From: Jakub Kicinski Date: Tue, 1 Dec 2015 14:55:22 +0000 (+0000) Subject: net: add driver for Netronome NFP4000/NFP6000 NIC VFs X-Git-Url: http://git.cdn.openwrt.org/?a=commitdiff_plain;h=4c3523623dc0b980158e34b64360603035239a71;p=openwrt%2Fstaging%2Fblogic.git net: add driver for Netronome NFP4000/NFP6000 NIC VFs Add driver for Virtual Functions for the Netronome's NFP-4000 and NFP-6000 based NICs. Signed-off-by: Jakub Kicinski Signed-off-by: Rolf Neugebauer Signed-off-by: David S. Miller --- diff --git a/MAINTAINERS b/MAINTAINERS index 9981db3f03ed..79029b4bfc80 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -7364,6 +7364,13 @@ F: include/net/netrom.h F: include/uapi/linux/netrom.h F: net/netrom/ +NETRONOME ETHERNET DRIVERS +M: Jakub Kicinski +M: Rolf Neugebauer +L: oss-drivers@netronome.com +S: Maintained +F: drivers/net/ethernet/netronome/ + NETWORK BLOCK DEVICE (NBD) M: Markus Pargmann S: Maintained diff --git a/drivers/net/ethernet/Kconfig b/drivers/net/ethernet/Kconfig index 955d06b9cdba..e57304397d33 100644 --- a/drivers/net/ethernet/Kconfig +++ b/drivers/net/ethernet/Kconfig @@ -121,6 +121,7 @@ config FEALNX cards. source "drivers/net/ethernet/natsemi/Kconfig" +source "drivers/net/ethernet/netronome/Kconfig" source "drivers/net/ethernet/8390/Kconfig" config NET_NETX diff --git a/drivers/net/ethernet/Makefile b/drivers/net/ethernet/Makefile index 4a2ee98738f0..99ede51adf10 100644 --- a/drivers/net/ethernet/Makefile +++ b/drivers/net/ethernet/Makefile @@ -52,6 +52,7 @@ obj-$(CONFIG_NET_VENDOR_MOXART) += moxa/ obj-$(CONFIG_NET_VENDOR_MYRI) += myricom/ obj-$(CONFIG_FEALNX) += fealnx.o obj-$(CONFIG_NET_VENDOR_NATSEMI) += natsemi/ +obj-$(CONFIG_NET_VENDOR_NETRONOME) += netronome/ obj-$(CONFIG_NET_NETX) += netx-eth.o obj-$(CONFIG_NET_VENDOR_NUVOTON) += nuvoton/ obj-$(CONFIG_NET_VENDOR_NVIDIA) += nvidia/ diff --git a/drivers/net/ethernet/netronome/Kconfig b/drivers/net/ethernet/netronome/Kconfig new file mode 100644 index 000000000000..9508ad782c30 --- /dev/null +++ b/drivers/net/ethernet/netronome/Kconfig @@ -0,0 +1,36 @@ +# +# Netronome device configuration +# + +config NET_VENDOR_NETRONOME + bool "Netronome(R) devices" + default y + ---help--- + If you have a Netronome(R) network (Ethernet) card or device, say Y. + + Note that the answer to this question doesn't directly affect the + kernel: saying N will just cause the configurator to skip all + the questions about Netronome(R) cards. If you say Y, you will be + asked for your specific card in the following questions. + +if NET_VENDOR_NETRONOME + +config NFP_NETVF + tristate "Netronome(R) NFP4000/NFP6000 VF NIC driver" + depends on PCI && PCI_MSI + depends on VXLAN || VXLAN=n + ---help--- + This driver supports SR-IOV virtual functions of + the Netronome(R) NFP4000/NFP6000 cards working as + a advanced Ethernet NIC. + +config NFP_NET_DEBUG + bool "Debug support for Netronome(R) NFP3200/NFP6000 NIC drivers" + depends on NFP_NET || NFP_NETVF + ---help--- + Enable extra sanity checks and debugfs support in + Netronome(R) NFP3200/NFP6000 NIC PF and VF drivers. + Note: selecting this option may adversely impact + performance. + +endif diff --git a/drivers/net/ethernet/netronome/Makefile b/drivers/net/ethernet/netronome/Makefile new file mode 100644 index 000000000000..dcb7b383f634 --- /dev/null +++ b/drivers/net/ethernet/netronome/Makefile @@ -0,0 +1,5 @@ +# +# Makefile for the Netronome network device drivers +# + +obj-$(CONFIG_NFP_NETVF) += nfp/ diff --git a/drivers/net/ethernet/netronome/nfp/Makefile b/drivers/net/ethernet/netronome/nfp/Makefile new file mode 100644 index 000000000000..68178819ff12 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/Makefile @@ -0,0 +1,8 @@ +obj-$(CONFIG_NFP_NETVF) += nfp_netvf.o + +nfp_netvf-objs := \ + nfp_net_common.o \ + nfp_net_ethtool.o \ + nfp_netvf_main.o + +nfp_netvf-$(CONFIG_NFP_NET_DEBUG) += nfp_net_debugfs.o diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net.h b/drivers/net/ethernet/netronome/nfp/nfp_net.h new file mode 100644 index 000000000000..ab264e1bccd0 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_net.h @@ -0,0 +1,748 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* + * nfp_net.h + * Declarations for Netronome network device driver. + * Authors: Jakub Kicinski + * Jason McMullan + * Rolf Neugebauer + */ + +#ifndef _NFP_NET_H_ +#define _NFP_NET_H_ + +#include +#include +#include +#include + +#include "nfp_net_ctrl.h" + +#define nn_err(nn, fmt, args...) netdev_err((nn)->netdev, fmt, ## args) +#define nn_warn(nn, fmt, args...) netdev_warn((nn)->netdev, fmt, ## args) +#define nn_info(nn, fmt, args...) netdev_info((nn)->netdev, fmt, ## args) +#define nn_dbg(nn, fmt, args...) netdev_dbg((nn)->netdev, fmt, ## args) +#define nn_warn_ratelimit(nn, fmt, args...) \ + do { \ + if (unlikely(net_ratelimit())) \ + netdev_warn((nn)->netdev, fmt, ## args); \ + } while (0) + +/* Max time to wait for NFP to respond on updates (in ms) */ +#define NFP_NET_POLL_TIMEOUT 5000 + +/* Bar allocation */ +#define NFP_NET_CRTL_BAR 0 +#define NFP_NET_Q0_BAR 2 +#define NFP_NET_Q1_BAR 4 /* OBSOLETE */ + +/* Max bits in DMA address */ +#define NFP_NET_MAX_DMA_BITS 40 + +/* Default size for MTU and freelist buffer sizes */ +#define NFP_NET_DEFAULT_MTU 1500 +#define NFP_NET_DEFAULT_RX_BUFSZ 2048 + +/* Maximum number of bytes prepended to a packet */ +#define NFP_NET_MAX_PREPEND 64 + +/* Interrupt definitions */ +#define NFP_NET_NON_Q_VECTORS 2 +#define NFP_NET_IRQ_LSC_IDX 0 +#define NFP_NET_IRQ_EXN_IDX 1 + +/* Queue/Ring definitions */ +#define NFP_NET_MAX_TX_RINGS 64 /* Max. # of Tx rings per device */ +#define NFP_NET_MAX_RX_RINGS 64 /* Max. # of Rx rings per device */ + +#define NFP_NET_MIN_TX_DESCS 256 /* Min. # of Tx descs per ring */ +#define NFP_NET_MIN_RX_DESCS 256 /* Min. # of Rx descs per ring */ +#define NFP_NET_MAX_TX_DESCS (256 * 1024) /* Max. # of Tx descs per ring */ +#define NFP_NET_MAX_RX_DESCS (256 * 1024) /* Max. # of Rx descs per ring */ + +#define NFP_NET_TX_DESCS_DEFAULT 4096 /* Default # of Tx descs per ring */ +#define NFP_NET_RX_DESCS_DEFAULT 4096 /* Default # of Rx descs per ring */ + +#define NFP_NET_FL_BATCH 16 /* Add freelist in this Batch size */ + +/* Offload definitions */ +#define NFP_NET_N_VXLAN_PORTS (NFP_NET_CFG_VXLAN_SZ / sizeof(__be16)) + +/* Forward declarations */ +struct nfp_net; +struct nfp_net_r_vector; + +/* Convenience macro for writing dma address into RX/TX descriptors */ +#define nfp_desc_set_dma_addr(desc, dma_addr) \ + do { \ + __typeof(desc) __d = (desc); \ + dma_addr_t __addr = (dma_addr); \ + \ + __d->dma_addr_lo = cpu_to_le32(lower_32_bits(__addr)); \ + __d->dma_addr_hi = upper_32_bits(__addr) & 0xff; \ + } while (0) + +/* TX descriptor format */ + +#define PCIE_DESC_TX_EOP BIT(7) +#define PCIE_DESC_TX_OFFSET_MASK GENMASK(6, 0) +#define PCIE_DESC_TX_MSS_MASK GENMASK(13, 0) + +/* Flags in the host TX descriptor */ +#define PCIE_DESC_TX_CSUM BIT(7) +#define PCIE_DESC_TX_IP4_CSUM BIT(6) +#define PCIE_DESC_TX_TCP_CSUM BIT(5) +#define PCIE_DESC_TX_UDP_CSUM BIT(4) +#define PCIE_DESC_TX_VLAN BIT(3) +#define PCIE_DESC_TX_LSO BIT(2) +#define PCIE_DESC_TX_ENCAP BIT(1) +#define PCIE_DESC_TX_O_IP4_CSUM BIT(0) + +struct nfp_net_tx_desc { + union { + struct { + u8 dma_addr_hi; /* High bits of host buf address */ + __le16 dma_len; /* Length to DMA for this desc */ + u8 offset_eop; /* Offset in buf where pkt starts + + * highest bit is eop flag. + */ + __le32 dma_addr_lo; /* Low 32bit of host buf addr */ + + __le16 mss; /* MSS to be used for LSO */ + u8 l4_offset; /* LSO, where the L4 data starts */ + u8 flags; /* TX Flags, see @PCIE_DESC_TX_* */ + + __le16 vlan; /* VLAN tag to add if indicated */ + __le16 data_len; /* Length of frame + meta data */ + } __packed; + __le32 vals[4]; + }; +}; + +/** + * struct nfp_net_tx_buf - software TX buffer descriptor + * @skb: sk_buff associated with this buffer + * @dma_addr: DMA mapping address of the buffer + * @fidx: Fragment index (-1 for the head and [0..nr_frags-1] for frags) + * @pkt_cnt: Number of packets to be produced out of the skb associated + * with this buffer (valid only on the head's buffer). + * Will be 1 for all non-TSO packets. + * @real_len: Number of bytes which to be produced out of the skb (valid only + * on the head's buffer). Equal to skb->len for non-TSO packets. + */ +struct nfp_net_tx_buf { + struct sk_buff *skb; + dma_addr_t dma_addr; + short int fidx; + u16 pkt_cnt; + u32 real_len; +}; + +/** + * struct nfp_net_tx_ring - TX ring structure + * @r_vec: Back pointer to ring vector structure + * @idx: Ring index from Linux's perspective + * @qcidx: Queue Controller Peripheral (QCP) queue index for the TX queue + * @qcp_q: Pointer to base of the QCP TX queue + * @cnt: Size of the queue in number of descriptors + * @wr_p: TX ring write pointer (free running) + * @rd_p: TX ring read pointer (free running) + * @qcp_rd_p: Local copy of QCP TX queue read pointer + * @wr_ptr_add: Accumulated number of buffers to add to QCP write pointer + * (used for .xmit_more delayed kick) + * @txbufs: Array of transmitted TX buffers, to free on transmit + * @txds: Virtual address of TX ring in host memory + * @dma: DMA address of the TX ring + * @size: Size, in bytes, of the TX ring (needed to free) + */ +struct nfp_net_tx_ring { + struct nfp_net_r_vector *r_vec; + + u32 idx; + int qcidx; + u8 __iomem *qcp_q; + + u32 cnt; + u32 wr_p; + u32 rd_p; + u32 qcp_rd_p; + + u32 wr_ptr_add; + + struct nfp_net_tx_buf *txbufs; + struct nfp_net_tx_desc *txds; + + dma_addr_t dma; + unsigned int size; +} ____cacheline_aligned; + +/* RX and freelist descriptor format */ + +#define PCIE_DESC_RX_DD BIT(7) +#define PCIE_DESC_RX_META_LEN_MASK GENMASK(6, 0) + +/* Flags in the RX descriptor */ +#define PCIE_DESC_RX_RSS cpu_to_le16(BIT(15)) +#define PCIE_DESC_RX_I_IP4_CSUM cpu_to_le16(BIT(14)) +#define PCIE_DESC_RX_I_IP4_CSUM_OK cpu_to_le16(BIT(13)) +#define PCIE_DESC_RX_I_TCP_CSUM cpu_to_le16(BIT(12)) +#define PCIE_DESC_RX_I_TCP_CSUM_OK cpu_to_le16(BIT(11)) +#define PCIE_DESC_RX_I_UDP_CSUM cpu_to_le16(BIT(10)) +#define PCIE_DESC_RX_I_UDP_CSUM_OK cpu_to_le16(BIT(9)) +#define PCIE_DESC_RX_SPARE cpu_to_le16(BIT(8)) +#define PCIE_DESC_RX_EOP cpu_to_le16(BIT(7)) +#define PCIE_DESC_RX_IP4_CSUM cpu_to_le16(BIT(6)) +#define PCIE_DESC_RX_IP4_CSUM_OK cpu_to_le16(BIT(5)) +#define PCIE_DESC_RX_TCP_CSUM cpu_to_le16(BIT(4)) +#define PCIE_DESC_RX_TCP_CSUM_OK cpu_to_le16(BIT(3)) +#define PCIE_DESC_RX_UDP_CSUM cpu_to_le16(BIT(2)) +#define PCIE_DESC_RX_UDP_CSUM_OK cpu_to_le16(BIT(1)) +#define PCIE_DESC_RX_VLAN cpu_to_le16(BIT(0)) + +#define PCIE_DESC_RX_CSUM_ALL (PCIE_DESC_RX_IP4_CSUM | \ + PCIE_DESC_RX_TCP_CSUM | \ + PCIE_DESC_RX_UDP_CSUM | \ + PCIE_DESC_RX_I_IP4_CSUM | \ + PCIE_DESC_RX_I_TCP_CSUM | \ + PCIE_DESC_RX_I_UDP_CSUM) +#define PCIE_DESC_RX_CSUM_OK_SHIFT 1 +#define __PCIE_DESC_RX_CSUM_ALL le16_to_cpu(PCIE_DESC_RX_CSUM_ALL) +#define __PCIE_DESC_RX_CSUM_ALL_OK (__PCIE_DESC_RX_CSUM_ALL >> \ + PCIE_DESC_RX_CSUM_OK_SHIFT) + +struct nfp_net_rx_desc { + union { + struct { + u8 dma_addr_hi; /* High bits of the buf address */ + __le16 reserved; /* Must be zero */ + u8 meta_len_dd; /* Must be zero */ + + __le32 dma_addr_lo; /* Low bits of the buffer address */ + } __packed fld; + + struct { + __le16 data_len; /* Length of the frame + meta data */ + u8 reserved; + u8 meta_len_dd; /* Length of meta data prepended + + * descriptor done flag. + */ + + __le16 flags; /* RX flags. See @PCIE_DESC_RX_* */ + __le16 vlan; /* VLAN if stripped */ + } __packed rxd; + + __le32 vals[2]; + }; +}; + +struct nfp_net_rx_hash { + __be32 hash_type; + __be32 hash; +}; + +/** + * struct nfp_net_rx_buf - software RX buffer descriptor + * @skb: sk_buff associated with this buffer + * @dma_addr: DMA mapping address of the buffer + */ +struct nfp_net_rx_buf { + struct sk_buff *skb; + dma_addr_t dma_addr; +}; + +/** + * struct nfp_net_rx_ring - RX ring structure + * @r_vec: Back pointer to ring vector structure + * @cnt: Size of the queue in number of descriptors + * @wr_p: FL/RX ring write pointer (free running) + * @rd_p: FL/RX ring read pointer (free running) + * @idx: Ring index from Linux's perspective + * @fl_qcidx: Queue Controller Peripheral (QCP) queue index for the freelist + * @rx_qcidx: Queue Controller Peripheral (QCP) queue index for the RX queue + * @qcp_fl: Pointer to base of the QCP freelist queue + * @qcp_rx: Pointer to base of the QCP RX queue + * @wr_ptr_add: Accumulated number of buffers to add to QCP write pointer + * (used for free list batching) + * @rxbufs: Array of transmitted FL/RX buffers + * @rxds: Virtual address of FL/RX ring in host memory + * @dma: DMA address of the FL/RX ring + * @size: Size, in bytes, of the FL/RX ring (needed to free) + */ +struct nfp_net_rx_ring { + struct nfp_net_r_vector *r_vec; + + u32 cnt; + u32 wr_p; + u32 rd_p; + + u16 idx; + u16 wr_ptr_add; + + int fl_qcidx; + int rx_qcidx; + u8 __iomem *qcp_fl; + u8 __iomem *qcp_rx; + + struct nfp_net_rx_buf *rxbufs; + struct nfp_net_rx_desc *rxds; + + dma_addr_t dma; + unsigned int size; +} ____cacheline_aligned; + +/** + * struct nfp_net_r_vector - Per ring interrupt vector configuration + * @nfp_net: Backpointer to nfp_net structure + * @napi: NAPI structure for this ring vec + * @tx_ring: Pointer to TX ring + * @rx_ring: Pointer to RX ring + * @irq_idx: Index into MSI-X table + * @rx_sync: Seqlock for atomic updates of RX stats + * @rx_pkts: Number of received packets + * @rx_bytes: Number of received bytes + * @rx_drops: Number of packets dropped on RX due to lack of resources + * @hw_csum_rx_ok: Counter of packets where the HW checksum was OK + * @hw_csum_rx_inner_ok: Counter of packets where the inner HW checksum was OK + * @hw_csum_rx_error: Counter of packets with bad checksums + * @tx_sync: Seqlock for atomic updates of TX stats + * @tx_pkts: Number of Transmitted packets + * @tx_bytes: Number of Transmitted bytes + * @hw_csum_tx: Counter of packets with TX checksum offload requested + * @hw_csum_tx_inner: Counter of inner TX checksum offload requests + * @tx_gather: Counter of packets with Gather DMA + * @tx_lso: Counter of LSO packets sent + * @tx_errors: How many TX errors were encountered + * @tx_busy: How often was TX busy (no space)? + * @handler: Interrupt handler for this ring vector + * @name: Name of the interrupt vector + * @affinity_mask: SMP affinity mask for this vector + * + * This structure ties RX and TX rings to interrupt vectors and a NAPI + * context. This currently only supports one RX and TX ring per + * interrupt vector but might be extended in the future to allow + * association of multiple rings per vector. + */ +struct nfp_net_r_vector { + struct nfp_net *nfp_net; + struct napi_struct napi; + + struct nfp_net_tx_ring *tx_ring; + struct nfp_net_rx_ring *rx_ring; + + int irq_idx; + + struct u64_stats_sync rx_sync; + u64 rx_pkts; + u64 rx_bytes; + u64 rx_drops; + u64 hw_csum_rx_ok; + u64 hw_csum_rx_inner_ok; + u64 hw_csum_rx_error; + + struct u64_stats_sync tx_sync; + u64 tx_pkts; + u64 tx_bytes; + u64 hw_csum_tx; + u64 hw_csum_tx_inner; + u64 tx_gather; + u64 tx_lso; + u64 tx_errors; + u64 tx_busy; + + irq_handler_t handler; + char name[IFNAMSIZ + 8]; + cpumask_t affinity_mask; +} ____cacheline_aligned; + +/* Firmware version as it is written in the 32bit value in the BAR */ +struct nfp_net_fw_version { + u8 minor; + u8 major; + u8 class; + u8 resv; +} __packed; + +static inline bool nfp_net_fw_ver_eq(struct nfp_net_fw_version *fw_ver, + u8 resv, u8 class, u8 major, u8 minor) +{ + return fw_ver->resv == resv && + fw_ver->class == class && + fw_ver->major == major && + fw_ver->minor == minor; +} + +/** + * struct nfp_net - NFP network device structure + * @pdev: Backpointer to PCI device + * @netdev: Backpointer to net_device structure + * @nfp_fallback: Is the driver used in fallback mode? + * @is_vf: Is the driver attached to a VF? + * @is_nfp3200: Is the driver for a NFP-3200 card? + * @fw_loaded: Is the firmware loaded? + * @ctrl: Local copy of the control register/word. + * @fl_bufsz: Currently configured size of the freelist buffers + * @rx_offset: Offset in the RX buffers where packet data starts + * @cpp: Pointer to the CPP handle + * @nfp_dev_cpp: Pointer to the NFP Device handle + * @ctrl_area: Pointer to the CPP area for the control BAR + * @tx_area: Pointer to the CPP area for the TX queues + * @rx_area: Pointer to the CPP area for the FL/RX queues + * @fw_ver: Firmware version + * @cap: Capabilities advertised by the Firmware + * @max_mtu: Maximum support MTU advertised by the Firmware + * @rss_cfg: RSS configuration + * @rss_key: RSS secret key + * @rss_itbl: RSS indirection table + * @max_tx_rings: Maximum number of TX rings supported by the Firmware + * @max_rx_rings: Maximum number of RX rings supported by the Firmware + * @num_tx_rings: Currently configured number of TX rings + * @num_rx_rings: Currently configured number of RX rings + * @txd_cnt: Size of the TX ring in number of descriptors + * @rxd_cnt: Size of the RX ring in number of descriptors + * @tx_rings: Array of pre-allocated TX ring structures + * @rx_rings: Array of pre-allocated