The ioctls were disabled a while back since they error out and allegedly
cause problem with X running in another VT.
Omitting the ioctls is not cool, even as a workaround.
If they fail, the user is supposed to fallback appropriately - use a suid
wrapper, logind-like daemon or otherwise.
As of kernel 5.10, they should just work in nearly all cases, so let's
just reinstate the calls.
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
The EGL stuff is really complicated because of historical reasons
(tl;dr: blame EGL). There was one edge case with EGL context creation
that lead to incorrect behavior. EGL_KHR_create_context was created with
EGL 1.4 (which mpv does support) but it is still possible for an EGL 1.4
device to not implement this extension. That means that none of the EGL
attrs that pass a specific opengl version work. So for this obscure
case, there is a fallback context creation at the very end which simply
creates an EGLContext without passing any special attrs.
This has another problem however. mpv has a hard requirement on at least
desktop opengl 2.1 or opengl ES 2.0 to function (we're not asking for
much here). Since the fallback EGL context creation has no version
checking, it is entirely possible to create an EGL context with a
desktop opengl version under 2.1. As you get further along in the code,
the user will encounter the hard opengl version check and then error
out. However, we're supposed to also actually check if GLES works
(that's what the opengl-es=auto option is for) so this is a bug.
The fix is to do a bit of code duplication and make a mpgl_check_version
specifically for if we hit the edge case of needing to create an EGL
context without the EGL_KHR_create_context extension. Grab the version
with the function pointer, check if it's under 210, if so destroy the
EGL context and set it to NULL. After that, if the user has set
opengl-es to auto, mpv will try GLES next. If it is set to no, then mpv
will simply fail as desired. Fixes#5915.
Sidenote: the ra_gl_ctx_test_version originally testing 140 doesn't make
any sense. Passing the version number in that function only does
something if the user has set opengl-restrict. What we need to do is to
pass the version of the created context to that function. If the version
is higher than the opengl-restrict option, then make this a failure.
This was explictly coded to avoid the autoprobe way back when in
2015[1] with no real explanation. In theory, there shouldn't be any
issues allowing this. If a user runs mpv in tty, this is probably what
they actually want in most cases. Perhaps something strange happens on
nvidia, but it should just fail anyway during the autoprobe.
[1]: f757163058
The wl_surface lives for the entire lifetime of the vo. It's only
neccesary to set the scale initially and when the output scaling changes
(the surface moves to a different output with a different scale or the
output itself changes it scale). All of the calls that were being made
in the egl/vulkan resize functions are not needed. vo_wlshm wasn't
correctly rescaling itself before this commit since it had no logic to
handle scale changes. This should all be shared, common code in the
surface/output listeners.
A subtle regression from c26d833. On sway if mpv was set to be a
floating window in the config, set_buffer_scale would actually get
applied twice according to the wayland log. That meant a 1920x1080
window would appear as a 960x540 window if the scale of the wl_output
was set to 2. This only affected egl on sway (didn't occur on weston and
was too lazy to try anything else; probably they were fine). Since
wl->render is initially false, that meant that the very first run
through the render loop returns false. This probably caused something
weird to happen with the set_buffer_scale calls (the egl window gets
created and everything but mpv doesn't write to it just yet) which makes
the set_buffer_scale call happen an extra time. Since it was always
intended for mpv to initally render, this is worth fixing. Just chnage
wl->render to wl->hidden (again) and flip the bools around. That way,
the initial false value results in render == true and mpv tries to draw
on the first pass. This fixes the weird scaling behavior because
reasons.
Mostly a cosmetic change that (hopefully) makes things look better. Some
functions and structs that were previously being exported in the wayland
header were made static to the wayland_common.c file (these shouldn't be
accessed by anyone else).
Take two. f4e89dd went wrong by moving vo_wayland_wait_frame before
start_frame was called. Whether or not this matters depends on the
compositor, but some weird things can happen. Basically, it's a
scheduling issue. vo_wayland_wait_frame queues all events and sends them
to the server to process (with no blocking if presentation time is
available). If mpv changes state while rendering (and this function is
called before every frame is drawn), then that event also gets
dispatched and sent to the compositor. This, in some cases, can cause
some funny behavior because the next frame gets attached to the surface
while the old buffer is getting released. It's safer to call this
function after the swap already happens and well before mpv calls its
next draw. There's no weird scheduling of events, and the compositor log
is more normal.
The second part of this is to fix some stuttering issues. This is mostly
just conjecture, but probably what was happening was this thing called
"composition". The easiest way to see this is to play a video on the
default audio sync mode (probably easiest to see on a typical 23.976
video). Have that in a window and float it over firefox (floating
windows are bloat on a tiling wm anyway). Then in firefox, do some short
bursts of smooth scrolling (likely uses egl). Some stutter in video
rendering could be observed, particularly in panning shots.
Compositors are supposed to prevent tearing so what likely was happening
was that the compositor was simply holding the buffer a wee bit longer
to make sure it happened in sync with the smooth scrolling. Because the
mpv code waits precisely on presentation time, the loop would timeout on
occasion instead of receiving the frame callback. This would then lead
to a skipped frame when rendering and thus causing stuttering.
The fix is simple: just only count consecutive timeouts as not receiving
frame callback. If a compositor holds the mpv buffer slightly longer to
avoid tearing, then we will definitely receive frame callback on the
next round of the render loop. This logic also appears to be sound for
plasma (funfact: Plasma always returns frame callback even when the
window is hidden. Not sure what's up with that, but luckily it doesn't
matter to us.), so get rid of the goofy 1/vblank_time thing and just
keep it a simple > 1 check.
This is actually a very nice simplification that should have been
thought of years ago (sue me). In a nutshell, the story with the
wayland code is that the frame callback and swap buffer behavior doesn't
fit very well with mpv's rendering loop. It's been refactored/changed
quite a few times over the years and works well enough but things could
be better. The current iteration works with an external swapchain to
check if we have frame callback before deciding whether or not to
render. This logic was implemented in both egl and vulkan.
