_interrupt = lio_cn6xxx_disable_interrupt;
+
+	oct->fn_list.enable_io_queues = lio_cn6xxx_enable_io_queues;
+	oct->fn_list.disable_io_queues = lio_cn6xxx_disable_io_queues;
+
+	lio_cn6xxx_setup_reg_address(oct, oct->chip, &oct->reg_list);
+
+	/* Determine variant of card */
+	if (lio_is_210nv(oct))
+		card_type = LIO_210NV;
+
+	cn68xx->conf = (struct octeon_config *)
+		       oct_get_config_info(oct, card_type);
+	if (!cn68xx->conf) {
+		dev_err(&oct->pci_dev->dev, "%s No Config found for CN68XX %s\n",
+			__func__,
+			(card_type == LIO_410NV) ? LIO_410NV_NAME :
+			LIO_210NV_NAME);
+		octeon_unmap_pci_barx(oct, 0);
+		octeon_unmap_pci_barx(oct, 1);
+		return 1;
+	}
+
+	oct->coproc_clock_rate = 1000000ULL * lio_cn6xxx_coprocessor_clock(oct);
+
+	lio_cn68xx_vendor_message_fix(oct);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(lio_setup_cn68xx_octeon_device);

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn68xx_device.h

@@ -0,0 +1,27 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file  cn68xx_device.h
+ *  \brief Host Driver: Routines that perform CN68XX specific operations.
+ */
+
+#ifndef __CN68XX_DEVICE_H__
+#define  __CN68XX_DEVICE_H__
+
+int lio_setup_cn68xx_octeon_device(struct octeon_device *oct);
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/cn68xx_regs.h

@@ -0,0 +1,45 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file cn68xx_regs.h
+ *  \brief Host Driver: Register Address and Register Mask values for
+ *  Octeon CN68XX devices. The register map for CN66XX is the same
+ *  for most registers. This file has the other registers that are
+ *  68XX-specific.
+ */
+
+#ifndef __CN68XX_REGS_H__
+#define __CN68XX_REGS_H__
+
+/*###################### REQUEST QUEUE #########################*/
+
+#define    CN68XX_SLI_IQ_PORT0_PKIND             0x0800
+
+#define    CN68XX_SLI_IQ_PORT_PKIND(iq)           \
+	(CN68XX_SLI_IQ_PORT0_PKIND + ((iq) * CN6XXX_IQ_OFFSET))
+
+/*############################ OUTPUT QUEUE #########################*/
+
+/* Starting pipe number and number of pipes used by the SLI packet output. */
+#define    CN68XX_SLI_TX_PIPE                    0x1230
+
+/*######################## INTERRUPTS #########################*/
+
+/*------------------ Interrupt Masks ----------------*/
+#define    CN68XX_INTR_PIPE_ERR                  BIT_ULL(61)
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/lio_vf_rep.c

@@ -0,0 +1,672 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2017 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/if_vlan.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_nic.h"
+#include "octeon_main.h"
+#include "octeon_network.h"
+#include "lio_vf_rep.h"
+
+static int lio_vf_rep_open(struct net_device *ndev);
+static int lio_vf_rep_stop(struct net_device *ndev);
+static netdev_tx_t lio_vf_rep_pkt_xmit(struct sk_buff *skb,
+				       struct net_device *ndev);
+static void lio_vf_rep_tx_timeout(struct net_device *netdev, unsigned int txqueue);
+static int lio_vf_rep_phys_port_name(struct net_device *dev,
+				     char *buf, size_t len);
+static void lio_vf_rep_get_stats64(struct net_device *dev,
+				   struct rtnl_link_stats64 *stats64);
+static int lio_vf_rep_change_mtu(struct net_device *ndev, int new_mtu);
+static int lio_vf_get_port_parent_id(struct net_device *dev,
+				     struct netdev_phys_item_id *ppid);
+
+static const struct net_device_ops lio_vf_rep_ndev_ops = {
+	.ndo_open = lio_vf_rep_open,
+	.ndo_stop = lio_vf_rep_stop,
+	.ndo_start_xmit = lio_vf_rep_pkt_xmit,
+	.ndo_tx_timeout = lio_vf_rep_tx_timeout,
+	.ndo_get_phys_port_name = lio_vf_rep_phys_port_name,
+	.ndo_get_stats64 = lio_vf_rep_get_stats64,
+	.ndo_change_mtu = lio_vf_rep_change_mtu,
+	.ndo_get_port_parent_id = lio_vf_get_port_parent_id,
+};
+
+static int
+lio_vf_rep_send_soft_command(struct octeon_device *oct,
+			     void *req, int req_size,
+			     void *resp, int resp_size)
+{
+	int tot_resp_size = sizeof(struct lio_vf_rep_resp) + resp_size;
+	struct octeon_soft_command *sc = NULL;
+	struct lio_vf_rep_resp *rep_resp;
+	void *sc_req;
+	int err;
+
+	sc = (struct octeon_soft_command *)
+		octeon_alloc_soft_command(oct, req_size,
+					  tot_resp_size, 0);
+	if (!sc)
+		return -ENOMEM;
+
+	init_completion(&sc->complete);
+	sc->sc_status = OCTEON_REQUEST_PENDING;
+
+	sc_req = (struct lio_vf_rep_req *)sc->virtdptr;
+	memcpy(sc_req, req, req_size);
+
+	rep_resp = (struct lio_vf_rep_resp *)sc->virtrptr;
+	memset(rep_resp, 0, tot_resp_size);
+	WRITE_ONCE(rep_resp->status, 1);
+
+	sc->iq_no = 0;
+	octeon_prepare_soft_command(oct, sc, OPCODE_NIC,
+				    OPCODE_NIC_VF_REP_CMD, 0, 0, 0);
+
+	err = octeon_send_soft_command(oct, sc);
+	if (err == IQ_SEND_FAILED)
+		goto free_buff;
+
+	err = wait_for_sc_completion_timeout(oct, sc, 0);
+	if (err)
+		return err;
+
+	err = READ_ONCE(rep_resp->status) ? -EBUSY : 0;
+	if (err)
+		dev_err(&oct->pci_dev->dev, "VF rep send config failed\n");
+	else if (resp)
+		memcpy(resp, (rep_resp + 1), resp_size);
+
+	WRITE_ONCE(sc->caller_is_done, true);
+	return err;
+
+free_buff:
+	octeon_free_soft_command(oct, sc);
+
+	return err;
+}
+
+static int
+lio_vf_rep_open(struct net_device *ndev)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(ndev);
+	struct lio_vf_rep_req rep_cfg;
+	struct octeon_device *oct;
+	int ret;
+
+	oct = vf_rep->oct;
+
+	memset(&rep_cfg, 0, sizeof(rep_cfg));
+	rep_cfg.req_type = LIO_VF_REP_REQ_STATE;
+	rep_cfg.ifidx = vf_rep->ifidx;
+	rep_cfg.rep_state.state = LIO_VF_REP_STATE_UP;
+
+	ret = lio_vf_rep_send_soft_command(oct, &rep_cfg,
+					   sizeof(rep_cfg), NULL, 0);
+
+	if (ret) {
+		dev_err(&oct->pci_dev->dev,
+			"VF_REP open failed with err %d\n", ret);
+		return -EIO;
+	}
+
+	atomic_set(&vf_rep->ifstate, (atomic_read(&vf_rep->ifstate) |
+				      LIO_IFSTATE_RUNNING));
+
+	netif_carrier_on(ndev);
+	netif_start_queue(ndev);
+
+	return 0;
+}
+
+static int
+lio_vf_rep_stop(struct net_device *ndev)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(ndev);
+	struct lio_vf_rep_req rep_cfg;
+	struct octeon_device *oct;
+	int ret;
+
+	oct = vf_rep->oct;
+
+	memset(&rep_cfg, 0, sizeof(rep_cfg));
+	rep_cfg.req_type = LIO_VF_REP_REQ_STATE;
+	rep_cfg.ifidx = vf_rep->ifidx;
+	rep_cfg.rep_state.state = LIO_VF_REP_STATE_DOWN;
+
+	ret = lio_vf_rep_send_soft_command(oct, &rep_cfg,
+					   sizeof(rep_cfg), NULL, 0);
+
+	if (ret) {
+		dev_err(&oct->pci_dev->dev,
+			"VF_REP dev stop failed with err %d\n", ret);
+		return -EIO;
+	}
+
+	atomic_set(&vf_rep->ifstate, (atomic_read(&vf_rep->ifstate) &
+				      ~LIO_IFSTATE_RUNNING));
+
+	netif_tx_disable(ndev);
+	netif_carrier_off(ndev);
+
+	return 0;
+}
+
+static void
+lio_vf_rep_tx_timeout(struct net_device *ndev, unsigned int txqueue)
+{
+	netif_trans_update(ndev);
+
+	netif_wake_queue(ndev);
+}
+
+static void
+lio_vf_rep_get_stats64(struct net_device *dev,
+		       struct rtnl_link_stats64 *stats64)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(dev);
+
+	/* Swap tx and rx stats as VF rep is a switch port */
+	stats64->tx_packets = vf_rep->stats.rx_packets;
+	stats64->tx_bytes   = vf_rep->stats.rx_bytes;
+	stats64->tx_dropped = vf_rep->stats.rx_dropped;
+
+	stats64->rx_packets = vf_rep->stats.tx_packets;
+	stats64->rx_bytes   = vf_rep->stats.tx_bytes;
+	stats64->rx_dropped = vf_rep->stats.tx_dropped;
+}
+
+static int
+lio_vf_rep_change_mtu(struct net_device *ndev, int new_mtu)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(ndev);
+	struct lio_vf_rep_req rep_cfg;
+	struct octeon_device *oct;
+	int ret;
+
+	oct = vf_rep->oct;
+
+	memset(&rep_cfg, 0, sizeof(rep_cfg));
+	rep_cfg.req_type = LIO_VF_REP_REQ_MTU;
+	rep_cfg.ifidx = vf_rep->ifidx;
+	rep_cfg.rep_mtu.mtu = cpu_to_be32(new_mtu);
+
+	ret = lio_vf_rep_send_soft_command(oct, &rep_cfg,
+					   sizeof(rep_cfg), NULL, 0);
+	if (ret) {
+		dev_err(&oct->pci_dev->dev,
+			"Change MTU failed with err %d\n", ret);
+		return -EIO;
+	}
+
+	WRITE_ONCE(ndev->mtu, new_mtu);
+
+	return 0;
+}
+
+static int
+lio_vf_rep_phys_port_name(struct net_device *dev,
+			  char *buf, size_t len)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(dev);
+	struct octeon_device *oct = vf_rep->oct;
+	int ret;
+
+	ret = snprintf(buf, len, "pf%dvf%d", oct->pf_num,
+		       vf_rep->ifidx - oct->pf_num * 64 - 1);
+	if (ret >= len)
+		return -EOPNOTSUPP;
+
+	return 0;
+}
+
+static struct net_device *
+lio_vf_rep_get_ndev(struct octeon_device *oct, int ifidx)
+{
+	int vf_id, max_vfs = CN23XX_MAX_VFS_PER_PF + 1;
+	int vfid_mask = max_vfs - 1;
+
+	if (ifidx <= oct->pf_num * max_vfs ||
+	    ifidx >= oct->pf_num * max_vfs + max_vfs)
+		return NULL;
+
+	/* ifidx 1-63 for PF0 VFs
+	 * ifidx 65-127 for PF1 VFs
+	 */
+	vf_id = (ifidx & vfid_mask) - 1;
+
+	return oct->vf_rep_list.ndev[vf_id];
+}
+
+static void
+lio_vf_rep_copy_packet(struct octeon_device *oct,
+		       struct sk_buff *skb,
+		       int len)
+{
+	if (likely(len > MIN_SKB_SIZE)) {
+		struct octeon_skb_page_info *pg_info;
+		unsigned char *va;
+
+		pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+		if (pg_info->page) {
+			va = page_address(pg_info->page) +
+				pg_info->page_offset;
+			memcpy(skb->data, va, MIN_SKB_SIZE);
+			skb_put(skb, MIN_SKB_SIZE);
+			skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags,
+					pg_info->page,
+					pg_info->page_offset + MIN_SKB_SIZE,
+					len - MIN_SKB_SIZE,
+					LIO_RXBUFFER_SZ);
+		}
+	} else {
+		struct octeon_skb_page_info *pg_info =
+			((struct octeon_skb_page_info *)(skb->cb));
+
+		skb_copy_to_linear_data(skb, page_address(pg_info->page) +
+					pg_info->page_offset, len);
+		skb_put(skb, len);
+		put_page(pg_info->page);
+	}
+}
+
+static int
+lio_vf_rep_pkt_recv(struct octeon_recv_info *recv_info, void *buf)
+{
+	struct octeon_recv_pkt *recv_pkt = recv_info->recv_pkt;
+	struct lio_vf_rep_desc *vf_rep;
+	struct net_device *vf_ndev;
+	struct octeon_device *oct;
+	union octeon_rh *rh;
+	struct sk_buff *skb;
+	int i, ifidx;
+
+	oct = lio_get_device(recv_pkt->octeon_id);
+	if (!oct)
+		goto free_buffers;
+
+	skb = recv_pkt->buffer_ptr[0];
+	rh = &recv_pkt->rh;
+	ifidx = rh->r.ossp;
+
+	vf_ndev = lio_vf_rep_get_ndev(oct, ifidx);
+	if (!vf_ndev)
+		goto free_buffers;
+
+	vf_rep = netdev_priv(vf_ndev);
+	if (!(atomic_read(&vf_rep->ifstate) & LIO_IFSTATE_RUNNING) ||
+	    recv_pkt->buffer_count > 1)
+		goto free_buffers;
+
+	skb->dev = vf_ndev;
+
+	/* Multiple buffers are not used for vf_rep packets.
+	 * So just buffer_size[0] is valid.
+	 */
+	lio_vf_rep_copy_packet(oct, skb, recv_pkt->buffer_size[0]);
+
+	skb_pull(skb, rh->r_dh.len * BYTES_PER_DHLEN_UNIT);
+	skb->protocol = eth_type_trans(skb, skb->dev);
+	skb->ip_summed = CHECKSUM_NONE;
+
+	netif_rx(skb);
+
+	octeon_free_recv_info(recv_info);
+
+	return 0;
+
+free_buffers:
+	for (i = 0; i < recv_pkt->buffer_count; i++)
+		recv_buffer_free(recv_pkt->buffer_ptr[i]);
+
+	octeon_free_recv_info(recv_info);
+
+	return 0;
+}
+
+static void
+lio_vf_rep_packet_sent_callback(struct octeon_device *oct,
+				u32 status, void *buf)
+{
+	struct octeon_soft_command *sc = (struct octeon_soft_command *)buf;
+	struct sk_buff *skb = sc->ctxptr;
+	struct net_device *ndev = skb->dev;
+	u32 iq_no;
+
+	dma_unmap_single(&oct->pci_dev->dev, sc->dmadptr,
+			 sc->datasize, DMA_TO_DEVICE);
+	dev_kfree_skb_any(skb);
+	iq_no = sc->iq_no;
+	octeon_free_soft_command(oct, sc);
+
+	if (octnet_iq_is_full(oct, iq_no))
+		return;
+
+	if (netif_queue_stopped(ndev))
+		netif_wake_queue(ndev);
+}
+
+static netdev_tx_t
+lio_vf_rep_pkt_xmit(struct sk_buff *skb, struct net_device *ndev)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(ndev);
+	struct net_device *parent_ndev = vf_rep->parent_ndev;
+	struct octeon_device *oct = vf_rep->oct;
+	struct octeon_instr_pki_ih3 *pki_ih3;
+	struct octeon_soft_command *sc;
+	struct lio *parent_lio;
+	int status;
+
+	parent_lio = GET_LIO(parent_ndev);
+
+	if (!(atomic_read(&vf_rep->ifstate) & LIO_IFSTATE_RUNNING) ||
+	    skb->len <= 0)
+		goto xmit_failed;
+
+	if (octnet_iq_is_full(vf_rep->oct, parent_lio->txq)) {
+		dev_err(&oct->pci_dev->dev, "VF rep: Device IQ full\n");
+		netif_stop_queue(ndev);
+		return NETDEV_TX_BUSY;
+	}
+
+	sc = (struct octeon_soft_command *)
+		octeon_alloc_soft_command(oct, 0, 16, 0);
+	if (!sc) {
+		dev_err(&oct->pci_dev->dev, "VF rep: Soft command alloc failed\n");
+		goto xmit_failed;
+	}
+
+	/* Multiple buffers are not used for vf_rep packets. */
+	if (skb_shinfo(skb)->nr_frags != 0) {
+		dev_err(&oct->pci_dev->dev, "VF rep: nr_frags != 0. Dropping packet\n");
+		octeon_free_soft_command(oct, sc);
+		goto xmit_failed;
+	}
+
+	sc->dmadptr = dma_map_single(&oct->pci_dev->dev,
+				     skb->data, skb->len, DMA_TO_DEVICE);
+	if (dma_mapping_error(&oct->pci_dev->dev, sc->dmadptr)) {
+		dev_err(&oct->pci_dev->dev, "VF rep: DMA mapping failed\n");
+		octeon_free_soft_command(oct, sc);
+		goto xmit_failed;
+	}
+
+	sc->virtdptr = skb->data;
+	sc->datasize = skb->len;
+	sc->ctxptr = skb;
+	sc->iq_no = parent_lio->txq;
+
+	octeon_prepare_soft_command(oct, sc, OPCODE_NIC, OPCODE_NIC_VF_REP_PKT,
+				    vf_rep->ifidx, 0, 0);
+	pki_ih3 = (struct octeon_instr_pki_ih3 *)&sc->cmd.cmd3.pki_ih3;
+	pki_ih3->tagtype = ORDERED_TAG;
+
+	sc->callback = lio_vf_rep_packet_sent_callback;
+	sc->callback_arg = sc;
+
+	status = octeon_send_soft_command(oct, sc);
+	if (status == IQ_SEND_FAILED) {
+		dma_unmap_single(&oct->pci_dev->dev, sc->dmadptr,
+				 sc->datasize, DMA_TO_DEVICE);
+		octeon_free_soft_command(oct, sc);
+		goto xmit_failed;
+	}
+
+	if (status == IQ_SEND_STOP)
+		netif_stop_queue(ndev);
+
+	netif_trans_update(ndev);
+
+	return NETDEV_TX_OK;
+
+xmit_failed:
+	dev_kfree_skb_any(skb);
+
+	return NETDEV_TX_OK;
+}
+
+static int lio_vf_get_port_parent_id(struct net_device *dev,
+				     struct netdev_phys_item_id *ppid)
+{
+	struct lio_vf_rep_desc *vf_rep = netdev_priv(dev);
+	struct net_device *parent_ndev = vf_rep->parent_ndev;
+	struct lio *lio = GET_LIO(parent_ndev);
+
+	ppid->id_len = ETH_ALEN;
+	ether_addr_copy(ppid->id, (void *)&lio->linfo.hw_addr + 2);
+
+	return 0;
+}
+
+static void
+lio_vf_rep_fetch_stats(struct work_struct *work)
+{
+	struct cavium_wk *wk = (struct cavium_wk *)work;
+	struct lio_vf_rep_desc *vf_rep = wk->ctxptr;
+	struct lio_vf_rep_stats stats;
+	struct lio_vf_rep_req rep_cfg;
+	struct octeon_device *oct;
+	int ret;
+
+	oct = vf_rep->oct;
+
+	memset(&rep_cfg, 0, sizeof(rep_cfg));
+	rep_cfg.req_type = LIO_VF_REP_REQ_STATS;
+	rep_cfg.ifidx = vf_rep->ifidx;
+
+	ret = lio_vf_rep_send_soft_command(oct, &rep_cfg, sizeof(rep_cfg),
+					   &stats, sizeof(stats));
+
+	if (!ret) {
+		octeon_swap_8B_data((u64 *)&stats, (sizeof(stats) >> 3));
+		memcpy(&vf_rep->stats, &stats, sizeof(stats));
+	}
+
+	schedule_delayed_work(&vf_rep->stats_wk.work,
+			      msecs_to_jiffies(LIO_VF_REP_STATS_POLL_TIME_MS));
+}
+
+int
+lio_vf_rep_create(struct octeon_device *oct)
+{
+	struct lio_vf_rep_desc *vf_rep;
+	struct net_device *ndev;
+	int i, num_vfs;
+
+	if (oct->eswitch_mode != DEVLINK_ESWITCH_MODE_SWITCHDEV)
+		return 0;
+
+	if (!oct->sriov_info.sriov_enabled)
+		return 0;
+
+	num_vfs = oct->sriov_info.num_vfs_alloced;
+
+	oct->vf_rep_list.num_vfs = 0;
+	for (i = 0; i < num_vfs; i++) {
+		ndev = alloc_etherdev(sizeof(struct lio_vf_rep_desc));
+
+		if (!ndev) {
+			dev_err(&oct->pci_dev->dev,
+				"VF rep device %d creation failed\n", i);
+			goto cleanup;
+		}
+
+		ndev->min_mtu = LIO_MIN_MTU_SIZE;
+		ndev->max_mtu = LIO_MAX_MTU_SIZE;
+		ndev->netdev_ops = &lio_vf_rep_ndev_ops;
+
+		vf_rep = netdev_priv(ndev);
+		memset(vf_rep, 0, sizeof(*vf_rep));
+
+		vf_rep->ndev = ndev;
+		vf_rep->oct = oct;
+		vf_rep->parent_ndev = oct->props[0].netdev;
+		vf_rep->ifidx = (oct->pf_num * 64) + i + 1;
+
+		eth_hw_addr_random(ndev);
+
+		if (register_netdev(ndev)) {
+			dev_err(&oct->pci_dev->dev, "VF rep nerdev registration failed\n");
+
+			free_netdev(ndev);
+			goto cleanup;
+		}
+
+		netif_carrier_off(ndev);
+
+		INIT_DELAYED_WORK(&vf_rep->stats_wk.work,
+				  lio_vf_rep_fetch_stats);
+		vf_rep->stats_wk.ctxptr = (void *)vf_rep;
+		schedule_delayed_work(&vf_rep->stats_wk.work,
+				      msecs_to_jiffies
+				      (LIO_VF_REP_STATS_POLL_TIME_MS));
+		oct->vf_rep_list.num_vfs++;
+		oct->vf_rep_list.ndev[i] = ndev;
+	}
+
+	if (octeon_register_dispatch_fn(oct, OPCODE_NIC,
+					OPCODE_NIC_VF_REP_PKT,
+					lio_vf_rep_pkt_recv, oct)) {
+		dev_err(&oct->pci_dev->dev, "VF rep Dispatch func registration failed\n");
+
+		goto cleanup;
+	}
+
+	return 0;
+
+cleanup:
+	for (i = 0; i < oct->vf_rep_list.num_vfs; i++) {
+		ndev = oct->vf_rep_list.ndev[i];
+		oct->vf_rep_list.ndev[i] = NULL;
+		if (ndev) {
+			vf_rep = netdev_priv(ndev);
+			cancel_delayed_work_sync
+				(&vf_rep->stats_wk.work);
+			unregister_netdev(ndev);
+			free_netdev(ndev);
+		}
+	}
+
+	oct->vf_rep_list.num_vfs = 0;
+
+	return -1;
+}
+
+void
+lio_vf_rep_destroy(struct octeon_device *oct)
+{
+	struct lio_vf_rep_desc *vf_rep;
+	struct net_device *ndev;
+	int i;
+
+	if (oct->eswitch_mode != DEVLINK_ESWITCH_MODE_SWITCHDEV)
+		return;
+
+	if (!oct->sriov_info.sriov_enabled)
+		return;
+
+	for (i = 0; i < oct->vf_rep_list.num_vfs; i++) {
+		ndev = oct->vf_rep_list.ndev[i];
+		oct->vf_rep_list.ndev[i] = NULL;
+		if (ndev) {
+			vf_rep = netdev_priv(ndev);
+			cancel_delayed_work_sync
+				(&vf_rep->stats_wk.work);
+			netif_tx_disable(ndev);
+			netif_carrier_off(ndev);
+
+			unregister_netdev(ndev);
+			free_netdev(ndev);
+		}
+	}
+
+	oct->vf_rep_list.num_vfs = 0;
+}
+
+static int
+lio_vf_rep_netdev_event(struct notifier_block *nb,
+			unsigned long event, void *ptr)
+{
+	struct net_device *ndev = netdev_notifier_info_to_dev(ptr);
+	struct lio_vf_rep_desc *vf_rep;
+	struct lio_vf_rep_req rep_cfg;
+	struct octeon_device *oct;
+	int ret;
+
+	switch (event) {
+	case NETDEV_REGISTER:
+	case NETDEV_CHANGENAME:
+		break;
+
+	default:
+		return NOTIFY_DONE;
+	}
+
+	if (ndev->netdev_ops != &lio_vf_rep_ndev_ops)
+		return NOTIFY_DONE;
+
+	vf_rep = netdev_priv(ndev);
+	oct = vf_rep->oct;
+
+	if (strlen(ndev->name) > LIO_IF_NAME_SIZE) {
+		dev_err(&oct->pci_dev->dev,
+			"Device name change sync failed as the size is > %d\n",
+			LIO_IF_NAME_SIZE);
+		return NOTIFY_DONE;
+	}
+
+	memset(&rep_cfg, 0, sizeof(rep_cfg));
+	rep_cfg.req_type = LIO_VF_REP_REQ_DEVNAME;
+	rep_cfg.ifidx = vf_rep->ifidx;
+	strscpy(rep_cfg.rep_name.name, ndev->name,
+		sizeof(rep_cfg.rep_name.name));
+
+	ret = lio_vf_rep_send_soft_command(oct, &rep_cfg,
+					   sizeof(rep_cfg), NULL, 0);
+	if (ret)
+		dev_err(&oct->pci_dev->dev,
+			"vf_rep netdev name change failed with err %d\n", ret);
+
+	return NOTIFY_DONE;
+}
+
+static struct notifier_block lio_vf_rep_netdev_notifier = {
+	.notifier_call = lio_vf_rep_netdev_event,
+};
+
+int
+lio_vf_rep_modinit(void)
+{
+	if (register_netdevice_notifier(&lio_vf_rep_netdev_notifier)) {
+		pr_err("netdev notifier registration failed\n");
+		return -EFAULT;
+	}
+
+	return 0;
+}
+
+void
+lio_vf_rep_modexit(void)
+{
+	if (unregister_netdevice_notifier(&lio_vf_rep_netdev_notifier))
+		pr_err("netdev notifier unregister failed\n");
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/lio_vf_rep.h

@@ -0,0 +1,49 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2017 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ *
+ * This file may also be available under a different license from Cavium.
+ * Contact Cavium, Inc. for more information
+ **********************************************************************/
+
+/*! \file octeon_vf_main.h
+ *  \brief Host Driver: This file defines vf_rep related macros and structures
+ */
+#ifndef __LIO_VF_REP_H__
+#define __LIO_VF_REP_H__
+#define LIO_VF_REP_REQ_TMO_MS 5000
+#define LIO_VF_REP_STATS_POLL_TIME_MS 200
+
+struct lio_vf_rep_desc {
+	struct net_device *parent_ndev;
+	struct net_device *ndev;
+	struct octeon_device *oct;
+	struct lio_vf_rep_stats stats;
+	struct cavium_wk stats_wk;
+	atomic_t ifstate;
+	int ifidx;
+};
+
+struct lio_vf_rep_sc_ctx {
+	struct completion complete;
+};
+
+int lio_vf_rep_create(struct octeon_device *oct);
+void lio_vf_rep_destroy(struct octeon_device *oct);
+int lio_vf_rep_modinit(void);
+void lio_vf_rep_modexit(void);
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/liquidio_image.h

@@ -0,0 +1,54 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#ifndef _LIQUIDIO_IMAGE_H_
+#define _LIQUIDIO_IMAGE_H_
+
+#define LIO_MAX_FW_TYPE_LEN     (8)
+#define LIO_MAX_FW_FILENAME_LEN (256)
+#define LIO_FW_DIR              "liquidio/"
+#define LIO_FW_BASE_NAME        "lio_"
+#define LIO_FW_NAME_SUFFIX      ".bin"
+#define LIO_FW_NAME_TYPE_NIC    "nic"
+#define LIO_FW_NAME_TYPE_AUTO   "auto"
+#define LIO_FW_NAME_TYPE_NONE   "none"
+#define LIO_MAX_FIRMWARE_VERSION_LEN 16
+
+#define LIO_MAX_BOOTCMD_LEN 1024
+#define LIO_MAX_IMAGES 16
+#define LIO_NIC_MAGIC 0x434E4943     /* "CNIC" */
+struct octeon_firmware_desc {
+	__be64 addr;
+	__be32 len;
+	__be32 crc32;         /* crc32 of image */
+};
+
+/* Following the header is a list of 64-bit aligned binary images,
+ * as described by the desc field.
+ * Numeric fields are in network byte order.
+ */
+struct octeon_firmware_file_header {
+	__be32 magic;
+	char version[LIO_MAX_FIRMWARE_VERSION_LEN];
+	char bootcmd[LIO_MAX_BOOTCMD_LEN];
+	__be32 num_images;
+	struct octeon_firmware_desc desc[LIO_MAX_IMAGES];
+	__be32 pad;
+	__be32 crc32;         /* header checksum */
+};
+
+#endif /* _LIQUIDIO_IMAGE_H_ */