RX ring structures + * @num_irqs: Number of allocated interrupt vectors + * @num_r_vecs: Number of used ring vectors + * @r_vecs: Pre-allocated array of ring vectors + * @irq_entries: Pre-allocated array of MSI-X entries + * @lsc_handler: Handler for Link State Change interrupt + * @lsc_name: Name for Link State Change interrupt + * @exn_handler: Handler for Exception interrupt + * @exn_name: Name for Exception interrupt + * @shared_handler: Handler for shared interrupts + * @shared_name: Name for shared interrupt + * @me_freq_mhz: ME clock_freq (MHz) + * @reconfig_lock: Protects HW reconfiguration request regs/machinery + * @link_up: Is the link up? + * @link_status_lock: Protects @link_up and ensures atomicity with BAR reading + * @rx_coalesce_usecs: RX interrupt moderation usecs delay parameter + * @rx_coalesce_max_frames: RX interrupt moderation frame count parameter + * @tx_coalesce_usecs: TX interrupt moderation usecs delay parameter + * @tx_coalesce_max_frames: TX interrupt moderation frame count parameter + * @vxlan_ports: VXLAN ports for RX inner csum offload communicated to HW + * @vxlan_usecnt: IPv4/IPv6 VXLAN port use counts + * @qcp_cfg: Pointer to QCP queue used for configuration notification + * @ctrl_bar: Pointer to mapped control BAR + * @tx_bar: Pointer to mapped TX queues + * @rx_bar: Pointer to mapped FL/RX queues + * @debugfs_dir: Device directory in debugfs + */ +struct nfp_net { + struct pci_dev *pdev; + struct net_device *netdev; + + unsigned nfp_fallback:1; + unsigned is_vf:1; + unsigned is_nfp3200:1; + unsigned fw_loaded:1; + + u32 ctrl; + u32 fl_bufsz; + + u32 rx_offset; + +#ifdef CONFIG_PCI_IOV + unsigned int num_vfs; + struct vf_data_storage *vfinfo; + int vf_rate_link_speed; +#endif + + struct nfp_cpp *cpp; + struct platform_device *nfp_dev_cpp; + struct nfp_cpp_area *ctrl_area; + struct nfp_cpp_area *tx_area; + struct nfp_cpp_area *rx_area; + + struct nfp_net_fw_version fw_ver; + u32 cap; + u32 max_mtu; + + u32 rss_cfg; + u8 rss_key[NFP_NET_CFG_RSS_KEY_SZ]; + u8 rss_itbl[NFP_NET_CFG_RSS_ITBL_SZ]; + + int max_tx_rings; + int max_rx_rings; + + int num_tx_rings; + int num_rx_rings; + + int stride_tx; + int stride_rx; + + int txd_cnt; + int rxd_cnt; + + struct nfp_net_tx_ring tx_rings[NFP_NET_MAX_TX_RINGS]; + struct nfp_net_rx_ring rx_rings[NFP_NET_MAX_RX_RINGS]; + + u8 num_irqs; + u8 num_r_vecs; + struct nfp_net_r_vector r_vecs[NFP_NET_MAX_TX_RINGS]; + struct msix_entry irq_entries[NFP_NET_NON_Q_VECTORS + + NFP_NET_MAX_TX_RINGS]; + + irq_handler_t lsc_handler; + char lsc_name[IFNAMSIZ + 8]; + + irq_handler_t exn_handler; + char exn_name[IFNAMSIZ + 8]; + + irq_handler_t shared_handler; + char shared_name[IFNAMSIZ + 8]; + + u32 me_freq_mhz; + + bool link_up; + spinlock_t link_status_lock; + + spinlock_t reconfig_lock; + + u32 rx_coalesce_usecs; + u32 rx_coalesce_max_frames; + u32 tx_coalesce_usecs; + u32 tx_coalesce_max_frames; + + __be16 vxlan_ports[NFP_NET_N_VXLAN_PORTS]; + u8 vxlan_usecnt[NFP_NET_N_VXLAN_PORTS]; + + u8 __iomem *qcp_cfg; + + u8 __iomem *ctrl_bar; + u8 __iomem *q_bar; + u8 __iomem *tx_bar; + u8 __iomem *rx_bar; + + struct dentry *debugfs_dir; +}; + +/* Functions to read/write from/to a BAR + * Performs any endian conversion necessary. + */ +static inline void nn_writeb(struct nfp_net *nn, int off, u8 val) +{ + writeb(val, nn->ctrl_bar + off); +} + +/* NFP-3200 can't handle 16-bit accesses too well - hence no readw/writew */ + +static inline u32 nn_readl(struct nfp_net *nn, int off) +{ + return readl(nn->ctrl_bar + off); +} + +static inline void nn_writel(struct nfp_net *nn, int off, u32 val) +{ + writel(val, nn->ctrl_bar + off); +} + +static inline u64 nn_readq(struct nfp_net *nn, int off) +{ + return readq(nn->ctrl_bar + off); +} + +static inline void nn_writeq(struct nfp_net *nn, int off, u64 val) +{ + writeq(val, nn->ctrl_bar + off); +} + +/* Flush posted PCI writes by reading something without side effects */ +static inline void nn_pci_flush(struct nfp_net *nn) +{ + nn_readl(nn, NFP_NET_CFG_VERSION); +} + +/* Queue Controller Peripheral access functions and definitions. + * + * Some of the BARs of the NFP are mapped to portions of the Queue + * Controller Peripheral (QCP) address space on the NFP. A QCP queue + * has a read and a write pointer (as well as a size and flags, + * indicating overflow etc). The QCP offers a number of different + * operation on queue pointers, but here we only offer function to + * either add to a pointer or to read the pointer value. + */ +#define NFP_QCP_QUEUE_ADDR_SZ 0x800 +#define NFP_QCP_QUEUE_OFF(_x) ((_x) * NFP_QCP_QUEUE_ADDR_SZ) +#define NFP_QCP_QUEUE_ADD_RPTR 0x0000 +#define NFP_QCP_QUEUE_ADD_WPTR 0x0004 +#define NFP_QCP_QUEUE_STS_LO 0x0008 +#define NFP_QCP_QUEUE_STS_LO_READPTR_mask 0x3ffff +#define NFP_QCP_QUEUE_STS_HI 0x000c +#define NFP_QCP_QUEUE_STS_HI_WRITEPTR_mask 0x3ffff + +/* The offset of a QCP queues in the PCIe Target (same on NFP3200 and NFP6000 */ +#define NFP_PCIE_QUEUE(_q) (0x80000 + (NFP_QCP_QUEUE_ADDR_SZ * ((_q) & 0xff))) + +/* nfp_qcp_ptr - Read or Write Pointer of a queue */ +enum nfp_qcp_ptr { + NFP_QCP_READ_PTR = 0, + NFP_QCP_WRITE_PTR +}; + +/* There appear to be an *undocumented* upper limit on the value which + * one can add to a queue and that value is either 0x3f or 0x7f. We + * go with 0x3f as a conservative measure. + */ +#define NFP_QCP_MAX_ADD 0x3f + +static inline void _nfp_qcp_ptr_add(u8 __iomem *q, + enum nfp_qcp_ptr ptr, u32 val) +{ + u32 off; + + if (ptr == NFP_QCP_READ_PTR) + off = NFP_QCP_QUEUE_ADD_RPTR; + else + off = NFP_QCP_QUEUE_ADD_WPTR; + + while (val > NFP_QCP_MAX_ADD) { + writel(NFP_QCP_MAX_ADD, q + off); + val -= NFP_QCP_MAX_ADD; + } + + writel(val, q + off); +} + +/** + * nfp_qcp_rd_ptr_add() - Add the value to the read pointer of a queue + * + * @q: Base address for queue structure + * @val: Value to add to the queue pointer + * + * If @val is greater than @NFP_QCP_MAX_ADD multiple writes are performed. + */ +static inline void nfp_qcp_rd_ptr_add(u8 __iomem *q, u32 val) +{ + _nfp_qcp_ptr_add(q, NFP_QCP_READ_PTR, val); +} + +/** + * nfp_qcp_wr_ptr_add() - Add the value to the write pointer of a queue + * + * @q: Base address for queue structure + * @val: Value to add to the queue pointer + * + * If @val is greater than @NFP_QCP_MAX_ADD multiple writes are performed. + */ +static inline void nfp_qcp_wr_ptr_add(u8 __iomem *q, u32 val) +{ + _nfp_qcp_ptr_add(q, NFP_QCP_WRITE_PTR, val); +} + +static inline u32 _nfp_qcp_read(u8 __iomem *q, enum nfp_qcp_ptr ptr) +{ + u32 off; + u32 val; + + if (ptr == NFP_QCP_READ_PTR) + off = NFP_QCP_QUEUE_STS_LO; + else + off = NFP_QCP_QUEUE_STS_HI; + + val = readl(q + off); + + if (ptr == NFP_QCP_READ_PTR) + return val & NFP_QCP_QUEUE_STS_LO_READPTR_mask; + else + return val & NFP_QCP_QUEUE_STS_HI_WRITEPTR_mask; +} + +/** + * nfp_qcp_rd_ptr_read() - Read the current read pointer value for a queue + * @q: Base address for queue structure + * + * Return: Value read. + */ +static inline u32 nfp_qcp_rd_ptr_read(u8 __iomem *q) +{ + return _nfp_qcp_read(q, NFP_QCP_READ_PTR); +} + +/** + * nfp_qcp_wr_ptr_read() - Read the current write pointer value for a queue + * @q: Base address for queue structure + * + * Return: Value read. + */ +static inline u32 nfp_qcp_wr_ptr_read(u8 __iomem *q) +{ + return _nfp_qcp_read(q, NFP_QCP_WRITE_PTR); +} + +/* Globals */ +extern const char nfp_net_driver_name[]; +extern const char nfp_net_driver_version[]; + +/* Prototypes */ +void nfp_net_get_fw_version(struct nfp_net_fw_version *fw_ver, + void __iomem *ctrl_bar); + +struct nfp_net *nfp_net_netdev_alloc(struct pci_dev *pdev, + int max_tx_rings, int max_rx_rings); +void nfp_net_netdev_free(struct nfp_net *nn); +int nfp_net_netdev_init(struct net_device *netdev); +void nfp_net_netdev_clean(struct net_device *netdev); +void nfp_net_set_ethtool_ops(struct net_device *netdev); +void nfp_net_info(struct nfp_net *nn); +int nfp_net_reconfig(struct nfp_net *nn, u32 update); +void nfp_net_rss_write_itbl(struct nfp_net *nn); +void nfp_net_rss_write_key(struct nfp_net *nn); +void nfp_net_coalesce_write_cfg(struct nfp_net *nn); +int nfp_net_irqs_alloc(struct nfp_net *nn); +void nfp_net_irqs_disable(struct nfp_net *nn); + +#ifdef CONFIG_NFP_NET_DEBUG +void nfp_net_debugfs_create(void); +void nfp_net_debugfs_destroy(void); +void nfp_net_debugfs_adapter_add(struct nfp_net *nn); +void nfp_net_debugfs_adapter_del(struct nfp_net *nn); +#else +static inline void nfp_net_debugfs_create(void) +{ +} + +static inline void nfp_net_debugfs_destroy(void) +{ +} + +static inline void nfp_net_debugfs_adapter_add(struct nfp_net *nn) +{ +} + +static inline void nfp_net_debugfs_adapter_del(struct nfp_net *nn) +{ +} +#endif /* CONFIG_NFP_NET_DEBUG */ + +#endif /* _NFP_NET_H_ */ diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_common.c b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c new file mode 100644 index 000000000000..038ac6b14a60 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_common.c @@ -0,0 +1,2432 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* + * nfp_net_common.c + * Netronome network device driver: Common functions between PF and VF + * Authors: Jakub Kicinski + * Jason McMullan + * Rolf Neugebauer + * Brad Petrus + * Chris Telfer + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include + +#include "nfp_net_ctrl.h" +#include "nfp_net.h" + +/** + * nfp_net_get_fw_version() - Read and parse the FW version + * @fw_ver: Output fw_version structure to read to + * @ctrl_bar: Mapped address of the control BAR + */ +void nfp_net_get_fw_version(struct nfp_net_fw_version *fw_ver, + void __iomem *ctrl_bar) +{ + u32 reg; + + reg = readl(ctrl_bar + NFP_NET_CFG_VERSION); + put_unaligned_le32(reg, fw_ver); +} + +/** + * nfp_net_reconfig() - Reconfigure the firmware + * @nn: NFP Net device to reconfigure + * @update: The value for the update field in the BAR config + * + * Write the update word to the BAR and ping the reconfig queue. The + * poll until the firmware has acknowledged the update by zeroing the + * update word. + * + * Return: Negative errno on error, 0 on success + */ +int nfp_net_reconfig(struct nfp_net *nn, u32 update) +{ + int cnt, ret = 0; + u32 new; + + spin_lock_bh(&nn->reconfig_lock); + + nn_writel(nn, NFP_NET_CFG_UPDATE, update); + /* ensure update is written before pinging HW */ + nn_pci_flush(nn); + nfp_qcp_wr_ptr_add(nn->qcp_cfg, 1); + + /* Poll update field, waiting for NFP to ack the config */ + for (cnt = 0; ; cnt++) { + new = nn_readl(nn, NFP_NET_CFG_UPDATE); + if (new == 0) + break; + if (new & NFP_NET_CFG_UPDATE_ERR) { + nn_err(nn, "Reconfig error: 0x%08x\n", new); + ret = -EIO; + break; + } else if (cnt >= NFP_NET_POLL_TIMEOUT) { + nn_err(nn, "Reconfig timeout for 0x%08x after %dms\n", + update, cnt); + ret = -EIO; + break; + } + mdelay(1); + } + + spin_unlock_bh(&nn->reconfig_lock); + return ret; +} + +/* Interrupt configuration and handling + */ + +/** + * nfp_net_irq_unmask_msix() - Unmask MSI-X after automasking + * @nn: NFP Network structure + * @entry_nr: MSI-X table entry + * + * Clear the MSI-X table mask bit for the given entry bypassing Linux irq + * handling subsystem. Use *only* to reenable automasked vectors. + */ +static void nfp_net_irq_unmask_msix(struct nfp_net *nn, unsigned int entry_nr) +{ + struct list_head *msi_head = &nn->pdev->dev.msi_list; + struct msi_desc *entry; + u32 off; + + /* All MSI-Xs have the same mask_base */ + entry = list_first_entry(msi_head, struct msi_desc, list); + + off = (PCI_MSIX_ENTRY_SIZE * entry_nr) + + PCI_MSIX_ENTRY_VECTOR_CTRL; + writel(0, entry->mask_base + off); + readl(entry->mask_base); +} + +/** + * nfp_net_irq_unmask() - Unmask automasked interrupt + * @nn: NFP Network structure + * @entry_nr: MSI-X table entry + * + * If MSI-X auto-masking is enabled clear the mask bit, otherwise + * clear the ICR for the entry. + */ +static void nfp_net_irq_unmask(struct nfp_net *nn, unsigned int entry_nr) +{ + if (nn->ctrl & NFP_NET_CFG_CTRL_MSIXAUTO) { + nfp_net_irq_unmask_msix(nn, entry_nr); + return; + } + + nn_writeb(nn, NFP_NET_CFG_ICR(entry_nr), NFP_NET_CFG_ICR_UNMASKED); + nn_pci_flush(nn); +} + +/** + * nfp_net_msix_alloc() - Try to allocate MSI-X irqs + * @nn: NFP Network structure + * @nr_vecs: Number of MSI-X vectors to allocate + * + * For MSI-X we want at least NFP_NET_NON_Q_VECTORS + 1 vectors. + * + * Return: Number of MSI-X vectors obtained or 0 on error. + */ +static int nfp_net_msix_alloc(struct nfp_net *nn, int nr_vecs) +{ + struct pci_dev *pdev = nn->pdev; + int nvecs; + int i; + + for (i = 0; i < nr_vecs; i++) + nn->irq_entries[i].entry = i; + + nvecs = pci_enable_msix_range(pdev, nn->irq_entries, + NFP_NET_NON_Q_VECTORS + 1, nr_vecs); + if (nvecs < 0) { + nn_warn(nn, "Failed to enable MSI-X. Wanted %d-%d (err=%d)\n", + NFP_NET_NON_Q_VECTORS + 1, nr_vecs, nvecs); + return 0; + } + + return nvecs; +} + +/** + * nfp_net_irqs_wanted() - Work out how many interrupt vectors we want + * @nn: NFP Network structure + * + * We want a vector per CPU (or ring), whatever is smaller plus + * NFP_NET_NON_Q_VECTORS for LSC etc. + * + * Return: Number of interrupts wanted + */ +static int nfp_net_irqs_wanted(struct nfp_net *nn) +{ + int ncpus; + int vecs; + + ncpus = num_online_cpus(); + + vecs = max_t(int, nn->num_tx_rings, nn->num_rx_rings); + vecs = min_t(int, vecs, ncpus); + + return vecs + NFP_NET_NON_Q_VECTORS; +} + +/** + * nfp_net_irqs_alloc() - allocates MSI-X irqs + * @nn: NFP Network structure + * + * Return: Number of irqs obtained or 0 on error. + */ +int nfp_net_irqs_alloc(struct nfp_net *nn) +{ + int wanted_irqs; + + wanted_irqs = nfp_net_irqs_wanted(nn); + + nn->num_irqs = nfp_net_msix_alloc(nn, wanted_irqs); + if (nn->num_irqs == 0) { + nn_err(nn, "Failed to allocate MSI-X IRQs\n"); + return 0; + } + + nn->num_r_vecs = nn->num_irqs - NFP_NET_NON_Q_VECTORS; + + if (nn->num_irqs < wanted_irqs) + nn_warn(nn, "Unable to allocate %d vectors. Got %d instead\n", + wanted_irqs, nn->num_irqs); + + return nn->num_irqs; +} + +/** + * nfp_net_irqs_disable() - Disable interrupts + * @nn: NFP Network structure + * + * Undoes what @nfp_net_irqs_alloc() does. + */ +void nfp_net_irqs_disable(struct nfp_net *nn) +{ + pci_disable_msix(nn->pdev); +} + +/** + * nfp_net_irq_rxtx() - Interrupt service routine for RX/TX rings. + * @irq: Interrupt + * @data: Opaque data structure + * + * Return: Indicate if the interrupt has been handled. + */ +static irqreturn_t nfp_net_irq_rxtx(int irq, void *data) +{ + struct nfp_net_r_vector *r_vec = data; + + napi_schedule_irqoff(&r_vec->napi); + + /* The FW auto-masks any interrupt, either via the MASK bit in + * the MSI-X table or via the per entry ICR field. So there + * is no need to disable interrupts here. + */ + return IRQ_HANDLED; +} + +/** + * nfp_net_read_link_status() - Reread link status from control BAR + * @nn: NFP Network structure + */ +static void nfp_net_read_link_status(struct nfp_net *nn) +{ + unsigned long flags; + bool link_up; + u32 sts; + + spin_lock_irqsave(&nn->link_status_lock, flags); + + sts = nn_readl(nn, NFP_NET_CFG_STS); + link_up = !!(sts & NFP_NET_CFG_STS_LINK); + + if (nn->link_up == link_up) + goto out; + + nn->link_up = link_up; + + if (nn->link_up) { + netif_carrier_on(nn->netdev); + netdev_info(nn->netdev, "NIC Link is Up\n"); + } else { + netif_carrier_off(nn->netdev); + netdev_info(nn->netdev, "NIC Link is Down\n"); + } +out: + spin_unlock_irqrestore(&nn->link_status_lock, flags); +} + +/** + * nfp_net_irq_lsc() - Interrupt service routine for link state changes + * @irq: Interrupt + * @data: Opaque data structure + * + * Return: Indicate if the interrupt has been handled. + */ +static irqreturn_t nfp_net_irq_lsc(int irq, void *data) +{ + struct nfp_net *nn = data; + + nfp_net_read_link_status(nn); + + nfp_net_irq_unmask(nn, NFP_NET_IRQ_LSC_IDX); + + return IRQ_HANDLED; +} + +/** + * nfp_net_irq_exn() - Interrupt service routine for exceptions + * @irq: Interrupt + * @data: Opaque data structure + * + * Return: Indicate if the interrupt has been handled. + */ +static irqreturn_t nfp_net_irq_exn(int irq, void *data) +{ + struct nfp_net *nn = data; + + nn_err(nn, "%s: UNIMPLEMENTED.\n", __func__); + /* XXX TO BE IMPLEMENTED */ + return IRQ_HANDLED; +} + +/** + * nfp_net_tx_ring_init() - Fill in the boilerplate for a TX ring + * @tx_ring: TX ring structure + */ +static void nfp_net_tx_ring_init(struct nfp_net_tx_ring *tx_ring) +{ + struct nfp_net_r_vector *r_vec = tx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + + tx_ring->qcidx = tx_ring->idx * nn->stride_tx; + tx_ring->qcp_q = nn->tx_bar + NFP_QCP_QUEUE_OFF(tx_ring->qcidx); +} + +/** + * nfp_net_rx_ring_init() - Fill in the boilerplate for a RX ring + * @rx_ring: RX ring structure + */ +static void nfp_net_rx_ring_init(struct nfp_net_rx_ring *rx_ring) +{ + struct nfp_net_r_vector *r_vec = rx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + + rx_ring->fl_qcidx = rx_ring->idx * nn->stride_rx; + rx_ring->rx_qcidx = rx_ring->fl_qcidx + (nn->stride_rx - 1); + + rx_ring->qcp_fl = nn->rx_bar + NFP_QCP_QUEUE_OFF(rx_ring->fl_qcidx); + rx_ring->qcp_rx = nn->rx_bar + NFP_QCP_QUEUE_OFF(rx_ring->rx_qcidx); +} + +/** + * nfp_net_irqs_assign() - Assign IRQs and setup rvecs. + * @netdev: netdev structure + */ +static void nfp_net_irqs_assign(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + struct nfp_net_r_vector *r_vec; + int r; + + /* Assumes nn->num_tx_rings == nn->num_rx_rings */ + if (nn->num_tx_rings > nn->num_r_vecs) { + nn_warn(nn, "More rings (%d) than vectors (%d).