| Message ID | 20200722181332.26995-1-knaerzche@gmail.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | drm: rockchip: various ports for older VOPs | expand |
Am 22.07.20 um 23:43 schrieb Daniel Vetter: > On Wed, Jul 22, 2020 at 8:13 PM Alex Bee <knaerzche@gmail.com> wrote: >> Hi, >> >> this series mainly ports existining functionality to older SoCs - most >> importantly enables alpha blending for RK3036, RK3066, RK3126 and >> RK3188. >> Besides that, it also changes the window type from DRM_PLANE_TYPE_CURSOR >> to DRM_PLANE_TYPE_OVERLAY for VOPs that have only one (1) overlay window. > This doesn't make much sense, the cursor overlay is really just a hint > for legacy ioctls that this is the overlay that should be used for > cursors. Compositors should try to use such planes as full overlays > (if they don't want to use them as a cursor). So sounds like a case of > "fix your compositor". I agree here - but: If HWC windows would have been implemented in this particular driver, their max size would be 128x128 on some SoCs - so they woudn't be really suitable to create an OSD overlay at 4K, for example. I don't know, but I guess other vendors implement their HWC windows on this reduced HW resources as well. I guess that is one of the reasons, why userspace, which aims to be cross-plattfrom, avoids DRM_PLANE_TYPE_CURSOR when its looking for an usable overlay plane. (a heuristic, indeed) > For atomic there's 0 difference between a overlay or a cursor (primary > plane is still treated somewhat special in the RMFB ioctl, but again > that's for backwards compat reasons with existing uapi, not because > the primary plane is different). > > What does happen though is that this breaks cursor for legacy > userspace, which is probably not really what you want. Indeed not. Beforhand I was submiiting this, I looked arround and couldn't find anything which relies or even depends of a cursor window to be available. Even if: as per spec only one DRM_PLANE_TYPE_PRIMARY is mandatory, everything else is optional. > -Daniel > > >> Regards, >> Alex >> >> Changes in v2: >> - drop not yet upstreamed dsp_data_swap from RK3188 regs >> - rephrase most commit messages >> >> Alex Bee (5): >> drm: rockchip: add scaling for RK3036 win1 >> drm: rockchip: add missing registers for RK3188 >> drm: rockchip: add alpha support for RK3036, RK3066, RK3126 and RK3188 >> drm: rockchip: set alpha_en to 0 if it is not used >> drm: rockchip: use overlay windows as such >> >> drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 1 + >> drivers/gpu/drm/rockchip/rockchip_vop_reg.c | 42 ++++++++++++++++++--- >> drivers/gpu/drm/rockchip/rockchip_vop_reg.h | 1 + >> 3 files changed, 38 insertions(+), 6 deletions(-) >> >> -- >> 2.17.1 >> > > -- > Daniel Vetter > Software Engineer, Intel Corporation > http://blog.ffwll.ch Best Regards
On Sat, Jul 25, 2020 at 3:52 PM Alex Bee <knaerzche@gmail.com> wrote: > > > Am 22.07.20 um 23:43 schrieb Daniel Vetter: > > On Wed, Jul 22, 2020 at 8:13 PM Alex Bee <knaerzche@gmail.com> wrote: > >> Hi, > >> > >> this series mainly ports existining functionality to older SoCs - most > >> importantly enables alpha blending for RK3036, RK3066, RK3126 and > >> RK3188. > >> Besides that, it also changes the window type from DRM_PLANE_TYPE_CURSOR > >> to DRM_PLANE_TYPE_OVERLAY for VOPs that have only one (1) overlay window. > > This doesn't make much sense, the cursor overlay is really just a hint > > for legacy ioctls that this is the overlay that should be used for > > cursors. Compositors should try to use such planes as full overlays > > (if they don't want to use them as a cursor). So sounds like a case of > > "fix your compositor". > I agree here - but: If HWC windows would have been implemented in this > particular driver, their max size would be 128x128 on some SoCs - so > they woudn't be really suitable to create an OSD overlay at 4K, for > example. I don't know, but I guess other vendors implement their HWC > windows on this reduced HW resources as well. I guess that is one of the > reasons, why userspace, which aims to be cross-plattfrom, avoids > DRM_PLANE_TYPE_CURSOR when its looking for an usable overlay plane. (a > heuristic, indeed) Which userspace does that? We should fix that, not try to work around that in all the drivers in upstream, that wont work. -Daniel > > For atomic there's 0 difference between a overlay or a cursor (primary > > plane is still treated somewhat special in the RMFB ioctl, but again > > that's for backwards compat reasons with existing uapi, not because > > the primary plane is different). > > > > What does happen though is that this breaks cursor for legacy > > userspace, which is probably not really what you want. > > Indeed not. > > Beforhand I was submiiting this, I looked arround and couldn't find > anything which relies or even depends of a cursor window to be > available. Even if: as per spec only one DRM_PLANE_TYPE_PRIMARY is > mandatory, everything else is optional. > > > -Daniel > > > > > >> Regards, > >> Alex > >> > >> Changes in v2: > >> - drop not yet upstreamed dsp_data_swap from RK3188 regs > >> - rephrase most commit messages > >> > >> Alex Bee (5): > >> drm: rockchip: add scaling for RK3036 win1 > >> drm: rockchip: add missing registers for RK3188 > >> drm: rockchip: add alpha support for RK3036, RK3066, RK3126 and RK3188 > >> drm: rockchip: set alpha_en to 0 if it is not used > >> drm: rockchip: use overlay windows as such > >> > >> drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 1 + > >> drivers/gpu/drm/rockchip/rockchip_vop_reg.c | 42 ++++++++++++++++++--- > >> drivers/gpu/drm/rockchip/rockchip_vop_reg.h | 1 + > >> 3 files changed, 38 insertions(+), 6 deletions(-) > >> > >> -- > >> 2.17.1 > >> > > > > -- > > Daniel Vetter > > Software Engineer, Intel Corporation > > http://blog.ffwll.ch > > Best Regards >
Am 25.07.20 um 17:54 schrieb Daniel Vetter: > On Sat, Jul 25, 2020 at 3:52 PM Alex Bee <knaerzche@gmail.com> wrote: >> >> Am 22.07.20 um 23:43 schrieb Daniel Vetter: >>> On Wed, Jul 22, 2020 at 8:13 PM Alex Bee <knaerzche@gmail.com> wrote: >>>> Hi, >>>> >>>> this series mainly ports existining functionality to older SoCs - most >>>> importantly enables alpha blending for RK3036, RK3066, RK3126 and >>>> RK3188. >>>> Besides that, it also changes the window type from DRM_PLANE_TYPE_CURSOR >>>> to DRM_PLANE_TYPE_OVERLAY for VOPs that have only one (1) overlay window. >>> This doesn't make much sense, the cursor overlay is really just a hint >>> for legacy ioctls that this is the overlay that should be used for >>> cursors. Compositors should try to use such planes as full overlays >>> (if they don't want to use them as a cursor). So sounds like a case of >>> "fix your compositor". >> I agree here - but: If HWC windows would have been implemented in this >> particular driver, their max size would be 128x128 on some SoCs - so >> they woudn't be really suitable to create an OSD overlay at 4K, for >> example. I don't know, but I guess other vendors implement their HWC >> windows on this reduced HW resources as well. I guess that is one of the >> reasons, why userspace, which aims to be cross-plattfrom, avoids >> DRM_PLANE_TYPE_CURSOR when its looking for an usable overlay plane. (a >> heuristic, indeed) > Which userspace does that? kodi-gbm: https://github.com/xbmc/xbmc/blob/master/xbmc/windowing/gbm/DRMUtils.cpp#L406 > We should fix that, not try to work around > that in all the drivers in upstream, that wont work. You're right I'll drop this part. > -Daniel > >>> For atomic there's 0 difference between a overlay or a cursor (primary >>> plane is still treated somewhat special in the RMFB ioctl, but again >>> that's for backwards compat reasons with existing uapi, not because >>> the primary plane is different). >>> >>> What does happen though is that this breaks cursor for legacy >>> userspace, which is probably not really what you want. >> Indeed not. >> >> Beforhand I was submiiting this, I looked arround and couldn't find >> anything which relies or even depends of a cursor window to be >> available. Even if: as per spec only one DRM_PLANE_TYPE_PRIMARY is >> mandatory, everything else is optional. >> >>> -Daniel >>> >>> >>>> Regards, >>>> Alex >>>> >>>> Changes in v2: >>>> - drop not yet upstreamed dsp_data_swap from RK3188 regs >>>> - rephrase most commit messages >>>> >>>> Alex Bee (5): >>>> drm: rockchip: add scaling for RK3036 win1 >>>> drm: rockchip: add missing registers for RK3188 >>>> drm: rockchip: add alpha support for RK3036, RK3066, RK3126 and RK3188 >>>> drm: rockchip: set alpha_en to 0 if it is not used >>>> drm: rockchip: use overlay windows as such >>>> >>>> drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 1 + >>>> drivers/gpu/drm/rockchip/rockchip_vop_reg.c | 42 ++++++++++++++++++--- >>>> drivers/gpu/drm/rockchip/rockchip_vop_reg.h | 1 + >>>> 3 files changed, 38 insertions(+), 6 deletions(-) >>>> >>>> -- >>>> 2.17.1 >>>> >>> -- >>> Daniel Vetter >>> Software Engineer, Intel Corporation >>> http://blog.ffwll.ch >> Best Regards >> > Best regards
