Even when a direct client buffer has a compatible format, stride and
modifier for direct scanout, drmModePageFlip() may still fail sometimes.
From testing, it has been observed that it may seemingly randomly fail
with ENOSPC, where all subsequent attempts later on the same CRTC
failing with EBUSY.
Handle this by falling back to flipping after having composited a full
frame again.
Closes: https://gitlab.gnome.org/GNOME/mutter/-/issues/1410
When using its EGLStream-based presentation path with the proprietary NVIDIA
driver, mutter will use a different function to process page flips -
custom_egl_stream_page_flip. If that fails due to an EBUSY error, it will
attempt to retry the flip. However, when retrying, it unconditionally uses the
libdrm-based path. In practice, this causes a segfault when attempting to
access plane_assignments->fb_id, since plane_assignments will be NULL in the
EGLStream case. The issue can be reproduced reliably by VT-switching away from
GNOME and back again while an EGL application is running.
This patch has mutter also use the custom page flip function when retrying the
failed flip.
https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1375
If drmModePageFlip() or custom_page_flip_func fails, process_page_flip() was
forgetting to undo the ref taken for that call. This would leak page_flip_data.
The reference counting works like this:
- when created, ref count is 1
- when calling drmModePageFlip, ref count is increased to 2
- new: if flip failed, ref count is decreased back to 1
- if calling schedule_retry_page_flip(), it takes a ref internally
- if calling mode_set_fallback(), it takes a ref internally
- all return FALSE paths have an explicit unref
- return TRUE path has an explicit unref
This issue was found by code inspection and while debugging an unrelated issue
with debug prints sprinkled around. I am not aware of any end-user visible
issues being fixed by this, as the leak is small and probably very rare.
https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1209
The transactional KMS API has been modelled after atomic KMS. Atomic KMS
currently doesn't support forwarding cursor hotspot metadata, thus it
was left out of the transactional KMS API having the user set the simply
create a plane assigment with the cursor sprite assigned to a cursor
plane using regular coordinates.
This, however, proved to be inadequate for virtual machines using
"seamless mouse mode" where they rely on the cursor position to
correspond to the actual cursor position of the virtual machine, not the
cursor plane. In effect, this caused cursor positions to look "shifted".
Fix this by adding back the hotspot metadata, right now as a optional
field to the plane assignment. In the legacy KMS implementation, this is
translated into drmModeSetCursor2() just as before, while still falling
back to drmModeSetCursor() with the plane coordinates, if either there
was no hotspot set, or if drmModeSetCursor2() failed.
Eventually, the atomic KMS API will learn about hotspots, but when
adding our own atomic KMS backend to the transacitonal KMS API, we must
until then still fall back to legacy KMS for virtual machines.
https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/1136
When a page flip fails with a certain error code, we've treated this as
a hint that page flipping is broken and we should try to use mode
setting instead.
On some drivers, it seems that this error is also reported when there
was no mode set, which means we'll have no cached mode set to use in the
fallback. The lack of prior mode set tends to happen when we hit a race
when the DRM objects change before we have the time to process a hotplug
event.
Handle the lack a missing mode set in the flip fallback path, with the
assumption that we'll get a hotplug event that'll fix things up for us
eventually.
Closes: https://gitlab.gnome.org/GNOME/mutter/issues/917
A cursor plane can now be assigned, and for the simple KMS
implementation, it'll translate into drmModeSetCursor() and
drmModeMoveCursor() calls.
When assignments failed, the cursor planes that failed to be assigned
are communicated via the feedback object.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/930
The current API as all synchronous, so they can be made to return
feedback immediately. This will be needed for the cursor renderer which
needs to know whether it should fall back to OpenGL cursor rendering.
https://gitlab.gnome.org/GNOME/mutter/merge_requests/930
If we end up trying to do a mode set on a DRM state that has already
changed behind our back without us yet having seen the hotplug event we
may fail with `EINVAL`. Since the renderer layer doesn't handle mode set
failure, it'll still try to page flip later on, which will then also
fail. When failing, it'll try to look up the cached mode set in order to
retry the mode set later on, as is needed to handle other error
conditions. However, if the mode set prior to the page flip failed, we
won't cache the mode set, and the page flip error handling code will get
confused.
Instead of asserting that a page flip always has a valid cached mode set
ready to look up, handle it being missing more gracefully by failing to
mode set. It is expected that things will correct themself as there
should be a hotplug event waiting around the the corner, to reconfigure
the monitor configuration setting new modes.
Fixes: https://gitlab.gnome.org/GNOME/mutter/issues/917https://gitlab.gnome.org/GNOME/mutter/merge_requests/1007
Presumably this function is supposed to be like
meta_kms_impl_simple_handle_page_flip_callback() but the condition in the
if-statement is inverted. Fix the inversion to make these two functions look
alike.
This is part 2 of 2 fixing a complete desktop freeze when drmModePageFlip()
fails with EINVAL and the fallback to drmModeSetCrtc() succeeds but the success
is not registered correctly as completed "flip". The freeze occurs under
wait_for_pending_flips() which calls down into meta_kms_impl_device_dispatch()
which ends up poll()'ing the DRM fd even though drmModeSetCrtc() will not
produce a DRM event, hence the poll() never returns. The freeze was observed
when hotplugging a DisplayLink dock for the first time on Ubuntu 19.10.