\n", + nn->num_tx_rings, nn->num_r_vecs); + nn->num_tx_rings = nn->num_r_vecs; + nn->num_rx_rings = nn->num_r_vecs; + } + + nn->lsc_handler = nfp_net_irq_lsc; + nn->exn_handler = nfp_net_irq_exn; + + for (r = 0; r < nn->num_r_vecs; r++) { + r_vec = &nn->r_vecs[r]; + r_vec->nfp_net = nn; + r_vec->handler = nfp_net_irq_rxtx; + r_vec->irq_idx = NFP_NET_NON_Q_VECTORS + r; + + cpumask_set_cpu(r, &r_vec->affinity_mask); + + r_vec->tx_ring = &nn->tx_rings[r]; + nn->tx_rings[r].idx = r; + nn->tx_rings[r].r_vec = r_vec; + nfp_net_tx_ring_init(r_vec->tx_ring); + + r_vec->rx_ring = &nn->rx_rings[r]; + nn->rx_rings[r].idx = r; + nn->rx_rings[r].r_vec = r_vec; + nfp_net_rx_ring_init(r_vec->rx_ring); + } +} + +/** + * nfp_net_aux_irq_request() - Request an auxiliary interrupt (LSC or EXN) + * @nn: NFP Network structure + * @ctrl_offset: Control BAR offset where IRQ configuration should be written + * @format: printf-style format to construct the interrupt name + * @name: Pointer to allocated space for interrupt name + * @name_sz: Size of space for interrupt name + * @vector_idx: Index of MSI-X vector used for this interrupt + * @handler: IRQ handler to register for this interrupt + */ +static int +nfp_net_aux_irq_request(struct nfp_net *nn, u32 ctrl_offset, + const char *format, char *name, size_t name_sz, + unsigned int vector_idx, irq_handler_t handler) +{ + struct msix_entry *entry; + int err; + + entry = &nn->irq_entries[vector_idx]; + + snprintf(name, name_sz, format, netdev_name(nn->netdev)); + err = request_irq(entry->vector, handler, 0, name, nn); + if (err) { + nn_err(nn, "Failed to request IRQ %d (err=%d).\n", + entry->vector, err); + return err; + } + nn_writeb(nn, ctrl_offset, vector_idx); + + return 0; +} + +/** + * nfp_net_aux_irq_free() - Free an auxiliary interrupt (LSC or EXN) + * @nn: NFP Network structure + * @ctrl_offset: Control BAR offset where IRQ configuration should be written + * @vector_idx: Index of MSI-X vector used for this interrupt + */ +static void nfp_net_aux_irq_free(struct nfp_net *nn, u32 ctrl_offset, + unsigned int vector_idx) +{ + nn_writeb(nn, ctrl_offset, 0xff); + free_irq(nn->irq_entries[vector_idx].vector, nn); +} + +/* Transmit + * + * One queue controller peripheral queue is used for transmit. The + * driver en-queues packets for transmit by advancing the write + * pointer. The device indicates that packets have transmitted by + * advancing the read pointer. The driver maintains a local copy of + * the read and write pointer in @struct nfp_net_tx_ring. The driver + * keeps @wr_p in sync with the queue controller write pointer and can + * determine how many packets have been transmitted by comparing its + * copy of the read pointer @rd_p with the read pointer maintained by + * the queue controller peripheral. + */ + +/** + * nfp_net_tx_full() - Check if the TX ring is full + * @tx_ring: TX ring to check + * @dcnt: Number of descriptors that need to be enqueued (must be >= 1) + * + * This function checks, based on the *host copy* of read/write + * pointer if a given TX ring is full. The real TX queue may have + * some newly made available slots. + * + * Return: True if the ring is full. + */ +static inline int nfp_net_tx_full(struct nfp_net_tx_ring *tx_ring, int dcnt) +{ + return (tx_ring->wr_p - tx_ring->rd_p) >= (tx_ring->cnt - dcnt); +} + +/* Wrappers for deciding when to stop and restart TX queues */ +static int nfp_net_tx_ring_should_wake(struct nfp_net_tx_ring *tx_ring) +{ + return !nfp_net_tx_full(tx_ring, MAX_SKB_FRAGS * 4); +} + +static int nfp_net_tx_ring_should_stop(struct nfp_net_tx_ring *tx_ring) +{ + return nfp_net_tx_full(tx_ring, MAX_SKB_FRAGS + 1); +} + +/** + * nfp_net_tx_ring_stop() - stop tx ring + * @nd_q: netdev queue + * @tx_ring: driver tx queue structure + * + * Safely stop TX ring. Remember that while we are running .start_xmit() + * someone else may be cleaning the TX ring completions so we need to be + * extra careful here. + */ +static void nfp_net_tx_ring_stop(struct netdev_queue *nd_q, + struct nfp_net_tx_ring *tx_ring) +{ + netif_tx_stop_queue(nd_q); + + /* We can race with the TX completion out of NAPI so recheck */ + smp_mb(); + if (unlikely(nfp_net_tx_ring_should_wake(tx_ring))) + netif_tx_start_queue(nd_q); +} + +/** + * nfp_net_tx_tso() - Set up Tx descriptor for LSO + * @nn: NFP Net device + * @r_vec: per-ring structure + * @txbuf: Pointer to driver soft TX descriptor + * @txd: Pointer to HW TX descriptor + * @skb: Pointer to SKB + * + * Set up Tx descriptor for LSO, do nothing for non-LSO skbs. + * Return error on packet header greater than maximum supported LSO header size. + */ +static void nfp_net_tx_tso(struct nfp_net *nn, struct nfp_net_r_vector *r_vec, + struct nfp_net_tx_buf *txbuf, + struct nfp_net_tx_desc *txd, struct sk_buff *skb) +{ + u32 hdrlen; + u16 mss; + + if (!skb_is_gso(skb)) + return; + + if (!skb->encapsulation) + hdrlen = skb_transport_offset(skb) + tcp_hdrlen(skb); + else + hdrlen = skb_inner_transport_header(skb) - skb->data + + inner_tcp_hdrlen(skb); + + txbuf->pkt_cnt = skb_shinfo(skb)->gso_segs; + txbuf->real_len += hdrlen * (txbuf->pkt_cnt - 1); + + mss = skb_shinfo(skb)->gso_size & PCIE_DESC_TX_MSS_MASK; + txd->l4_offset = hdrlen; + txd->mss = cpu_to_le16(mss); + txd->flags |= PCIE_DESC_TX_LSO; + + u64_stats_update_begin(&r_vec->tx_sync); + r_vec->tx_lso++; + u64_stats_update_end(&r_vec->tx_sync); +} + +/** + * nfp_net_tx_csum() - Set TX CSUM offload flags in TX descriptor + * @nn: NFP Net device + * @r_vec: per-ring structure + * @txbuf: Pointer to driver soft TX descriptor + * @txd: Pointer to TX descriptor + * @skb: Pointer to SKB + * + * This function sets the TX checksum flags in the TX descriptor based + * on the configuration and the protocol of the packet to be transmitted. + */ +static void nfp_net_tx_csum(struct nfp_net *nn, struct nfp_net_r_vector *r_vec, + struct nfp_net_tx_buf *txbuf, + struct nfp_net_tx_desc *txd, struct sk_buff *skb) +{ + struct ipv6hdr *ipv6h; + struct iphdr *iph; + u8 l4_hdr; + + if (!(nn->ctrl & NFP_NET_CFG_CTRL_TXCSUM)) + return; + + if (skb->ip_summed != CHECKSUM_PARTIAL) + return; + + txd->flags |= PCIE_DESC_TX_CSUM; + if (skb->encapsulation) + txd->flags |= PCIE_DESC_TX_ENCAP; + + iph = skb->encapsulation ? inner_ip_hdr(skb) : ip_hdr(skb); + ipv6h = skb->encapsulation ? inner_ipv6_hdr(skb) : ipv6_hdr(skb); + + if (iph->version == 4) { + txd->flags |= PCIE_DESC_TX_IP4_CSUM; + l4_hdr = iph->protocol; + } else if (ipv6h->version == 6) { + l4_hdr = ipv6h->nexthdr; + } else { + nn_warn_ratelimit(nn, "partial checksum but ipv=%x!\n", + iph->version); + return; + } + + switch (l4_hdr) { + case IPPROTO_TCP: + txd->flags |= PCIE_DESC_TX_TCP_CSUM; + break; + case IPPROTO_UDP: + txd->flags |= PCIE_DESC_TX_UDP_CSUM; + break; + default: + nn_warn_ratelimit(nn, "partial checksum but l4 proto=%x!\n", + l4_hdr); + return; + } + + u64_stats_update_begin(&r_vec->tx_sync); + if (skb->encapsulation) + r_vec->hw_csum_tx_inner += txbuf->pkt_cnt; + else + r_vec->hw_csum_tx += txbuf->pkt_cnt; + u64_stats_update_end(&r_vec->tx_sync); +} + +/** + * nfp_net_tx() - Main transmit entry point + * @skb: SKB to transmit + * @netdev: netdev structure + * + * Return: NETDEV_TX_OK on success. + */ +static int nfp_net_tx(struct sk_buff *skb, struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + const struct skb_frag_struct *frag; + struct nfp_net_r_vector *r_vec; + struct nfp_net_tx_desc *txd, txdg; + struct nfp_net_tx_buf *txbuf; + struct nfp_net_tx_ring *tx_ring; + struct netdev_queue *nd_q; + dma_addr_t dma_addr; + unsigned int fsize; + int f, nr_frags; + int wr_idx; + u16 qidx; + + qidx = skb_get_queue_mapping(skb); + tx_ring = &nn->tx_rings[qidx]; + r_vec = tx_ring->r_vec; + nd_q = netdev_get_tx_queue(nn->netdev, qidx); + + nr_frags = skb_shinfo(skb)->nr_frags; + + if (unlikely(nfp_net_tx_full(tx_ring, nr_frags + 1))) { + nn_warn_ratelimit(nn, "TX ring %d busy. wrp=%u rdp=%u\n", + qidx, tx_ring->wr_p, tx_ring->rd_p); + netif_tx_stop_queue(nd_q); + u64_stats_update_begin(&r_vec->tx_sync); + r_vec->tx_busy++; + u64_stats_update_end(&r_vec->tx_sync); + return NETDEV_TX_BUSY; + } + + /* Start with the head skbuf */ + dma_addr = dma_map_single(&nn->pdev->dev, skb->data, skb_headlen(skb), + DMA_TO_DEVICE); + if (dma_mapping_error(&nn->pdev->dev, dma_addr)) + goto err_free; + + wr_idx = tx_ring->wr_p % tx_ring->cnt; + + /* Stash the soft descriptor of the head then initialize it */ + txbuf = &tx_ring->txbufs[wr_idx]; + txbuf->skb = skb; + txbuf->dma_addr = dma_addr; + txbuf->fidx = -1; + txbuf->pkt_cnt = 1; + txbuf->real_len = skb->len; + + /* Build TX descriptor */ + txd = &tx_ring->txds[wr_idx]; + txd->offset_eop = (nr_frags == 0) ? PCIE_DESC_TX_EOP : 0; + txd->dma_len = cpu_to_le16(skb_headlen(skb)); + nfp_desc_set_dma_addr(txd, dma_addr); + txd->data_len = cpu_to_le16(skb->len); + + txd->flags = 0; + txd->mss = 0; + txd->l4_offset = 0; + + nfp_net_tx_tso(nn, r_vec, txbuf, txd, skb); + + nfp_net_tx_csum(nn, r_vec, txbuf, txd, skb); + + if (skb_vlan_tag_present(skb) && nn->ctrl & NFP_NET_CFG_CTRL_TXVLAN) { + txd->flags |= PCIE_DESC_TX_VLAN; + txd->vlan = cpu_to_le16(skb_vlan_tag_get(skb)); + } + + /* Gather DMA */ + if (nr_frags > 0) { + /* all descs must match except for in addr, length and eop */ + txdg = *txd; + + for (f = 0; f < nr_frags; f++) { + frag = &skb_shinfo(skb)->frags[f]; + fsize = skb_frag_size(frag); + + dma_addr = skb_frag_dma_map(&nn->pdev->dev, frag, 0, + fsize, DMA_TO_DEVICE); + if (dma_mapping_error(&nn->pdev->dev, dma_addr)) + goto err_unmap; + + wr_idx = (wr_idx + 1) % tx_ring->cnt; + tx_ring->txbufs[wr_idx].skb = skb; + tx_ring->txbufs[wr_idx].dma_addr = dma_addr; + tx_ring->txbufs[wr_idx].fidx = f; + + txd = &tx_ring->txds[wr_idx]; + *txd = txdg; + txd->dma_len = cpu_to_le16(fsize); + nfp_desc_set_dma_addr(txd, dma_addr); + txd->offset_eop = + (f == nr_frags - 1) ? PCIE_DESC_TX_EOP : 0; + } + + u64_stats_update_begin(&r_vec->tx_sync); + r_vec->tx_gather++; + u64_stats_update_end(&r_vec->tx_sync); + } + + netdev_tx_sent_queue(nd_q, txbuf->real_len); + + tx_ring->wr_p += nr_frags + 1; + if (nfp_net_tx_ring_should_stop(tx_ring)) + nfp_net_tx_ring_stop(nd_q, tx_ring); + + tx_ring->wr_ptr_add += nr_frags + 1; + if (!skb->xmit_more || netif_xmit_stopped(nd_q)) { + /* force memory write before we let HW know */ + wmb(); + nfp_qcp_wr_ptr_add(tx_ring->qcp_q, tx_ring->wr_ptr_add); + tx_ring->wr_ptr_add = 0; + } + + skb_tx_timestamp(skb); + + return NETDEV_TX_OK; + +err_unmap: + --f; + while (f >= 0) { + frag = &skb_shinfo(skb)->frags[f]; + dma_unmap_page(&nn->pdev->dev, + tx_ring->txbufs[wr_idx].dma_addr, + skb_frag_size(frag), DMA_TO_DEVICE); + tx_ring->txbufs[wr_idx].skb = NULL; + tx_ring->txbufs[wr_idx].dma_addr = 0; + tx_ring->txbufs[wr_idx].fidx = -2; + wr_idx = wr_idx - 1; + if (wr_idx < 0) + wr_idx += tx_ring->cnt; + } + dma_unmap_single(&nn->pdev->dev, tx_ring->txbufs[wr_idx].dma_addr, + skb_headlen(skb), DMA_TO_DEVICE); + tx_ring->txbufs[wr_idx].skb = NULL; + tx_ring->txbufs[wr_idx].dma_addr = 0; + tx_ring->txbufs[wr_idx].fidx = -2; +err_free: + nn_warn_ratelimit(nn, "Failed to map DMA TX buffer\n"); + u64_stats_update_begin(&r_vec->tx_sync); + r_vec->tx_errors++; + u64_stats_update_end(&r_vec->tx_sync); + dev_kfree_skb_any(skb); + return NETDEV_TX_OK; +} + +/** + * nfp_net_tx_complete() - Handled completed TX packets + * @tx_ring: TX ring structure + * + * Return: Number of completed TX descriptors + */ +static void nfp_net_tx_complete(struct nfp_net_tx_ring *tx_ring) +{ + struct nfp_net_r_vector *r_vec = tx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + const struct skb_frag_struct *frag; + struct netdev_queue *nd_q; + u32 done_pkts = 0, done_bytes = 0; + struct sk_buff *skb; + int todo, nr_frags; + u32 qcp_rd_p; + int fidx; + int idx; + + /* Work out how many descriptors have been transmitted */ + qcp_rd_p = nfp_qcp_rd_ptr_read(tx_ring->qcp_q); + + if (qcp_rd_p == tx_ring->qcp_rd_p) + return; + + if (qcp_rd_p > tx_ring->qcp_rd_p) + todo = qcp_rd_p - tx_ring->qcp_rd_p; + else + todo = qcp_rd_p + tx_ring->cnt - tx_ring->qcp_rd_p; + + while (todo--) { + idx = tx_ring->rd_p % tx_ring->cnt; + tx_ring->rd_p++; + + skb = tx_ring->txbufs[idx].skb; + if (!skb) + continue; + + nr_frags = skb_shinfo(skb)->nr_frags; + fidx = tx_ring->txbufs[idx].fidx; + + if (fidx == -1) { + /* unmap head */ + dma_unmap_single(&nn->pdev->dev, + tx_ring->txbufs[idx].dma_addr, + skb_headlen(skb), DMA_TO_DEVICE); + + done_pkts += tx_ring->txbufs[idx].pkt_cnt; + done_bytes += tx_ring->txbufs[idx].real_len; + } else { + /* unmap fragment */ + frag = &skb_shinfo(skb)->frags[fidx]; + dma_unmap_page(&nn->pdev->dev, + tx_ring->txbufs[idx].dma_addr, + skb_frag_size(frag), DMA_TO_DEVICE); + } + + /* check for last gather fragment */ + if (fidx == nr_frags - 1) + dev_kfree_skb_any(skb); + + tx_ring->txbufs[idx].dma_addr = 0; + tx_ring->txbufs[idx].skb = NULL; + tx_ring->txbufs[idx].fidx = -2; + } + + tx_ring->qcp_rd_p = qcp_rd_p; + + u64_stats_update_begin(&r_vec->tx_sync); + r_vec->tx_bytes += done_bytes; + r_vec->tx_pkts += done_pkts; + u64_stats_update_end(&r_vec->tx_sync); + + nd_q = netdev_get_tx_queue(nn->netdev, tx_ring->idx); + netdev_tx_completed_queue(nd_q, done_pkts, done_bytes); + if (nfp_net_tx_ring_should_wake(tx_ring)) { + /* Make sure TX thread will see updated tx_ring->rd_p */ + smp_mb(); + + if (unlikely(netif_tx_queue_stopped(nd_q))) + netif_tx_wake_queue(nd_q); + } + + WARN_ONCE(tx_ring->wr_p - tx_ring->rd_p > tx_ring->cnt, + "TX ring corruption rd_p=%u wr_p=%u cnt=%u\n", + tx_ring->rd_p, tx_ring->wr_p, tx_ring->cnt); +} + +/** + * nfp_net_tx_flush() - Free any untransmitted buffers currently on the TX ring + * @tx_ring: TX ring structure + * + * Assumes that the device is stopped + */ +static void nfp_net_tx_flush(struct nfp_net_tx_ring *tx_ring) +{ + struct nfp_net_r_vector *r_vec = tx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + const struct skb_frag_struct *frag; + struct netdev_queue *nd_q; + struct sk_buff *skb; + int nr_frags; + int fidx; + int idx; + + while (tx_ring->rd_p != tx_ring->wr_p) { + idx = tx_ring->rd_p % tx_ring->cnt; + + skb = tx_ring->txbufs[idx].skb; + if (skb) { + nr_frags = skb_shinfo(skb)->nr_frags; + fidx = tx_ring->txbufs[idx].fidx; + + if (fidx == -1) { + /* unmap head */ + dma_unmap_single(&pdev->dev, + tx_ring->txbufs[idx].dma_addr, + skb_headlen(skb), + DMA_TO_DEVICE); + } else { + /* unmap fragment */ + frag = &skb_shinfo(skb)->frags[fidx]; + dma_unmap_page(&pdev->dev, + tx_ring->txbufs[idx].dma_addr, + skb_frag_size(frag), + DMA_TO_DEVICE); + } + + /* check for last gather fragment */ + if (fidx == nr_frags - 1) + dev_kfree_skb_any(skb); + + tx_ring->txbufs[idx].dma_addr = 0; + tx_ring->txbufs[idx].skb = NULL; + tx_ring->txbufs[idx].fidx = -2; + } + + memset(&tx_ring->txds[idx], 0, sizeof(tx_ring->txds[idx])); + + tx_ring->qcp_rd_p++; + tx_ring->rd_p++; + } + + nd_q = netdev_get_tx_queue(nn->netdev, tx_ring->idx); + netdev_tx_reset_queue(nd_q); +} + +static void nfp_net_tx_timeout(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + int i; + + for (i = 0; i < nn->num_tx_rings; i++) { + if (!netif_tx_queue_stopped(netdev_get_tx_queue(netdev, i))) + continue; + nn_warn(nn, "TX timeout on ring: %d\n", i); + } + nn_warn(nn, "TX watchdog timeout\n"); +} + +/* Receive processing + */ + +/** + * nfp_net_rx_space() - return the number of free slots on the RX ring + * @rx_ring: RX ring structure + * + * Make sure we leave at least one slot free. + * + * Return: True if there is space on the RX ring + */ +static inline int nfp_net_rx_space(struct nfp_net_rx_ring *rx_ring) +{ + return (rx_ring->cnt - 1) - (rx_ring->wr_p - rx_ring->rd_p); +} + +/** + * nfp_net_rx_alloc_one() - Allocate and map skb for RX + * @rx_ring: RX ring structure of the skb + * @dma_addr: Pointer to storage for DMA address (output param) + * + * This function will allcate a new skb, map it for DMA. + * + * Return: allocated skb or NULL on failure. + */ +static struct sk_buff * +nfp_net_rx_alloc_one(struct nfp_net_rx_ring *rx_ring, dma_addr_t *dma_addr) +{ + struct nfp_net *nn = rx_ring->r_vec->nfp_net; + struct sk_buff *skb; + + skb = netdev_alloc_skb(nn->netdev, nn->fl_bufsz); + if (!skb) { + nn_warn_ratelimit(nn, "Failed to alloc receive SKB\n"); + return NULL; + } + + *dma_addr = dma_map_single(&nn->pdev->dev, skb->data, + nn->fl_bufsz, DMA_FROM_DEVICE); + if (dma_mapping_error(&nn->pdev->dev, *dma_addr)) { + dev_kfree_skb_any(skb); + nn_warn_ratelimit(nn, "Failed to map DMA RX buffer\n"); + return NULL; + } + + return skb; +} + +/** + * nfp_net_rx_give_one() - Put mapped skb on the software and hardware rings + * @rx_ring: RX ring structure + * @skb: Skb to put on