Message ID | 20231115131549.2191589-5-javierm@redhat.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | drm: Allow the damage helpers to handle buffer damage | expand |
Hi Am 15.11.23 um 14:15 schrieb Javier Martinez Canillas: > The "Damage Tracking Properties" section in the documentation doesn't have > info about the two type of damage handling: frame damage vs buffer damage. > > Add it to the section and mention that helpers only support frame damage, > and how drivers handling buffer damage can indicate that the damage clips > should be ignored. > > Also add references to further documentation about the two damage types. > > Suggested-by: Simon Ser <contact@emersion.fr> > Signed-off-by: Javier Martinez Canillas <javierm@redhat.com> > --- > - Dropped Simon Ser's Reviwed-by tag because the text changed to adapt > to the approach Thomas Zimmermann suggested for v2. > > (no changes since v1) > > drivers/gpu/drm/drm_plane.c | 20 ++++++++++++++++++++ > 1 file changed, 20 insertions(+) > > diff --git a/drivers/gpu/drm/drm_plane.c b/drivers/gpu/drm/drm_plane.c > index 24e7998d1731..3b1b8bca3065 100644 > --- a/drivers/gpu/drm/drm_plane.c > +++ b/drivers/gpu/drm/drm_plane.c > @@ -1442,6 +1442,26 @@ int drm_mode_page_flip_ioctl(struct drm_device *dev, > * Drivers implementing damage can use drm_atomic_helper_damage_iter_init() and > * drm_atomic_helper_damage_iter_next() helper iterator function to get damage > * rectangles clipped to &drm_plane_state.src. > + * > + * Note that there are two types of damage handling: frame damage and buffer > + * damage. The type of damage handling implemented depends on a driver's upload > + * target. Drivers implementing a per-plane or per-CRTC upload target need to > + * handle frame damage while drivers implementing a per-buffer upload target > + * need to handle buffer damage. > + * > + * The existing damage helpers only support the frame damage type, there is no > + * buffer age support or similar damage accumulation algorithm implemented yet. > + * > + * Only drivers handling frame damage can use the mentiored damage helpers to > + * iterate over the damaged regions. Drivers that handle buffer damage, need to > + * set &struct drm_plane_state.ignore_damage_clips as an indication to > + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. > + * In that case, the returned damage rectangle is the &drm_plane_state.src since > + * a full plane update should happen. > + * > + * For more information about the two type of damage, see: > + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt > + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ One thought you might want to consider. These URLs are helpful. The only issue I have is that frame damage and buffer damage are user-space concepts. The kernel bug is that damage handling expects the backing storage/upload buffer not to change for a given plane. If the upload buffer changes between page flips, the new upload buffer has to be updated as a whole. Hence no damage handling then. And then maybe say that the effects of frame damage and buffer damage in userspace illustrate this problem and give the URLs. Best regards Thomas > */ > > /**
On Thursday, November 16th, 2023 at 13:06, Thomas Zimmermann <tzimmermann@suse.de> wrote: > > + * Note that there are two types of damage handling: frame damage and buffer > > + * damage. The type of damage handling implemented depends on a driver's upload > > + * target. Drivers implementing a per-plane or per-CRTC upload target need to > > + * handle frame damage while drivers implementing a per-buffer upload target > > + * need to handle buffer damage. > > + * > > + * The existing damage helpers only support the frame damage type, there is no > > + * buffer age support or similar damage accumulation algorithm implemented yet. > > + * > > + * Only drivers handling frame damage can use the mentiored damage helpers to Typo: mentioned > > + * iterate over the damaged regions. Drivers that handle buffer damage, need to > > + * set &struct drm_plane_state.ignore_damage_clips as an indication to > > + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. > > + * In that case, the returned damage rectangle is the &drm_plane_state.src since > > + * a full plane update should happen. > > + * > > + * For more information about the two type of damage, see: > > + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt > > + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ > > One thought you might want to consider. > > These URLs are helpful. The only issue I have is that frame damage and > buffer damage are user-space concepts. The kernel bug is that damage > handling expects the backing storage/upload buffer not to change for a > given plane. If the upload buffer changes between page flips, the new > upload buffer has to be updated as a whole. Hence no damage handling then. Why would these concepts be specific to user-space? The kernel could better handle buffer damage instead of forcing full damage, by doing something similar to what user-space does. Anyways: Reviewed-by: Simon Ser <contact@emersion.fr>