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_config.h

@@ -0,0 +1,472 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file  octeon_config.h
+ *  \brief Host Driver: Configuration data structures for the host driver.
+ */
+
+#ifndef __OCTEON_CONFIG_H__
+#define __OCTEON_CONFIG_H__
+
+/*--------------------------CONFIG VALUES------------------------*/
+
+/* The following macros affect the way the driver data structures
+ * are generated for Octeon devices.
+ * They can be modified.
+ */
+
+/* Maximum octeon devices defined as MAX_OCTEON_NICIF to support
+ * multiple(<= MAX_OCTEON_NICIF) Miniports
+ */
+#define   MAX_OCTEON_NICIF             128
+#define   MAX_OCTEON_DEVICES           MAX_OCTEON_NICIF
+#define   MAX_OCTEON_LINKS	       MAX_OCTEON_NICIF
+#define   MAX_OCTEON_MULTICAST_ADDR    32
+
+#define   MAX_OCTEON_FILL_COUNT        8
+
+/* CN6xxx IQ configuration macros */
+#define   CN6XXX_MAX_INPUT_QUEUES      32
+#define   CN6XXX_MAX_IQ_DESCRIPTORS    2048
+#define   CN6XXX_DB_MIN                1
+#define   CN6XXX_DB_MAX                8
+#define   CN6XXX_DB_TIMEOUT            1
+
+/* CN6xxx OQ configuration macros */
+#define   CN6XXX_MAX_OUTPUT_QUEUES     32
+#define   CN6XXX_MAX_OQ_DESCRIPTORS    2048
+#define   CN6XXX_OQ_BUF_SIZE           1664
+#define   CN6XXX_OQ_PKTSPER_INTR       ((CN6XXX_MAX_OQ_DESCRIPTORS < 512) ? \
+					(CN6XXX_MAX_OQ_DESCRIPTORS / 4) : 128)
+#define   CN6XXX_OQ_REFIL_THRESHOLD    ((CN6XXX_MAX_OQ_DESCRIPTORS < 512) ? \
+					(CN6XXX_MAX_OQ_DESCRIPTORS / 4) : 128)
+
+#define   CN6XXX_OQ_INTR_PKT           64
+#define   CN6XXX_OQ_INTR_TIME          100
+#define   DEFAULT_NUM_NIC_PORTS_66XX   2
+#define   DEFAULT_NUM_NIC_PORTS_68XX   4
+#define   DEFAULT_NUM_NIC_PORTS_68XX_210NV  2
+
+/* CN23xx  IQ configuration macros */
+#define   CN23XX_MAX_VFS_PER_PF_PASS_1_0 8
+#define   CN23XX_MAX_VFS_PER_PF_PASS_1_1 31
+#define   CN23XX_MAX_VFS_PER_PF          63
+#define   CN23XX_MAX_RINGS_PER_VF        8
+
+#define   CN23XX_MAX_RINGS_PER_PF_PASS_1_0 12
+#define   CN23XX_MAX_RINGS_PER_PF_PASS_1_1 32
+#define   CN23XX_MAX_RINGS_PER_PF          64
+#define   CN23XX_MAX_RINGS_PER_VF          8
+
+#define   CN23XX_MAX_INPUT_QUEUES	CN23XX_MAX_RINGS_PER_PF
+#define   CN23XX_MAX_IQ_DESCRIPTORS	2048
+#define   CN23XX_DEFAULT_IQ_DESCRIPTORS	512
+#define   CN23XX_MIN_IQ_DESCRIPTORS	128
+#define   CN23XX_DB_MIN                 1
+#define   CN23XX_DB_MAX                 8
+#define   CN23XX_DB_TIMEOUT             1
+
+#define   CN23XX_MAX_OUTPUT_QUEUES	CN23XX_MAX_RINGS_PER_PF
+#define   CN23XX_MAX_OQ_DESCRIPTORS	2048
+#define   CN23XX_DEFAULT_OQ_DESCRIPTORS	512
+#define   CN23XX_MIN_OQ_DESCRIPTORS	128
+#define   CN23XX_OQ_BUF_SIZE		1664
+#define   CN23XX_OQ_PKTSPER_INTR	128
+/*#define CAVIUM_ONLY_CN23XX_RX_PERF*/
+#define   CN23XX_OQ_REFIL_THRESHOLD	16
+
+#define   CN23XX_OQ_INTR_PKT		64
+#define   CN23XX_OQ_INTR_TIME		100
+#define   DEFAULT_NUM_NIC_PORTS_23XX	1
+
+#define   CN23XX_CFG_IO_QUEUES		CN23XX_MAX_RINGS_PER_PF
+/* PEMs count */
+#define   CN23XX_MAX_MACS		4
+
+#define   CN23XX_DEF_IQ_INTR_THRESHOLD	32
+#define   CN23XX_DEF_IQ_INTR_BYTE_THRESHOLD   (64 * 1024)
+/* common OCTEON configuration macros */
+#define   CN6XXX_CFG_IO_QUEUES         32
+#define   OCTEON_32BYTE_INSTR          32
+#define   OCTEON_64BYTE_INSTR          64
+#define   OCTEON_MAX_BASE_IOQ          4
+
+#define   OCTEON_DMA_INTR_PKT          64
+#define   OCTEON_DMA_INTR_TIME         1000
+
+#define MAX_TXQS_PER_INTF  8
+#define MAX_RXQS_PER_INTF  8
+#define DEF_TXQS_PER_INTF  4
+#define DEF_RXQS_PER_INTF  4
+
+#define INVALID_IOQ_NO          0xff
+
+#define   DEFAULT_POW_GRP       0
+
+/* Macros to get octeon config params */
+#define CFG_GET_IQ_CFG(cfg)                      ((cfg)->iq)
+#define CFG_GET_IQ_MAX_Q(cfg)                    ((cfg)->iq.max_iqs)
+#define CFG_GET_IQ_PENDING_LIST_SIZE(cfg)        ((cfg)->iq.pending_list_size)
+#define CFG_GET_IQ_INSTR_TYPE(cfg)               ((cfg)->iq.instr_type)
+#define CFG_GET_IQ_DB_MIN(cfg)                   ((cfg)->iq.db_min)
+#define CFG_GET_IQ_DB_TIMEOUT(cfg)               ((cfg)->iq.db_timeout)
+
+#define CFG_GET_IQ_INTR_PKT(cfg)                 ((cfg)->iq.iq_intr_pkt)
+#define CFG_SET_IQ_INTR_PKT(cfg, val)            (cfg)->iq.iq_intr_pkt = val
+
+#define CFG_GET_OQ_MAX_Q(cfg)                    ((cfg)->oq.max_oqs)
+#define CFG_GET_OQ_PKTS_PER_INTR(cfg)            ((cfg)->oq.pkts_per_intr)
+#define CFG_GET_OQ_REFILL_THRESHOLD(cfg)         ((cfg)->oq.refill_threshold)
+#define CFG_GET_OQ_INTR_PKT(cfg)                 ((cfg)->oq.oq_intr_pkt)
+#define CFG_GET_OQ_INTR_TIME(cfg)                ((cfg)->oq.oq_intr_time)
+#define CFG_SET_OQ_INTR_PKT(cfg, val)            (cfg)->oq.oq_intr_pkt = val
+#define CFG_SET_OQ_INTR_TIME(cfg, val)           (cfg)->oq.oq_intr_time = val
+
+#define CFG_GET_DMA_INTR_PKT(cfg)                ((cfg)->dma.dma_intr_pkt)
+#define CFG_GET_DMA_INTR_TIME(cfg)               ((cfg)->dma.dma_intr_time)
+#define CFG_GET_NUM_NIC_PORTS(cfg)               ((cfg)->num_nic_ports)
+#define CFG_GET_NUM_DEF_TX_DESCS(cfg)            ((cfg)->num_def_tx_descs)
+#define CFG_GET_NUM_DEF_RX_DESCS(cfg)            ((cfg)->num_def_rx_descs)
+#define CFG_GET_DEF_RX_BUF_SIZE(cfg)             ((cfg)->def_rx_buf_size)
+
+#define CFG_GET_MAX_TXQS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].max_txqs)
+#define CFG_GET_NUM_TXQS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].num_txqs)
+#define CFG_GET_MAX_RXQS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].max_rxqs)
+#define CFG_GET_NUM_RXQS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].num_rxqs)
+#define CFG_GET_NUM_RX_DESCS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].num_rx_descs)
+#define CFG_GET_NUM_TX_DESCS_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].num_tx_descs)
+#define CFG_GET_NUM_RX_BUF_SIZE_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].rx_buf_size)
+#define CFG_GET_BASE_QUE_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].base_queue)
+#define CFG_GET_GMXID_NIC_IF(cfg, idx) \
+				((cfg)->nic_if_cfg[idx].gmx_port_id)
+
+#define CFG_GET_CTRL_Q_GRP(cfg)                  ((cfg)->misc.ctrlq_grp)
+#define CFG_GET_HOST_LINK_QUERY_INTERVAL(cfg) \
+				((cfg)->misc.host_link_query_interval)
+#define CFG_GET_OCT_LINK_QUERY_INTERVAL(cfg) \
+				((cfg)->misc.oct_link_query_interval)
+#define CFG_GET_IS_SLI_BP_ON(cfg)                ((cfg)->misc.enable_sli_oq_bp)
+
+#define CFG_SET_NUM_RX_DESCS_NIC_IF(cfg, idx, value) \
+				((cfg)->nic_if_cfg[idx].num_rx_descs = value)
+#define CFG_SET_NUM_TX_DESCS_NIC_IF(cfg, idx, value) \
+				((cfg)->nic_if_cfg[idx].num_tx_descs = value)
+
+/* Max IOQs per OCTEON Link */
+#define MAX_IOQS_PER_NICIF              64
+
+enum lio_card_type {
+	LIO_210SV = 0, /* Two port, 66xx */
+	LIO_210NV,     /* Two port, 68xx */
+	LIO_410NV,     /* Four port, 68xx */
+	LIO_23XX       /* 23xx */
+};
+
+#define LIO_210SV_NAME "210sv"
+#define LIO_210NV_NAME "210nv"
+#define LIO_410NV_NAME "410nv"
+#define LIO_23XX_NAME  "23xx"
+
+/** Structure to define the configuration attributes for each Input queue.
+ *  Applicable to all Octeon processors
+ **/
+struct octeon_iq_config {
+#ifdef __BIG_ENDIAN_BITFIELD
+	u64 reserved:16;
+
+	/** Tx interrupt packets. Applicable to 23xx only */
+	u64 iq_intr_pkt:16;
+
+	/** Minimum ticks to wait before checking for pending instructions. */
+	u64 db_timeout:16;
+
+	/** Minimum number of commands pending to be posted to Octeon
+	 *  before driver hits the Input queue doorbell.
+	 */
+	u64 db_min:8;
+
+	/** Command size - 32 or 64 bytes */
+	u64 instr_type:32;
+
+	/** Pending list size (usually set to the sum of the size of all Input
+	 *  queues)
+	 */
+	u64 pending_list_size:32;
+
+	/* Max number of IQs available */
+	u64 max_iqs:8;
+#else
+	/* Max number of IQs available */
+	u64 max_iqs:8;
+
+	/** Pending list size (usually set to the sum of the size of all Input
+	 *  queues)
+	 */
+	u64 pending_list_size:32;
+
+	/** Command size - 32 or 64 bytes */
+	u64 instr_type:32;
+
+	/** Minimum number of commands pending to be posted to Octeon
+	 *  before driver hits the Input queue doorbell.
+	 */
+	u64 db_min:8;
+
+	/** Minimum ticks to wait before checking for pending instructions. */
+	u64 db_timeout:16;
+
+	/** Tx interrupt packets. Applicable to 23xx only */
+	u64 iq_intr_pkt:16;
+
+	u64 reserved:16;
+#endif
+};
+
+/** Structure to define the configuration attributes for each Output queue.
+ *  Applicable to all Octeon processors
+ **/
+struct octeon_oq_config {
+#ifdef __BIG_ENDIAN_BITFIELD
+	u64 reserved:16;
+
+	u64 pkts_per_intr:16;
+
+	/** Interrupt Coalescing (Time Interval). Octeon will interrupt the
+	 *  host if atleast one packet was sent in the time interval specified
+	 *  by this field. The driver uses time interval interrupt coalescing
+	 *  by default. The time is specified in microseconds.
+	 */
+	u64 oq_intr_time:16;
+
+	/** Interrupt Coalescing (Packet Count). Octeon will interrupt the host
+	 *  only if it sent as many packets as specified by this field.
+	 *  The driver
+	 *  usually does not use packet count interrupt coalescing.
+	 */
+	u64 oq_intr_pkt:16;
+
+	/** The number of buffers that were consumed during packet processing by
+	 *   the driver on this Output queue before the driver attempts to
+	 *   replenish
+	 *   the descriptor ring with new buffers.
+	 */
+	u64 refill_threshold:16;
+
+	/* Max number of OQs available */
+	u64 max_oqs:8;
+
+#else
+	/* Max number of OQs available */
+	u64 max_oqs:8;
+
+	/** The number of buffers that were consumed during packet processing by
+	 *   the driver on this Output queue before the driver attempts to
+	 *   replenish
+	 *   the descriptor ring with new buffers.
+	 */
+	u64 refill_threshold:16;
+
+	/** Interrupt Coalescing (Packet Count). Octeon will interrupt the host
+	 *  only if it sent as many packets as specified by this field.
+	 *  The driver
+	 *  usually does not use packet count interrupt coalescing.
+	 */
+	u64 oq_intr_pkt:16;
+
+	/** Interrupt Coalescing (Time Interval). Octeon will interrupt the
+	 *  host if atleast one packet was sent in the time interval specified
+	 *  by this field. The driver uses time interval interrupt coalescing
+	 *  by default.  The time is specified in microseconds.
+	 */
+	u64 oq_intr_time:16;
+
+	u64 pkts_per_intr:16;
+
+	u64 reserved:16;
+#endif
+
+};
+
+/** This structure conatins the NIC link configuration attributes,
+ *  common for all the OCTEON Modles.
+ */
+struct octeon_nic_if_config {
+#ifdef __BIG_ENDIAN_BITFIELD
+	u64 reserved:56;
+
+	u64 base_queue:16;
+
+	u64 gmx_port_id:8;
+
+	/* SKB size, We need not change buf size even for Jumbo frames.
+	 * Octeon can send jumbo frames in 4 consecutive descriptors,
+	 */
+	u64 rx_buf_size:16;
+
+	/* Num of desc for tx rings */
+	u64 num_tx_descs:16;
+
+	/* Num of desc for rx rings */
+	u64 num_rx_descs:16;
+
+	/* Actual configured value. Range could be: 1...max_rxqs */
+	u64 num_rxqs:16;
+
+	/* Max Rxqs: Half for each of the two ports :max_oq/2  */
+	u64 max_rxqs:16;
+
+	/* Actual configured value. Range could be: 1...max_txqs */
+	u64 num_txqs:16;
+
+	/* Max Txqs: Half for each of the two ports :max_iq/2 */
+	u64 max_txqs:16;
+#else
+	/* Max Txqs: Half for each of the two ports :max_iq/2 */
+	u64 max_txqs:16;
+
+	/* Actual configured value. Range could be: 1...max_txqs */
+	u64 num_txqs:16;
+
+	/* Max Rxqs: Half for each of the two ports :max_oq/2  */
+	u64 max_rxqs:16;
+
+	/* Actual configured value. Range could be: 1...max_rxqs */
+	u64 num_rxqs:16;
+
+	/* Num of desc for rx rings */
+	u64 num_rx_descs:16;
+
+	/* Num of desc for tx rings */
+	u64 num_tx_descs:16;
+
+	/* SKB size, We need not change buf size even for Jumbo frames.
+	 * Octeon can send jumbo frames in 4 consecutive descriptors,
+	 */
+	u64 rx_buf_size:16;
+
+	u64 gmx_port_id:8;
+
+	u64 base_queue:16;
+
+	u64 reserved:56;
+#endif
+
+};
+
+/** Structure to define the configuration attributes for meta data.
+ *  Applicable to all Octeon processors.
+ */
+
+struct octeon_misc_config {
+#ifdef __BIG_ENDIAN_BITFIELD
+	/** Host link status polling period */
+	u64 host_link_query_interval:32;
+	/** Oct link status polling period */
+	u64 oct_link_query_interval:32;
+
+	u64 enable_sli_oq_bp:1;
+	/** Control IQ Group */
+	u64 ctrlq_grp:4;
+#else
+	/** Control IQ Group */
+	u64 ctrlq_grp:4;
+	/** BP for SLI OQ */
+	u64 enable_sli_oq_bp:1;
+	/** Host link status polling period */
+	u64 oct_link_query_interval:32;
+	/** Oct link status polling period */
+	u64 host_link_query_interval:32;
+#endif
+};
+
+/** Structure to define the configuration for all OCTEON processors. */
+struct octeon_config {
+	u16 card_type;
+	char *card_name;
+
+	/** Input Queue attributes. */
+	struct octeon_iq_config iq;
+
+	/** Output Queue attributes. */
+	struct octeon_oq_config oq;
+
+	/** NIC Port Configuration */
+	struct octeon_nic_if_config nic_if_cfg[MAX_OCTEON_NICIF];
+
+	/** Miscellaneous attributes */
+	struct octeon_misc_config misc;
+
+	int num_nic_ports;
+
+	int num_def_tx_descs;
+
+	/* Num of desc for rx rings */
+	int num_def_rx_descs;
+
+	int def_rx_buf_size;
+
+};
+
+/* The following config values are fixed and should not be modified. */
+
+#define  BAR1_INDEX_DYNAMIC_MAP          2
+#define  BAR1_INDEX_STATIC_MAP          15
+#define  OCTEON_BAR1_ENTRY_SIZE         (4 * 1024 * 1024)
+
+#define  MAX_BAR1_IOREMAP_SIZE  (16 * OCTEON_BAR1_ENTRY_SIZE)
+
+/* Response lists - 1 ordered, 1 unordered-blocking, 1 unordered-nonblocking
+ *                  1 process done list, 1 zombie lists(timeouted sc list)
+ * NoResponse Lists are now maintained with each IQ. (Dec' 2007).
+ */
+#define MAX_RESPONSE_LISTS           6
+
+/* Opcode hash bits. The opcode is hashed on the lower 6-bits to lookup the
+ * dispatch table.
+ */
+#define OPCODE_MASK_BITS             6
+
+/* Mask for the 6-bit lookup hash */
+#define OCTEON_OPCODE_MASK           0x3f
+
+/* Size of the dispatch table. The 6-bit hash can index into 2^6 entries */
+#define DISPATCH_LIST_SIZE                      BIT(OPCODE_MASK_BITS)
+
+/* Maximum number of Octeon Instruction (command) queues */
+#define MAX_OCTEON_INSTR_QUEUES(oct)		\
+		(OCTEON_CN23XX_PF(oct) ? CN23XX_MAX_INPUT_QUEUES : \
+					CN6XXX_MAX_INPUT_QUEUES)
+
+/* Maximum number of Octeon Instruction (command) queues */
+#define MAX_OCTEON_OUTPUT_QUEUES(oct)		\
+		(OCTEON_CN23XX_PF(oct) ? CN23XX_MAX_OUTPUT_QUEUES : \
+					CN6XXX_MAX_OUTPUT_QUEUES)
+
+#define MAX_POSSIBLE_OCTEON_INSTR_QUEUES	CN23XX_MAX_INPUT_QUEUES
+#define MAX_POSSIBLE_OCTEON_OUTPUT_QUEUES	CN23XX_MAX_OUTPUT_QUEUES
+
+#define MAX_POSSIBLE_VFS			64
+
+#endif /* __OCTEON_CONFIG_H__  */

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_console.c

@@ -0,0 +1,919 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*
+ * @file octeon_console.c
+ */
+#include <linux/moduleparam.h>
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/crc32.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "liquidio_image.h"
+#include "octeon_mem_ops.h"
+
+static void octeon_remote_lock(void);
+static void octeon_remote_unlock(void);
+static u64 cvmx_bootmem_phy_named_block_find(struct octeon_device *oct,
+					     const char *name,
+					     u32 flags);
+static int octeon_console_read(struct octeon_device *oct, u32 console_num,
+			       char *buffer, u32 buf_size);
+
+#define BOOTLOADER_PCI_READ_BUFFER_DATA_ADDR    0x0006c008
+#define BOOTLOADER_PCI_READ_BUFFER_LEN_ADDR     0x0006c004
+#define BOOTLOADER_PCI_READ_BUFFER_OWNER_ADDR   0x0006c000
+#define BOOTLOADER_PCI_READ_DESC_ADDR           0x0006c100
+#define BOOTLOADER_PCI_WRITE_BUFFER_STR_LEN     248
+
+#define OCTEON_PCI_IO_BUF_OWNER_OCTEON    0x00000001
+#define OCTEON_PCI_IO_BUF_OWNER_HOST      0x00000002
+
+/** Can change without breaking ABI */
+#define CVMX_BOOTMEM_NUM_NAMED_BLOCKS 64
+
+/** minimum alignment of bootmem alloced blocks */
+#define CVMX_BOOTMEM_ALIGNMENT_SIZE     (16ull)
+
+/** CVMX bootmem descriptor major version */
+#define CVMX_BOOTMEM_DESC_MAJ_VER   3
+/* CVMX bootmem descriptor minor version */
+#define CVMX_BOOTMEM_DESC_MIN_VER   0
+
+/* Current versions */
+#define OCTEON_PCI_CONSOLE_MAJOR_VERSION    1
+#define OCTEON_PCI_CONSOLE_MINOR_VERSION    0
+#define OCTEON_PCI_CONSOLE_BLOCK_NAME   "__pci_console"
+#define OCTEON_CONSOLE_POLL_INTERVAL_MS  100    /* 10 times per second */
+
+/* First three members of cvmx_bootmem_desc are left in original
+ * positions for backwards compatibility.
+ * Assumes big endian target
+ */
+struct cvmx_bootmem_desc {
+	/** spinlock to control access to list */
+	u32 lock;
+
+	/** flags for indicating various conditions */
+	u32 flags;
+
+	u64 head_addr;
+
+	/** incremented changed when incompatible changes made */
+	u32 major_version;
+
+	/** incremented changed when compatible changes made,
+	 *  reset to zero when major incremented
+	 */
+	u32 minor_version;
+
+	u64 app_data_addr;
+	u64 app_data_size;
+
+	/** number of elements in named blocks array */
+	u32 nb_num_blocks;
+
+	/** length of name array in bootmem blocks */
+	u32 named_block_name_len;
+
+	/** address of named memory block descriptors */
+	u64 named_block_array_addr;
+};
+
+/* Structure that defines a single console.
+ *
+ * Note: when read_index == write_index, the buffer is empty.
+ * The actual usable size of each console is console_buf_size -1;
+ */
+struct octeon_pci_console {
+	u64 input_base_addr;
+	u32 input_read_index;
+	u32 input_write_index;
+	u64 output_base_addr;
+	u32 output_read_index;
+	u32 output_write_index;
+	u32 lock;
+	u32 buf_size;
+};
+
+/* This is the main container structure that contains all the information
+ * about all PCI consoles.  The address of this structure is passed to various
+ * routines that operation on PCI consoles.
+ */
+struct octeon_pci_console_desc {
+	u32 major_version;
+	u32 minor_version;
+	u32 lock;
+	u32 flags;
+	u32 num_consoles;
+	u32 pad;
+	/* must be 64 bit aligned here... */
+	/* Array of addresses of octeon_pci_console structures */
+	u64 console_addr_array[];
+	/* Implicit storage for console_addr_array */
+};
+
+/*
+ * This function is the implementation of the get macros defined
+ * for individual structure members. The argument are generated
+ * by the macros inorder to read only the needed memory.
+ *
+ * @param oct    Pointer to current octeon device
+ * @param base   64bit physical address of the complete structure
+ * @param offset Offset from the beginning of the structure to the member being
+ *               accessed.
+ * @param size   Size of the structure member.
+ *
+ * @return Value of the structure member promoted into a u64.
+ */
+static inline u64 __cvmx_bootmem_desc_get(struct octeon_device *oct,
+					  u64 base,
+					  u32 offset,
+					  u32 size)
+{
+	base = (1ull << 63) | (base + offset);
+	switch (size) {
+	case 4:
+		return octeon_read_device_mem32(oct, base);
+	case 8:
+		return octeon_read_device_mem64(oct, base);
+	default:
+		return 0;
+	}
+}
+
+/*
+ * This function retrieves the string name of a named block. It is
+ * more complicated than a simple memcpy() since the named block
+ * descriptor may not be directly accessible.
+ *
+ * @param addr   Physical address of the named block descriptor
+ * @param str    String to receive the named block string name
+ * @param len    Length of the string buffer, which must match the length
+ *               stored in the bootmem descriptor.
+ */
+static void CVMX_BOOTMEM_NAMED_GET_NAME(struct octeon_device *oct,
+					u64 addr,
+					char *str,
+					u32 len)
+{
+	addr += offsetof(struct cvmx_bootmem_named_block_desc, name);
+	octeon_pci_read_core_mem(oct, addr, (u8 *)str, len);
+	str[len] = 0;
+}
+
+/* See header file for descriptions of functions */
+
+/*
+ * Check the version information on the bootmem descriptor
+ *
+ * @param exact_match
+ *               Exact major version to check against. A zero means
+ *               check that the version supports named blocks.
+ *
+ * @return Zero if the version is correct. Negative if the version is
+ *         incorrect. Failures also cause a message to be displayed.
+ */
+static int __cvmx_bootmem_check_version(struct octeon_device *oct,
+					u32 exact_match)
+{
+	u32 major_version;
+	u32 minor_version;
+
+	if (!oct->bootmem_desc_addr)
+		oct->bootmem_desc_addr =
+			octeon_read_device_mem64(oct,
+						 BOOTLOADER_PCI_READ_DESC_ADDR);
+	major_version = (u32)__cvmx_bootmem_desc_get(
+			oct, oct->bootmem_desc_addr,
+			offsetof(struct cvmx_bootmem_desc, major_version),
+			sizeof_field(struct cvmx_bootmem_desc, major_version));
+	minor_version = (u32)__cvmx_bootmem_desc_get(
+			oct, oct->bootmem_desc_addr,
+			offsetof(struct cvmx_bootmem_desc, minor_version),
+			sizeof_field(struct cvmx_bootmem_desc, minor_version));
+
+	dev_dbg(&oct->pci_dev->dev, "%s: major_version=%d\n", __func__,
+		major_version);
+	if ((major_version > 3) ||
+	    (exact_match && major_version != exact_match)) {
+		dev_err(&oct->pci_dev->dev, "bootmem ver mismatch %d.%d addr:0x%llx\n",
+			major_version, minor_version,
+			(long long)oct->bootmem_desc_addr);
+		return -1;
+	} else {
+		return 0;
+	}
+}
+
+static const struct cvmx_bootmem_named_block_desc
+*__cvmx_bootmem_find_named_block_flags(struct octeon_device *oct,
+					const char *name, u32 flags)
+{
+	struct cvmx_bootmem_named_block_desc *desc =
+		&oct->bootmem_named_block_desc;
+	u64 named_addr = cvmx_bootmem_phy_named_block_find(oct, name, flags);
+
+	if (named_addr) {
+		desc->base_addr = __cvmx_bootmem_desc_get(
+				oct, named_addr,
+				offsetof(struct cvmx_bootmem_named_block_desc,
+					 base_addr),
+				sizeof_field(
+					struct cvmx_bootmem_named_block_desc,
+					base_addr));
+		desc->size = __cvmx_bootmem_desc_get(oct, named_addr,
+				offsetof(struct cvmx_bootmem_named_block_desc,
+					 size),
+				sizeof_field(
+					struct cvmx_bootmem_named_block_desc,
+					size));
+
+		strscpy(desc->name, name, sizeof(desc->name));
+		return &oct->bootmem_named_block_desc;
+	} else {
+		return NULL;
+	}
+}
+
+static u64 cvmx_bootmem_phy_named_block_find(struct octeon_device *oct,
+					     const char *name,
+					     u32 flags)
+{
+	u64 result = 0;
+
+	if (!__cvmx_bootmem_check_version(oct, 3)) {
+		u32 i;
+
+		u64 named_block_array_addr = __cvmx_bootmem_desc_get(
+					oct, oct->bootmem_desc_addr,
+					offsetof(struct cvmx_bootmem_desc,
+						 named_block_array_addr),
+					sizeof_field(struct cvmx_bootmem_desc,
+						     named_block_array_addr));
+		u32 num_blocks = (u32)__cvmx_bootmem_desc_get(
+					oct, oct->bootmem_desc_addr,
+					offsetof(struct cvmx_bootmem_desc,
+						 nb_num_blocks),
+					sizeof_field(struct cvmx_bootmem_desc,
+						     nb_num_blocks));
+
+		u32 name_length = (u32)__cvmx_bootmem_desc_get(
+					oct, oct->bootmem_desc_addr,
+					offsetof(struct cvmx_bootmem_desc,
+						 named_block_name_len),
+					sizeof_field(struct cvmx_bootmem_desc,
+						     named_block_name_len));
+
+		u64 named_addr = named_block_array_addr;
+
+		for (i = 0; i < num_blocks; i++) {
+			u64 named_size = __cvmx_bootmem_desc_get(
+					oct, named_addr,
+					 offsetof(
+					struct cvmx_bootmem_named_block_desc,
+					size),
+					 sizeof_field(
+					struct cvmx_bootmem_named_block_desc,
+					size));
+
+			if (name && named_size) {
+				char *name_tmp =
+					kmalloc(name_length + 1, GFP_KERNEL);
+				if (!name_tmp)
+					break;
+
+				CVMX_BOOTMEM_NAMED_GET_NAME(oct, named_addr,
+							    name_tmp,
+							    name_length);
+				if (!strncmp(name, name_tmp, name_length)) {
+					result = named_addr;
+					kfree(name_tmp);
+					break;
+				}
+				kfree(name_tmp);
+			} else if (!name && !named_size) {
+				result = named_addr;
+				break;
+			}
+
+			named_addr +=
+				sizeof(struct cvmx_bootmem_named_block_desc);
+		}
+	}
+	return result;
+}
+
+/*
+ * Find a named block on the remote Octeon
+ *
+ * @param name      Name of block to find
+ * @param base_addr Address the block is at (OUTPUT)
+ * @param size      The size of the block (OUTPUT)
+ *
+ * @return Zero on success, One on failure.
+ */
+static int octeon_named_block_find(struct octeon_device *oct, const char *name,
+				   u64 *base_addr, u64 *size)
+{
+	const struct cvmx_bootmem_named_block_desc *named_block;
+
+	octeon_remote_lock();
+	named_block = __cvmx_bootmem_find_named_block_flags(oct, name, 0);
+	octeon_remote_unlock();
+	if (named_block) {
+		*base_addr = named_block->base_addr;
+		*size = named_block->size;
+		return 0;
+	}
+	return 1;
+}
+
+static void octeon_remote_lock(void)
+{
+	/* fill this in if any sharing is needed */
+}
+
+static void octeon_remote_unlock(void)
+{
+	/* fill this in if any sharing is needed */
+}
+
+int octeon_console_send_cmd(struct octeon_device *oct, char *cmd_str,
+			    u32 wait_hundredths)
+{
+	u32 len = (u32)strlen(cmd_str);
+
+	dev_dbg(&oct->pci_dev->dev, "sending \"%s\" to bootloader\n", cmd_str);
+
+	if (len > BOOTLOADER_PCI_WRITE_BUFFER_STR_LEN - 1) {
+		dev_err(&oct->pci_dev->dev, "Command string too long, max length is: %d\n",
+			BOOTLOADER_PCI_WRITE_BUFFER_STR_LEN - 1);
+		return -1;
+	}
+
+	if (octeon_wait_for_bootloader(oct, wait_hundredths) != 0) {
+		dev_err(&oct->pci_dev->dev, "Bootloader not ready for command.\n");
+		return -1;
+	}
+
+	/* Write command to bootloader */
+	octeon_remote_lock();
+	octeon_pci_write_core_mem(oct, BOOTLOADER_PCI_READ_BUFFER_DATA_ADDR,
+				  (u8 *)cmd_str, len);
+	octeon_write_device_mem32(oct, BOOTLOADER_PCI_READ_BUFFER_LEN_ADDR,
+				  len);
+	octeon_write_device_mem32(oct, BOOTLOADER_PCI_READ_BUFFER_OWNER_ADDR,
+				  OCTEON_PCI_IO_BUF_OWNER_OCTEON);
+
+	/* Bootloader should accept command very quickly
+	 * if it really was ready
+	 */
+	if (octeon_wait_for_bootloader(oct, 200) != 0) {
+		octeon_remote_unlock();
+		dev_err(&oct->pci_dev->dev, "Bootloader did not accept command.\n");
+		return -1;
+	}
+	octeon_remote_unlock();
+	return 0;
+}
+
+int octeon_wait_for_bootloader(struct octeon_device *oct,
+			       u32 wait_time_hundredths)
+{
+	dev_dbg(&oct->pci_dev->dev, "waiting %d0 ms for bootloader\n",
+		wait_time_hundredths);
+
+	if (octeon_mem_access_ok(oct))
+		return -1;
+
+	while (wait_time_hundredths > 0 &&
+	       octeon_read_device_mem32(oct,
+					BOOTLOADER_PCI_READ_BUFFER_OWNER_ADDR)
+	       != OCTEON_PCI_IO_BUF_OWNER_HOST) {
+		if (--wait_time_hundredths <= 0)
+			return -1;
+		schedule_timeout_uninterruptible(HZ / 100);
+	}
+	return 0;
+}
+
+static void octeon_console_handle_result(struct octeon_device *oct,
+					 size_t console_num)
+{
+	struct octeon_console *console;
+
+	console = &oct->console[console_num];
+
+	console->waiting = 0;
+}
+
+static char console_buffer[OCTEON_CONSOLE_MAX_READ_BYTES];
+
+static void output_console_line(struct octeon_device *oct,
+				struct octeon_console *console,
+				size_t console_num,
+				char *console_buffer,
+				s32 bytes_read)
+{
+	char *line;
+	s32 i;
+	size_t len;
+
+	line = console_buffer;
+	for (i = 0; i < bytes_read; i++) {
+		/* Output a line at a time, prefixed */
+		if (console_buffer[i] == '\n') {
+			console_buffer[i] = '\0';
+			/* We need to output 'line', prefaced by 'leftover'.
+			 * However, it is possible we're being called to
+			 * output 'leftover' by itself (in the case of nothing
+			 * having been read from the console).
+			 *
+			 * To avoid duplication, check for this condition.
+			 */
+			if (console->leftover[0] &&
+			    (line != console->leftover)) {
+				if (console->print)
+					(*console->print)(oct, (u32)console_num,
+							  console->leftover,
+							  line);
+				console->leftover[0] = '\0';
+			} else {
+				if (console->print)
+					(*console->print)(oct, (u32)console_num,
+							  line, NULL);
+			}
+			line = &console_buffer[i + 1];
+		}
+	}
+
+	/* Save off any leftovers */
+	if (line != &console_buffer[bytes_read]) {
+		console_buffer[bytes_read] = '\0';
+		len = strlen(console->leftover);
+		strscpy(&console->leftover[len], line,
+			sizeof(console->leftover) - len + 1);
+	}
+}
+
+static void check_console(struct work_struct *work)
+{
+	s32 bytes_read, tries, total_read;
+	size_t len;
+	struct octeon_console *console;
+	struct cavium_wk *wk = (struct cavium_wk *)work;
+	struct octeon_device *oct = (struct octeon_device *)wk->ctxptr;
+	u32 console_num = (u32)wk->ctxul;
+	u32 delay;
+
+	console = &oct->console[console_num];
+	tries = 0;
+	total_read = 0;
+
+	do {
+		/* Take console output regardless of whether it will
+		 * be logged
+		 */
+		bytes_read =
+			octeon_console_read(oct, console_num, console_buffer,
+					    sizeof(console_buffer) - 1);
+		if (bytes_read > 0) {
+			total_read += bytes_read;
+			if (console->waiting)
+				octeon_console_handle_result(oct, console_num);
+			if (console->print) {
+				output_console_line(oct, console, console_num,
+						    console_buffer, bytes_read);
+			}
+		} else if (bytes_read < 0) {
+			dev_err(&oct->pci_dev->dev, "Error reading console %u, ret=%d\n",
+				console_num, bytes_read);
+		}
+
+		tries++;
+	} while ((bytes_read > 0) && (tries < 16));
+
+	/* If nothing is read after polling the console,
+	 * output any leftovers if any
+	 */
+	if (console->print && (total_read == 0) &&
+	    (console->leftover[0])) {
+		/* append '\n' as terminator for 'output_console_line' */
+		len = strlen(console->leftover);
+		console->leftover[len] = '\n';
+		output_console_line(oct, console, console_num,
+				    console->leftover, (s32)(len + 1));
+		console->leftover[0] = '\0';
+	}
+
+	delay = OCTEON_CONSOLE_POLL_INTERVAL_MS;
+
+	schedule_delayed_work(&wk->work, msecs_to_jiffies(delay));
+}
+
+int octeon_init_consoles(struct octeon_device *oct)
+{
+	int ret = 0;
+	u64 addr, size;
+
+	ret = octeon_mem_access_ok(oct);
+	if (ret) {
+		dev_err(&oct->pci_dev->dev, "Memory access not okay'\n");
+		return ret;
+	}
+
+	ret = octeon_named_block_find(oct, OCTEON_PCI_CONSOLE_BLOCK_NAME, &addr,
+				      &size);
+	if (ret) {
+		dev_err(&oct->pci_dev->dev, "Could not find console '%s'\n",
+			OCTEON_PCI_CONSOLE_BLOCK_NAME);
+		return ret;
+	}
+
+	/* Dedicate one of Octeon's BAR1 index registers to create a static
+	 * mapping to a region of Octeon DRAM that contains the PCI console
+	 * named block.
+	 */
+	oct->console_nb_info.bar1_index = BAR1_INDEX_STATIC_MAP;
+	oct->fn_list.bar1_idx_setup(oct, addr, oct->console_nb_info.bar1_index,
+				    true);
+	oct->console_nb_info.dram_region_base = addr
+		& ~(OCTEON_BAR1_ENTRY_SIZE - 1ULL);
+
+	/* num_consoles > 0, is an indication that the consoles
+	 * are accessible
+	 */
+	oct->num_consoles = octeon_read_device_mem32(oct,
+		addr + offsetof(struct octeon_pci_console_desc,
+			num_consoles));
+	oct->console_desc_addr = addr;
+
+	dev_dbg(&oct->pci_dev->dev, "Initialized consoles. %d available\n",
+		oct->num_consoles);
+
+	return ret;
+}
+
+static void octeon_get_uboot_version(struct octeon_device *oct)
+{
+	s32 bytes_read, tries, total_read;
+	struct octeon_console *console;
+	u32 console_num = 0;
+	char *uboot_ver;
+	char *buf;
+	char *p;
+
+#define OCTEON_UBOOT_VER_BUF_SIZE 512
+	buf = kmalloc(OCTEON_UBOOT_VER_BUF_SIZE, GFP_KERNEL);
+	if (!buf)
+		return;
+
+	if (octeon_console_send_cmd(oct, "setenv stdout pci\n", 50)) {
+		kfree(buf);
+		return;
+	}
+
+	if (octeon_console_send_cmd(oct, "version\n", 1)) {
+		kfree(buf);
+		return;
+	}
+
+	console = &oct->console[console_num];
+	tries = 0;
+	total_read = 0;
+
+	do {
+		/* Take console output regardless of whether it will
+		 * be logged
+		 */
+		bytes_read =
+			octeon_console_read(oct,
+					    console_num, buf + total_read,
+					    OCTEON_UBOOT_VER_BUF_SIZE - 1 -
+					    total_read);
+		if (bytes_read > 0) {
+			buf[bytes_read] = '\0';
+
+			total_read += bytes_read;
+			if (console->waiting)
+				octeon_console_handle_result(oct, console_num);
+		} else if (bytes_read < 0) {
+			dev_err(&oct->pci_dev->dev, "Error reading console %u, ret=%d\n",
+				console_num, bytes_read);
+		}
+
+		tries++;
+	} while ((bytes_read > 0) && (tries < 16));
+
+	/* If nothing is read after polling the console,
+	 * output any leftovers if any
+	 */
+	if ((total_read == 0) && (console->leftover[0])) {
+		dev_dbg(&oct->pci_dev->dev, "%u: %s\n",
+			console_num, console->leftover);
+		console->leftover[0] = '\0';
+	}
+
+	buf[OCTEON_UBOOT_VER_BUF_SIZE - 1] = '\0';
+
+	uboot_ver = strstr(buf, "U-Boot");
+	if (uboot_ver) {
+		p = strstr(uboot_ver, "mips");
+		if (p) {
+			p--;
+			*p = '\0';
+			dev_info(&oct->pci_dev->dev, "%s\n", uboot_ver);
+		}
+	}
+
+	kfree(buf);
+	octeon_console_send_cmd(oct, "setenv stdout serial\n", 50);
+}
+
+int octeon_add_console(struct octeon_device *oct, u32 console_num,
+		       char *dbg_enb)
+{
+	int ret = 0;
+	u32 delay;
+	u64 coreaddr;
+	struct delayed_work *work;
+	struct octeon_console *console;
+
+	if (console_num >= oct->num_consoles) {
+		dev_err(&oct->pci_dev->dev,
+			"trying to read from console number %d when only 0 to %d exist\n",
+			console_num, oct->num_consoles);
+	} else {
+		console = &oct->console[console_num];
+
+		console->waiting = 0;
+
+		coreaddr = oct->console_desc_addr + console_num * 8 +
+			offsetof(struct octeon_pci_console_desc,
+				 console_addr_array);
+		console->addr = octeon_read_device_mem64(oct, coreaddr);
+		coreaddr = console->addr + offsetof(struct octeon_pci_console,
+						    buf_size);
+		console->buffer_size = octeon_read_device_mem32(oct, coreaddr);
+		coreaddr = console->addr + offsetof(struct octeon_pci_console,
+						    input_base_addr);
+		console->input_base_addr =
+			octeon_read_device_mem64(oct, coreaddr);
+		coreaddr = console->addr + offsetof(struct octeon_pci_console,
+						    output_base_addr);
+		console->output_base_addr =
+			octeon_read_device_mem64(oct, coreaddr);
+		console->leftover[0] = '\0';
+
+		work = &oct->console_poll_work[console_num].work;
+
+		octeon_get_uboot_version(oct);
+
+		INIT_DELAYED_WORK(work, check_console);
+		oct->console_poll_work[console_num].ctxptr = (void *)oct;
+		oct->console_poll_work[console_num].ctxul = console_num;
+		delay = OCTEON_CONSOLE_POLL_INTERVAL_MS;
+		schedule_delayed_work(work, msecs_to_jiffies(delay));
+
+		/* an empty string means use default debug console enablement */
+		if (dbg_enb && !dbg_enb[0])
+			dbg_enb = "setenv pci_console_active 1";
+		if (dbg_enb)
+			ret = octeon_console_send_cmd(oct, dbg_enb, 2000);
+
+		console->active = 1;
+	}
+
+	return ret;
+}
+
+/*
+ * Removes all consoles
+ *
+ * @param oct         octeon device
+ */
+void octeon_remove_consoles(struct octeon_device *oct)
+{
+	u32 i;
+	struct octeon_console *console;
+
+	for (i = 0; i < oct->num_consoles; i++) {
+		console = &oct->console[i];
+
+		if (!console->active)
+			continue;
+
+		cancel_delayed_work_sync(&oct->console_poll_work[i].
+						work);
+		console->addr = 0;
+		console->buffer_size = 0;
+		console->input_base_addr = 0;
+		console->output_base_addr = 0;
+	}
+
+	oct->num_consoles = 0;
+}
+
+static inline int octeon_console_free_bytes(u32 buffer_size,
+					    u32 wr_idx,
+					    u32 rd_idx)
+{
+	if (rd_idx >= buffer_size || wr_idx >= buffer_size)
+		return -1;
+
+	return ((buffer_size - 1) - (wr_idx - rd_idx)) % buffer_size;
+}
+
+static inline int octeon_console_avail_bytes(u32 buffer_size,
+					     u32 wr_idx,
+					     u32 rd_idx)
+{
+	if (rd_idx >= buffer_size || wr_idx >= buffer_size)
+		return -1;
+
+	return buffer_size - 1 -
+	       octeon_console_free_bytes(buffer_size, wr_idx, rd_idx);
+}
+
+static int octeon_console_read(struct octeon_device *oct, u32 console_num,
+			       char *buffer, u32 buf_size)
+{
+	int bytes_to_read;
+	u32 rd_idx, wr_idx;
+	struct octeon_console *console;
+
+	if (console_num >= oct->num_consoles) {
+		dev_err(&oct->pci_dev->dev, "Attempted to read from disabled console %d\n",
+			console_num);
+		return 0;
+	}
+
+	console = &oct->console[console_num];
+
+	/* Check to see if any data is available.
+	 * Maybe optimize this with 64-bit read.
+	 */
+	rd_idx = octeon_read_device_mem32(oct, console->addr +
+		offsetof(struct octeon_pci_console, output_read_index));
+	wr_idx = octeon_read_device_mem32(oct, console->addr +
+		offsetof(struct octeon_pci_console, output_write_index));
+
+	bytes_to_read = octeon_console_avail_bytes(console->buffer_size,
+						   wr_idx, rd_idx);
+	if (bytes_to_read <= 0)
+		return bytes_to_read;
+
+	bytes_to_read = min_t(s32, bytes_to_read, buf_size);
+
+	/* Check to see if what we want to read is not contiguous, and limit
+	 * ourselves to the contiguous block
+	 */
+	if (rd_idx + bytes_to_read >= console->buffer_size)
+		bytes_to_read = console->buffer_size - rd_idx;
+
+	octeon_pci_read_core_mem(oct, console->output_base_addr + rd_idx,
+				 (u8 *)buffer, bytes_to_read);
+	octeon_write_device_mem32(oct, console->addr +
+				  offsetof(struct octeon_pci_console,
+					   output_read_index),
+				  (rd_idx + bytes_to_read) %
+				  console->buffer_size);
+
+	return bytes_to_read;
+}
+
+#define FBUF_SIZE	(4 * 1024 * 1024)
+#define MAX_BOOTTIME_SIZE    80
+
+int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
+			     size_t size)
+{
+	struct octeon_firmware_file_header *h;
+	char boottime[MAX_BOOTTIME_SIZE];
+	struct timespec64 ts;
+	u32 crc32_result;
+	u64 load_addr;
+	u32 image_len;
+	int ret = 0;
+	u32 i, rem;
+
+	if (size < sizeof(struct octeon_firmware_file_header)) {
+		dev_err(&oct->pci_dev->dev, "Firmware file too small (%d < %d).\n",
+			(u32)size,
+			(u32)sizeof(struct octeon_firmware_file_header));
+		return -EINVAL;
+	}
+
+	h = (struct octeon_firmware_file_header *)data;
+
+	if (be32_to_cpu(h->magic) != LIO_NIC_MAGIC) {
+		dev_err(&oct->pci_dev->dev, "Unrecognized firmware file.\n");
+		return -EINVAL;
+	}
+
+	crc32_result = crc32((unsigned int)~0, data,
+			     sizeof(struct octeon_firmware_file_header) -
+			     sizeof(u32)) ^ ~0U;
+	if (crc32_result != be32_to_cpu(h->crc32)) {
+		dev_err(&oct->pci_dev->dev, "Firmware CRC mismatch (0x%08x != 0x%08x).\n",
+			crc32_result, be32_to_cpu(h->crc32));
+		return -EINVAL;
+	}
+
+	if (memcmp(LIQUIDIO_BASE_VERSION, h->version,
+		   strlen(LIQUIDIO_BASE_VERSION))) {
+		dev_err(&oct->pci_dev->dev, "Unmatched firmware version. Expected %s.x, got %s.\n",
+			LIQUIDIO_BASE_VERSION,
+			h->version);
+		return -EINVAL;
+	}
+
+	if (be32_to_cpu(h->num_images) > LIO_MAX_IMAGES) {
+		dev_err(&oct->pci_dev->dev, "Too many images in firmware file (%d).\n",
+			be32_to_cpu(h->num_images));
+		return -EINVAL;
+	}
+
+	dev_info(&oct->pci_dev->dev, "Firmware version: %s\n", h->version);
+	snprintf(oct->fw_info.liquidio_firmware_version, 32, "LIQUIDIO: %s",
+		 h->version);
+
+	data += sizeof(struct octeon_firmware_file_header);
+
+	dev_info(&oct->pci_dev->dev, "%s: Loading %d images\n", __func__,
+		 be32_to_cpu(h->num_images));
+	/* load all images */
+	for (i = 0; i < be32_to_cpu(h->num_images); i++) {
+		load_addr = be64_to_cpu(h->desc[i].addr);
+		image_len = be32_to_cpu(h->desc[i].len);
+
+		dev_info(&oct->pci_dev->dev, "Loading firmware %d at %llx\n",
+			 image_len, load_addr);
+
+		/* Write in 4MB chunks*/
+		rem = image_len;
+
+		while (rem) {
+			if (rem < FBUF_SIZE)
+				size = rem;
+			else
+				size = FBUF_SIZE;
+
+			/* download the image */
+			octeon_pci_write_core_mem(oct, load_addr, data, (u32)size);
+
+			data += size;
+			rem -= (u32)size;
+			load_addr += size;
+		}
+	}
+
+	/* Pass date and time information to NIC at the time of loading
+	 * firmware and periodically update the host time to NIC firmware.
+	 * This is to make NIC firmware use the same time reference as Host,
+	 * so that it is easy to correlate logs from firmware and host for
+	 * debugging.
+	 *
+	 * Octeon always uses UTC time. so timezone information is not sent.
+	 */
+	ktime_get_real_ts64(&ts);
+	ret = snprintf(boottime, MAX_BOOTTIME_SIZE,
+		       " time_sec=%lld time_nsec=%ld",
+		       (s64)ts.tv_sec, ts.tv_nsec);
+	if ((sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd))) <
+		ret) {
+		dev_err(&oct->pci_dev->dev, "Boot command buffer too small\n");
+		return -EINVAL;
+	}
+	strncat(h->bootcmd, boottime,
+		sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd)));
+
+	dev_info(&oct->pci_dev->dev, "Writing boot command: %s\n",
+		 h->bootcmd);
+
+	/* Invoke the bootcmd */
+	ret = octeon_console_send_cmd(oct, h->bootcmd, 50);
+	if (ret)
+		dev_info(&oct->pci_dev->dev, "Boot command send failed\n");
+
+	return ret;
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_device.h