rings + * @dma_addr: DMA address of skb mapping + */ +static void nfp_net_rx_give_one(struct nfp_net_rx_ring *rx_ring, + struct sk_buff *skb, dma_addr_t dma_addr) +{ + unsigned int wr_idx; + + wr_idx = rx_ring->wr_p % rx_ring->cnt; + + /* Stash SKB and DMA address away */ + rx_ring->rxbufs[wr_idx].skb = skb; + rx_ring->rxbufs[wr_idx].dma_addr = dma_addr; + + /* Fill freelist descriptor */ + rx_ring->rxds[wr_idx].fld.reserved = 0; + rx_ring->rxds[wr_idx].fld.meta_len_dd = 0; + nfp_desc_set_dma_addr(&rx_ring->rxds[wr_idx].fld, dma_addr); + + rx_ring->wr_p++; + rx_ring->wr_ptr_add++; + if (rx_ring->wr_ptr_add >= NFP_NET_FL_BATCH) { + /* Update write pointer of the freelist queue. Make + * sure all writes are flushed before telling the hardware. + */ + wmb(); + nfp_qcp_wr_ptr_add(rx_ring->qcp_fl, rx_ring->wr_ptr_add); + rx_ring->wr_ptr_add = 0; + } +} + +/** + * nfp_net_rx_flush() - Free any buffers currently on the RX ring + * @rx_ring: RX ring to remove buffers from + * + * Assumes that the device is stopped + */ +static void nfp_net_rx_flush(struct nfp_net_rx_ring *rx_ring) +{ + struct nfp_net *nn = rx_ring->r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + int idx; + + while (rx_ring->rd_p != rx_ring->wr_p) { + idx = rx_ring->rd_p % rx_ring->cnt; + + if (rx_ring->rxbufs[idx].skb) { + dma_unmap_single(&pdev->dev, + rx_ring->rxbufs[idx].dma_addr, + nn->fl_bufsz, DMA_FROM_DEVICE); + dev_kfree_skb_any(rx_ring->rxbufs[idx].skb); + rx_ring->rxbufs[idx].dma_addr = 0; + rx_ring->rxbufs[idx].skb = NULL; + } + + memset(&rx_ring->rxds[idx], 0, sizeof(rx_ring->rxds[idx])); + + rx_ring->rd_p++; + } +} + +/** + * nfp_net_rx_fill_freelist() - Attempt filling freelist with RX buffers + * @rx_ring: RX ring to fill + * + * Try to fill as many buffers as possible into freelist. Return + * number of buffers added. + * + * Return: Number of freelist buffers added. + */ +static int nfp_net_rx_fill_freelist(struct nfp_net_rx_ring *rx_ring) +{ + struct sk_buff *skb; + dma_addr_t dma_addr; + + while (nfp_net_rx_space(rx_ring)) { + skb = nfp_net_rx_alloc_one(rx_ring, &dma_addr); + if (!skb) { + nfp_net_rx_flush(rx_ring); + return -ENOMEM; + } + nfp_net_rx_give_one(rx_ring, skb, dma_addr); + } + + return 0; +} + +/** + * nfp_net_rx_csum_has_errors() - group check if rxd has any csum errors + * @flags: RX descriptor flags field in CPU byte order + */ +static int nfp_net_rx_csum_has_errors(u16 flags) +{ + u16 csum_all_checked, csum_all_ok; + + csum_all_checked = flags & __PCIE_DESC_RX_CSUM_ALL; + csum_all_ok = flags & __PCIE_DESC_RX_CSUM_ALL_OK; + + return csum_all_checked != (csum_all_ok << PCIE_DESC_RX_CSUM_OK_SHIFT); +} + +/** + * nfp_net_rx_csum() - set SKB checksum field based on RX descriptor flags + * @nn: NFP Net device + * @r_vec: per-ring structure + * @rxd: Pointer to RX descriptor + * @skb: Pointer to SKB + */ +static void nfp_net_rx_csum(struct nfp_net *nn, struct nfp_net_r_vector *r_vec, + struct nfp_net_rx_desc *rxd, struct sk_buff *skb) +{ + skb_checksum_none_assert(skb); + + if (!(nn->netdev->features & NETIF_F_RXCSUM)) + return; + + if (nfp_net_rx_csum_has_errors(le16_to_cpu(rxd->rxd.flags))) { + u64_stats_update_begin(&r_vec->rx_sync); + r_vec->hw_csum_rx_error++; + u64_stats_update_end(&r_vec->rx_sync); + return; + } + + /* Assume that the firmware will never report inner CSUM_OK unless outer + * L4 headers were successfully parsed. FW will always report zero UDP + * checksum as CSUM_OK. + */ + if (rxd->rxd.flags & PCIE_DESC_RX_TCP_CSUM_OK || + rxd->rxd.flags & PCIE_DESC_RX_UDP_CSUM_OK) { + __skb_incr_checksum_unnecessary(skb); + u64_stats_update_begin(&r_vec->rx_sync); + r_vec->hw_csum_rx_ok++; + u64_stats_update_end(&r_vec->rx_sync); + } + + if (rxd->rxd.flags & PCIE_DESC_RX_I_TCP_CSUM_OK || + rxd->rxd.flags & PCIE_DESC_RX_I_UDP_CSUM_OK) { + __skb_incr_checksum_unnecessary(skb); + u64_stats_update_begin(&r_vec->rx_sync); + r_vec->hw_csum_rx_inner_ok++; + u64_stats_update_end(&r_vec->rx_sync); + } +} + +/** + * nfp_net_set_hash() - Set SKB hash data + * @netdev: adapter's net_device structure + * @skb: SKB to set the hash data on + * @rxd: RX descriptor + * + * The RSS hash and hash-type are pre-pended to the packet data. + * Extract and decode it and set the skb fields. + */ +static void nfp_net_set_hash(struct net_device *netdev, struct sk_buff *skb, + struct nfp_net_rx_desc *rxd) +{ + struct nfp_net_rx_hash *rx_hash; + + if (!(rxd->rxd.flags & PCIE_DESC_RX_RSS) || + !(netdev->features & NETIF_F_RXHASH)) + return; + + rx_hash = (struct nfp_net_rx_hash *)(skb->data - sizeof(*rx_hash)); + + switch (be32_to_cpu(rx_hash->hash_type)) { + case NFP_NET_RSS_IPV4: + case NFP_NET_RSS_IPV6: + case NFP_NET_RSS_IPV6_EX: + skb_set_hash(skb, be32_to_cpu(rx_hash->hash), PKT_HASH_TYPE_L3); + break; + default: + skb_set_hash(skb, be32_to_cpu(rx_hash->hash), PKT_HASH_TYPE_L4); + break; + } +} + +/** + * nfp_net_rx() - receive up to @budget packets on @rx_ring + * @rx_ring: RX ring to receive from + * @budget: NAPI budget + * + * Note, this function is separated out from the napi poll function to + * more cleanly separate packet receive code from other bookkeeping + * functions performed in the napi poll function. + * + * There are differences between the NFP-3200 firmware and the + * NFP-6000 firmware. The NFP-3200 firmware uses a dedicated RX queue + * to indicate that new packets have arrived. The NFP-6000 does not + * have this queue and uses the DD bit in the RX descriptor. This + * method cannot be used on the NFP-3200 as it causes a race + * condition: The RX ring write pointer on the NFP-3200 is updated + * after packets (and descriptors) have been DMAed. If the DD bit is + * used and subsequently the read pointer is updated this may lead to + * the RX queue to underflow (if the firmware has not yet update the + * write pointer). Therefore we use slightly ugly conditional code + * below to handle the differences. We may, in the future update the + * NFP-3200 firmware to behave the same as the firmware on the + * NFP-6000. + * + * Return: Number of packets received. + */ +static int nfp_net_rx(struct nfp_net_rx_ring *rx_ring, int budget) +{ + struct nfp_net_r_vector *r_vec = rx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + unsigned int data_len, meta_len; + int avail = 0, pkts_polled = 0; + struct sk_buff *skb, *new_skb; + struct nfp_net_rx_desc *rxd; + dma_addr_t new_dma_addr; + u32 qcp_wr_p; + int idx; + + if (nn->is_nfp3200) { + /* Work out how many packets arrived */ + qcp_wr_p = nfp_qcp_wr_ptr_read(rx_ring->qcp_rx); + idx = rx_ring->rd_p % rx_ring->cnt; + + if (qcp_wr_p == idx) + /* No new packets */ + return 0; + + if (qcp_wr_p > idx) + avail = qcp_wr_p - idx; + else + avail = qcp_wr_p + rx_ring->cnt - idx; + } else { + avail = budget + 1; + } + + while (avail > 0 && pkts_polled < budget) { + idx = rx_ring->rd_p % rx_ring->cnt; + + rxd = &rx_ring->rxds[idx]; + if (!(rxd->rxd.meta_len_dd & PCIE_DESC_RX_DD)) { + if (nn->is_nfp3200) + nn_dbg(nn, "RX descriptor not valid (DD)%d:%u rxd[0]=%#x rxd[1]=%#x\n", + rx_ring->idx, idx, + rxd->vals[0], rxd->vals[1]); + break; + } + /* Memory barrier to ensure that we won't do other reads + * before the DD bit. + */ + dma_rmb(); + + rx_ring->rd_p++; + pkts_polled++; + avail--; + + skb = rx_ring->rxbufs[idx].skb; + + new_skb = nfp_net_rx_alloc_one(rx_ring, &new_dma_addr); + if (!new_skb) { + nfp_net_rx_give_one(rx_ring, rx_ring->rxbufs[idx].skb, + rx_ring->rxbufs[idx].dma_addr); + u64_stats_update_begin(&r_vec->rx_sync); + r_vec->rx_drops++; + u64_stats_update_end(&r_vec->rx_sync); + continue; + } + + dma_unmap_single(&nn->pdev->dev, + rx_ring->rxbufs[idx].dma_addr, + nn->fl_bufsz, DMA_FROM_DEVICE); + + nfp_net_rx_give_one(rx_ring, new_skb, new_dma_addr); + + meta_len = rxd->rxd.meta_len_dd & PCIE_DESC_RX_META_LEN_MASK; + data_len = le16_to_cpu(rxd->rxd.data_len); + + if (WARN_ON_ONCE(data_len > nn->fl_bufsz)) { + dev_kfree_skb_any(skb); + continue; + } + + if (nn->rx_offset == NFP_NET_CFG_RX_OFFSET_DYNAMIC) { + /* The packet data starts after the metadata */ + skb_reserve(skb, meta_len); + } else { + /* The packet data starts at a fixed offset */ + skb_reserve(skb, nn->rx_offset); + } + + /* Adjust the SKB for the dynamic meta data pre-pended */ + skb_put(skb, data_len - meta_len); + + nfp_net_set_hash(nn->netdev, skb, rxd); + + /* Pad small frames to minimum */ + if (skb_put_padto(skb, 60)) + break; + + /* Stats update */ + u64_stats_update_begin(&r_vec->rx_sync); + r_vec->rx_pkts++; + r_vec->rx_bytes += skb->len; + u64_stats_update_end(&r_vec->rx_sync); + + skb_record_rx_queue(skb, rx_ring->idx); + skb->protocol = eth_type_trans(skb, nn->netdev); + + nfp_net_rx_csum(nn, r_vec, rxd, skb); + + if (rxd->rxd.flags & PCIE_DESC_RX_VLAN) + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), + le16_to_cpu(rxd->rxd.vlan)); + + napi_gro_receive(&rx_ring->r_vec->napi, skb); + } + + if (nn->is_nfp3200) + nfp_qcp_rd_ptr_add(rx_ring->qcp_rx, pkts_polled); + + return pkts_polled; +} + +/** + * nfp_net_poll() - napi poll function + * @napi: NAPI structure + * @budget: NAPI budget + * + * Return: number of packets polled. + */ +static int nfp_net_poll(struct napi_struct *napi, int budget) +{ + struct nfp_net_r_vector *r_vec = + container_of(napi, struct nfp_net_r_vector, napi); + struct nfp_net_rx_ring *rx_ring = r_vec->rx_ring; + struct nfp_net_tx_ring *tx_ring = r_vec->tx_ring; + struct nfp_net *nn = r_vec->nfp_net; + struct netdev_queue *txq; + unsigned int pkts_polled; + + tx_ring = &nn->tx_rings[rx_ring->idx]; + txq = netdev_get_tx_queue(nn->netdev, tx_ring->idx); + nfp_net_tx_complete(tx_ring); + + pkts_polled = nfp_net_rx(rx_ring, budget); + + if (pkts_polled < budget) { + napi_complete_done(napi, pkts_polled); + nfp_net_irq_unmask(nn, r_vec->irq_idx); + } + + return pkts_polled; +} + +/* Setup and Configuration + */ + +/** + * nfp_net_tx_ring_free() - Free resources allocated to a TX ring + * @tx_ring: TX ring to free + */ +static void nfp_net_tx_ring_free(struct nfp_net_tx_ring *tx_ring) +{ + struct nfp_net_r_vector *r_vec = tx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + + nn_writeq(nn, NFP_NET_CFG_TXR_ADDR(tx_ring->idx), 0); + nn_writeb(nn, NFP_NET_CFG_TXR_SZ(tx_ring->idx), 0); + nn_writeb(nn, NFP_NET_CFG_TXR_VEC(tx_ring->idx), 0); + + kfree(tx_ring->txbufs); + + if (tx_ring->txds) + dma_free_coherent(&pdev->dev, tx_ring->size, + tx_ring->txds, tx_ring->dma); + + tx_ring->cnt = 0; + tx_ring->wr_p = 0; + tx_ring->rd_p = 0; + tx_ring->qcp_rd_p = 0; + + tx_ring->txbufs = NULL; + tx_ring->txds = NULL; + tx_ring->dma = 0; + tx_ring->size = 0; +} + +/** + * nfp_net_tx_ring_alloc() - Allocate resource for a TX ring + * @tx_ring: TX Ring structure to allocate + * + * Return: 0 on success, negative errno otherwise. + */ +static int nfp_net_tx_ring_alloc(struct nfp_net_tx_ring *tx_ring) +{ + struct nfp_net_r_vector *r_vec = tx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + int sz; + + tx_ring->cnt = nn->txd_cnt; + + tx_ring->size = sizeof(*tx_ring->txds) * tx_ring->cnt; + tx_ring->txds = dma_zalloc_coherent(&pdev->dev, tx_ring->size, + &tx_ring->dma, GFP_KERNEL); + if (!tx_ring->txds) + goto err_alloc; + + sz = sizeof(*tx_ring->txbufs) * tx_ring->cnt; + tx_ring->txbufs = kzalloc(sz, GFP_KERNEL); + if (!tx_ring->txbufs) + goto err_alloc; + + /* Write the DMA address, size and MSI-X info to the device */ + nn_writeq(nn, NFP_NET_CFG_TXR_ADDR(tx_ring->idx), tx_ring->dma); + nn_writeb(nn, NFP_NET_CFG_TXR_SZ(tx_ring->idx), ilog2(tx_ring->cnt)); + nn_writeb(nn, NFP_NET_CFG_TXR_VEC(tx_ring->idx), r_vec->irq_idx); + + netif_set_xps_queue(nn->netdev, &r_vec->affinity_mask, tx_ring->idx); + + nn_dbg(nn, "TxQ%02d: QCidx=%02d cnt=%d dma=%#llx host=%p\n", + tx_ring->idx, tx_ring->qcidx, + tx_ring->cnt, (unsigned long long)tx_ring->dma, tx_ring->txds); + + return 0; + +err_alloc: + nfp_net_tx_ring_free(tx_ring); + return -ENOMEM; +} + +/** + * nfp_net_rx_ring_free() - Free resources allocated to a RX ring + * @rx_ring: RX ring to free + */ +static void nfp_net_rx_ring_free(struct nfp_net_rx_ring *rx_ring) +{ + struct nfp_net_r_vector *r_vec = rx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + + nn_writeq(nn, NFP_NET_CFG_RXR_ADDR(rx_ring->idx), 0); + nn_writeb(nn, NFP_NET_CFG_RXR_SZ(rx_ring->idx), 0); + nn_writeb(nn, NFP_NET_CFG_RXR_VEC(rx_ring->idx), 0); + + kfree(rx_ring->rxbufs); + + if (rx_ring->rxds) + dma_free_coherent(&pdev->dev, rx_ring->size, + rx_ring->rxds, rx_ring->dma); + + rx_ring->cnt = 0; + rx_ring->wr_p = 0; + rx_ring->rd_p = 0; + + rx_ring->rxbufs = NULL; + rx_ring->rxds = NULL; + rx_ring->dma = 0; + rx_ring->size = 0; +} + +/** + * nfp_net_rx_ring_alloc() - Allocate resource for a RX ring + * @rx_ring: RX ring to allocate + * + * Return: 0 on success, negative errno otherwise. + */ +static int nfp_net_rx_ring_alloc(struct nfp_net_rx_ring *rx_ring) +{ + struct nfp_net_r_vector *r_vec = rx_ring->r_vec; + struct nfp_net *nn = r_vec->nfp_net; + struct pci_dev *pdev = nn->pdev; + int sz; + + rx_ring->cnt = nn->rxd_cnt; + + rx_ring->size = sizeof(*rx_ring->rxds) * rx_ring->cnt; + rx_ring->rxds = dma_zalloc_coherent(&pdev->dev, rx_ring->size, + &rx_ring->dma, GFP_KERNEL); + if (!rx_ring->rxds) + goto err_alloc; + + sz = sizeof(*rx_ring->rxbufs) * rx_ring->cnt; + rx_ring->rxbufs = kzalloc(sz, GFP_KERNEL); + if (!rx_ring->rxbufs) + goto err_alloc; + + /* Write the DMA address, size and MSI-X info to the device */ + nn_writeq(nn, NFP_NET_CFG_RXR_ADDR(rx_ring->idx), rx_ring->dma); + nn_writeb(nn, NFP_NET_CFG_RXR_SZ(rx_ring->idx), ilog2(rx_ring->cnt)); + nn_writeb(nn, NFP_NET_CFG_RXR_VEC(rx_ring->idx), r_vec->irq_idx); + + nn_dbg(nn, "RxQ%02d: FlQCidx=%02d RxQCidx=%02d cnt=%d dma=%#llx host=%p\n", + rx_ring->idx, rx_ring->fl_qcidx, rx_ring->rx_qcidx, + rx_ring->cnt, (unsigned long long)rx_ring->dma, rx_ring->rxds); + + return 0; + +err_alloc: + nfp_net_rx_ring_free(rx_ring); + return -ENOMEM; +} + +static void __nfp_net_free_rings(struct nfp_net *nn, unsigned int n_free) +{ + struct nfp_net_r_vector *r_vec; + struct msix_entry *entry; + + while (n_free--) { + r_vec = &nn->r_vecs[n_free]; + entry = &nn->irq_entries[r_vec->irq_idx]; + + nfp_net_rx_ring_free(r_vec->rx_ring); + nfp_net_tx_ring_free(r_vec->tx_ring); + + irq_set_affinity_hint(entry->vector, NULL); + free_irq(entry->vector, r_vec); + + netif_napi_del(&r_vec->napi); + } +} + +/** + * nfp_net_free_rings() - Free all ring resources + * @nn: NFP Net device to reconfigure + */ +static void nfp_net_free_rings(struct nfp_net *nn) +{ + __nfp_net_free_rings(nn, nn->num_r_vecs); +} + +/** + * nfp_net_alloc_rings() - Allocate resources for RX and TX rings + * @nn: NFP Net device to reconfigure + * + * Return: 0 on success or negative errno on error. + */ +static int nfp_net_alloc_rings(struct nfp_net *nn) +{ + struct nfp_net_r_vector *r_vec; + struct msix_entry *entry; + int err; + int r; + + for (r = 0; r < nn->num_r_vecs; r++) { + r_vec = &nn->r_vecs[r]; + entry = &nn->irq_entries[r_vec->irq_idx]; + + /* Setup NAPI */ + netif_napi_add(nn->netdev, &r_vec->napi, + nfp_net_poll, NAPI_POLL_WEIGHT); + + snprintf(r_vec->name, sizeof(r_vec->name), + "%s-rxtx-%d", nn->netdev->name, r); + err = request_irq(entry->vector, r_vec->handler, 0, + r_vec->name, r_vec); + if (err) { + nn_dbg(nn, "Error requesting IRQ %d\n", entry->vector); + goto err_napi_del; + } + + irq_set_affinity_hint(entry->vector, &r_vec->affinity_mask); + + nn_dbg(nn, "RV%02d: irq=%03d/%03d\n", + r, entry->vector, entry->entry); + + /* Allocate TX ring resources */ + err = nfp_net_tx_ring_alloc(r_vec->tx_ring); + if (err) + goto err_free_irq; + + /* Allocate RX ring resources */ + err = nfp_net_rx_ring_alloc(r_vec->rx_ring); + if (err) + goto err_free_tx; + } + + return 0; + +err_free_tx: + nfp_net_tx_ring_free(r_vec->tx_ring); +err_free_irq: + irq_set_affinity_hint(entry->vector, NULL); + free_irq(entry->vector, r_vec); +err_napi_del: + netif_napi_del(&r_vec->napi); + __nfp_net_free_rings(nn, r); + return err; +} + +/** + * nfp_net_rss_write_itbl() - Write RSS indirection table to device + * @nn: NFP Net device to reconfigure + */ +void nfp_net_rss_write_itbl(struct nfp_net *nn) +{ + int i; + + for (i = 0; i < NFP_NET_CFG_RSS_ITBL_SZ; i += 4) + nn_writel(nn, NFP_NET_CFG_RSS_ITBL + i, + get_unaligned_le32(nn->rss_itbl + i)); +} + +/** + * nfp_net_rss_write_key() - Write RSS hash key to device + * @nn: NFP Net device to reconfigure + */ +void nfp_net_rss_write_key(struct nfp_net *nn) +{ + int i; + + for (i = 0; i < NFP_NET_CFG_RSS_KEY_SZ; i += 4) + nn_writel(nn, NFP_NET_CFG_RSS_KEY + i, + get_unaligned_le32(nn->rss_key + i)); +} + +/** + * nfp_net_coalesce_write_cfg() - Write irq coalescence configuration to HW + * @nn: NFP Net device to reconfigure + */ +void nfp_net_coalesce_write_cfg(struct nfp_net *nn) +{ + u8 i; + u32 factor; + u32 value; + + /* Compute factor used to convert coalesce '_usecs' parameters to + * ME timestamp ticks. There are 16 ME clock cycles for each timestamp + * count. + */ + factor = nn->me_freq_mhz / 16; + + /* copy RX interrupt coalesce parameters */ + value = (nn->rx_coalesce_max_frames << 16) | + (factor * nn->rx_coalesce_usecs); + for (i = 0; i < nn->num_r_vecs; i++) + nn_writel(nn, NFP_NET_CFG_RXR_IRQ_MOD(i), value); + + /* copy TX interrupt coalesce parameters */ + value = (nn->tx_coalesce_max_frames << 16) | + (factor * nn->tx_coalesce_usecs); + for (i = 0; i < nn->num_r_vecs; i++) + nn_writel(nn, NFP_NET_CFG_TXR_IRQ_MOD(i), value); +} + +/** + * nfp_net_write_mac_addr() - Write mac address to device registers + * @nn: NFP Net device to reconfigure + * @mac: Six-byte MAC address to be written + * + * We do a bit of byte swapping dance because firmware is LE. + */ +static void nfp_net_write_mac_addr(struct nfp_net *nn, const u8 *mac) +{ + nn_writel(nn, NFP_NET_CFG_MACADDR + 0, + get_unaligned_be32(nn->netdev->dev_addr)); + /* We can't do writew for NFP-3200 compatibility */ + nn_writel(nn, NFP_NET_CFG_MACADDR + 4, + get_unaligned_be16(nn->netdev->dev_addr + 4) << 16); +} + +/** + * nfp_net_clear_config_and_disable() - Clear control BAR and disable NFP + * @nn: NFP Net device to reconfigure + */ +static void nfp_net_clear_config_and_disable(struct nfp_net *nn) +{ + u32 new_ctrl, update; + int err; + + new_ctrl = nn->ctrl; + new_ctrl &= ~NFP_NET_CFG_CTRL_ENABLE; + update = NFP_NET_CFG_UPDATE_GEN; + update |= NFP_NET_CFG_UPDATE_MSIX; + update |= NFP_NET_CFG_UPDATE_RING; + + if (nn->cap & NFP_NET_CFG_CTRL_RINGCFG) + new_ctrl &= ~NFP_NET_CFG_CTRL_RINGCFG; + + nn_writeq(nn, NFP_NET_CFG_TXRS_ENABLE, 0); + nn_writeq(nn, NFP_NET_CFG_RXRS_ENABLE, 0); + + nn_writel(nn, NFP_NET_CFG_CTRL, new_ctrl); + err = nfp_net_reconfig(nn, update); + if (err) { + nn_err(nn, "Could not disable device: %d\n", err); + return; + } + + nn->ctrl = new_ctrl; +} + +/** + * nfp_net_start_vec() - Start ring vector + * @nn: NFP Net device structure + * @r_vec: Ring vector to be started + */ +static int nfp_net_start_vec(struct nfp_net *nn, struct nfp_net_r_vector *r_vec) +{ + unsigned int irq_vec; + int err = 0; + + irq_vec = nn->irq_entries[r_vec->irq_idx].vector; + + disable_irq(irq_vec); + + err = nfp_net_rx_fill_freelist(r_vec->rx_ring); + if (err) { + nn_err(nn, "RV%02d: couldn't allocate enough buffers\n", + r_vec->irq_idx); + goto out; + } + + napi_enable(&r_vec->napi); +out: + enable_irq(irq_vec); + + return err; +} + +static int nfp_net_netdev_open(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + int err, r; + u32 update = 0; + u32 new_ctrl; + + if (nn->ctrl & NFP_NET_CFG_CTRL_ENABLE) { + nn_err(nn, "Dev is already enabled: 0x%08x\n", nn->ctrl); + return -EBUSY; + } + + new_ctrl = nn->ctrl; + + /* Step 1: Allocate resources for rings and the like + * - Request interrupts + * - Allocate RX and TX ring resources + * - Setup initial RSS table + */ + err = nfp_net_aux_irq_request(nn, NFP_NET_CFG_EXN, "%s-exn", + nn->exn_name, sizeof(nn->exn_name), + NFP_NET_IRQ_EXN_IDX, nn->exn_handler); + if (err) + return err; + + err = nfp_net_alloc_rings(nn); + if (err) + goto err_free_exn; + + err = netif_set_real_num_tx_queues(netdev, nn->num_tx_rings); + if (err) + goto err_free_rings; + + err = netif_set_real_num_rx_queues(netdev, nn->num_rx_rings); + if (err) + goto err_free_rings; + + if (nn->cap & NFP_NET_CFG_CTRL_RSS) { + nfp_net_rss_write_key(nn); + nfp_net_rss_write_itbl(nn); + nn_writel(nn, NFP_NET_CFG_RSS_CTRL, nn->rss_cfg); + update |= NFP_NET_CFG_UPDATE_RSS; + } + + if (nn->cap & NFP_NET_CFG_CTRL_IRQMOD) { + nfp_net_coalesce_write_cfg(nn); + + new_ctrl |= NFP_NET_CFG_CTRL_IRQMOD; + update |= NFP_NET_CFG_UPDATE_IRQMOD; + } + + /* Step 2: Configure the NFP + * - Enable rings from 0 to tx_rings/rx_rings - 1. + * - Write MAC address (in case it changed) + * - Set the MTU + * - Set the Freelist buffer size + * - Enable the FW + */ + nn_writeq(nn, NFP_NET_CFG_TXRS_ENABLE, nn->num_tx_rings == 64 ? + 0xffffffffffffffffULL : ((u64)1 << nn->num_tx_rings) - 1); + + nn_writeq(nn, NFP_NET_CFG_RXRS_ENABLE, nn->num_rx_rings == 64 ? + 0xffffffffffffffffULL : ((u64)1 << nn->num_rx_rings) - 1); + + nfp_net_write_mac_addr(nn, netdev->dev_addr); + + nn_writel(nn, NFP_NET_CFG_MTU, netdev->mtu); + nn_writel(nn, NFP_NET_CFG_FLBUFSZ, nn->fl_bufsz); + + /* Enable device */ + new_ctrl |= NFP_NET_CFG_CTRL_ENABLE; + update |= NFP_NET_CFG_UPDATE_GEN; + update |= NFP_NET_CFG_UPDATE_MSIX; + update |= NFP_NET_CFG_UPDATE_RING; + if (nn->cap & NFP_NET_CFG_CTRL_RINGCFG) + new_ctrl |= NFP_NET_CFG_CTRL_RINGCFG; + + nn_writel(nn, NFP_NET_CFG_CTRL, new_ctrl); + err = nfp_net_reconfig(nn, update); + if (err) + goto err_clear_config; + + nn->ctrl = new_ctrl; + + /* Since reconfiguration requests while NFP is down are ignored we + * have to wipe the entire VXLAN configuration and reinitialize it. + */ + if (nn->ctrl & NFP_NET_CFG_CTRL_VXLAN) { + memset(&nn->vxlan_ports, 0, sizeof(nn->vxlan_ports)); + memset(&nn->vxlan_usecnt, 0, sizeof(nn->vxlan_usecnt)); + vxlan_get_rx_port(netdev); + } + + /* Step 3: Enable for kernel + * - put some freelist descriptors on each RX ring + * - enable NAPI on each ring + * - enable all TX queues + * - set link state + */ + for (r = 0; r < nn->num_r_vecs; r++) { + err = nfp_net_start_vec(nn, &nn->r_vecs[r]); + if (err) + goto err_disable_napi; + } + + netif_tx_wake_all_queues(netdev); + + err = nfp_net_aux_irq_request(nn, NFP_NET_CFG_LSC, "%s-lsc", + nn->lsc_name, sizeof(nn->lsc_name), + NFP_NET_IRQ_LSC_IDX, nn->lsc_handler); + if (err) + goto err_stop_tx; + nfp_net_read_link_status(nn); + + return 0; + +err_stop_tx: + netif_tx_disable(netdev); + for (r = 0; r < nn->num_r_vecs; r++) + nfp_net_tx_flush(nn->r_vecs[r].tx_ring); +err_disable_napi: + while (r--) { + napi_disable(&nn->r_vecs[r].napi); + nfp_net_rx_flush(nn->r_vecs[r].rx_ring); + } +err_clear_config: + nfp_net_clear_config_and_disable(nn); +err_free_rings: + nfp_net_free_rings(nn); +err_free_exn: + nfp_net_aux_irq_free(nn, NFP_NET_CFG_EXN, NFP_NET_IRQ_EXN_IDX); + return err; +} + +/** + * nfp_net_netdev_close() - Called when the device is downed + * @netdev: netdev structure + */ +static int nfp_net_netdev_close(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + int r; + + if (!(nn->ctrl & NFP_NET_CFG_CTRL_ENABLE)) { + nn_err(nn, "Dev is not up: 0x%08x\n", nn->ctrl); + return 0; + } + + /* Step 1: Disable RX and TX rings from the Linux kernel perspective + */ + nfp_net_aux_irq_free(nn, NFP_NET_CFG_LSC, NFP_NET_IRQ_LSC_IDX); + netif_carrier_off(netdev); + nn->link_up = false; + + for (r = 0; r < nn->num_r_vecs; r++) + napi_disable(&nn->r_vecs[r].napi); + + netif_tx_disable(netdev); + + /* Step 2: Tell NFP + */ + nfp_net_clear_config_and_disable(nn); + + /* Step 3: Free resources + */ + for (r = 0; r < nn->num_r_vecs; r++) { + nfp_net_rx_flush(nn->r_vecs[r].rx_ring); + nfp_net_tx_flush(nn->r_vecs[r].tx_ring); + } + + nfp_net_free_rings(nn); + nfp_net_aux_irq_free(nn, NFP_NET_CFG_EXN, NFP_NET_IRQ_EXN_IDX); + + nn_dbg(nn, "%s down", netdev->name); + return 0; +} + +static void nfp_net_set_rx_mode(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + u32 new_ctrl; + + new_ctrl = nn->ctrl; + + if (netdev->flags & IFF_PROMISC) { + if (nn->cap & NFP_NET_CFG_CTRL_PROMISC) + new_ctrl |= NFP_NET_CFG_CTRL_PROMISC; + else + nn_warn(nn, "FW does not support promiscuous mode\n"); + } else { + new_ctrl &= ~NFP_NET_CFG_CTRL_PROMISC; + } + + if (new_ctrl == nn->ctrl) + return; + + nn_writel(nn, NFP_NET_CFG_CTRL, new_ctrl); + if (nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_GEN)) + return; + + nn->ctrl = new_ctrl; +} + +static int nfp_net_change_mtu(struct net_device *netdev, int new_mtu) +{ + struct nfp_net *nn = netdev_priv(netdev); + u32 tmp; + + nn_dbg(nn, "New MTU = %d\n", new_mtu); + + if (new_mtu < 68 || new_mtu > nn->max_mtu) { + nn_err(nn, "New MTU (%d) is not valid\n", new_mtu); + return -EINVAL; + } + + netdev->mtu = new_mtu; + + /* Freelist buffer size rounded up to the nearest 1K */ + tmp = new_mtu + ETH_HLEN + VLAN_HLEN + NFP_NET_MAX_PREPEND; + nn->fl_bufsz = roundup(tmp, 1024); + + /* restart if running */ + if (netif_running(netdev)) { + nfp_net_netdev_close(netdev); + nfp_net_netdev_open(netdev); + } + + return 0; +} + +static struct rtnl_link_stats64 *nfp_net_stat64(struct net_device *netdev, + struct rtnl_link_stats64 *stats) +{ + struct nfp_net *nn = netdev_priv(netdev); + int r; + + for (r = 0; r < nn->num_r_vecs; r++) { + struct nfp_net_r_vector *r_vec = &nn->r_vecs[r]; + u64 data[3]; + unsigned int start; + + do { + start = u64_stats_fetch_begin(&r_vec->rx_sync); + data[0] = r_vec->rx_pkts; + data[1] = r_vec->rx_bytes; + data[2] = r_vec->rx_drops; + } while (u64_stats_fetch_retry(&r_vec->rx_sync, start)); + stats->rx_packets += data[0]; + stats->rx_bytes += data[1]; + stats->rx_dropped += data[2]; + + do { + start = u64_stats_fetch_begin(&r_vec->tx_sync); + data[0] = r_vec->tx_pkts; + data[1] = r_vec->tx_bytes; + data[2] = r_vec->tx_errors; + } while (u64_stats_fetch_retry(&r_vec->tx_sync, start)); + stats->tx_packets += data[0]; + stats->tx_bytes += data[1]; + stats->tx_errors += data[2]; + } + + return stats; +} + +static int nfp_net_set_features(struct net_device *netdev, + netdev_features_t features) +{ + netdev_features_t changed = netdev->features ^ features; + struct nfp_net *nn = netdev_priv(netdev); + u32 new_ctrl; + int err; + + /* Assume this is not called with features we have not advertised */ + + new_ctrl = nn->ctrl; + + if (changed & NETIF_F_RXCSUM) { + if (features & NETIF_F_RXCSUM) + new_ctrl |= NFP_NET_CFG_CTRL_RXCSUM; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_RXCSUM; + } + + if (changed & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)) { + if (features & (NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM)) + new_ctrl |= NFP_NET_CFG_CTRL_TXCSUM; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_TXCSUM; + } + + if (changed & (NETIF_F_TSO | NETIF_F_TSO6)) { + if (features & (NETIF_F_TSO | NETIF_F_TSO6)) + new_ctrl |= NFP_NET_CFG_CTRL_LSO; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_LSO; + } + + if (changed & NETIF_F_HW_VLAN_CTAG_RX) { + if (features & NETIF_F_HW_VLAN_CTAG_RX) + new_ctrl |= NFP_NET_CFG_CTRL_RXVLAN; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_RXVLAN; + } + + if (changed & NETIF_F_HW_VLAN_CTAG_TX) { + if (features & NETIF_F_HW_VLAN_CTAG_TX) + new_ctrl |= NFP_NET_CFG_CTRL_TXVLAN; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_TXVLAN; + } + + if (changed & NETIF_F_SG) { + if (features & NETIF_F_SG) + new_ctrl |= NFP_NET_CFG_CTRL_GATHER; + else + new_ctrl &= ~NFP_NET_CFG_CTRL_GATHER; + } + + nn_dbg(nn, "Feature change 0x%llx -> 0x%llx (changed=0x%llx)\n", + netdev->features, features, changed); + + if (new_ctrl == nn->ctrl) + return 0; + + nn_dbg(nn, "NIC ctrl: 0x%x -> 0x%x\n", nn->ctrl, new_ctrl); + nn_writel(nn, NFP_NET_CFG_CTRL, new_ctrl); + err = nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_GEN); + if (err) + return err; + + nn->ctrl = new_ctrl; + + return 0; +} + +static netdev_features_t +nfp_net_features_check(struct sk_buff *skb, struct net_device *dev, + netdev_features_t features) +{ + u8 l4_hdr; + + /* We can't do TSO over double tagged packets (802.1AD) */ + features &= vlan_features_check(skb, features); + + if (!skb->encapsulation) + return features; + + /* Ensure that inner L4 header offset fits into TX descriptor field */ + if (skb_is_gso(skb)) { + u32 hdrlen; + + hdrlen = skb_inner_transport_header(skb) - skb->data + + inner_tcp_hdrlen(skb); + + if (unlikely(hdrlen > NFP_NET_LSO_MAX_HDR_SZ)) + features &= ~NETIF_F_GSO_MASK; + } + + /* VXLAN/GRE check */ + switch (vlan_get_protocol(skb)) { + case htons(ETH_P_IP): + l4_hdr = ip_hdr(skb)->protocol; + break; + case htons(ETH_P_IPV6): + l4_hdr = ipv6_hdr(skb)->nexthdr; + break; + default: + return features & ~(NETIF_F_ALL_CSUM | NETIF_F_GSO_MASK); + } + + if (skb->inner_protocol_type != ENCAP_TYPE_ETHER || + skb->inner_protocol != htons(ETH_P_TEB) || + (l4_hdr != IPPROTO_UDP && l4_hdr != IPPROTO_GRE) || + (l4_hdr == IPPROTO_UDP && + (skb_inner_mac_header(skb) - skb_transport_header(skb) != + sizeof(struct udphdr) + sizeof(struct vxlanhdr)))) + return features & ~(NETIF_F_ALL_CSUM | NETIF_F_GSO_MASK); + + return features; +} + +/** + * nfp_net_set_vxlan_port() - set vxlan port in SW and reconfigure HW + * @nn: NFP Net device to reconfigure + * @idx: Index into the port table where new port should be written + * @port: UDP port to configure (pass zero to remove VXLAN port) + */ +static void nfp_net_set_vxlan_port(struct nfp_net *nn, int idx, __be16 port) +{ + int i; + + nn->vxlan_ports[idx] = port; + + if (!(nn->ctrl & NFP_NET_CFG_CTRL_VXLAN)) + return; + + BUILD_BUG_ON(NFP_NET_N_VXLAN_PORTS & 1); + for (i = 0; i < NFP_NET_N_VXLAN_PORTS; i += 2) + nn_writel(nn, NFP_NET_CFG_VXLAN_PORT + i * sizeof(port), + be16_to_cpu(nn->vxlan_ports[i + 1]) << 16 | + be16_to_cpu(nn->vxlan_ports[i])); + + nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_VXLAN); +} + +/** + * nfp_net_find_vxlan_idx() - find table entry of the port or a free one + * @nn: NFP Network structure + * @port: UDP port to look for + * + * Return: if the port is already in the table -- it's position; + * if the port is not in the table -- free position to use; + * if the table is full -- -ENOSPC. + */ +static int nfp_net_find_vxlan_idx(struct nfp_net *nn, __be16 port) +{ + int i, free_idx = -ENOSPC; + + for (i = 0; i < NFP_NET_N_VXLAN_PORTS; i++) { + if (nn->vxlan_ports[i] == port) + return i; + if (!nn->vxlan_usecnt[i]) + free_idx = i; + } + + return free_idx; +} + +static void nfp_net_add_vxlan_port(struct net_device *netdev, + sa_family_t sa_family, __be16 port) +{ + struct nfp_net *nn = netdev_priv(netdev); + int idx; + + idx = nfp_net_find_vxlan_idx(nn, port); + if (idx == -ENOSPC) + return; + + if (!nn->vxlan_usecnt[idx]++) + nfp_net_set_vxlan_port(nn, idx, port); +} + +static void nfp_net_del_vxlan_port(struct net_device *netdev, + sa_family_t sa_family, __be16 port) +{ + struct nfp_net *nn = netdev_priv(netdev); + int idx; + + idx = nfp_net_find_vxlan_idx(nn, port); + if (!nn->vxlan_usecnt[idx] || idx == -ENOSPC) + return; + + if (!--nn->vxlan_usecnt[idx]) + nfp_net_set_vxlan_port(nn, idx, 0); +} + +static const struct net_device_ops nfp_net_netdev_ops = { + .ndo_open = nfp_net_netdev_open, + .ndo_stop = nfp_net_netdev_close, + .ndo_start_xmit = nfp_net_tx, + .ndo_get_stats64 = nfp_net_stat64, + .ndo_tx_timeout = nfp_net_tx_timeout, + .ndo_set_rx_mode = nfp_net_set_rx_mode, + .ndo_change_mtu = nfp_net_change_mtu, + .ndo_set_mac_address = eth_mac_addr, + .ndo_set_features = nfp_net_set_features, + .ndo_features_check = nfp_net_features_check, + .ndo_add_vxlan_port = nfp_net_add_vxlan_port, + .ndo_del_vxlan_port = nfp_net_del_vxlan_port, +}; + +/** + * nfp_net_info() - Print general info about the NIC + * @nn: NFP Net device to reconfigure + */ +void nfp_net_info(struct nfp_net *nn) +{ + nn_info(nn, "Netronome %s %sNetdev: TxQs=%d/%d RxQs=%d/%d\n", + nn->is_nfp3200 ? "NFP-32xx" : "NFP-6xxx", + nn->is_vf ? "VF " : "", + nn->num_tx_rings, nn->max_tx_rings, + nn->num_rx_rings, nn->max_rx_rings); + nn_info(nn, "VER: %d.%d.%d.