| Message ID | 20230815185710.159779-7-andrealmeid@igalia.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm: Add support for atomic async page-flip | expand |
On 8/15/23 20:57, André Almeida wrote: > From: Pekka Paalanen <pekka.paalanen@collabora.com> > > Specify how the atomic state is maintained between userspace and > kernel, plus the special case for async flips. > > Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> > Signed-off-by: André Almeida <andrealmeid@igalia.com> [...] > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > +effectively change only the FB_ID property on any planes. No-operation changes > +are ignored as always. [...] During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation.
On 8/17/23 12:37, Michel Dänzer wrote: > On 8/15/23 20:57, André Almeida wrote: >> From: Pekka Paalanen <pekka.paalanen@collabora.com> >> >> Specify how the atomic state is maintained between userspace and >> kernel, plus the special case for async flips. >> >> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >> Signed-off-by: André Almeida <andrealmeid@igalia.com> > > [...] > >> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >> +effectively change only the FB_ID property on any planes. No-operation changes >> +are ignored as always. [...] > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. That said, it doesn't make too much sense to use DRM_MODE_PAGE_FLIP_ASYNC for this, since it won't have any effect outside of vertical blank anyway.
Hi Michel, Em 17/08/2023 07:37, Michel Dänzer escreveu: > On 8/15/23 20:57, André Almeida wrote: >> From: Pekka Paalanen <pekka.paalanen@collabora.com> >> >> Specify how the atomic state is maintained between userspace and >> kernel, plus the special case for async flips. >> >> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >> Signed-off-by: André Almeida <andrealmeid@igalia.com> > > [...] > >> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >> +effectively change only the FB_ID property on any planes. No-operation changes >> +are ignored as always. [...] > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > I believe the documentation already addresses that sending redundant information may not lead to the desired behavior during an async flip. Do you think adding a note about using the same FB_ID would be helpful?
On 8/21/23 22:02, André Almeida wrote: > Em 17/08/2023 07:37, Michel Dänzer escreveu: >> On 8/15/23 20:57, André Almeida wrote: >>> From: Pekka Paalanen <pekka.paalanen@collabora.com> >>> >>> Specify how the atomic state is maintained between userspace and >>> kernel, plus the special case for async flips. >>> >>> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >>> Signed-off-by: André Almeida <andrealmeid@igalia.com> >> >> [...] >> >>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >>> +effectively change only the FB_ID property on any planes. No-operation changes >>> +are ignored as always. [...] >> >> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. >> > > I believe the documentation already addresses that sending redundant information may not lead to the desired behavior during an async flip. Do you think adding a note about using the same FB_ID would be helpful? Maybe not.
Hi Michel, On 8/17/23 12:37, Michel Dänzer wrote: > On 8/15/23 20:57, André Almeida wrote: >> From: Pekka Paalanen <pekka.paalanen@collabora.com> >> >> Specify how the atomic state is maintained between userspace and >> kernel, plus the special case for async flips. >> >> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >> Signed-off-by: André Almeida <andrealmeid@igalia.com> > [...] > >> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >> +effectively change only the FB_ID property on any planes. No-operation changes >> +are ignored as always. [...] > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor and it worked as expected, so this shouldn't be a concern. Thanks, André
On Mon, 16 Oct 2023 12:52:32 +0200 André Almeida <andrealmeid@igalia.com> wrote: > Hi Michel, > > On 8/17/23 12:37, Michel Dänzer wrote: > > On 8/15/23 20:57, André Almeida wrote: > >> From: Pekka Paalanen <pekka.paalanen@collabora.com> > >> > >> Specify how the atomic state is maintained between userspace and > >> kernel, plus the special case for async flips. > >> > >> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> > >> Signed-off-by: André Almeida <andrealmeid@igalia.com> > > [...] > > > >> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > >> +effectively change only the FB_ID property on any planes. No-operation changes > >> +are ignored as always. [...] > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > and it worked as expected, so this shouldn't be a concern. Right, so it must have some effect. It cannot be simply ignored like in the proposed doc wording. Do we special-case re-setting the same FB_ID as "not a no-op" or "not ignored" or some other way? Thanks, pq
Hi Pekka, On 10/16/23 14:18, Pekka Paalanen wrote: > On Mon, 16 Oct 2023 12:52:32 +0200 > André Almeida <andrealmeid@igalia.com> wrote: > >> Hi Michel, >> >> On 8/17/23 12:37, Michel Dänzer wrote: >>> On 8/15/23 20:57, André Almeida wrote: >>>> From: Pekka Paalanen <pekka.paalanen@collabora.com> >>>> >>>> Specify how the atomic state is maintained between userspace and >>>> kernel, plus the special case for async flips. >>>> >>>> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >>>> Signed-off-by: André Almeida <andrealmeid@igalia.com> >>> [...] >>> >>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >>>> +effectively change only the FB_ID property on any planes. No-operation changes >>>> +are ignored as always. [...] >>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. >>> >> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor >> and it worked as expected, so this shouldn't be a concern. > Right, so it must have some effect. It cannot be simply ignored like in > the proposed doc wording. Do we special-case re-setting the same FB_ID > as "not a no-op" or "not ignored" or some other way? There's an effect in the refresh rate, the image won't change but it will report that a flip had happened asynchronously so the reported framerate will be increased. Maybe an additional wording could be like: Flipping to the same FB_ID will result in a immediate flip as if it was changing to a different one, with no effect on the image but effecting the reported frame rate. > > Thanks, > pq