@@ -0,0 +1,894 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file octeon_device.h
+ *  \brief Host Driver: This file defines the octeon device structure.
+ */
+
+#ifndef _OCTEON_DEVICE_H_
+#define  _OCTEON_DEVICE_H_
+
+#include <linux/interrupt.h>
+#include <net/devlink.h>
+
+/** PCI VendorId Device Id */
+#define  OCTEON_CN68XX_PCIID          0x91177d
+#define  OCTEON_CN66XX_PCIID          0x92177d
+#define  OCTEON_CN23XX_PCIID_PF       0x9702177d
+/** Driver identifies chips by these Ids, created by clubbing together
+ *  DeviceId+RevisionId; Where Revision Id is not used to distinguish
+ *  between chips, a value of 0 is used for revision id.
+ */
+#define  OCTEON_CN68XX                0x0091
+#define  OCTEON_CN66XX                0x0092
+#define  OCTEON_CN23XX_PF_VID         0x9702
+#define  OCTEON_CN23XX_VF_VID         0x9712
+
+/**RevisionId for the chips */
+#define  OCTEON_CN23XX_REV_1_0        0x00
+#define  OCTEON_CN23XX_REV_1_1        0x01
+#define  OCTEON_CN23XX_REV_2_0        0x80
+
+/**SubsystemId for the chips */
+#define	 OCTEON_CN2350_10GB_SUBSYS_ID_1	0X3177d
+#define	 OCTEON_CN2350_10GB_SUBSYS_ID_2	0X4177d
+#define	 OCTEON_CN2360_10GB_SUBSYS_ID	0X5177d
+#define	 OCTEON_CN2350_25GB_SUBSYS_ID	0X7177d
+#define	 OCTEON_CN2360_25GB_SUBSYS_ID	0X6177d
+
+/** Endian-swap modes supported by Octeon. */
+enum octeon_pci_swap_mode {
+	OCTEON_PCI_PASSTHROUGH = 0,
+	OCTEON_PCI_64BIT_SWAP = 1,
+	OCTEON_PCI_32BIT_BYTE_SWAP = 2,
+	OCTEON_PCI_32BIT_LW_SWAP = 3
+};
+
+enum lio_fw_state {
+	FW_IS_PRELOADED = 0,
+	FW_NEEDS_TO_BE_LOADED = 1,
+	FW_IS_BEING_LOADED = 2,
+	FW_HAS_BEEN_LOADED = 3,
+};
+
+enum {
+	OCTEON_CONFIG_TYPE_DEFAULT = 0,
+	NUM_OCTEON_CONFS,
+};
+
+#define  OCTEON_INPUT_INTR    (1)
+#define  OCTEON_OUTPUT_INTR   (2)
+#define  OCTEON_MBOX_INTR     (4)
+#define  OCTEON_ALL_INTR      0xff
+
+/*---------------   PCI BAR1 index registers -------------*/
+
+/* BAR1 Mask */
+#define    PCI_BAR1_ENABLE_CA            1
+#define    PCI_BAR1_ENDIAN_MODE          OCTEON_PCI_64BIT_SWAP
+#define    PCI_BAR1_ENTRY_VALID          1
+#define    PCI_BAR1_MASK                 ((PCI_BAR1_ENABLE_CA << 3)   \
+					    | (PCI_BAR1_ENDIAN_MODE << 1) \
+					    | PCI_BAR1_ENTRY_VALID)
+
+/** Octeon Device state.
+ *  Each octeon device goes through each of these states
+ *  as it is initialized.
+ */
+#define    OCT_DEV_BEGIN_STATE            0x0
+#define    OCT_DEV_PCI_ENABLE_DONE        0x1
+#define    OCT_DEV_PCI_MAP_DONE           0x2
+#define    OCT_DEV_DISPATCH_INIT_DONE     0x3
+#define    OCT_DEV_INSTR_QUEUE_INIT_DONE  0x4
+#define    OCT_DEV_SC_BUFF_POOL_INIT_DONE 0x5
+#define    OCT_DEV_RESP_LIST_INIT_DONE    0x6
+#define    OCT_DEV_DROQ_INIT_DONE         0x7
+#define    OCT_DEV_MBOX_SETUP_DONE        0x8
+#define    OCT_DEV_MSIX_ALLOC_VECTOR_DONE 0x9
+#define    OCT_DEV_INTR_SET_DONE          0xa
+#define    OCT_DEV_IO_QUEUES_DONE         0xb
+#define    OCT_DEV_CONSOLE_INIT_DONE      0xc
+#define    OCT_DEV_HOST_OK                0xd
+#define    OCT_DEV_CORE_OK                0xe
+#define    OCT_DEV_RUNNING                0xf
+#define    OCT_DEV_IN_RESET               0x10
+#define    OCT_DEV_STATE_INVALID          0x11
+
+#define    OCT_DEV_STATES                 OCT_DEV_STATE_INVALID
+
+/** Octeon Device interrupts
+ * These interrupt bits are set in int_status filed of
+ * octeon_device structure
+ */
+#define	   OCT_DEV_INTR_DMA0_FORCE	  0x01
+#define	   OCT_DEV_INTR_DMA1_FORCE	  0x02
+#define	   OCT_DEV_INTR_PKT_DATA	  0x04
+
+#define LIO_RESET_SECS (3)
+
+/*---------------------------DISPATCH LIST-------------------------------*/
+
+/** The dispatch list entry.
+ *  The driver keeps a record of functions registered for each
+ *  response header opcode in this structure. Since the opcode is
+ *  hashed to index into the driver's list, more than one opcode
+ *  can hash to the same entry, in which case the list field points
+ *  to a linked list with the other entries.
+ */
+struct octeon_dispatch {
+	/** List head for this entry */
+	struct list_head list;
+
+	/** The opcode for which the dispatch function & arg should be used */
+	u16 opcode;
+
+	/** The function to be called for a packet received by the driver */
+	octeon_dispatch_fn_t dispatch_fn;
+
+	/* The application specified argument to be passed to the above
+	 * function along with the received packet
+	 */
+	void *arg;
+};
+
+/** The dispatch list structure. */
+struct octeon_dispatch_list {
+	/** access to dispatch list must be atomic */
+	spinlock_t lock;
+
+	/** Count of dispatch functions currently registered */
+	u32 count;
+
+	/** The list of dispatch functions */
+	struct octeon_dispatch *dlist;
+};
+
+/*-----------------------  THE OCTEON DEVICE  ---------------------------*/
+
+#define OCT_MEM_REGIONS     3
+/** PCI address space mapping information.
+ *  Each of the 3 address spaces given by BAR0, BAR2 and BAR4 of
+ *  Octeon gets mapped to different physical address spaces in
+ *  the kernel.
+ */
+struct octeon_mmio {
+	/** PCI address to which the BAR is mapped. */
+	u64 start;
+
+	/** Length of this PCI address space. */
+	u32 len;
+
+	/** Length that has been mapped to phys. address space. */
+	u32 mapped_len;
+
+	/** The physical address to which the PCI address space is mapped. */
+	u8 __iomem *hw_addr;
+
+	/** Flag indicating the mapping was successful. */
+	u32 done;
+};
+
+#define   MAX_OCTEON_MAPS    32
+
+struct octeon_io_enable {
+	u64 iq;
+	u64 oq;
+	u64 iq64B;
+};
+
+struct octeon_reg_list {
+	u32 __iomem *pci_win_wr_addr_hi;
+	u32 __iomem *pci_win_wr_addr_lo;
+	u64 __iomem *pci_win_wr_addr;
+
+	u32 __iomem *pci_win_rd_addr_hi;
+	u32 __iomem *pci_win_rd_addr_lo;
+	u64 __iomem *pci_win_rd_addr;
+
+	u32 __iomem *pci_win_wr_data_hi;
+	u32 __iomem *pci_win_wr_data_lo;
+	u64 __iomem *pci_win_wr_data;
+
+	u32 __iomem *pci_win_rd_data_hi;
+	u32 __iomem *pci_win_rd_data_lo;
+	u64 __iomem *pci_win_rd_data;
+};
+
+#define OCTEON_CONSOLE_MAX_READ_BYTES 512
+typedef int (*octeon_console_print_fn)(struct octeon_device *oct,
+				       u32 num, char *pre, char *suf);
+struct octeon_console {
+	u32 active;
+	u32 waiting;
+	u64 addr;
+	u32 buffer_size;
+	u64 input_base_addr;
+	u64 output_base_addr;
+	octeon_console_print_fn print;
+	char leftover[OCTEON_CONSOLE_MAX_READ_BYTES];
+};
+
+struct octeon_board_info {
+	char name[OCT_BOARD_NAME];
+	char serial_number[OCT_SERIAL_LEN];
+	u64 major;
+	u64 minor;
+};
+
+struct octeon_fn_list {
+	void (*setup_iq_regs)(struct octeon_device *, u32);
+	void (*setup_oq_regs)(struct octeon_device *, u32);
+
+	irqreturn_t (*process_interrupt_regs)(void *);
+	u64 (*msix_interrupt_handler)(void *);
+
+	int (*setup_mbox)(struct octeon_device *);
+	int (*free_mbox)(struct octeon_device *);
+
+	int (*soft_reset)(struct octeon_device *);
+	int (*setup_device_regs)(struct octeon_device *);
+	void (*bar1_idx_setup)(struct octeon_device *, u64, u32, int);
+	void (*bar1_idx_write)(struct octeon_device *, u32, u32);
+	u32 (*bar1_idx_read)(struct octeon_device *, u32);
+	u32 (*update_iq_read_idx)(struct octeon_instr_queue *);
+
+	void (*enable_oq_pkt_time_intr)(struct octeon_device *, u32);
+	void (*disable_oq_pkt_time_intr)(struct octeon_device *, u32);
+
+	void (*enable_interrupt)(struct octeon_device *, u8);
+	void (*disable_interrupt)(struct octeon_device *, u8);
+
+	int (*enable_io_queues)(struct octeon_device *);
+	void (*disable_io_queues)(struct octeon_device *);
+};
+
+/* Must be multiple of 8, changing breaks ABI */
+#define CVMX_BOOTMEM_NAME_LEN 128
+
+/* Structure for named memory blocks
+ * Number of descriptors
+ * available can be changed without affecting compatibility,
+ * but name length changes require a bump in the bootmem
+ * descriptor version
+ * Note: This structure must be naturally 64 bit aligned, as a single
+ * memory image will be used by both 32 and 64 bit programs.
+ */
+struct cvmx_bootmem_named_block_desc {
+	/** Base address of named block */
+	u64 base_addr;
+
+	/** Size actually allocated for named block */
+	u64 size;
+
+	/** name of named block */
+	char name[CVMX_BOOTMEM_NAME_LEN];
+};
+
+struct oct_fw_info {
+	u32 max_nic_ports;      /** max nic ports for the device */
+	u32 num_gmx_ports;      /** num gmx ports */
+	u64 app_cap_flags;      /** firmware cap flags */
+
+	/** The core application is running in this mode.
+	 * See octeon-drv-opcodes.h for values.
+	 */
+	u32 app_mode;
+	char   liquidio_firmware_version[32];
+	/* Fields extracted from legacy string 'liquidio_firmware_version' */
+	struct {
+		u8  maj;
+		u8  min;
+		u8  rev;
+	} ver;
+};
+
+#define OCT_FW_VER(maj, min, rev) \
+	(((u32)(maj) << 16) | ((u32)(min) << 8) | ((u32)(rev)))
+
+/* wrappers around work structs */
+struct cavium_wk {
+	struct delayed_work work;
+	void *ctxptr;
+	u64 ctxul;
+};
+
+struct cavium_wq {
+	struct workqueue_struct *wq;
+	struct cavium_wk wk;
+};
+
+struct octdev_props {
+	/* Each interface in the Octeon device has a network
+	 * device pointer (used for OS specific calls).
+	 */
+	int    rx_on;
+	int    fec;
+	int    fec_boot;
+	int    napi_enabled;
+	int    gmxport;
+	struct net_device *netdev;
+};
+
+#define LIO_FLAG_MSIX_ENABLED	0x1
+#define MSIX_PO_INT		0x1
+#define MSIX_PI_INT		0x2
+#define MSIX_MBOX_INT		0x4
+
+struct octeon_pf_vf_hs_word {
+#ifdef __LITTLE_ENDIAN_BITFIELD
+	/** PKIND value assigned for the DPI interface */
+	u64        pkind : 8;
+
+	/** OCTEON core clock multiplier   */
+	u64        core_tics_per_us : 16;
+
+	/** OCTEON coprocessor clock multiplier  */
+	u64        coproc_tics_per_us : 16;
+
+	/** app that currently running on OCTEON  */
+	u64        app_mode : 8;
+
+	/** RESERVED */
+	u64 reserved : 16;
+
+#else
+
+	/** RESERVED */
+	u64 reserved : 16;
+
+	/** app that currently running on OCTEON  */
+	u64        app_mode : 8;
+
+	/** OCTEON coprocessor clock multiplier  */
+	u64        coproc_tics_per_us : 16;
+
+	/** OCTEON core clock multiplier   */
+	u64        core_tics_per_us : 16;
+
+	/** PKIND value assigned for the DPI interface */
+	u64        pkind : 8;
+#endif
+};
+
+struct octeon_sriov_info {
+	/* Number of rings assigned to VF */
+	u32	rings_per_vf;
+
+	/** Max Number of VF devices that can be enabled. This variable can
+	 *  specified during load time or it will be derived after allocating
+	 *  PF queues. When max_vfs is derived then each VF will get one queue
+	 **/
+	u32	max_vfs;
+
+	/** Number of VF devices enabled using sysfs. */
+	u32	num_vfs_alloced;
+
+	/* Actual rings left for PF device */
+	u32	num_pf_rings;
+
+	/* SRN of PF usable IO queues */
+	u32	pf_srn;
+
+	/* total pf rings */
+	u32	trs;
+
+	u32	sriov_enabled;
+
+	struct lio_trusted_vf	trusted_vf;
+
+	u64	vf_macaddr[MAX_POSSIBLE_VFS];
+
+	u16	vf_vlantci[MAX_POSSIBLE_VFS];
+
+	int	vf_linkstate[MAX_POSSIBLE_VFS];
+
+	bool    vf_spoofchk[MAX_POSSIBLE_VFS];
+
+	u64	vf_drv_loaded_mask;
+};
+
+struct octeon_ioq_vector {
+	struct octeon_device   *oct_dev;
+	int		        iq_index;
+	int		        droq_index;
+	int			vector;
+	struct octeon_mbox     *mbox;
+	struct cpumask		affinity_mask;
+	u32			ioq_num;
+};
+
+struct lio_vf_rep_list {
+	int num_vfs;
+	struct net_device *ndev[CN23XX_MAX_VFS_PER_PF];
+};
+
+struct lio_devlink_priv {
+	struct octeon_device *oct;
+};
+
+/** The Octeon device.
+ *  Each Octeon device has this structure to represent all its
+ *  components.
+ */
+struct octeon_device {
+	/** Lock for PCI window configuration accesses */
+	spinlock_t pci_win_lock;
+
+	/** Lock for memory accesses */
+	spinlock_t mem_access_lock;
+
+	/** PCI device pointer */
+	struct pci_dev *pci_dev;
+
+	/** Chip specific information. */
+	void *chip;
+
+	/** Number of interfaces detected in this octeon device. */
+	u32 ifcount;
+
+	struct octdev_props props[MAX_OCTEON_LINKS];
+
+	/** Octeon Chip type. */
+	u16 chip_id;
+
+	u16 rev_id;
+
+	u32 subsystem_id;
+
+	u16 pf_num;
+
+	u16 vf_num;
+
+	/** This device's id - set by the driver. */
+	u32 octeon_id;
+
+	/** This device's PCIe port used for traffic. */
+	u16 pcie_port;
+
+	u16 flags;
+#define LIO_FLAG_MSI_ENABLED                  (u32)(1 << 1)
+
+	/** The state of this device */
+	atomic_t status;
+
+	/** memory mapped io range */
+	struct octeon_mmio mmio[OCT_MEM_REGIONS];
+
+	struct octeon_reg_list reg_list;
+
+	struct octeon_fn_list fn_list;
+
+	struct octeon_board_info boardinfo;
+
+	u32 num_iqs;
+
+	/* The pool containing pre allocated buffers used for soft commands */
+	struct octeon_sc_buffer_pool	sc_buf_pool;
+
+	/** The input instruction queues */
+	struct octeon_instr_queue *instr_queue
+		[MAX_POSSIBLE_OCTEON_INSTR_QUEUES];
+
+	/** The doubly-linked list of instruction response */
+	struct octeon_response_list response_list[MAX_RESPONSE_LISTS];
+
+	u32 num_oqs;
+
+	/** The DROQ output queues  */
+	struct octeon_droq *droq[MAX_POSSIBLE_OCTEON_OUTPUT_QUEUES];
+
+	struct octeon_io_enable io_qmask;
+
+	/** List of dispatch functions */
+	struct octeon_dispatch_list dispatch;
+
+	u32 int_status;
+
+	u64 droq_intr;
+
+	/** Physical location of the cvmx_bootmem_desc_t in octeon memory */
+	u64 bootmem_desc_addr;
+
+	/** Placeholder memory for named blocks.
+	 * Assumes single-threaded access
+	 */
+	struct cvmx_bootmem_named_block_desc bootmem_named_block_desc;
+
+	/** Address of consoles descriptor */
+	u64 console_desc_addr;
+
+	/** Number of consoles available. 0 means they are inaccessible */
+	u32 num_consoles;
+
+	/* Console caches */
+	struct octeon_console console[MAX_OCTEON_MAPS];
+
+	/* Console named block info */
+	struct {
+		u64 dram_region_base;
+		int bar1_index;
+	} console_nb_info;
+
+	/* Coprocessor clock rate. */
+	u64 coproc_clock_rate;
+
+	/** The core application is running in this mode. See liquidio_common.h
+	 * for values.
+	 */
+	u32 app_mode;
+
+	struct oct_fw_info fw_info;
+
+	/** The name given to this device. */
+	char device_name[32];
+
+	/** Application Context */
+	void *app_ctx;
+
+	struct cavium_wq dma_comp_wq;
+
+	/** Lock for dma response list */
+	spinlock_t cmd_resp_wqlock;
+	u32 cmd_resp_state;
+
+	struct cavium_wq check_db_wq[MAX_POSSIBLE_OCTEON_INSTR_QUEUES];
+
+	struct cavium_wk nic_poll_work;
+
+	struct cavium_wk console_poll_work[MAX_OCTEON_MAPS];
+
+	void *priv;
+
+	int num_msix_irqs;
+
+	void *msix_entries;
+
+	/* when requesting IRQs, the names are stored here */
+	void *irq_name_storage;
+
+	struct octeon_sriov_info sriov_info;
+
+	struct octeon_pf_vf_hs_word pfvf_hsword;
+
+	int msix_on;
+
+	/** Mail Box details of each octeon queue. */
+	struct octeon_mbox  *mbox[MAX_POSSIBLE_VFS];
+
+	/** IOq information of it's corresponding MSI-X interrupt. */
+	struct octeon_ioq_vector    *ioq_vector;
+
+	int rx_pause;
+	int tx_pause;
+
+	struct oct_link_stats link_stats; /*stastics from firmware*/
+
+	/* private flags to control driver-specific features through ethtool */
+	u32 priv_flags;
+
+	void *watchdog_task;
+
+	u32 rx_coalesce_usecs;
+	u32 rx_max_coalesced_frames;
+	u32 tx_max_coalesced_frames;
+
+	bool cores_crashed;
+
+	struct {
+		int bus;
+		int dev;
+		int func;
+	} loc;
+
+	atomic_t *adapter_refcount; /* reference count of adapter */
+
+	atomic_t *adapter_fw_state; /* per-adapter, lio_fw_state */
+
+	bool ptp_enable;
+
+	struct lio_vf_rep_list vf_rep_list;
+	struct devlink *devlink;
+	enum devlink_eswitch_mode eswitch_mode;
+
+	/* for 25G NIC speed change */
+	u8  speed_boot;
+	u8  speed_setting;
+	u8  no_speed_setting;
+
+	u32    vfstats_poll;
+#define LIO_VFSTATS_POLL 10
+};
+
+#define  OCT_DRV_ONLINE 1
+#define  OCT_DRV_OFFLINE 2
+#define  OCTEON_CN6XXX(oct)	({					\
+				 typeof(oct) _oct = (oct);		\
+				 ((_oct->chip_id == OCTEON_CN66XX) ||	\
+				  (_oct->chip_id == OCTEON_CN68XX));	})
+#define  OCTEON_CN23XX_PF(oct)        ((oct)->chip_id == OCTEON_CN23XX_PF_VID)
+#define  OCTEON_CN23XX_VF(oct)        ((oct)->chip_id == OCTEON_CN23XX_VF_VID)
+#define CHIP_CONF(oct, TYPE)             \
+	(((struct octeon_ ## TYPE  *)((oct)->chip))->conf)
+
+#define MAX_IO_PENDING_PKT_COUNT 100
+
+/*------------------ Function Prototypes ----------------------*/
+
+/** Initialize device list memory */
+void octeon_init_device_list(int conf_type);
+
+/** Free memory for Input and Output queue structures for a octeon device */
+void octeon_free_device_mem(struct octeon_device *oct);
+
+/* Look up a free entry in the octeon_device table and allocate resources
+ * for the octeon_device structure for an octeon device. Called at init
+ * time.
+ */
+struct octeon_device *octeon_allocate_device(u32 pci_id,
+					     u32 priv_size);
+
+/** Register a device's bus location at initialization time.
+ *  @param octeon_dev - pointer to the octeon device structure.
+ *  @param bus        - PCIe bus #
+ *  @param dev        - PCIe device #
+ *  @param func       - PCIe function #
+ *  @param is_pf      - TRUE for PF, FALSE for VF
+ *  @return reference count of device's adapter
+ */
+int octeon_register_device(struct octeon_device *oct,
+			   int bus, int dev, int func, int is_pf);
+
+/** Deregister a device at de-initialization time.
+ *  @param octeon_dev - pointer to the octeon device structure.
+ *  @return reference count of device's adapter
+ */
+int octeon_deregister_device(struct octeon_device *oct);
+
+/**  Initialize the driver's dispatch list which is a mix of a hash table
+ *  and a linked list. This is done at driver load time.
+ *  @param octeon_dev - pointer to the octeon device structure.
+ *  @return 0 on success, else -ve error value
+ */
+int octeon_init_dispatch_list(struct octeon_device *octeon_dev);
+
+/**  Delete the driver's dispatch list and all registered entries.
+ * This is done at driver unload time.
+ *  @param octeon_dev - pointer to the octeon device structure.
+ */
+void octeon_delete_dispatch_list(struct octeon_device *octeon_dev);
+
+/** Initialize the core device fields with the info returned by the FW.
+ * @param recv_info - Receive info structure
+ * @param buf       - Receive buffer
+ */
+int octeon_core_drv_init(struct octeon_recv_info *recv_info, void *buf);
+
+/** Gets the dispatch function registered to receive packets with a
+ *  given opcode/subcode.
+ *  @param  octeon_dev  - the octeon device pointer.
+ *  @param  opcode      - the opcode for which the dispatch function
+ *                        is to checked.
+ *  @param  subcode     - the subcode for which the dispatch function
+ *                        is to checked.
+ *
+ *  @return Success: octeon_dispatch_fn_t (dispatch function pointer)
+ *  @return Failure: NULL
+ *
+ *  Looks up the dispatch list to get the dispatch function for a
+ *  given opcode.
+ */
+octeon_dispatch_fn_t
+octeon_get_dispatch(struct octeon_device *octeon_dev, u16 opcode,
+		    u16 subcode);
+
+/** Get the octeon device pointer.
+ *  @param octeon_id  - The id for which the octeon device pointer is required.
+ *  @return Success: Octeon device pointer.
+ *  @return Failure: NULL.
+ */
+struct octeon_device *lio_get_device(u32 octeon_id);
+
+/** Read windowed register.
+ *  @param  oct   -  pointer to the Octeon device.
+ *  @param  addr  -  Address of the register to read.
+ *
+ *  This routine is called to read from the indirectly accessed
+ *  Octeon registers that are visible through a PCI BAR0 mapped window
+ *  register.
+ *  @return  - 64 bit value read from the register.
+ */
+
+u64 lio_pci_readq(struct octeon_device *oct, u64 addr);
+
+/** Write windowed register.
+ *  @param  oct  -  pointer to the Octeon device.
+ *  @param  val  -  Value to write
+ *  @param  addr -  Address of the register to write
+ *
+ *  This routine is called to write to the indirectly accessed
+ *  Octeon registers that are visible through a PCI BAR0 mapped window
+ *  register.
+ *  @return   Nothing.
+ */
+void lio_pci_writeq(struct octeon_device *oct, u64 val, u64 addr);
+
+/* Routines for reading and writing CSRs */
+#define   octeon_write_csr(oct_dev, reg_off, value) \
+		writel(value, (oct_dev)->mmio[0].hw_addr + (reg_off))
+
+#define   octeon_write_csr64(oct_dev, reg_off, val64) \
+		writeq(val64, (oct_dev)->mmio[0].hw_addr + (reg_off))
+
+#define   octeon_read_csr(oct_dev, reg_off)         \
+		readl((oct_dev)->mmio[0].hw_addr + (reg_off))
+
+#define   octeon_read_csr64(oct_dev, reg_off)         \
+		readq((oct_dev)->mmio[0].hw_addr + (reg_off))
+
+/**
+ * Checks if memory access is okay
+ *
+ * @param oct which octeon to send to
+ * @return Zero on success, negative on failure.
+ */
+int octeon_mem_access_ok(struct octeon_device *oct);
+
+/**
+ * Waits for DDR initialization.
+ *
+ * @param oct which octeon to send to
+ * @param timeout_in_ms pointer to how long to wait until DDR is initialized
+ * in ms.
+ *                      If contents are 0, it waits until contents are non-zero
+ *                      before starting to check.
+ * @return Zero on success, negative on failure.
+ */
+int octeon_wait_for_ddr_init(struct octeon_device *oct,
+			     u32 *timeout_in_ms);
+
+/**
+ * Wait for u-boot to boot and be waiting for a command.
+ *
+ * @param wait_time_hundredths
+ *               Maximum time to wait
+ *
+ * @return Zero on success, negative on failure.
+ */
+int octeon_wait_for_bootloader(struct octeon_device *oct,
+			       u32 wait_time_hundredths);
+
+/**
+ * Initialize console access
+ *
+ * @param oct which octeon initialize
+ * @return Zero on success, negative on failure.
+ */
+int octeon_init_consoles(struct octeon_device *oct);
+
+/**
+ * Adds access to a console to the device.
+ *
+ * @param oct:          which octeon to add to
+ * @param console_num:  which console
+ * @param dbg_enb:      ptr to debug enablement string, one of:
+ *                    * NULL for no debug output (i.e. disabled)
+ *                    * empty string enables debug output (via default method)
+ *                    * specific string to enable debug console output
+ *
+ * @return Zero on success, negative on failure.
+ */
+int octeon_add_console(struct octeon_device *oct, u32 console_num,
+		       char *dbg_enb);
+
+/** Removes all attached consoles. */
+void octeon_remove_consoles(struct octeon_device *oct);
+
+/**
+ * Send a string to u-boot on console 0 as a command.
+ *
+ * @param oct which octeon to send to
+ * @param cmd_str String to send
+ * @param wait_hundredths Time to wait for u-boot to accept the command.
+ *
+ * @return Zero on success, negative on failure.
+ */
+int octeon_console_send_cmd(struct octeon_device *oct, char *cmd_str,
+			    u32 wait_hundredths);
+
+/** Parses, validates, and downloads firmware, then boots associated cores.
+ *  @param oct which octeon to download firmware to
+ *  @param data  - The complete firmware file image
+ *  @param size  - The size of the data
+ *
+ *  @return 0 if success.
+ *         -EINVAL if file is incompatible or badly formatted.
+ *         -ENODEV if no handler was found for the application type or an
+ *         invalid octeon id was passed.
+ */
+int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
+			     size_t size);
+
+char *lio_get_state_string(atomic_t *state_ptr);
+
+/** Sets up instruction queues for the device
+ *  @param oct which octeon to setup
+ *
+ *  @return 0 if success. 1 if fails
+ */
+int octeon_setup_instr_queues(struct octeon_device *oct);
+
+/** Sets up output queues for the device
+ *  @param oct which octeon to setup
+ *
+ *  @return 0 if success. 1 if fails
+ */
+int octeon_setup_output_queues(struct octeon_device *oct);
+
+int octeon_get_tx_qsize(struct octeon_device *oct, u32 q_no);
+
+int octeon_get_rx_qsize(struct octeon_device *oct, u32 q_no);
+
+/** Turns off the input and output queues for the device
+ *  @param oct which octeon to disable
+ */
+int octeon_set_io_queues_off(struct octeon_device *oct);
+
+/** Turns on or off the given output queue for the device
+ *  @param oct which octeon to change
+ *  @param q_no which queue
+ *  @param enable 1 to enable, 0 to disable
+ */
+void octeon_set_droq_pkt_op(struct octeon_device *oct, u32 q_no, u32 enable);
+
+/** Retrieve the config for the device
+ *  @param oct which octeon
+ *  @param card_type type of card
+ *
+ *  @returns pointer to configuration
+ */
+void *oct_get_config_info(struct octeon_device *oct, u16 card_type);
+
+/** Gets the octeon device configuration
+ *  @return - pointer to the octeon configuration struture
+ */
+struct octeon_config *octeon_get_conf(struct octeon_device *oct);
+
+void octeon_free_ioq_vector(struct octeon_device *oct);
+int octeon_allocate_ioq_vector(struct octeon_device  *oct, u32 num_ioqs);
+void lio_enable_irq(struct octeon_droq *droq, struct octeon_instr_queue *iq);
+
+/* LiquidIO driver pivate flags */
+enum {
+	OCT_PRIV_FLAG_TX_BYTES = 0, /* Tx interrupts by pending byte count */
+};
+
+#define OCT_PRIV_FLAG_DEFAULT 0x0
+
+static inline u32 lio_get_priv_flag(struct octeon_device *octdev, u32 flag)
+{
+	return !!(octdev->priv_flags & (0x1 << flag));
+}
+
+static inline void lio_set_priv_flag(struct octeon_device *octdev,
+				     u32 flag, u32 val)
+{
+	if (val)
+		octdev->priv_flags |= (0x1 << flag);
+	else
+		octdev->priv_flags &= ~(0x1 << flag);
+}
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_droq.c