Hi Am 16.11.23 um 13:14 schrieb Simon Ser: > On Thursday, November 16th, 2023 at 13:06, Thomas Zimmermann <tzimmermann@suse.de> wrote: > >>> + * Note that there are two types of damage handling: frame damage and buffer >>> + * damage. The type of damage handling implemented depends on a driver's upload >>> + * target. Drivers implementing a per-plane or per-CRTC upload target need to >>> + * handle frame damage while drivers implementing a per-buffer upload target >>> + * need to handle buffer damage. >>> + * >>> + * The existing damage helpers only support the frame damage type, there is no >>> + * buffer age support or similar damage accumulation algorithm implemented yet. >>> + * >>> + * Only drivers handling frame damage can use the mentiored damage helpers to > > Typo: mentioned > >>> + * iterate over the damaged regions. Drivers that handle buffer damage, need to >>> + * set &struct drm_plane_state.ignore_damage_clips as an indication to >>> + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. >>> + * In that case, the returned damage rectangle is the &drm_plane_state.src since >>> + * a full plane update should happen. >>> + * >>> + * For more information about the two type of damage, see: >>> + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt >>> + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ >> >> One thought you might want to consider. >> >> These URLs are helpful. The only issue I have is that frame damage and >> buffer damage are user-space concepts. The kernel bug is that damage >> handling expects the backing storage/upload buffer not to change for a >> given plane. If the upload buffer changes between page flips, the new >> upload buffer has to be updated as a whole. Hence no damage handling then. > > Why would these concepts be specific to user-space? The kernel could > better handle buffer damage instead of forcing full damage, by doing > something similar to what user-space does. The terms 'frame damage' and 'buffer damage' do not exist in the kernel. The problem can be better described in wording that is common within the context of the kernel drivers. Best regards Thomas > > Anyways: > > Reviewed-by: Simon Ser <contact@emersion.fr>
On Thu, 16 Nov 2023 13:34:07 +0100 Thomas Zimmermann <tzimmermann@suse.de> wrote: > Hi > > Am 16.11.23 um 13:14 schrieb Simon Ser: > > On Thursday, November 16th, 2023 at 13:06, Thomas Zimmermann <tzimmermann@suse.de> wrote: > > > >>> + * Note that there are two types of damage handling: frame damage and buffer > >>> + * damage. The type of damage handling implemented depends on a driver's upload > >>> + * target. Drivers implementing a per-plane or per-CRTC upload target need to > >>> + * handle frame damage while drivers implementing a per-buffer upload target > >>> + * need to handle buffer damage. > >>> + * > >>> + * The existing damage helpers only support the frame damage type, there is no > >>> + * buffer age support or similar damage accumulation algorithm implemented yet. > >>> + * > >>> + * Only drivers handling frame damage can use the mentiored damage helpers to > > > > Typo: mentioned > > > >>> + * iterate over the damaged regions. Drivers that handle buffer damage, need to > >>> + * set &struct drm_plane_state.ignore_damage_clips as an indication to > >>> + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. > >>> + * In that case, the returned damage rectangle is the &drm_plane_state.src since > >>> + * a full plane update should happen. > >>> + * > >>> + * For more information about the two type of damage, see: > >>> + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt > >>> + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ > >> > >> One thought you might want to consider. > >> > >> These URLs are helpful. The only issue I have is that frame damage and > >> buffer damage are user-space concepts. The kernel bug is that damage > >> handling expects the backing storage/upload buffer not to change for a > >> given plane. If the upload buffer changes between page flips, the new > >> upload buffer has to be updated as a whole. Hence no damage handling then. > > > > Why would these concepts be specific to user-space? The kernel could > > better handle buffer damage instead of forcing full damage, by doing > > something similar to what user-space does. > > The terms 'frame damage' and 'buffer damage' do not exist in the kernel. > The problem can be better described in wording that is common within the > context of the kernel drivers. What terms does the kernel use for these two different concepts of damage? Thanks, pq