This does have its warts however. There's some hidden state detection
logic which works but is kind of ugly. Since wayland doesn't allow
clients to know if they are actually visible (questionable but
whatever), you can just reasonably assume that if a bunch of callbacks
are missed in a row, you're probably not visible. That's fine, but it is
indeed less than ideal since the threshold is basically entirely
arbitrary and mpv does do a few wasteful renders before it decides that
the window is actually hidden.
The biggest urk in the vo_wayland_wait_frame is the use of
wl_display_roundtrip. Wayland developers would probably be offended by
the way mpv abuses that function, but essentially it was a way to have
semi-blocking behavior needed for display-resample to work. Since the
swap interval must be 0 on wayland (otherwise it will block the entire
player's rendering loop), we need some other way to wait on vsync. The
idea here was to dispatch and poll a bunch of wayland events, wait (with
a timeout) until we get frame callback, and then wait for the compositor
to process it. That pretty much perfectly waits on vsync and lets us
keep all the good timings and all that jazz that we want for mpv. The
problem is that wl_display_roundtrip is conceptually a bad function. It
can internally call wl_display_dispatch which in certain instances,
empty event queue, will block forever. Now strictly speaking, this
probably will never, ever happen (once I was able to to trigger it by
hardcoding an error into a compositor), but ideally
vo_wayland_wait_frame should never infinitely block and stall the
player. Unfortunately, removing that function always lead to problems
with timings and unsteady vsync intervals so it survived many refactors.
Until now, of course. In wayland, the ideal is to never do wasteful
rendering (i.e. don't render if the window isn't visible). Instead of
wrestling around with hidden states and possible missed vblanks, let's
rearrange the wayland rendering logic so we only ever draw a frame when
the frame callback is returned to use (within a reasonable timeout to
avoid blocking forever).
This slight rearrangement of the wait allows for several simplifications
to be made. Namely, wl_display_roundtrip stops being needed. Instead, we
can rely entirely on totally nonblocking calls (dispatch_pending, flush,
and so on). We still need to poll the fd here to actually get the frame
callback event from the compositor, but there's no longer any reason to
do extra waiting. As soon as we get the callback, we immediately draw.
This works quite well and has stable vsync (display-resample and audio).
Additionally, all of the logic about hidden states is no longer needed.
If vo_wayland_wait_frame times out, it's okay to assume immediately that
the window is not visible and skip rendering.
Unfortunately, there's one limitation on this new approach. It will only
work correctly if the compositor implements presentation time. That
means a reduced version of the old way still has to be carried around in
vo_wayland_wait_frame. So if the compositor has no presentation time,
then we are forced to use wl_display_roundtrip and juggle some funny
assumptions about whether or not the window is hidden or not. Plasma is
the only real notable compositor without presentation time at this stage
so perhaps this "legacy" mechanism could be removed in the future.
For some reason, this never existed before. Add VOCTRL_GET_DISPLAY_RES
and use it to obtain the current display's resolution from each
vo/windowing backend if applicable. Users can then access the current
display resolution as display-width and display-height as per the client
api. Note that macOS/cocoa was not attempted in this commit since the
author has no clue how to write swift.
Originally when presentation time was implemented, the frame callback
and presentation feedback functions were called in each rendering api's
separate backend (egl and vulkan). This meant that their respective
structs were basically copy and pasted across both files. Plus later
vo_wlshm started using frame callbacks too. Things got refactored a few
times and it turns out there's actually no need to have these things
separate anymore. The frame callback can just be initialized in
vo_wayland_init and then everything else will follow from there. Just
move all of this code to wayland_common and get rid of the duplication.
Sidenote: This means that vo_wlshm can actually receive presentation
feedback now. It's really simple to do so might as well. See the next
commit.
30dcfbc is a workaround for incorrect border sizes that could occur on
sway/wlroots in certain edge cases. This seemed harmless enough, but it
turns out that on mutter the extra wl_surface_commit somehow causes the
window always go to the top left of the screen after you leave
fullscreen. No idea why this occurs, but the original commit is a
workaround a sway bug and causing regressions for other users isn't
right despite the author being biased towards sway/wlroots.
This reverts commit 30dcfbc9cb.
7fb972f fixed transparency on x11/EGL/Mesa but happened to also break it
for wayland and nvidia. Ideally on wayland, you should just be able to
pick the right EGLConfig that has alpha but this doesn't seem to work
because reasons. So just go back to setting the EGL_ALPHA_SIZE bit if
the user asks for alpha. Apparently this worked before for nvidia as
well. The hack is to just run an eglQueryString in the x11egl context.
If it picks up Mesa as the EGL_VENDOR, then force ctx->opts.want_alpha
to 0 and let pick_xrgba_config take care of the rest.
Made possible with 00b9c81. 34b8adc let the wayland surface set an
opaque region depending on if alpha was set by the user or not. However,
there was no attempted detection for runtime changes and it is possible
(at least in wayland vulkan) to toggle the alpha on and off. So this
meant, we could be incorrectly signalling an opaque region if the user
happened to change the alpha. Additionally, add a helper function for
this and use it everywhere we want to set the opaque region.
efb0c5c changed the rendering logic of mpv on wayland and made it skip
rendering when it did not receive frame callback in time. The idea was
to skip rendering when the surface was hidden and be less wasteful. This
unfortunately had issues in certain instances where a frame callback
could be missed (but the window was still in view) due to imprecise
rendering (like the default audio video-sync mode). This would lead to
the video appearing to stutter since mpv would skip rendering in those
cases.
