On 2017-10-13 04:36 PM, Andrey Grodzovsky wrote:
>
>
> On 10/13/2017 03:29 PM, sunpeng.li at amd.com wrote:
>> From: "Leo (Sunpeng) Li" <sunpeng.li at amd.com>
>>
>> Use the correct for_each_new/old_* iterators instead of for_each_*
>>
>> The following functions were considered:
>>
>> amdgpu_dm_find_first_crtc_matching_connector: use for_each_new
>> Â Â Â Â - Old from_state_var flag was always choosing the new state
>>
>> amdgpu_dm_display_resume: use for_each_new
>> Â Â Â Â - drm_atomic_helper_duplicate_state is called during suspend to
>> Â Â Â Â Â Â cache the state
>> Â Â Â Â - It sets 'state' within the state triplet to 'new_state'
>>
>> amdgpu_dm_commit_planes: use for_each_old
>> Â Â Â Â - Called after the state was swapped (via atomic commit tail)
>>
>> amdgpu_dm_atomic_commit: use for_each_new
>> Â Â Â Â - Called before the state is swapped
>>
>> amdgpu_dm_atomic_commit_tail: use for_each_old
>> Â Â Â Â - Called after the state was swapped
>>
>> dm_update_crtcs_state: use for_each_new
>> Â Â Â Â - Called before the state is swapped (via atomic check)
>>
>> amdgpu_dm_atomic_check: use for_each_new
>> Â Â Â Â - Called before the state is swapped
>>
>> v2: Split out typo fixes to a new patch.
>>
>> v3: Say "functions considered" instead of "affected functions". The
>> Â Â Â Â latter implies that changes are made to each.
>>
>> Signed-off-by: Leo (Sunpeng) Li <sunpeng.li at amd.com>
>> ---
>> Â drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 32
>> ++++++++---------------
>> Â drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h |Â 3 +--
>> Â 2 files changed, 12 insertions(+), 23 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 9bfe1f9..cc024ab 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>> @@ -570,23 +570,15 @@ static int dm_suspend(void *handle)
>> Â struct amdgpu_dm_connector
>> *amdgpu_dm_find_first_crct_matching_connector(
>> Â Â Â Â Â struct drm_atomic_state *state,
>> -Â Â Â struct drm_crtc *crtc,
>> -Â Â Â bool from_state_var)
>> +Â Â Â struct drm_crtc *crtc)
>> Â {
>> Â Â Â Â Â uint32_t i;
>> Â Â Â Â Â struct drm_connector_state *conn_state;
>> Â Â Â Â Â struct drm_connector *connector;
>> Â Â Â Â Â struct drm_crtc *crtc_from_state;
>> -Â Â Â for_each_new_connector_in_state(
>> -Â Â Â Â Â Â Â state,
>> -Â Â Â Â Â Â Â connector,
>> -Â Â Â Â Â Â Â conn_state,
>> -Â Â Â Â Â Â Â i) {
>> -Â Â Â Â Â Â Â crtc_from_state =
>> -Â Â Â Â Â Â Â Â Â Â Â from_state_var ?
>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â conn_state->crtc :
>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â connector->state->crtc;
>> +Â Â Â for_each_new_connector_in_state(state, connector, conn_state, i) {
>> +Â Â Â Â Â Â Â crtc_from_state = conn_state->crtc;
>> Â Â Â Â Â Â Â Â Â if (crtc_from_state == crtc)
>> Â Â Â Â Â Â Â Â Â Â Â Â Â return to_amdgpu_dm_connector(connector);
>> @@ -3890,7 +3882,7 @@ static void amdgpu_dm_commit_planes(struct
>> drm_atomic_state *state,
>> Â Â Â Â Â unsigned long flags;
>> Â Â Â Â Â /* update planes when needed */
>> -Â Â Â for_each_new_plane_in_state(state, plane, old_plane_state, i) {
>> +Â Â Â for_each_old_plane_in_state(state, plane, old_plane_state, i) {
>> Â Â Â Â Â Â Â Â Â struct drm_plane_state *plane_state = plane->state;
>> Â Â Â Â Â Â Â Â Â struct drm_crtc *crtc = plane_state->crtc;
>> Â Â Â Â Â Â Â Â Â struct drm_framebuffer *fb = plane_state->fb;
>> @@ -4024,7 +4016,7 @@ void amdgpu_dm_atomic_commit_tail(
>> Â Â Â Â Â dm_state = to_dm_atomic_state(state);
>> Â Â Â Â Â /* update changed items */
>> -Â Â Â for_each_new_crtc_in_state(state, crtc, old_crtc_state, i) {
>> +Â Â Â for_each_old_crtc_in_state(state, crtc, old_crtc_state, i) {
>
> Better just use for_each_new_old here, so you can get both old and new
> from the iterator.
