dm_test_ism_...e_defers_sso(9)

💅 dm_test_ism_...e_defers_sso(9) — from the Kernel API Manual, bestie 🎀

okay so before we even start: yes, the title up there is literally cut off mid-word (dm_test_ism_...e_defers_sso), that's just how the Kernel API Manual rolls, it's giving mysterious ✨. Anyway. Welcome to the group chat, grab a strawberry 🍓, we're learning about a KUnit test today.

---

🌸 NAME 🌸

dm_test_ism_dispatch_optimized_idle_defers_sso — the test that covers the OPTIMIZED_IDLE dispatch path without touching DC at all. Not even a little. Zero DC contact, we're social distancing from DC fr fr 💀.

---

⭐ SYNOPSIS ⭐

void dm_test_ism_dispatch_optimized_idle_defers_sso (struct kunit *test
);

that's it. one function. one argument. iconic, honestly 👑

---

🦋 ARGUMENTS 🦋

• test — the KUnit test context. it's the lil' sandbox our test lives in, the standard KUnit vibe-passed-in-by-the-framework situation. main character energy 🫶

---

💜 DESCRIPTION 💜

okay so here's the tea ☕:

when sso_num_frames < filter_num_frames, the OPTIMIZED_IDLE current-state arm skips the idle-optimization commit and only schedules the SSO worker. it does not do the commit. it defers. hence the name, bestie, the whole thing makes sense now 🔥

and the way this test wakes that arm up? it drives the state machine HYSTERESIS_WAITING → OPTIMIZED_IDLE via TIMER_ELAPSED. so one moment we're waiting, the next the timer's elapsed and we've slid into optimized idle, and that's exactly what pokes the arm we care about 👀

so what does that actually exercise? three lil' bits of code:

• get_sso_delay — gets the delay, obviously 🕐
• the skip branch — the branch that skips, yes I'm being redundant, it's fine ✨
• mod_delayed_work — schedules the delayed work, slay

and all of that with no DC access whatsoever. it's a clean, self-contained unit test. no hardware, no DC, no stress. just vibes and state transitions 🌈

lowkey the point of this test is that the arm defers the SSO work instead of committing an idle optimization — and we prove it happens safely with zero DC involvement. no cap 💅

---

✨ SEE ALSO ✨

so you want the whole friend group? bet, here's everybody 🎀:

• the Kernel file itself (iykyk): ./drivers/gpu/drm/amd/display/amdgpu_dm/tests/amdgpu_dm_ism_test.c
• dm_test_ism_idle_delay_entry_count_exceeds_history_size(9)
• dm_test_ism_init_sets_initial_state(9)
• dm_test_ism_fini_after_init(9)
• dm_test_ism_set_last_idle_ts_updates_timestamp(9)
• dm_test_ism_insert_record_basic(9)
• dm_test_ism_insert_record_wraps_around(9)
• dm_test_ism_trigger_event_valid_transition(9)
• dm_test_ism_trigger_event_invalid_transition(9)
• dm_test_dispatch_next_event_hyst_wait_no_delay(9)
• dm_test_dispatch_next_event_hyst_wait_with_delay(9)
• dm_test_dispatch_next_event_opt_idle_no_sso_delay(9)
• dm_test_dispatch_next_event_opt_idle_with_sso_delay(9)
• dm_test_dispatch_next_event_timer_aborted(9)
• dm_test_dispatch_next_event_no_action_state(9)
• dm_test_ism_commit_event_no_state(9)
• dm_test_ism_commit_event_cursor_transition(9)
• dm_test_ism_commit_event_invalid_event(9)
• dm_test_ism_force_full_power(9)
• dm_test_ism_disable_enable_cycle(9)
• dm_test_ism_dispatch_hysteresis_schedule_and_cancel(9)

twenty-two references, all accounted for, not one dropped. we keep every single one 💖

---

and that's the page! this cutie came out of ./drivers/gpu/drm/amd/display/... in October 2026, under the dm_test_ism_...e_defers_sso(9) name in the Kernel API Manual. go forth and dispatch, bestie — see you in the next manpage 🧚💜