To account for this case, simply re-add an old heuristic for detecting
if a window is hidden or not since the goal is to simply not render when
a window is hidden. If the wait on the frame callback times out enough
times in a row, then we consider the window hidden and thus begin to
skip rendering then. The actual threshold to consider a surface as
hidden is completely arbitrary (greater than your monitor's refresh
rate), but it's safe enough since realistically you're not going to miss
60+ frame callbacks in a row unless the surface actually is hidden.
Fixes#8169.
Apparently a part of the wayland spec. A compositor may use a surface
that has set part of itself as opaque for various optimizations. For
mpv, we simply set the entire surface as opaque as long as the user has
not set alpha=yes (note: alpha is technically broken in the wayland EGL
backend at the time of this commit but oh well). wlshm is always opaque.
Fixes#8125.
Back in the olden days, mpv's wayland backend was driven by the frame
callback. This had several issues and was removed in favor of the
current approach which allowed some advanced features (like
display-resample and presentation time) to actually work properly.
However as a consequence, it meant that mpv always rendered, even if the
surface was hidden. Wayland people consider this "wasteful" (and well
they aren't wrong). This commit aims to avoid wasteful rendering by
doing some additional checks in the swapchain. There's three main parts
to this.
1. Wayland EGL now uses an external swapchain (like the drm context).
Before we start a new frame, we check to see if we are waiting on a
callback from the compositor. If there is no wait, then go ahead and
proceed to render the frame, swap buffers, and then initiate
vo_wayland_wait_frame to poll (with a timeout) for the next potential
callback. If we are still waiting on callback from the compositor when
starting a new frame, then we simple skip rendering it entirely until
the surface comes back into view.
2. Wayland on vulkan has essentially the same approach although the
details are a little different. The ra_vk_ctx does not have support for
an external swapchain and although such a mechanism could theoretically
be added, it doesn't make much sense with libplacebo. Instead,
start_frame was added as a param and used to check for callback.
3. For wlshm, it's simply a matter of adding frame callback to it,
leveraging vo_wayland_wait_frame, and using the frame callback value to
whether or not to draw the image.
Transparent windows on X11/EGL/native Mesa GL didn't work for various
reasons. From what I remember, the current code did work with nvidia at
least. Mesa has made attempts to fix this, but they never really made it
in.
But it turns out you can make EGL/Mesa list the EGLConfigs that use X11
RGBA visuals, and context_x11egl.c contains code that explicitly selects
them if alpha is requested (see pick_xrgba_config()).
The reason EGL/Mesa did not list them (and thus breaking transparency)
is because we requested a EGL_ALPHA_SIZE != 0 if alpha is requested. But
the transparent EGLConfigs use EGL_ALPHA_SIZE == 0. That's because EGL
doesn't actually support the concept of transparent windows; the alpha
size parameter is something else (memory rendering without FBOs or
something, I don't care enough to look up the real reasons).
This still won't work on Wayland. Every EGL backend needs platform
specific code. (Good job, EGL, such an awesome platform independent
standard.)
Fixes: #6590
It was possible for sway to get incorrectly sized borders if you resized
the mpv window in a creative manner (e.g. open a video in a non-floating
mode, set window scale to 2, then float it and witness wrong border
sizes). This is possibly a sway bug (Plasma doesn't have these border
issues at least), but there's a reasonable workaround for this.
The reason for the incorrect border size is because it is possible for
mpv to ignore the width/height from the toplevel listener and set its
own size. This new size can differ from what sway/wlroots believes the
size is which is what causes the sever side decorations to be drawn on
incorrect dimensions.
A simple trick is to just explicitly commit the surface after a resize
is performed. This is only done if mpv is not fullscreened or maximized
since we always obey the compositor widths/heights in those cases.
Sending the commit signals the compositor of the new change in the
surface and thus sway/wlroots updates its internal coordinates
appropriately and borders are no longer broken.
When using presentation time, we have to be sure to update the ust when
no presentation events are received to make sure playback is still
smooth and in sync. Part of the recent presentation time refactor was to
use the presentation discarded event to signal that the window is
hidden. Evidently, this doesn't work the same everywhere for whatever
reason (drivers?? hardware??) and at least one user experienced issues
with playback getting out of sync since (presumably) the discarded event
didn't occur when hiding the window. Instead, let's just go back to the
old way of checking if the last_ust is equal to the ust value of the
last member in the wayland sync queue. Fixes#8010.
The motivation for this change was a segfault caused by e107342 which
has complicated reasons for occuring (i.e. I'm not 100% sure but I think
it is a really weird race). The major part of this commit is moving the
initialization of presentation listener to the frame_callback function.
Calling it in swap_buffers worked fine but in practice it meant a lot of
meaningless function calls if a window was hidden (the presentation
would just be immediately discarded). By calling it in frame_callback,
we ensure the listener is only created when it is possible to receive a
presentation event.
Of course calling the presentation listener in feedback_presented or
feedback_discarded was considered, but ultimately these events are too
slow. Receiving the ust/msc/sbc triplet here and then passing it to mpv
results in higher vsync judder since there is (likely) not enough time
before the next pageflip. By design, the frame callback is meant to give
us as much time as possible before the next repaint so calling it here
is probably optimal.
Additionally, we can make better use of the feedback_discarded event.
The wp_presentation_feedback should not be destroyed here. It will be
taken care of either when we get feedback again or when the player
quits. Instead what we can do is set a bool that tells wayland_sync_swap
to update itself based on mp_time delta. In practice, the result is not
any different than before, but it should be more understandable what is
going on now.
Of course, the segfault mentioned at the beginning is fixed with this as
well.
Was broken with a zimg wrapper refucktor before the previous commit. In
addition, it seems this didn't match the vo_drm format, or the format
naming convention. So the order actually changes, and the format is
redefined. (The img_format.h comment was probably wrong.)
Change vo_gpu to the new format as well, so we can still test it.