On Mon, 16 Oct 2023 15:42:16 +0200 André Almeida <andrealmeid@igalia.com> wrote: > Hi Pekka, > > On 10/16/23 14:18, Pekka Paalanen wrote: > > On Mon, 16 Oct 2023 12:52:32 +0200 > > André Almeida <andrealmeid@igalia.com> wrote: > > > >> Hi Michel, > >> > >> On 8/17/23 12:37, Michel Dänzer wrote: > >>> On 8/15/23 20:57, André Almeida wrote: > >>>> From: Pekka Paalanen <pekka.paalanen@collabora.com> > >>>> > >>>> Specify how the atomic state is maintained between userspace and > >>>> kernel, plus the special case for async flips. > >>>> > >>>> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> > >>>> Signed-off-by: André Almeida <andrealmeid@igalia.com> > >>> [...] > >>> > >>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > >>>> +effectively change only the FB_ID property on any planes. No-operation changes > >>>> +are ignored as always. [...] > >>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > >>> > >> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > >> and it worked as expected, so this shouldn't be a concern. > > Right, so it must have some effect. It cannot be simply ignored like in > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > as "not a no-op" or "not ignored" or some other way? > There's an effect in the refresh rate, the image won't change but it > will report that a flip had happened asynchronously so the reported > framerate will be increased. Maybe an additional wording could be like: > > Flipping to the same FB_ID will result in a immediate flip as if it was > changing to a different one, with no effect on the image but effecting > the reported frame rate. Re-setting FB_ID to its current value is a special case regardless of PAGE_FLIP_ASYNC, is it not? So it should be called out somewhere that applies regardless of PAGE_FLIP_ASYNC. Maybe to the end of the earlier paragraph: > +The changes recorded in an atomic commit apply on top the current KMS state in > +the kernel. Hence, the complete new KMS state is the complete old KMS state with > +the committed property settings done on top. The kernel will try to avoid > +no-operation changes, so it is safe for userspace to send redundant property > +settings. However, not every situation allows for no-op changes, due to the > +need to acquire locks for some attributes. Userspace needs to be aware that some > +redundant information might result in oversynchronization issues. No-operation > +changes do not count towards actually needed changes, e.g. setting MODE_ID to a > +different blob with identical contents as the current KMS state shall not be a > +modeset on its own. +As a special exception for VRR needs, explicitly setting FB_ID to its +current value is not a no-op. Would that work? I'd like to try to avoid being more specific about what it does exactly, because that's not the topic here. Such things can be documented with the property itself. This is a summary of what is or is not a no-op or a modeset. Thanks, pq
On 10/16/23 16:52, Pekka Paalanen wrote: > On Mon, 16 Oct 2023 15:42:16 +0200 > André Almeida <andrealmeid@igalia.com> wrote: > >> Hi Pekka, >> >> On 10/16/23 14:18, Pekka Paalanen wrote: >>> On Mon, 16 Oct 2023 12:52:32 +0200 >>> André Almeida <andrealmeid@igalia.com> wrote: >>> >>>> Hi Michel, >>>> >>>> On 8/17/23 12:37, Michel Dänzer wrote: >>>>> On 8/15/23 20:57, André Almeida wrote: >>>>>> From: Pekka Paalanen <pekka.paalanen@collabora.com> >>>>>> >>>>>> Specify how the atomic state is maintained between userspace and >>>>>> kernel, plus the special case for async flips. >>>>>> >>>>>> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> >>>>>> Signed-off-by: André Almeida <andrealmeid@igalia.com> >>>>> [...] >>>>> >>>>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >>>>>> +effectively change only the FB_ID property on any planes. No-operation changes >>>>>> +are ignored as always. [...] >>>>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. >>>>> >>>> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor >>>> and it worked as expected, so this shouldn't be a concern. >>> Right, so it must have some effect. It cannot be simply ignored like in >>> the proposed doc wording. Do we special-case re-setting the same FB_ID >>> as "not a no-op" or "not ignored" or some other way? >> There's an effect in the refresh rate, the image won't change but it >> will report that a flip had happened asynchronously so the reported >> framerate will be increased. Maybe an additional wording could be like: >> >> Flipping to the same FB_ID will result in a immediate flip as if it was >> changing to a different one, with no effect on the image but effecting >> the reported frame rate. > Re-setting FB_ID to its current value is a special case regardless of > PAGE_FLIP_ASYNC, is it not? > > So it should be called out somewhere that applies regardless of > PAGE_FLIP_ASYNC. Maybe to the end of the earlier paragraph: > >> +The changes recorded in an atomic commit apply on top the current KMS state in >> +the kernel. Hence, the complete new KMS state is the complete old KMS state with >> +the committed property settings done on top. The kernel will try to avoid >> +no-operation changes, so it is safe for userspace to send redundant property >> +settings. However, not every situation allows for no-op changes, due to the >> +need to acquire locks for some attributes. Userspace needs to be aware that some >> +redundant information might result in oversynchronization issues. No-operation >> +changes do not count towards actually needed changes, e.g. setting MODE_ID to a >> +different blob with identical contents as the current KMS state shall not be a >> +modeset on its own. > +As a special exception for VRR needs, explicitly setting FB_ID to its > +current value is not a no-op. > > Would that work? I liked this suggestion, thanks! I'll wrap up a v7 > I'd like to try to avoid being more specific about what it does > exactly, because that's not the topic here. Such things can be > documented with the property itself. This is a summary of what is or is > not a no-op or a modeset. > > > Thanks, > pq
On Mon, Oct 16, 2023 at 05:52:22PM +0300, Pekka Paalanen wrote: > On Mon, 16 Oct 2023 15:42:16 +0200 > André Almeida <andrealmeid@igalia.com> wrote: > > > Hi Pekka, > > > > On 10/16/23 14:18, Pekka Paalanen wrote: > > > On Mon, 16 Oct 2023 12:52:32 +0200 > > > André Almeida <andrealmeid@igalia.com> wrote: > > > > > >> Hi Michel, > > >> > > >> On 8/17/23 12:37, Michel Dänzer wrote: > > >>> On 8/15/23 20:57, André Almeida wrote: > > >>>> From: Pekka Paalanen <pekka.paalanen@collabora.com> > > >>>> > > >>>> Specify how the atomic state is maintained between userspace and > > >>>> kernel, plus the special case for async flips. > > >>>> > > >>>> Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com> > > >>>> Signed-off-by: André Almeida <andrealmeid@igalia.com> > > >>> [...] > > >>> > > >>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > >>>> +effectively change only the FB_ID property on any planes. No-operation changes > > >>>> +are ignored as always. [...] > > >>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > >>> > > >> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > >> and it worked as expected, so this shouldn't be a concern. > > > Right, so it must have some effect. It cannot be simply ignored like in > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > as "not a no-op" or "not ignored" or some other way? > > There's an effect in the refresh rate, the image won't change but it > > will report that a flip had happened asynchronously so the reported > > framerate will be increased. Maybe an additional wording could be like: > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > changing to a different one, with no effect on the image but effecting > > the reported frame rate. > > Re-setting FB_ID to its current value is a special case regardless of > PAGE_FLIP_ASYNC, is it not? No. The rule has so far been that all side effects are observed even if you flip to the same fb. And that is one of my annoyances with this proposal. The rules will now be different for async flips vs. everything else. The other issues (mainly relating to hardware where not all planes support async flips) I've already highlighted in some earlier mail.
