The previous logic didn't work correctly at least when priority-based
preeption wasn't supported by the DRM driver, such as in the case
of amdgpu. The call to glGetQueryObjecti64v would block on client
work which is already in progress (most likely for the next frame)
and delay notifying the ClutterFrameClock about presentation.
Conveniently, the Wayland transactions mechanism guarantees that all
fences of a dma-buf buffer are signalled before the buffer is
included in a frame, which means that dma-buf buffers are ready for
presentation when being directly scanned-out.
Direct scanout is only supported for dma-buf buffers too, which means
that all buffers going through direct scanout are effectively ready
and require no GPU rendering before presentation.
Assuming zero rendering time for dma-buf buffers going through direct
scanout simplifies the code and removes the need for
glGetQueryObjecti64v, thus avoiding the aforementioned issue where it
could block for longer than expected.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2766
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3080>
The existence of the KMS property just means that we can send an
InfoFrame but we also have to make sure the sink actually supports the
metadata type 1 and the selected transfer function.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2879>
We relied on them being valid longer to keep track of used GPUs. If we
don't have the CRTC (or output) we don't have a way to fetch the pointer
to the MetaGpu that drives the associated monitor.
This avoids a crash when trying to fetch said pointer from what would be
the NULL MetaCrtc pointer.
Fixes: 08593ea872 ("onscreen/native: Hold ref to the output and CRTC until detached")
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2667
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2887>
When an onscreen is "attached" it means it has an active CRTC and output
it interacts with, e.g. listens to configuration changes to update gamma
and privacy screen state.
MetaOutput and MetaCrtc are rather short lived objects meaning they are
disposed of and regenerated each time the compositor reloads monitor
resources, and while MetaOutput are indirectly kept alive due to the
MetaMonitor holding on to them during reloading, the same does not apply
to MetaCrtc, so to avoid trying to disconnect our signals from
disappeared outputs and CRTCs when we dispatch, hold our own references
to these objects.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2665
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2885>
On exit, explicitly detach the onscreens during disposal. This means no
functional changes, but allows for doing more cleanup on detach that
doesn't need to be repeated on disposal.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2885>
In order to make things more and more asynchronus and to each time we
paint be an isolated event, that can be potentially be applied
individually or together with other updates, make it so that each time
we draw, we use the transient MetaFrameNative (ClutterFrame) instance to
carry a KMS update for us.
For this to work, we also need to restructure how we apply mode sets.
Previously we'd amend the same KMS update each frame during mode set,
then after the last CRTC was composited, we'd apply the update that
contained updates for all CRTC.
Now each CRTC has its own KMS update, and instead we put them in a per
device table, and whenever we finished painting, we'll merge the new
update into any existing one, and then finally once all CRTCs have been
composited, we'll apply an update that contains all the mode sets for all
relevant CRTCs on a device.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2855>
Instead of using the "discarded" page flip callback when the
"discarding" happened during actual immediate processing, communicate
the same via the KMS update feedback.
The "discarded" page flip callback is instead used only for when a
posted page flip is discarded. In the atomic backend, this only happens
on shutdown, while in the simple backend, this also happens when a
asynchronous retry sequence eventually is abandoned.
This allows further improvements making KMS handling fully async.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2854>
We test direct client buffer scanout using a TEST_ONLY commit on atomic,
and with various conditions in non-atomic, but if we end up failing to
actually commit despite this, handle the fallout asynchronously. What
this means is that we'll reschedule a new frame immediately.
For this to work, the same scanout buffer needs to be avoided for the
same CRTC. This is done by using the newly added signal on the
CoglScanout object to let the MetaWaylandBuffer object mark the current
buffer as non-working for the onsrceen that it failed on. This allows to
re-try buffers on the same onscreen when new ones are attached.