One not-so-nice hack in the wayland code is the assumption of when a
window is hidden (out of view from the compositor) and an arbitrary
delay for enabling/disabling the usage of presentation time. Since you
do not receive any presentation feedback when a window is hidden on
wayland (a feature or misfeature depending on who you ask), the ust is
updated based on the refresh_nsec statistic gathered from the previous
feedback event.
The flaw with this is that refresh_nsec basically just reports back the
display's refresh rate (1 / refresh_rate * 10^9). It doesn't tell you
how long the vsync interval really was. So as a video is left playing
out of view, the wl->last_queue_display_time becomes increasingly
inaccurate. This led to a vsync spike when bringing the mpv window back
into sight after it was hidden for a period of time. The hack for
working around this is to just wait a while before enabling presentation
time again. The discrepancy between the "bogus"
wl->last_queue_display_time and the actual value you get from the
feedback only happens initially after a switch. If you just discard
those values, you avoid the dramatic vsync spike.
It turns out that there's a smarter way to do this. Just use mp_time_us
deltas. The whole reason for these hacks is because
wl->last_queue_display_time wasn't close enough to how long it would
take for a frame to actually display if it wasn't hidden. Instead, mpv's
internal timer can be used, and the difference between wayland_sync_swap
calls is a close enough proxy for the vsync interval (certainly better
than using the monitor's refresh rate). This avoids the entire conundrum
of massive vsync spikes when bringing the player back into view, and it
means we can get rid of extra crap like wl->hidden.
In theory this mostly happens automatically, especially after the 5
vsync limit disables this already. But if we uninit before 5 vsyncs are
rendered, this can get left in a dangling 'enabled' state, which leaks a
debug report callback.
Always explicitly disable it just to be on the safe side.
Change all OPT_* macros such that they don't define the entire m_option
initializer, and instead expand only to a part of it, which sets certain
fields. This requires changing almost every option declaration, because
they all use these macros. A declaration now always starts with
{"name", ...
followed by designated initializers only (possibly wrapped in macros).
The OPT_* macros now initialize the .offset and .type fields only,
sometimes also .priv and others.
I think this change makes the option macros less tricky. The old code
had to stuff everything into macro arguments (and attempted to allow
setting arbitrary fields by letting the user pass designated
initializers in the vararg parts). Some of this was made messy due to
C99 and C11 not allowing 0-sized varargs with ',' removal. It's also
possible that this change is pointless, other than cosmetic preferences.
Not too happy about some things. For example, the OPT_CHOICE()
indentation I applied looks a bit ugly.
Much of this change was done with regex search&replace, but some places
required manual editing. In particular, code in "obscure" areas (which I
didn't include in compilation) might be broken now.
In wayland_common.c the author of some option declarations confused the
flags parameter with the default value (though the default value was
also properly set below). I fixed this with this change.
This was mostly unused, and has certain problems. Just get rid of it.
It was still used in CDDA (--cdda-span) and a debug option for OpenGL
(--opengl-check-pattern). Replace both of these with 2 options, where
each sets the start/end values of the former span. Both were
undocumented somehow (normally we require all options to be documented),
so I'm not caring about compatibility, and not bothering to add it to
the API changelog.
This commit fixes a bug where handle for a framebuffer gets double
freed.
It seems to happen that the same prime fd gets two framebuffers.
As the prime fd is the same the resulting prime handle is also the
same.
This means one handle but 2 framebuffers and can lead to the following
chain:
1. The first framebuffer gets deleted the handle gets also freed via
the ioctl.
2. In startup phase not all 4 dumb buffers for overlay drawing
are set up. It can happen that the last dumb buffer gets the
handle we freed above.
3. The second framebuffer gets freed and the handle will be
freed again resulting that the 4's dumb buffer handle is not
backed by a buffer.
4. Drm prime continues to assign handles to its prime fds an
will lead to have this handle which was just freed to
reassign again but to an prime buffer.
5.Now the overlay should be drawn into dumb buffer 4 which
still has the same handle but is backed by the wrong buffer.
This leads to two different behaviors:
- MPV crashes as the drm prime buffers size als calculated
by the decoder output format. The overlay output format
differs and it takes more space. SO the size check
in kernel fails.
- MPV is continuing play. This happens when the decoders
allocates a bigger buffer than needed for the overlay.
For example overlay is Full HD and decoder output is 4k.
This leads to the behavior das the overlay wil be drawn
into the wrong buffer as its a drm prime buffer and results
in a flicker every fourth step.
This originally existed as a hack for weston. In certain scenarios, a
frame taking too long to render would cause vo_wayland_wait_frame to
timeout which would result in a ton of dropped frames. The naive
solution was to just to add a slight delay to the time value. If a
frame took too long, it would likely to fall under the timeout value and
all was well. This was exposed to the user since the default delay
(1000) was completely arbitrary.
However with presentation time, this doesn't appear to be neccesary.
Fresh frames that take longer than the display's refresh rate (16.666 ms
in most cases) behave well in Weston. In the other two main compositors
without presentation time (GNOME and Plasma), they also do not
experience any ill effects. It's better not to overcomplicate things, so
this "feature" can be removed now.
this deprecates the old cocoa backend only option and moves it to the
general macos ones. add support for the new option in the cocoa-cb
layer creation and use the new option in the olde cocoa backend.
Fixes#7272
As we are less and less interested in vpdpau, with nvdec and vaapi
being better choices in general on nvidia and AMD respectively, we
might consider removing direct_mode, where we bypass the vdpau
mixer and work directly with yuv textures. Normally, working with
yuv textures would be great, but vdpau built in an assumption that
all frames are delivered as separate fields, causing us to have
to re-interleave them.
nvidia then introduces a new OpenGL extension that can return the
yuv frames as frames, but we can't just unconditionally switch to
that as we'd want to keep supporting older hardware where the drivers
are no longer getting new features. The end result is that we
wouldn't be able to get rid of the old code paths.