On Monday, October 16th, 2023 at 17:10, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote: > On Mon, Oct 16, 2023 at 05:52:22PM +0300, Pekka Paalanen wrote: > > > On Mon, 16 Oct 2023 15:42:16 +0200 > > André Almeida andrealmeid@igalia.com wrote: > > > > > Hi Pekka, > > > > > > On 10/16/23 14:18, Pekka Paalanen wrote: > > > > > > > On Mon, 16 Oct 2023 12:52:32 +0200 > > > > André Almeida andrealmeid@igalia.com wrote: > > > > > > > > > Hi Michel, > > > > > > > > > > On 8/17/23 12:37, Michel Dänzer wrote: > > > > > > > > > > > On 8/15/23 20:57, André Almeida wrote: > > > > > > > > > > > > > From: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > > > > > > > > Specify how the atomic state is maintained between userspace and > > > > > > > kernel, plus the special case for async flips. > > > > > > > > > > > > > > Signed-off-by: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > Signed-off-by: André Almeida andrealmeid@igalia.com > > > > > > > [...] > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > +are ignored as always. [...] > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > will report that a flip had happened asynchronously so the reported > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > changing to a different one, with no effect on the image but effecting > > > the reported frame rate. > > > > Re-setting FB_ID to its current value is a special case regardless of > > PAGE_FLIP_ASYNC, is it not? > > No. The rule has so far been that all side effects are observed > even if you flip to the same fb. And that is one of my annoyances > with this proposal. The rules will now be different for async flips > vs. everything else. Well with the patches the async page-flip case is exactly the same as the non-async page-flip case. In both cases, if a FB_ID is included in an atomic commit then the side effects are triggered even if the property value didn't change. The rules are the same for everything.
On Mon, Oct 16, 2023 at 10:00:51PM +0000, Simon Ser wrote: > On Monday, October 16th, 2023 at 17:10, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote: > > > On Mon, Oct 16, 2023 at 05:52:22PM +0300, Pekka Paalanen wrote: > > > > > On Mon, 16 Oct 2023 15:42:16 +0200 > > > André Almeida andrealmeid@igalia.com wrote: > > > > > > > Hi Pekka, > > > > > > > > On 10/16/23 14:18, Pekka Paalanen wrote: > > > > > > > > > On Mon, 16 Oct 2023 12:52:32 +0200 > > > > > André Almeida andrealmeid@igalia.com wrote: > > > > > > > > > > > Hi Michel, > > > > > > > > > > > > On 8/17/23 12:37, Michel Dänzer wrote: > > > > > > > > > > > > > On 8/15/23 20:57, André Almeida wrote: > > > > > > > > > > > > > > > From: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > > > > > > > > > > Specify how the atomic state is maintained between userspace and > > > > > > > > kernel, plus the special case for async flips. > > > > > > > > > > > > > > > > Signed-off-by: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > > Signed-off-by: André Almeida andrealmeid@igalia.com > > > > > > > > [...] > > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > +are ignored as always. [...] > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > changing to a different one, with no effect on the image but effecting > > > > the reported frame rate. > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > PAGE_FLIP_ASYNC, is it not? > > > > No. The rule has so far been that all side effects are observed > > even if you flip to the same fb. And that is one of my annoyances > > with this proposal. The rules will now be different for async flips > > vs. everything else. > > Well with the patches the async page-flip case is exactly the same as > the non-async page-flip case. In both cases, if a FB_ID is included in > an atomic commit then the side effects are triggered even if the property > value didn't change. The rules are the same for everything. I see it only checking if FB_ID changes or not. If it doesn't change then the implication is that the side effects will in fact be skipped as not all planes may even support async flips.