This patch makes meta_set_fallback_feedback_idle() actually end up calling into
notify_view_crtc_presented() which decrements
secondary_gpu_state->pending_flips so that wait_for_pending_flips() can finish.
CC stable: gnome-3-34
https://gitlab.gnome.org/GNOME/mutter/merge_requests/953
mode_set_fallback() schedules a call to mode_set_fallback_feedback_idle(), but
it is possible for Mutter to repaint before the idle callbacks are dispatched.
If that happens, mode_set_fallback_feedback_idle() does not get called before
Mutter enters wait_for_pending_flips(), leading to a deadlock.
Add the needed interfaces so that meta_kms_device_dispatch_sync() can flush all
the implementation idle callbacks before it checks if any "events" are
available. This prevents the deadlock by ensuring
mode_set_fallback_feedback_idle() does get called before potentially waiting
for actual DRM events.
Presumably this call would not be needed if the implementation was running in
its own thread, since it would eventually dispatch its idle callbacks before
going to sleep polling on the DRM fd. This call might even be unnecessary
overhead in that case, synchronizing with the implementation thread needlessly.
But the thread does not exist yet, so this is needed for now.
This is part 1 of 2 fixing a complete desktop freeze when drmModePageFlip()
fails with EINVAL and the fallback to drmModeSetCrtc() succeeds but the success
is not registered correctly as completed "flip". The freeze occurs under
wait_for_pending_flips() which calls down into meta_kms_impl_device_dispatch()
which ends up poll()'ing the DRM fd even though drmModeSetCrtc() will not
produce a DRM event, hence the poll() never returns. The freeze was observed
when hotplugging a DisplayLink dock for the first time on Ubuntu 19.10.
CC stable: gnome-3-34
https://gitlab.gnome.org/GNOME/mutter/merge_requests/953
This commit introduces, and makes use of, a transactional API used for
setting up KMS state, later to be applied, potentially atomically. From
an API point of view, so is always the case, but in the current
implementation, it still uses legacy drmMode* API to apply the state
non-atomically.
The API consists of various buliding blocks:
* MetaKmsUpdate - a set of configuration changes, the higher level
handle for handing over configuration to the impl backend. It's used to
set mode, assign framebuffers to planes, queue page flips and set
connector properties.
* MetaKmsPlaneAssignment - the assignment of a framebuffer to a plane.
Currently used to map a framebuffer to the primary plane of a CRTC. In
the legacy KMS implementation, the plane assignment is used to derive
the framebuffer used for mode setting and page flipping.
This also means various high level changes:
State, excluding configuring the cursor plane and creating/destroying
DRM framebuffer handles, are applied in the end of a clutter frame, in
one go. From an API point of view, this is done atomically, but as
mentioned, only the non-atomic implementation exists so far.
From MetaRendererNative's point of view, a page flip now initially
always succeeds; the handling of EBUSY errors are done asynchronously in
the MetaKmsImpl backend (still by retrying at refresh rate, but
postponing flip callbacks instead of manipulating the frame clock).
Handling of falling back to mode setting instead of page flipping is
notified after the fact by a more precise page flip feedback API.
EGLStream based page flipping relies on the impl backend not being
atomic, as the page flipping is done in the EGLStream backend (e.g.
nvidia driver). It uses a 'custom' page flip queueing method, keeping
the EGLStream logic inside meta-renderer-native.c.
Page flip handling is moved to meta-kms-impl-device.c from
meta-gpu-kms.c. It goes via an extra idle callback before reaching
meta-renderer-native.c to make sure callbacks are invoked outside of the
impl context.
While dummy power save page flipping is kept in meta-renderer-native.c, the
EBUSY handling is moved to meta-kms-impl-simple.c. Instead of freezing the
frame clock, actual page flip callbacks are postponed until all EBUSY retries
have either succeeded or failed due to some other error than EBUSY. This
effectively inhibits new frames to be drawn, meaning we won't stall waiting on
the file descriptor for pending page flips.
https://gitlab.gnome.org/GNOME/mutter/issues/548https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
The intention with KMS abstraction is to hide away accessing the drm
functions behind an API that allows us to have different kind of KMS
implementations, including legacy non-atomic and atomic. The intention
is also that the code interacting with the drm device should be able to
be run in a different thread than the main thread. This means that we
need to make sure that all drm*() API usage must only occur from within
tasks that eventually can be run in the dedicated thread.
The idea here is that MetaKms provides a outward facing API other places
of mutter can use (e.g. MetaGpuKms and friends), while MetaKmsImpl is
an internal implementation that only gets interacted with via "tasks"
posted via the MetaKms object. These tasks will in the future
potentially be run on the dedicated KMS thread. Initially, we don't
create any new threads.
Likewise, MetaKmsDevice is a outward facing representation of a KMS
device, while MetaKmsImplDevice is the corresponding implementation,
which only runs from within the MetaKmsImpl tasks.
This commit only moves opening and closing the device to this new API,
while leaking the fd outside of the impl enclosure, effectively making
the isolation for drm*() calls pointless. This, however, is necessary to
allow gradual porting of drm interaction, and eventually the file
descriptor in MetaGpuKms will be removed. For now, it's harmless, since
everything still run in the main thread.
https://gitlab.gnome.org/GNOME/mutter/issues/548https://gitlab.gnome.org/GNOME/mutter/merge_requests/525