>
>> Â Â Â Â Â Â Â Â Â struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
>> Â Â Â Â Â Â Â Â Â struct drm_crtc_state *new_state = crtc->state;
>> @@ -4113,11 +4105,9 @@ void amdgpu_dm_atomic_commit_tail(
>> Â Â Â Â Â Â Â Â Â Â Â Â Â new_acrtc_state =
>> to_dm_crtc_state(new_crtcs[i]->base.state);
>> Â Â Â Â Â Â Â Â Â Â Â Â Â new_stream = new_acrtc_state->stream;
>> -Â Â Â Â Â Â Â Â Â Â Â aconnector =
>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â amdgpu_dm_find_first_crct_matching_connector(
>> +Â Â Â Â Â Â Â Â Â Â Â aconnector = amdgpu_dm_find_first_crct_matching_connector(
>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â state,
>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â &new_crtcs[i]->base,
>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â false);
>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â &new_crtcs[i]->base);
>> Â Â Â Â Â Â Â Â Â Â Â Â Â if (!aconnector) {
>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â DRM_DEBUG_DRIVER("Atomic commit: Failed to find
>> connector for acrtc id:%d "
>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â "skipping freesync init\n",
>> @@ -4151,7 +4141,7 @@ void amdgpu_dm_atomic_commit_tail(
>> Â Â Â Â Â }
>> Â Â Â Â Â /* Handle scaling and undersacn changes*/
>> -Â Â Â for_each_new_connector_in_state(state, connector, old_conn_state,
>> i) {
>> +Â Â Â for_each_old_connector_in_state(state, connector, old_conn_state,
>> i) {
>> Â Â Â Â Â Â Â Â Â struct amdgpu_dm_connector *aconnector =
>> to_amdgpu_dm_connector(connector);
>> Â Â Â Â Â Â Â Â Â struct dm_connector_state *con_new_state =
>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â to_dm_connector_state(aconnector->base.state);
>> @@ -4205,7 +4195,7 @@ void amdgpu_dm_atomic_commit_tail(
>> Â Â Â Â Â }
>> Â Â Â Â Â /* update planes when needed per crtc*/
>> -Â Â Â for_each_new_crtc_in_state(state, pcrtc, old_crtc_state, j) {
>> +Â Â Â for_each_old_crtc_in_state(state, pcrtc, old_crtc_state, j) {
>> Â Â Â Â Â Â Â Â Â new_acrtc_state = to_dm_crtc_state(pcrtc->state);
>
> Why did you switch to for_each_old iterator here ? if you use the
> for_each_new
> iterator you can get rid of new_acrtc_state =
> to_dm_crtc_state(pcrtc->state);
> and just use the iterator's state
>> Â Â Â Â Â Â Â Â Â if (new_acrtc_state->stream)
>> @@ -4218,7 +4208,7 @@ void amdgpu_dm_atomic_commit_tail(
>> Â Â Â Â Â Â * mark consumed event for drm_atomic_helper_commit_hw_done
>> Â Â Â Â Â Â */
>> Â Â Â Â Â spin_lock_irqsave(&adev->ddev->event_lock, flags);
>> -Â Â Â for_each_new_crtc_in_state(state, crtc, old_crtc_state, i) {
>> +Â Â Â for_each_old_crtc_in_state(state, crtc, old_crtc_state, i) {
>> Â Â Â Â Â Â Â Â Â struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
>
> Same as above
>
> Thanks,
> Andrey
>
Patches 1/6 and 2/6 should really be squashed, sorry for the confusion.
What you've pointed out is addressed in 2/6 :)
Leo
>> Â Â Â Â Â Â Â Â Â if (acrtc->base.state->event)
>> @@ -4402,7 +4392,7 @@ static int dm_update_crtcs_state(
>> Â Â Â Â Â Â Â Â Â new_acrtc_state = to_dm_crtc_state(crtc_state);
>> Â Â Â Â Â Â Â Â Â acrtc = to_amdgpu_crtc(crtc);
>> -Â Â Â Â Â Â Â aconnector =
>> amdgpu_dm_find_first_crct_matching_connector(state, crtc, true);
>> +Â Â Â Â Â Â Â aconnector =
>> amdgpu_dm_find_first_crct_matching_connector(state, crtc);
>> Â Â Â Â Â Â Â Â Â /* TODO This hack should go away */
>> Â Â Â Â Â Â Â Â Â if (aconnector) {
>> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
>> b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
>> index 630e6cd..1c55a0b 100644
>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h
>> @@ -228,8 +228,7 @@ void amdgpu_dm_update_connector_after_detect(
>> Â struct amdgpu_dm_connector
>> *amdgpu_dm_find_first_crct_matching_connector(
>> Â Â Â Â Â struct drm_atomic_state *state,
>> -Â Â Â struct drm_crtc *crtc,
>> -Â Â Â bool from_state_var);
>> +Â Â Â struct drm_crtc *crtc);
>> Â struct amdgpu_framebuffer;
>