On Tuesday, October 17th, 2023 at 14:10, Ville Syrjälä <ville.syrjala@linux.intel.com> wrote: > On Mon, Oct 16, 2023 at 10:00:51PM +0000, Simon Ser wrote: > > > On Monday, October 16th, 2023 at 17:10, Ville Syrjälä ville.syrjala@linux.intel.com wrote: > > > > > On Mon, Oct 16, 2023 at 05:52:22PM +0300, Pekka Paalanen wrote: > > > > > > > On Mon, 16 Oct 2023 15:42:16 +0200 > > > > André Almeida andrealmeid@igalia.com wrote: > > > > > > > > > Hi Pekka, > > > > > > > > > > On 10/16/23 14:18, Pekka Paalanen wrote: > > > > > > > > > > > On Mon, 16 Oct 2023 12:52:32 +0200 > > > > > > André Almeida andrealmeid@igalia.com wrote: > > > > > > > > > > > > > Hi Michel, > > > > > > > > > > > > > > On 8/17/23 12:37, Michel Dänzer wrote: > > > > > > > > > > > > > > > On 8/15/23 20:57, André Almeida wrote: > > > > > > > > > > > > > > > > > From: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > > > > > > > > > > > > Specify how the atomic state is maintained between userspace and > > > > > > > > > kernel, plus the special case for async flips. > > > > > > > > > > > > > > > > > > Signed-off-by: Pekka Paalanen pekka.paalanen@collabora.com > > > > > > > > > Signed-off-by: André Almeida andrealmeid@igalia.com > > > > > > > > > [...] > > > > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > +are ignored as always. [...] > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > changing to a different one, with no effect on the image but effecting > > > > > the reported frame rate. > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > No. The rule has so far been that all side effects are observed > > > even if you flip to the same fb. And that is one of my annoyances > > > with this proposal. The rules will now be different for async flips > > > vs. everything else. > > > > Well with the patches the async page-flip case is exactly the same as > > the non-async page-flip case. In both cases, if a FB_ID is included in > > an atomic commit then the side effects are triggered even if the property > > value didn't change. The rules are the same for everything. > > I see it only checking if FB_ID changes or not. If it doesn't > change then the implication is that the side effects will in > fact be skipped as not all planes may even support async flips. Hm right. So the problem is that setting any prop = same value as previous one will result in a new page-flip for asynchronous page-flips, but will not result in any side-effect for asynchronous page-flips. Does it actually matter though? For async page-flips, I don't think this would result in any actual difference in behavior?
On Monday, October 23rd, 2023 at 10:25, Simon Ser <contact@emersion.fr> wrote: > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > the reported frame rate. > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > No. The rule has so far been that all side effects are observed > > > > even if you flip to the same fb. And that is one of my annoyances > > > > with this proposal. The rules will now be different for async flips > > > > vs. everything else. > > > > > > Well with the patches the async page-flip case is exactly the same as > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > an atomic commit then the side effects are triggered even if the property > > > value didn't change. The rules are the same for everything. > > > > I see it only checking if FB_ID changes or not. If it doesn't > > change then the implication is that the side effects will in > > fact be skipped as not all planes may even support async flips. > > Hm right. So the problem is that setting any prop = same value as > previous one will result in a new page-flip for asynchronous page-flips, > but will not result in any side-effect for asynchronous page-flips. > > Does it actually matter though? For async page-flips, I don't think this > would result in any actual difference in behavior? To sum this up, here is a matrix of behavior as seen by user-space: - Sync atomic page-flip - Set FB_ID to different value: programs hw for page-flip, sends uevent - Set FB_ID to same value: same (important for VRR) - Set another plane prop to same value: same - Set another plane prop to different value: maybe rejected if modeset required - Async atomic page-flip - Set FB_ID to different value: updates hw with new FB address, sends immediate uevent - Set FB_ID to same value: same (no-op for the hw) - Set another plane prop to same value: ignored, sends immediate uevent (special codepath) - Set another plane prop to different value: always rejected To me sync and async look consistent.
On Mon, 13 Nov 2023 09:18:39 +0000 Simon Ser <contact@emersion.fr> wrote: > On Monday, October 23rd, 2023 at 10:25, Simon Ser <contact@emersion.fr> wrote: > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > > the reported frame rate. > > > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > > > No. The rule has so far been that all side effects are observed > > > > > even if you flip to the same fb. And that is one of my annoyances > > > > > with this proposal. The rules will now be different for async flips > > > > > vs. everything else. > > > > > > > > Well with the patches the async page-flip case is exactly the same as > > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > > an atomic commit then the side effects are triggered even if the property > > > > value didn't change. The rules are the same for everything. > > > > > > I see it only checking if FB_ID changes or not. If it doesn't > > > change then the implication is that the side effects will in > > > fact be skipped as not all planes may even support async flips. > > > > Hm right. So the problem is that setting any prop = same value as > > previous one will result in a new page-flip for asynchronous page-flips, > > but will not result in any side-effect for asynchronous page-flips. > > > > Does it actually matter though? For async page-flips, I don't think this > > would result in any actual difference in behavior? Hi Simon, a fly-by question... > To sum this up, here is a matrix of behavior as seen by user-space: > > - Sync atomic page-flip > - Set FB_ID to different value: programs hw for page-flip, sends uevent > - Set FB_ID to same value: same (important for VRR) > - Set another plane prop to same value: same > - Set another plane prop to different value: maybe rejected if modeset required > - Async atomic page-flip > - Set FB_ID to different value: updates hw with new FB address, sends > immediate uevent > - Set FB_ID to same value: same (no-op for the hw) It should not be a no-op for the hw, because the hw might be in the middle of a VRR front-porch waiting period, and the commit needs to end the waiting immediately rather than time out? > - Set another plane prop to same value: ignored, sends immediate uevent > (special codepath) If the sync case says "same", then shouldn't this say "same" as well to be consistent? > - Set another plane prop to different value: always rejected > > To me sync and async look consistent. Thanks, pq
