STRUCT INTEL_GUC_STATE_CAPTURE(9)

✨ struct intel_guc_state_capture — Kernel API Manual ✨

okay so bestie, this is the kernel API manual for struct intel_guc_state_capture, and we're gonna get into it 💅 this one lives in the Kernel API Manual, and honestly it's giving "backstage pass to the GuC error-capture machine" 🎀

---

🌸 NAME 🌸

struct intel_guc_state_capture — that's the struct, that's the name, no notes 💖

---

🦋 SYNOPSIS 🦋

struct intel_guc_state_capture {
    const struct __guc_mmio_reg_descr_group *reglists;
    struct __guc_mmio_reg_descr_group *extlists;
    struct __guc_capture_ads_cache ads_cache[GUC_CAPTURE_LIST_INDEX_MAX]
[GUC_CAPTURE_LIST_TYPE_MAX] [GUC_MAX_ENGINE_CLASSES];
    void *ads_null_cache;
    struct list_head cachelist;
  #define PREALLOC_NODES_MAX_COUNT (3 * GUC_MAX_ENGINE_CLASSES *
GUC_MAX_INSTANCES_PER_CLASS);
  #define PREALLOC_NODES_DEFAULT_NUMREGS 64;
    int max_mmio_per_node;
    struct list_head outlist;
 };

like, look at her. iconic struct definition, no cap 👑

---

🎀 Members 🎀

• reglists — the static table of register lists that gets used for error-capture state. it just sits there being static and reliable, we love her 🫶
• extlists — the allocated table of steered register lists used for error-capture state. allocated, not static, okay? she's built at runtime, main character energy ✨

NOTE: steered registers have multiple instances depending on the HW configuration (slices or dual-sub-slices) and thus depends on HW fuses discovered at startup 💀 like, the hardware literally tells you at boot how many of these you get, and it's not always the same, so yeah 🙈

• ads_cache — cached register lists that is ADS format ready 🔥 pre-formatted, pre-cached, ready to go, she ate
• ads_null_cache — the ADS null cache. that's the whole description bestie, it's a null cache, it's doing null cache things 🧚
• cachelist — the pool of pre-allocated nodes for error capture output 🌈 and okay, here's the why, because this part is actually important:

we need this pool of pre-allocated nodes because we cannot dynamically allocate new nodes when receiving the G2H notification — because the event handlers for all G2H event-processing is called by the ct processing worker queue, and when that queue is being processed, there is no absolute guarantee that we are not in the midst of a GT reset operation (which doesn't allow allocations) 😭 so like. no allocations. not allowed. pre-allocated pool it is. periodt.

• max_mmio_per_node — max MMIO per node. yes. that's it. that's the function 💅
• outlist — pool of pre-allocated nodes for error capture output 👑

a linked list of parsed GuC error-capture output data before reporting with formatting via i915_gpu_coredump. each node in this linked list shall contain a single engine-capture including global, engine-class and engine-instance register dumps as per guc_capture_parsed_output_node ⭐ like, one node = one full engine capture, with all three layers of register dump, all cozy in there together 🍓

---

💜 Description 💜

internal context of the intel_guc_capture module. literally just the internal context. that's her job. that's what she does. and she does it well 🎀

---

🌟 SEE ALSO 🌟

• Kernel file ./drivers/gpu/drm/i915/gt/uc/guc_capture_fwif.h — go read the header bestie, iykyk 📖✨

---

so yeah, that's the whole page 😭💖 no extra lore, no hidden options, just struct fields and the very good reasons they exist.

okay and before i go — this one came to you from ./drivers/gpu/drm/i915/gt/uc and it's dated October 2026, so like, mark your calendars or whatever 🗓️ love u, bye 🫶💅🌸