linux-stable.git/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_ism.c, branch v7.2

drm/amd/display: Export symbols for KUnit test modules

2026-06-03T17:42:01+00:00

Add missing EXPORT_IF_KUNIT() calls for amdgpu_dm_psr_set_event,
amdgpu_dm_ism_init, and amdgpu_dm_ism_fini so that the KUnit test
modules can resolve these symbols when built as modules, i.e.,
CONFIG_DRM_AMD_DC_KUNIT_TEST=m.

Fixes: 34f281489976 ("drm/amd/display: Add KUnit tests for amdgpu_dm_psr_set_event")
Fixes: a3142b13fe9f ("drm/amd/display: Add more KUnit tests for amdgpu_dm_ism")
Assisted-by: Copilot:Claude-Opus-4.6
Reviewed-by: Aurabindo Pillai 
Acked-by: Alex Deucher 
Signed-off-by: Alex Hung 
Signed-off-by: Alex Deucher

drm/amd/display: Extract dm_ism_dispatch_next_event

2026-06-03T17:41:44+00:00

[Why & How]
Separate the "should we emit IMMEDIATE?" decision into a pure,
side-effect-free helper so it can be tested in isolation without a
full DRM context.

- dm_ism_dispatch_next_event(current_state, delay_ns, sso_delay_ns)
  returns DM_ISM_EVENT_IMMEDIATE when the state requires an immediate
  follow-up (HYSTERESIS_WAITING with zero delay, OPTIMIZED_IDLE with
  zero SSO delay, or TIMER_ABORTED), and DM_ISM_NUM_EVENTS otherwise.

- Removes the passthrough event parameter that was always
  DM_ISM_NUM_EVENTS at the call site, making the sentinel explicit.

- Drops the now-unused event parameter from dm_ism_dispatch_power_state.

Assisted-by: Copilot:Claude-Sonnet-4.6
Reviewed-by: Ray Wu 
Signed-off-by: Alex Hung 
Signed-off-by: Ray Wu 
Tested-by: Daniel Wheeler 
Signed-off-by: Alex Deucher

drm/amd/display: Add more KUnit tests for amdgpu_dm_ism

2026-06-03T17:41:38+00:00

[Why & How]
Add 8 more KUnit tests:

- dm_ism_get_idle_allow_delay: add a case where filter_entry_count
  exceeds filter_history_size, exercising the max() branch in the
  history_size calculation.
- amdgpu_dm_ism_init: verify all fields are initialised to
  expected values after construction.
- amdgpu_dm_ism_fini: smoke-test cancellation of never-scheduled
  delayed work items paired with a preceding init.
- dm_ism_set_last_idle_ts: verify last_idle_timestamp_ns is
  updated to at least the value captured before the call.
- dm_ism_insert_record: verify index increment and duration
  calculation; verify out-of-bounds index wraps to slot 0.
- dm_ism_trigger_event: verify current_state and previous_state
  are updated on a valid transition and left unchanged on an
  invalid one.

Assisted-by: Copilot:Claude-Sonnet-4.6
Reviewed-by: Ray Wu 
Signed-off-by: Alex Hung 
Signed-off-by: Ray Wu 
Tested-by: Daniel Wheeler 
Signed-off-by: Alex Deucher

drm/amd/display: Replace use of system_unbound_wq with system_dfl_wq

2026-05-27T14:26:36+00:00

This patch continues the effort to refactor workqueue APIs, which has begun
with the changes introducing new workqueues and a new alloc_workqueue flag:

   commit 128ea9f6ccfb ("workqueue: Add system_percpu_wq and system_dfl_wq")
   commit 930c2ea566af ("workqueue: Add new WQ_PERCPU flag")

The point of the refactoring is to eventually alter the default behavior of
workqueues to become unbound by default so that their workload placement is
optimized by the scheduler.

Before that to happen, workqueue users must be converted to the better named
new workqueues with no intended behaviour changes:

   system_wq -> system_percpu_wq
   system_unbound_wq -> system_dfl_wq

This way the old obsolete workqueues (system_wq, system_unbound_wq) can be
removed in the future.

