We're already inhibiting real time scheduling when reading new KMS state
after hot plugs, as well as when during mode sets, due to the kernel not
being able to reliably handle these within the 250 ms limit. However, we
didn't do this during initial probing, which meant that occasionally
we'd run into these kind of issues during startup.
Handle this by always inhibiting real time scheduling up front, and
don't uninhibit until all initially discovered device have finished
processing their initial mode set.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/3628
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3960>
And take it into account in meta_kms_crtc_get_deadline_evasion.
This uses the same fundamental approach as clutter frame clock scheduling:
Measure the deadline timer dispatch duration, keep track of the longest
duration, and set the timer to fire such that the longest measured
dispatch duration would result in it completing shortly before start of
vblank.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/3612
v2:
* Move DEADLINE_EVASION_CONSTANT_US addition from
meta_kms_crtc_determine_deadline to meta_kms_crtc_get_deadline_evasion.
* Calculate how long before start of vblank dispatch completed for
debug output in crtc_frame_deadline_dispatch.
* Shorten over-long lines in crtc_frame_deadline_dispatch.
v3:
* Take VRR into account in crtc_frame_deadline_dispatch &
meta_kms_crtc_update_shortterm_max_dispatch_duration. (Robert Mader)
v4:
* Check if deadline has already passed in meta_kms_crtc_determine_deadline,
set the deadline for one refresh interval later if so.
* Fix indentation in crtc_frame_deadline_dispatch.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3934>
We don't want to do the work of the layout mode detection and conversion
every time we read the monitors.xml file.
Instead, when the detection logic is used, set a flag to automatically
update the config files after the parsing is finished.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
Introduce some "best effort" conversion code to migrate monitor configurations
from PHYSICAL (the old default) to LOGICAL (the new default on wayland)
layout mode.
This conversion will only be used when the old PHYSICAL layout-mode
configuration is not compatible with the new LOGICAL layout-mode one.
This only applies if 1) there's a monitor that needs scaling in the
layout, and 2) the scaled monitor comes before other monitors in the
coordinate system (ie. it's not the rightmost or bottommost monitor).
There are two algorithms added here to convert monitor layouts:
- One for "simple" 1-dimensional monitor layouts, where all monitors are
aligned on a vertical or horizontal strip.
Here's a few (inaccurate) examples of how this would look with different
layouts (left side is PHYSICAL, right side is LOGICAL, x is the origin of
the coordinate system, the numbers are scales of the monitors):
```
x──────┬──────┬──────┐ x ┌──────┐
│ 2 │ 1 │ 2 │ 2┌──┤ 1 ├──┐2
│ │ │ │ ──► └──┤ ├──┘
└──────┴──────┴──────┘ └──────┘
x ┌──────┐ x ┌──────┐
┌────┤ 1 │ ┌──┤ 1 │
│ 2 │ │ ──► └──┤ │
└────┴──────┘ 2 └──────┘
x ┌────┐
┌──────┤ │ x──────┐
│ │ │ │ ├─┬────┐
│ 1 │ 3 │ ──► │ 1 │3│ 1 │
│ │ │ │ ├─┴────┘
└──────┤ │ └──────┘
│ ├────┐
│ │ 1 │
└────┴────┘
```
- A second more complex algorithm for 2-dimensional monitor layouts with
a common baseline that all monitors are aligned to.
And examples for this one:
```
x ┌──────┐
┌──────┤ │
│ 1 │ 2 │ x──────┐
│ │ │ │ 1 ├────┐
└──┬───┴───┬──┘ ──► │ │ 2 │
│ 3 │ ├──┬───┴────┘
│ │ └──┘3
└───────┘
x ┌──────┬──────┐
│ 1 │ │
│ │ │ x──────┬──────┐
┌─────┴──────┤ 1 │ │ 1 │ │
│ │ │ │ │ │
│ │ │ ──► └──┬───┤ 1 │
│ 3 │ │ │ 3 │ │
│ ├──────┘ └───┤ │
│ │ │ │
│ │ └──────┘
└────────────┘
x ┌───────┐
┌──────┐ │ │ x ┌───────┐
│ 2 │ ┌──────┤ 1 │ │ │
│ │ │ 1 │ │ 2 ┌──────┤ 1 │
└────┬─┴────┴─┬────┴───────┘ ──► ┌──┤ 1 │ │
│ │ ├──┴┬─────┴───────┘
│ 2 │ │ 2 │
│ │ └───┘
└────────┘
```
These algorithms will fail for any more complex 2d monitor layout, eg.