@@ -0,0 +1,968 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/vmalloc.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+#include "octeon_network.h"
+#include "cn66xx_regs.h"
+#include "cn66xx_device.h"
+#include "cn23xx_pf_device.h"
+#include "cn23xx_vf_device.h"
+
+struct __dispatch {
+	struct list_head list;
+	struct octeon_recv_info *rinfo;
+	octeon_dispatch_fn_t disp_fn;
+};
+
+/** Get the argument that the user set when registering dispatch
+ *  function for a given opcode/subcode.
+ *  @param  octeon_dev - the octeon device pointer.
+ *  @param  opcode     - the opcode for which the dispatch argument
+ *                       is to be checked.
+ *  @param  subcode    - the subcode for which the dispatch argument
+ *                       is to be checked.
+ *  @return  Success: void * (argument to the dispatch function)
+ *  @return  Failure: NULL
+ *
+ */
+void *octeon_get_dispatch_arg(struct octeon_device *octeon_dev,
+			      u16 opcode, u16 subcode)
+{
+	int idx;
+	struct list_head *dispatch;
+	void *fn_arg = NULL;
+	u16 combined_opcode = OPCODE_SUBCODE(opcode, subcode);
+
+	idx = combined_opcode & OCTEON_OPCODE_MASK;
+
+	spin_lock_bh(&octeon_dev->dispatch.lock);
+
+	if (octeon_dev->dispatch.count == 0) {
+		spin_unlock_bh(&octeon_dev->dispatch.lock);
+		return NULL;
+	}
+
+	if (octeon_dev->dispatch.dlist[idx].opcode == combined_opcode) {
+		fn_arg = octeon_dev->dispatch.dlist[idx].arg;
+	} else {
+		list_for_each(dispatch,
+			      &octeon_dev->dispatch.dlist[idx].list) {
+			if (((struct octeon_dispatch *)dispatch)->opcode ==
+			    combined_opcode) {
+				fn_arg = ((struct octeon_dispatch *)
+					  dispatch)->arg;
+				break;
+			}
+		}
+	}
+
+	spin_unlock_bh(&octeon_dev->dispatch.lock);
+	return fn_arg;
+}
+
+/** Check for packets on Droq. This function should be called with lock held.
+ *  @param  droq - Droq on which count is checked.
+ *  @return Returns packet count.
+ */
+u32 octeon_droq_check_hw_for_pkts(struct octeon_droq *droq)
+{
+	u32 pkt_count = 0;
+	u32 last_count;
+
+	pkt_count = readl(droq->pkts_sent_reg);
+
+	last_count = pkt_count - droq->pkt_count;
+	droq->pkt_count = pkt_count;
+
+	/* we shall write to cnts  at napi irq enable or end of droq tasklet */
+	if (last_count)
+		atomic_add(last_count, &droq->pkts_pending);
+
+	return last_count;
+}
+EXPORT_SYMBOL_GPL(octeon_droq_check_hw_for_pkts);
+
+static void octeon_droq_compute_max_packet_bufs(struct octeon_droq *droq)
+{
+	u32 count = 0;
+
+	/* max_empty_descs is the max. no. of descs that can have no buffers.
+	 * If the empty desc count goes beyond this value, we cannot safely
+	 * read in a 64K packet sent by Octeon
+	 * (64K is max pkt size from Octeon)
+	 */
+	droq->max_empty_descs = 0;
+
+	do {
+		droq->max_empty_descs++;
+		count += droq->buffer_size;
+	} while (count < (64 * 1024));
+
+	droq->max_empty_descs = droq->max_count - droq->max_empty_descs;
+}
+
+static void octeon_droq_reset_indices(struct octeon_droq *droq)
+{
+	droq->read_idx = 0;
+	droq->write_idx = 0;
+	droq->refill_idx = 0;
+	droq->refill_count = 0;
+	atomic_set(&droq->pkts_pending, 0);
+}
+
+static void
+octeon_droq_destroy_ring_buffers(struct octeon_device *oct,
+				 struct octeon_droq *droq)
+{
+	u32 i;
+	struct octeon_skb_page_info *pg_info;
+
+	for (i = 0; i < droq->max_count; i++) {
+		pg_info = &droq->recv_buf_list[i].pg_info;
+		if (!pg_info)
+			continue;
+
+		if (pg_info->dma)
+			lio_unmap_ring(oct->pci_dev,
+				       (u64)pg_info->dma);
+		pg_info->dma = 0;
+
+		if (pg_info->page)
+			recv_buffer_destroy(droq->recv_buf_list[i].buffer,
+					    pg_info);
+
+		droq->recv_buf_list[i].buffer = NULL;
+	}
+
+	octeon_droq_reset_indices(droq);
+}
+
+static int
+octeon_droq_setup_ring_buffers(struct octeon_device *oct,
+			       struct octeon_droq *droq)
+{
+	u32 i;
+	void *buf;
+	struct octeon_droq_desc *desc_ring = droq->desc_ring;
+
+	for (i = 0; i < droq->max_count; i++) {
+		buf = recv_buffer_alloc(oct, &droq->recv_buf_list[i].pg_info);
+
+		if (!buf) {
+			dev_err(&oct->pci_dev->dev, "%s buffer alloc failed\n",
+				__func__);
+			droq->stats.rx_alloc_failure++;
+			return -ENOMEM;
+		}
+
+		droq->recv_buf_list[i].buffer = buf;
+		droq->recv_buf_list[i].data = get_rbd(buf);
+		desc_ring[i].info_ptr = 0;
+		desc_ring[i].buffer_ptr =
+			lio_map_ring(droq->recv_buf_list[i].buffer);
+	}
+
+	octeon_droq_reset_indices(droq);
+
+	octeon_droq_compute_max_packet_bufs(droq);
+
+	return 0;
+}
+
+int octeon_delete_droq(struct octeon_device *oct, u32 q_no)
+{
+	struct octeon_droq *droq = oct->droq[q_no];
+
+	dev_dbg(&oct->pci_dev->dev, "%s[%d]\n", __func__, q_no);
+
+	octeon_droq_destroy_ring_buffers(oct, droq);
+	vfree(droq->recv_buf_list);
+
+	if (droq->desc_ring)
+		lio_dma_free(oct, (droq->max_count * OCT_DROQ_DESC_SIZE),
+			     droq->desc_ring, droq->desc_ring_dma);
+
+	memset(droq, 0, OCT_DROQ_SIZE);
+	oct->io_qmask.oq &= ~(1ULL << q_no);
+	vfree(oct->droq[q_no]);
+	oct->droq[q_no] = NULL;
+	oct->num_oqs--;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_delete_droq);
+
+int octeon_init_droq(struct octeon_device *oct,
+		     u32 q_no,
+		     u32 num_descs,
+		     u32 desc_size,
+		     void *app_ctx)
+{
+	struct octeon_droq *droq;
+	u32 desc_ring_size = 0, c_num_descs = 0, c_buf_size = 0;
+	u32 c_pkts_per_intr = 0, c_refill_threshold = 0;
+	int numa_node = dev_to_node(&oct->pci_dev->dev);
+
+	dev_dbg(&oct->pci_dev->dev, "%s[%d]\n", __func__, q_no);
+
+	droq = oct->droq[q_no];
+	memset(droq, 0, OCT_DROQ_SIZE);
+
+	droq->oct_dev = oct;
+	droq->q_no = q_no;
+	if (app_ctx)
+		droq->app_ctx = app_ctx;
+	else
+		droq->app_ctx = (void *)(size_t)q_no;
+
+	c_num_descs = num_descs;
+	c_buf_size = desc_size;
+	if (OCTEON_CN6XXX(oct)) {
+		struct octeon_config *conf6x = CHIP_CONF(oct, cn6xxx);
+
+		c_pkts_per_intr = (u32)CFG_GET_OQ_PKTS_PER_INTR(conf6x);
+		c_refill_threshold =
+			(u32)CFG_GET_OQ_REFILL_THRESHOLD(conf6x);
+	} else if (OCTEON_CN23XX_PF(oct)) {
+		struct octeon_config *conf23 = CHIP_CONF(oct, cn23xx_pf);
+
+		c_pkts_per_intr = (u32)CFG_GET_OQ_PKTS_PER_INTR(conf23);
+		c_refill_threshold = (u32)CFG_GET_OQ_REFILL_THRESHOLD(conf23);
+	} else if (OCTEON_CN23XX_VF(oct)) {
+		struct octeon_config *conf23 = CHIP_CONF(oct, cn23xx_vf);
+
+		c_pkts_per_intr = (u32)CFG_GET_OQ_PKTS_PER_INTR(conf23);
+		c_refill_threshold = (u32)CFG_GET_OQ_REFILL_THRESHOLD(conf23);
+	} else {
+		return 1;
+	}
+
+	droq->max_count = c_num_descs;
+	droq->buffer_size = c_buf_size;
+
+	desc_ring_size = droq->max_count * OCT_DROQ_DESC_SIZE;
+	droq->desc_ring = lio_dma_alloc(oct, desc_ring_size,
+					(dma_addr_t *)&droq->desc_ring_dma);
+
+	if (!droq->desc_ring) {
+		dev_err(&oct->pci_dev->dev,
+			"Output queue %d ring alloc failed\n", q_no);
+		return 1;
+	}
+
+	dev_dbg(&oct->pci_dev->dev, "droq[%d]: desc_ring: virt: 0x%p, dma: %lx\n",
+		q_no, droq->desc_ring, droq->desc_ring_dma);
+	dev_dbg(&oct->pci_dev->dev, "droq[%d]: num_desc: %d\n", q_no,
+		droq->max_count);
+
+	droq->recv_buf_list = vzalloc_node(array_size(droq->max_count, OCT_DROQ_RECVBUF_SIZE),
+					   numa_node);
+	if (!droq->recv_buf_list)
+		droq->recv_buf_list = vzalloc(array_size(droq->max_count, OCT_DROQ_RECVBUF_SIZE));
+	if (!droq->recv_buf_list) {
+		dev_err(&oct->pci_dev->dev, "Output queue recv buf list alloc failed\n");
+		goto init_droq_fail;
+	}
+
+	if (octeon_droq_setup_ring_buffers(oct, droq))
+		goto init_droq_fail;
+
+	droq->pkts_per_intr = c_pkts_per_intr;
+	droq->refill_threshold = c_refill_threshold;
+
+	dev_dbg(&oct->pci_dev->dev, "DROQ INIT: max_empty_descs: %d\n",
+		droq->max_empty_descs);
+
+	INIT_LIST_HEAD(&droq->dispatch_list);
+
+	/* For 56xx Pass1, this function won't be called, so no checks. */
+	oct->fn_list.setup_oq_regs(oct, q_no);
+
+	oct->io_qmask.oq |= BIT_ULL(q_no);
+
+	return 0;
+
+init_droq_fail:
+	octeon_delete_droq(oct, q_no);
+	return 1;
+}
+
+/* octeon_create_recv_info
+ * Parameters:
+ *  octeon_dev - pointer to the octeon device structure
+ *  droq       - droq in which the packet arrived.
+ *  buf_cnt    - no. of buffers used by the packet.
+ *  idx        - index in the descriptor for the first buffer in the packet.
+ * Description:
+ *  Allocates a recv_info_t and copies the buffer addresses for packet data
+ *  into the recv_pkt space which starts at an 8B offset from recv_info_t.
+ *  Flags the descriptors for refill later. If available descriptors go
+ *  below the threshold to receive a 64K pkt, new buffers are first allocated
+ *  before the recv_pkt_t is created.
+ *  This routine will be called in interrupt context.
+ * Returns:
+ *  Success: Pointer to recv_info_t
+ *  Failure: NULL.
+ */
+static inline struct octeon_recv_info *octeon_create_recv_info(
+		struct octeon_device *octeon_dev,
+		struct octeon_droq *droq,
+		u32 buf_cnt,
+		u32 idx)
+{
+	struct octeon_droq_info *info;
+	struct octeon_recv_pkt *recv_pkt;
+	struct octeon_recv_info *recv_info;
+	u32 i, bytes_left;
+	struct octeon_skb_page_info *pg_info;
+
+	info = (struct octeon_droq_info *)droq->recv_buf_list[idx].data;
+
+	recv_info = octeon_alloc_recv_info(sizeof(struct __dispatch));
+	if (!recv_info)
+		return NULL;
+
+	recv_pkt = recv_info->recv_pkt;
+	recv_pkt->rh = info->rh;
+	recv_pkt->length = (u32)info->length;
+	recv_pkt->buffer_count = (u16)buf_cnt;
+	recv_pkt->octeon_id = (u16)octeon_dev->octeon_id;
+
+	i = 0;
+	bytes_left = (u32)info->length;
+
+	while (buf_cnt) {
+		{
+			pg_info = &droq->recv_buf_list[idx].pg_info;
+
+			lio_unmap_ring(octeon_dev->pci_dev,
+				       (u64)pg_info->dma);
+			pg_info->page = NULL;
+			pg_info->dma = 0;
+		}
+
+		recv_pkt->buffer_size[i] =
+			(bytes_left >=
+			 droq->buffer_size) ? droq->buffer_size : bytes_left;
+
+		recv_pkt->buffer_ptr[i] = droq->recv_buf_list[idx].buffer;
+		droq->recv_buf_list[idx].buffer = NULL;
+
+		idx = incr_index(idx, 1, droq->max_count);
+		bytes_left -= droq->buffer_size;
+		i++;
+		buf_cnt--;
+	}
+
+	return recv_info;
+}
+
+/* If we were not able to refill all buffers, try to move around
+ * the buffers that were not dispatched.
+ */
+static inline u32
+octeon_droq_refill_pullup_descs(struct octeon_droq *droq,
+				struct octeon_droq_desc *desc_ring)
+{
+	u32 desc_refilled = 0;
+
+	u32 refill_index = droq->refill_idx;
+
+	while (refill_index != droq->read_idx) {
+		if (droq->recv_buf_list[refill_index].buffer) {
+			droq->recv_buf_list[droq->refill_idx].buffer =
+				droq->recv_buf_list[refill_index].buffer;
+			droq->recv_buf_list[droq->refill_idx].data =
+				droq->recv_buf_list[refill_index].data;
+			desc_ring[droq->refill_idx].buffer_ptr =
+				desc_ring[refill_index].buffer_ptr;
+			droq->recv_buf_list[refill_index].buffer = NULL;
+			desc_ring[refill_index].buffer_ptr = 0;
+			do {
+				droq->refill_idx = incr_index(droq->refill_idx,
+							      1,
+							      droq->max_count);
+				desc_refilled++;
+				droq->refill_count--;
+			} while (droq->recv_buf_list[droq->refill_idx].buffer);
+		}
+		refill_index = incr_index(refill_index, 1, droq->max_count);
+	}                       /* while */
+	return desc_refilled;
+}
+
+/* octeon_droq_refill
+ * Parameters:
+ *  droq       - droq in which descriptors require new buffers.
+ * Description:
+ *  Called during normal DROQ processing in interrupt mode or by the poll
+ *  thread to refill the descriptors from which buffers were dispatched
+ *  to upper layers. Attempts to allocate new buffers. If that fails, moves
+ *  up buffers (that were not dispatched) to form a contiguous ring.
+ * Returns:
+ *  No of descriptors refilled.
+ */
+static u32
+octeon_droq_refill(struct octeon_device *octeon_dev, struct octeon_droq *droq)
+{
+	struct octeon_droq_desc *desc_ring;
+	void *buf = NULL;
+	u8 *data;
+	u32 desc_refilled = 0;
+	struct octeon_skb_page_info *pg_info;
+
+	desc_ring = droq->desc_ring;
+
+	while (droq->refill_count && (desc_refilled < droq->max_count)) {
+		/* If a valid buffer exists (happens if there is no dispatch),
+		 * reuse the buffer, else allocate.
+		 */
+		if (!droq->recv_buf_list[droq->refill_idx].buffer) {
+			pg_info =
+				&droq->recv_buf_list[droq->refill_idx].pg_info;
+			/* Either recycle the existing pages or go for
+			 * new page alloc
+			 */
+			if (pg_info->page)
+				buf = recv_buffer_reuse(octeon_dev, pg_info);
+			else
+				buf = recv_buffer_alloc(octeon_dev, pg_info);
+			/* If a buffer could not be allocated, no point in
+			 * continuing
+			 */
+			if (!buf) {
+				droq->stats.rx_alloc_failure++;
+				break;
+			}
+			droq->recv_buf_list[droq->refill_idx].buffer =
+				buf;
+			data = get_rbd(buf);
+		} else {
+			data = get_rbd(droq->recv_buf_list
+				       [droq->refill_idx].buffer);
+		}
+
+		droq->recv_buf_list[droq->refill_idx].data = data;
+
+		desc_ring[droq->refill_idx].buffer_ptr =
+			lio_map_ring(droq->recv_buf_list[
+				     droq->refill_idx].buffer);
+
+		droq->refill_idx = incr_index(droq->refill_idx, 1,
+					      droq->max_count);
+		desc_refilled++;
+		droq->refill_count--;
+	}
+
+	if (droq->refill_count)
+		desc_refilled +=
+			octeon_droq_refill_pullup_descs(droq, desc_ring);
+
+	/* if droq->refill_count
+	 * The refill count would not change in pass two. We only moved buffers
+	 * to close the gap in the ring, but we would still have the same no. of
+	 * buffers to refill.
+	 */
+	return desc_refilled;
+}
+
+/** check if we can allocate packets to get out of oom.
+ *  @param  droq - Droq being checked.
+ *  @return 1 if fails to refill minimum
+ */
+int octeon_retry_droq_refill(struct octeon_droq *droq)
+{
+	struct octeon_device *oct = droq->oct_dev;
+	int desc_refilled, reschedule = 1;
+	u32 pkts_credit;
+
+	pkts_credit = readl(droq->pkts_credit_reg);
+	desc_refilled = octeon_droq_refill(oct, droq);
+	if (desc_refilled) {
+		/* Flush the droq descriptor data to memory to be sure
+		 * that when we update the credits the data in memory
+		 * is accurate.
+		 */
+		wmb();
+		writel(desc_refilled, droq->pkts_credit_reg);
+
+		if (pkts_credit + desc_refilled >= CN23XX_SLI_DEF_BP)
+			reschedule = 0;
+	}
+
+	return reschedule;
+}
+
+static inline u32
+octeon_droq_get_bufcount(u32 buf_size, u32 total_len)
+{
+	return DIV_ROUND_UP(total_len, buf_size);
+}
+
+static int
+octeon_droq_dispatch_pkt(struct octeon_device *oct,
+			 struct octeon_droq *droq,
+			 union octeon_rh *rh,
+			 struct octeon_droq_info *info)
+{
+	u32 cnt;
+	octeon_dispatch_fn_t disp_fn;
+	struct octeon_recv_info *rinfo;
+
+	cnt = octeon_droq_get_bufcount(droq->buffer_size, (u32)info->length);
+
+	disp_fn = octeon_get_dispatch(oct, (u16)rh->r.opcode,
+				      (u16)rh->r.subcode);
+	if (disp_fn) {
+		rinfo = octeon_create_recv_info(oct, droq, cnt, droq->read_idx);
+		if (rinfo) {
+			struct __dispatch *rdisp = rinfo->rsvd;
+
+			rdisp->rinfo = rinfo;
+			rdisp->disp_fn = disp_fn;
+			rinfo->recv_pkt->rh = *rh;
+			list_add_tail(&rdisp->list,
+				      &droq->dispatch_list);
+		} else {
+			droq->stats.dropped_nomem++;
+		}
+	} else {
+		dev_err(&oct->pci_dev->dev, "DROQ: No dispatch function (opcode %u/%u)\n",
+			(unsigned int)rh->r.opcode,
+			(unsigned int)rh->r.subcode);
+		droq->stats.dropped_nodispatch++;
+	}
+
+	return cnt;
+}
+
+static inline void octeon_droq_drop_packets(struct octeon_device *oct,
+					    struct octeon_droq *droq,
+					    u32 cnt)
+{
+	u32 i = 0, buf_cnt;
+	struct octeon_droq_info *info;
+
+	for (i = 0; i < cnt; i++) {
+		info = (struct octeon_droq_info *)
+			droq->recv_buf_list[droq->read_idx].data;
+		octeon_swap_8B_data((u64 *)info, 2);
+
+		if (info->length) {
+			info->length += OCTNET_FRM_LENGTH_SIZE;
+			droq->stats.bytes_received += info->length;
+			buf_cnt = octeon_droq_get_bufcount(droq->buffer_size,
+							   (u32)info->length);
+		} else {
+			dev_err(&oct->pci_dev->dev, "DROQ: In drop: pkt with len 0\n");
+			buf_cnt = 1;
+		}
+
+		droq->read_idx = incr_index(droq->read_idx, buf_cnt,
+					    droq->max_count);
+		droq->refill_count += buf_cnt;
+	}
+}
+
+static u32
+octeon_droq_fast_process_packets(struct octeon_device *oct,
+				 struct octeon_droq *droq,
+				 u32 pkts_to_process)
+{
+	u32 pkt, total_len = 0, pkt_count, retval;
+	struct octeon_droq_info *info;
+	union octeon_rh *rh;
+
+	pkt_count = pkts_to_process;
+
+	for (pkt = 0; pkt < pkt_count; pkt++) {
+		u32 pkt_len = 0;
+		struct sk_buff *nicbuf = NULL;
+		struct octeon_skb_page_info *pg_info;
+		void *buf;
+
+		info = (struct octeon_droq_info *)
+			droq->recv_buf_list[droq->read_idx].data;
+		octeon_swap_8B_data((u64 *)info, 2);
+
+		if (!info->length) {
+			dev_err(&oct->pci_dev->dev,
+				"DROQ[%d] idx: %d len:0, pkt_cnt: %d\n",
+				droq->q_no, droq->read_idx, pkt_count);
+			print_hex_dump_bytes("", DUMP_PREFIX_ADDRESS,
+					     (u8 *)info,
+					     OCT_DROQ_INFO_SIZE);
+			break;
+		}
+
+		/* Len of resp hdr in included in the received data len. */
+		rh = &info->rh;
+
+		info->length += OCTNET_FRM_LENGTH_SIZE;
+		rh->r_dh.len += (ROUNDUP8(OCT_DROQ_INFO_SIZE) / sizeof(u64));
+		total_len += (u32)info->length;
+		if (opcode_slow_path(rh)) {
+			u32 buf_cnt;
+
+			buf_cnt = octeon_droq_dispatch_pkt(oct, droq, rh, info);
+			droq->read_idx = incr_index(droq->read_idx,
+						    buf_cnt, droq->max_count);
+			droq->refill_count += buf_cnt;
+		} else {
+			if (info->length <= droq->buffer_size) {
+				pkt_len = (u32)info->length;
+				nicbuf = droq->recv_buf_list[
+					droq->read_idx].buffer;
+				pg_info = &droq->recv_buf_list[
+					droq->read_idx].pg_info;
+				if (recv_buffer_recycle(oct, pg_info))
+					pg_info->page = NULL;
+				droq->recv_buf_list[droq->read_idx].buffer =
+					NULL;
+
+				droq->read_idx = incr_index(droq->read_idx, 1,
+							    droq->max_count);
+				droq->refill_count++;
+			} else {
+				nicbuf = octeon_fast_packet_alloc((u32)
+								  info->length);
+				pkt_len = 0;
+				/* nicbuf allocation can fail. We'll handle it
+				 * inside the loop.
+				 */
+				while (pkt_len < info->length) {
+					int cpy_len, idx = droq->read_idx;
+
+					cpy_len = ((pkt_len + droq->buffer_size)
+						   > info->length) ?
+						((u32)info->length - pkt_len) :
+						droq->buffer_size;
+
+					if (nicbuf) {
+						octeon_fast_packet_next(droq,
+									nicbuf,
+									cpy_len,
+									idx);
+						buf = droq->recv_buf_list[
+							idx].buffer;
+						recv_buffer_fast_free(buf);
+						droq->recv_buf_list[idx].buffer
+							= NULL;
+					} else {
+						droq->stats.rx_alloc_failure++;
+					}
+
+					pkt_len += cpy_len;
+					droq->read_idx =
+						incr_index(droq->read_idx, 1,
+							   droq->max_count);
+					droq->refill_count++;
+				}
+			}
+
+			if (nicbuf) {
+				if (droq->ops.fptr) {
+					droq->ops.fptr(oct->octeon_id,
+						       nicbuf, pkt_len,
+						       rh, &droq->napi,
+						       droq->ops.farg);
+				} else {
+					recv_buffer_free(nicbuf);
+				}
+			}
+		}
+
+		if (droq->refill_count >= droq->refill_threshold) {
+			int desc_refilled = octeon_droq_refill(oct, droq);
+
+			if (desc_refilled) {
+				/* Flush the droq descriptor data to memory to
+				 * be sure that when we update the credits the
+				 * data in memory is accurate.
+				 */
+				wmb();
+				writel(desc_refilled, droq->pkts_credit_reg);
+			}
+		}
+	}                       /* for (each packet)... */
+
+	/* Increment refill_count by the number of buffers processed. */
+	droq->stats.pkts_received += pkt;
+	droq->stats.bytes_received += total_len;
+
+	retval = pkt;
+	if ((droq->ops.drop_on_max) && (pkts_to_process - pkt)) {
+		octeon_droq_drop_packets(oct, droq, (pkts_to_process - pkt));
+
+		droq->stats.dropped_toomany += (pkts_to_process - pkt);
+		retval = pkts_to_process;
+	}
+
+	atomic_sub(retval, &droq->pkts_pending);
+
+	if (droq->refill_count >= droq->refill_threshold &&
+	    readl(droq->pkts_credit_reg) < CN23XX_SLI_DEF_BP) {
+		octeon_droq_check_hw_for_pkts(droq);
+
+		/* Make sure there are no pkts_pending */
+		if (!atomic_read(&droq->pkts_pending))
+			octeon_schedule_rxq_oom_work(oct, droq);
+	}
+
+	return retval;
+}
+
+int
+octeon_droq_process_packets(struct octeon_device *oct,
+			    struct octeon_droq *droq,
+			    u32 budget)
+{
+	u32 pkt_count = 0;
+	struct list_head *tmp, *tmp2;
+
+	octeon_droq_check_hw_for_pkts(droq);
+	pkt_count = atomic_read(&droq->pkts_pending);
+
+	if (!pkt_count)
+		return 0;
+
+	if (pkt_count > budget)
+		pkt_count = budget;
+
+	octeon_droq_fast_process_packets(oct, droq, pkt_count);
+
+	list_for_each_safe(tmp, tmp2, &droq->dispatch_list) {
+		struct __dispatch *rdisp = (struct __dispatch *)tmp;
+
+		list_del(tmp);
+		rdisp->disp_fn(rdisp->rinfo,
+			       octeon_get_dispatch_arg
+			       (oct,
+				(u16)rdisp->rinfo->recv_pkt->rh.r.opcode,
+				(u16)rdisp->rinfo->recv_pkt->rh.r.subcode));
+	}
+
+	/* If there are packets pending. schedule tasklet again */
+	if (atomic_read(&droq->pkts_pending))
+		return 1;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_droq_process_packets);
+
+/*
+ * Utility function to poll for packets. check_hw_for_packets must be
+ * called before calling this routine.
+ */
+
+int
+octeon_droq_process_poll_pkts(struct octeon_device *oct,
+			      struct octeon_droq *droq, u32 budget)
+{
+	struct list_head *tmp, *tmp2;
+	u32 pkts_available = 0, pkts_processed = 0;
+	u32 total_pkts_processed = 0;
+
+	if (budget > droq->max_count)
+		budget = droq->max_count;
+
+	while (total_pkts_processed < budget) {
+		octeon_droq_check_hw_for_pkts(droq);
+
+		pkts_available = min((budget - total_pkts_processed),
+				     (u32)(atomic_read(&droq->pkts_pending)));
+
+		if (pkts_available == 0)
+			break;
+
+		pkts_processed =
+			octeon_droq_fast_process_packets(oct, droq,
+							 pkts_available);
+
+		total_pkts_processed += pkts_processed;
+	}
+
+	list_for_each_safe(tmp, tmp2, &droq->dispatch_list) {
+		struct __dispatch *rdisp = (struct __dispatch *)tmp;
+
+		list_del(tmp);
+		rdisp->disp_fn(rdisp->rinfo,
+			       octeon_get_dispatch_arg
+			       (oct,
+				(u16)rdisp->rinfo->recv_pkt->rh.r.opcode,
+				(u16)rdisp->rinfo->recv_pkt->rh.r.subcode));
+	}
+
+	return total_pkts_processed;
+}
+
+/* Enable Pkt Interrupt */
+int
+octeon_enable_irq(struct octeon_device *oct, u32 q_no)
+{
+	switch (oct->chip_id) {
+	case OCTEON_CN66XX:
+	case OCTEON_CN68XX: {
+		struct octeon_cn6xxx *cn6xxx =
+			(struct octeon_cn6xxx *)oct->chip;
+		unsigned long flags;
+		u32 value;
+
+		spin_lock_irqsave
+			(&cn6xxx->lock_for_droq_int_enb_reg, flags);
+		value = octeon_read_csr(oct, CN6XXX_SLI_PKT_TIME_INT_ENB);
+		value |= (1 << q_no);
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_TIME_INT_ENB, value);
+		value = octeon_read_csr(oct, CN6XXX_SLI_PKT_CNT_INT_ENB);
+		value |= (1 << q_no);
+		octeon_write_csr(oct, CN6XXX_SLI_PKT_CNT_INT_ENB, value);
+
+		/* don't bother flushing the enables */
+
+		spin_unlock_irqrestore
+			(&cn6xxx->lock_for_droq_int_enb_reg, flags);
+	}
+		break;
+	case OCTEON_CN23XX_PF_VID:
+		lio_enable_irq(oct->droq[q_no], oct->instr_queue[q_no]);
+		break;
+
+	case OCTEON_CN23XX_VF_VID:
+		lio_enable_irq(oct->droq[q_no], oct->instr_queue[q_no]);
+		break;
+	default:
+		dev_err(&oct->pci_dev->dev, "%s Unknown Chip\n", __func__);
+		return 1;
+	}
+
+	return 0;
+}
+
+int octeon_register_droq_ops(struct octeon_device *oct, u32 q_no,
+			     struct octeon_droq_ops *ops)
+{
+	struct octeon_config *oct_cfg = NULL;
+	struct octeon_droq *droq;
+
+	oct_cfg = octeon_get_conf(oct);
+
+	if (!oct_cfg)
+		return -EINVAL;
+
+	if (!(ops)) {
+		dev_err(&oct->pci_dev->dev, "%s: droq_ops pointer is NULL\n",
+			__func__);
+		return -EINVAL;
+	}
+
+	if (q_no >= CFG_GET_OQ_MAX_Q(oct_cfg)) {
+		dev_err(&oct->pci_dev->dev, "%s: droq id (%d) exceeds MAX (%d)\n",
+			__func__, q_no, (oct->num_oqs - 1));
+		return -EINVAL;
+	}
+
+	droq = oct->droq[q_no];
+	memcpy(&droq->ops, ops, sizeof(struct octeon_droq_ops));
+
+	return 0;
+}
+
+int octeon_unregister_droq_ops(struct octeon_device *oct, u32 q_no)
+{
+	struct octeon_config *oct_cfg = NULL;
+	struct octeon_droq *droq;
+
+	oct_cfg = octeon_get_conf(oct);
+
+	if (!oct_cfg)
+		return -EINVAL;
+
+	if (q_no >= CFG_GET_OQ_MAX_Q(oct_cfg)) {
+		dev_err(&oct->pci_dev->dev, "%s: droq id (%d) exceeds MAX (%d)\n",
+			__func__, q_no, oct->num_oqs - 1);
+		return -EINVAL;
+	}
+
+	droq = oct->droq[q_no];
+
+	if (!droq) {
+		dev_info(&oct->pci_dev->dev,
+			 "Droq id (%d) not available.\n", q_no);
+		return 0;
+	}
+
+	droq->ops.fptr = NULL;
+	droq->ops.farg = NULL;
+	droq->ops.drop_on_max = 0;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_unregister_droq_ops);
+
+int octeon_create_droq(struct octeon_device *oct,
+		       u32 q_no, u32 num_descs,
+		       u32 desc_size, void *app_ctx)
+{
+	struct octeon_droq *droq;
+	int numa_node = dev_to_node(&oct->pci_dev->dev);
+
+	if (oct->droq[q_no]) {
+		dev_dbg(&oct->pci_dev->dev, "Droq already in use. Cannot create droq %d again\n",
+			q_no);
+		return 1;
+	}
+
+	/* Allocate the DS for the new droq. */
+	droq = vmalloc_node(sizeof(*droq), numa_node);
+	if (!droq)
+		droq = vmalloc(sizeof(*droq));
+	if (!droq)
+		return -1;
+
+	memset(droq, 0, sizeof(struct octeon_droq));
+
+	/*Disable the pkt o/p for this Q  */
+	octeon_set_droq_pkt_op(oct, q_no, 0);
+	oct->droq[q_no] = droq;
+
+	/* Initialize the Droq */
+	if (octeon_init_droq(oct, q_no, num_descs, desc_size, app_ctx)) {
+		vfree(oct->droq[q_no]);
+		oct->droq[q_no] = NULL;
+		return -1;
+	}
+
+	oct->num_oqs++;
+
+	dev_dbg(&oct->pci_dev->dev, "%s: Total number of OQ: %d\n", __func__,
+		oct->num_oqs);
+
+	/* Global Droq register settings */
+
+	/* As of now not required, as setting are done for all 32 Droqs at
+	 * the same time.
+	 */
+	return 0;
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_droq.h