This queues a full damage, since we consumed the qeued redraw rect. The
redraw rect wasn't lost - it was accumulated to make sure the whole
primary plane was redrawed according to the damage region, whenever we
would end up no longer doing direct scanout, but this accumulation only
works when we're not intentionally stopping to scanout. For now, lets
just damage the whole view, it's just an graceful fallback in response
to an unexpected error anyway.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2854>
If we get a "ready" page flip feedback, it means the page flip was
symbolic, i.e. not real, e.g. as a result of an update that didn't
change the state of the primary plane. Warn if there is a "next fb"
meaning we expected to have a new buffer that we flipped to.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2854>
With detach meaning having the onscreen stop listening on configuration
changes on the corresponding backing mode setting objects. We need to do
this as there is a time between rebuilding the views, and that the new
mode sets are called, where the old onscreen is kept alive, but the
stage view is gone. At this point in time, if privacy screen or gamma
configuration changes, e.g. by the night light temperature changing, the
onscreen would attempt to schedule an update on the now gone stage view.
This commit also renames the "keep onscreen alive" to "detached
onscreens" to more clearly communicate that it's detached onscreens from
their corresponding mode setting objects.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2621
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2863>
So we can remove the additional `next_fb` and `current_fb` pointers from
`MetaOnscreenNativeSecondaryGpuState`.
Some non-scanout buffers also need to be held in the case of GL blitting
which completes in the background. Those are referenced from the scanout
buffers themselves to ensure the source buffers live just as long.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2087>
As with GAMMA_LUT, track whether privacy screen state has been pushed to
KMS in the onscreen. This leaves MetaOutput and MetaCrtc to be about
configuration, and not application.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2814>
As with CRTC GAMMA_LUT, we're moving towards making the entity managing
KMS updates aware if there are any changes to be made, and whether KMS
updates are actually needed or not, and for privacy screen changes, this
means we need to communicate whether the privacy screen state is valid
or not. This allows the caller to create any needed MetaKmsUpdate.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2814>
We're moving towards making the entity managing KMS updates aware if
there are any changes to be made, and whether KMS updates are actually
needed or not, and for GAMMA_LUT changes, this means we need to
communicate whether the GAMMA_LUT state is valid or not. This allows the
caller to create any needed MetaKmsUpdate.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2814>
We rather confusingly still call a secondary display card that is
GPU-less (DisplayLink or other basic KMS device) a "secondary GPU",
so just because secondary_gpu_state is non-NULL doesn't mean we
can use it for rendering. The clearest indication of this is when
there is no EGL surface.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2341>
At first glance the `goto` looks like a loop, or potentially an infinite
loop. It's not a loop because the mode has changed at that point to
`META_SHARED_FRAMEBUFFER_COPY_MODE_PRIMARY`. But we can make it more
obvious and avoid the need for a goto.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2240>
There was a sanity check that complained if there was still a "next
framebuffer" when disposing an onscreen. This is correct to complain
about under normal operation, as we always wait until receiving the page
flip callback before cleaning up the onscreen and their state.
However, when there are many hotplugs occurring, we might end up with
race conditions when the above sanity check is not valid: when we have
more than one monitor active, paint 1 one of them, but receive a hotplug
event before we paint the other(s), we will discard the already painted
onscreen before really issuing a page flip.
