On 2017-10-30 12:36 PM, Leo wrote:
>
>
> On 2017-10-26 11:06 PM, Andrey Grodzovsky wrote:
>>
>>
>> On 2017-10-26 02:34 PM, Harry Wentland wrote:
>>> From: "Leo (Sunpeng) Li" <sunpeng.li at amd.com>
>>>
>>> Abandon new_crtcs array and use for_each_new iterator to acquire new
>>> crtcs.
>>>
>>> Signed-off-by: Leo (Sunpeng) Li <sunpeng.li at amd.com>
>>> Reviewed-by: Harry Wentland <Harry.Wentland at amd.com>
>>> ---
>>> Â drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 36
>>> +++++++++--------------
>>> Â 1 file changed, 14 insertions(+), 22 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 442b399a9400..590f80d29b56 100644
>>> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
>>> @@ -4013,10 +4013,8 @@ static void
>>> amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
>>> Â Â Â Â Â struct amdgpu_display_manager *dm = &adev->dm;
>>> Â Â Â Â Â struct dm_atomic_state *dm_state;
>>> Â Â Â Â Â uint32_t i, j;
>>> -Â Â Â uint32_t new_crtcs_count = 0;
>>> Â Â Â Â Â struct drm_crtc *crtc;
>>> Â Â Â Â Â struct drm_crtc_state *old_crtc_state, *new_crtc_state;
>>> -Â Â Â struct amdgpu_crtc *new_crtcs[MAX_STREAMS];
>>> Â Â Â Â Â unsigned long flags;
>>> Â Â Â Â Â bool wait_for_vblank = true;
>>> Â Â Â Â Â struct drm_connector *connector;
>>> @@ -4075,25 +4073,9 @@ static void
>>> amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â continue;
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â }
>>> -
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â if (dm_old_crtc_state->stream)
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â remove_stream(adev, acrtc, dm_old_crtc_state->stream);
>>> -
>>> -Â Â Â Â Â Â Â Â Â Â Â /*
>>> -Â Â Â Â Â Â Â Â Â Â Â Â * this loop saves set mode crtcs
>>> -Â Â Â Â Â Â Â Â Â Â Â Â * we needed to enable vblanks once all
>>> -Â Â Â Â Â Â Â Â Â Â Â Â * resources acquired in dc after dc_commit_streams
>>> -Â Â Â Â Â Â Â Â Â Â Â Â */
>>> -
>>> -Â Â Â Â Â Â Â Â Â Â Â /*TODO move all this into dm_crtc_state, get rid of
>>> -Â Â Â Â Â Â Â Â Â Â Â Â * new_crtcs array and use old and new atomic states
>>> -Â Â Â Â Â Â Â Â Â Â Â Â * instead
>>> -Â Â Â Â Â Â Â Â Â Â Â Â */
>>> -Â Â Â Â Â Â Â Â Â Â Â new_crtcs[new_crtcs_count] = acrtc;
>>> -Â Â Â Â Â Â Â Â Â Â Â new_crtcs_count++;
>>> -
>>> -Â Â Â Â Â Â Â Â Â Â Â new_crtc_state = drm_atomic_get_new_crtc_state(state,
>>> crtc);
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â acrtc->enabled = true;
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â acrtc->hw_mode = new_crtc_state->mode;
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â crtc->hwmode = new_crtc_state->mode;
>>> @@ -4221,18 +4203,28 @@ static void
>>> amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â dm_error("%s: Failed to update stream scaling!\n",
>>> __func__);
>>> Â Â Â Â Â }
>>> -Â Â Â for (i = 0; i < new_crtcs_count; i++) {
>>> +Â Â Â for_each_oldnew_crtc_in_state(state, crtc, old_crtc_state,
>>> +Â Â Â Â Â Â Â Â Â Â Â new_crtc_state, i) {
>>> Â Â Â Â Â Â Â Â Â /*
>>> Â Â Â Â Â Â Â Â Â Â * loop to enable interrupts on newly arrived crtc
>>> Â Â Â Â Â Â Â Â Â Â */
>>> -Â Â Â Â Â Â Â struct amdgpu_crtc *acrtc = new_crtcs[i];
>>> +Â Â Â Â Â Â Â struct amdgpu_crtc *acrtc = to_amdgpu_crtc(crtc);
>>> +Â Â Â Â Â Â Â bool modeset_needed;
>>> -Â Â Â Â Â Â Â new_crtc_state = drm_atomic_get_new_crtc_state(state,
>>> &acrtc->base);
>>> Â Â Â Â Â Â Â Â Â dm_new_crtc_state = to_dm_crtc_state(new_crtc_state);
>>> +Â Â Â Â Â Â Â dm_old_crtc_state = to_dm_crtc_state(old_crtc_state);
>>> +Â Â Â Â Â Â Â modeset_needed = modeset_required(
>>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â new_crtc_state,
>>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â dm_new_crtc_state->stream,
>>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â dm_old_crtc_state->stream);
>>> +
>>> +Â Â Â Â Â Â Â if (dm_new_crtc_state->stream == NULL || !modeset_needed)
>>> +Â Â Â Â Â Â Â Â Â Â Â continue;
>>
>> I feel it's a bit future bug prone to repeat the 2 checks above from
>> the initial
>> for_each_crtc_in_state loop, somebody makes changes there and forget
>> about this loop.
>> The array is ugly but it avoids logic duplication.
>
> I think a better fix would be to combine this for_each section with the
> one above it. The only thing keeping them separate (not shown in the
> above context) is an `if (adev->dm.freesync_module)` guard, which can be
> pulled into the first for_each loop. It would also eliminate a lot of
> duplicated initialization code.
>
> Thoughts?
>
> Leo
>
Ignore what I said, It seems I was looking at the wrong context :) The
fact this blob is being done after dc_commit_state() is annoying...
The maintainability of this duplicated logic is a good point. Although
I'm not sure if going back to a shared array using custom iterator logic
is the best approach. Perhapse we can pull this duplicated logic into a
seperate function?
Thanks,
Leo
>>
>> Thanks,
>> Andrey
>>
>>> Â Â Â Â Â Â Â Â Â if (adev->dm.freesync_module)
>>> Â Â Â Â Â Â Â Â Â Â Â Â Â mod_freesync_notify_mode_change(
>>> -Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â adev->dm.freesync_module,
>>> &dm_new_crtc_state->stream, 1);
>>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â adev->dm.freesync_module,
>>> +Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â &dm_new_crtc_state->stream, 1);
>>> Â Â Â Â Â Â Â Â Â manage_dm_interrupts(adev, acrtc, true);
>>> Â Â Â Â Â }
>>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