@@ -0,0 +1,414 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*!  \file  octeon_droq.h
+ *   \brief Implementation of Octeon Output queues. "Output" is with
+ *   respect to the Octeon device on the NIC. From this driver's point of
+ *   view they are ingress queues.
+ */
+
+#ifndef __OCTEON_DROQ_H__
+#define __OCTEON_DROQ_H__
+
+/* Default number of packets that will be processed in one iteration. */
+#define MAX_PACKET_BUDGET 0xFFFFFFFF
+
+/** Octeon descriptor format.
+ *  The descriptor ring is made of descriptors which have 2 64-bit values:
+ *  -# Physical (bus) address of the data buffer.
+ *  -# Physical (bus) address of a octeon_droq_info structure.
+ *  The Octeon device DMA's incoming packets and its information at the address
+ *  given by these descriptor fields.
+ */
+struct octeon_droq_desc {
+	/** The buffer pointer */
+	u64 buffer_ptr;
+
+	/** The Info pointer */
+	u64 info_ptr;
+};
+
+#define OCT_DROQ_DESC_SIZE    (sizeof(struct octeon_droq_desc))
+
+/** Information about packet DMA'ed by Octeon.
+ *  The format of the information available at Info Pointer after Octeon
+ *  has posted a packet. Not all descriptors have valid information. Only
+ *  the Info field of the first descriptor for a packet has information
+ *  about the packet.
+ */
+struct octeon_droq_info {
+	/** The Length of the packet. */
+	u64 length;
+
+	/** The Output Receive Header. */
+	union octeon_rh rh;
+};
+
+#define OCT_DROQ_INFO_SIZE   (sizeof(struct octeon_droq_info))
+
+struct octeon_skb_page_info {
+	/* DMA address for the page */
+	dma_addr_t dma;
+
+	/* Page for the rx dma  **/
+	struct page *page;
+
+	/** which offset into page */
+	unsigned int page_offset;
+};
+
+/** Pointer to data buffer.
+ *  Driver keeps a pointer to the data buffer that it made available to
+ *  the Octeon device. Since the descriptor ring keeps physical (bus)
+ *  addresses, this field is required for the driver to keep track of
+ *  the virtual address pointers.
+ */
+struct octeon_recv_buffer {
+	/** Packet buffer, including metadata. */
+	void *buffer;
+
+	/** Data in the packet buffer.  */
+	u8 *data;
+
+	/** pg_info **/
+	struct octeon_skb_page_info pg_info;
+};
+
+#define OCT_DROQ_RECVBUF_SIZE    (sizeof(struct octeon_recv_buffer))
+
+/** Output Queue statistics. Each output queue has four stats fields. */
+struct oct_droq_stats {
+	/** Number of packets received in this queue. */
+	u64 pkts_received;
+
+	/** Bytes received by this queue. */
+	u64 bytes_received;
+
+	/** Packets dropped due to no dispatch function. */
+	u64 dropped_nodispatch;
+
+	/** Packets dropped due to no memory available. */
+	u64 dropped_nomem;
+
+	/** Packets dropped due to large number of pkts to process. */
+	u64 dropped_toomany;
+
+	/** Number of packets  sent to stack from this queue. */
+	u64 rx_pkts_received;
+
+	/** Number of Bytes sent to stack from this queue. */
+	u64 rx_bytes_received;
+
+	/** Num of Packets dropped due to receive path failures. */
+	u64 rx_dropped;
+
+	u64 rx_vxlan;
+
+	/** Num of failures of recv_buffer_alloc() */
+	u64 rx_alloc_failure;
+
+};
+
+/* The maximum number of buffers that can be dispatched from the
+ * output/dma queue. Set to 64 assuming 1K buffers in DROQ and the fact that
+ * max packet size from DROQ is 64K.
+ */
+#define    MAX_RECV_BUFS    64
+
+/** Receive Packet format used when dispatching output queue packets
+ *  with non-raw opcodes.
+ *  The received packet will be sent to the upper layers using this
+ *  structure which is passed as a parameter to the dispatch function
+ */
+struct octeon_recv_pkt {
+	/**  Number of buffers in this received packet */
+	u16 buffer_count;
+
+	/** Id of the device that is sending the packet up */
+	u16 octeon_id;
+
+	/** Length of data in the packet buffer */
+	u32 length;
+
+	/** The receive header */
+	union octeon_rh rh;
+
+	/** Pointer to the OS-specific packet buffer */
+	void *buffer_ptr[MAX_RECV_BUFS];
+
+	/** Size of the buffers pointed to by ptr's in buffer_ptr */
+	u32 buffer_size[MAX_RECV_BUFS];
+};
+
+#define OCT_RECV_PKT_SIZE    (sizeof(struct octeon_recv_pkt))
+
+/** The first parameter of a dispatch function.
+ *  For a raw mode opcode, the driver dispatches with the device
+ *  pointer in this structure.
+ *  For non-raw mode opcode, the driver dispatches the recv_pkt
+ *  created to contain the buffers with data received from Octeon.
+ *  ---------------------
+ *  |     *recv_pkt ----|---
+ *  |-------------------|   |
+ *  | 0 or more bytes   |   |
+ *  | reserved by driver|   |
+ *  |-------------------|<-/
+ *  | octeon_recv_pkt   |
+ *  |                   |
+ *  |___________________|
+ */
+struct octeon_recv_info {
+	void *rsvd;
+	struct octeon_recv_pkt *recv_pkt;
+};
+
+#define  OCT_RECV_INFO_SIZE    (sizeof(struct octeon_recv_info))
+
+/** Allocate a recv_info structure. The recv_pkt pointer in the recv_info
+ *  structure is filled in before this call returns.
+ *  @param extra_bytes - extra bytes to be allocated at the end of the recv info
+ *                       structure.
+ *  @return - pointer to a newly allocated recv_info structure.
+ */
+static inline struct octeon_recv_info *octeon_alloc_recv_info(int extra_bytes)
+{
+	struct octeon_recv_info *recv_info;
+	u8 *buf;
+
+	buf = kmalloc(OCT_RECV_PKT_SIZE + OCT_RECV_INFO_SIZE +
+		      extra_bytes, GFP_ATOMIC);
+	if (!buf)
+		return NULL;
+
+	recv_info = (struct octeon_recv_info *)buf;
+	recv_info->recv_pkt =
+		(struct octeon_recv_pkt *)(buf + OCT_RECV_INFO_SIZE);
+	recv_info->rsvd = NULL;
+	if (extra_bytes)
+		recv_info->rsvd = buf + OCT_RECV_INFO_SIZE + OCT_RECV_PKT_SIZE;
+
+	return recv_info;
+}
+
+/** Free a recv_info structure.
+ *  @param recv_info - Pointer to receive_info to be freed
+ */
+static inline void octeon_free_recv_info(struct octeon_recv_info *recv_info)
+{
+	kfree(recv_info);
+}
+
+typedef int (*octeon_dispatch_fn_t)(struct octeon_recv_info *, void *);
+
+/** Used by NIC module to register packet handler and to get device
+ * information for each octeon device.
+ */
+struct octeon_droq_ops {
+	/** This registered function will be called by the driver with
+	 *  the octeon id, pointer to buffer from droq and length of
+	 *  data in the buffer. The receive header gives the port
+	 *  number to the caller.  Function pointer is set by caller.
+	 */
+	void (*fptr)(u32, void *, u32, union octeon_rh *, void *, void *);
+	void *farg;
+
+	/* This function will be called by the driver for all NAPI related
+	 * events. The first param is the octeon id. The second param is the
+	 * output queue number. The third is the NAPI event that occurred.
+	 */
+	void (*napi_fn)(void *);
+
+	u32 poll_mode;
+
+	/** Flag indicating if the DROQ handler should drop packets that
+	 *  it cannot handle in one iteration. Set by caller.
+	 */
+	u32 drop_on_max;
+};
+
+/** The Descriptor Ring Output Queue structure.
+ *  This structure has all the information required to implement a
+ *  Octeon DROQ.
+ */
+struct octeon_droq {
+	u32 q_no;
+
+	u32 pkt_count;
+
+	struct octeon_droq_ops ops;
+
+	struct octeon_device *oct_dev;
+
+	/** The 8B aligned descriptor ring starts at this address. */
+	struct octeon_droq_desc *desc_ring;
+
+	/** Index in the ring where the driver should read the next packet */
+	u32 read_idx;
+
+	/** Index in the ring where Octeon will write the next packet */
+	u32 write_idx;
+
+	/** Index in the ring where the driver will refill the descriptor's
+	 * buffer
+	 */
+	u32 refill_idx;
+
+	/** Packets pending to be processed */
+	atomic_t pkts_pending;
+
+	/** Number of  descriptors in this ring. */
+	u32 max_count;
+
+	/** The number of descriptors pending refill. */
+	u32 refill_count;
+
+	u32 pkts_per_intr;
+	u32 refill_threshold;
+
+	/** The max number of descriptors in DROQ without a buffer.
+	 * This field is used to keep track of empty space threshold. If the
+	 * refill_count reaches this value, the DROQ cannot accept a max-sized
+	 * (64K) packet.
+	 */
+	u32 max_empty_descs;
+
+	/** The receive buffer list. This list has the virtual addresses of the
+	 * buffers.
+	 */
+	struct octeon_recv_buffer *recv_buf_list;
+
+	/** The size of each buffer pointed by the buffer pointer. */
+	u32 buffer_size;
+
+	/** Pointer to the mapped packet credit register.
+	 * Host writes number of info/buffer ptrs available to this register
+	 */
+	void  __iomem *pkts_credit_reg;
+
+	/** Pointer to the mapped packet sent register.
+	 * Octeon writes the number of packets DMA'ed to host memory
+	 * in this register.
+	 */
+	void __iomem *pkts_sent_reg;
+
+	struct list_head dispatch_list;
+
+	/** Statistics for this DROQ. */
+	struct oct_droq_stats stats;
+
+	/** DMA mapped address of the DROQ descriptor ring. */
+	size_t desc_ring_dma;
+
+	/** application context */
+	void *app_ctx;
+
+	struct napi_struct napi;
+
+	u32 cpu_id;
+
+	call_single_data_t csd;
+};
+
+#define OCT_DROQ_SIZE   (sizeof(struct octeon_droq))
+
+/**
+ *  Allocates space for the descriptor ring for the droq and sets the
+ *   base addr, num desc etc in Octeon registers.
+ *
+ * @param  oct_dev    - pointer to the octeon device structure
+ * @param  q_no       - droq no. ranges from 0 - 3.
+ * @param app_ctx     - pointer to application context
+ * @return Success: 0    Failure: 1
+ */
+int octeon_init_droq(struct octeon_device *oct_dev,
+		     u32 q_no,
+		     u32 num_descs,
+		     u32 desc_size,
+		     void *app_ctx);
+
+/**
+ *  Frees the space for descriptor ring for the droq.
+ *
+ *  @param oct_dev - pointer to the octeon device structure
+ *  @param q_no    - droq no. ranges from 0 - 3.
+ *  @return:    Success: 0    Failure: 1
+ */
+int octeon_delete_droq(struct octeon_device *oct_dev, u32 q_no);
+
+/** Register a change in droq operations. The ops field has a pointer to a
+ * function which will called by the DROQ handler for all packets arriving
+ * on output queues given by q_no irrespective of the type of packet.
+ * The ops field also has a flag which if set tells the DROQ handler to
+ * drop packets if it receives more than what it can process in one
+ * invocation of the handler.
+ * @param oct       - octeon device
+ * @param q_no      - octeon output queue number (0 <= q_no <= MAX_OCTEON_DROQ-1
+ * @param ops       - the droq_ops settings for this queue
+ * @return          - 0 on success, -ENODEV or -EINVAL on error.
+ */
+int
+octeon_register_droq_ops(struct octeon_device *oct,
+			 u32 q_no,
+			 struct octeon_droq_ops *ops);
+
+/** Resets the function pointer and flag settings made by
+ * octeon_register_droq_ops(). After this routine is called, the DROQ handler
+ * will lookup dispatch function for each arriving packet on the output queue
+ * given by q_no.
+ * @param oct       - octeon device
+ * @param q_no      - octeon output queue number (0 <= q_no <= MAX_OCTEON_DROQ-1
+ * @return          - 0 on success, -ENODEV or -EINVAL on error.
+ */
+int octeon_unregister_droq_ops(struct octeon_device *oct, u32 q_no);
+
+/**   Register a dispatch function for a opcode/subcode. The driver will call
+ *    this dispatch function when it receives a packet with the given
+ *    opcode/subcode in its output queues along with the user specified
+ *    argument.
+ *    @param  oct        - the octeon device to register with.
+ *    @param  opcode     - the opcode for which the dispatch will be registered.
+ *    @param  subcode    - the subcode for which the dispatch will be registered
+ *    @param  fn         - the dispatch function.
+ *    @param  fn_arg     - user specified that will be passed along with the
+ *                         dispatch function by the driver.
+ *    @return Success: 0; Failure: 1
+ */
+int octeon_register_dispatch_fn(struct octeon_device *oct,
+				u16 opcode,
+				u16 subcode,
+				octeon_dispatch_fn_t fn, void *fn_arg);
+
+void *octeon_get_dispatch_arg(struct octeon_device *oct,
+			      u16 opcode, u16 subcode);
+
+u32 octeon_droq_check_hw_for_pkts(struct octeon_droq *droq);
+
+int octeon_create_droq(struct octeon_device *oct, u32 q_no,
+		       u32 num_descs, u32 desc_size, void *app_ctx);
+
+int octeon_droq_process_packets(struct octeon_device *oct,
+				struct octeon_droq *droq,
+				u32 budget);
+
+int octeon_droq_process_poll_pkts(struct octeon_device *oct,
+				  struct octeon_droq *droq, u32 budget);
+
+int octeon_enable_irq(struct octeon_device *oct, u32 q_no);
+
+int octeon_retry_droq_refill(struct octeon_droq *droq);
+
+#endif	/*__OCTEON_DROQ_H__ */

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_iq.h

@@ -0,0 +1,396 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*!  \file  octeon_iq.h
+ *   \brief Host Driver: Implementation of Octeon input queues. "Input" is
+ *   with respect to the Octeon device on the NIC. From this driver's
+ *   point of view they are egress queues.
+ */
+
+#ifndef __OCTEON_IQ_H__
+#define  __OCTEON_IQ_H__
+
+#define IQ_STATUS_RUNNING   1
+
+#define IQ_SEND_OK          0
+#define IQ_SEND_STOP        1
+#define IQ_SEND_FAILED     -1
+
+/*-------------------------  INSTRUCTION QUEUE --------------------------*/
+
+/* \cond */
+
+#define REQTYPE_NONE                 0
+#define REQTYPE_NORESP_NET           1
+#define REQTYPE_NORESP_NET_SG        2
+#define REQTYPE_RESP_NET             3
+#define REQTYPE_RESP_NET_SG          4
+#define REQTYPE_SOFT_COMMAND         5
+#define REQTYPE_LAST                 5
+
+struct octeon_request_list {
+	u32 reqtype;
+	void *buf;
+};
+
+/* \endcond */
+
+/** Input Queue statistics. Each input queue has four stats fields. */
+struct oct_iq_stats {
+	u64 instr_posted; /**< Instructions posted to this queue. */
+	u64 instr_processed; /**< Instructions processed in this queue. */
+	u64 instr_dropped; /**< Instructions that could not be processed */
+	u64 bytes_sent;  /**< Bytes sent through this queue. */
+	u64 sgentry_sent;/**< Gather entries sent through this queue. */
+	u64 tx_done;/**< Num of packets sent to network. */
+	u64 tx_iq_busy;/**< Numof times this iq was found to be full. */
+	u64 tx_dropped;/**< Numof pkts dropped dueto xmitpath errors. */
+	u64 tx_tot_bytes;/**< Total count of bytes sento to network. */
+	u64 tx_gso;  /* count of tso */
+	u64 tx_vxlan; /* tunnel */
+	u64 tx_dmamap_fail; /* Number of times dma mapping failed */
+	u64 tx_restart; /* Number of times this queue restarted */
+};
+
+#define OCT_IQ_STATS_SIZE   (sizeof(struct oct_iq_stats))
+
+/** The instruction (input) queue.
+ *  The input queue is used to post raw (instruction) mode data or packet
+ *  data to Octeon device from the host. Each input queue (upto 4) for
+ *  a Octeon device has one such structure to represent it.
+ */
+struct octeon_instr_queue {
+	struct octeon_device *oct_dev;
+
+	/** A spinlock to protect access to the input ring.  */
+	spinlock_t lock;
+
+	/** A spinlock to protect while posting on the ring.  */
+	spinlock_t post_lock;
+
+	/** This flag indicates if the queue can be used for soft commands.
+	 *  If this flag is set, post_lock must be acquired before posting
+	 *  a command to the queue.
+	 *  If this flag is clear, post_lock is invalid for the queue.
+	 *  All control commands (soft commands) will go through only Queue 0
+	 *  (control and data queue). So only queue-0 needs post_lock,
+	 *  other queues are only data queues and does not need post_lock
+	 */
+	bool allow_soft_cmds;
+
+	u32 pkt_in_done;
+
+	u32 pkts_processed;
+
+	/** A spinlock to protect access to the input ring.*/
+	spinlock_t iq_flush_running_lock;
+
+	/** Flag that indicates if the queue uses 64 byte commands. */
+	u32 iqcmd_64B:1;
+
+	/** Queue info. */
+	union oct_txpciq txpciq;
+
+	u32 rsvd:17;
+
+	/* Controls whether extra flushing of IQ is done on Tx */
+	u32 do_auto_flush:1;
+
+	u32 status:8;
+
+	/** Maximum no. of instructions in this queue. */
+	u32 max_count;
+
+	/** Index in input ring where the driver should write the next packet */
+	u32 host_write_index;
+
+	/** Index in input ring where Octeon is expected to read the next
+	 * packet.
+	 */
+	u32 octeon_read_index;
+
+	/** This index aids in finding the window in the queue where Octeon
+	 *  has read the commands.
+	 */
+	u32 flush_index;
+
+	/** This field keeps track of the instructions pending in this queue. */
+	atomic_t instr_pending;
+
+	u32 reset_instr_cnt;
+
+	/** Pointer to the Virtual Base addr of the input ring. */
+	u8 *base_addr;
+
+	struct octeon_request_list *request_list;
+
+	/** Octeon doorbell register for the ring. */
+	void __iomem *doorbell_reg;
+
+	/** Octeon instruction count register for this ring. */
+	void __iomem *inst_cnt_reg;
+
+	/** Number of instructions pending to be posted to Octeon. */
+	u32 fill_cnt;
+
+	/** The max. number of instructions that can be held pending by the
+	 * driver.
+	 */
+	u32 fill_threshold;
+
+	/** The last time that the doorbell was rung. */
+	u64 last_db_time;
+
+	/** The doorbell timeout. If the doorbell was not rung for this time and
+	 * fill_cnt is non-zero, ring the doorbell again.
+	 */
+	u32 db_timeout;
+
+	/** Statistics for this input queue. */
+	struct oct_iq_stats stats;
+
+	/** DMA mapped base address of the input descriptor ring. */
+	dma_addr_t base_addr_dma;
+
+	/** Application context */
+	void *app_ctx;
+
+	/* network stack queue index */
+	int q_index;
+
+	/*os ifidx associated with this queue */
+	int ifidx;
+
+};
+
+/*----------------------  INSTRUCTION FORMAT ----------------------------*/
+
+/** 32-byte instruction format.
+ *  Format of instruction for a 32-byte mode input queue.
+ */
+struct octeon_instr_32B {
+	/** Pointer where the input data is available. */
+	u64 dptr;
+
+	/** Instruction Header.  */
+	u64 ih;
+
+	/** Pointer where the response for a RAW mode packet will be written
+	 * by Octeon.
+	 */
+	u64 rptr;
+
+	/** Input Request Header. Additional info about the input. */
+	u64 irh;
+
+};
+
+#define OCT_32B_INSTR_SIZE     (sizeof(struct octeon_instr_32B))
+
+/** 64-byte instruction format.
+ *  Format of instruction for a 64-byte mode input queue.
+ */
+struct octeon_instr2_64B {
+	/** Pointer where the input data is available. */
+	u64 dptr;
+
+	/** Instruction Header. */
+	u64 ih2;
+
+	/** Input Request Header. */
+	u64 irh;
+
+	/** opcode/subcode specific parameters */
+	u64 ossp[2];
+
+	/** Return Data Parameters */
+	u64 rdp;
+
+	/** Pointer where the response for a RAW mode packet will be written
+	 * by Octeon.
+	 */
+	u64 rptr;
+
+	u64 reserved;
+};
+
+struct octeon_instr3_64B {
+	/** Pointer where the input data is available. */
+	u64 dptr;
+
+	/** Instruction Header. */
+	u64 ih3;
+
+	/** Instruction Header. */
+	u64 pki_ih3;
+
+	/** Input Request Header. */
+	u64 irh;
+
+	/** opcode/subcode specific parameters */
+	u64 ossp[2];
+
+	/** Return Data Parameters */
+	u64 rdp;
+
+	/** Pointer where the response for a RAW mode packet will be written
+	 * by Octeon.
+	 */
+	u64 rptr;
+
+};
+
+union octeon_instr_64B {
+	struct octeon_instr2_64B cmd2;
+	struct octeon_instr3_64B cmd3;
+};
+
+#define OCT_64B_INSTR_SIZE     (sizeof(union octeon_instr_64B))
+
+/** The size of each buffer in soft command buffer pool
+ */
+#define  SOFT_COMMAND_BUFFER_SIZE	2048
+
+struct octeon_soft_command {
+	/** Soft command buffer info. */
+	struct list_head node;
+	u64 dma_addr;
+	u32 size;
+
+	/** Command and return status */
+	union octeon_instr_64B cmd;
+
+#define COMPLETION_WORD_INIT    0xffffffffffffffffULL
+	u64 *status_word;
+
+	/** Data buffer info */
+	void *virtdptr;
+	u64 dmadptr;
+	u32 datasize;
+
+	/** Return buffer info */
+	void *virtrptr;
+	u64 dmarptr;
+	u32 rdatasize;
+
+	/** Context buffer info */
+	void *ctxptr;
+	u32  ctxsize;
+
+	/** Time out and callback */
+	size_t expiry_time;
+	u32 iq_no;
+	void (*callback)(struct octeon_device *, u32, void *);
+	void *callback_arg;
+
+	int caller_is_done;
+	u32 sc_status;
+	struct completion complete;
+};
+
+/* max timeout (in milli sec) for soft request */
+#define LIO_SC_MAX_TMO_MS       60000
+
+/** Maximum number of buffers to allocate into soft command buffer pool
+ */
+#define  MAX_SOFT_COMMAND_BUFFERS	256
+
+/** Head of a soft command buffer pool.
+ */
+struct octeon_sc_buffer_pool {
+	/** List structure to add delete pending entries to */
+	struct list_head head;
+
+	/** A lock for this response list */
+	spinlock_t lock;
+
+	atomic_t alloc_buf_count;
+};
+
+#define INCR_INSTRQUEUE_PKT_COUNT(octeon_dev_ptr, iq_no, field, count)  \
+		(((octeon_dev_ptr)->instr_queue[iq_no]->stats.field) += count)
+
+int octeon_setup_sc_buffer_pool(struct octeon_device *oct);
+int octeon_free_sc_done_list(struct octeon_device *oct);
+int octeon_free_sc_zombie_list(struct octeon_device *oct);
+int octeon_free_sc_buffer_pool(struct octeon_device *oct);
+struct octeon_soft_command *
+	octeon_alloc_soft_command(struct octeon_device *oct,
+				  u32 datasize, u32 rdatasize,
+				  u32 ctxsize);
+void octeon_free_soft_command(struct octeon_device *oct,
+			      struct octeon_soft_command *sc);
+
+/**
+ *  octeon_init_instr_queue()
+ *  @param octeon_dev      - pointer to the octeon device structure.
+ *  @param txpciq          - queue to be initialized (0 <= q_no <= 3).
+ *
+ *  Called at driver init time for each input queue. iq_conf has the
+ *  configuration parameters for the queue.
+ *
+ *  @return  Success: 0   Failure: 1
+ */
+int octeon_init_instr_queue(struct octeon_device *octeon_dev,
+			    union oct_txpciq txpciq,
+			    u32 num_descs);
+
+/**
+ *  octeon_delete_instr_queue()
+ *  @param octeon_dev      - pointer to the octeon device structure.
+ *  @param iq_no           - queue to be deleted (0 <= q_no <= 3).
+ *
+ *  Called at driver unload time for each input queue. Deletes all
+ *  allocated resources for the input queue.
+ *
+ *  @return  Success: 0   Failure: 1
+ */
+int octeon_delete_instr_queue(struct octeon_device *octeon_dev, u32 iq_no);
+
+int lio_wait_for_instr_fetch(struct octeon_device *oct);
+
+void
+octeon_ring_doorbell_locked(struct octeon_device *oct, u32 iq_no);
+
+int
+octeon_register_reqtype_free_fn(struct octeon_device *oct, int reqtype,
+				void (*fn)(void *));
+
+int
+lio_process_iq_request_list(struct octeon_device *oct,
+			    struct octeon_instr_queue *iq, u32 napi_budget);
+
+int octeon_send_command(struct octeon_device *oct, u32 iq_no,
+			u32 force_db, void *cmd, void *buf,
+			u32 datasize, u32 reqtype);
+
+void octeon_prepare_soft_command(struct octeon_device *oct,
+				 struct octeon_soft_command *sc,
+				 u8 opcode, u8 subcode,
+				 u32 irh_ossp, u64 ossp0,
+				 u64 ossp1);
+
+int octeon_send_soft_command(struct octeon_device *oct,
+			     struct octeon_soft_command *sc);
+
+int octeon_setup_iq(struct octeon_device *oct, int ifidx,
+		    int q_index, union oct_txpciq iq_no, u32 num_descs,
+		    void *app_ctx);
+int
+octeon_flush_iq(struct octeon_device *oct, struct octeon_instr_queue *iq,
+		u32 napi_budget);
+#endif				/* __OCTEON_IQ_H__ */

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_mailbox.c

@@ -0,0 +1,406 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+#include "octeon_mailbox.h"
+#include "cn23xx_pf_device.h"
+
+static struct pci_dev *lio_vf_pci_dev_by_qno(struct octeon_device *oct, u32 q_no)
+{
+	struct pci_dev *vfdev = NULL;
+	int vfidx;
+
+	if (!oct->sriov_info.rings_per_vf)
+		return NULL;
+
+	if (q_no % oct->sriov_info.rings_per_vf)
+		return NULL;
+
+	vfidx = q_no / oct->sriov_info.rings_per_vf;
+	if (vfidx >= oct->sriov_info.num_vfs_alloced)
+		return NULL;
+
+	while ((vfdev = pci_get_device(PCI_VENDOR_ID_CAVIUM,
+				       OCTEON_CN23XX_VF_VID, vfdev))) {
+		if (pci_physfn(vfdev) && pci_physfn(vfdev) == oct->pci_dev &&
+		    pci_iov_vf_id(vfdev) == vfidx)
+			return vfdev;
+	}
+
+	return NULL;
+}
+
+/**
+ * octeon_mbox_read:
+ * @mbox: Pointer mailbox
+ *
+ * Reads the 8-bytes of data from the mbox register
+ * Writes back the acknowldgement inidcating completion of read
+ */
+int octeon_mbox_read(struct octeon_mbox *mbox)
+{
+	union octeon_mbox_message msg;
+	int ret = 0;
+
+	spin_lock(&mbox->lock);
+
+	msg.u64 = readq(mbox->mbox_read_reg);
+
+	if ((msg.u64 == OCTEON_PFVFACK) || (msg.u64 == OCTEON_PFVFSIG)) {
+		spin_unlock(&mbox->lock);
+		return 0;
+	}
+
+	if (mbox->state & OCTEON_MBOX_STATE_REQUEST_RECEIVING) {
+		mbox->mbox_req.data[mbox->mbox_req.recv_len - 1] = msg.u64;
+		mbox->mbox_req.recv_len++;
+	} else {
+		if (mbox->state & OCTEON_MBOX_STATE_RESPONSE_RECEIVING) {
+			mbox->mbox_resp.data[mbox->mbox_resp.recv_len - 1] =
+				msg.u64;
+			mbox->mbox_resp.recv_len++;
+		} else {
+			if ((mbox->state & OCTEON_MBOX_STATE_IDLE) &&
+			    (msg.s.type == OCTEON_MBOX_REQUEST)) {
+				mbox->state &= ~OCTEON_MBOX_STATE_IDLE;
+				mbox->state |=
+				    OCTEON_MBOX_STATE_REQUEST_RECEIVING;
+				mbox->mbox_req.msg.u64 = msg.u64;
+				mbox->mbox_req.q_no = mbox->q_no;
+				mbox->mbox_req.recv_len = 1;
+			} else {
+				if ((mbox->state &
+				     OCTEON_MBOX_STATE_RESPONSE_PENDING) &&
+				    (msg.s.type == OCTEON_MBOX_RESPONSE)) {
+					mbox->state &=
+					    ~OCTEON_MBOX_STATE_RESPONSE_PENDING;
+					mbox->state |=
+					    OCTEON_MBOX_STATE_RESPONSE_RECEIVING
+					    ;
+					mbox->mbox_resp.msg.u64 = msg.u64;
+					mbox->mbox_resp.q_no = mbox->q_no;
+					mbox->mbox_resp.recv_len = 1;
+				} else {
+					writeq(OCTEON_PFVFERR,
+					       mbox->mbox_read_reg);
+					mbox->state |= OCTEON_MBOX_STATE_ERROR;
+					spin_unlock(&mbox->lock);
+					return 1;
+				}
+			}
+		}
+	}
+
+	if (mbox->state & OCTEON_MBOX_STATE_REQUEST_RECEIVING) {
+		if (mbox->mbox_req.recv_len < mbox->mbox_req.msg.s.len) {
+			ret = 0;
+		} else {
+			mbox->state &= ~OCTEON_MBOX_STATE_REQUEST_RECEIVING;
+			mbox->state |= OCTEON_MBOX_STATE_REQUEST_RECEIVED;
+			ret = 1;
+		}
+	} else {
+		if (mbox->state & OCTEON_MBOX_STATE_RESPONSE_RECEIVING) {
+			if (mbox->mbox_resp.recv_len <
+			    mbox->mbox_resp.msg.s.len) {
+				ret = 0;
+			} else {
+				mbox->state &=
+				    ~OCTEON_MBOX_STATE_RESPONSE_RECEIVING;
+				mbox->state |=
+				    OCTEON_MBOX_STATE_RESPONSE_RECEIVED;
+				ret = 1;
+			}
+		} else {
+			WARN_ON(1);
+		}
+	}
+
+	writeq(OCTEON_PFVFACK, mbox->mbox_read_reg);
+
+	spin_unlock(&mbox->lock);
+
+	return ret;
+}
+
+/**
+ * octeon_mbox_write:
+ * @oct: Pointer Octeon Device
+ * @mbox_cmd: Cmd to send to mailbox.
+ *
+ * Populates the queue specific mbox structure
+ * with cmd information.
+ * Write the cmd to mbox register
+ */
+int octeon_mbox_write(struct octeon_device *oct,
+		      struct octeon_mbox_cmd *mbox_cmd)
+{
+	struct octeon_mbox *mbox = oct->mbox[mbox_cmd->q_no];
+	u32 count, i, ret = OCTEON_MBOX_STATUS_SUCCESS;
+	long timeout = LIO_MBOX_WRITE_WAIT_TIME;
+	unsigned long flags;
+
+	spin_lock_irqsave(&mbox->lock, flags);
+
+	if ((mbox_cmd->msg.s.type == OCTEON_MBOX_RESPONSE) &&
+	    !(mbox->state & OCTEON_MBOX_STATE_REQUEST_RECEIVED)) {
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		return OCTEON_MBOX_STATUS_FAILED;
+	}
+
+	if ((mbox_cmd->msg.s.type == OCTEON_MBOX_REQUEST) &&
+	    !(mbox->state & OCTEON_MBOX_STATE_IDLE)) {
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		return OCTEON_MBOX_STATUS_BUSY;
+	}
+
+	if (mbox_cmd->msg.s.type == OCTEON_MBOX_REQUEST) {
+		memcpy(&mbox->mbox_resp, mbox_cmd,
+		       sizeof(struct octeon_mbox_cmd));
+		mbox->state = OCTEON_MBOX_STATE_RESPONSE_PENDING;
+	}
+
+	spin_unlock_irqrestore(&mbox->lock, flags);
+
+	count = 0;
+
+	while (readq(mbox->mbox_write_reg) != OCTEON_PFVFSIG) {
+		schedule_timeout_uninterruptible(timeout);
+		if (count++ == LIO_MBOX_WRITE_WAIT_CNT) {
+			ret = OCTEON_MBOX_STATUS_FAILED;
+			break;
+		}
+	}
+
+	if (ret == OCTEON_MBOX_STATUS_SUCCESS) {
+		writeq(mbox_cmd->msg.u64, mbox->mbox_write_reg);
+		for (i = 0; i < (u32)(mbox_cmd->msg.s.len - 1); i++) {
+			count = 0;
+			while (readq(mbox->mbox_write_reg) !=
+			       OCTEON_PFVFACK) {
+				schedule_timeout_uninterruptible(timeout);
+				if (count++ == LIO_MBOX_WRITE_WAIT_CNT) {
+					ret = OCTEON_MBOX_STATUS_FAILED;
+					break;
+				}
+			}
+			if (ret == OCTEON_MBOX_STATUS_SUCCESS)
+				writeq(mbox_cmd->data[i], mbox->mbox_write_reg);
+			else
+				break;
+		}
+	}
+
+	spin_lock_irqsave(&mbox->lock, flags);
+	if (mbox_cmd->msg.s.type == OCTEON_MBOX_RESPONSE) {
+		mbox->state = OCTEON_MBOX_STATE_IDLE;
+		writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+	} else {
+		if ((!mbox_cmd->msg.s.resp_needed) ||
+		    (ret == OCTEON_MBOX_STATUS_FAILED)) {
+			mbox->state &= ~OCTEON_MBOX_STATE_RESPONSE_PENDING;
+			if (!(mbox->state &
+			      (OCTEON_MBOX_STATE_REQUEST_RECEIVING |
+			       OCTEON_MBOX_STATE_REQUEST_RECEIVED)))
+				mbox->state = OCTEON_MBOX_STATE_IDLE;
+		}
+	}
+	spin_unlock_irqrestore(&mbox->lock, flags);
+
+	return ret;
+}
+
+static void get_vf_stats(struct octeon_device *oct,
+			 struct oct_vf_stats *stats)
+{
+	int i;
+
+	for (i = 0; i < oct->num_iqs; i++) {
+		if (!oct->instr_queue[i])
+			continue;
+		stats->tx_packets += oct->instr_queue[i]->stats.tx_done;
+		stats->tx_bytes += oct->instr_queue[i]->stats.tx_tot_bytes;
+	}
+
+	for (i = 0; i < oct->num_oqs; i++) {
+		if (!oct->droq[i])
+			continue;
+		stats->rx_packets += oct->droq[i]->stats.rx_pkts_received;
+		stats->rx_bytes += oct->droq[i]->stats.rx_bytes_received;
+	}
+}
+
+/**
+ * octeon_mbox_process_cmd:
+ * @mbox: Pointer mailbox
+ * @mbox_cmd: Pointer to command received
+ *
+ * Process the cmd received in mbox
+ */
+static int octeon_mbox_process_cmd(struct octeon_mbox *mbox,
+				   struct octeon_mbox_cmd *mbox_cmd)
+{
+	struct octeon_device *oct = mbox->oct_dev;
+	struct pci_dev *vfdev;
+
+	switch (mbox_cmd->msg.s.cmd) {
+	case OCTEON_VF_ACTIVE:
+		dev_dbg(&oct->pci_dev->dev, "got vfactive sending data back\n");
+		mbox_cmd->msg.s.type = OCTEON_MBOX_RESPONSE;
+		mbox_cmd->msg.s.resp_needed = 1;
+		mbox_cmd->msg.s.len = 2;
+		mbox_cmd->data[0] = 0; /* VF version is in mbox_cmd->data[0] */
+		((struct lio_version *)&mbox_cmd->data[0])->major =
+			LIQUIDIO_BASE_MAJOR_VERSION;
+		((struct lio_version *)&mbox_cmd->data[0])->minor =
+			LIQUIDIO_BASE_MINOR_VERSION;
+		((struct lio_version *)&mbox_cmd->data[0])->micro =
+			LIQUIDIO_BASE_MICRO_VERSION;
+		memcpy(mbox_cmd->msg.s.params, (uint8_t *)&oct->pfvf_hsword, 6);
+		/* Sending core cofig info to the corresponding active VF.*/
+		octeon_mbox_write(oct, mbox_cmd);
+		break;
+
+	case OCTEON_VF_FLR_REQUEST:
+		dev_info(&oct->pci_dev->dev,
+			 "got a request for FLR from VF that owns DPI ring %u\n",
+			 mbox->q_no);
+		vfdev = lio_vf_pci_dev_by_qno(oct, mbox->q_no);
+		if (!vfdev)
+			break;
+
+		pcie_flr(vfdev);
+		pci_dev_put(vfdev);
+		break;
+
+	case OCTEON_PF_CHANGED_VF_MACADDR:
+		if (OCTEON_CN23XX_VF(oct))
+			octeon_pf_changed_vf_macaddr(oct,
+						     mbox_cmd->msg.s.params);
+		break;
+
+	case OCTEON_GET_VF_STATS:
+		dev_dbg(&oct->pci_dev->dev, "Got VF stats request. Sending data back\n");
+		mbox_cmd->msg.s.type = OCTEON_MBOX_RESPONSE;
+		mbox_cmd->msg.s.resp_needed = 1;
+		mbox_cmd->msg.s.len = 1 +
+			sizeof(struct oct_vf_stats) / sizeof(u64);
+		get_vf_stats(oct, (struct oct_vf_stats *)mbox_cmd->data);
+		octeon_mbox_write(oct, mbox_cmd);
+		break;
+	default:
+		break;
+	}
+	return 0;
+}
+
+/**
+ * octeon_mbox_process_message
+ * @mbox: mailbox
+ *
+ * Process the received mbox message.
+ */
+int octeon_mbox_process_message(struct octeon_mbox *mbox)
+{
+	struct octeon_mbox_cmd mbox_cmd;
+	unsigned long flags;
+
+	spin_lock_irqsave(&mbox->lock, flags);
+
+	if (mbox->state & OCTEON_MBOX_STATE_ERROR) {
+		if (mbox->state & (OCTEON_MBOX_STATE_RESPONSE_PENDING |
+				   OCTEON_MBOX_STATE_RESPONSE_RECEIVING)) {
+			memcpy(&mbox_cmd, &mbox->mbox_resp,
+			       sizeof(struct octeon_mbox_cmd));
+			mbox->state = OCTEON_MBOX_STATE_IDLE;
+			writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+			spin_unlock_irqrestore(&mbox->lock, flags);
+			mbox_cmd.recv_status = 1;
+			if (mbox_cmd.fn)
+				mbox_cmd.fn(mbox->oct_dev, &mbox_cmd,
+					    mbox_cmd.fn_arg);
+			return 0;
+		}
+
+		mbox->state = OCTEON_MBOX_STATE_IDLE;
+		writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		return 0;
+	}
+
+	if (mbox->state & OCTEON_MBOX_STATE_RESPONSE_RECEIVED) {
+		memcpy(&mbox_cmd, &mbox->mbox_resp,
+		       sizeof(struct octeon_mbox_cmd));
+		mbox->state = OCTEON_MBOX_STATE_IDLE;
+		writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		mbox_cmd.recv_status = 0;
+		if (mbox_cmd.fn)
+			mbox_cmd.fn(mbox->oct_dev, &mbox_cmd, mbox_cmd.fn_arg);
+		return 0;
+	}
+
+	if (mbox->state & OCTEON_MBOX_STATE_REQUEST_RECEIVED) {
+		memcpy(&mbox_cmd, &mbox->mbox_req,
+		       sizeof(struct octeon_mbox_cmd));
+		if (!mbox_cmd.msg.s.resp_needed) {
+			mbox->state &= ~OCTEON_MBOX_STATE_REQUEST_RECEIVED;
+			if (!(mbox->state &
+			      OCTEON_MBOX_STATE_RESPONSE_PENDING))
+				mbox->state = OCTEON_MBOX_STATE_IDLE;
+			writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+		}
+
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		octeon_mbox_process_cmd(mbox, &mbox_cmd);
+		return 0;
+	}
+
+	spin_unlock_irqrestore(&mbox->lock, flags);
+	WARN_ON(1);
+
+	return 0;
+}
+
+int octeon_mbox_cancel(struct octeon_device *oct, int q_no)
+{
+	struct octeon_mbox *mbox = oct->mbox[q_no];
+	struct octeon_mbox_cmd *mbox_cmd;
+	unsigned long flags = 0;
+
+	spin_lock_irqsave(&mbox->lock, flags);
+	mbox_cmd = &mbox->mbox_resp;
+
+	if (!(mbox->state & OCTEON_MBOX_STATE_RESPONSE_PENDING)) {
+		spin_unlock_irqrestore(&mbox->lock, flags);
+		return 1;
+	}
+
+	mbox->state = OCTEON_MBOX_STATE_IDLE;
+	memset(mbox_cmd, 0, sizeof(*mbox_cmd));
+	writeq(OCTEON_PFVFSIG, mbox->mbox_read_reg);
+	spin_unlock_irqrestore(&mbox->lock, flags);
+
+	return 0;
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_mailbox.h

