STRUCT IWL_MEI(9)

✨ struct iwl_mei ✨

okay so this is from the Kernel API Manual and it's literally the struct that holds all the private data for iwl_mei 💅 (yes the original NAME line says "private date" — we're choosing to believe they meant data, but we're reporting the typo honestly because iykyk 🫶).

Think of it as the main character's group chat: every wait queue, every work item, every bool, all vibing in one struct. Let's get into it 🔥

🌸 SYNOPSIS 🌸

struct iwl_mei {
    wait_queue_head_t get_nvm_wq;
    struct work_struct send_csa_msg_wk;
    wait_queue_head_t get_ownership_wq;
    struct iwl_mei_shared_mem_ptrs shared_mem;
    struct mei_cl_device *cldev;
    struct iwl_mei_nvm *nvm;
    struct iwl_mei_filters *filters;
    bool got_ownership;
    bool amt_enabled;
    bool csa_throttled;
    bool csme_taking_ownership;
    bool link_prot_state;
    bool device_down;
    struct delayed_work csa_throttle_end_wk;
    wait_queue_head_t pldr_wq;
    bool pldr_active;
    spinlock_t data_q_lock;
    struct work_struct netdev_work;
    struct delayed_work ownership_dwork;
    atomic_t sap_seq_no;
    atomic_t seq_no;
    struct dentry *dbgfs_dir;
};

copy it exactly bestie, don't be cute with it 🙈

🎀 Members 🎀

• get_nvm_wq — the wait queue for the get_nvm flow. it sits here and waits. iconic behaviour honestly.
• send_csa_msg_wk — used to defer the transmission of the CHECK_SHARED_AREA message. the whole reason it exists? so we can send CHECK_SHARED_AREA from atomic contexts. no cap, that's the flex ⚡
• get_ownership_wq — the wait queue for the get_ownership_flow. yep, another one. she queues, she waits, she slays 💅
• shared_mem — the memory that is shared between CSME and the host 🫶
• cldev — the pointer to the MEI client device
• nvm — the data returned by the CSME for the NVM
• filters — the filters sent by CSME
• got_ownership — true if we own the device 👑
• amt_enabled — true if CSME has wireless enabled
• csa_throttled — when true, we can't send CHECK_SHARED_AREA over the MEI bus, but rather need to wait until send_csa_msg_wk runs. it's giving "hold please, I'm busy" 😭
• csme_taking_ownership — true when CSME is taking ownership. used to remember to send CSME_OWNERSHIP_CONFIRMED when the driver completes its down flow. she remembers. she always remembers. 🧠
• link_prot_state — true when we are in link protection PASSIVE
• device_down — true if the device is down. used to remember to send CSME_OWNERSHIP_CONFIRMED when the driver is already down. same energy as above, different bool fr fr
• csa_throttle_end_wk — used when csa_throttled is true. yes it's a delayed work. yes it's here for the throttle ✨
• pldr_wq — the wait queue for PLDR flow
• pldr_active — PLDR flow is in progress 🌀
• data_q_lock — protects the access to the data queues which are accessed without the mutex. lowkey important, don't sleep on the lock 🔒
• netdev_work — used to defer registering and unregistering of the netdev to avoid taking the rtnl lock in the SAP messages handlers. very "I'll do it later so I don't deadlock" of her 🙈
• ownership_dwork — used to re-ask for NIC ownership after ownership was taken by CSME or when a previous ownership request failed. she's persistent, she's a delayed work, she's not giving up 💖
• sap_seq_no — the sequence number for the SAP messages
• seq_no — the sequence number for the SAP messages. yes. the exact same description. no we didn't mix them up, that's what it says 🫠
• dbgfs_dir — the debugfs dir entry ⭐

🦋 SEE ALSO 🦋

• Kernel file ./drivers/net/wireless/intel/iwlwifi/mei/main.c struct iwl_mei_cache(9)

ok that's the whole struct, go read that main.c file if you want the full lore 🍓

this masterpiece is dated October 2026, lives at ./drivers/net/wireless/intel/i... (truncated by the man page itself, not by me 💀), and the whole thing is filed under struct iwl_mei(9). mwah, periodt 💜