Removing direct_mode means we always use the mixer, and work with
rgba frame textures. There are some theoretical limitations to
this, but in practice they probably don't matter much - unsupported
colourspaces don't matter because without 10bit decoding support,
we can't use them anyway, and apparently we're not doing separate
chroma scaling these days, so scaling the rbga doesn't really lose
anything (and the vdpau hq scaling option remains available).
This tries to deal with the crazy EGL situation. The summary is:
- using eglGetDisplay() with multiple windowing platforms doesn't really
work, but Mesa had an awful hack for it
- this hack can be disabled at build time, and some distros sometimes
accidentally or intentionally do so
- Mesa will probably eventually disable it by default
- we switched to eglGetPlatformDisplay(), but this requires EGL 1.5
- the very regrettable graphics company (also known as Nvidia) ships
drivers (for old hardware I think) that are EGL 1.4 only
- that means even though we "require" EGL 1.5 and link against it, the
runtime EGL may be 1.4
- trying to run mpv there crashes in the dynamic linker
- so we have to go through some more awful compatibility hacks
This commit tries to do it "properly", but using EGL 1.4 as base. The
plaform selection mechanism is a messy extension there, which got
elevated to core API in 1.5 (but OF COURSE in incompatible ways).
I'm not sure whether the EGL 1.5 code path (by parsing the EGL_VERSION)
is really needed, but if you ask me, it feels slightly saner not to rely
on an EGL 1.4 kludge forever. But maybe this is just an instance of
self-harm, since they will most likely never drop or not provide this
API.
Also, unlike before, we actually check the extension string for the
individual platform extensions, because who knows, some EGL
implementations might curse us if we pass unknown platform parameters.
(But actually, the more I think about this, the more bullshit it is.)
X11 and Wayland were the only ones trying to call eglGetPlatformDisplay,
so they're the only ones which are adjusted in this commit.
Unfortunately, correct function of this commit is unconfirmed. It's
possible that it crashes with the old drivers mentioned above.
Why didn't they solve it like this:
struct native_display {
int platform_type;
void *native_display;
};
Could have kept eglGetDisplay() without all the obnoxious extension BS.
...probably.
The EGL backend had a strange problem: when recreating the window, EGL
surface creation sometimes mysteriously failed. For example, keeping the
"_" key down (cycles video by default) destroys and recreates the window
in rapid succession, which will often enough show the "Could not create
EGL surface!" message.
This was puzzling because due to mpv's architecture, the X11 Window and
even the X11 Display were fully destroyed, the thread on which they ran
was destroyed, and then everything was recreated. There shouldn't have
been any state that could make subsequent EGL initialization fail.
It turns out mpv forgot to free EGLSurfaces in the x11 code. EGL is a
pretty crazy API (full of thread local and global state with weird
lifetime requirements), and for example it seems EGLDisplay cannot be
explicitly released, but apparently implicitly dies when the native
display is closed (at least EGL 1.5 claims eglTerminate() does _not_
invalidate the display, only certain objects linked to it). It appears
that Mesa still referenced at least EGLSurface in some form, and either
some pointer or some X11 ID was dangling, and when it randomly matched
when eglCreateWindowSurface() was called, it failed.
Fix this by calling eglTerminate(), which supposedly destroys (or rather
unreferences) contexts and surfaces created from the display (but
absurdly not the display itself).
Now why can't you just destroy the display? If it's implicitly
invalidated, why can't it just call eglTerminate() implicitly when this
happens? Did Mesa do something wrong when they somehow didn't
automatically remove the dangling object (so I could claim not to be
responsible for the bug)? Who the fuck knows, and I'm too tired to
figure this out (both because it's late, and because I'm tired of this
EGL crap API).
Still not sure if the code is correct now. I think EGL was designed to
maximize implementation and API-use complications. How else could you
possibly come up with something like the EGLDisplay life cycle? Or am I
just making a fuss? Anyway, fuck EGL, fuck computers, fuck technology.
Fixes: #7129
Get rid of the legacy VOCTRL (which will be removed later). I'm not sure
what exactly fullscreen was supposed to do (toggling between using the
entire display, and what --geometry forced?), but I don't care, just get
rid of the VOCTRL. PRs to fix regressions caused by this will be
accepted, but personally I don't care since this is excessively fringe
and obscure.
Seems like some drivers only increment msc every other page flip when
running in interlaced mode (I'm looking at you nouveau). I.e. it seems
to be incremented at the frame rate, rather than the field rate.
Obviously we can't work with this, so shame the driver and bail.
On intel this isn't an issue, as msc is incremented at field rate
there.
This means presentation feedback won't work correctly in interlaced
modes with those drivers, but who in their right mind uses an
interlaced mode these days, anyway?
glx.h recursively includes gl.h, and there is no way to prevent this.
Old Mesa defines some GL symbols, but not all which mpv needs. In
particular, one user who was too lazy to update his ancient Ubuntu and
preferred to bother us with obscure bug reports, had Mesa headers which
did not define GL 3.2, so GLsync was not defined.
All in all I still think the idea of providing the GL API definitions
ourselves was a good idea; just GLX should have been isolated better.
But isolating GLX now is too much effort.
Not sure why I'm bothering with this at all.
Fixes: #7201 (unconfirmed)
This function always expects the GL struct pointer to be a talloc
allocation. So far so bad. But the terrible thing is that _lots_ of code
in mpv didn't quite get this (including the code which introduced the
way it is used this way). For example, in context_glx.c you see this:
struct priv {
GL gl;
...
GL is not a talloc allocation, but since it's at the start of a talloc
allocation, it works anyway. So far so bad. But the really terrible
thing is that mpgl_load_functions2() calls talloc_free_children() on the
GL pointer, which means that all of priv's. This would be unintentional
and could create dangling pointers. And this happens at the about 1
dozen of callers. I'm amazed it didn't broke yet anywhere.