@@ -0,0 +1,125 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+#ifndef __MAILBOX_H__
+#define __MAILBOX_H__
+
+/* Macros for Mail Box Communication */
+
+#define OCTEON_MBOX_DATA_MAX		32
+
+#define OCTEON_VF_ACTIVE		0x1
+#define OCTEON_VF_FLR_REQUEST		0x2
+#define OCTEON_PF_CHANGED_VF_MACADDR	0x4
+#define OCTEON_GET_VF_STATS		0x8
+
+/*Macro for Read acknowldgement*/
+#define OCTEON_PFVFACK			0xffffffffffffffffULL
+#define OCTEON_PFVFSIG			0x1122334455667788ULL
+#define OCTEON_PFVFERR			0xDEADDEADDEADDEADULL
+
+#define LIO_MBOX_WRITE_WAIT_CNT         1000
+#define LIO_MBOX_WRITE_WAIT_TIME        msecs_to_jiffies(1)
+
+enum octeon_mbox_cmd_status {
+	OCTEON_MBOX_STATUS_SUCCESS = 0,
+	OCTEON_MBOX_STATUS_FAILED = 1,
+	OCTEON_MBOX_STATUS_BUSY = 2
+};
+
+enum octeon_mbox_message_type {
+	OCTEON_MBOX_REQUEST = 0,
+	OCTEON_MBOX_RESPONSE = 1
+};
+
+union octeon_mbox_message {
+	u64 u64;
+	struct {
+		u16 type : 1;
+		u16 resp_needed : 1;
+		u16 cmd : 6;
+		u16 len : 8;
+		u8 params[6];
+	} s;
+};
+
+struct octeon_mbox_cmd;
+
+typedef void (*octeon_mbox_callback_t)(struct octeon_device *,
+				       struct octeon_mbox_cmd *, void *);
+
+struct octeon_mbox_cmd {
+	union octeon_mbox_message msg;
+	u64 data[OCTEON_MBOX_DATA_MAX];
+	u32 q_no;
+	u32 recv_len;
+	u32 recv_status;
+	octeon_mbox_callback_t fn;
+	void *fn_arg;
+};
+
+enum octeon_mbox_state {
+	OCTEON_MBOX_STATE_IDLE = 1,
+	OCTEON_MBOX_STATE_REQUEST_RECEIVING = 2,
+	OCTEON_MBOX_STATE_REQUEST_RECEIVED = 4,
+	OCTEON_MBOX_STATE_RESPONSE_PENDING = 8,
+	OCTEON_MBOX_STATE_RESPONSE_RECEIVING = 16,
+	OCTEON_MBOX_STATE_RESPONSE_RECEIVED = 32,
+	OCTEON_MBOX_STATE_ERROR = 64
+};
+
+struct octeon_mbox {
+	/** A spinlock to protect access to this q_mbox. */
+	spinlock_t lock;
+
+	struct octeon_device *oct_dev;
+
+	u32 q_no;
+
+	enum octeon_mbox_state state;
+
+	struct cavium_wk mbox_poll_wk;
+
+	/** SLI_MAC_PF_MBOX_INT for PF, SLI_PKT_MBOX_INT for VF. */
+	void *mbox_int_reg;
+
+	/** SLI_PKT_PF_VF_MBOX_SIG(0) for PF, SLI_PKT_PF_VF_MBOX_SIG(1) for VF.
+	 */
+	void *mbox_write_reg;
+
+	/** SLI_PKT_PF_VF_MBOX_SIG(1) for PF, SLI_PKT_PF_VF_MBOX_SIG(0) for VF.
+	 */
+	void *mbox_read_reg;
+
+	struct octeon_mbox_cmd mbox_req;
+
+	struct octeon_mbox_cmd mbox_resp;
+
+};
+
+struct oct_vf_stats_ctx {
+	atomic_t status;
+	struct oct_vf_stats *stats;
+};
+
+int octeon_mbox_read(struct octeon_mbox *mbox);
+int octeon_mbox_write(struct octeon_device *oct,
+		      struct octeon_mbox_cmd *mbox_cmd);
+int octeon_mbox_process_message(struct octeon_mbox *mbox);
+int octeon_mbox_cancel(struct octeon_device *oct, int q_no);
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_main.h

@@ -0,0 +1,236 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more details.
+ ***********************************************************************/
+/*! \file octeon_main.h
+ *  \brief Host Driver: This file is included by all host driver source files
+ *  to include common definitions.
+ */
+
+#ifndef _OCTEON_MAIN_H_
+#define  _OCTEON_MAIN_H_
+
+#include <linux/sched/signal.h>
+
+#if BITS_PER_LONG == 32
+#define CVM_CAST64(v) ((long long)(v))
+#elif BITS_PER_LONG == 64
+#define CVM_CAST64(v) ((long long)(long)(v))
+#else
+#error "Unknown system architecture"
+#endif
+
+#define DRV_NAME "LiquidIO"
+
+struct octeon_device_priv {
+	/** Tasklet structures for this device. */
+	struct tasklet_struct droq_tasklet;
+	unsigned long napi_mask;
+	struct octeon_device *dev;
+};
+
+/** This structure is used by NIC driver to store information required
+ * to free the sk_buff when the packet has been fetched by Octeon.
+ * Bytes offset below assume worst-case of a 64-bit system.
+ */
+struct octnet_buf_free_info {
+	/** Bytes 1-8.  Pointer to network device private structure. */
+	struct lio *lio;
+
+	/** Bytes 9-16.  Pointer to sk_buff. */
+	struct sk_buff *skb;
+
+	/** Bytes 17-24.  Pointer to gather list. */
+	struct octnic_gather *g;
+
+	/** Bytes 25-32. Physical address of skb->data or gather list. */
+	u64 dptr;
+
+	/** Bytes 33-47. Piggybacked soft command, if any */
+	struct octeon_soft_command *sc;
+};
+
+/* BQL-related functions */
+int octeon_report_sent_bytes_to_bql(void *buf, int reqtype);
+void octeon_update_tx_completion_counters(void *buf, int reqtype,
+					  unsigned int *pkts_compl,
+					  unsigned int *bytes_compl);
+void octeon_report_tx_completion_to_bql(void *txq, unsigned int pkts_compl,
+					unsigned int bytes_compl);
+void octeon_pf_changed_vf_macaddr(struct octeon_device *oct, u8 *mac);
+
+void octeon_schedule_rxq_oom_work(struct octeon_device *oct,
+				  struct octeon_droq *droq);
+
+/** Swap 8B blocks */
+static inline void octeon_swap_8B_data(u64 *data, u32 blocks)
+{
+	while (blocks) {
+		cpu_to_be64s(data);
+		blocks--;
+		data++;
+	}
+}
+
+/**
+ * \brief unmaps a PCI BAR
+ * @param oct Pointer to Octeon device
+ * @param baridx bar index
+ */
+static inline void octeon_unmap_pci_barx(struct octeon_device *oct, int baridx)
+{
+	dev_dbg(&oct->pci_dev->dev, "Freeing PCI mapped regions for Bar%d\n",
+		baridx);
+
+	if (oct->mmio[baridx].done)
+		iounmap(oct->mmio[baridx].hw_addr);
+
+	if (oct->mmio[baridx].start)
+		pci_release_region(oct->pci_dev, baridx * 2);
+}
+
+/**
+ * \brief maps a PCI BAR
+ * @param oct Pointer to Octeon device
+ * @param baridx bar index
+ * @param max_map_len maximum length of mapped memory
+ */
+static inline int octeon_map_pci_barx(struct octeon_device *oct,
+				      int baridx, int max_map_len)
+{
+	u32 mapped_len = 0;
+
+	if (pci_request_region(oct->pci_dev, baridx * 2, DRV_NAME)) {
+		dev_err(&oct->pci_dev->dev, "pci_request_region failed for bar %d\n",
+			baridx);
+		return 1;
+	}
+
+	oct->mmio[baridx].start = pci_resource_start(oct->pci_dev, baridx * 2);
+	oct->mmio[baridx].len = pci_resource_len(oct->pci_dev, baridx * 2);
+
+	mapped_len = oct->mmio[baridx].len;
+	if (!mapped_len)
+		goto err_release_region;
+
+	if (max_map_len && (mapped_len > max_map_len))
+		mapped_len = max_map_len;
+
+	oct->mmio[baridx].hw_addr =
+		ioremap(oct->mmio[baridx].start, mapped_len);
+	oct->mmio[baridx].mapped_len = mapped_len;
+
+	dev_dbg(&oct->pci_dev->dev, "BAR%d start: 0x%llx mapped %u of %u bytes\n",
+		baridx, oct->mmio[baridx].start, mapped_len,
+		oct->mmio[baridx].len);
+
+	if (!oct->mmio[baridx].hw_addr) {
+		dev_err(&oct->pci_dev->dev, "error ioremap for bar %d\n",
+			baridx);
+		goto err_release_region;
+	}
+	oct->mmio[baridx].done = 1;
+
+	return 0;
+
+err_release_region:
+	pci_release_region(oct->pci_dev, baridx * 2);
+	return 1;
+}
+
+/* input parameter:
+ * sc: pointer to a soft request
+ * timeout: milli sec which an application wants to wait for the
+	    response of the request.
+ *          0: the request will wait until its response gets back
+ *	       from the firmware within LIO_SC_MAX_TMO_MS milli sec.
+ *	       If the response does not return within
+ *	       LIO_SC_MAX_TMO_MS milli sec, lio_process_ordered_list()
+ *	       will move the request to zombie response list.
+ *
+ * return value:
+ * 0: got the response from firmware for the sc request.
+ * errno -EINTR: user abort the command.
+ * errno -ETIME: user spefified timeout value has been expired.
+ * errno -EBUSY: the response of the request does not return in
+ *               resonable time (LIO_SC_MAX_TMO_MS).
+ *               the sc wll be move to zombie response list by
+ *               lio_process_ordered_list()
+ *
+ * A request with non-zero return value, the sc->caller_is_done
+ *  will be marked 1.
+ * When getting a request with zero return value, the requestor
+ *  should mark sc->caller_is_done with 1 after examing the
+ *  response of sc.
+ * lio_process_ordered_list() will free the soft command on behalf
+ * of the soft command requestor.
+ * This is to fix the possible race condition of both timeout process
+ * and lio_process_ordered_list()/callback function to free a
+ * sc strucutre.
+ */
+static inline int
+wait_for_sc_completion_timeout(struct octeon_device *oct_dev,
+			       struct octeon_soft_command *sc,
+			       unsigned long timeout)
+{
+	int errno = 0;
+	long timeout_jiff;
+
+	if (timeout)
+		timeout_jiff = msecs_to_jiffies(timeout);
+	else
+		timeout_jiff = MAX_SCHEDULE_TIMEOUT;
+
+	timeout_jiff =
+		wait_for_completion_interruptible_timeout(&sc->complete,
+							  timeout_jiff);
+	if (timeout_jiff == 0) {
+		dev_err(&oct_dev->pci_dev->dev, "%s: sc is timeout\n",
+			__func__);
+		WRITE_ONCE(sc->caller_is_done, true);
+		errno = -ETIME;
+	} else if (timeout_jiff == -ERESTARTSYS) {
+		dev_err(&oct_dev->pci_dev->dev, "%s: sc is interrupted\n",
+			__func__);
+		WRITE_ONCE(sc->caller_is_done, true);
+		errno = -EINTR;
+	} else  if (sc->sc_status == OCTEON_REQUEST_TIMEOUT) {
+		dev_err(&oct_dev->pci_dev->dev, "%s: sc has fatal timeout\n",
+			__func__);
+		WRITE_ONCE(sc->caller_is_done, true);
+		errno = -EBUSY;
+	}
+
+	return errno;
+}
+
+#ifndef ROUNDUP4
+#define ROUNDUP4(val) (((val) + 3) & 0xfffffffc)
+#endif
+
+#ifndef ROUNDUP8
+#define ROUNDUP8(val) (((val) + 7) & 0xfffffff8)
+#endif
+
+#ifndef ROUNDUP16
+#define ROUNDUP16(val) (((val) + 15) & 0xfffffff0)
+#endif
+
+#ifndef ROUNDUP128
+#define ROUNDUP128(val) (((val) + 127) & 0xffffff80)
+#endif
+
+#endif /* _OCTEON_MAIN_H_ */

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_mem_ops.c

@@ -0,0 +1,206 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_mem_ops.h"
+
+#define MEMOPS_IDX   BAR1_INDEX_DYNAMIC_MAP
+
+#ifdef __BIG_ENDIAN_BITFIELD
+static inline void
+octeon_toggle_bar1_swapmode(struct octeon_device *oct, u32 idx)
+{
+	u32 mask;
+
+	mask = oct->fn_list.bar1_idx_read(oct, idx);
+	mask = (mask & 0x2) ? (mask & ~2) : (mask | 2);
+	oct->fn_list.bar1_idx_write(oct, idx, mask);
+}
+#else
+#define octeon_toggle_bar1_swapmode(oct, idx)
+#endif
+
+static void
+octeon_pci_fastwrite(struct octeon_device *oct, u8 __iomem *mapped_addr,
+		     u8 *hostbuf, u32 len)
+{
+	while ((len) && ((unsigned long)mapped_addr) & 7) {
+		writeb(*(hostbuf++), mapped_addr++);
+		len--;
+	}
+
+	octeon_toggle_bar1_swapmode(oct, MEMOPS_IDX);
+
+	while (len >= 8) {
+		writeq(*((u64 *)hostbuf), mapped_addr);
+		mapped_addr += 8;
+		hostbuf += 8;
+		len -= 8;
+	}
+
+	octeon_toggle_bar1_swapmode(oct, MEMOPS_IDX);
+
+	while (len--)
+		writeb(*(hostbuf++), mapped_addr++);
+}
+
+static void
+octeon_pci_fastread(struct octeon_device *oct, u8 __iomem *mapped_addr,
+		    u8 *hostbuf, u32 len)
+{
+	while ((len) && ((unsigned long)mapped_addr) & 7) {
+		*(hostbuf++) = readb(mapped_addr++);
+		len--;
+	}
+
+	octeon_toggle_bar1_swapmode(oct, MEMOPS_IDX);
+
+	while (len >= 8) {
+		*((u64 *)hostbuf) = readq(mapped_addr);
+		mapped_addr += 8;
+		hostbuf += 8;
+		len -= 8;
+	}
+
+	octeon_toggle_bar1_swapmode(oct, MEMOPS_IDX);
+
+	while (len--)
+		*(hostbuf++) = readb(mapped_addr++);
+}
+
+/* Core mem read/write with temporary bar1 settings. */
+/* op = 1 to read, op = 0 to write. */
+static void
+__octeon_pci_rw_core_mem(struct octeon_device *oct, u64 addr,
+			 u8 *hostbuf, u32 len, u32 op)
+{
+	u32 copy_len = 0, index_reg_val = 0;
+	unsigned long flags;
+	u8 __iomem *mapped_addr;
+	u64 static_mapping_base;
+
+	static_mapping_base = oct->console_nb_info.dram_region_base;
+
+	if (static_mapping_base &&
+	    static_mapping_base == (addr & ~(OCTEON_BAR1_ENTRY_SIZE - 1ULL))) {
+		int bar1_index = oct->console_nb_info.bar1_index;
+
+		mapped_addr = oct->mmio[1].hw_addr
+			+ (bar1_index << ilog2(OCTEON_BAR1_ENTRY_SIZE))
+			+ (addr & (OCTEON_BAR1_ENTRY_SIZE - 1ULL));
+
+		if (op)
+			octeon_pci_fastread(oct, mapped_addr, hostbuf, len);
+		else
+			octeon_pci_fastwrite(oct, mapped_addr, hostbuf, len);
+
+		return;
+	}
+
+	spin_lock_irqsave(&oct->mem_access_lock, flags);
+
+	/* Save the original index reg value. */
+	index_reg_val = oct->fn_list.bar1_idx_read(oct, MEMOPS_IDX);
+	do {
+		oct->fn_list.bar1_idx_setup(oct, addr, MEMOPS_IDX, 1);
+		mapped_addr = oct->mmio[1].hw_addr
+		    + (MEMOPS_IDX << 22) + (addr & 0x3fffff);
+
+		/* If operation crosses a 4MB boundary, split the transfer
+		 * at the 4MB
+		 * boundary.
+		 */
+		if (((addr + len - 1) & ~(0x3fffff)) != (addr & ~(0x3fffff))) {
+			copy_len = (u32)(((addr & ~(0x3fffff)) +
+				   (MEMOPS_IDX << 22)) - addr);
+		} else {
+			copy_len = len;
+		}
+
+		if (op) {	/* read from core */
+			octeon_pci_fastread(oct, mapped_addr, hostbuf,
+					    copy_len);
+		} else {
+			octeon_pci_fastwrite(oct, mapped_addr, hostbuf,
+					     copy_len);
+		}
+
+		len -= copy_len;
+		addr += copy_len;
+		hostbuf += copy_len;
+
+	} while (len);
+
+	oct->fn_list.bar1_idx_write(oct, MEMOPS_IDX, index_reg_val);
+
+	spin_unlock_irqrestore(&oct->mem_access_lock, flags);
+}
+
+void
+octeon_pci_read_core_mem(struct octeon_device *oct,
+			 u64 coreaddr,
+			 u8 *buf,
+			 u32 len)
+{
+	__octeon_pci_rw_core_mem(oct, coreaddr, buf, len, 1);
+}
+EXPORT_SYMBOL_GPL(octeon_pci_read_core_mem);
+
+void
+octeon_pci_write_core_mem(struct octeon_device *oct,
+			  u64 coreaddr,
+			  const u8 *buf,
+			  u32 len)
+{
+	__octeon_pci_rw_core_mem(oct, coreaddr, (u8 *)buf, len, 0);
+}
+EXPORT_SYMBOL_GPL(octeon_pci_write_core_mem);
+
+u64 octeon_read_device_mem64(struct octeon_device *oct, u64 coreaddr)
+{
+	__be64 ret;
+
+	__octeon_pci_rw_core_mem(oct, coreaddr, (u8 *)&ret, 8, 1);
+
+	return be64_to_cpu(ret);
+}
+EXPORT_SYMBOL_GPL(octeon_read_device_mem64);
+
+u32 octeon_read_device_mem32(struct octeon_device *oct, u64 coreaddr)
+{
+	__be32 ret;
+
+	__octeon_pci_rw_core_mem(oct, coreaddr, (u8 *)&ret, 4, 1);
+
+	return be32_to_cpu(ret);
+}
+EXPORT_SYMBOL_GPL(octeon_read_device_mem32);
+
+void octeon_write_device_mem32(struct octeon_device *oct, u64 coreaddr,
+			       u32 val)
+{
+	__be32 t = cpu_to_be32(val);
+
+	__octeon_pci_rw_core_mem(oct, coreaddr, (u8 *)&t, 4, 0);
+}
+EXPORT_SYMBOL_GPL(octeon_write_device_mem32);

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_mem_ops.h

@@ -0,0 +1,72 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+
+/*!  \file octeon_mem_ops.h
+ *   \brief Host Driver: Routines used to read/write Octeon memory.
+ */
+
+#ifndef __OCTEON_MEM_OPS_H__
+#define __OCTEON_MEM_OPS_H__
+
+/**  Read a 64-bit value from a BAR1 mapped core memory address.
+ *   @param  oct        -  pointer to the octeon device.
+ *   @param  core_addr  -  the address to read from.
+ *
+ *   The range_idx gives the BAR1 index register for the range of address
+ *   in which core_addr is mapped.
+ *
+ *   @return  64-bit value read from Core memory
+ */
+u64 octeon_read_device_mem64(struct octeon_device *oct, u64 core_addr);
+
+/**  Read a 32-bit value from a BAR1 mapped core memory address.
+ *   @param  oct        -  pointer to the octeon device.
+ *   @param  core_addr  -  the address to read from.
+ *
+ *   @return  32-bit value read from Core memory
+ */
+u32 octeon_read_device_mem32(struct octeon_device *oct, u64 core_addr);
+
+/**  Write a 32-bit value to a BAR1 mapped core memory address.
+ *   @param  oct        -  pointer to the octeon device.
+ *   @param  core_addr  -  the address to write to.
+ *   @param  val        -  32-bit value to write.
+ */
+void
+octeon_write_device_mem32(struct octeon_device *oct,
+			  u64 core_addr,
+			  u32 val);
+
+/** Read multiple bytes from Octeon memory.
+ */
+void
+octeon_pci_read_core_mem(struct octeon_device *oct,
+			 u64 coreaddr,
+			 u8 *buf,
+			 u32 len);
+
+/** Write multiple bytes into Octeon memory.
+ */
+void
+octeon_pci_write_core_mem(struct octeon_device *oct,
+			  u64 coreaddr,
+			  const u8 *buf,
+			  u32 len);
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_network.h

@@ -0,0 +1,626 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+
+/*!  \file  octeon_network.h
+ *   \brief Host NIC Driver: Structure and Macro definitions used by NIC Module.
+ */
+
+#ifndef __OCTEON_NETWORK_H__
+#define __OCTEON_NETWORK_H__
+#include <linux/ptp_clock_kernel.h>
+
+#define LIO_MAX_MTU_SIZE (OCTNET_MAX_FRM_SIZE - OCTNET_FRM_HEADER_SIZE)
+#define LIO_MIN_MTU_SIZE ETH_MIN_MTU
+
+/* Bit mask values for lio->ifstate */
+#define   LIO_IFSTATE_DROQ_OPS             0x01
+#define   LIO_IFSTATE_REGISTERED           0x02
+#define   LIO_IFSTATE_RUNNING              0x04
+#define   LIO_IFSTATE_RX_TIMESTAMP_ENABLED 0x08
+#define   LIO_IFSTATE_RESETTING		   0x10
+
+struct liquidio_if_cfg_resp {
+	u64 rh;
+	struct liquidio_if_cfg_info cfg_info;
+	u64 status;
+};
+
+#define LIO_IFCFG_WAIT_TIME    3000 /* In milli seconds */
+#define LIQUIDIO_NDEV_STATS_POLL_TIME_MS 200
+
+/* Structure of a node in list of gather components maintained by
+ * NIC driver for each network device.
+ */
+struct octnic_gather {
+	/* List manipulation. Next and prev pointers. */
+	struct list_head list;
+
+	/* Size of the gather component at sg in bytes. */
+	int sg_size;
+
+	/* Number of bytes that sg was adjusted to make it 8B-aligned. */
+	int adjust;
+
+	/* Gather component that can accommodate max sized fragment list
+	 * received from the IP layer.
+	 */
+	struct octeon_sg_entry *sg;
+
+	dma_addr_t sg_dma_ptr;
+};
+
+struct oct_nic_stats_resp {
+	u64     rh;
+	struct oct_link_stats stats;
+	u64     status;
+};
+
+struct oct_nic_vf_stats_resp {
+	u64     rh;
+	u64	spoofmac_cnt;
+	u64     status;
+};
+
+struct oct_nic_stats_ctrl {
+	struct completion complete;
+	struct net_device *netdev;
+};
+
+struct oct_nic_seapi_resp {
+	u64 rh;
+	union {
+		u32 fec_setting;
+		u32 speed;
+	};
+	u64 status;
+};
+
+/** LiquidIO per-interface network private data */
+struct lio {
+	/** State of the interface. Rx/Tx happens only in the RUNNING state.  */
+	atomic_t ifstate;
+
+	/** Octeon Interface index number. This device will be represented as
+	 *  oct<ifidx> in the system.
+	 */
+	int ifidx;
+
+	/** Octeon Input queue to use to transmit for this network interface. */
+	int txq;
+
+	/** Octeon Output queue from which pkts arrive
+	 * for this network interface.
+	 */
+	int rxq;
+
+	/** Guards each glist */
+	spinlock_t *glist_lock;
+
+	/** Array of gather component linked lists */
+	struct list_head *glist;
+	void **glists_virt_base;
+	dma_addr_t *glists_dma_base;
+	u32 glist_entry_size;
+
+	/** Pointer to the NIC properties for the Octeon device this network
+	 *  interface is associated with.
+	 */
+	struct octdev_props *octprops;
+
+	/** Pointer to the octeon device structure. */
+	struct octeon_device *oct_dev;
+
+	struct net_device *netdev;
+
+	/** Link information sent by the core application for this interface. */
+	struct oct_link_info linfo;
+
+	/** counter of link changes */
+	u64 link_changes;
+
+	/** Size of Tx queue for this octeon device. */
+	u32 tx_qsize;
+
+	/** Size of Rx queue for this octeon device. */
+	u32 rx_qsize;
+
+	/** Size of MTU this octeon device. */
+	u32 mtu;
+
+	/** msg level flag per interface. */
+	u32 msg_enable;
+
+	/** Copy of Interface capabilities: TSO, TSO6, LRO, Chescksums . */
+	u64 dev_capability;
+
+	/* Copy of transmit encapsulation capabilities:
+	 * TSO, TSO6, Checksums for this device for Kernel
+	 * 3.10.0 onwards
+	 */
+	u64 enc_dev_capability;
+
+	/** Copy of beacaon reg in phy */
+	u32 phy_beacon_val;
+
+	/** Copy of ctrl reg in phy */
+	u32 led_ctrl_val;
+
+	/* PTP clock information */
+	struct ptp_clock_info ptp_info;
+	struct ptp_clock *ptp_clock;
+	s64 ptp_adjust;
+
+	/* for atomic access to Octeon PTP reg and data struct */
+	spinlock_t ptp_lock;
+
+	/* Interface info */
+	u32	intf_open;
+
+	/* work queue for  txq status */
+	struct cavium_wq	txq_status_wq;
+
+	/* work queue for  rxq oom status */
+	struct cavium_wq rxq_status_wq[MAX_POSSIBLE_OCTEON_OUTPUT_QUEUES];
+
+	/* work queue for  link status */
+	struct cavium_wq	link_status_wq;
+
+	/* work queue to regularly send local time to octeon firmware */
+	struct cavium_wq	sync_octeon_time_wq;
+
+	int netdev_uc_count;
+	struct cavium_wk stats_wk;
+};
+
+#define LIO_SIZE         (sizeof(struct lio))
+#define GET_LIO(netdev)  ((struct lio *)netdev_priv(netdev))
+
+#define LIO_MAX_CORES                16
+
+/**
+ * \brief Enable or disable feature
+ * @param netdev    pointer to network device
+ * @param cmd       Command that just requires acknowledgment
+ * @param param1    Parameter to command
+ */
+int liquidio_set_feature(struct net_device *netdev, int cmd, u16 param1);
+
+int setup_rx_oom_poll_fn(struct net_device *netdev);
+
+void cleanup_rx_oom_poll_fn(struct net_device *netdev);
+
+/**
+ * \brief Link control command completion callback
+ * @param nctrl_ptr pointer to control packet structure
+ *
+ * This routine is called by the callback function when a ctrl pkt sent to
+ * core app completes. The nctrl_ptr contains a copy of the command type
+ * and data sent to the core app. This routine is only called if the ctrl
+ * pkt was sent successfully to the core app.
+ */
+void liquidio_link_ctrl_cmd_completion(void *nctrl_ptr);
+
+int liquidio_setup_io_queues(struct octeon_device *octeon_dev, int ifidx,
+			     u32 num_iqs, u32 num_oqs);
+
+irqreturn_t liquidio_msix_intr_handler(int irq __attribute__((unused)),
+				       void *dev);
+
+int octeon_setup_interrupt(struct octeon_device *oct, u32 num_ioqs);
+
+void lio_fetch_stats(struct work_struct *work);
+
+int lio_wait_for_clean_oq(struct octeon_device *oct);
+/**
+ * \brief Register ethtool operations
+ * @param netdev    pointer to network device
+ */
+void liquidio_set_ethtool_ops(struct net_device *netdev);
+
+void lio_delete_glists(struct lio *lio);
+
+int lio_setup_glists(struct octeon_device *oct, struct lio *lio, int num_qs);
+
+int liquidio_get_speed(struct lio *lio);
+int liquidio_set_speed(struct lio *lio, int speed);
+int liquidio_get_fec(struct lio *lio);
+int liquidio_set_fec(struct lio *lio, int on_off);
+
+/**
+ * \brief Net device change_mtu
+ * @param netdev network device
+ */
+int liquidio_change_mtu(struct net_device *netdev, int new_mtu);
+#define LIO_CHANGE_MTU_SUCCESS 1
+#define LIO_CHANGE_MTU_FAIL    2
+
+#define SKB_ADJ_MASK  0x3F
+#define SKB_ADJ       (SKB_ADJ_MASK + 1)
+
+#define MIN_SKB_SIZE       256 /* 8 bytes and more - 8 bytes for PTP */
+#define LIO_RXBUFFER_SZ    2048
+
+static inline void
+*recv_buffer_alloc(struct octeon_device *oct,
+		   struct octeon_skb_page_info *pg_info)
+{
+	struct page *page;
+	struct sk_buff *skb;
+	struct octeon_skb_page_info *skb_pg_info;
+
+	page = alloc_page(GFP_ATOMIC);
+	if (unlikely(!page))
+		return NULL;
+
+	skb = dev_alloc_skb(MIN_SKB_SIZE + SKB_ADJ);
+	if (unlikely(!skb)) {
+		__free_page(page);
+		pg_info->page = NULL;
+		return NULL;
+	}
+
+	if ((unsigned long)skb->data & SKB_ADJ_MASK) {
+		u32 r = SKB_ADJ - ((unsigned long)skb->data & SKB_ADJ_MASK);
+
+		skb_reserve(skb, r);
+	}
+
+	skb_pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+	/* Get DMA info */
+	pg_info->dma = dma_map_page(&oct->pci_dev->dev, page, 0,
+				    PAGE_SIZE, DMA_FROM_DEVICE);
+
+	/* Mapping failed!! */
+	if (dma_mapping_error(&oct->pci_dev->dev, pg_info->dma)) {
+		__free_page(page);
+		dev_kfree_skb_any((struct sk_buff *)skb);
+		pg_info->page = NULL;
+		return NULL;
+	}
+
+	pg_info->page = page;
+	pg_info->page_offset = 0;
+	skb_pg_info->page = page;
+	skb_pg_info->page_offset = 0;
+	skb_pg_info->dma = pg_info->dma;
+
+	return (void *)skb;
+}
+
+static inline void
+*recv_buffer_fast_alloc(u32 size)
+{
+	struct sk_buff *skb;
+	struct octeon_skb_page_info *skb_pg_info;
+
+	skb = dev_alloc_skb(size + SKB_ADJ);
+	if (unlikely(!skb))
+		return NULL;
+
+	if ((unsigned long)skb->data & SKB_ADJ_MASK) {
+		u32 r = SKB_ADJ - ((unsigned long)skb->data & SKB_ADJ_MASK);
+
+		skb_reserve(skb, r);
+	}
+
+	skb_pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+	skb_pg_info->page = NULL;
+	skb_pg_info->page_offset = 0;
+	skb_pg_info->dma = 0;
+
+	return skb;
+}
+
+static inline int
+recv_buffer_recycle(struct octeon_device *oct, void *buf)
+{
+	struct octeon_skb_page_info *pg_info = buf;
+
+	if (!pg_info->page) {
+		dev_err(&oct->pci_dev->dev, "%s: pg_info->page NULL\n",
+			__func__);
+		return -ENOMEM;
+	}
+
+	if (unlikely(page_count(pg_info->page) != 1) ||
+	    unlikely(page_to_nid(pg_info->page)	!= numa_node_id())) {
+		dma_unmap_page(&oct->pci_dev->dev,
+			       pg_info->dma, (PAGE_SIZE << 0),
+			       DMA_FROM_DEVICE);
+		pg_info->dma = 0;
+		pg_info->page = NULL;
+		pg_info->page_offset = 0;
+		return -ENOMEM;
+	}
+
+	/* Flip to other half of the buffer */
+	if (pg_info->page_offset == 0)
+		pg_info->page_offset = LIO_RXBUFFER_SZ;
+	else
+		pg_info->page_offset = 0;
+	page_ref_inc(pg_info->page);
+
+	return 0;
+}
+
+static inline void
+*recv_buffer_reuse(struct octeon_device *oct, void *buf)
+{
+	struct octeon_skb_page_info *pg_info = buf, *skb_pg_info;
+	struct sk_buff *skb;
+
+	skb = dev_alloc_skb(MIN_SKB_SIZE + SKB_ADJ);
+	if (unlikely(!skb)) {
+		dma_unmap_page(&oct->pci_dev->dev,
+			       pg_info->dma, (PAGE_SIZE << 0),
+			       DMA_FROM_DEVICE);
+		return NULL;
+	}
+
+	if ((unsigned long)skb->data & SKB_ADJ_MASK) {
+		u32 r = SKB_ADJ - ((unsigned long)skb->data & SKB_ADJ_MASK);
+
+		skb_reserve(skb, r);
+	}
+
+	skb_pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+	skb_pg_info->page = pg_info->page;
+	skb_pg_info->page_offset = pg_info->page_offset;
+	skb_pg_info->dma = pg_info->dma;
+
+	return skb;
+}
+
+static inline void
+recv_buffer_destroy(void *buffer, struct octeon_skb_page_info *pg_info)
+{
+	struct sk_buff *skb = (struct sk_buff *)buffer;
+
+	put_page(pg_info->page);
+	pg_info->dma = 0;
+	pg_info->page = NULL;
+	pg_info->page_offset = 0;
+
+	if (skb)
+		dev_kfree_skb_any(skb);
+}
+
+static inline void recv_buffer_free(void *buffer)
+{
+	struct sk_buff *skb = (struct sk_buff *)buffer;
+	struct octeon_skb_page_info *pg_info;
+
+	pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+
+	if (pg_info->page) {
+		put_page(pg_info->page);
+		pg_info->dma = 0;
+		pg_info->page = NULL;
+		pg_info->page_offset = 0;
+	}
+
+	dev_kfree_skb_any((struct sk_buff *)buffer);
+}
+
+static inline void
+recv_buffer_fast_free(void *buffer)
+{
+	dev_kfree_skb_any((struct sk_buff *)buffer);
+}
+
+static inline void tx_buffer_free(void *buffer)
+{
+	dev_kfree_skb_any((struct sk_buff *)buffer);
+}
+
+#define lio_dma_alloc(oct, size, dma_addr) \
+	dma_alloc_coherent(&(oct)->pci_dev->dev, size, dma_addr, GFP_KERNEL)
+#define lio_dma_free(oct, size, virt_addr, dma_addr) \
+	dma_free_coherent(&(oct)->pci_dev->dev, size, virt_addr, dma_addr)
+
+static inline
+void *get_rbd(struct sk_buff *skb)
+{
+	struct octeon_skb_page_info *pg_info;
+	unsigned char *va;
+
+	pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+	va = page_address(pg_info->page) + pg_info->page_offset;
+
+	return va;
+}
+
+static inline u64
+lio_map_ring(void *buf)
+{
+	dma_addr_t dma_addr;
+
+	struct sk_buff *skb = (struct sk_buff *)buf;
+	struct octeon_skb_page_info *pg_info;
+
+	pg_info = ((struct octeon_skb_page_info *)(skb->cb));
+	if (!pg_info->page) {
+		pr_err("%s: pg_info->page NULL\n", __func__);
+		WARN_ON(1);
+	}
+
+	/* Get DMA info */
+	dma_addr = pg_info->dma;
+	if (!pg_info->dma) {
+		pr_err("%s: ERROR it should be already available\n",
+		       __func__);
+		WARN_ON(1);
+	}
+	dma_addr += pg_info->page_offset;
+
+	return (u64)dma_addr;
+}
+
+static inline void
+lio_unmap_ring(struct pci_dev *pci_dev,
+	       u64 buf_ptr)
+
+{
+	dma_unmap_page(&pci_dev->dev,
+		       buf_ptr, (PAGE_SIZE << 0),
+		       DMA_FROM_DEVICE);
+}
+
+static inline void *octeon_fast_packet_alloc(u32 size)
+{
+	return recv_buffer_fast_alloc(size);
+}
+
+static inline void octeon_fast_packet_next(struct octeon_droq *droq,
+					   struct sk_buff *nicbuf,
+					   int copy_len,
+					   int idx)
+{
+	skb_put_data(nicbuf, get_rbd(droq->recv_buf_list[idx].buffer),
+		     copy_len);
+}
+
+/**
+ * \brief check interface state
+ * @param lio per-network private data
+ * @param state_flag flag state to check
+ */
+static inline int ifstate_check(struct lio *lio, int state_flag)
+{
+	return atomic_read(&lio->ifstate) & state_flag;
+}
+
+/**
+ * \brief set interface state
+ * @param lio per-network private data
+ * @param state_flag flag state to set
+ */
+static inline void ifstate_set(struct lio *lio, int state_flag)
+{
+	atomic_set(&lio->ifstate, (atomic_read(&lio->ifstate) | state_flag));
+}
+
+/**
+ * \brief clear interface state
+ * @param lio per-network private data
+ * @param state_flag flag state to clear
+ */
+static inline void ifstate_reset(struct lio *lio, int state_flag)
+{
+	atomic_set(&lio->ifstate, (atomic_read(&lio->ifstate) & ~(state_flag)));
+}
+
+/**
+ * \brief wait for all pending requests to complete
+ * @param oct Pointer to Octeon device
+ *
+ * Called during shutdown sequence
+ */
+static inline int wait_for_pending_requests(struct octeon_device *oct)
+{
+	int i, pcount = 0;
+
+	for (i = 0; i < MAX_IO_PENDING_PKT_COUNT; i++) {
+		pcount = atomic_read(
+		    &oct->response_list[OCTEON_ORDERED_SC_LIST]
+			 .pending_req_count);
+		if (pcount)
+			schedule_timeout_uninterruptible(HZ / 10);
+		else
+			break;
+	}
+
+	if (pcount)
+		return 1;
+
+	return 0;
+}
+
+/**
+ * \brief Stop Tx queues
+ * @param netdev network device
+ */
+static inline void stop_txqs(struct net_device *netdev)
+{
+	int i;
+
+	for (i = 0; i < netdev->real_num_tx_queues; i++)
+		netif_stop_subqueue(netdev, i);
+}
+
+/**
+ * \brief Wake Tx queues
+ * @param netdev network device
+ */
+static inline void wake_txqs(struct net_device *netdev)
+{
+	struct lio *lio = GET_LIO(netdev);
+	int i, qno;
+
+	for (i = 0; i < netdev->real_num_tx_queues; i++) {
+		qno = lio->linfo.txpciq[i % lio->oct_dev->num_iqs].s.q_no;
+
+		if (__netif_subqueue_stopped(netdev, i)) {
+			INCR_INSTRQUEUE_PKT_COUNT(lio->oct_dev, qno,
+						  tx_restart, 1);
+			netif_wake_subqueue(netdev, i);
+		}
+	}
+}
+
+/**
+ * \brief Start Tx queues
+ * @param netdev network device
+ */
+static inline void start_txqs(struct net_device *netdev)
+{
+	struct lio *lio = GET_LIO(netdev);
+	int i;
+
+	if (lio->linfo.link.s.link_up) {
+		for (i = 0; i < netdev->real_num_tx_queues; i++)
+			netif_start_subqueue(netdev, i);
+	}
+}
+
+static inline int skb_iq(struct octeon_device *oct, struct sk_buff *skb)
+{
+	return skb->queue_mapping % oct->num_iqs;
+}
+
+/**
+ * Remove the node at the head of the list. The list would be empty at
+ * the end of this call if there are no more nodes in the list.
+ */
+static inline struct list_head *lio_list_delete_head(struct list_head *root)
+{
+	struct list_head *node;
+
+	if (list_empty_careful(root))
+		node = NULL;
+	else
+		node = root->next;
+
+	if (node)
+		list_del(node);
+
+	return node;
+}
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_nic.c