On 11/13/23 10:18, Simon Ser wrote: > On Monday, October 23rd, 2023 at 10:25, Simon Ser <contact@emersion.fr> wrote: > >>>>>>>>>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >>>>>>>>>>> +effectively change only the FB_ID property on any planes. No-operation changes >>>>>>>>>>> +are ignored as always. [...] >>>>>>>>>>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. >>>>>>>>> >>>>>>>>> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor >>>>>>>>> and it worked as expected, so this shouldn't be a concern. >>>>>>>>> Right, so it must have some effect. It cannot be simply ignored like in >>>>>>>>> the proposed doc wording. Do we special-case re-setting the same FB_ID >>>>>>>>> as "not a no-op" or "not ignored" or some other way? >>>>>>>>> There's an effect in the refresh rate, the image won't change but it >>>>>>>>> will report that a flip had happened asynchronously so the reported >>>>>>>>> framerate will be increased. Maybe an additional wording could be like: >>>>>>> >>>>>>> Flipping to the same FB_ID will result in a immediate flip as if it was >>>>>>> changing to a different one, with no effect on the image but effecting >>>>>>> the reported frame rate. >>>>>> >>>>>> Re-setting FB_ID to its current value is a special case regardless of >>>>>> PAGE_FLIP_ASYNC, is it not? >>>>> >>>>> No. The rule has so far been that all side effects are observed >>>>> even if you flip to the same fb. And that is one of my annoyances >>>>> with this proposal. The rules will now be different for async flips >>>>> vs. everything else. >>>> >>>> Well with the patches the async page-flip case is exactly the same as >>>> the non-async page-flip case. In both cases, if a FB_ID is included in >>>> an atomic commit then the side effects are triggered even if the property >>>> value didn't change. The rules are the same for everything. >>> >>> I see it only checking if FB_ID changes or not. If it doesn't >>> change then the implication is that the side effects will in >>> fact be skipped as not all planes may even support async flips. >> >> Hm right. So the problem is that setting any prop = same value as >> previous one will result in a new page-flip for asynchronous page-flips, >> but will not result in any side-effect for asynchronous page-flips. >> >> Does it actually matter though? For async page-flips, I don't think this >> would result in any actual difference in behavior? > > To sum this up, here is a matrix of behavior as seen by user-space: > > - Sync atomic page-flip > - Set FB_ID to different value: programs hw for page-flip, sends uevent > - Set FB_ID to same value: same (important for VRR) > - Set another plane prop to same value: same A page flip is programmed even if FB_ID isn't touched? > - Set another plane prop to different value: maybe rejected if modeset required > - Async atomic page-flip > - Set FB_ID to different value: updates hw with new FB address, sends > immediate uevent > - Set FB_ID to same value: same (no-op for the hw) No-op implies it doesn't trigger scanning out a frame with VRR, if scanout is currently in vertical blank. Is that the case? If so, async flips can't reliably trigger scanning out a frame with VRR.
On Monday, November 13th, 2023 at 10:38, Pekka Paalanen <ppaalanen@gmail.com> wrote: > On Mon, 13 Nov 2023 09:18:39 +0000 > Simon Ser contact@emersion.fr wrote: > > > On Monday, October 23rd, 2023 at 10:25, Simon Ser contact@emersion.fr wrote: > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > > > the reported frame rate. > > > > > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > > > > > No. The rule has so far been that all side effects are observed > > > > > > even if you flip to the same fb. And that is one of my annoyances > > > > > > with this proposal. The rules will now be different for async flips > > > > > > vs. everything else. > > > > > > > > > > Well with the patches the async page-flip case is exactly the same as > > > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > > > an atomic commit then the side effects are triggered even if the property > > > > > value didn't change. The rules are the same for everything. > > > > > > > > I see it only checking if FB_ID changes or not. If it doesn't > > > > change then the implication is that the side effects will in > > > > fact be skipped as not all planes may even support async flips. > > > > > > Hm right. So the problem is that setting any prop = same value as > > > previous one will result in a new page-flip for asynchronous page-flips, > > > but will not result in any side-effect for asynchronous page-flips. > > > > > > Does it actually matter though? For async page-flips, I don't think this > > > would result in any actual difference in behavior? > > > Hi Simon, > > a fly-by question... > > > To sum this up, here is a matrix of behavior as seen by user-space: > > > > - Sync atomic page-flip > > - Set FB_ID to different value: programs hw for page-flip, sends uevent > > - Set FB_ID to same value: same (important for VRR) > > - Set another plane prop to same value: same > > - Set another plane prop to different value: maybe rejected if modeset required > > - Async atomic page-flip > > - Set FB_ID to different value: updates hw with new FB address, sends > > immediate uevent > > - Set FB_ID to same value: same (no-op for the hw) > > It should not be a no-op for the hw, because the hw might be in the > middle of a VRR front-porch waiting period, and the commit needs to end > the waiting immediately rather than time out? I'm not sure > > - Set another plane prop to same value: ignored, sends immediate uevent > > (special codepath) > > If the sync case says "same", then shouldn't this say "same" as well to > be consistent? Okay, I think I chose my words badly. By "same" I meant "same as previous item in the list". Here I tried to be more explicit and explain why it's the same behavior. We have a special path in the kernel code that ignores the change, but the effective result is that it doesn't differ from the sync case. Here's a fixed matrix where I don't use confusing words: - Sync atomic page-flip - Set FB_ID to different value: programs hw for page-flip, sends uevent - Set FB_ID to same value: programs hw for page-flip, sends uevent (important for VRR) - Set another plane prop to same value: programs hw for page-flip, sends uevent - Set another plane prop to different value: maybe rejected if modeset required - Async atomic page-flip - Set FB_ID to different value: updates hw with new FB address, sends immediate uevent - Set FB_ID to same value: updates hw with new FB address (no-op for the hw), sends immediate uevent - Set another plane prop to same value: ignored, sends immediate uevent (special codepath) - Set another plane prop to different value: always rejected
On Monday, November 13th, 2023 at 10:41, Michel Dänzer <michel.daenzer@mailbox.org> wrote: > On 11/13/23 10:18, Simon Ser wrote: > > > On Monday, October 23rd, 2023 at 10:25, Simon Ser contact@emersion.fr wrote: > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > > > the reported frame rate. > > > > > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > > > > > No. The rule has so far been that all side effects are observed > > > > > > even if you flip to the same fb. And that is one of my annoyances > > > > > > with this proposal. The rules will now be different for async flips > > > > > > vs. everything else. > > > > > > > > > > Well with the patches the async page-flip case is exactly the same as > > > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > > > an atomic commit then the side effects are triggered even if the property > > > > > value didn't change. The rules are the same for everything. > > > > > > > > I see it only checking if FB_ID changes or not. If it doesn't > > > > change then the implication is that the side effects will in > > > > fact be skipped as not all planes may even support async flips. > > > > > > Hm right. So the problem is that setting any prop = same value as > > > previous one will result in a new page-flip for asynchronous page-flips, > > > but will not result in any side-effect for asynchronous page-flips. > > > > > > Does it actually matter though? For async page-flips, I don't think this > > > would result in any actual difference in behavior? > > > > To sum this up, here is a matrix of behavior as seen by user-space: > > > > - Sync atomic page-flip > > - Set FB_ID to different value: programs hw for page-flip, sends uevent > > - Set FB_ID to same value: same (important for VRR) > > - Set another plane prop to same value: same > > A page flip is programmed even if FB_ID isn't touched? I believe so. Set CRTC_X on a plane to the same value as before, and the CRTC gets implicitly included in the atomic commit? > > - Set another plane prop to different value: maybe rejected if modeset required > > - Async atomic page-flip > > - Set FB_ID to different value: updates hw with new FB address, sends > > immediate uevent > > - Set FB_ID to same value: same (no-op for the hw) > > No-op implies it doesn't trigger scanning out a frame with VRR, if > scanout is currently in vertical blank. Is that the case? If so, async > flips can't reliably trigger scanning out a frame with VRR. By no-op I mean that the hw is programmed for an immediate async flip with the same buffer addr as the previous one. So this doesn't actually change anything.