On Sat, Jul 25, 2020 at 8:48 PM Alex Bee <knaerzche@gmail.com> wrote: > > > Am 25.07.20 um 17:54 schrieb Daniel Vetter: > > On Sat, Jul 25, 2020 at 3:52 PM Alex Bee <knaerzche@gmail.com> wrote: > >> > >> Am 22.07.20 um 23:43 schrieb Daniel Vetter: > >>> On Wed, Jul 22, 2020 at 8:13 PM Alex Bee <knaerzche@gmail.com> wrote: > >>>> Hi, > >>>> > >>>> this series mainly ports existining functionality to older SoCs - most > >>>> importantly enables alpha blending for RK3036, RK3066, RK3126 and > >>>> RK3188. > >>>> Besides that, it also changes the window type from DRM_PLANE_TYPE_CURSOR > >>>> to DRM_PLANE_TYPE_OVERLAY for VOPs that have only one (1) overlay window. > >>> This doesn't make much sense, the cursor overlay is really just a hint > >>> for legacy ioctls that this is the overlay that should be used for > >>> cursors. Compositors should try to use such planes as full overlays > >>> (if they don't want to use them as a cursor). So sounds like a case of > >>> "fix your compositor". > >> I agree here - but: If HWC windows would have been implemented in this > >> particular driver, their max size would be 128x128 on some SoCs - so > >> they woudn't be really suitable to create an OSD overlay at 4K, for > >> example. I don't know, but I guess other vendors implement their HWC > >> windows on this reduced HW resources as well. I guess that is one of the > >> reasons, why userspace, which aims to be cross-plattfrom, avoids > >> DRM_PLANE_TYPE_CURSOR when its looking for an usable overlay plane. (a > >> heuristic, indeed) > > Which userspace does that? > kodi-gbm: > https://github.com/xbmc/xbmc/blob/master/xbmc/windowing/gbm/DRMUtils.cpp#L406 Can you pls file a bug report, linking to this thread here? Maybe also link to the docs: https://dri.freedesktop.org/docs/drm/gpu/drm-kms.html?highlight=drm_plane#c.drm_plane_type "For userspace which is universal plane aware and which is using that atomic IOCTL there’s no difference between these planes (beyong what the driver and hardware can support of course)." Also feel free to cc me on the github issue (I'm @danvet over there) in case there's more questions. Cheers, Daniel > > > We should fix that, not try to work around > > that in all the drivers in upstream, that wont work. > You're right I'll drop this part. > > -Daniel > > > >>> For atomic there's 0 difference between a overlay or a cursor (primary > >>> plane is still treated somewhat special in the RMFB ioctl, but again > >>> that's for backwards compat reasons with existing uapi, not because > >>> the primary plane is different). > >>> > >>> What does happen though is that this breaks cursor for legacy > >>> userspace, which is probably not really what you want. > >> Indeed not. > >> > >> Beforhand I was submiiting this, I looked arround and couldn't find > >> anything which relies or even depends of a cursor window to be > >> available. Even if: as per spec only one DRM_PLANE_TYPE_PRIMARY is > >> mandatory, everything else is optional. > >> > >>> -Daniel > >>> > >>> > >>>> Regards, > >>>> Alex > >>>> > >>>> Changes in v2: > >>>> - drop not yet upstreamed dsp_data_swap from RK3188 regs > >>>> - rephrase most commit messages > >>>> > >>>> Alex Bee (5): > >>>> drm: rockchip: add scaling for RK3036 win1 > >>>> drm: rockchip: add missing registers for RK3188 > >>>> drm: rockchip: add alpha support for RK3036, RK3066, RK3126 and RK3188 > >>>> drm: rockchip: set alpha_en to 0 if it is not used > >>>> drm: rockchip: use overlay windows as such > >>>> > >>>> drivers/gpu/drm/rockchip/rockchip_drm_vop.c | 1 + > >>>> drivers/gpu/drm/rockchip/rockchip_vop_reg.c | 42 ++++++++++++++++++--- > >>>> drivers/gpu/drm/rockchip/rockchip_vop_reg.h | 1 + > >>>> 3 files changed, 38 insertions(+), 6 deletions(-) > >>>> > >>>> -- > >>>> 2.17.1 > >>>> > >>> -- > >>> Daniel Vetter > >>> Software Engineer, Intel Corporation > >>> http://blog.ffwll.ch > >> Best Regards > >> > > > Best regards