@@ -0,0 +1,201 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_nic.h"
+#include "octeon_main.h"
+
+void *
+octeon_alloc_soft_command_resp(struct octeon_device    *oct,
+			       union octeon_instr_64B *cmd,
+			       u32		       rdatasize)
+{
+	struct octeon_soft_command *sc;
+	struct octeon_instr_ih3  *ih3;
+	struct octeon_instr_ih2  *ih2;
+	struct octeon_instr_irh *irh;
+	struct octeon_instr_rdp *rdp;
+
+	sc = (struct octeon_soft_command *)
+		octeon_alloc_soft_command(oct, 0, rdatasize, 0);
+
+	if (!sc)
+		return NULL;
+
+	/* Copy existing command structure into the soft command */
+	memcpy(&sc->cmd, cmd, sizeof(union octeon_instr_64B));
+
+	/* Add in the response related fields. Opcode and Param are already
+	 * there.
+	 */
+	if (OCTEON_CN23XX_PF(oct) || OCTEON_CN23XX_VF(oct)) {
+		ih3      = (struct octeon_instr_ih3 *)&sc->cmd.cmd3.ih3;
+		rdp     = (struct octeon_instr_rdp *)&sc->cmd.cmd3.rdp;
+		irh     = (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
+		/*pkiih3 + irh + ossp[0] + ossp[1] + rdp + rptr = 40 bytes */
+		ih3->fsz = LIO_SOFTCMDRESP_IH3;
+	} else {
+		ih2      = (struct octeon_instr_ih2 *)&sc->cmd.cmd2.ih2;
+		rdp     = (struct octeon_instr_rdp *)&sc->cmd.cmd2.rdp;
+		irh     = (struct octeon_instr_irh *)&sc->cmd.cmd2.irh;
+		/* irh + ossp[0] + ossp[1] + rdp + rptr = 40 bytes */
+		ih2->fsz = LIO_SOFTCMDRESP_IH2;
+	}
+
+	irh->rflag = 1; /* a response is required */
+
+	rdp->pcie_port = oct->pcie_port;
+	rdp->rlen      = rdatasize;
+
+	*sc->status_word = COMPLETION_WORD_INIT;
+
+	if (OCTEON_CN23XX_PF(oct) || OCTEON_CN23XX_VF(oct))
+		sc->cmd.cmd3.rptr =  sc->dmarptr;
+	else
+		sc->cmd.cmd2.rptr =  sc->dmarptr;
+
+	sc->expiry_time = jiffies + msecs_to_jiffies(LIO_SC_MAX_TMO_MS);
+
+	return sc;
+}
+EXPORT_SYMBOL_GPL(octeon_alloc_soft_command_resp);
+
+int octnet_send_nic_data_pkt(struct octeon_device *oct,
+			     struct octnic_data_pkt *ndata,
+			     int xmit_more)
+{
+	int ring_doorbell = !xmit_more;
+
+	return octeon_send_command(oct, ndata->q_no, ring_doorbell, &ndata->cmd,
+				   ndata->buf, ndata->datasize,
+				   ndata->reqtype);
+}
+EXPORT_SYMBOL_GPL(octnet_send_nic_data_pkt);
+
+static inline struct octeon_soft_command
+*octnic_alloc_ctrl_pkt_sc(struct octeon_device *oct,
+			  struct octnic_ctrl_pkt *nctrl)
+{
+	struct octeon_soft_command *sc = NULL;
+	u8 *data;
+	u32 rdatasize;
+	u32 uddsize = 0, datasize = 0;
+
+	uddsize = (u32)(nctrl->ncmd.s.more * 8);
+
+	datasize = OCTNET_CMD_SIZE + uddsize;
+	rdatasize = 16;
+
+	sc = (struct octeon_soft_command *)
+		octeon_alloc_soft_command(oct, datasize, rdatasize, 0);
+
+	if (!sc)
+		return NULL;
+
+	data = (u8 *)sc->virtdptr;
+
+	memcpy(data, &nctrl->ncmd, OCTNET_CMD_SIZE);
+
+	octeon_swap_8B_data((u64 *)data, (OCTNET_CMD_SIZE >> 3));
+
+	if (uddsize) {
+		/* Endian-Swap for UDD should have been done by caller. */
+		memcpy(data + OCTNET_CMD_SIZE, nctrl->udd, uddsize);
+	}
+
+	sc->iq_no = (u32)nctrl->iq_no;
+
+	octeon_prepare_soft_command(oct, sc, OPCODE_NIC, OPCODE_NIC_CMD,
+				    0, 0, 0);
+
+	init_completion(&sc->complete);
+	sc->sc_status = OCTEON_REQUEST_PENDING;
+
+	return sc;
+}
+
+int
+octnet_send_nic_ctrl_pkt(struct octeon_device *oct,
+			 struct octnic_ctrl_pkt *nctrl)
+{
+	int retval;
+	struct octeon_soft_command *sc = NULL;
+
+	spin_lock_bh(&oct->cmd_resp_wqlock);
+	/* Allow only rx ctrl command to stop traffic on the chip
+	 * during offline operations
+	 */
+	if ((oct->cmd_resp_state == OCT_DRV_OFFLINE) &&
+	    (nctrl->ncmd.s.cmd != OCTNET_CMD_RX_CTL)) {
+		spin_unlock_bh(&oct->cmd_resp_wqlock);
+		dev_err(&oct->pci_dev->dev,
+			"%s cmd:%d not processed since driver offline\n",
+			__func__, nctrl->ncmd.s.cmd);
+		return -1;
+	}
+
+	sc = octnic_alloc_ctrl_pkt_sc(oct, nctrl);
+	if (!sc) {
+		dev_err(&oct->pci_dev->dev, "%s soft command alloc failed\n",
+			__func__);
+		spin_unlock_bh(&oct->cmd_resp_wqlock);
+		return -1;
+	}
+
+	retval = octeon_send_soft_command(oct, sc);
+	if (retval == IQ_SEND_FAILED) {
+		octeon_free_soft_command(oct, sc);
+		dev_err(&oct->pci_dev->dev, "%s pf_num:%d soft command:%d send failed status: %x\n",
+			__func__, oct->pf_num, nctrl->ncmd.s.cmd, retval);
+		spin_unlock_bh(&oct->cmd_resp_wqlock);
+		return -1;
+	}
+
+	spin_unlock_bh(&oct->cmd_resp_wqlock);
+
+	if (nctrl->ncmd.s.cmdgroup == 0) {
+		switch (nctrl->ncmd.s.cmd) {
+			/* caller holds lock, can not sleep */
+		case OCTNET_CMD_CHANGE_DEVFLAGS:
+		case OCTNET_CMD_SET_MULTI_LIST:
+		case OCTNET_CMD_SET_UC_LIST:
+			WRITE_ONCE(sc->caller_is_done, true);
+			return retval;
+		}
+	}
+
+	retval = wait_for_sc_completion_timeout(oct, sc, 0);
+	if (retval)
+		return (retval);
+
+	nctrl->sc_status = sc->sc_status;
+	retval = nctrl->sc_status;
+	if (nctrl->cb_fn)
+		nctrl->cb_fn(nctrl);
+
+	WRITE_ONCE(sc->caller_is_done, true);
+
+	return retval;
+}
+EXPORT_SYMBOL_GPL(octnet_send_nic_ctrl_pkt);

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/octeon_nic.h

@@ -0,0 +1,288 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+
+/*!  \file octeon_nic.h
+ *   \brief Host NIC Driver: Routine to send network data &
+ *   control packet to Octeon.
+ */
+
+#ifndef __OCTEON_NIC_H__
+#define  __OCTEON_NIC_H__
+
+/* Maximum number of 8-byte words can be sent in a NIC control message.
+ */
+#define  MAX_NCTRL_UDD  32
+
+typedef void (*octnic_ctrl_pkt_cb_fn_t) (void *);
+
+/* Structure of control information passed by the NIC module to the OSI
+ * layer when sending control commands to Octeon device software.
+ */
+struct octnic_ctrl_pkt {
+	/** Command to be passed to the Octeon device software. */
+	union octnet_cmd ncmd;
+
+	/** Send buffer  */
+	void *data;
+	u64 dmadata;
+
+	/** Response buffer */
+	void *rdata;
+	u64 dmardata;
+
+	/** Additional data that may be needed by some commands. */
+	u64 udd[MAX_NCTRL_UDD];
+
+	/** Input queue to use to send this command. */
+	u64 iq_no;
+
+	/** The network device that issued the control command. */
+	u64 netpndev;
+
+	/** Callback function called when the command has been fetched */
+	octnic_ctrl_pkt_cb_fn_t cb_fn;
+
+	u32 sc_status;
+};
+
+#define MAX_UDD_SIZE(nctrl) (sizeof((nctrl)->udd))
+
+/** Structure of data information passed by the NIC module to the OSI
+ * layer when forwarding data to Octeon device software.
+ */
+struct octnic_data_pkt {
+	/** Pointer to information maintained by NIC module for this packet. The
+	 *  OSI layer passes this as-is to the driver.
+	 */
+	void *buf;
+
+	/** Type of buffer passed in "buf" above. */
+	u32 reqtype;
+
+	/** Total data bytes to be transferred in this command. */
+	u32 datasize;
+
+	/** Command to be passed to the Octeon device software. */
+	union octeon_instr_64B cmd;
+
+	/** Input queue to use to send this command. */
+	u32 q_no;
+
+};
+
+/** Structure passed by NIC module to OSI layer to prepare a command to send
+ * network data to Octeon.
+ */
+union octnic_cmd_setup {
+	struct {
+		u32 iq_no:8;
+		u32 gather:1;
+		u32 timestamp:1;
+		u32 ip_csum:1;
+		u32 transport_csum:1;
+		u32 tnl_csum:1;
+		u32 rsvd:19;
+
+		union {
+			u32 datasize;
+			u32 gatherptrs;
+		} u;
+	} s;
+
+	u64 u64;
+
+};
+
+static inline int octnet_iq_is_full(struct octeon_device *oct, u32 q_no)
+{
+	return ((u32)atomic_read(&oct->instr_queue[q_no]->instr_pending)
+		>= (oct->instr_queue[q_no]->max_count - 2));
+}
+
+static inline void
+octnet_prepare_pci_cmd_o2(struct octeon_device *oct,
+			  union octeon_instr_64B *cmd,
+			  union octnic_cmd_setup *setup, u32 tag)
+{
+	struct octeon_instr_ih2 *ih2;
+	struct octeon_instr_irh *irh;
+	union octnic_packet_params packet_params;
+	int port;
+
+	memset(cmd, 0, sizeof(union octeon_instr_64B));
+
+	ih2 = (struct octeon_instr_ih2 *)&cmd->cmd2.ih2;
+
+	/* assume that rflag is cleared so therefore front data will only have
+	 * irh and ossp[0], ossp[1] for a total of 32 bytes
+	 */
+	ih2->fsz = LIO_PCICMD_O2;
+
+	ih2->tagtype = ORDERED_TAG;
+	ih2->grp = DEFAULT_POW_GRP;
+
+	port = (int)oct->instr_queue[setup->s.iq_no]->txpciq.s.port;
+
+	if (tag)
+		ih2->tag = tag;
+	else
+		ih2->tag = LIO_DATA(port);
+
+	ih2->raw = 1;
+	ih2->qos = (port & 3) + 4;	/* map qos based on interface */
+
+	if (!setup->s.gather) {
+		ih2->dlengsz = setup->s.u.datasize;
+	} else {
+		ih2->gather = 1;
+		ih2->dlengsz = setup->s.u.gatherptrs;
+	}
+
+	irh = (struct octeon_instr_irh *)&cmd->cmd2.irh;
+
+	irh->opcode = OPCODE_NIC;
+	irh->subcode = OPCODE_NIC_NW_DATA;
+
+	packet_params.u32 = 0;
+
+	packet_params.s.ip_csum = setup->s.ip_csum;
+	packet_params.s.transport_csum = setup->s.transport_csum;
+	packet_params.s.tnl_csum = setup->s.tnl_csum;
+	packet_params.s.tsflag = setup->s.timestamp;
+
+	irh->ossp = packet_params.u32;
+}
+
+static inline void
+octnet_prepare_pci_cmd_o3(struct octeon_device *oct,
+			  union octeon_instr_64B *cmd,
+			  union octnic_cmd_setup *setup, u32 tag)
+{
+	struct octeon_instr_irh *irh;
+	struct octeon_instr_ih3     *ih3;
+	struct octeon_instr_pki_ih3 *pki_ih3;
+	union octnic_packet_params packet_params;
+	int port;
+
+	memset(cmd, 0, sizeof(union octeon_instr_64B));
+
+	ih3 = (struct octeon_instr_ih3 *)&cmd->cmd3.ih3;
+	pki_ih3 = (struct octeon_instr_pki_ih3 *)&cmd->cmd3.pki_ih3;
+
+	/* assume that rflag is cleared so therefore front data will only have
+	 * irh and ossp[1] and ossp[2] for a total of 24 bytes
+	 */
+	ih3->pkind       = oct->instr_queue[setup->s.iq_no]->txpciq.s.pkind;
+	/*PKI IH*/
+	ih3->fsz = LIO_PCICMD_O3;
+
+	if (!setup->s.gather) {
+		ih3->dlengsz = setup->s.u.datasize;
+	} else {
+		ih3->gather = 1;
+		ih3->dlengsz = setup->s.u.gatherptrs;
+	}
+
+	pki_ih3->w       = 1;
+	pki_ih3->raw     = 1;
+	pki_ih3->utag    = 1;
+	pki_ih3->utt     = 1;
+	pki_ih3->uqpg    = oct->instr_queue[setup->s.iq_no]->txpciq.s.use_qpg;
+
+	port = (int)oct->instr_queue[setup->s.iq_no]->txpciq.s.port;
+
+	if (tag)
+		pki_ih3->tag = tag;
+	else
+		pki_ih3->tag     = LIO_DATA(port);
+
+	pki_ih3->tagtype = ORDERED_TAG;
+	pki_ih3->qpg     = oct->instr_queue[setup->s.iq_no]->txpciq.s.qpg;
+	pki_ih3->pm      = 0x7; /*0x7 - meant for Parse nothing, uninterpreted*/
+	pki_ih3->sl      = 8;   /* sl will be sizeof(pki_ih3)*/
+
+	irh = (struct octeon_instr_irh *)&cmd->cmd3.irh;
+
+	irh->opcode = OPCODE_NIC;
+	irh->subcode = OPCODE_NIC_NW_DATA;
+
+	packet_params.u32 = 0;
+
+	packet_params.s.ip_csum = setup->s.ip_csum;
+	packet_params.s.transport_csum = setup->s.transport_csum;
+	packet_params.s.tnl_csum = setup->s.tnl_csum;
+	packet_params.s.tsflag = setup->s.timestamp;
+
+	irh->ossp = packet_params.u32;
+}
+
+/** Utility function to prepare a 64B NIC instruction based on a setup command
+ * @param cmd - pointer to instruction to be filled in.
+ * @param setup - pointer to the setup structure
+ * @param q_no - which queue for back pressure
+ *
+ * Assumes the cmd instruction is pre-allocated, but no fields are filled in.
+ */
+static inline void
+octnet_prepare_pci_cmd(struct octeon_device *oct, union octeon_instr_64B *cmd,
+		       union octnic_cmd_setup *setup, u32 tag)
+{
+	if (OCTEON_CN6XXX(oct))
+		octnet_prepare_pci_cmd_o2(oct, cmd, setup, tag);
+	else
+		octnet_prepare_pci_cmd_o3(oct, cmd, setup, tag);
+}
+
+/** Allocate and a soft command with space for a response immediately following
+ * the commnad.
+ * @param oct - octeon device pointer
+ * @param cmd - pointer to the command structure, pre-filled for everything
+ * except the response.
+ * @param rdatasize - size in bytes of the response.
+ *
+ * @returns pointer to allocated buffer with command copied into it, and
+ * response space immediately following.
+ */
+void *
+octeon_alloc_soft_command_resp(struct octeon_device    *oct,
+			       union octeon_instr_64B *cmd,
+			       u32		       rdatasize);
+
+/** Send a NIC data packet to the device
+ * @param oct - octeon device pointer
+ * @param ndata - control structure with queueing, and buffer information
+ *
+ * @returns IQ_FAILED if it failed to add to the input queue. IQ_STOP if the
+ * queue should be stopped, and IQ_SEND_OK if it sent okay.
+ */
+int octnet_send_nic_data_pkt(struct octeon_device *oct,
+			     struct octnic_data_pkt *ndata,
+			     int xmit_more);
+
+/** Send a NIC control packet to the device
+ * @param oct - octeon device pointer
+ * @param nctrl - control structure with command, timout, and callback info
+ * @returns IQ_FAILED if it failed to add to the input queue. IQ_STOP if the
+ * queue should be stopped, and IQ_SEND_OK if it sent okay.
+ */
+int
+octnet_send_nic_ctrl_pkt(struct octeon_device *oct,
+			 struct octnic_ctrl_pkt *nctrl);
+
+#endif