In this situation, we will have the "next framebuffer", but having that
is not a bug, it's a race condition, thus to not leak in this situation,
make sure to clean up the next framebuffer here too.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/2081
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2225>
This significantly increases the chance of a fullscreen surface buffer
being scanned out instead of being painted via composition. This is
assuming the client supports the DMA buffer feedback Wayland protocol.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2146>
An incorrect assumption that after mode set there would be no pending
page flips was made. This meant that if there was a mode set, followed
by a page flip, if that page flip was for a CRTC on a now unused GPU,
we'd crash due to the renderer GPU data having already been freed. This
commit avoids that by keeping it alive as long as the page flips are
still in the air. It fixes crashes with backtraces such as
0) meta_render_device_get_egl_display (render_device=0x0)
at ../src/backends/native/meta-render-device.c:320
1) secondary_gpu_state_free (secondary_gpu_state=0x1c8cc30)
at ../src/backends/native/meta-onscreen-native.c:560
2) meta_onscreen_native_dispose (object=0x1cb65e0)
at ../src/backends/native/meta-onscreen-native.c:2168
3) g_object_unref (_object=<optimized out>)
at ../gobject/gobject.c:3540
4) g_object_unref (_object=0x1cb65e0)
at ../gobject/gobject.c:3470
5) clutter_stage_view_finalize (object=0x1cbb450)
at ../clutter/clutter/clutter-stage-view.c:1412
6) g_object_unref (_object=<optimized out>)
at ../gobject/gobject.c:3578
7) g_object_unref (_object=0x1cbb450)
at ../gobject/gobject.c:3470
8) meta_kms_page_flip_closure_free (closure=0x1d47e60)
at ../src/backends/native/meta-kms-page-flip.c:76
9) g_list_foreach (list=<optimized out>, func=0x7fb3ada67111 <meta_kms_page_flip_closure_free>, user_data=0x0)
at ../glib/glist.c:1090
10) g_list_free_full (list=0x1cb4d20 = {...}, free_func=<optimized out>)
at ../glib/glist.c:244
11) meta_kms_page_flip_data_unref (page_flip_data=0x1c65510)
at ../src/backends/native/meta-kms-page-flip.c:109
12) meta_kms_callback_data_free (callback_data=0x227ebf0)
at ../src/backends/native/meta-kms.c:372
13) flush_callbacks (kms=0x18e2630)
at ../src/backends/native/meta-kms.c:391
14) callback_idle (user_data=0x18e2630)
at ../src/backends/native/meta-kms.c
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2147>
It was dropping to zero after the first frame because it hadn't been
incremented high enough. So the second frame would crash with:
```
#0 g_type_check_instance_cast
#1 META_DRM_BUFFER
#2 copy_shared_framebuffer_cpu
```
That's the CPU-copy path (fallback-fallback) that probably no one is using
but it does work after this fix. Exactly the same issue as was fixed
in `copy_shared_framebuffer_primary_gpu` by 36352f44f9.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2104>
Systems with AMD GPUs do not take advantage of Mutter's zero-copy path
when driving DisplayLink screens. This is due to a very slow CPU access
to the zero-copy texture. Instead they fall back on primary GPU doing a
copy of the texture for fast CPU access. This commit accelerates texture
copy by working through damage regions only.
Tests on a 4K screen with windowed applications show significant
reduction of GPU utilisation.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/2033>
This replaces functionality that MetaRenderDevice and friends has
learned, e.g. buffer allocation, EGLDisplay creation, with the usage of
those helper objects. The main objective is to shrink
meta-renderer-native.c and by extension meta-onscreen-native.c, moving
its functionality into more isolated objects.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1854>
Scanout doesn't go through the usual path of compositing and doing
eglSwapBuffers, therefore it doesn't hit the timestamp query placed in
that path. Instead, get the timings by binding the scanout buffer to an
FBO and doing a timestamp query on the FBO.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1762>
The DRM buffers aren't really tied to mode setting, so they shouldn't
need to have an associated mode setting device. Now that we have a
device file level object that can fill this role, port over
MetaDrmBuffer and friends away from MetaKmsDevice to MetaDeviceFile.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1828>
When drmModePageFlip() or drmModeAtomicCommit() unexpectedly failed (e.g.
ENOSPC, which has been seen in the wild), this failure was not handled
very gracefully. The page flip listener for the scanout was left in the
MetaKmsUpdate, meaning when the primary plane composition was later page
flipped, two page flip listeners were added, one for the primary plane,
and one for the scanout. This caused the 'page-flipped' event to be
handled twice, the second time being fatal.
Handle this by making 'no-discard' listener flag be somewhat reversed,
and say 'drop-on-error', and then drop all 'drop-on-error' listeners
when a MetaKmsUpdate failed to be processed.
Also for a "preserve" flagged update, don't ever trigger "discard"
callbacks just yet, as preserved updates are used again for the primary
plane composition, in order to not miss e.g. CRTC gamma updates, or
cursor plane updates, which were added separately.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1809
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1910>
We already swapped the front buffer, and even if it didn't get
presented, we should still swap our representation of the state, to not
get into a confused buffer tracking state.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1822>