On 11/13/23 10:47, Simon Ser wrote: > On Monday, November 13th, 2023 at 10:41, Michel Dänzer <michel.daenzer@mailbox.org> wrote: > >> On 11/13/23 10:18, Simon Ser wrote: >> >>> On Monday, October 23rd, 2023 at 10:25, Simon Ser contact@emersion.fr wrote: >>> >>>>>>>>>>>>> +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to >>>>>>>>>>>>> +effectively change only the FB_ID property on any planes. No-operation changes >>>>>>>>>>>>> +are ignored as always. [...] >>>>>>>>>>>>> During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. >>>>>>>>>>> >>>>>>>>>>> Xaver tested this hypothesis in a flipping the same fb in a VRR monitor >>>>>>>>>>> and it worked as expected, so this shouldn't be a concern. >>>>>>>>>>> Right, so it must have some effect. It cannot be simply ignored like in >>>>>>>>>>> the proposed doc wording. Do we special-case re-setting the same FB_ID >>>>>>>>>>> as "not a no-op" or "not ignored" or some other way? >>>>>>>>>>> There's an effect in the refresh rate, the image won't change but it >>>>>>>>>>> will report that a flip had happened asynchronously so the reported >>>>>>>>>>> framerate will be increased. Maybe an additional wording could be like: >>>>>>>>> >>>>>>>>> Flipping to the same FB_ID will result in a immediate flip as if it was >>>>>>>>> changing to a different one, with no effect on the image but effecting >>>>>>>>> the reported frame rate. >>>>>>>> >>>>>>>> Re-setting FB_ID to its current value is a special case regardless of >>>>>>>> PAGE_FLIP_ASYNC, is it not? >>>>>>> >>>>>>> No. The rule has so far been that all side effects are observed >>>>>>> even if you flip to the same fb. And that is one of my annoyances >>>>>>> with this proposal. The rules will now be different for async flips >>>>>>> vs. everything else. >>>>>> >>>>>> Well with the patches the async page-flip case is exactly the same as >>>>>> the non-async page-flip case. In both cases, if a FB_ID is included in >>>>>> an atomic commit then the side effects are triggered even if the property >>>>>> value didn't change. The rules are the same for everything. >>>>> >>>>> I see it only checking if FB_ID changes or not. If it doesn't >>>>> change then the implication is that the side effects will in >>>>> fact be skipped as not all planes may even support async flips. >>>> >>>> Hm right. So the problem is that setting any prop = same value as >>>> previous one will result in a new page-flip for asynchronous page-flips, >>>> but will not result in any side-effect for asynchronous page-flips. >>>> >>>> Does it actually matter though? For async page-flips, I don't think this >>>> would result in any actual difference in behavior? >>> >>> To sum this up, here is a matrix of behavior as seen by user-space: >>> >>> - Sync atomic page-flip >>> - Set FB_ID to different value: programs hw for page-flip, sends uevent >>> - Set FB_ID to same value: same (important for VRR) >>> - Set another plane prop to same value: same >> >> A page flip is programmed even if FB_ID isn't touched? > > I believe so. Set CRTC_X on a plane to the same value as before, and the > CRTC gets implicitly included in the atomic commit? > >>> - Set another plane prop to different value: maybe rejected if modeset required >>> - Async atomic page-flip >>> - Set FB_ID to different value: updates hw with new FB address, sends >>> immediate uevent >>> - Set FB_ID to same value: same (no-op for the hw) >> >> No-op implies it doesn't trigger scanning out a frame with VRR, if >> scanout is currently in vertical blank. Is that the case? If so, async >> flips can't reliably trigger scanning out a frame with VRR. > > By no-op I mean that the hw is programmed for an immediate async flip > with the same buffer addr as the previous one. So this doesn't actually > change anything. If a flip is programmed to the HW, it's not a no-op any more than in the sync case, in particular not with VRR.