%d, Maximum supported MTU: %d\n", + nn->fw_ver.resv, nn->fw_ver.class, + nn->fw_ver.major, nn->fw_ver.minor, + nn->max_mtu); + nn_info(nn, "CAP: %#x %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s\n", + nn->cap, + nn->cap & NFP_NET_CFG_CTRL_PROMISC ? "PROMISC " : "", + nn->cap & NFP_NET_CFG_CTRL_L2BC ? "L2BCFILT " : "", + nn->cap & NFP_NET_CFG_CTRL_L2MC ? "L2MCFILT " : "", + nn->cap & NFP_NET_CFG_CTRL_RXCSUM ? "RXCSUM " : "", + nn->cap & NFP_NET_CFG_CTRL_TXCSUM ? "TXCSUM " : "", + nn->cap & NFP_NET_CFG_CTRL_RXVLAN ? "RXVLAN " : "", + nn->cap & NFP_NET_CFG_CTRL_TXVLAN ? "TXVLAN " : "", + nn->cap & NFP_NET_CFG_CTRL_SCATTER ? "SCATTER " : "", + nn->cap & NFP_NET_CFG_CTRL_GATHER ? "GATHER " : "", + nn->cap & NFP_NET_CFG_CTRL_LSO ? "TSO " : "", + nn->cap & NFP_NET_CFG_CTRL_RSS ? "RSS " : "", + nn->cap & NFP_NET_CFG_CTRL_L2SWITCH ? "L2SWITCH " : "", + nn->cap & NFP_NET_CFG_CTRL_MSIXAUTO ? "AUTOMASK " : "", + nn->cap & NFP_NET_CFG_CTRL_IRQMOD ? "IRQMOD " : "", + nn->cap & NFP_NET_CFG_CTRL_VXLAN ? "VXLAN " : "", + nn->cap & NFP_NET_CFG_CTRL_NVGRE ? "NVGRE " : ""); +} + +/** + * nfp_net_netdev_alloc() - Allocate netdev and related structure + * @pdev: PCI device + * @max_tx_rings: Maximum number of TX rings supported by device + * @max_rx_rings: Maximum number of RX rings supported by device + * + * This function allocates a netdev device and fills in the initial + * part of the @struct nfp_net structure. + * + * Return: NFP Net device structure, or ERR_PTR on error. + */ +struct nfp_net *nfp_net_netdev_alloc(struct pci_dev *pdev, + int max_tx_rings, int max_rx_rings) +{ + struct net_device *netdev; + struct nfp_net *nn; + int nqs; + + netdev = alloc_etherdev_mqs(sizeof(struct nfp_net), + max_tx_rings, max_rx_rings); + if (!netdev) + return ERR_PTR(-ENOMEM); + + SET_NETDEV_DEV(netdev, &pdev->dev); + nn = netdev_priv(netdev); + + nn->netdev = netdev; + nn->pdev = pdev; + + nn->max_tx_rings = max_tx_rings; + nn->max_rx_rings = max_rx_rings; + + nqs = netif_get_num_default_rss_queues(); + nn->num_tx_rings = min_t(int, nqs, max_tx_rings); + nn->num_rx_rings = min_t(int, nqs, max_rx_rings); + + nn->txd_cnt = NFP_NET_TX_DESCS_DEFAULT; + nn->rxd_cnt = NFP_NET_RX_DESCS_DEFAULT; + + spin_lock_init(&nn->reconfig_lock); + spin_lock_init(&nn->link_status_lock); + + return nn; +} + +/** + * nfp_net_netdev_free() - Undo what @nfp_net_netdev_alloc() did + * @nn: NFP Net device to reconfigure + */ +void nfp_net_netdev_free(struct nfp_net *nn) +{ + free_netdev(nn->netdev); +} + +/** + * nfp_net_rss_init() - Set the initial RSS parameters + * @nn: NFP Net device to reconfigure + */ +static void nfp_net_rss_init(struct nfp_net *nn) +{ + int i; + + netdev_rss_key_fill(nn->rss_key, NFP_NET_CFG_RSS_KEY_SZ); + + for (i = 0; i < sizeof(nn->rss_itbl); i++) + nn->rss_itbl[i] = + ethtool_rxfh_indir_default(i, nn->num_rx_rings); + + /* Enable IPv4/IPv6 TCP by default */ + nn->rss_cfg = NFP_NET_CFG_RSS_IPV4_TCP | + NFP_NET_CFG_RSS_IPV6_TCP | + NFP_NET_CFG_RSS_TOEPLITZ | + NFP_NET_CFG_RSS_MASK; +} + +/** + * nfp_net_irqmod_init() - Set the initial IRQ moderation parameters + * @nn: NFP Net device to reconfigure + */ +static void nfp_net_irqmod_init(struct nfp_net *nn) +{ + nn->rx_coalesce_usecs = 50; + nn->rx_coalesce_max_frames = 64; + nn->tx_coalesce_usecs = 50; + nn->tx_coalesce_max_frames = 64; +} + +/** + * nfp_net_netdev_init() - Initialise/finalise the netdev structure + * @netdev: netdev structure + * + * Return: 0 on success or negative errno on error. + */ +int nfp_net_netdev_init(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + int err; + + /* Get some of the read-only fields from the BAR */ + nn->cap = nn_readl(nn, NFP_NET_CFG_CAP); + nn->max_mtu = nn_readl(nn, NFP_NET_CFG_MAX_MTU); + + nfp_net_write_mac_addr(nn, nn->netdev->dev_addr); + + /* Set default MTU and Freelist buffer size */ + if (nn->max_mtu < NFP_NET_DEFAULT_MTU) + netdev->mtu = nn->max_mtu; + else + netdev->mtu = NFP_NET_DEFAULT_MTU; + nn->fl_bufsz = NFP_NET_DEFAULT_RX_BUFSZ; + + /* Advertise/enable offloads based on capabilities + * + * Note: netdev->features show the currently enabled features + * and netdev->hw_features advertises which features are + * supported. By default we enable most features. + */ + netdev->hw_features = NETIF_F_HIGHDMA; + if (nn->cap & NFP_NET_CFG_CTRL_RXCSUM) { + netdev->hw_features |= NETIF_F_RXCSUM; + nn->ctrl |= NFP_NET_CFG_CTRL_RXCSUM; + } + if (nn->cap & NFP_NET_CFG_CTRL_TXCSUM) { + netdev->hw_features |= NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM; + nn->ctrl |= NFP_NET_CFG_CTRL_TXCSUM; + } + if (nn->cap & NFP_NET_CFG_CTRL_GATHER) { + netdev->hw_features |= NETIF_F_SG; + nn->ctrl |= NFP_NET_CFG_CTRL_GATHER; + } + if ((nn->cap & NFP_NET_CFG_CTRL_LSO) && nn->fw_ver.major > 2) { + netdev->hw_features |= NETIF_F_TSO | NETIF_F_TSO6; + nn->ctrl |= NFP_NET_CFG_CTRL_LSO; + } + if (nn->cap & NFP_NET_CFG_CTRL_RSS) { + netdev->hw_features |= NETIF_F_RXHASH; + nfp_net_rss_init(nn); + nn->ctrl |= NFP_NET_CFG_CTRL_RSS; + } + if (nn->cap & NFP_NET_CFG_CTRL_VXLAN && + nn->cap & NFP_NET_CFG_CTRL_NVGRE) { + if (nn->cap & NFP_NET_CFG_CTRL_LSO) + netdev->hw_features |= NETIF_F_GSO_GRE | + NETIF_F_GSO_UDP_TUNNEL; + nn->ctrl |= NFP_NET_CFG_CTRL_VXLAN | NFP_NET_CFG_CTRL_NVGRE; + + netdev->hw_enc_features = netdev->hw_features; + } + + netdev->vlan_features = netdev->hw_features; + + if (nn->cap & NFP_NET_CFG_CTRL_RXVLAN) { + netdev->hw_features |= NETIF_F_HW_VLAN_CTAG_RX; + nn->ctrl |= NFP_NET_CFG_CTRL_RXVLAN; + } + if (nn->cap & NFP_NET_CFG_CTRL_TXVLAN) { + netdev->hw_features |= NETIF_F_HW_VLAN_CTAG_TX; + nn->ctrl |= NFP_NET_CFG_CTRL_TXVLAN; + } + + netdev->features = netdev->hw_features; + + /* Advertise but disable TSO by default. */ + netdev->features &= ~(NETIF_F_TSO | NETIF_F_TSO6); + + /* Allow L2 Broadcast and Multicast through by default, if supported */ + if (nn->cap & NFP_NET_CFG_CTRL_L2BC) + nn->ctrl |= NFP_NET_CFG_CTRL_L2BC; + if (nn->cap & NFP_NET_CFG_CTRL_L2MC) + nn->ctrl |= NFP_NET_CFG_CTRL_L2MC; + + /* Allow IRQ moderation, if supported */ + if (nn->cap & NFP_NET_CFG_CTRL_IRQMOD) { + nfp_net_irqmod_init(nn); + nn->ctrl |= NFP_NET_CFG_CTRL_IRQMOD; + } + + /* On NFP-3200 enable MSI-X auto-masking, if supported and the + * interrupts are not shared. + */ + if (nn->is_nfp3200 && nn->cap & NFP_NET_CFG_CTRL_MSIXAUTO) + nn->ctrl |= NFP_NET_CFG_CTRL_MSIXAUTO; + + /* On NFP4000/NFP6000, determine RX packet/metadata boundary offset */ + if (nn->fw_ver.major >= 2) + nn->rx_offset = nn_readl(nn, NFP_NET_CFG_RX_OFFSET); + else + nn->rx_offset = NFP_NET_RX_OFFSET; + + /* Stash the re-configuration queue away. First odd queue in TX Bar */ + nn->qcp_cfg = nn->tx_bar + NFP_QCP_QUEUE_ADDR_SZ; + + /* Make sure the FW knows the netdev is supposed to be disabled here */ + nn_writel(nn, NFP_NET_CFG_CTRL, 0); + nn_writeq(nn, NFP_NET_CFG_TXRS_ENABLE, 0); + nn_writeq(nn, NFP_NET_CFG_RXRS_ENABLE, 0); + err = nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_RING | + NFP_NET_CFG_UPDATE_GEN); + if (err) + return err; + + /* Finalise the netdev setup */ + ether_setup(netdev); + netdev->netdev_ops = &nfp_net_netdev_ops; + netdev->watchdog_timeo = msecs_to_jiffies(5 * 1000); + + nfp_net_set_ethtool_ops(netdev); + nfp_net_irqs_assign(netdev); + + return register_netdev(netdev); +} + +/** + * nfp_net_netdev_clean() - Undo what nfp_net_netdev_init() did. + * @netdev: netdev structure + */ +void nfp_net_netdev_clean(struct net_device *netdev) +{ + unregister_netdev(netdev); +} diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h new file mode 100644 index 000000000000..8692003aeed8 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h @@ -0,0 +1,323 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* + * nfp_net_ctrl.h + * Netronome network device driver: Control BAR layout + * Authors: Jakub Kicinski + * Jason McMullan + * Rolf Neugebauer + * Brad Petrus + */ + +#ifndef _NFP_NET_CTRL_H_ +#define _NFP_NET_CTRL_H_ + +/* IMPORTANT: This header file is shared with the FW, + * no OS specific constructs, please! + */ + +/** + * Configuration BAR size. + * + * The configuration BAR is 8K in size, but on the NFP6000, due to + * THB-350, 32k needs to be reserved. + */ +#define NFP_NET_CFG_BAR_SZ (32 * 1024) + +/** + * Offset in Freelist buffer where packet starts on RX + */ +#define NFP_NET_RX_OFFSET 32 + +/** + * Maximum header size supported for LSO frames + */ +#define NFP_NET_LSO_MAX_HDR_SZ 255 + +/** + * Hash type pre-pended when a RSS hash was computed + */ +#define NFP_NET_RSS_NONE 0 +#define NFP_NET_RSS_IPV4 1 +#define NFP_NET_RSS_IPV6 2 +#define NFP_NET_RSS_IPV6_EX 3 +#define NFP_NET_RSS_IPV4_TCP 4 +#define NFP_NET_RSS_IPV6_TCP 5 +#define NFP_NET_RSS_IPV6_EX_TCP 6 +#define NFP_NET_RSS_IPV4_UDP 7 +#define NFP_NET_RSS_IPV6_UDP 8 +#define NFP_NET_RSS_IPV6_EX_UDP 9 + +/** + * @NFP_NET_TXR_MAX: Maximum number of TX rings + * @NFP_NET_TXR_MASK: Mask for TX rings + * @NFP_NET_RXR_MAX: Maximum number of RX rings + * @NFP_NET_RXR_MASK: Mask for RX rings + */ +#define NFP_NET_TXR_MAX 64 +#define NFP_NET_TXR_MASK (NFP_NET_TXR_MAX - 1) +#define NFP_NET_RXR_MAX 64 +#define NFP_NET_RXR_MASK (NFP_NET_RXR_MAX - 1) + +/** + * Read/Write config words (0x0000 - 0x002c) + * @NFP_NET_CFG_CTRL: Global control + * @NFP_NET_CFG_UPDATE: Indicate which fields are updated + * @NFP_NET_CFG_TXRS_ENABLE: Bitmask of enabled TX rings + * @NFP_NET_CFG_RXRS_ENABLE: Bitmask of enabled RX rings + * @NFP_NET_CFG_MTU: Set MTU size + * @NFP_NET_CFG_FLBUFSZ: Set freelist buffer size (must be larger than MTU) + * @NFP_NET_CFG_EXN: MSI-X table entry for exceptions + * @NFP_NET_CFG_LSC: MSI-X table entry for link state changes + * @NFP_NET_CFG_MACADDR: MAC address + * + * TODO: + * - define Error details in UPDATE + */ +#define NFP_NET_CFG_CTRL 0x0000 +#define NFP_NET_CFG_CTRL_ENABLE (0x1 << 0) /* Global enable */ +#define NFP_NET_CFG_CTRL_PROMISC (0x1 << 1) /* Enable Promisc mode */ +#define NFP_NET_CFG_CTRL_L2BC (0x1 << 2) /* Allow L2 Broadcast */ +#define NFP_NET_CFG_CTRL_L2MC (0x1 << 3) /* Allow L2 Multicast */ +#define NFP_NET_CFG_CTRL_RXCSUM (0x1 << 4) /* Enable RX Checksum */ +#define NFP_NET_CFG_CTRL_TXCSUM (0x1 << 5) /* Enable TX Checksum */ +#define NFP_NET_CFG_CTRL_RXVLAN (0x1 << 6) /* Enable VLAN strip */ +#define NFP_NET_CFG_CTRL_TXVLAN (0x1 << 7) /* Enable VLAN insert */ +#define NFP_NET_CFG_CTRL_SCATTER (0x1 << 8) /* Scatter DMA */ +#define NFP_NET_CFG_CTRL_GATHER (0x1 << 9) /* Gather DMA */ +#define NFP_NET_CFG_CTRL_LSO (0x1 << 10) /* LSO/TSO */ +#define NFP_NET_CFG_CTRL_RINGCFG (0x1 << 16) /* Ring runtime changes */ +#define NFP_NET_CFG_CTRL_RSS (0x1 << 17) /* RSS */ +#define NFP_NET_CFG_CTRL_IRQMOD (0x1 << 18) /* Interrupt moderation */ +#define NFP_NET_CFG_CTRL_RINGPRIO (0x1 << 19) /* Ring priorities */ +#define NFP_NET_CFG_CTRL_MSIXAUTO (0x1 << 20) /* MSI-X auto-masking */ +#define NFP_NET_CFG_CTRL_TXRWB (0x1 << 21) /* Write-back of TX ring*/ +#define NFP_NET_CFG_CTRL_L2SWITCH (0x1 << 22) /* L2 Switch */ +#define NFP_NET_CFG_CTRL_L2SWITCH_LOCAL (0x1 << 23) /* Switch to local */ +#define NFP_NET_CFG_CTRL_VXLAN (0x1 << 24) /* VXLAN tunnel support */ +#define NFP_NET_CFG_CTRL_NVGRE (0x1 << 25) /* NVGRE tunnel support */ +#define NFP_NET_CFG_UPDATE 0x0004 +#define NFP_NET_CFG_UPDATE_GEN (0x1 << 0) /* General update */ +#define NFP_NET_CFG_UPDATE_RING (0x1 << 1) /* Ring config change */ +#define NFP_NET_CFG_UPDATE_RSS (0x1 << 2) /* RSS config change */ +#define NFP_NET_CFG_UPDATE_TXRPRIO (0x1 << 3) /* TX Ring prio change */ +#define NFP_NET_CFG_UPDATE_RXRPRIO (0x1 << 4) /* RX Ring prio change */ +#define NFP_NET_CFG_UPDATE_MSIX (0x1 << 5) /* MSI-X change */ +#define NFP_NET_CFG_UPDATE_L2SWITCH (0x1 << 6) /* Switch changes */ +#define NFP_NET_CFG_UPDATE_RESET (0x1 << 7) /* Update due to FLR */ +#define NFP_NET_CFG_UPDATE_IRQMOD (0x1 << 8) /* IRQ mod change */ +#define NFP_NET_CFG_UPDATE_VXLAN (0x1 << 9) /* VXLAN port change */ +#define NFP_NET_CFG_UPDATE_ERR (0x1 << 31) /* A error occurred */ +#define NFP_NET_CFG_TXRS_ENABLE 0x0008 +#define NFP_NET_CFG_RXRS_ENABLE 0x0010 +#define NFP_NET_CFG_MTU 0x0018 +#define NFP_NET_CFG_FLBUFSZ 0x001c +#define NFP_NET_CFG_EXN 0x001f +#define NFP_NET_CFG_LSC 0x0020 +#define NFP_NET_CFG_MACADDR 0x0024 + +/** + * Read-only words (0x0030 - 0x0050): + * @NFP_NET_CFG_VERSION: Firmware version number + * @NFP_NET_CFG_STS: Status + * @NFP_NET_CFG_CAP: Capabilities (same bits as @NFP_NET_CFG_CTRL) + * @NFP_NET_MAX_TXRINGS: Maximum number of TX rings + * @NFP_NET_MAX_RXRINGS: Maximum number of RX rings + * @NFP_NET_MAX_MTU: Maximum support MTU + * @NFP_NET_CFG_START_TXQ: Start Queue Control Queue to use for TX (PF only) + * @NFP_NET_CFG_START_RXQ: Start Queue Control Queue to use for RX (PF only) + * + * TODO: + * - define more STS bits + */ +#define NFP_NET_CFG_VERSION 0x0030 +#define NFP_NET_CFG_VERSION_RESERVED_MASK (0xff << 24) +#define NFP_NET_CFG_VERSION_CLASS_MASK (0xff << 16) +#define NFP_NET_CFG_VERSION_CLASS(x) (((x) & 0xff) << 16) +#define NFP_NET_CFG_VERSION_CLASS_GENERIC 0 +#define NFP_NET_CFG_VERSION_MAJOR_MASK (0xff << 8) +#define NFP_NET_CFG_VERSION_MAJOR(x) (((x) & 0xff) << 8) +#define NFP_NET_CFG_VERSION_MINOR_MASK (0xff << 0) +#define NFP_NET_CFG_VERSION_MINOR(x) (((x) & 0xff) << 0) +#define NFP_NET_CFG_STS 0x0034 +#define NFP_NET_CFG_STS_LINK (0x1 << 0) /* Link up or down */ +#define NFP_NET_CFG_CAP 0x0038 +#define NFP_NET_CFG_MAX_TXRINGS 0x003c +#define NFP_NET_CFG_MAX_RXRINGS 0x0040 +#define NFP_NET_CFG_MAX_MTU 0x0044 +/* Next two words are being used by VFs for solving THB350 issue */ +#define NFP_NET_CFG_START_TXQ 0x0048 +#define NFP_NET_CFG_START_RXQ 0x004c + +/** + * NFP-3200 workaround (0x0050 - 0x0058) + * @NFP_NET_CFG_SPARE_ADDR: DMA address for ME code to use (e.g. YDS-155 fix) + */ +#define NFP_NET_CFG_SPARE_ADDR 0x0050 +/** + * NFP6000/NFP4000 - Prepend configuration + */ +#define NFP_NET_CFG_RX_OFFSET 0x0050 +#define NFP_NET_CFG_RX_OFFSET_DYNAMIC 0 /* Prepend mode */ + +/** + * NFP6000/NFP4000 - VXLAN/UDP encap configuration + * @NFP_NET_CFG_VXLAN_PORT: Base address of table of tunnels' UDP dst ports + * @NFP_NET_CFG_VXLAN_SZ: Size of the UDP port table in bytes + */ +#define NFP_NET_CFG_VXLAN_PORT 0x0060 +#define NFP_NET_CFG_VXLAN_SZ 0x0008 + +/** + * 64B reserved for future use (0x0080 - 0x00c0) + */ +#define NFP_NET_CFG_RESERVED 0x0080 +#define NFP_NET_CFG_RESERVED_SZ 0x0040 + +/** + * RSS configuration (0x0100 - 0x01ac): + * Used only when NFP_NET_CFG_CTRL_RSS is enabled + * @NFP_NET_CFG_RSS_CFG: RSS configuration word + * @NFP_NET_CFG_RSS_KEY: RSS "secret" key + * @NFP_NET_CFG_RSS_ITBL: RSS indirection table + */ +#define NFP_NET_CFG_RSS_BASE 0x0100 +#define NFP_NET_CFG_RSS_CTRL NFP_NET_CFG_RSS_BASE +#define NFP_NET_CFG_RSS_MASK (0x7f) +#define NFP_NET_CFG_RSS_MASK_of(_x) ((_x) & 0x7f) +#define NFP_NET_CFG_RSS_IPV4 (1 << 8) /* RSS for IPv4 */ +#define NFP_NET_CFG_RSS_IPV6 (1 << 9) /* RSS for IPv6 */ +#define NFP_NET_CFG_RSS_IPV4_TCP (1 << 10) /* RSS for IPv4/TCP */ +#define NFP_NET_CFG_RSS_IPV4_UDP (1 << 11) /* RSS for IPv4/UDP */ +#define NFP_NET_CFG_RSS_IPV6_TCP (1 << 12) /* RSS for IPv6/TCP */ +#define NFP_NET_CFG_RSS_IPV6_UDP (1 << 13) /* RSS for IPv6/UDP */ +#define NFP_NET_CFG_RSS_TOEPLITZ (1 << 24) /* Use Toeplitz hash */ +#define NFP_NET_CFG_RSS_KEY (NFP_NET_CFG_RSS_BASE + 0x4) +#define NFP_NET_CFG_RSS_KEY_SZ 0x28 +#define NFP_NET_CFG_RSS_ITBL (NFP_NET_CFG_RSS_BASE + 0x4 + \ + NFP_NET_CFG_RSS_KEY_SZ) +#define NFP_NET_CFG_RSS_ITBL_SZ 0x80 + +/** + * TX ring configuration (0x200 - 0x800) + * @NFP_NET_CFG_TXR_BASE: Base offset for TX ring configuration + * @NFP_NET_CFG_TXR_ADDR: Per TX ring DMA address (8B entries) + * @NFP_NET_CFG_TXR_WB_ADDR: Per TX ring write back DMA address (8B entries) + * @NFP_NET_CFG_TXR_SZ: Per TX ring ring size (1B entries) + * @NFP_NET_CFG_TXR_VEC: Per TX ring MSI-X table entry (1B entries) + * @NFP_NET_CFG_TXR_PRIO: Per TX ring priority (1B entries) + * @NFP_NET_CFG_TXR_IRQ_MOD: Per TX ring interrupt moderation packet + */ +#define NFP_NET_CFG_TXR_BASE 0x0200 +#define NFP_NET_CFG_TXR_ADDR(_x) (NFP_NET_CFG_TXR_BASE + ((_x) * 0x8)) +#define NFP_NET_CFG_TXR_WB_ADDR(_x) (NFP_NET_CFG_TXR_BASE + 0x200 + \ + ((_x) * 0x8)) +#define NFP_NET_CFG_TXR_SZ(_x) (NFP_NET_CFG_TXR_BASE + 0x400 + (_x)) +#define NFP_NET_CFG_TXR_VEC(_x) (NFP_NET_CFG_TXR_BASE + 0x440 + (_x)) +#define NFP_NET_CFG_TXR_PRIO(_x) (NFP_NET_CFG_TXR_BASE + 0x480 + (_x)) +#define NFP_NET_CFG_TXR_IRQ_MOD(_x) (NFP_NET_CFG_TXR_BASE + 0x500 + \ + ((_x) * 0x4)) + +/** + * RX ring configuration (0x0800 - 0x0c00) + * @NFP_NET_CFG_RXR_BASE: Base offset for RX ring configuration + * @NFP_NET_CFG_RXR_ADDR: Per RX ring DMA address (8B entries) + * @NFP_NET_CFG_RXR_SZ: Per RX ring ring size (1B entries) + * @NFP_NET_CFG_RXR_VEC: Per RX ring MSI-X table entry (1B entries) + * @NFP_NET_CFG_RXR_PRIO: Per RX ring priority (1B entries) + * @NFP_NET_CFG_RXR_IRQ_MOD: Per RX ring interrupt moderation (4B entries) + */ +#define NFP_NET_CFG_RXR_BASE 0x0800 +#define NFP_NET_CFG_RXR_ADDR(_x) (NFP_NET_CFG_RXR_BASE + ((_x) * 0x8)) +#define NFP_NET_CFG_RXR_SZ(_x) (NFP_NET_CFG_RXR_BASE + 0x200 + (_x)) +#define NFP_NET_CFG_RXR_VEC(_x) (NFP_NET_CFG_RXR_BASE + 0x240 + (_x)) +#define NFP_NET_CFG_RXR_PRIO(_x) (NFP_NET_CFG_RXR_BASE + 0x280 + (_x)) +#define NFP_NET_CFG_RXR_IRQ_MOD(_x) (NFP_NET_CFG_RXR_BASE + 0x300 + \ + ((_x) * 0x4)) + +/** + * Interrupt Control/Cause registers (0x0c00 - 0x0d00) + * These registers are only used when MSI-X auto-masking is not + * enabled (@NFP_NET_CFG_CTRL_MSIXAUTO not set). The array is index + * by MSI-X entry and are 1B in size. If an entry is zero, the + * corresponding entry is enabled. If the FW generates an interrupt, + * it writes a cause into the corresponding field. This also masks + * the MSI-X entry and the host driver must clear the register to + * re-enable the interrupt. + */ +#define NFP_NET_CFG_ICR_BASE 0x0c00 +#define NFP_NET_CFG_ICR(_x) (NFP_NET_CFG_ICR_BASE + (_x)) +#define NFP_NET_CFG_ICR_UNMASKED 0x0 +#define NFP_NET_CFG_ICR_RXTX 0x1 +#define NFP_NET_CFG_ICR_LSC 0x2 + +/** + * General device stats (0x0d00 - 0x0d90) + * all counters are 64bit. + */ +#define NFP_NET_CFG_STATS_BASE 0x0d00 +#define NFP_NET_CFG_STATS_RX_DISCARDS (NFP_NET_CFG_STATS_BASE + 0x00) +#define NFP_NET_CFG_STATS_RX_ERRORS (NFP_NET_CFG_STATS_BASE + 0x08) +#define NFP_NET_CFG_STATS_RX_OCTETS (NFP_NET_CFG_STATS_BASE + 0x10) +#define NFP_NET_CFG_STATS_RX_UC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x18) +#define NFP_NET_CFG_STATS_RX_MC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x20) +#define NFP_NET_CFG_STATS_RX_BC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x28) +#define NFP_NET_CFG_STATS_RX_FRAMES (NFP_NET_CFG_STATS_BASE + 0x30) +#define NFP_NET_CFG_STATS_RX_MC_FRAMES (NFP_NET_CFG_STATS_BASE + 0x38) +#define NFP_NET_CFG_STATS_RX_BC_FRAMES (NFP_NET_CFG_STATS_BASE + 0x40) + +#define NFP_NET_CFG_STATS_TX_DISCARDS (NFP_NET_CFG_STATS_BASE + 0x48) +#define NFP_NET_CFG_STATS_TX_ERRORS (NFP_NET_CFG_STATS_BASE + 0x50) +#define NFP_NET_CFG_STATS_TX_OCTETS (NFP_NET_CFG_STATS_BASE + 0x58) +#define NFP_NET_CFG_STATS_TX_UC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x60) +#define NFP_NET_CFG_STATS_TX_MC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x68) +#define NFP_NET_CFG_STATS_TX_BC_OCTETS (NFP_NET_CFG_STATS_BASE + 