Removing this anti-pattern with making GL "implicitly" a talloc
allocation would be too much effort at this point. So just manually free
the only allocation that the function attached to GL.
Surprisingly, we've managed to get this far without context_glx ever
adding the X11 display as a native resource. But with the recent change
to attempt to enable vdpau when using EGL, the hwdec now requires the
display to be added. So let's add it.
eglGetPlatform() is a broken API, since it takes a windowing specific
argument, yet is supposed to work for multiple APIs at the same time. On
Linux, it can take both a X11 "Display" and a "wl_display". Obviously
there is no way to specify what kind of display the argument is (it's
just a void*).
Mesa has _eglNativePlatformDetectNativeDisplay, which does funny stuff
to try to guess the display type, including trying to call mincore() to
determine whether the pointer can be accessed at all. I guess this
recently accidentally broke (as a bug), but on the other hand, maybe
it's time to do this properly.
The fix is using eglGetPlaformDisplay(). This requires EGL 1.5, plus
Mesa needs to support the associated platform extension
(EGL_KHR_platform_x11).
Since I see no reasonable way to do this in a compatible way, just
require that EGL 1.5 is available. The problem is that EGL 1.4 seems to
require you to create a display to query EGL version and extension, and
you have a chicken-and-egg problem. It's very stupid. Maybe you could
jump through some more hoops to get something compatible, but fuck that.
Users on "too old" Mesa will fall back to GLX (which we keep around for
a regrettable company known by the name of Nvidia).
I think Wayland and GBM should do the same. They're sufficiently
bleeding-edge that you can expect them to have EGL 1.5. On the other
hand, the cursed RPI code will have to stay with a eglGetDisplay().
Speculative fix for #7154.
(Rant about EGL follows. Actually I deleted it.)
The use of glXGetCurrentDisplay() restricted this to the GLX backend.
But actually it works under EGL as well. Removing the GLX-specific call
and using the general mpv-internal method to get the X "Display" makes
it work in mpv.
I didn't know this. Nvidia didn't list this as extension in the EGL
context when I still used their GPUs.
Note that this might in theory break use of vdpau in some libmpv clients
using the render API. But only if MPV_RENDER_PARAM_X11_DISPLAY is not
used, and they relied on mpv using glXGetCurrentDisplay(). EGL does not
provide such an API, and hwdec_vaapi.c also uses what hwdec_vdpau.c uses
now. Considering that vaapi is preferable these days, it's not bad at
all if these clients get "broken". They can be easily fixed by passing
the display to mpv correctly.
This integrates it as "special" format, with no alpha component, as the
equivalent IMGFMT_RGB30 isn't meant to contain any.
Nothing can produce this format in the video chain yet, so the next
commits are needed to make this actually work.
There's 2 stupid things here that need to be fixed. First of all,
vulkan wasn't actually using presentation time because somehow the
get_vsync function in context.c disappeared. Secondly, if the mpv window
was hidden it was updating the ust time based on the refresh_usec but
really it should simply just not feed any information to the vsync info
structure. So this adds some logic to assume whether or not a window is
hidden.
This was obviously missing from the recent commit, which probably broke
10 bit decoding. The original commit didn't test this for lack of
working hardware; this commit isn't tested either.
Fixes: a1c7d61393
2 years ago, ANGLE removed the old NV12-specific extension, and added
a new one that supports a number of formats, including P010. Actually
they just renamed it and removed their initial annoying and obvious
design error (bravo, Google).
Since it broke 2 years ago, nobody should give a shit about this code,
and it should just be removed. But for some reason I still dived the
shit-tank (Windows development).
I guess Intel code monkeys can't write drivers (or maybe the issue is
because we're doing zero-copy, which probably maybe is not actually
allowed by D3D11 due to array textures, see --d3d11va-zero-copy), so
the P010 path is completely untested. It doesn't work, I'll delete all
this ANGLE hwdec code.
Fixes: #7054
The old way of using wayland in mpv relied on an external renderloop for
semi-accurate timings. This had multiple issues though. Display sync
would break whenever the window was hidden (since the frame callback
stopped being executed) which was really annoying. Also the entire
external renderloop logic was kind of fragile and didn't play well with
mpv's internal structure (i.e. using presentation time in that old
paradigm breaks stats.lua).
Basically the problem is that swap buffers blocks on wayland which is
crap whenever you hide the mpv window since it looks up the entire
player. So you have to make swap buffers not block, but this has a
different problem. Timings will be terrible if you use the unblocked
swap buffers call.
Based on some discussion in #wayland, the trick here is relatively
simple and works well enough for our purposes. Instead we basically
build a way to block with a timeout in the wayland buffer swap
functions.
A bool is set in the frame callback function that indicates whether or
not mpv is waiting for a frame to be displayed. In the actual buffer
swap function, we enter into a while loop waiting for this flag to be
set. At the same time, the wl_display is polled to block the thread and
wakeup if it receives any events from the compositor. This loop only
breaks if enough time has passed or if the frame callback bool is
received.
In the near future, it is better to set whether or not frame a frame has
been displayed in the presentation feedback. However as a first pass,
doing it in the frame callback is more than good enough.
The "downside" is that we render frames that aren't actually shown on
screen when the player is hidden (it seems like wayland people don't
like that). But who cares. Accurate timings are way more important. It's
probably not too hard to add that behavior back in the player though.
This is the proper fix to the memory leak @wm4 pointed out. It turns out
that when you autoprobe opengl and vo_wayland_init returns false,
vo_wayland_uninit is never actually executed. So you have a leftover
pointer. The vulkan context does this correctly which was why my old,
dumb "fix" broke it.
This allows to use drm hwaccels that require a hwdevice.