Cc: Ray Wu 
Cc: Harry Wentland 
Cc: Leo Li 
Cc: Rodrigo Siqueira 
Link: https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/
Suggested-by: Tejun Heo 
Signed-off-by: Marco Crivellari 
Signed-off-by: Alex Deucher

drm/amd/display: Add KUnit test for ISM functions

2026-05-19T15:54:37+00:00

Add KUnit tests for three static functions in amdgpu_dm_ism.c:
dm_ism_next_state, dm_ism_get_sso_delay, and
dm_ism_get_idle_allow_delay.

The 32 test cases cover the full FSM transition table,
SSO delay calculation with various timings, and
hysteresis-based idle allow delay including circular
buffer wraparound and old history cutoff logic.

Conditionally remove static linkage and export the three
functions under CONFIG_DRM_AMD_DC_KUNIT_TEST so the test
module can call them.

Assisted-by: Copilot:Claude-Opus-4.6
Reviewed-by: Harry Wentland 
Signed-off-by: Alex Hung 
Signed-off-by: Ivan Lipski 
Tested-by: Dan Wheeler 
Signed-off-by: Alex Deucher

drm/amd/display: Use lockdep_assert_held() for dc_lock check

2026-05-19T15:45:41+00:00

[Why]
mutex_is_locked() only tells whether *some* task holds the mutex, not
the current one, so the existing ASSERT can silently pass when the
caller violates the contract.

[How]
Use the kernel's lockdep debugging utility (include/linux/lockdep.h)
and replace ASSERT(mutex_is_locked(&dm->dc_lock)) with
lockdep_assert_held(&dm->dc_lock), which checks the current task's
held-lock stack.

Reviewed-by: Sun peng (Leo) Li 
Signed-off-by: Ray Wu 
Signed-off-by: Ivan Lipski 
Tested-by: Dan Wheeler 
Signed-off-by: Alex Deucher

drm/amd/display: Fix ISM dc_lock deadlock during suspend

2026-05-19T15:45:40+00:00

[Why]
System hang observed during suspend/resume while video is playing.
amdgpu_dm_ism_disable() is called under dc_lock and waits for ISM
delayed work via disable_delayed_work_sync(). The work handlers
themselves take dc_lock, producing an ABBA deadlock when a worker is
in flight at suspend time.

[How]
Split the disable path into two phases with opposite locking
contracts:
  1. amdgpu_dm_ism_disable() -- quiesces workers, must NOT hold
     dc_lock.
  2. amdgpu_dm_ism_force_full_power() (new) -- drives the ISM FSM
     back to FULL_POWER_RUNNING, must hold dc_lock.

Reviewed-by: Sun peng (Leo) Li 
Signed-off-by: Ray Wu 
Signed-off-by: Ivan Lipski 
Tested-by: Dan Wheeler 
Signed-off-by: Alex Deucher

drm/amd/display: Undo accidental fix revert in amdgpu_dm_ism.c

2026-04-21T20:49:04+00:00

[Why]

Pausing DPM power profiles during static screen caused a bunch of
audio/performance/clock issues that were addressed in this fix:
'commit 1412482b7143 ("Revert "drm/amd/display: pause the workload setting in dm"")'

This logic in function amdgpu_dm_crtc_vblank_control_worker() was moved
to amdgpu_dm_ism.c, but the fix was lost in the process.

[How]

Reapply the fix to amdgpu_dm_ism.c

Fixes: 754003486c3c ("drm/amd/display: Add Idle state manager(ISM)")
Reviewed-by: Alex Deucher 
Reviewed-by: Mario Limonciello (AMD) 
Signed-off-by: Leo Li 
Signed-off-by: Alex Deucher

drm/amd/display: Add power module on Linux

2026-04-17T19:41:16+00:00

[Why & How]
Refactors dm to utilize the power module for managing
replay, PSR, and backlight control functionalities.

Key changes:
- Introduced replay / PSR events to enable / disable replay / PSR.
- Implemented replay rate control and power option
- Refactored backlight control by using the power module.
- Enhanced handling of VRR within replay and PSR logic.