On Mon, 13 Nov 2023 09:44:15 +0000 Simon Ser <contact@emersion.fr> wrote: > On Monday, November 13th, 2023 at 10:38, Pekka Paalanen <ppaalanen@gmail.com> wrote: > > > On Mon, 13 Nov 2023 09:18:39 +0000 > > Simon Ser contact@emersion.fr wrote: > > > > > On Monday, October 23rd, 2023 at 10:25, Simon Ser contact@emersion.fr wrote: > > > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > > > > the reported frame rate. > > > > > > > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > > > > > > > No. The rule has so far been that all side effects are observed > > > > > > > even if you flip to the same fb. And that is one of my annoyances > > > > > > > with this proposal. The rules will now be different for async flips > > > > > > > vs. everything else. > > > > > > > > > > > > Well with the patches the async page-flip case is exactly the same as > > > > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > > > > an atomic commit then the side effects are triggered even if the property > > > > > > value didn't change. The rules are the same for everything. > > > > > > > > > > I see it only checking if FB_ID changes or not. If it doesn't > > > > > change then the implication is that the side effects will in > > > > > fact be skipped as not all planes may even support async flips. > > > > > > > > Hm right. So the problem is that setting any prop = same value as > > > > previous one will result in a new page-flip for asynchronous page-flips, > > > > but will not result in any side-effect for asynchronous page-flips. > > > > > > > > Does it actually matter though? For async page-flips, I don't think this > > > > would result in any actual difference in behavior? > > > > > > Hi Simon, > > > > a fly-by question... > > > > > To sum this up, here is a matrix of behavior as seen by user-space: > > > > > > - Sync atomic page-flip > > > - Set FB_ID to different value: programs hw for page-flip, sends uevent > > > - Set FB_ID to same value: same (important for VRR) > > > - Set another plane prop to same value: same > > > - Set another plane prop to different value: maybe rejected if modeset required > > > - Async atomic page-flip > > > - Set FB_ID to different value: updates hw with new FB address, sends > > > immediate uevent > > > - Set FB_ID to same value: same (no-op for the hw) > > > > It should not be a no-op for the hw, because the hw might be in the > > middle of a VRR front-porch waiting period, and the commit needs to end > > the waiting immediately rather than time out? > > I'm not sure Would people not want to use async commits to trigger new VRR scanout cycles without content updates? I seem to recall previous comments that switching between sync and async commit modes may take a moment (intel's one last sync flip), so using sync once in a while then using async otherwise is probably not a good idea. > > > - Set another plane prop to same value: ignored, sends immediate uevent > > > (special codepath) > > > > If the sync case says "same", then shouldn't this say "same" as well to > > be consistent? > > Okay, I think I chose my words badly. By "same" I meant "same as > previous item in the list". > > Here I tried to be more explicit and explain why it's the same behavior. > We have a special path in the kernel code that ignores the change, but > the effective result is that it doesn't differ from the sync case. > > Here's a fixed matrix where I don't use confusing words: > > - Sync atomic page-flip > - Set FB_ID to different value: programs hw for page-flip, sends uevent > - Set FB_ID to same value: programs hw for page-flip, sends uevent (important > for VRR) > - Set another plane prop to same value: programs hw for page-flip, sends > uevent Programming hw for page-flip probably triggers a new VRR scanout cycle, even if the FB address didn't change. > - Set another plane prop to different value: maybe rejected if modeset required > - Async atomic page-flip > - Set FB_ID to different value: updates hw with new FB address, sends > immediate uevent > - Set FB_ID to same value: updates hw with new FB address (no-op for the hw), > sends immediate uevent > - Set another plane prop to same value: ignored, sends immediate uevent > (special codepath) Just like Michel points out: if this case has a special case ignoring the set, then this case will not trigger a new VRR scanout cycle like the corresponding sync case does. > - Set another plane prop to different value: always rejected Thanks, pq
On Monday, November 13th, 2023 at 11:15, Pekka Paalanen <ppaalanen@gmail.com> wrote: > > > On Mon, 13 Nov 2023 09:44:15 +0000 > Simon Ser contact@emersion.fr wrote: > > > On Monday, November 13th, 2023 at 10:38, Pekka Paalanen ppaalanen@gmail.com wrote: > > > > > On Mon, 13 Nov 2023 09:18:39 +0000 > > > Simon Ser contact@emersion.fr wrote: > > > > > > > On Monday, October 23rd, 2023 at 10:25, Simon Ser contact@emersion.fr wrote: > > > > > > > > > > > > > > > > > > +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to > > > > > > > > > > > > > > +effectively change only the FB_ID property on any planes. No-operation changes > > > > > > > > > > > > > > +are ignored as always. [...] > > > > > > > > > > > > > > During the hackfest in Brno, it was mentioned that a commit which re-sets the same FB_ID could actually have an effect with VRR: It could trigger scanout of the next frame before vertical blank has reached its maximum duration. Some kind of mechanism is required for this in order to allow user space to perform low frame rate compensation. > > > > > > > > > > > > > > > > > > > > > > > > Xaver tested this hypothesis in a flipping the same fb in a VRR monitor > > > > > > > > > > > > and it worked as expected, so this shouldn't be a concern. > > > > > > > > > > > > Right, so it must have some effect. It cannot be simply ignored like in > > > > > > > > > > > > the proposed doc wording. Do we special-case re-setting the same FB_ID > > > > > > > > > > > > as "not a no-op" or "not ignored" or some other way? > > > > > > > > > > > > There's an effect in the refresh rate, the image won't change but it > > > > > > > > > > > > will report that a flip had happened asynchronously so the reported > > > > > > > > > > > > framerate will be increased. Maybe an additional wording could be like: > > > > > > > > > > > > > > > > > > > > Flipping to the same FB_ID will result in a immediate flip as if it was > > > > > > > > > > changing to a different one, with no effect on the image but effecting > > > > > > > > > > the reported frame rate. > > > > > > > > > > > > > > > > > > Re-setting FB_ID to its current value is a special case regardless of > > > > > > > > > PAGE_FLIP_ASYNC, is it not? > > > > > > > > > > > > > > > > No. The rule has so far been that all side effects are observed > > > > > > > > even if you flip to the same fb. And that is one of my annoyances > > > > > > > > with this proposal. The rules will now be different for async flips > > > > > > > > vs. everything else. > > > > > > > > > > > > > > Well with the patches the async page-flip case is exactly the same as > > > > > > > the non-async page-flip case. In both cases, if a FB_ID is included in > > > > > > > an atomic commit then the side effects are triggered even if the property > > > > > > > value didn't change. The rules are the same for everything. > > > > > > > > > > > > I see it only checking if FB_ID changes or not. If it doesn't > > > > > > change then the implication is that the side effects will in > > > > > > fact be skipped as not all planes may even support async flips. > > > > > > > > > > Hm right. So the problem is that setting any prop = same value as > > > > > previous one will result in a new page-flip for asynchronous page-flips, > > > > > but will not result in any side-effect for asynchronous page-flips. > > > > > > > > > > Does it actually matter though? For async page-flips, I don't think this > > > > > would result in any actual difference in behavior? > > > > > > Hi Simon, > > > > > > a fly-by question... > > > > > > > To sum this up, here is a matrix of behavior as seen by user-space: > > > > > > > > - Sync atomic page-flip > > > > - Set FB_ID to different value: programs hw for page-flip, sends uevent > > > > - Set FB_ID to same value: same (important for VRR) > > > > - Set another plane prop to same value: same > > > > - Set another plane prop to different value: maybe rejected if modeset required > > > > - Async atomic page-flip > > > > - Set FB_ID to different value: updates hw with new FB address, sends > > > > immediate uevent > > > > - Set FB_ID to same value: same (no-op for the hw) > > > > > > It should not be a no-op for the hw, because the hw might be in the > > > middle of a VRR front-porch waiting period, and the commit needs to end > > > the waiting immediately rather than time out? > > > > I'm not sure > > Would people not want to use async commits to trigger new VRR scanout > cycles without content updates? > > I seem to recall previous comments that switching between sync and > async commit modes may take a moment (intel's one last sync flip), so > using sync once in a while then using async otherwise is probably not a > good idea. Sorry, my line got cut off. I meant: "I'm not sure what you mean". > > > > - Set another plane prop to same value: ignored, sends immediate uevent > > > > (special codepath) > > > > > > If the sync case says "same", then shouldn't this say "same" as well to > > > be consistent? > > > > Okay, I think I chose my words badly. By "same" I meant "same as > > previous item in the list". > > > > Here I tried to be more explicit and explain why it's the same behavior. > > We have a special path in the kernel code that ignores the change, but > > the effective result is that it doesn't differ from the sync case. > > > > Here's a fixed matrix where I don't use confusing words: > > > > - Sync atomic page-flip > > - Set FB_ID to different value: programs hw for page-flip, sends uevent > > - Set FB_ID to same value: programs hw for page-flip, sends uevent (important > > for VRR) > > - Set another plane prop to same value: programs hw for page-flip, sends > > uevent > > Programming hw for page-flip probably triggers a new VRR scanout cycle, > even if the FB address didn't change. > > > - Set another plane prop to different value: maybe rejected if modeset required > > - Async atomic page-flip > > - Set FB_ID to different value: updates hw with new FB address, sends > > immediate uevent > > - Set FB_ID to same value: updates hw with new FB address (no-op for the hw), > > sends immediate uevent > > - Set another plane prop to same value: ignored, sends immediate uevent > > (special codepath) > > Just like Michel points out: if this case has a special case ignoring > the set, then this case will not trigger a new VRR scanout cycle like > the corresponding sync case does. Ignore the prop change, but still include the CRTC in the commit, so everything is fine.
diff --git a/Documentation/gpu/drm-uapi.rst b/Documentation/gpu/drm-uapi.rst index 65fb3036a580..b91ccaddeeb9 100644 --- a/Documentation/gpu/drm-uapi.rst +++ b/Documentation/gpu/drm-uapi.rst @@ -486,3 +486,47 @@ and the CRTC index is its position in this array. .. kernel-doc:: include/uapi/drm/drm_mode.h :internal: + +KMS atomic state +================ + +An atomic commit can change multiple KMS properties in an atomic fashion, +without ever applying intermediate or partial state changes. Either the whole +commit succeeds or fails, and it will never be applied partially. This is the +fundamental improvement of the atomic API over the older non-atomic API which is +referred to as the "legacy API". Applying intermediate state could unexpectedly +fail, cause visible glitches, or delay reaching the final state. + +An atomic commit can be flagged with DRM_MODE_ATOMIC_TEST_ONLY, which means the +complete state change is validated but not applied. Userspace should use this +flag to validate any state change before asking to apply it. If validation fails +for any reason, userspace should attempt to fall back to another, perhaps +simpler, final state. This allows userspace to probe for various configurations +without causing visible glitches on screen and without the need to undo a +probing change. + +The changes recorded in an atomic commit apply on top the current KMS state in +the kernel. Hence, the complete new KMS state is the complete old KMS state with +the committed property settings done on top. The kernel will try to avoid +no-operation changes, so it is safe for userspace to send redundant property +settings. However, not every situation allows for no-op changes, due to the +need to acquire locks for some attributes. Userspace needs to be aware that some +redundant information might result in oversynchronization issues. No-operation +changes do not count towards actually needed changes, e.g. setting MODE_ID to a +different blob with identical contents as the current KMS state shall not be a +modeset on its own. + +A "modeset" is a change in KMS state that might enable, disable, or temporarily +disrupt the emitted video signal, possibly causing visible glitches on screen. A +modeset may also take considerably more time to complete than other kinds of +changes, and the video sink might also need time to adapt to the new signal +properties. Therefore a modeset must be explicitly allowed with the flag +DRM_MODE_ATOMIC_ALLOW_MODESET. This in combination with +DRM_MODE_ATOMIC_TEST_ONLY allows userspace to determine if a state change is +likely to cause visible disruption on screen and avoid such changes when end +users do not expect them. + +An atomic commit with the flag DRM_MODE_PAGE_FLIP_ASYNC is allowed to +effectively change only the FB_ID property on any planes. No-operation changes +are ignored as always. Changing any other property will cause the commit to be +rejected.