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/request_manager.c

@@ -0,0 +1,941 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/vmalloc.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+#include "octeon_network.h"
+#include "cn66xx_device.h"
+#include "cn23xx_pf_device.h"
+#include "cn23xx_vf_device.h"
+
+struct iq_post_status {
+	int status;
+	int index;
+};
+
+static void check_db_timeout(struct work_struct *work);
+static void  __check_db_timeout(struct octeon_device *oct, u64 iq_no);
+
+static void (*reqtype_free_fn[MAX_OCTEON_DEVICES][REQTYPE_LAST + 1]) (void *);
+
+/* Define this to return the request status comaptible to old code */
+/*#define OCTEON_USE_OLD_REQ_STATUS*/
+
+/* Return 0 on success, 1 on failure */
+int octeon_init_instr_queue(struct octeon_device *oct,
+			    union oct_txpciq txpciq,
+			    u32 num_descs)
+{
+	struct octeon_instr_queue *iq;
+	struct octeon_iq_config *conf = NULL;
+	u32 iq_no = (u32)txpciq.s.q_no;
+	u32 q_size;
+	struct cavium_wq *db_wq;
+	int numa_node = dev_to_node(&oct->pci_dev->dev);
+
+	if (OCTEON_CN6XXX(oct))
+		conf = &(CFG_GET_IQ_CFG(CHIP_CONF(oct, cn6xxx)));
+	else if (OCTEON_CN23XX_PF(oct))
+		conf = &(CFG_GET_IQ_CFG(CHIP_CONF(oct, cn23xx_pf)));
+	else if (OCTEON_CN23XX_VF(oct))
+		conf = &(CFG_GET_IQ_CFG(CHIP_CONF(oct, cn23xx_vf)));
+
+	if (!conf) {
+		dev_err(&oct->pci_dev->dev, "Unsupported Chip %x\n",
+			oct->chip_id);
+		return 1;
+	}
+
+	q_size = (u32)conf->instr_type * num_descs;
+
+	iq = oct->instr_queue[iq_no];
+
+	iq->oct_dev = oct;
+
+	iq->base_addr = lio_dma_alloc(oct, q_size, &iq->base_addr_dma);
+	if (!iq->base_addr) {
+		dev_err(&oct->pci_dev->dev, "Cannot allocate memory for instr queue %d\n",
+			iq_no);
+		return 1;
+	}
+
+	iq->max_count = num_descs;
+
+	/* Initialize a list to holds requests that have been posted to Octeon
+	 * but has yet to be fetched by octeon
+	 */
+	iq->request_list = vzalloc_node(array_size(num_descs, sizeof(*iq->request_list)),
+					numa_node);
+	if (!iq->request_list)
+		iq->request_list = vzalloc(array_size(num_descs, sizeof(*iq->request_list)));
+	if (!iq->request_list) {
+		lio_dma_free(oct, q_size, iq->base_addr, iq->base_addr_dma);
+		dev_err(&oct->pci_dev->dev, "Alloc failed for IQ[%d] nr free list\n",
+			iq_no);
+		return 1;
+	}
+
+	dev_dbg(&oct->pci_dev->dev, "IQ[%d]: base: %p basedma: %pad count: %d\n",
+		iq_no, iq->base_addr, &iq->base_addr_dma, iq->max_count);
+
+	iq->txpciq.u64 = txpciq.u64;
+	iq->fill_threshold = (u32)conf->db_min;
+	iq->fill_cnt = 0;
+	iq->host_write_index = 0;
+	iq->octeon_read_index = 0;
+	iq->flush_index = 0;
+	iq->last_db_time = 0;
+	iq->do_auto_flush = 1;
+	iq->db_timeout = (u32)conf->db_timeout;
+	atomic_set(&iq->instr_pending, 0);
+	iq->pkts_processed = 0;
+
+	/* Initialize the spinlock for this instruction queue */
+	spin_lock_init(&iq->lock);
+	if (iq_no == 0) {
+		iq->allow_soft_cmds = true;
+		spin_lock_init(&iq->post_lock);
+	} else {
+		iq->allow_soft_cmds = false;
+	}
+
+	spin_lock_init(&iq->iq_flush_running_lock);
+
+	oct->io_qmask.iq |= BIT_ULL(iq_no);
+
+	/* Set the 32B/64B mode for each input queue */
+	oct->io_qmask.iq64B |= ((u64)(conf->instr_type == 64) << iq_no);
+	iq->iqcmd_64B = (conf->instr_type == 64);
+
+	oct->fn_list.setup_iq_regs(oct, iq_no);
+
+	oct->check_db_wq[iq_no].wq = alloc_workqueue("check_iq_db",
+						     WQ_MEM_RECLAIM | WQ_PERCPU,
+						     0);
+	if (!oct->check_db_wq[iq_no].wq) {
+		vfree(iq->request_list);
+		iq->request_list = NULL;
+		lio_dma_free(oct, q_size, iq->base_addr, iq->base_addr_dma);
+		dev_err(&oct->pci_dev->dev, "check db wq create failed for iq %d\n",
+			iq_no);
+		return 1;
+	}
+
+	db_wq = &oct->check_db_wq[iq_no];
+
+	INIT_DELAYED_WORK(&db_wq->wk.work, check_db_timeout);
+	db_wq->wk.ctxptr = oct;
+	db_wq->wk.ctxul = iq_no;
+	queue_delayed_work(db_wq->wq, &db_wq->wk.work, msecs_to_jiffies(1));
+
+	return 0;
+}
+
+int octeon_delete_instr_queue(struct octeon_device *oct, u32 iq_no)
+{
+	u64 desc_size = 0, q_size;
+	struct octeon_instr_queue *iq = oct->instr_queue[iq_no];
+
+	cancel_delayed_work_sync(&oct->check_db_wq[iq_no].wk.work);
+	destroy_workqueue(oct->check_db_wq[iq_no].wq);
+
+	if (OCTEON_CN6XXX(oct))
+		desc_size =
+		    CFG_GET_IQ_INSTR_TYPE(CHIP_CONF(oct, cn6xxx));
+	else if (OCTEON_CN23XX_PF(oct))
+		desc_size =
+		    CFG_GET_IQ_INSTR_TYPE(CHIP_CONF(oct, cn23xx_pf));
+	else if (OCTEON_CN23XX_VF(oct))
+		desc_size =
+		    CFG_GET_IQ_INSTR_TYPE(CHIP_CONF(oct, cn23xx_vf));
+
+	vfree(iq->request_list);
+
+	if (iq->base_addr) {
+		q_size = iq->max_count * desc_size;
+		lio_dma_free(oct, (u32)q_size, iq->base_addr,
+			     iq->base_addr_dma);
+		oct->io_qmask.iq &= ~(1ULL << iq_no);
+		vfree(oct->instr_queue[iq_no]);
+		oct->instr_queue[iq_no] = NULL;
+		oct->num_iqs--;
+		return 0;
+	}
+	return 1;
+}
+EXPORT_SYMBOL_GPL(octeon_delete_instr_queue);
+
+/* Return 0 on success, 1 on failure */
+int octeon_setup_iq(struct octeon_device *oct,
+		    int ifidx,
+		    int q_index,
+		    union oct_txpciq txpciq,
+		    u32 num_descs,
+		    void *app_ctx)
+{
+	u32 iq_no = (u32)txpciq.s.q_no;
+	int numa_node = dev_to_node(&oct->pci_dev->dev);
+
+	if (oct->instr_queue[iq_no]) {
+		dev_dbg(&oct->pci_dev->dev, "IQ is in use. Cannot create the IQ: %d again\n",
+			iq_no);
+		oct->instr_queue[iq_no]->txpciq.u64 = txpciq.u64;
+		oct->instr_queue[iq_no]->app_ctx = app_ctx;
+		return 0;
+	}
+	oct->instr_queue[iq_no] =
+	    vzalloc_node(sizeof(struct octeon_instr_queue), numa_node);
+	if (!oct->instr_queue[iq_no])
+		oct->instr_queue[iq_no] =
+		    vzalloc(sizeof(struct octeon_instr_queue));
+	if (!oct->instr_queue[iq_no])
+		return 1;
+
+
+	oct->instr_queue[iq_no]->q_index = q_index;
+	oct->instr_queue[iq_no]->app_ctx = app_ctx;
+	oct->instr_queue[iq_no]->ifidx = ifidx;
+
+	if (octeon_init_instr_queue(oct, txpciq, num_descs)) {
+		vfree(oct->instr_queue[iq_no]);
+		oct->instr_queue[iq_no] = NULL;
+		return 1;
+	}
+
+	oct->num_iqs++;
+	if (oct->fn_list.enable_io_queues(oct)) {
+		octeon_delete_instr_queue(oct, iq_no);
+		return 1;
+	}
+
+	return 0;
+}
+
+int lio_wait_for_instr_fetch(struct octeon_device *oct)
+{
+	int i, retry = 1000, pending, instr_cnt = 0;
+
+	do {
+		instr_cnt = 0;
+
+		for (i = 0; i < MAX_OCTEON_INSTR_QUEUES(oct); i++) {
+			if (!(oct->io_qmask.iq & BIT_ULL(i)))
+				continue;
+			pending =
+			    atomic_read(&oct->instr_queue[i]->instr_pending);
+			if (pending)
+				__check_db_timeout(oct, i);
+			instr_cnt += pending;
+		}
+
+		if (instr_cnt == 0)
+			break;
+
+		schedule_timeout_uninterruptible(1);
+
+	} while (retry-- && instr_cnt);
+
+	return instr_cnt;
+}
+EXPORT_SYMBOL_GPL(lio_wait_for_instr_fetch);
+
+static inline void
+ring_doorbell(struct octeon_device *oct, struct octeon_instr_queue *iq)
+{
+	if (atomic_read(&oct->status) == OCT_DEV_RUNNING) {
+		writel(iq->fill_cnt, iq->doorbell_reg);
+		/* make sure doorbell write goes through */
+		iq->fill_cnt = 0;
+		iq->last_db_time = jiffies;
+		return;
+	}
+}
+
+void
+octeon_ring_doorbell_locked(struct octeon_device *oct, u32 iq_no)
+{
+	struct octeon_instr_queue *iq;
+
+	iq = oct->instr_queue[iq_no];
+	spin_lock(&iq->post_lock);
+	if (iq->fill_cnt)
+		ring_doorbell(oct, iq);
+	spin_unlock(&iq->post_lock);
+}
+EXPORT_SYMBOL_GPL(octeon_ring_doorbell_locked);
+
+static inline void __copy_cmd_into_iq(struct octeon_instr_queue *iq,
+				      u8 *cmd)
+{
+	u8 *iqptr, cmdsize;
+
+	cmdsize = ((iq->iqcmd_64B) ? 64 : 32);
+	iqptr = iq->base_addr + (cmdsize * iq->host_write_index);
+
+	memcpy(iqptr, cmd, cmdsize);
+}
+
+static inline struct iq_post_status
+__post_command2(struct octeon_instr_queue *iq, u8 *cmd)
+{
+	struct iq_post_status st;
+
+	st.status = IQ_SEND_OK;
+
+	/* This ensures that the read index does not wrap around to the same
+	 * position if queue gets full before Octeon could fetch any instr.
+	 */
+	if (atomic_read(&iq->instr_pending) >= (s32)(iq->max_count - 1)) {
+		st.status = IQ_SEND_FAILED;
+		st.index = -1;
+		return st;
+	}
+
+	if (atomic_read(&iq->instr_pending) >= (s32)(iq->max_count - 2))
+		st.status = IQ_SEND_STOP;
+
+	__copy_cmd_into_iq(iq, cmd);
+
+	/* "index" is returned, host_write_index is modified. */
+	st.index = iq->host_write_index;
+	iq->host_write_index = incr_index(iq->host_write_index, 1,
+					  iq->max_count);
+	iq->fill_cnt++;
+
+	/* Flush the command into memory. We need to be sure the data is in
+	 * memory before indicating that the instruction is pending.
+	 */
+	wmb();
+
+	atomic_inc(&iq->instr_pending);
+
+	return st;
+}
+
+int
+octeon_register_reqtype_free_fn(struct octeon_device *oct, int reqtype,
+				void (*fn)(void *))
+{
+	if (reqtype > REQTYPE_LAST) {
+		dev_err(&oct->pci_dev->dev, "%s: Invalid reqtype: %d\n",
+			__func__, reqtype);
+		return -EINVAL;
+	}
+
+	reqtype_free_fn[oct->octeon_id][reqtype] = fn;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_register_reqtype_free_fn);
+
+static inline void
+__add_to_request_list(struct octeon_instr_queue *iq,
+		      int idx, void *buf, int reqtype)
+{
+	iq->request_list[idx].buf = buf;
+	iq->request_list[idx].reqtype = reqtype;
+}
+
+/* Can only run in process context */
+int
+lio_process_iq_request_list(struct octeon_device *oct,
+			    struct octeon_instr_queue *iq, u32 napi_budget)
+{
+	struct cavium_wq *cwq = &oct->dma_comp_wq;
+	int reqtype;
+	void *buf;
+	u32 old = iq->flush_index;
+	u32 inst_count = 0;
+	unsigned int pkts_compl = 0, bytes_compl = 0;
+	struct octeon_soft_command *sc;
+	unsigned long flags;
+
+	while (old != iq->octeon_read_index) {
+		reqtype = iq->request_list[old].reqtype;
+		buf     = iq->request_list[old].buf;
+
+		if (reqtype == REQTYPE_NONE)
+			goto skip_this;
+
+		octeon_update_tx_completion_counters(buf, reqtype, &pkts_compl,
+						     &bytes_compl);
+
+		switch (reqtype) {
+		case REQTYPE_NORESP_NET:
+		case REQTYPE_NORESP_NET_SG:
+		case REQTYPE_RESP_NET_SG:
+			reqtype_free_fn[oct->octeon_id][reqtype](buf);
+			break;
+		case REQTYPE_RESP_NET:
+		case REQTYPE_SOFT_COMMAND:
+			sc = buf;
+			/* We're expecting a response from Octeon.
+			 * It's up to lio_process_ordered_list() to
+			 * process  sc. Add sc to the ordered soft
+			 * command response list because we expect
+			 * a response from Octeon.
+			 */
+			spin_lock_irqsave(&oct->response_list
+					  [OCTEON_ORDERED_SC_LIST].lock, flags);
+			atomic_inc(&oct->response_list
+				   [OCTEON_ORDERED_SC_LIST].pending_req_count);
+			list_add_tail(&sc->node, &oct->response_list
+				[OCTEON_ORDERED_SC_LIST].head);
+			spin_unlock_irqrestore(&oct->response_list
+					       [OCTEON_ORDERED_SC_LIST].lock,
+					       flags);
+			break;
+		default:
+			dev_err(&oct->pci_dev->dev,
+				"%s Unknown reqtype: %d buf: %p at idx %d\n",
+				__func__, reqtype, buf, old);
+		}
+
+		iq->request_list[old].buf = NULL;
+		iq->request_list[old].reqtype = 0;
+
+ skip_this:
+		inst_count++;
+		old = incr_index(old, 1, iq->max_count);
+
+		if ((napi_budget) && (inst_count >= napi_budget))
+			break;
+	}
+	if (bytes_compl)
+		octeon_report_tx_completion_to_bql(iq->app_ctx, pkts_compl,
+						   bytes_compl);
+	iq->flush_index = old;
+
+	if (atomic_read(&oct->response_list
+			[OCTEON_ORDERED_SC_LIST].pending_req_count))
+		queue_work(cwq->wq, &cwq->wk.work.work);
+
+	return inst_count;
+}
+EXPORT_SYMBOL_GPL(lio_process_iq_request_list);
+
+/* Can only be called from process context */
+int
+octeon_flush_iq(struct octeon_device *oct, struct octeon_instr_queue *iq,
+		u32 napi_budget)
+{
+	u32 inst_processed = 0;
+	u32 tot_inst_processed = 0;
+	int tx_done = 1;
+
+	if (!spin_trylock(&iq->iq_flush_running_lock))
+		return tx_done;
+
+	spin_lock_bh(&iq->lock);
+
+	iq->octeon_read_index = oct->fn_list.update_iq_read_idx(iq);
+
+	do {
+		/* Process any outstanding IQ packets. */
+		if (iq->flush_index == iq->octeon_read_index)
+			break;
+
+		if (napi_budget)
+			inst_processed =
+				lio_process_iq_request_list(oct, iq,
+							    napi_budget -
+							    tot_inst_processed);
+		else
+			inst_processed =
+				lio_process_iq_request_list(oct, iq, 0);
+
+		if (inst_processed) {
+			iq->pkts_processed += inst_processed;
+			atomic_sub(inst_processed, &iq->instr_pending);
+			iq->stats.instr_processed += inst_processed;
+		}
+
+		tot_inst_processed += inst_processed;
+	} while (tot_inst_processed < napi_budget);
+
+	if (napi_budget && (tot_inst_processed >= napi_budget))
+		tx_done = 0;
+
+	iq->last_db_time = jiffies;
+
+	spin_unlock_bh(&iq->lock);
+
+	spin_unlock(&iq->iq_flush_running_lock);
+
+	return tx_done;
+}
+
+/* Process instruction queue after timeout.
+ * This routine gets called from a workqueue or when removing the module.
+ */
+static void __check_db_timeout(struct octeon_device *oct, u64 iq_no)
+{
+	struct octeon_instr_queue *iq;
+	u64 next_time;
+
+	if (!oct)
+		return;
+
+	iq = oct->instr_queue[iq_no];
+	if (!iq)
+		return;
+
+	/* return immediately, if no work pending */
+	if (!atomic_read(&iq->instr_pending))
+		return;
+	/* If jiffies - last_db_time < db_timeout do nothing  */
+	next_time = iq->last_db_time + iq->db_timeout;
+	if (!time_after(jiffies, (unsigned long)next_time))
+		return;
+	iq->last_db_time = jiffies;
+
+	/* Flush the instruction queue */
+	octeon_flush_iq(oct, iq, 0);
+
+	lio_enable_irq(NULL, iq);
+}
+
+/* Called by the Poll thread at regular intervals to check the instruction
+ * queue for commands to be posted and for commands that were fetched by Octeon.
+ */
+static void check_db_timeout(struct work_struct *work)
+{
+	struct cavium_wk *wk = (struct cavium_wk *)work;
+	struct octeon_device *oct = (struct octeon_device *)wk->ctxptr;
+	u64 iq_no = wk->ctxul;
+	struct cavium_wq *db_wq = &oct->check_db_wq[iq_no];
+	u32 delay = 10;
+
+	__check_db_timeout(oct, iq_no);
+	queue_delayed_work(db_wq->wq, &db_wq->wk.work, msecs_to_jiffies(delay));
+}
+
+int
+octeon_send_command(struct octeon_device *oct, u32 iq_no,
+		    u32 force_db, void *cmd, void *buf,
+		    u32 datasize, u32 reqtype)
+{
+	int xmit_stopped;
+	struct iq_post_status st;
+	struct octeon_instr_queue *iq = oct->instr_queue[iq_no];
+
+	/* Get the lock and prevent other tasks and tx interrupt handler from
+	 * running.
+	 */
+	if (iq->allow_soft_cmds)
+		spin_lock_bh(&iq->post_lock);
+
+	st = __post_command2(iq, cmd);
+
+	if (st.status != IQ_SEND_FAILED) {
+		xmit_stopped = octeon_report_sent_bytes_to_bql(buf, reqtype);
+		__add_to_request_list(iq, st.index, buf, reqtype);
+		INCR_INSTRQUEUE_PKT_COUNT(oct, iq_no, bytes_sent, datasize);
+		INCR_INSTRQUEUE_PKT_COUNT(oct, iq_no, instr_posted, 1);
+
+		if (iq->fill_cnt >= MAX_OCTEON_FILL_COUNT || force_db ||
+		    xmit_stopped || st.status == IQ_SEND_STOP)
+			ring_doorbell(oct, iq);
+	} else {
+		INCR_INSTRQUEUE_PKT_COUNT(oct, iq_no, instr_dropped, 1);
+	}
+
+	if (iq->allow_soft_cmds)
+		spin_unlock_bh(&iq->post_lock);
+
+	/* This is only done here to expedite packets being flushed
+	 * for cases where there are no IQ completion interrupts.
+	 */
+
+	return st.status;
+}
+EXPORT_SYMBOL_GPL(octeon_send_command);
+
+void
+octeon_prepare_soft_command(struct octeon_device *oct,
+			    struct octeon_soft_command *sc,
+			    u8 opcode,
+			    u8 subcode,
+			    u32 irh_ossp,
+			    u64 ossp0,
+			    u64 ossp1)
+{
+	struct octeon_config *oct_cfg;
+	struct octeon_instr_ih2 *ih2;
+	struct octeon_instr_ih3 *ih3;
+	struct octeon_instr_pki_ih3 *pki_ih3;
+	struct octeon_instr_irh *irh;
+	struct octeon_instr_rdp *rdp;
+
+	WARN_ON(opcode > 15);
+	WARN_ON(subcode > 127);
+
+	oct_cfg = octeon_get_conf(oct);
+
+	if (OCTEON_CN23XX_PF(oct) || OCTEON_CN23XX_VF(oct)) {
+		ih3 = (struct octeon_instr_ih3 *)&sc->cmd.cmd3.ih3;
+
+		ih3->pkind = oct->instr_queue[sc->iq_no]->txpciq.s.pkind;
+
+		pki_ih3 = (struct octeon_instr_pki_ih3 *)&sc->cmd.cmd3.pki_ih3;
+
+		pki_ih3->w           = 1;
+		pki_ih3->raw         = 1;
+		pki_ih3->utag        = 1;
+		pki_ih3->uqpg        =
+			oct->instr_queue[sc->iq_no]->txpciq.s.use_qpg;
+		pki_ih3->utt         = 1;
+		pki_ih3->tag     = LIO_CONTROL;
+		pki_ih3->tagtype = ATOMIC_TAG;
+		pki_ih3->qpg         =
+			oct->instr_queue[sc->iq_no]->txpciq.s.ctrl_qpg;
+
+		pki_ih3->pm          = 0x7;
+		pki_ih3->sl          = 8;
+
+		if (sc->datasize)
+			ih3->dlengsz = sc->datasize;
+
+		irh            = (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
+		irh->opcode    = opcode;
+		irh->subcode   = subcode;
+
+		/* opcode/subcode specific parameters (ossp) */
+		irh->ossp       = irh_ossp;
+		sc->cmd.cmd3.ossp[0] = ossp0;
+		sc->cmd.cmd3.ossp[1] = ossp1;
+
+		if (sc->rdatasize) {
+			rdp = (struct octeon_instr_rdp *)&sc->cmd.cmd3.rdp;
+			rdp->pcie_port = oct->pcie_port;
+			rdp->rlen      = sc->rdatasize;
+
+			irh->rflag =  1;
+			/*PKI IH3*/
+			/* pki_ih3 irh+ossp[0]+ossp[1]+rdp+rptr = 48 bytes */
+			ih3->fsz    = LIO_SOFTCMDRESP_IH3;
+		} else {
+			irh->rflag =  0;
+			/*PKI IH3*/
+			/* pki_h3 + irh + ossp[0] + ossp[1] = 32 bytes */
+			ih3->fsz    = LIO_PCICMD_O3;
+		}
+
+	} else {
+		ih2          = (struct octeon_instr_ih2 *)&sc->cmd.cmd2.ih2;
+		ih2->tagtype = ATOMIC_TAG;
+		ih2->tag     = LIO_CONTROL;
+		ih2->raw     = 1;
+		ih2->grp     = CFG_GET_CTRL_Q_GRP(oct_cfg);
+
+		if (sc->datasize) {
+			ih2->dlengsz = sc->datasize;
+			ih2->rs = 1;
+		}
+
+		irh            = (struct octeon_instr_irh *)&sc->cmd.cmd2.irh;
+		irh->opcode    = opcode;
+		irh->subcode   = subcode;
+
+		/* opcode/subcode specific parameters (ossp) */
+		irh->ossp       = irh_ossp;
+		sc->cmd.cmd2.ossp[0] = ossp0;
+		sc->cmd.cmd2.ossp[1] = ossp1;
+
+		if (sc->rdatasize) {
+			rdp = (struct octeon_instr_rdp *)&sc->cmd.cmd2.rdp;
+			rdp->pcie_port = oct->pcie_port;
+			rdp->rlen      = sc->rdatasize;
+
+			irh->rflag =  1;
+			/* irh+ossp[0]+ossp[1]+rdp+rptr = 40 bytes */
+			ih2->fsz   = LIO_SOFTCMDRESP_IH2;
+		} else {
+			irh->rflag =  0;
+			/* irh + ossp[0] + ossp[1] = 24 bytes */
+			ih2->fsz   = LIO_PCICMD_O2;
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(octeon_prepare_soft_command);
+
+int octeon_send_soft_command(struct octeon_device *oct,
+			     struct octeon_soft_command *sc)
+{
+	struct octeon_instr_queue *iq;
+	struct octeon_instr_ih2 *ih2;
+	struct octeon_instr_ih3 *ih3;
+	struct octeon_instr_irh *irh;
+	u32 len;
+
+	iq = oct->instr_queue[sc->iq_no];
+	if (!iq->allow_soft_cmds) {
+		dev_err(&oct->pci_dev->dev, "Soft commands are not allowed on Queue %d\n",
+			sc->iq_no);
+		INCR_INSTRQUEUE_PKT_COUNT(oct, sc->iq_no, instr_dropped, 1);
+		return IQ_SEND_FAILED;
+	}
+
+	if (OCTEON_CN23XX_PF(oct) || OCTEON_CN23XX_VF(oct)) {
+		ih3 =  (struct octeon_instr_ih3 *)&sc->cmd.cmd3.ih3;
+		if (ih3->dlengsz) {
+			WARN_ON(!sc->dmadptr);
+			sc->cmd.cmd3.dptr = sc->dmadptr;
+		}
+		irh = (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
+		if (irh->rflag) {
+			WARN_ON(!sc->dmarptr);
+			WARN_ON(!sc->status_word);
+			*sc->status_word = COMPLETION_WORD_INIT;
+			sc->cmd.cmd3.rptr = sc->dmarptr;
+		}
+		len = (u32)ih3->dlengsz;
+	} else {
+		ih2 = (struct octeon_instr_ih2 *)&sc->cmd.cmd2.ih2;
+		if (ih2->dlengsz) {
+			WARN_ON(!sc->dmadptr);
+			sc->cmd.cmd2.dptr = sc->dmadptr;
+		}
+		irh = (struct octeon_instr_irh *)&sc->cmd.cmd2.irh;
+		if (irh->rflag) {
+			WARN_ON(!sc->dmarptr);
+			WARN_ON(!sc->status_word);
+			*sc->status_word = COMPLETION_WORD_INIT;
+			sc->cmd.cmd2.rptr = sc->dmarptr;
+		}
+		len = (u32)ih2->dlengsz;
+	}
+
+	sc->expiry_time = jiffies + msecs_to_jiffies(LIO_SC_MAX_TMO_MS);
+
+	return (octeon_send_command(oct, sc->iq_no, 1, &sc->cmd, sc,
+				    len, REQTYPE_SOFT_COMMAND));
+}
+EXPORT_SYMBOL_GPL(octeon_send_soft_command);
+
+int octeon_setup_sc_buffer_pool(struct octeon_device *oct)
+{
+	int i;
+	u64 dma_addr;
+	struct octeon_soft_command *sc;
+
+	INIT_LIST_HEAD(&oct->sc_buf_pool.head);
+	spin_lock_init(&oct->sc_buf_pool.lock);
+	atomic_set(&oct->sc_buf_pool.alloc_buf_count, 0);
+
+	for (i = 0; i < MAX_SOFT_COMMAND_BUFFERS; i++) {
+		sc = (struct octeon_soft_command *)
+			lio_dma_alloc(oct,
+				      SOFT_COMMAND_BUFFER_SIZE,
+					  (dma_addr_t *)&dma_addr);
+		if (!sc) {
+			octeon_free_sc_buffer_pool(oct);
+			return 1;
+		}
+
+		sc->dma_addr = dma_addr;
+		sc->size = SOFT_COMMAND_BUFFER_SIZE;
+
+		list_add_tail(&sc->node, &oct->sc_buf_pool.head);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_setup_sc_buffer_pool);
+
+int octeon_free_sc_done_list(struct octeon_device *oct)
+{
+	struct octeon_response_list *done_sc_list, *zombie_sc_list;
+	struct octeon_soft_command *sc;
+	struct list_head *tmp, *tmp2;
+	spinlock_t *sc_lists_lock; /* lock for response_list */
+
+	done_sc_list = &oct->response_list[OCTEON_DONE_SC_LIST];
+	zombie_sc_list = &oct->response_list[OCTEON_ZOMBIE_SC_LIST];
+
+	if (!atomic_read(&done_sc_list->pending_req_count))
+		return 0;
+
+	sc_lists_lock = &oct->response_list[OCTEON_ORDERED_SC_LIST].lock;
+
+	spin_lock_bh(sc_lists_lock);
+
+	list_for_each_safe(tmp, tmp2, &done_sc_list->head) {
+		sc = list_entry(tmp, struct octeon_soft_command, node);
+
+		if (READ_ONCE(sc->caller_is_done)) {
+			list_del(&sc->node);
+			atomic_dec(&done_sc_list->pending_req_count);
+
+			if (*sc->status_word == COMPLETION_WORD_INIT) {
+				/* timeout; move sc to zombie list */
+				list_add_tail(&sc->node, &zombie_sc_list->head);
+				atomic_inc(&zombie_sc_list->pending_req_count);
+			} else {
+				octeon_free_soft_command(oct, sc);
+			}
+		}
+	}
+
+	spin_unlock_bh(sc_lists_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_free_sc_done_list);
+
+int octeon_free_sc_zombie_list(struct octeon_device *oct)
+{
+	struct octeon_response_list *zombie_sc_list;
+	struct octeon_soft_command *sc;
+	struct list_head *tmp, *tmp2;
+	spinlock_t *sc_lists_lock; /* lock for response_list */
+
+	zombie_sc_list = &oct->response_list[OCTEON_ZOMBIE_SC_LIST];
+	sc_lists_lock = &oct->response_list[OCTEON_ORDERED_SC_LIST].lock;
+
+	spin_lock_bh(sc_lists_lock);
+
+	list_for_each_safe(tmp, tmp2, &zombie_sc_list->head) {
+		list_del(tmp);
+		atomic_dec(&zombie_sc_list->pending_req_count);
+		sc = list_entry(tmp, struct octeon_soft_command, node);
+		octeon_free_soft_command(oct, sc);
+	}
+
+	spin_unlock_bh(sc_lists_lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_free_sc_zombie_list);
+
+int octeon_free_sc_buffer_pool(struct octeon_device *oct)
+{
+	struct list_head *tmp, *tmp2;
+	struct octeon_soft_command *sc;
+
+	octeon_free_sc_zombie_list(oct);
+
+	spin_lock_bh(&oct->sc_buf_pool.lock);
+
+	list_for_each_safe(tmp, tmp2, &oct->sc_buf_pool.head) {
+		list_del(tmp);
+
+		sc = (struct octeon_soft_command *)tmp;
+
+		lio_dma_free(oct, sc->size, sc, sc->dma_addr);
+	}
+
+	INIT_LIST_HEAD(&oct->sc_buf_pool.head);
+
+	spin_unlock_bh(&oct->sc_buf_pool.lock);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(octeon_free_sc_buffer_pool);
+
+struct octeon_soft_command *octeon_alloc_soft_command(struct octeon_device *oct,
+						      u32 datasize,
+						      u32 rdatasize,
+						      u32 ctxsize)
+{
+	u64 dma_addr;
+	u32 size;
+	u32 offset = sizeof(struct octeon_soft_command);
+	struct octeon_soft_command *sc = NULL;
+	struct list_head *tmp;
+
+	if (!rdatasize)
+		rdatasize = 16;
+
+	WARN_ON((offset + datasize + rdatasize + ctxsize) >
+	       SOFT_COMMAND_BUFFER_SIZE);
+
+	spin_lock_bh(&oct->sc_buf_pool.lock);
+
+	if (list_empty(&oct->sc_buf_pool.head)) {
+		spin_unlock_bh(&oct->sc_buf_pool.lock);
+		return NULL;
+	}
+
+	list_for_each(tmp, &oct->sc_buf_pool.head)
+		break;
+
+	list_del(tmp);
+
+	atomic_inc(&oct->sc_buf_pool.alloc_buf_count);
+
+	spin_unlock_bh(&oct->sc_buf_pool.lock);
+
+	sc = (struct octeon_soft_command *)tmp;
+
+	dma_addr = sc->dma_addr;
+	size = sc->size;
+
+	memset(sc, 0, sc->size);
+
+	sc->dma_addr = dma_addr;
+	sc->size = size;
+
+	if (ctxsize) {
+		sc->ctxptr = (u8 *)sc + offset;
+		sc->ctxsize = ctxsize;
+	}
+
+	/* Start data at 128 byte boundary */
+	offset = (offset + ctxsize + 127) & 0xffffff80;
+
+	if (datasize) {
+		sc->virtdptr = (u8 *)sc + offset;
+		sc->dmadptr = dma_addr + offset;
+		sc->datasize = datasize;
+	}
+
+	/* Start rdata at 128 byte boundary */
+	offset = (offset + datasize + 127) & 0xffffff80;
+
+	if (rdatasize) {
+		WARN_ON(rdatasize < 16);
+		sc->virtrptr = (u8 *)sc + offset;
+		sc->dmarptr = dma_addr + offset;
+		sc->rdatasize = rdatasize;
+		sc->status_word = (u64 *)((u8 *)(sc->virtrptr) + rdatasize - 8);
+	}
+
+	return sc;
+}
+EXPORT_SYMBOL_GPL(octeon_alloc_soft_command);
+
+void octeon_free_soft_command(struct octeon_device *oct,
+			      struct octeon_soft_command *sc)
+{
+	spin_lock_bh(&oct->sc_buf_pool.lock);
+
+	list_add_tail(&sc->node, &oct->sc_buf_pool.head);
+
+	atomic_dec(&oct->sc_buf_pool.alloc_buf_count);
+
+	spin_unlock_bh(&oct->sc_buf_pool.lock);
+}
+EXPORT_SYMBOL_GPL(octeon_free_soft_command);

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/response_manager.c

@@ -0,0 +1,238 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include "liquidio_common.h"
+#include "octeon_droq.h"
+#include "octeon_iq.h"
+#include "response_manager.h"
+#include "octeon_device.h"
+#include "octeon_main.h"
+
+static void oct_poll_req_completion(struct work_struct *work);
+
+int octeon_setup_response_list(struct octeon_device *oct)
+{
+	int i, ret = 0;
+	struct cavium_wq *cwq;
+
+	for (i = 0; i < MAX_RESPONSE_LISTS; i++) {
+		INIT_LIST_HEAD(&oct->response_list[i].head);
+		spin_lock_init(&oct->response_list[i].lock);
+		atomic_set(&oct->response_list[i].pending_req_count, 0);
+	}
+	spin_lock_init(&oct->cmd_resp_wqlock);
+
+	oct->dma_comp_wq.wq = alloc_workqueue("dma-comp",
+					      WQ_MEM_RECLAIM | WQ_PERCPU, 0);
+	if (!oct->dma_comp_wq.wq) {
+		dev_err(&oct->pci_dev->dev, "failed to create wq thread\n");
+		return -ENOMEM;
+	}
+
+	cwq = &oct->dma_comp_wq;
+	INIT_DELAYED_WORK(&cwq->wk.work, oct_poll_req_completion);
+	cwq->wk.ctxptr = oct;
+	oct->cmd_resp_state = OCT_DRV_ONLINE;
+
+	return ret;
+}
+EXPORT_SYMBOL_GPL(octeon_setup_response_list);
+
+void octeon_delete_response_list(struct octeon_device *oct)
+{
+	cancel_delayed_work_sync(&oct->dma_comp_wq.wk.work);
+	destroy_workqueue(oct->dma_comp_wq.wq);
+}
+EXPORT_SYMBOL_GPL(octeon_delete_response_list);
+
+int lio_process_ordered_list(struct octeon_device *octeon_dev,
+			     u32 force_quit)
+{
+	struct octeon_response_list *ordered_sc_list;
+	struct octeon_soft_command *sc;
+	int request_complete = 0;
+	int resp_to_process = MAX_ORD_REQS_TO_PROCESS;
+	u32 status;
+	u64 status64;
+
+	octeon_free_sc_done_list(octeon_dev);
+
+	ordered_sc_list = &octeon_dev->response_list[OCTEON_ORDERED_SC_LIST];
+
+	do {
+		spin_lock_bh(&ordered_sc_list->lock);
+
+		if (list_empty(&ordered_sc_list->head)) {
+			spin_unlock_bh(&ordered_sc_list->lock);
+			return 1;
+		}
+
+		sc = list_first_entry(&ordered_sc_list->head,
+				      struct octeon_soft_command, node);
+
+		status = OCTEON_REQUEST_PENDING;
+
+		/* check if octeon has finished DMA'ing a response
+		 * to where rptr is pointing to
+		 */
+		status64 = *sc->status_word;
+
+		if (status64 != COMPLETION_WORD_INIT) {
+			/* This logic ensures that all 64b have been written.
+			 * 1. check byte 0 for non-FF
+			 * 2. if non-FF, then swap result from BE to host order
+			 * 3. check byte 7 (swapped to 0) for non-FF
+			 * 4. if non-FF, use the low 32-bit status code
+			 * 5. if either byte 0 or byte 7 is FF, don't use status
+			 */
+			if ((status64 & 0xff) != 0xff) {
+				octeon_swap_8B_data(&status64, 1);
+				if (((status64 & 0xff) != 0xff)) {
+					/* retrieve 16-bit firmware status */
+					status = (u32)(status64 & 0xffffULL);
+					if (status) {
+						status =
+						  FIRMWARE_STATUS_CODE(status);
+					} else {
+						/* i.e. no error */
+						status = OCTEON_REQUEST_DONE;
+					}
+				}
+			}
+		} else if (unlikely(force_quit) || (sc->expiry_time &&
+			time_after(jiffies, (unsigned long)sc->expiry_time))) {
+			struct octeon_instr_irh *irh =
+				(struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
+
+			dev_err(&octeon_dev->pci_dev->dev, "%s: ", __func__);
+			dev_err(&octeon_dev->pci_dev->dev,
+				"cmd %x/%x/%llx/%llx failed, ",
+				irh->opcode, irh->subcode,
+				sc->cmd.cmd3.ossp[0], sc->cmd.cmd3.ossp[1]);
+			dev_err(&octeon_dev->pci_dev->dev,
+				"timeout (%ld, %ld)\n",
+				(long)jiffies, (long)sc->expiry_time);
+			status = OCTEON_REQUEST_TIMEOUT;
+		}
+
+		if (status != OCTEON_REQUEST_PENDING) {
+			sc->sc_status = status;
+
+			/* we have received a response or we have timed out */
+			/* remove node from linked list */
+			list_del(&sc->node);
+			atomic_dec(&octeon_dev->response_list
+				   [OCTEON_ORDERED_SC_LIST].
+				   pending_req_count);
+
+			if (!sc->callback) {
+				atomic_inc(&octeon_dev->response_list
+					   [OCTEON_DONE_SC_LIST].
+					   pending_req_count);
+				list_add_tail(&sc->node,
+					      &octeon_dev->response_list
+					      [OCTEON_DONE_SC_LIST].head);
+
+				if (unlikely(READ_ONCE(sc->caller_is_done))) {
+					/* caller does not wait for response
+					 * from firmware
+					 */
+					if (status != OCTEON_REQUEST_DONE) {
+						struct octeon_instr_irh *irh;
+
+						irh =
+						    (struct octeon_instr_irh *)
+						    &sc->cmd.cmd3.irh;
+						dev_dbg
+						    (&octeon_dev->pci_dev->dev,
+						    "%s: sc failed: opcode=%x, ",
+						    __func__, irh->opcode);
+						dev_dbg
+						    (&octeon_dev->pci_dev->dev,
+						    "subcode=%x, ossp[0]=%llx, ",
+						    irh->subcode,
+						    sc->cmd.cmd3.ossp[0]);
+						dev_dbg
+						    (&octeon_dev->pci_dev->dev,
+						    "ossp[1]=%llx, status=%d\n",
+						    sc->cmd.cmd3.ossp[1],
+						    status);
+					}
+				} else {
+					complete(&sc->complete);
+				}
+
+				spin_unlock_bh(&ordered_sc_list->lock);
+			} else {
+				/* sc with callback function */
+				if (status == OCTEON_REQUEST_TIMEOUT) {
+					atomic_inc(&octeon_dev->response_list
+						   [OCTEON_ZOMBIE_SC_LIST].
+						   pending_req_count);
+					list_add_tail(&sc->node,
+						      &octeon_dev->response_list
+						      [OCTEON_ZOMBIE_SC_LIST].
+						      head);
+				}
+
+				spin_unlock_bh(&ordered_sc_list->lock);
+
+				sc->callback(octeon_dev, status,
+					     sc->callback_arg);
+				/* sc is freed by caller */
+			}
+
+			request_complete++;
+
+		} else {
+			/* no response yet */
+			request_complete = 0;
+			spin_unlock_bh
+			    (&ordered_sc_list->lock);
+		}
+
+		/* If we hit the Max Ordered requests to process every loop,
+		 * we quit
+		 * and let this function be invoked the next time the poll
+		 * thread runs
+		 * to process the remaining requests. This function can take up
+		 * the entire CPU if there is no upper limit to the requests
+		 * processed.
+		 */
+		if (request_complete >= resp_to_process)
+			break;
+	} while (request_complete);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(lio_process_ordered_list);
+
+static void oct_poll_req_completion(struct work_struct *work)
+{
+	struct cavium_wk *wk = (struct cavium_wk *)work;
+	struct octeon_device *oct = (struct octeon_device *)wk->ctxptr;
+	struct cavium_wq *cwq = &oct->dma_comp_wq;
+
+	lio_process_ordered_list(oct, 0);
+
+	if (atomic_read(&oct->response_list
+			[OCTEON_ORDERED_SC_LIST].pending_req_count))
+		queue_delayed_work(cwq->wq, &cwq->wk.work, msecs_to_jiffies(1));
+}

linux-7.2.3/drivers/net/ethernet/cavium/liquidio/response_manager.h

@@ -0,0 +1,143 @@
+/**********************************************************************
+ * Author: Cavium, Inc.
+ *
+ * Contact: support@cavium.com
+ *          Please include "LiquidIO" in the subject.
+ *
+ * Copyright (c) 2003-2016 Cavium, Inc.
+ *
+ * This file is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License, Version 2, as
+ * published by the Free Software Foundation.
+ *
+ * This file is distributed in the hope that it will be useful, but
+ * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
+ * NONINFRINGEMENT.  See the GNU General Public License for more
+ * details.
+ **********************************************************************/
+
+/*! \file response_manager.h
+ *  \brief Host Driver:  Response queues for host instructions.
+ */
+
+#ifndef __RESPONSE_MANAGER_H__
+#define __RESPONSE_MANAGER_H__
+
+/** Maximum ordered requests to process in every invocation of
+ * lio_process_ordered_list(). The function will continue to process requests
+ * as long as it can find one that has finished processing. If it keeps
+ * finding requests that have completed, the function can run for ever. The
+ * value defined here sets an upper limit on the number of requests it can
+ * process before it returns control to the poll thread.
+ */
+#define  MAX_ORD_REQS_TO_PROCESS   4096
+
+/** Head of a response list. There are several response lists in the
+ *  system. One for each response order- Unordered, ordered
+ *  and 1 for noresponse entries on each instruction queue.
+ */
+struct octeon_response_list {
+	/** List structure to add delete pending entries to */
+	struct list_head head;
+
+	/** A lock for this response list */
+	spinlock_t lock;
+
+	atomic_t pending_req_count;
+};
+
+/** The type of response list.
+ */
+enum {
+	OCTEON_ORDERED_LIST = 0,
+	OCTEON_UNORDERED_NONBLOCKING_LIST = 1,
+	OCTEON_UNORDERED_BLOCKING_LIST = 2,
+	OCTEON_ORDERED_SC_LIST = 3,
+	OCTEON_DONE_SC_LIST = 4,
+	OCTEON_ZOMBIE_SC_LIST = 5
+};
+
+/** Response Order values for a Octeon Request. */
+enum {
+	OCTEON_RESP_ORDERED = 0,
+	OCTEON_RESP_UNORDERED = 1,
+	OCTEON_RESP_NORESPONSE = 2
+};
+
+/** Error codes  used in Octeon Host-Core communication.
+ *
+ *   31            16 15            0
+ *   ---------------------------------
+ *   |               |               |
+ *   ---------------------------------
+ *   Error codes are 32-bit wide. The upper 16-bits, called Major Error Number,
+ *   are reserved to identify the group to which the error code belongs. The
+ *   lower 16-bits, called Minor Error Number, carry the actual code.
+ *
+ *   So error codes are (MAJOR NUMBER << 16)| MINOR_NUMBER.
+ */
+
+/*------------   Error codes used by host driver   -----------------*/
+#define DRIVER_MAJOR_ERROR_CODE           0x0000
+/*------   Error codes used by firmware (bits 15..0 set by firmware */
+#define FIRMWARE_MAJOR_ERROR_CODE         0x0001
+
+/**  A value of 0x00000000 indicates no error i.e. success */
+#define DRIVER_ERROR_NONE                 0x00000000
+
+#define DRIVER_ERROR_REQ_PENDING          0x00000001
+#define DRIVER_ERROR_REQ_TIMEOUT          0x00000003
+#define DRIVER_ERROR_REQ_EINTR            0x00000004
+#define DRIVER_ERROR_REQ_ENXIO            0x00000006
+#define DRIVER_ERROR_REQ_ENOMEM           0x0000000C
+#define DRIVER_ERROR_REQ_EINVAL           0x00000016
+#define DRIVER_ERROR_REQ_FAILED           0x000000ff
+
+/** Status for a request.
+ * If a request is not queued to Octeon by the driver, the driver returns
+ * an error condition that's describe by one of the OCTEON_REQ_ERR_* value
+ * below. If the request is successfully queued, the driver will return
+ * a OCTEON_REQUEST_PENDING status. OCTEON_REQUEST_TIMEOUT and
+ * OCTEON_REQUEST_INTERRUPTED are only returned by the driver if the
+ * response for request failed to arrive before a time-out period or if
+ * the request processing * got interrupted due to a signal respectively.
+ */
+enum {
+	OCTEON_REQUEST_DONE = (DRIVER_ERROR_NONE),
+	OCTEON_REQUEST_PENDING = (DRIVER_ERROR_REQ_PENDING),
+	OCTEON_REQUEST_TIMEOUT = (DRIVER_ERROR_REQ_TIMEOUT),
+	OCTEON_REQUEST_INTERRUPTED = (DRIVER_ERROR_REQ_EINTR),
+	OCTEON_REQUEST_NO_DEVICE = (0x00000021),
+	OCTEON_REQUEST_NOT_RUNNING,
+	OCTEON_REQUEST_INVALID_IQ,
+	OCTEON_REQUEST_INVALID_BUFCNT,
+	OCTEON_REQUEST_INVALID_RESP_ORDER,
+	OCTEON_REQUEST_NO_MEMORY,
+	OCTEON_REQUEST_INVALID_BUFSIZE,
+	OCTEON_REQUEST_NO_PENDING_ENTRY,
+	OCTEON_REQUEST_NO_IQ_SPACE = (0x7FFFFFFF)
+
+};
+
+#define FIRMWARE_STATUS_CODE(status) \
+	((FIRMWARE_MAJOR_ERROR_CODE << 16) | (status))
+
+/** Initialize the response lists. The number of response lists to create is
+ * given by count.
+ * @param octeon_dev      - the octeon device structure.
+ */
+int octeon_setup_response_list(struct octeon_device *octeon_dev);
+
+void octeon_delete_response_list(struct octeon_device *octeon_dev);
+
+/** Check the status of first entry in the ordered list. If the instruction at
+ * that entry finished processing or has timed-out, the entry is cleaned.
+ * @param octeon_dev  - the octeon device structure.
+ * @param force_quit - the request is forced to timeout if this is 1
+ * @return 1 if the ordered list is empty, 0 otherwise.
+ */
+int lio_process_ordered_list(struct octeon_device *octeon_dev,
+			     u32 force_quit);
+
+#endif