```
x ┌───┬────┐
│ 2 │ 1 │
│ ├────┘
┌───┴┬──┘
│ 1 │
└────┘
x───┬───┬───┐
│ 1 │ 2 │ 1 │
├───┼───┼───┤
│ 1 │ 1 │ 1 │
├───┼───┼───┤
│ 1 │ 1 │ 1 │
└───┴───┴───┘
```
In those cases where the conversion failed, we fall back to aligning
the monitors on a horizontal line, preserving the scale, the primary
monitor and the disabled monitors.
Note that we also need to convert the scale factor in some cases,
because LOGICAL layout mode also behaves different here:
When the scale results in a fractional logical monitor size (eg.
the native monitor width is 2560px, and a scale of 3 is set =>
2560px / 3 = 853.333px), in LOGICAL mode we won't use that scale.
Instead we have an algorithm (see
meta_monitor_get_closest_scale_factor_for_resolution()) to find
the nearest fractional scale factor which doesn't result in
fractional logical monitor size. We reuse this algorithm here for
the conversion.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
We'll reuse meta_monitor_get_closest_scale_factor_for_resolution() for the
conversion of monitor configs, and during those conversions, we probably don't
want to impose the same limits to fractional scales that we usually impose.
This means that we can even convert physical layout configs where the user
manually changed to a value higher than what our fractional scale calculations
would allow.
Move this check into the calling function so that it's not imposed by
meta_monitor_get_closest_scale_factor_for_resolution() directly.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
When there is no layout mode set in a logical monitor config, we currently just
assume the configuration matches the mode that the system expects. This blows
up when the layout mode expected by the system changes (eg. by turning on
"scale-monitor-framebuffers" in mutter): Suddenly configs fail the validation
check and get thrown away.
Since we now can add one configuration for each layout mode to the config store,
we can do better here: Let's only add configurations to the store where we
verified beforehand that the monitor layout is compatible with that mode, either
because we set it ourselves using the <layout_mode> key, or by detecting which
modes the layout is compatible with.
Also update monitor config ifiles to adjust for the new layout_mode, as
they all are assumed to be "logical".
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
Verify even more assumptions we make about logical monitor configs:
- Have a more explicit check that the monitor modes in the logical monitor are
all equal
- Complain if scale factor with physical layout mode is fractional
- Make sure that scale factor with logical layout mode actually scales to a
non-fractional width and height
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
We'll need a few of those things from the monitor config store soon, also it's
generally useful to have a prefix which makes it clear where functions are
defined.
So factor some things out into a new monitor-config-utils.c file.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
Store and load the layout mode for each logical monitor configuration in
monitors.xml by introducing a new <layoutmode> element. The value of the
element can be either "logical" or "physical". The layout mode is also
made part of the monitor configuration key.
Right now this isn't doing a lot:
When no <layoutmode> is found on the config (this is the case with all
existing configs), we'll keep using the layout mode expected by the system,
without updating the config file.
When changing an existing, or introducing a new configuration, we'll now
store the current layout mode with the config though, and load it again
on the next start of mutter. This is still not problematic as long as
mutters expected layout mode doesn't change (eg. by turning on/off
"scale-monitor-framebuffers").
When the expected layout mode of mutter switches between
restarts, the monitor config is now still loaded but remains unused,
and mutter will create (and store) a new one with the other layout mode.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
We'll introduce some new migration code with the next few commits to introduce
a layout_mode property in monitors.xml. This will be significantly easier
without keeping around the old monitor migration code, so drop it.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
We have meta_verify_logical_monitor_config() already, and it does a few checks that
meta_verify_monitors_config() doesn't do yet, so let's also call
meta_verify_logical_monitor_config() when verifying the whole config.
We'll rely on this being part of meta_verify_monitors_config() soon, because we'll
stop calling meta_verify_logical_monitor_config() from the config parser.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
We forgot to check whether multiple groups of monitors are actually
all connected with each other, so fix that.
[jadahl: Rewrote algorithm to detect split groups]
[jadahl: Added test case]
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3596>
On a hybrid machine with i915 primary and nvidia-drm (470) secondary,
`meta_render_device_egl_stream_initable_init` calls
`meta_kms_inhibit_kernel_thread` to change from the default 'kernel'
thread type to 'user'. And soon after that it would
`meta_render_device_egl_stream_finalize` because I'm not actually
using that GPU, and calls `meta_kms_uninhibit_kernel_thread`.
So during startup Mutter would default to a realtime kernel thread,
switch to a user thread (which doesn't support realtime), and then
switch back to a realtime kernel thread.