Hi Am 16.11.23 um 16:24 schrieb Pekka Paalanen: > On Thu, 16 Nov 2023 13:34:07 +0100 > Thomas Zimmermann <tzimmermann@suse.de> wrote: > >> Hi >> >> Am 16.11.23 um 13:14 schrieb Simon Ser: >>> On Thursday, November 16th, 2023 at 13:06, Thomas Zimmermann <tzimmermann@suse.de> wrote: >>> >>>>> + * Note that there are two types of damage handling: frame damage and buffer >>>>> + * damage. The type of damage handling implemented depends on a driver's upload >>>>> + * target. Drivers implementing a per-plane or per-CRTC upload target need to >>>>> + * handle frame damage while drivers implementing a per-buffer upload target >>>>> + * need to handle buffer damage. >>>>> + * >>>>> + * The existing damage helpers only support the frame damage type, there is no >>>>> + * buffer age support or similar damage accumulation algorithm implemented yet. >>>>> + * >>>>> + * Only drivers handling frame damage can use the mentiored damage helpers to >>> >>> Typo: mentioned >>> >>>>> + * iterate over the damaged regions. Drivers that handle buffer damage, need to >>>>> + * set &struct drm_plane_state.ignore_damage_clips as an indication to >>>>> + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. >>>>> + * In that case, the returned damage rectangle is the &drm_plane_state.src since >>>>> + * a full plane update should happen. >>>>> + * >>>>> + * For more information about the two type of damage, see: >>>>> + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt >>>>> + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ >>>> >>>> One thought you might want to consider. >>>> >>>> These URLs are helpful. The only issue I have is that frame damage and >>>> buffer damage are user-space concepts. The kernel bug is that damage >>>> handling expects the backing storage/upload buffer not to change for a >>>> given plane. If the upload buffer changes between page flips, the new >>>> upload buffer has to be updated as a whole. Hence no damage handling then. >>> >>> Why would these concepts be specific to user-space? The kernel could >>> better handle buffer damage instead of forcing full damage, by doing >>> something similar to what user-space does. >> >> The terms 'frame damage' and 'buffer damage' do not exist in the kernel. >> The problem can be better described in wording that is common within the >> context of the kernel drivers. > > What terms does the kernel use for these two different concepts of > damage? None AFAIK. Damage is relative to the plane's backing storage/upload buffer. That's frame damage. We never needed a name for buffer damage as we don't support it. Best regards Thomas > > > Thanks, > pq
diff --git a/drivers/gpu/drm/drm_plane.c b/drivers/gpu/drm/drm_plane.c index 24e7998d1731..3b1b8bca3065 100644 --- a/drivers/gpu/drm/drm_plane.c +++ b/drivers/gpu/drm/drm_plane.c @@ -1442,6 +1442,26 @@ int drm_mode_page_flip_ioctl(struct drm_device *dev, * Drivers implementing damage can use drm_atomic_helper_damage_iter_init() and * drm_atomic_helper_damage_iter_next() helper iterator function to get damage * rectangles clipped to &drm_plane_state.src. + * + * Note that there are two types of damage handling: frame damage and buffer + * damage. The type of damage handling implemented depends on a driver's upload + * target. Drivers implementing a per-plane or per-CRTC upload target need to + * handle frame damage while drivers implementing a per-buffer upload target + * need to handle buffer damage. + * + * The existing damage helpers only support the frame damage type, there is no + * buffer age support or similar damage accumulation algorithm implemented yet. + * + * Only drivers handling frame damage can use the mentiored damage helpers to + * iterate over the damaged regions. Drivers that handle buffer damage, need to + * set &struct drm_plane_state.ignore_damage_clips as an indication to + * drm_atomic_helper_damage_iter_init() that the damage clips should be ignored. + * In that case, the returned damage rectangle is the &drm_plane_state.src since + * a full plane update should happen. + * + * For more information about the two type of damage, see: + * https://registry.khronos.org/EGL/extensions/KHR/EGL_KHR_swap_buffers_with_damage.txt + * https://emersion.fr/blog/2019/intro-to-damage-tracking/ */ /**
The "Damage Tracking Properties" section in the documentation doesn't have info about the two type of damage handling: frame damage vs buffer damage. Add it to the section and mention that helpers only support frame damage, and how drivers handling buffer damage can indicate that the damage clips should be ignored. Also add references to further documentation about the two damage types. Suggested-by: Simon Ser <contact@emersion.fr> Signed-off-by: Javier Martinez Canillas <javierm@redhat.com> --- - Dropped Simon Ser's Reviwed-by tag because the text changed to adapt to the approach Thomas Zimmermann suggested for v2. (no changes since v1) drivers/gpu/drm/drm_plane.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)