Op 14-10-15 om 14:59 schreef Ander Conselvan De Oliveira:
> On Wed, 2015-09-23 at 13:27 +0200, Maarten Lankhorst wrote:
>> atomic->disabled_planes is a hack that had to exist because
>> prepare_fb was only called when a new fb was set. This messed
>> up fb tracking in some circumstances like aborts from
>> interruptible waits. As a result interruptible waiting in
>> prepare_plane_fb was forbidden, but other errors could still
>> cause frontbuffer tracking to be messed up.
>>
>> Now that prepare_fb is always called, this hack is no longer
>> required and prepare_fb may fail without consequences.
>>
>> Changes since v1:
>> - Clean up a few fb tracking warnings by changing plane->fb to
>> plane->state->fb.
>>
>> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>
>> ---
>> drivers/gpu/drm/i915/intel_display.c | 47 ++++++++++++++----------------------
>> drivers/gpu/drm/i915/intel_drv.h | 1 -
>> 2 files changed, 18 insertions(+), 30 deletions(-)
>>
>> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
>> index fc0086748b71..ac97af69be62 100644
>> --- a/drivers/gpu/drm/i915/intel_display.c
>> +++ b/drivers/gpu/drm/i915/intel_display.c
>> @@ -4752,17 +4752,6 @@ static void intel_pre_plane_update(struct intel_crtc *crtc)
>> struct drm_device *dev = crtc->base.dev;
>> struct drm_i915_private *dev_priv = dev->dev_private;
>> struct intel_crtc_atomic_commit *atomic = &crtc->atomic;
>> - struct drm_plane *p;
>> -
>> - /* Track fb's for any planes being disabled */
>> - drm_for_each_plane_mask(p, dev, atomic->disabled_planes) {
>> - struct intel_plane *plane = to_intel_plane(p);
>> -
>> - mutex_lock(&dev->struct_mutex);
>> - i915_gem_track_fb(intel_fb_obj(plane->base.fb), NULL,
>> - plane->frontbuffer_bit);
>> - mutex_unlock(&dev->struct_mutex);
>> - }
>>
>> if (atomic->wait_for_flips)
>> intel_crtc_wait_for_pending_flips(&crtc->base);
>> @@ -11561,14 +11550,6 @@ int intel_plane_atomic_calc_changes(struct drm_crtc_state *crtc_state,
>> return ret;
>> }
>>
>> - /*
>> - * Disabling a plane is always okay; we just need to update
>> - * fb tracking in a special way since cleanup_fb() won't
>> - * get called by the plane helpers.
>> - */
>> - if (old_plane_state->base.fb && !fb)
>> - intel_crtc->atomic.disabled_planes |= 1 << i;
>> -
>> was_visible = old_plane_state->visible;
>> visible = to_intel_plane_state(plane_state)->visible;
>>
>> @@ -13318,15 +13299,17 @@ intel_prepare_plane_fb(struct drm_plane *plane,
>> struct drm_framebuffer *fb = new_state->fb;
>> struct intel_plane *intel_plane = to_intel_plane(plane);
>> struct drm_i915_gem_object *obj = intel_fb_obj(fb);
>> - struct drm_i915_gem_object *old_obj = intel_fb_obj(plane->fb);
>> + struct drm_i915_gem_object *old_obj = intel_fb_obj(plane->state->fb);
>> int ret = 0;
>>
>> - if (!obj)
>> + if (!obj && !old_obj)
>> return 0;
>>
>> mutex_lock(&dev->struct_mutex);
>>
>> - if (plane->type == DRM_PLANE_TYPE_CURSOR &&
>> + if (!obj) {
>> + ret = 0;
>> + } else if (plane->type == DRM_PLANE_TYPE_CURSOR &&
>> INTEL_INFO(dev)->cursor_needs_physical) {
>> int align = IS_I830(dev) ? 16 * 1024 : 256;
>> ret = i915_gem_object_attach_phys(obj, align);
>> @@ -13356,17 +13339,23 @@ intel_cleanup_plane_fb(struct drm_plane *plane,
>> const struct drm_plane_state *old_state)
>> {
>> struct drm_device *dev = plane->dev;
>> - struct drm_i915_gem_object *obj = intel_fb_obj(old_state->fb);
>> + struct intel_plane *intel_plane = to_intel_plane(plane);
>> + struct drm_i915_gem_object *old_obj = intel_fb_obj(old_state->fb);
>> + struct drm_i915_gem_object *obj = intel_fb_obj(plane->state->fb);
>>
>> - if (!obj)
>> + if (!obj && !old_obj)
>> return;
>>
>> - if (plane->type != DRM_PLANE_TYPE_CURSOR ||
>> - !INTEL_INFO(dev)->cursor_needs_physical) {
>> - mutex_lock(&dev->struct_mutex);
>> + mutex_lock(&dev->struct_mutex);
>> + if (old_obj && (plane->type != DRM_PLANE_TYPE_CURSOR ||
>> + !INTEL_INFO(dev)->cursor_needs_physical))
>> intel_unpin_fb_obj(old_state->fb, old_state);
>> - mutex_unlock(&dev->struct_mutex);
>> - }
>> +
>> + /* prepare_fb aborted? */
>> + if ((old_obj && (old_obj->frontbuffer_bits & intel_plane->frontbuffer_bit)) ||
>> + (obj && !(obj->frontbuffer_bits & intel_plane->frontbuffer_bit)))
>> + i915_gem_track_fb(old_obj, obj, intel_plane->frontbuffer_bit);
> I'm not a fan of this big condition. Would it make sense to add a new parameter to cleanup_fb() that
> tells us if this is an abort clean up?
>
> But in any case, that can be done later.
>
> Reviewed-by: Ander Conselvan de Oliveira <conselvan2@xxxxxxxxx>
>
This is only temporarily though. I plan to make unpin async in the future. When that happens cleanup_fb will only be called when aborting so it will become a moot point. :)
We could also add a force flag to i915_gem_track_fb to ignore warnings in cleanup_plane_fb later on.
~Maarten
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
