STRUCT NFP_NET_DP(9)

🌸 NAME 🌸

struct nfp_net_dp — the NFP network device datapath data structure, aka the main character of your NIC's whole life 💅

So like, quick vibe check: this is from the kernel API manual, page struct nfp_net_dp(9). Grab a boba, bestie, we're going in 🧋

✨ SYNOPSIS ✨

struct nfp_net_dp {
    struct device *dev;
    struct net_device *netdev;
    u8 is_vf:1;
    u8 chained_metadata_format:1;
    u8 ktls_tx:1;
    u8 rx_dma_dir;
    u8 rx_offset;
    u32 rx_dma_off;
    u32 ctrl;
    u32 ctrl_w1;
    u32 fl_bufsz;
    struct bpf_prog *xdp_prog;
    struct nfp_net_tx_ring *tx_rings;
    struct nfp_net_rx_ring *rx_rings;
    u8 __iomem *ctrl_bar;
    const struct nfp_dp_ops *ops;
    u64 *txrwb;
    dma_addr_t txrwb_dma;
    unsigned int txd_cnt;
    unsigned int rxd_cnt;
    unsigned int num_r_vecs;
    unsigned int num_tx_rings;
    unsigned int num_stack_tx_rings;
    unsigned int num_rx_rings;
    unsigned int mtu;
    struct xsk_buff_pool **xsk_pools;
};

No cap, that's the whole struct. Every field, every bitfield, exactly as it lives in the kernel 🫶

💖 Members 💖

Okay so here's the group chat roll call — every field gets its moment 🔥

• dev — the backpointer to struct device. It's giving "who's your parent" energy.
• netdev — backpointer to the net_device structure. Your classic "go ask my netdev" situation.
• is_vf — is the driver attached to a VF? One bit, huge implications 💀
• chained_metadata_format — Firemware will use the new metadata format. (Yeah that's the original's spelling bestie — the firmware is fine, we promise 🌸)
• ktls_tx — is kTLS TX enabled? Yes or no, one single bit, iconic.
• rx_dma_dir — the mapping direction for RX buffers. Direction matters, always 🧭
• rx_offset — offset in the RX buffers where the packet data actually starts.
• rx_dma_off — offset at which DMA packets (for XDP headroom). Headroom is self-care ✨
• ctrl — local copy of the control register/word.
• ctrl_w1 — local copy of the control register/word1. (Yes there's a second one. There's always a second one.)
• fl_bufsz — the currently configured size of the freelist buffers.
• xdp_prog — the installed XDP program. Your eBPF bestie is in the building 👑
• tx_rings — an array of pre-allocated TX ring structures.
• rx_rings — an array of pre-allocated RX ring structures. Matching set, slay 🎀
• ctrl_bar — pointer to the mapped control BAR.
• ops — the callbacks and parameters for this vNIC's NFD version.
• txrwb — TX pointer write back area, indexed by queue id.
• txrwb_dma — the TX pointer write back area's DMA address.
• txd_cnt — size of the TX ring in number of min size packets.
• rxd_cnt — size of the RX ring in number of min size packets.
• num_r_vecs — number of used ring vectors.
• num_tx_rings — the currently configured number of TX rings.
• num_stack_tx_rings — number of TX rings used by the stack (not XDP). Stack rings stay in their own lane, periodt.
• num_rx_rings — the currently configured number of RX rings.
• mtu — device MTU. Big packets, big energy.
• xsk_pools — XSK buffer pools, max_r_vecs in size (for AF_XDP). Serving AF_XDP looks 💅

🦋 SEE ALSO 🦋

Kernel file ./drivers/net/ethernet/netronome/nfp/nfp_net.h, plus these absolute legends:

• struct nfp_net_tx_ring(9)
• struct nfp_net_rx_buf(9)
• struct nfp_net_xsk_rx_buf(9)
• struct nfp_net_rx_ring(9)
• struct nfp_net_r_vector(9)
• struct nfp_net(9)
• nfp_qcp_rd_ptr_add(9)
• nfp_qcp_wr_ptr_add(9)
• nfp_qcp_rd_ptr_read(9)
• nfp_qcp_wr_ptr_read(9)

Go read them. All of them. For real, fr fr — iykyk 📚✨

and that's a wrap bestie!! 💜 This page came from the source/version line ./drivers/net/ethernet/netrono... (yep, it's genuinely truncated like that in the original — not me being lazy 💀) and it's dated October 2026. Stay hydrated, keep your rings pre-allocated, and I'll see you in the next (9) page 🎀🫶💖