Tested with v4l2request hwaccel and cedrus driver on an allwinner device
running mpv with --vo=gpu --gpu-context=drm --hwdec=drm.
Check if eglGetPlatformDisplayEXT is available and try to
use it to obtain the display connection. Fall back to eglGetDisplay
if eglGetPlatformDisplayEXT is not available or failing.
From PR #5992
Useless garbage.
This was once added to test whether vdpau presentation feedback could be
used. Results were always unsatisfactory, and now vdpau is dead.
Extending the client-allocated mpv_opengl_drm_params struct
constituted a break of ABI that could cause UB.
Create a clean break by deprecating "drm_params" and related structs
and enum values, and replacing it with "drm_params_v2".
Also fix some comments and code that wrongly assumed that open could
return any other negative number than -1 for failure.
This commit updates the libmpv version to 1.104
Originally, vo_gpu/vo_opengl considered the case of Nvidia proprietary
drivers, which required vdpau/GLX, and Intel open source drivers, which
require vaapi/EGL. Since window creation and GPU context creation are
inseparable in mpv's internal API, it had to pick the correct API very
early, or hardware decoding wouldn't work. "x11probe" was introduced for
this reason. It created a GLX context (without showing the window yet),
and checked whether vdpau was available. If yes, it used GLX, if not, it
continued probing x11/EGL. (Obviously it couldn't always fail on GLX
without vdpau, which is why it was a separate "probe" backend.)
Years passed, and now the situation is different. Vdpau is dead. Nvidia
drivers and libavcodec now provide CUDA interop, which requires EGL, and
fixes some of the vdpau problems. AMD drivers now provide vaapi, which
generally works better than vdpau. Intel didn't change.
In particular, vaapi provides working HEVC Main10 support. In theory, it
should work on vdpau too, with quality reduction (no 10 bit surfaces),
but I couldn't get it to work.
So always prefer EGL. And suddenly hardware decoding works. This is
actually rather important, because HEVC is unfortunately on the rise,
despite shitty encoders and unoptimized decoders. The latter may mean
that hardware decoding works better than libavcodec.
This should have been done a long, long time ago.
Mesa supports the EGL_CHROMIUM_sync_control extension, and it's
available out of the box with AMD drivers. In practice, this is exactly
the same as GLX_OML_sync_control, but for EGL. The extension
specification is separate from the GLX one though, and buried somewhere
in the Chromium code.
This appears to work, although I don't know if it really works.
In theory, this could be useful for other EGL targets. Support code for
it could have been added to egl_helpers.c to avoid some minor duplicated
glue code if another EGL target were to provide this extension. I didn't
bother with that. ANGLE on Windows can't support it, because the
extension spec. explicitly requires POSIX timers. ANGLE on Linux/OSX is
actively harmful for mpv and hopefully won't ever use it. Wayland uses
EGL, but has its own fancy presentation feedback stuff (and besides, I
don't think basic video player functionality works on Wayland at all).
context_drm_egl maybe? But I think DRM has its own stuff.
So the next commit can make EGL use it. EGL has a quite similar
function, that practically works the same. Although it's relatively
trivial, it's still tricky, and probably shouldn't end up as duplicated
code.
There are no functional changes, except initialization, and how failure
of the glXGetSyncValues call is handled. Also, some comments mention the
EGL extension.
Note that there's no intention for this code to handle anything else
than the very specific OML sync extension (and its EGL equivalent). This
is just too weirdly specific to the weird idiosyncrasies of the
extension, and it makes no sense to extend it to handle anything else.
(Such as Wayland or DXGI presentation feedback.)
New releases of VDPAU support decoding 4:4:4 content, and that comes
back as NV24 when using 'direct mode' in OpenGL Interop. That means we
need to be a little bit smarter about how we set up the OpenGL
textures.
If the compositor sends a configure event before the surface is initially
mapped, resize gets called before the egl_window gets created, resulting
in a crash in wl_egl_window_resize.
This was fixed back in 618361c697, but was reintroduced when the wayland
code was rewritten in 68f9ee7e0b.
While `ra` supports the concept of a texture as a storage
destination, it does not support the concept of a texture format
being usable for a storage texture. This can lead to us attempting
to create a texture from an incompatible format, with undefined
results.
So, let's introduce an explicit format flag for storage and use
it. In `ra_pl` we can simply reflect the `storable` flag. For
GL and D3D, we'll need to write some new code to do the compatibility
checks. I'm not going to do it here because it's not a regression;
we were already implicitly assuming all formats were storable.
Fixes#6657
This was implemented by using OPT_STRING_VALIDATE for drm-mode,
instead of OPT_INT. Using a string here also prepares for future
additions to drm-mode that aim to allow specifying a mode by its
resolution.
It is useful when debugging to be able to force atomic off, or as a
workaround if atomic breaks for some user. Legacy modesetting is less
likely to break by virtue of being a less complex API.
The amount of code now present that's specific to Vulkan or OpenGL
has reached the point where we really want to split it out to
avoid a mess of #ifdefs.
At the same time, I'm moving the code to an api neutral location.
This change updates the vulkan interop code to work with the
libplacebo based ra_vk, but also introduces direct VkImage
sharing to avoid the use of the intermediate buffer.
It is also necessary and desirable to introduce explicit
semaphore bsed synchronisation for operations on the shared
images.
Synchronisation means we can safely reuse the same VkImage for every
mapped frame, by ensuring the frame is copied to the VkImage before
mapping the next frame.
This functionality requires a 417.xx or newer nvidia driver, due to
bugs in the VkImage interop in the earlier 411 and 415 drivers.
It's definitely worth the effort, as the raw throughput is about
twice that of implementation using an intermediate buffer.
This commit rips out the entire mpv vulkan implementation in favor of
exposing lightweight wrappers on top of libplacebo instead, which
provides much of the same except in a more up-to-date and polished form.
