STRUCT PCC_CHAN_INFO(9)

📦 version 7.3.1, in nixpkgs unstable

👑 struct pcc_chan_info(9) 👑

okay so this is a page straight out of the Kernel API Manual, and bestie, it's literally the whole personality of one PCC channel in a single struct. one (1) channel. that's it. that's the vibe. ✨

🌸 NAME 🌸

struct pcc_chan_info — the PCC channel specific information. all the little details that belong to one channel, no sharing, no cap. 🎀

🦋 SYNOPSIS 🦋

struct pcc_chan_info {
    struct pcc_mbox_chan chan;
    struct pcc_chan_reg db;
    struct pcc_chan_reg plat_irq_ack;
    struct pcc_chan_reg cmd_complete;
    struct pcc_chan_reg cmd_update;
    struct pcc_chan_reg error;
    int plat_irq;
    u8 type;
    unsigned int plat_irq_flags;
    bool chan_in_use;
 };

look at her. ten members. every single one iconic. 💅

💖 Members 💖

okay so let's go through the whole cast, one by one, because they all matter:

  • chan — the PCC channel information, and yes it comes with the Shared Memory Region info too. the whole package. 🫶
  • db — the PCC register bundle for the doorbell register. ring ring, bestie 🔔
  • plat_irq_ack — the PCC register bundle for the platform interrupt acknowledge register. acknowledging, but make it platform. ✨
  • cmd_complete — the PCC register bundle for the command complete check register. did it finish? this one knows. 👀
  • cmd_update — the PCC register bundle for the command complete update register. the "update" half of the complete situation. 🔥
  • error — the PCC register bundle for the error status register. the one nobody wants a notification from. 💀
  • plat_irq — the platform interrupt. just an int. living its simple life. 🌈
  • type — the PCC subspace type, chillin' as a u8. ⭐
  • plat_irq_flags — the platform interrupt flags. unspecified unsigned int energy. 🧚
  • chan_in_use — okay this one's the main character, so listen up: it's a lockless flag, and it's used by type 3 initiator subspaces to filter platform interrupts. only one transfer can occur at a time, BUT the interrupt handler may sample the flag on another CPU — so all accesses must use READ_ONCE or WRITE_ONCE. not optional, not a suggestion, literally must. 😭 other subspace types do not test it at all, so if that's you, you can lowkey ignore it. 🙈
bestie I'm serious about the READ_ONCE / WRITE_ONCE thing. lockless means the compiler will absolutely betray you if you let it. that's giving undefined behaviour, and undefined behaviour is not the aesthetic we're going for. 💀

🔥 SEE ALSO 🔥

the whole friend group, all in the kernel file ./drivers/mailbox/pcc.c:

aaaand that's the page, bestie 🎀 this one comes to you from ./drivers/mailbox, dated October 2026 — go forth and don't forget your WRITE_ONCE 💜🦋

📦 where this page comes from