In the middle of all that, while the thread type was 'user' and
realtime disabled, something was invoking `ensure_crtc_frame` which
created a `CrtcFrame` without a deadline timer. Soon after that the
thread type changed back to 'kernel' with deadline timers expected, but
our existing `CrtcFrame` has no deadline timer associated with it. And
so it would never fire, causing the cursor to freeze whenever the primary
plane isn't changing. And the problem was permanent, not just the first
frame because each `CrtcFrame` gets repeatedly reused (maybe shouldn't
be called a "Frame"?).
Now we adapt to switching between kernel and user thread types by adding
and removing the deadline timer as required.
Close: https://gitlab.gnome.org/GNOME/mutter/-/issues/3464
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3950>
Colord is a system service which will result in a polkit dialog showing
up when connecting a remote session.
We want to get rid of colord eventually anyway, so disconnecting virtual
monitors from colord isn't an issue.
Fixes: f5ce2ddf3c ("color-manager: Create color devices also for virtual monitors")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3942>
If we finish compositing in time, the composited result will be
submitted prior to the deadline timer is triggered, and we'll be fine,
and if not, at least the cursor updates will be smooth, which makes it
appear smoother than not.
There is a risk that this can negatively impact composited updates when
moving the cursor, so make it possible to toggle a paint-debug flag for
now until this has been more tested.
This also mean we need to disarm the deadline timer after handling
update, as there might be a scheduled cursor update pending, but we
already handled it, so disarm the timer.
Here is an illustration of the difference.
In the following scenario, with disarming, the composited frame E, and
the cursor movement C gets presented. With this branch, only the cursor
movement C gets presented.
```
* A: beginning of composited frame
* B: begin notification reaches KMS thread
* C: cursor moved
* D: calculated deadline dispatch time (disabled with the branch)
* E: KMS update posted
* F: KMS update reaches KMS thread
* G: actual deadline (and with branch and gets committed)
Compositor thread: --------A---------------E---------
\ \
\ \
KMS thread: -----------B------C----D---F-G----
```
In the following scenario, by not disarming, the cursor update C will be
presented, and the would-be-delayed composited frame E would be delayed
anyway, i.e. fixing cursor stutter.
```
* A: beginning of composited frame
* B: begin notification reaches KMS thread
* C: cursor moved
* D: calculated deadline dispatch time (and with branch will be dispatched)
* E: KMS update posted
* F: actual deadline
* G: KMS update reaches KMS thread (and with branch gets postponed)
Compositor thread: --------A---------------E---------
\ \
\ \
KMS thread: -----------B------C----D-F-G------
```
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3184>
Although we track updates for EGL_DEVICE, they are often empty because
the primary plane has a custom page flip method. That means there's
no CRTC latched yet, but we do know exactly which CRTC is associated
with the flip. Set it so the update can still be processed.
Fixes: 27ed069766 ("kms/impl-device: Add deadline based KMS commit scheduling")
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3939>
The relationship between MetaKmsConnector and MetaDrmLease is already
stored in MetaDrmLeaseManager::leased_connectors.
Change the type of MetaDrmLeaseManager::connectors to a GList.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3922>
As in the protocol definition for wp_drm_lease_connector_v1::withdrawn:
Sent to indicate that the compositor will no longer honor requests
for DRM leases which include this connector. [...] Compositors are
encouraged to send this event when [...] the connector gets leased
to a client.
Withdrawn the leased connectors and, if they are available once the
lease finishes, advertise them again.
Related to: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/322/
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3922>
And stop passing in the color states from the RendererNative. We also
keep the color states updated by listening for changes in the color
device.
The RendererX11Cm has a single view and no mapping to a specific color
device, so we handle the absense of a color device as well and rely on
ClutterStageView to have the default color states.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3930>
This allows us to destroy and create a new offscreen dynamically, when
the rotation or color state changes.
An idle gsource with priority higher than CLUTTER_PRIORITY_REDRAW is
used to ensure the an offscreen exists when required without having to
allocate in the redraw process.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3930>
Currently, we would only disable a11y if a certain flag is passed
but the function is always called with NONE flag. Instead
drop the flag, use a new environment variable for that
That value is then used by actors to short-circuit get_accessible
implementation and return NULL if the accessibility is not enabled
Also clean the other accessibility functions
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3917>
Instead of doing that in both MetaStage & CallyStage.
This allows ClutterStage to also emits the relavant acessibility
bits directly without having a roundtrip through Cally
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3917>
The generic term updated can mean anything. This is specifically about
calibration related updates like changing the sink colorimetry
(Colorspace, HDR metadata) and changes to the white point for night
light etc.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3904>
Previously the color device was destroyed when it was attached to a
monitor that was going away. However, the MetaMonitor objects are
ref-counted and can stay around for longer, even if the underlying
resources went away. We need color devices for as long as the
MetaMonitors are alive.
Part-of: <https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3904>