0x70) +#define NFP_NET_CFG_STATS_TX_FRAMES (NFP_NET_CFG_STATS_BASE + 0x78) +#define NFP_NET_CFG_STATS_TX_MC_FRAMES (NFP_NET_CFG_STATS_BASE + 0x80) +#define NFP_NET_CFG_STATS_TX_BC_FRAMES (NFP_NET_CFG_STATS_BASE + 0x88) + +/** + * Per ring stats (0x1000 - 0x1800) + * options, 64bit per entry + * @NFP_NET_CFG_TXR_STATS: TX ring statistics (Packet and Byte count) + * @NFP_NET_CFG_RXR_STATS: RX ring statistics (Packet and Byte count) + */ +#define NFP_NET_CFG_TXR_STATS_BASE 0x1000 +#define NFP_NET_CFG_TXR_STATS(_x) (NFP_NET_CFG_TXR_STATS_BASE + \ + ((_x) * 0x10)) +#define NFP_NET_CFG_RXR_STATS_BASE 0x1400 +#define NFP_NET_CFG_RXR_STATS(_x) (NFP_NET_CFG_RXR_STATS_BASE + \ + ((_x) * 0x10)) + +#endif /* _NFP_NET_CTRL_H_ */ diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_debugfs.c b/drivers/net/ethernet/netronome/nfp/nfp_net_debugfs.c new file mode 100644 index 000000000000..4c97c713121c --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_debugfs.c @@ -0,0 +1,235 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#include +#include +#include + +#include "nfp_net.h" + +static struct dentry *nfp_dir; + +static int nfp_net_debugfs_rx_q_read(struct seq_file *file, void *data) +{ + struct nfp_net_rx_ring *rx_ring = file->private; + int fl_rd_p, fl_wr_p, rx_rd_p, rx_wr_p, rxd_cnt; + struct nfp_net_rx_desc *rxd; + struct sk_buff *skb; + struct nfp_net *nn; + int i; + + rtnl_lock(); + + if (!rx_ring->r_vec || !rx_ring->r_vec->nfp_net) + goto out; + nn = rx_ring->r_vec->nfp_net; + if (!netif_running(nn->netdev)) + goto out; + + rxd_cnt = rx_ring->cnt; + + fl_rd_p = nfp_qcp_rd_ptr_read(rx_ring->qcp_fl); + fl_wr_p = nfp_qcp_wr_ptr_read(rx_ring->qcp_fl); + rx_rd_p = nfp_qcp_rd_ptr_read(rx_ring->qcp_rx); + rx_wr_p = nfp_qcp_wr_ptr_read(rx_ring->qcp_rx); + + seq_printf(file, "RX[%02d]: H_RD=%d H_WR=%d FL_RD=%d FL_WR=%d RX_RD=%d RX_WR=%d\n", + rx_ring->idx, rx_ring->rd_p, rx_ring->wr_p, + fl_rd_p, fl_wr_p, rx_rd_p, rx_wr_p); + + for (i = 0; i < rxd_cnt; i++) { + rxd = &rx_ring->rxds[i]; + seq_printf(file, "%04d: 0x%08x 0x%08x", i, + rxd->vals[0], rxd->vals[1]); + + skb = READ_ONCE(rx_ring->rxbufs[i].skb); + if (skb) + seq_printf(file, " skb->head=%p skb->data=%p", + skb->head, skb->data); + + if (rx_ring->rxbufs[i].dma_addr) + seq_printf(file, " dma_addr=%pad", + &rx_ring->rxbufs[i].dma_addr); + + if (i == rx_ring->rd_p % rxd_cnt) + seq_puts(file, " H_RD "); + if (i == rx_ring->wr_p % rxd_cnt) + seq_puts(file, " H_WR "); + if (i == fl_rd_p % rxd_cnt) + seq_puts(file, " FL_RD"); + if (i == fl_wr_p % rxd_cnt) + seq_puts(file, " FL_WR"); + if (i == rx_rd_p % rxd_cnt) + seq_puts(file, " RX_RD"); + if (i == rx_wr_p % rxd_cnt) + seq_puts(file, " RX_WR"); + + seq_putc(file, '\n'); + } +out: + rtnl_unlock(); + return 0; +} + +static int nfp_net_debugfs_rx_q_open(struct inode *inode, struct file *f) +{ + return single_open(f, nfp_net_debugfs_rx_q_read, inode->i_private); +} + +static const struct file_operations nfp_rx_q_fops = { + .owner = THIS_MODULE, + .open = nfp_net_debugfs_rx_q_open, + .release = single_release, + .read = seq_read, + .llseek = seq_lseek +}; + +static int nfp_net_debugfs_tx_q_read(struct seq_file *file, void *data) +{ + struct nfp_net_tx_ring *tx_ring = file->private; + struct nfp_net_tx_desc *txd; + int d_rd_p, d_wr_p, txd_cnt; + struct sk_buff *skb; + struct nfp_net *nn; + int i; + + rtnl_lock(); + + if (!tx_ring->r_vec || !tx_ring->r_vec->nfp_net) + goto out; + nn = tx_ring->r_vec->nfp_net; + if (!netif_running(nn->netdev)) + goto out; + + txd_cnt = tx_ring->cnt; + + d_rd_p = nfp_qcp_rd_ptr_read(tx_ring->qcp_q); + d_wr_p = nfp_qcp_wr_ptr_read(tx_ring->qcp_q); + + seq_printf(file, "TX[%02d]: H_RD=%d H_WR=%d D_RD=%d D_WR=%d\n", + tx_ring->idx, tx_ring->rd_p, tx_ring->wr_p, d_rd_p, d_wr_p); + + for (i = 0; i < txd_cnt; i++) { + txd = &tx_ring->txds[i]; + seq_printf(file, "%04d: 0x%08x 0x%08x 0x%08x 0x%08x", i, + txd->vals[0], txd->vals[1], + txd->vals[2], txd->vals[3]); + + skb = READ_ONCE(tx_ring->txbufs[i].skb); + if (skb) + seq_printf(file, " skb->head=%p skb->data=%p", + skb->head, skb->data); + if (tx_ring->txbufs[i].dma_addr) + seq_printf(file, " dma_addr=%pad", + &tx_ring->txbufs[i].dma_addr); + + if (i == tx_ring->rd_p % txd_cnt) + seq_puts(file, " H_RD"); + if (i == tx_ring->wr_p % txd_cnt) + seq_puts(file, " H_WR"); + if (i == d_rd_p % txd_cnt) + seq_puts(file, " D_RD"); + if (i == d_wr_p % txd_cnt) + seq_puts(file, " D_WR"); + + seq_putc(file, '\n'); + } +out: + rtnl_unlock(); + return 0; +} + +static int nfp_net_debugfs_tx_q_open(struct inode *inode, struct file *f) +{ + return single_open(f, nfp_net_debugfs_tx_q_read, inode->i_private); +} + +static const struct file_operations nfp_tx_q_fops = { + .owner = THIS_MODULE, + .open = nfp_net_debugfs_tx_q_open, + .release = single_release, + .read = seq_read, + .llseek = seq_lseek +}; + +void nfp_net_debugfs_adapter_add(struct nfp_net *nn) +{ + static struct dentry *queues, *tx, *rx; + char int_name[16]; + int i; + + if (IS_ERR_OR_NULL(nfp_dir)) + return; + + nn->debugfs_dir = debugfs_create_dir(pci_name(nn->pdev), nfp_dir); + if (IS_ERR_OR_NULL(nn->debugfs_dir)) + return; + + /* Create queue debugging sub-tree */ + queues = debugfs_create_dir("queue", nn->debugfs_dir); + if (IS_ERR_OR_NULL(nn->debugfs_dir)) + return; + + rx = debugfs_create_dir("rx", queues); + tx = debugfs_create_dir("tx", queues); + if (IS_ERR_OR_NULL(rx) || IS_ERR_OR_NULL(tx)) + return; + + for (i = 0; i < nn->num_rx_rings; i++) { + sprintf(int_name, "%d", i); + debugfs_create_file(int_name, S_IRUSR, rx, + &nn->rx_rings[i], &nfp_rx_q_fops); + } + + for (i = 0; i < nn->num_tx_rings; i++) { + sprintf(int_name, "%d", i); + debugfs_create_file(int_name, S_IRUSR, tx, + &nn->tx_rings[i], &nfp_tx_q_fops); + } +} + +void nfp_net_debugfs_adapter_del(struct nfp_net *nn) +{ + debugfs_remove_recursive(nn->debugfs_dir); + nn->debugfs_dir = NULL; +} + +void nfp_net_debugfs_create(void) +{ + nfp_dir = debugfs_create_dir("nfp_net", NULL); +} + +void nfp_net_debugfs_destroy(void) +{ + debugfs_remove_recursive(nfp_dir); + nfp_dir = NULL; +} diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c new file mode 100644 index 000000000000..9a4084a68db5 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c @@ -0,0 +1,640 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* + * nfp_net_ethtool.c + * Netronome network device driver: ethtool support + * Authors: Jakub Kicinski + * Jason McMullan + * Rolf Neugebauer + * Brad Petrus + */ + +#include +#include +#include +#include +#include +#include +#include + +#include "nfp_net_ctrl.h" +#include "nfp_net.h" + +/* Support for stats. Returns netdev, driver, and device stats */ +enum { NETDEV_ET_STATS, NFP_NET_DRV_ET_STATS, NFP_NET_DEV_ET_STATS }; +struct _nfp_net_et_stats { + char name[ETH_GSTRING_LEN]; + int type; + int sz; + int off; +}; + +#define NN_ET_NETDEV_STAT(m) NETDEV_ET_STATS, \ + FIELD_SIZEOF(struct net_device_stats, m), \ + offsetof(struct net_device_stats, m) +/* For stats in the control BAR (other than Q stats) */ +#define NN_ET_DEV_STAT(m) NFP_NET_DEV_ET_STATS, \ + sizeof(u64), \ + (m) +static const struct _nfp_net_et_stats nfp_net_et_stats[] = { + /* netdev stats */ + {"rx_packets", NN_ET_NETDEV_STAT(rx_packets)}, + {"tx_packets", NN_ET_NETDEV_STAT(tx_packets)}, + {"rx_bytes", NN_ET_NETDEV_STAT(rx_bytes)}, + {"tx_bytes", NN_ET_NETDEV_STAT(tx_bytes)}, + {"rx_errors", NN_ET_NETDEV_STAT(rx_errors)}, + {"tx_errors", NN_ET_NETDEV_STAT(tx_errors)}, + {"rx_dropped", NN_ET_NETDEV_STAT(rx_dropped)}, + {"tx_dropped", NN_ET_NETDEV_STAT(tx_dropped)}, + {"multicast", NN_ET_NETDEV_STAT(multicast)}, + {"collisions", NN_ET_NETDEV_STAT(collisions)}, + {"rx_over_errors", NN_ET_NETDEV_STAT(rx_over_errors)}, + {"rx_crc_errors", NN_ET_NETDEV_STAT(rx_crc_errors)}, + {"rx_frame_errors", NN_ET_NETDEV_STAT(rx_frame_errors)}, + {"rx_fifo_errors", NN_ET_NETDEV_STAT(rx_fifo_errors)}, + {"rx_missed_errors", NN_ET_NETDEV_STAT(rx_missed_errors)}, + {"tx_aborted_errors", NN_ET_NETDEV_STAT(tx_aborted_errors)}, + {"tx_carrier_errors", NN_ET_NETDEV_STAT(tx_carrier_errors)}, + {"tx_fifo_errors", NN_ET_NETDEV_STAT(tx_fifo_errors)}, + /* Stats from the device */ + {"dev_rx_discards", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_DISCARDS)}, + {"dev_rx_errors", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_ERRORS)}, + {"dev_rx_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_OCTETS)}, + {"dev_rx_uc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_UC_OCTETS)}, + {"dev_rx_mc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_MC_OCTETS)}, + {"dev_rx_bc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_BC_OCTETS)}, + {"dev_rx_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_FRAMES)}, + {"dev_rx_mc_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_MC_FRAMES)}, + {"dev_rx_bc_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_RX_BC_FRAMES)}, + + {"dev_tx_discards", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_DISCARDS)}, + {"dev_tx_errors", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_ERRORS)}, + {"dev_tx_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_OCTETS)}, + {"dev_tx_uc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_UC_OCTETS)}, + {"dev_tx_mc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_MC_OCTETS)}, + {"dev_tx_bc_bytes", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_BC_OCTETS)}, + {"dev_tx_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_FRAMES)}, + {"dev_tx_mc_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_MC_FRAMES)}, + {"dev_tx_bc_pkts", NN_ET_DEV_STAT(NFP_NET_CFG_STATS_TX_BC_FRAMES)}, +}; + +#define NN_ET_GLOBAL_STATS_LEN ARRAY_SIZE(nfp_net_et_stats) +#define NN_ET_RVEC_STATS_LEN (nn->num_r_vecs * 3) +#define NN_ET_RVEC_GATHER_STATS 7 +#define NN_ET_QUEUE_STATS_LEN ((nn->num_tx_rings + nn->num_rx_rings) * 2) +#define NN_ET_STATS_LEN (NN_ET_GLOBAL_STATS_LEN + NN_ET_RVEC_GATHER_STATS + \ + NN_ET_RVEC_STATS_LEN + NN_ET_QUEUE_STATS_LEN) + +static void nfp_net_get_drvinfo(struct net_device *netdev, + struct ethtool_drvinfo *drvinfo) +{ + struct nfp_net *nn = netdev_priv(netdev); + + strlcpy(drvinfo->driver, nfp_net_driver_name, sizeof(drvinfo->driver)); + strlcpy(drvinfo->version, nfp_net_driver_version, + sizeof(drvinfo->version)); + + snprintf(drvinfo->fw_version, sizeof(drvinfo->fw_version), + "%d.%d.%d.%d", + nn->fw_ver.resv, nn->fw_ver.class, + nn->fw_ver.major, nn->fw_ver.minor); + strlcpy(drvinfo->bus_info, pci_name(nn->pdev), + sizeof(drvinfo->bus_info)); + + drvinfo->n_stats = NN_ET_STATS_LEN; + drvinfo->regdump_len = NFP_NET_CFG_BAR_SZ; +} + +static void nfp_net_get_ringparam(struct net_device *netdev, + struct ethtool_ringparam *ring) +{ + struct nfp_net *nn = netdev_priv(netdev); + + ring->rx_max_pending = NFP_NET_MAX_RX_DESCS; + ring->tx_max_pending = NFP_NET_MAX_TX_DESCS; + ring->rx_pending = nn->rxd_cnt; + ring->tx_pending = nn->txd_cnt; +} + +static int nfp_net_set_ringparam(struct net_device *netdev, + struct ethtool_ringparam *ring) +{ + struct nfp_net *nn = netdev_priv(netdev); + u32 rxd_cnt, txd_cnt; + + if (netif_running(netdev)) { + /* Some NIC drivers allow reconfiguration on the fly, + * some down the interface, change and then up it + * again. For now we don't allow changes when the + * device is up. + */ + nn_warn(nn, "Can't change rings while device is up\n"); + return -EBUSY; + } + + /* We don't have separate queues/rings for small/large frames. */ + if (ring->rx_mini_pending || ring->rx_jumbo_pending) + return -EINVAL; + + /* Round up to supported values */ + rxd_cnt = roundup_pow_of_two(ring->rx_pending); + rxd_cnt = max_t(u32, rxd_cnt, NFP_NET_MIN_RX_DESCS); + rxd_cnt = min_t(u32, rxd_cnt, NFP_NET_MAX_RX_DESCS); + + txd_cnt = roundup_pow_of_two(ring->tx_pending); + txd_cnt = max_t(u32, txd_cnt, NFP_NET_MIN_TX_DESCS); + txd_cnt = min_t(u32, txd_cnt, NFP_NET_MAX_TX_DESCS); + + if (nn->rxd_cnt != rxd_cnt || nn->txd_cnt != txd_cnt) + nn_dbg(nn, "Change ring size: RxQ %u->%u, TxQ %u->%u\n", + nn->rxd_cnt, rxd_cnt, nn->txd_cnt, txd_cnt); + + nn->rxd_cnt = rxd_cnt; + nn->txd_cnt = txd_cnt; + + return 0; +} + +static void nfp_net_get_strings(struct net_device *netdev, + u32 stringset, u8 *data) +{ + struct nfp_net *nn = netdev_priv(netdev); + u8 *p = data; + int i; + + switch (stringset) { + case ETH_SS_STATS: + for (i = 0; i < NN_ET_GLOBAL_STATS_LEN; i++) { + memcpy(p, nfp_net_et_stats[i].name, ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + } + for (i = 0; i < nn->num_r_vecs; i++) { + sprintf(p, "rvec_%u_rx_pkts", i); + p += ETH_GSTRING_LEN; + sprintf(p, "rvec_%u_tx_pkts", i); + p += ETH_GSTRING_LEN; + sprintf(p, "rvec_%u_tx_busy", i); + p += ETH_GSTRING_LEN; + } + strncpy(p, "hw_rx_csum_ok", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "hw_rx_csum_inner_ok", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "hw_rx_csum_err", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "hw_tx_csum", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "hw_tx_inner_csum", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "tx_gather", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + strncpy(p, "tx_lso", ETH_GSTRING_LEN); + p += ETH_GSTRING_LEN; + for (i = 0; i < nn->num_tx_rings; i++) { + sprintf(p, "txq_%u_pkts", i); + p += ETH_GSTRING_LEN; + sprintf(p, "txq_%u_bytes", i); + p += ETH_GSTRING_LEN; + } + for (i = 0; i < nn->num_rx_rings; i++) { + sprintf(p, "rxq_%u_pkts", i); + p += ETH_GSTRING_LEN; + sprintf(p, "rxq_%u_bytes", i); + p += ETH_GSTRING_LEN; + } + break; + } +} + +static void nfp_net_get_stats(struct net_device *netdev, + struct ethtool_stats *stats, u64 *data) +{ + u64 gathered_stats[NN_ET_RVEC_GATHER_STATS] = {}; + struct nfp_net *nn = netdev_priv(netdev); + struct rtnl_link_stats64 *netdev_stats; + struct rtnl_link_stats64 temp = {}; + u64 tmp[NN_ET_RVEC_GATHER_STATS]; + u8 __iomem *io_p; + int i, j, k; + u8 *p; + + netdev_stats = dev_get_stats(netdev, &temp); + + for (i = 0; i < NN_ET_GLOBAL_STATS_LEN; i++) { + switch (nfp_net_et_stats[i].type) { + case NETDEV_ET_STATS: + p = (char *)netdev_stats + nfp_net_et_stats[i].off; + data[i] = nfp_net_et_stats[i].sz == sizeof(u64) ? + *(u64 *)p : *(u32 *)p; + break; + + case NFP_NET_DEV_ET_STATS: + io_p = nn->ctrl_bar + nfp_net_et_stats[i].off; + data[i] = readq(io_p); + break; + } + } + for (j = 0; j < nn->num_r_vecs; j++) { + unsigned int start; + + do { + start = u64_stats_fetch_begin(&nn->r_vecs[j].rx_sync); + data[i++] = nn->r_vecs[j].rx_pkts; + tmp[0] = nn->r_vecs[j].hw_csum_rx_ok; + tmp[1] = nn->r_vecs[j].hw_csum_rx_inner_ok; + tmp[2] = nn->r_vecs[j].hw_csum_rx_error; + } while (u64_stats_fetch_retry(&nn->r_vecs[j].rx_sync, start)); + + do { + start = u64_stats_fetch_begin(&nn->r_vecs[j].tx_sync); + data[i++] = nn->r_vecs[j].tx_pkts; + data[i++] = nn->r_vecs[j].tx_busy; + tmp[3] = nn->r_vecs[j].hw_csum_tx; + tmp[4] = nn->r_vecs[j].hw_csum_tx_inner; + tmp[5] = nn->r_vecs[j].tx_gather; + tmp[6] = nn->r_vecs[j].tx_lso; + } while (u64_stats_fetch_retry(&nn->r_vecs[j].tx_sync, start)); + + for (k = 0; k < NN_ET_RVEC_GATHER_STATS; k++) + gathered_stats[k] += tmp[k]; + } + for (j = 0; j < NN_ET_RVEC_GATHER_STATS; j++) + data[i++] = gathered_stats[j]; + for (j = 0; j < nn->num_tx_rings; j++) { + io_p = nn->ctrl_bar + NFP_NET_CFG_TXR_STATS(j); + data[i++] = readq(io_p); + io_p = nn->ctrl_bar + NFP_NET_CFG_TXR_STATS(j) + 8; + data[i++] = readq(io_p); + } + for (j = 0; j < nn->num_rx_rings; j++) { + io_p = nn->ctrl_bar + NFP_NET_CFG_RXR_STATS(j); + data[i++] = readq(io_p); + io_p = nn->ctrl_bar + NFP_NET_CFG_RXR_STATS(j) + 8; + data[i++] = readq(io_p); + } +} + +static int nfp_net_get_sset_count(struct net_device *netdev, int sset) +{ + struct nfp_net *nn = netdev_priv(netdev); + + switch (sset) { + case ETH_SS_STATS: + return NN_ET_STATS_LEN; + default: + return -EOPNOTSUPP; + } +} + +/* RX network flow classification (RSS, filters, etc) + */ +static u32 ethtool_flow_to_nfp_flag(u32 flow_type) +{ + static const u32 xlate_ethtool_to_nfp[IPV6_FLOW + 1] = { + [TCP_V4_FLOW] = NFP_NET_CFG_RSS_IPV4_TCP, + [TCP_V6_FLOW] = NFP_NET_CFG_RSS_IPV6_TCP, + [UDP_V4_FLOW] = NFP_NET_CFG_RSS_IPV4_UDP, + [UDP_V6_FLOW] = NFP_NET_CFG_RSS_IPV6_UDP, + [IPV4_FLOW] = NFP_NET_CFG_RSS_IPV4, + [IPV6_FLOW] = NFP_NET_CFG_RSS_IPV6, + }; + + if (flow_type >= ARRAY_SIZE(xlate_ethtool_to_nfp)) + return 0; + + return xlate_ethtool_to_nfp[flow_type]; +} + +static int nfp_net_get_rss_hash_opts(struct nfp_net *nn, + struct ethtool_rxnfc *cmd) +{ + u32 nfp_rss_flag; + + cmd->data = 0; + + if (!(nn->cap & NFP_NET_CFG_CTRL_RSS)) + return -EOPNOTSUPP; + + nfp_rss_flag = ethtool_flow_to_nfp_flag(cmd->flow_type); + if (!nfp_rss_flag) + return -EINVAL; + + cmd->data |= RXH_IP_SRC | RXH_IP_DST; + if (nn->rss_cfg & nfp_rss_flag) + cmd->data |= RXH_L4_B_0_1 | RXH_L4_B_2_3; + + return 0; +} + +static int nfp_net_get_rxnfc(struct net_device *netdev, + struct ethtool_rxnfc *cmd, u32 *rule_locs) +{ + struct nfp_net *nn = netdev_priv(netdev); + + switch (cmd->cmd) { + case ETHTOOL_GRXRINGS: + cmd->data = nn->num_rx_rings; + return 0; + case ETHTOOL_GRXFH: + return nfp_net_get_rss_hash_opts(nn, cmd); + default: + return -EOPNOTSUPP; + } +} + +static int nfp_net_set_rss_hash_opt(struct nfp_net *nn, + struct ethtool_rxnfc *nfc) +{ + u32 new_rss_cfg = nn->rss_cfg; + u32 nfp_rss_flag; + int err; + + if (!