[PATCH 21/33] drm/amd/display: Fix race condition when turning off an output alone
Alan Liu
HaoPing.Liu at amd.com
Wed Jul 12 17:11:25 UTC 2023
From: Alan Liu <haoping.liu at amd.com>
[Why]
When 2 threads are doing commit_tail parallelly, one thread could
commit new streams to dc state but another thread remove it from dc
right away.
[How]
If we don't have new dm state change from commit_check, then we should
not call dc_commit_streams() in commit_tail. A new function
amdgpu_dm_commit_streams() is introduced to refator dc_commit_stream()
adjacent code and fix this issue.
Reviewed-by: Harry Wentland <harry.wentland at amd.com>
Acked-by: Alan Liu <haoping.liu at amd.com>
Signed-off-by: Alan Liu <haoping.liu at amd.com>
---
.../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 114 +++++++++---------
1 file changed, 55 insertions(+), 59 deletions(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 88057641972d..1e1a38014475 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -7937,7 +7937,6 @@ static inline uint32_t get_mem_type(struct drm_framebuffer *fb)
}
static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
- struct dc_state *dc_state,
struct drm_device *dev,
struct amdgpu_display_manager *dm,
struct drm_crtc *pcrtc,
@@ -8417,52 +8416,17 @@ static void amdgpu_dm_crtc_copy_transient_flags(struct drm_crtc_state *crtc_stat
stream_state->mode_changed = drm_atomic_crtc_needs_modeset(crtc_state);
}
-/**
- * amdgpu_dm_atomic_commit_tail() - AMDgpu DM's commit tail implementation.
- * @state: The atomic state to commit
- *
- * This will tell DC to commit the constructed DC state from atomic_check,
- * programming the hardware. Any failures here implies a hardware failure, since
- * atomic check should have filtered anything non-kosher.
- */
-static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
+static void amdgpu_dm_commit_streams(struct drm_atomic_state *state,
+ struct dc_state *dc_state)
{
struct drm_device *dev = state->dev;
struct amdgpu_device *adev = drm_to_adev(dev);
struct amdgpu_display_manager *dm = &adev->dm;
- struct dm_atomic_state *dm_state;
- struct dc_state *dc_state = NULL, *dc_state_temp = NULL;
- u32 i, j;
struct drm_crtc *crtc;
struct drm_crtc_state *old_crtc_state, *new_crtc_state;
- unsigned long flags;
- bool wait_for_vblank = true;
- struct drm_connector *connector;
- struct drm_connector_state *old_con_state, *new_con_state;
struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
- int crtc_disable_count = 0;
bool mode_set_reset_required = false;
- int r;
-
- trace_amdgpu_dm_atomic_commit_tail_begin(state);
-
- r = drm_atomic_helper_wait_for_fences(dev, state, false);
- if (unlikely(r))
- DRM_ERROR("Waiting for fences timed out!");
-
- drm_atomic_helper_update_legacy_modeset_state(dev, state);
- drm_dp_mst_atomic_wait_for_dependencies(state);
-
- dm_state = dm_atomic_get_new_state(state);
- if (dm_state && dm_state->context) {
- dc_state = dm_state->context;
- } else {
- /* No state changes, retain current state. */
- dc_state_temp = dc_create_state(dm->dc);
- ASSERT(dc_state_temp);
- dc_state = dc_state_temp;
- dc_resource_state_copy_construct_current(dm->dc, dc_state);
- }
+ u32 i;
for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state,
new_crtc_state, i) {
@@ -8561,24 +8525,22 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
}
} /* for_each_crtc_in_state() */
- if (dc_state) {
- /* if there mode set or reset, disable eDP PSR */
- if (mode_set_reset_required) {