Reviewed-by: Leo Li 
Signed-off-by: Ray Wu 
Signed-off-by: Leo Li 
Signed-off-by: Chenyu Chen 
Signed-off-by: Alex Deucher

drm/amd/display: Fix ISM teardown crash from NULL dc dereference

2026-04-17T19:40:04+00:00

The Idle State Manager (ISM) uses delayed work to apply display idle
optimizations later, instead of immediately. This helps avoid rapid idle
transitions that can hurt power or performance.

A crash was seen during driver teardown. The system boots normally and
the driver loads successfully. Later, when the GPU is being stopped, the
log shows:

  amdgpu 0000:0e:00.0: finishing device.
  Workqueue: events_unbound dm_ism_sso_delayed_work_func [amdgpu]

After this, delayed ISM work still runs and reaches:

  dm_ism_sso_delayed_work_func()
    -> amdgpu_dm_ism_commit_event()
    -> dm_ism_commit_idle_optimization_state()
    -> dc_allow_idle_optimizations_internal()

The crash report showed:
  KASAN: null-ptr-deref in range [0x690-0x697]

Signature:
	[22601.113316] KASAN: null-ptr-deref in range [0x0000000000000690-0x0000000000000697]
	...
	[22601.113368] Workqueue: events_unbound dm_ism_sso_delayed_work_func [amdgpu]
	[22601.113930] RIP: 0010:dc_allow_idle_optimizations_internal+0xa6/0xc40 [amdgpu]
	...
	[22601.114491] RDX: dffffc0000000000 RSI: 0000000000000000 RDI: 0000000000000690
	...
	[22601.114561] Call Trace:
	[22601.114566]  
	[22601.114572]  ? srso_alias_return_thunk+0x5/0xfbef5
	[22601.114582]  ? update_load_avg+0x1b6/0x20b0
	[22601.114593]  ? __pfx_dc_allow_idle_optimizations_internal+0x10/0x10 [amdgpu]
	[22601.114932]  ? psi_group_change+0x4ed/0x8d0
	[22601.114942]  dm_ism_commit_idle_optimization_state+0x214/0x570 [amdgpu]
	[22601.115268]  amdgpu_dm_ism_commit_event+0xe1d/0x15a0 [amdgpu]
	[22601.115588]  ? srso_alias_return_thunk+0x5/0xfbef5
	[22601.115595]  ? __kasan_check_write+0x18/0x20
	[22601.115603]  ? srso_alias_return_thunk+0x5/0xfbef5
	[22601.115610]  ? mutex_lock+0x83/0xc0
	[22601.115620]  dm_ism_sso_delayed_work_func+0x64/0x90 [amdgpu]

GDB resolved dc_allow_idle_optimizations_internal+0xa6 to:

  struct dc_state *context = dc->current_state;

The matching disassembly showed:

  mov %rdi, %r12
  mov 0x690(%r12), %r13

where r12 holds the dc pointer. A GDB layout dump of struct dc showed:

  /* 1680 | 8 */ struct dc_state *current_state;

Since 1680 decimal is 0x690, this confirms that current_state is at
offset 0x690. The faulting access was effectively:

  dc + 0x690

which indicates that dc was NULL at the time of dereference.

This shows that ISM work can still run during teardown after dc has
been cleared.

ISM is not expected to run after dc is destroyed. Fix this by disabling
ISM under dc_lock in amdgpu_dm_fini() before dc_destroy(), ensuring no
further ISM work runs after dc teardown.

Also add ASSERT(dm->dc) in amdgpu_dm_ism_commit_event() to enforce this
invariant, and ASSERT(mutex_is_locked(&dm->dc_lock)) in
amdgpu_dm_ism_disable() to clarify the locking requirement.

Fixes: 754003486c3c ("drm/amd/display: Add Idle state manager(ISM)")
Suggested-by: Leo Li 
Cc: Ray Wu 
Cc: Roman Li 
Cc: Alex Hung 
Cc: Tom Chung 
Cc: Harry Wentland 
Cc: Aurabindo Pillai 
Cc: Mario Limonciello (AMD) 
Signed-off-by: Srinivasan Shanmugam 
Reviewed-by: Leo Li 
Signed-off-by: Alex Deucher