(nn->cap & NFP_NET_CFG_CTRL_RSS)) + return -EOPNOTSUPP; + + /* RSS only supports IP SA/DA and L4 src/dst ports */ + if (nfc->data & ~(RXH_IP_SRC | RXH_IP_DST | + RXH_L4_B_0_1 | RXH_L4_B_2_3)) + return -EINVAL; + + /* We need at least the IP SA/DA fields for hashing */ + if (!(nfc->data & RXH_IP_SRC) || + !(nfc->data & RXH_IP_DST)) + return -EINVAL; + + nfp_rss_flag = ethtool_flow_to_nfp_flag(nfc->flow_type); + if (!nfp_rss_flag) + return -EINVAL; + + switch (nfc->data & (RXH_L4_B_0_1 | RXH_L4_B_2_3)) { + case 0: + new_rss_cfg &= ~nfp_rss_flag; + break; + case (RXH_L4_B_0_1 | RXH_L4_B_2_3): + new_rss_cfg |= nfp_rss_flag; + break; + default: + return -EINVAL; + } + + new_rss_cfg |= NFP_NET_CFG_RSS_TOEPLITZ; + new_rss_cfg |= NFP_NET_CFG_RSS_MASK; + + if (new_rss_cfg == nn->rss_cfg) + return 0; + + writel(new_rss_cfg, nn->ctrl_bar + NFP_NET_CFG_RSS_CTRL); + err = nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_RSS); + if (err) + return err; + + nn->rss_cfg = new_rss_cfg; + + nn_dbg(nn, "Changed RSS config to 0x%x\n", nn->rss_cfg); + return 0; +} + +static int nfp_net_set_rxnfc(struct net_device *netdev, + struct ethtool_rxnfc *cmd) +{ + struct nfp_net *nn = netdev_priv(netdev); + + switch (cmd->cmd) { + case ETHTOOL_SRXFH: + return nfp_net_set_rss_hash_opt(nn, cmd); + default: + return -EOPNOTSUPP; + } +} + +static u32 nfp_net_get_rxfh_indir_size(struct net_device *netdev) +{ + struct nfp_net *nn = netdev_priv(netdev); + + if (!(nn->cap & NFP_NET_CFG_CTRL_RSS)) + return 0; + + return ARRAY_SIZE(nn->rss_itbl); +} + +static u32 nfp_net_get_rxfh_key_size(struct net_device *netdev) +{ + return NFP_NET_CFG_RSS_KEY_SZ; +} + +static int nfp_net_get_rxfh(struct net_device *netdev, u32 *indir, u8 *key, + u8 *hfunc) +{ + struct nfp_net *nn = netdev_priv(netdev); + int i; + + if (!(nn->cap & NFP_NET_CFG_CTRL_RSS)) + return -EOPNOTSUPP; + + if (indir) + for (i = 0; i < ARRAY_SIZE(nn->rss_itbl); i++) + indir[i] = nn->rss_itbl[i]; + if (key) + memcpy(key, nn->rss_key, NFP_NET_CFG_RSS_KEY_SZ); + if (hfunc) + *hfunc = ETH_RSS_HASH_TOP; + + return 0; +} + +static int nfp_net_set_rxfh(struct net_device *netdev, + const u32 *indir, const u8 *key, + const u8 hfunc) +{ + struct nfp_net *nn = netdev_priv(netdev); + int i; + + if (!(nn->cap & NFP_NET_CFG_CTRL_RSS) || + !(hfunc == ETH_RSS_HASH_NO_CHANGE || hfunc == ETH_RSS_HASH_TOP)) + return -EOPNOTSUPP; + + if (!key && !indir) + return 0; + + if (key) { + memcpy(nn->rss_key, key, NFP_NET_CFG_RSS_KEY_SZ); + nfp_net_rss_write_key(nn); + } + if (indir) { + for (i = 0; i < ARRAY_SIZE(nn->rss_itbl); i++) + nn->rss_itbl[i] = indir[i]; + + nfp_net_rss_write_itbl(nn); + } + + return nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_RSS); +} + +/* Dump BAR registers + */ +static int nfp_net_get_regs_len(struct net_device *netdev) +{ + return NFP_NET_CFG_BAR_SZ; +} + +static void nfp_net_get_regs(struct net_device *netdev, + struct ethtool_regs *regs, void *p) +{ + struct nfp_net *nn = netdev_priv(netdev); + u32 *regs_buf = p; + int i; + + regs->version = nn_readl(nn, NFP_NET_CFG_VERSION); + + for (i = 0; i < NFP_NET_CFG_BAR_SZ / sizeof(u32); i++) + regs_buf[i] = readl(nn->ctrl_bar + (i * sizeof(u32))); +} + +static int nfp_net_get_coalesce(struct net_device *netdev, + struct ethtool_coalesce *ec) +{ + struct nfp_net *nn = netdev_priv(netdev); + + if (!(nn->cap & NFP_NET_CFG_CTRL_IRQMOD)) + return -EINVAL; + + ec->rx_coalesce_usecs = nn->rx_coalesce_usecs; + ec->rx_max_coalesced_frames = nn->rx_coalesce_max_frames; + ec->tx_coalesce_usecs = nn->tx_coalesce_usecs; + ec->tx_max_coalesced_frames = nn->tx_coalesce_max_frames; + + return 0; +} + +static int nfp_net_set_coalesce(struct net_device *netdev, + struct ethtool_coalesce *ec) +{ + struct nfp_net *nn = netdev_priv(netdev); + unsigned int factor; + + if (ec->rx_coalesce_usecs_irq || + ec->rx_max_coalesced_frames_irq || + ec->tx_coalesce_usecs_irq || + ec->tx_max_coalesced_frames_irq || + ec->stats_block_coalesce_usecs || + ec->use_adaptive_rx_coalesce || + ec->use_adaptive_tx_coalesce || + ec->pkt_rate_low || + ec->rx_coalesce_usecs_low || + ec->rx_max_coalesced_frames_low || + ec->tx_coalesce_usecs_low || + ec->tx_max_coalesced_frames_low || + ec->pkt_rate_high || + ec->rx_coalesce_usecs_high || + ec->rx_max_coalesced_frames_high || + ec->tx_coalesce_usecs_high || + ec->tx_max_coalesced_frames_high || + ec->rate_sample_interval) + return -ENOTSUPP; + + /* Compute factor used to convert coalesce '_usecs' parameters to + * ME timestamp ticks. There are 16 ME clock cycles for each timestamp + * count. + */ + factor = nn->me_freq_mhz / 16; + + /* Each pair of (usecs, max_frames) fields specifies that interrupts + * should be coalesced until + * (usecs > 0 && time_since_first_completion >= usecs) || + * (max_frames > 0 && completed_frames >= max_frames) + * + * It is illegal to set both usecs and max_frames to zero as this would + * cause interrupts to never be generated. To disable coalescing, set + * usecs = 0 and max_frames = 1. + * + * Some implementations ignore the value of max_frames and use the + * condition time_since_first_completion >= usecs + */ + + if (!(nn->cap & NFP_NET_CFG_CTRL_IRQMOD)) + return -EINVAL; + + /* ensure valid configuration */ + if (!ec->rx_coalesce_usecs && !ec->rx_max_coalesced_frames) + return -EINVAL; + + if (!ec->tx_coalesce_usecs && !ec->tx_max_coalesced_frames) + return -EINVAL; + + if (ec->rx_coalesce_usecs * factor >= ((1 << 16) - 1)) + return -EINVAL; + + if (ec->tx_coalesce_usecs * factor >= ((1 << 16) - 1)) + return -EINVAL; + + if (ec->rx_max_coalesced_frames >= ((1 << 16) - 1)) + return -EINVAL; + + if (ec->tx_max_coalesced_frames >= ((1 << 16) - 1)) + return -EINVAL; + + /* configuration is valid */ + nn->rx_coalesce_usecs = ec->rx_coalesce_usecs; + nn->rx_coalesce_max_frames = ec->rx_max_coalesced_frames; + nn->tx_coalesce_usecs = ec->tx_coalesce_usecs; + nn->tx_coalesce_max_frames = ec->tx_max_coalesced_frames; + + /* write configuration to device */ + nfp_net_coalesce_write_cfg(nn); + return nfp_net_reconfig(nn, NFP_NET_CFG_UPDATE_IRQMOD); +} + +static const struct ethtool_ops nfp_net_ethtool_ops = { + .get_drvinfo = nfp_net_get_drvinfo, + .get_ringparam = nfp_net_get_ringparam, + .set_ringparam = nfp_net_set_ringparam, + .get_strings = nfp_net_get_strings, + .get_ethtool_stats = nfp_net_get_stats, + .get_sset_count = nfp_net_get_sset_count, + .get_rxnfc = nfp_net_get_rxnfc, + .set_rxnfc = nfp_net_set_rxnfc, + .get_rxfh_indir_size = nfp_net_get_rxfh_indir_size, + .get_rxfh_key_size = nfp_net_get_rxfh_key_size, + .get_rxfh = nfp_net_get_rxfh, + .set_rxfh = nfp_net_set_rxfh, + .get_regs_len = nfp_net_get_regs_len, + .get_regs = nfp_net_get_regs, + .get_coalesce = nfp_net_get_coalesce, + .set_coalesce = nfp_net_set_coalesce, +}; + +void nfp_net_set_ethtool_ops(struct net_device *netdev) +{ + netdev->ethtool_ops = &nfp_net_ethtool_ops; +} diff --git a/drivers/net/ethernet/netronome/nfp/nfp_netvf_main.c b/drivers/net/ethernet/netronome/nfp/nfp_netvf_main.c new file mode 100644 index 000000000000..e2b22b8a20f1 --- /dev/null +++ b/drivers/net/ethernet/netronome/nfp/nfp_netvf_main.c @@ -0,0 +1,385 @@ +/* + * Copyright (C) 2015 Netronome Systems, Inc. + * + * This software is dual licensed under the GNU General License Version 2, + * June 1991 as shown in the file COPYING in the top-level directory of this + * source tree or the BSD 2-Clause License provided below. You have the + * option to license this software under the complete terms of either license. + * + * The BSD 2-Clause License: + * + * Redistribution and use in source and binary forms, with or + * without modification, are permitted provided that the following + * conditions are met: + * + * 1. Redistributions of source code must retain the above + * copyright notice, this list of conditions and the following + * disclaimer. + * + * 2. Redistributions in binary form must reproduce the above + * copyright notice, this list of conditions and the following + * disclaimer in the documentation and/or other materials + * provided with the distribution. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS + * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN + * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ + +/* + * nfp_netvf_main.c + * Netronome virtual function network device driver: Main entry point + * Author: Jason McMullan + * Rolf Neugebauer + */ + +#include +#include +#include +#include +#include + +#include "nfp_net_ctrl.h" +#include "nfp_net.h" + +const char nfp_net_driver_name[] = "nfp_netvf"; +const char nfp_net_driver_version[] = "0.1"; +#define PCI_DEVICE_NFP6000VF 0x6003 +static const struct pci_device_id nfp_netvf_pci_device_ids[] = { + { PCI_VENDOR_ID_NETRONOME, PCI_DEVICE_NFP6000VF, + PCI_VENDOR_ID_NETRONOME, PCI_ANY_ID, + PCI_ANY_ID, 0, + }, + { 0, } /* Required last entry. */ +}; +MODULE_DEVICE_TABLE(pci, nfp_netvf_pci_device_ids); + +static void nfp_netvf_get_mac_addr(struct nfp_net *nn) +{ + u8 mac_addr[ETH_ALEN]; + + put_unaligned_be32(nn_readl(nn, NFP_NET_CFG_MACADDR + 0), &mac_addr[0]); + /* We can't do readw for NFP-3200 compatibility */ + put_unaligned_be16(nn_readl(nn, NFP_NET_CFG_MACADDR + 4) >> 16, + &mac_addr[4]); + + if (!is_valid_ether_addr(mac_addr)) { + eth_hw_addr_random(nn->netdev); + return; + } + + ether_addr_copy(nn->netdev->dev_addr, mac_addr); + ether_addr_copy(nn->netdev->perm_addr, mac_addr); +} + +static int nfp_netvf_pci_probe(struct pci_dev *pdev, + const struct pci_device_id *pci_id) +{ + struct nfp_net_fw_version fw_ver; + int max_tx_rings, max_rx_rings; + u32 tx_bar_off, rx_bar_off; + u32 tx_bar_sz, rx_bar_sz; + int tx_bar_no, rx_bar_no; + u8 __iomem *ctrl_bar; + struct nfp_net *nn; + int is_nfp3200; + u32 startq; + int stride; + int err; + + err = pci_enable_device_mem(pdev); + if (err) + return err; + + err = pci_request_regions(pdev, nfp_net_driver_name); + if (err) { + dev_err(&pdev->dev, "Unable to allocate device memory.\n"); + goto err_pci_disable; + } + + switch (pdev->device) { + case PCI_DEVICE_NFP6000VF: + is_nfp3200 = 0; + break; + default: + err = -ENODEV; + goto err_pci_regions; + } + + pci_set_master(pdev); + + err = dma_set_mask_and_coherent(&pdev->dev, + DMA_BIT_MASK(NFP_NET_MAX_DMA_BITS)); + if (err) + goto err_pci_regions; + + /* Map the Control BAR. + * + * Irrespective of the advertised BAR size we only map the + * first NFP_NET_CFG_BAR_SZ of the BAR. This keeps the code + * the identical for PF and VF drivers. + */ + ctrl_bar = ioremap_nocache(pci_resource_start(pdev, NFP_NET_CRTL_BAR), + NFP_NET_CFG_BAR_SZ); + if (!ctrl_bar) { + dev_err(&pdev->dev, + "Failed to map resource %d\n", NFP_NET_CRTL_BAR); + err = -EIO; + goto err_pci_regions; + } + + nfp_net_get_fw_version(&fw_ver, ctrl_bar); + if (fw_ver.class != NFP_NET_CFG_VERSION_CLASS_GENERIC) { + dev_err(&pdev->dev, "Unknown Firmware ABI %d.%d.%d.%d\n", + fw_ver.resv, fw_ver.class, fw_ver.major, fw_ver.minor); + err = -EINVAL; + goto err_ctrl_unmap; + } + + /* Determine stride */ + if (nfp_net_fw_ver_eq(&fw_ver, 0, 0, 0, 0) || + nfp_net_fw_ver_eq(&fw_ver, 0, 0, 0, 1) || + nfp_net_fw_ver_eq(&fw_ver, 0, 0, 0x12, 0x48)) { + stride = 2; + tx_bar_no = NFP_NET_Q0_BAR; + rx_bar_no = NFP_NET_Q1_BAR; + dev_warn(&pdev->dev, "OBSOLETE Firmware detected - VF isolation not available\n"); + } else { + switch (fw_ver.major) { + case 1 ... 3: + if (is_nfp3200) { + stride = 2; + tx_bar_no = NFP_NET_Q0_BAR; + rx_bar_no = NFP_NET_Q1_BAR; + } else { + stride = 4; + tx_bar_no = NFP_NET_Q0_BAR; + rx_bar_no = tx_bar_no; + } + break; + default: + dev_err(&pdev->dev, "Unsupported Firmware ABI %d.%d.%d.%d\n", + fw_ver.resv, fw_ver.class, + fw_ver.major, fw_ver.minor); + err = -EINVAL; + goto err_ctrl_unmap; + } + } + + /* Find out how many rings are supported */ + max_tx_rings = readl(ctrl_bar + NFP_NET_CFG_MAX_TXRINGS); + max_rx_rings = readl(ctrl_bar + NFP_NET_CFG_MAX_RXRINGS); + + tx_bar_sz = NFP_QCP_QUEUE_ADDR_SZ * max_tx_rings * stride; + rx_bar_sz = NFP_QCP_QUEUE_ADDR_SZ * max_rx_rings * stride; + + /* Sanity checks */ + if (tx_bar_sz > pci_resource_len(pdev, tx_bar_no)) { + dev_err(&pdev->dev, + "TX BAR too small for number of TX rings. Adjusting\n"); + tx_bar_sz = pci_resource_len(pdev, tx_bar_no); + max_tx_rings = (tx_bar_sz / NFP_QCP_QUEUE_ADDR_SZ) / 2; + } + if (rx_bar_sz > pci_resource_len(pdev, rx_bar_no)) { + dev_err(&pdev->dev, + "RX BAR too small for number of RX rings. Adjusting\n"); + rx_bar_sz = pci_resource_len(pdev, rx_bar_no); + max_rx_rings = (rx_bar_sz / NFP_QCP_QUEUE_ADDR_SZ) / 2; + } + + /* XXX Implement a workaround for THB-350 here. Ideally, we + * have a different PCI ID for A rev VFs. + */ + switch (pdev->device) { + case PCI_DEVICE_NFP6000VF: + startq = readl(ctrl_bar + NFP_NET_CFG_START_TXQ); + tx_bar_off = NFP_PCIE_QUEUE(startq); + startq = readl(ctrl_bar + NFP_NET_CFG_START_RXQ); + rx_bar_off = NFP_PCIE_QUEUE(startq); + break; + default: + err = -ENODEV; + goto err_ctrl_unmap; + } + + /* Allocate and initialise the netdev */ + nn = nfp_net_netdev_alloc(pdev, max_tx_rings, max_rx_rings); + if (IS_ERR(nn)) { + err = PTR_ERR(nn); + goto err_ctrl_unmap; + } + + nn->fw_ver = fw_ver; + nn->ctrl_bar = ctrl_bar; + nn->is_vf = 1; + nn->is_nfp3200 = is_nfp3200; + nn->stride_tx = stride; + nn->stride_rx = stride; + + if (rx_bar_no == tx_bar_no) { + u32 bar_off, bar_sz; + resource_size_t map_addr; + + /* Make a single overlapping BAR mapping */ + if (tx_bar_off < rx_bar_off) + bar_off = tx_bar_off; + else + bar_off = rx_bar_off; + + if ((tx_bar_off + tx_bar_sz) > (rx_bar_off + rx_bar_sz)) + bar_sz = (tx_bar_off + tx_bar_sz) - bar_off; + else + bar_sz = (rx_bar_off + rx_bar_sz) - bar_off; + + map_addr = pci_resource_start(pdev, tx_bar_no) + bar_off; + nn->q_bar = ioremap_nocache(map_addr, bar_sz); + if (!nn->q_bar) { + nn_err(nn, "Failed to map resource %d\n", tx_bar_no); + err = -EIO; + goto err_netdev_free; + } + + /* TX queues */ + nn->tx_bar = nn->q_bar + (tx_bar_off - bar_off); + /* RX queues */ + nn->rx_bar = nn->q_bar + (rx_bar_off - bar_off); + } else { + resource_size_t map_addr; + + /* TX queues */ + map_addr = pci_resource_start(pdev, tx_bar_no) + tx_bar_off; + nn->tx_bar = ioremap_nocache(map_addr, tx_bar_sz); + if (!nn->tx_bar) { + nn_err(nn, "Failed to map resource %d\n", tx_bar_no); + err = -EIO; + goto err_netdev_free; + } + + /* RX queues */ + map_addr = pci_resource_start(pdev, rx_bar_no) + rx_bar_off; + nn->rx_bar = ioremap_nocache(map_addr, rx_bar_sz); + if (!nn->rx_bar) { + nn_err(nn, "Failed to map resource %d\n", rx_bar_no); + err = -EIO; + goto err_unmap_tx; + } + } + + nfp_netvf_get_mac_addr(nn); + + err = nfp_net_irqs_alloc(nn); + if (!err) { + nn_warn(nn, "Unable to allocate MSI-X Vectors. Exiting\n"); + err = -EIO; + goto err_unmap_rx; + } + + /* Get ME clock frequency from ctrl BAR + * XXX for now frequency is hardcoded until we figure out how + * to get the value from nfp-hwinfo into ctrl bar + */ + nn->me_freq_mhz = 1200; + + err = nfp_net_netdev_init(nn->netdev); + if (err) + goto err_irqs_disable; + + pci_set_drvdata(pdev, nn); + + nfp_net_info(nn); + nfp_net_debugfs_adapter_add(nn); + + return 0; + +err_irqs_disable: + nfp_net_irqs_disable(nn); +err_unmap_rx: + if (!nn->q_bar) + iounmap(nn->rx_bar); +err_unmap_tx: + if (!nn->q_bar) + iounmap(nn->tx_bar); + else + iounmap(nn->q_bar); +err_netdev_free: + pci_set_drvdata(pdev, NULL); + nfp_net_netdev_free(nn); +err_ctrl_unmap: + iounmap(ctrl_bar); +err_pci_regions: + pci_release_regions(pdev); +err_pci_disable: + pci_disable_device(pdev); + return err; +} + +static void nfp_netvf_pci_remove(struct pci_dev *pdev) +{ + struct nfp_net *nn = pci_get_drvdata(pdev); + + /* Note, the order is slightly different from above as we need + * to keep the nn pointer around till we have freed everything. + */ + nfp_net_debugfs_adapter_del(nn); + + nfp_net_netdev_clean(nn->netdev); + + nfp_net_irqs_disable(nn); + + if (!nn->q_bar) { + iounmap(nn->rx_bar); + iounmap(nn->tx_bar); + } else { + iounmap(nn->q_bar); + } + iounmap(nn->ctrl_bar); + + pci_set_drvdata(pdev, NULL); + + nfp_net_netdev_free(nn); + + pci_release_regions(pdev); + pci_disable_device(pdev); +} + +static struct pci_driver nfp_netvf_pci_driver = { + .name = nfp_net_driver_name, + .id_table = nfp_netvf_pci_device_ids, + .probe = nfp_netvf_pci_probe, + .remove = nfp_netvf_pci_remove, +}; + +static int __init nfp_netvf_init(void) +{ + int err; + + pr_info("%s: NFP VF Network driver, Copyright (C) 2014-2015 Netronome Systems\n", + nfp_net_driver_name); + + nfp_net_debugfs_create(); + err = pci_register_driver(&nfp_netvf_pci_driver); + if (err) { + nfp_net_debugfs_destroy(); + return err; + } + + return 0; +} + +static void __exit nfp_netvf_exit(void) +{ + pci_unregister_driver(&nfp_netvf_pci_driver); + nfp_net_debugfs_destroy(); +} + +module_init(nfp_netvf_init); +module_exit(nfp_netvf_exit); + +MODULE_AUTHOR("Netronome Systems "); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("NFP VF network device driver");