- if (dm->vblank_control_workqueue)
- flush_workqueue(dm->vblank_control_workqueue);
+ /* if there mode set or reset, disable eDP PSR */
+ if (mode_set_reset_required) {
+ if (dm->vblank_control_workqueue)
+ flush_workqueue(dm->vblank_control_workqueue);
- amdgpu_dm_psr_disable_all(dm);
- }
+ amdgpu_dm_psr_disable_all(dm);
+ }
- dm_enable_per_frame_crtc_master_sync(dc_state);
- mutex_lock(&dm->dc_lock);
- WARN_ON(!dc_commit_streams(dm->dc, dc_state->streams, dc_state->stream_count));
+ dm_enable_per_frame_crtc_master_sync(dc_state);
+ mutex_lock(&dm->dc_lock);
+ WARN_ON(!dc_commit_streams(dm->dc, dc_state->streams, dc_state->stream_count));
- /* Allow idle optimization when vblank count is 0 for display off */
- if (dm->active_vblank_irq_count == 0)
- dc_allow_idle_optimizations(dm->dc, true);
- mutex_unlock(&dm->dc_lock);
- }
+ /* Allow idle optimization when vblank count is 0 for display off */
+ if (dm->active_vblank_irq_count == 0)
+ dc_allow_idle_optimizations(dm->dc, true);
+ mutex_unlock(&dm->dc_lock);
for_each_new_crtc_in_state(state, crtc, new_crtc_state, i) {
struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
@@ -8598,6 +8560,44 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
acrtc->otg_inst = status->primary_otg_inst;
}
}
+}
+
+/**
+ * amdgpu_dm_atomic_commit_tail() - AMDgpu DM's commit tail implementation.
+ * @state: The atomic state to commit
+ *
+ * This will tell DC to commit the constructed DC state from atomic_check,
+ * programming the hardware. Any failures here implies a hardware failure, since
+ * atomic check should have filtered anything non-kosher.
+ */
+static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
+{
+ struct drm_device *dev = state->dev;
+ struct amdgpu_device *adev = drm_to_adev(dev);
+ struct amdgpu_display_manager *dm = &adev->dm;
+ struct dm_atomic_state *dm_state;
+ struct dc_state *dc_state = NULL;
+ u32 i, j;
+ struct drm_crtc *crtc;
+ struct drm_crtc_state *old_crtc_state, *new_crtc_state;
+ unsigned long flags;
+ bool wait_for_vblank = true;
+ struct drm_connector *connector;
+ struct drm_connector_state *old_con_state, *new_con_state;
+ struct dm_crtc_state *dm_old_crtc_state, *dm_new_crtc_state;
+ int crtc_disable_count = 0;
+
+ trace_amdgpu_dm_atomic_commit_tail_begin(state);
+
+ drm_atomic_helper_update_legacy_modeset_state(dev, state);
+ drm_dp_mst_atomic_wait_for_dependencies(state);
+
+ dm_state = dm_atomic_get_new_state(state);
+ if (dm_state && dm_state->context) {
+ dc_state = dm_state->context;
+ amdgpu_dm_commit_streams(state, dc_state);
+ }
+
for_each_oldnew_connector_in_state(state, connector, old_con_state, new_con_state, i) {
struct dm_connector_state *dm_new_con_state = to_dm_connector_state(new_con_state);
struct amdgpu_crtc *acrtc = to_amdgpu_crtc(dm_new_con_state->base.crtc);
@@ -8875,8 +8875,7 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
if (dm_new_crtc_state->stream)
- amdgpu_dm_commit_planes(state, dc_state, dev,
- dm, crtc, wait_for_vblank);
+ amdgpu_dm_commit_planes(state, dev, dm, crtc, wait_for_vblank);
}
/* Update audio instances for each connector. */
@@ -8931,9 +8930,6 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
for (i = 0; i < crtc_disable_count; i++)
pm_runtime_put_autosuspend(dev->dev);
pm_runtime_mark_last_busy(dev->dev);
-
- if (dc_state_temp)
- dc_release_state(dc_state_temp);
}
static int dm_force_atomic_commit(struct drm_connector *connector)
--
2.34.1
More information about the amd-gfx
mailing list