This (finally) unifies the code base between mpv and libplacebo, which
is something I've been hoping to do for a long time.
Note: The ra_pl wrappers are abstract enough from the actual libplacebo
device type that we can in theory re-use them for other devices like
d3d11 or even opengl in the future, so I moved them to a separate
directory for the time being. However, the rest of the code is still
vulkan-specific, so I've kept the "vulkan" naming and file paths, rather
than introducing a new `--gpu-api` type. (Which would have been ended up
with significantly more code duplicaiton)
Plus, the code and functionality is similar enough that for most users
this should just be a straight-up drop-in replacement.
Note: This commit excludes some changes; specifically, the updates to
context_win and hwdec_cuda are deferred to separate commits for
authorship reasons.
Manual changes done:
* Merged the interface-changes under the already master'd changes.
* Moved the hwdec-related option changes to video/decode/vd_lavc.c.
This allows context_drm_egl to use as many buffers as libgbm or the
swapchain_depth setting allows (whichever is smaller).
On pause and on still images (cover art etc.) to make sure that output does not
lag behind user input, the swapchain is drained and reverts to working in a dual
buffered (equivalent to swapchain-depth=1) manner.
When possible (swapchain-depth>=2), the wait on the page flip event is now not
done immediately after queueing, but is deferred to the next invocation of
swap_buffers. Which should give us more CPU time between invocations.
Although, since gbm_surface_has_free_buffers() can only tell us a boolean value
and not how many buffers we have left, we are forced to do this contortionist
dance where we first overshoot until gbm_surface_has_free_buffers() reports 0,
followed by immediately waiting so we can free a buffer, to be able to get the
deferred wait on page flip rolling.
With this commit we do not rely on the default vsync fences/latency emulation of
video/out/opengl/context.c, but supply our own, since the places we create and
wait for the fences needs to be somewhat different for best performance.
Minor fixes:
* According to GBM documentation all BO:s gotten with
gbm_surface_lock_front_buffer must be released before gbm_surface_destroy is
called on the surface.
* We let the page flip handler function handle the waiting_for_flip flag.
I misunderstood how this extension works. If I understand it correctly
now, it's worse than I thought. They key thing is that the (ust, msc,
sbc) tripple is not for a single swap event. Instead, (ust, msc) run
independently from sbc. Assuming a CFR display/compositor, this means
you can at best know the vsync phase and frequency, but not the exact
time a sbc changed value.
There is GLX_INTEL_swap_event, which might work as expected, but it has
no EGL equivalent (while GLX_OML_sync_control does, in theory).
Redo the context_glx sync code. Now it's either more correct or less
correct. I wanted to add proper skip detection (if a vsync gets skipped
due to rendering taking too long and other problems), but it turned out
to be too complex, so only some unused fields in vo.h are left of it.
The "generic" skip detection has to do.
The vsync_duration field is also unused by vo.c.
Actually this seems to be an improvement. In cases where the flip call
timing is off, but the real driver-level timing apparently still works,
this will not report vsync skips or higher vsync jitter anymore. I could
observe this with screenshots and fullscreen switching. On the other
hand, maybe it just introduces an A/V offset or so.
Why the fuck can't there be a proper API for retrieving these
statistics? I'm not even asking for much.
Use the extension to compute the (hopefully correct) video delay and
vsync phase.
This is very fuzzy, because the latency will suddenly be applied after
some frames have already been shown. This means there _will_ be "jumps"
in the time accounting, which can lead to strange effects at start of
playback (such as making initial "dropped" etc. frames worse). The only
reasonable way to fix this would be running a few dummy frame swaps at
start of playback until the latency is known. The same happens when
unpausing.
This only affects display-sync mode.
Correct function was not confirmed. It only "looks right". I don't have
the equipment to make scientifically correct measurements.
A potentially bad thing is that we trust the timestamps we're receiving.
Out of bounds timestamps could wreak havoc. On the other hand, this will
probably cause the higher level code to panic and just disable DS.
As a further caveat, this makes a bunch of assumptions about UST
timestamps. If there are delayed frames (i.e. we skipped one or more
vsyncs), the latency logic is mostly reset. There is no attempt to make
the vo.c skipped vsync logic to use this. Also, the latency computation
determines a vsync duration, and there's no effort to reconcile or share
the vo.c logic for determining vsync duration.
This commit bumps the libmpv version to 1.102
drm-osd-plane -> drm-draw-plane
drm-video-plane -> drm-drmprime-video-plane
drm-osd-size -> drm-draw-surface-size
"draw plane", as in the plane that OpenGL draws to, whether it be
video + OSD or just OSD.
"drmprime video plane", as in the plane used for hwdec video imported
via drmprime.
"draw surface size", as in the size of the surface used for the draw plane
The new names are invariant whether or not hwdec_drmprime_drm is being
used or not. The original naming was very confusing, as when doing
regular rendering (swdec or vaapi) the video would be displayed on the
"OSD plane", and the "Video plane" would remain unused.
We are currently unnecessarily including vulkan headers even when
not building with vulkan support. I also guarded the GL header
inclusion even though this doesn't appear to break anything today.
Fixes#6330.
Currently, the error paths in init() are a bit confusing, and we can
end up trying to pop the current context when there is no context,
which leads to distracting error messages.
I also added an explicit path to return early if the GPU backend is
not OpenGL or Vulkan. It's pointless to do any other cuda init
after that point. (Of course, someone could write more interops.)
Fixes#6256
Since 810acf32d6 video_plane can be NULL
under some circumstances. While there is a check in init, init treats
this as an error condition and would call uninit, which in turn calls
disable_video_plane, which would then segfault. Fix this by including
a NULL check inside disable_video_plane, so that it doesn't try to
disable what isnt' there.
Despite their place in the tree, hwdecs can be loaded and used just
fine by the vulkan GPU backend.
In this change we add Vulkan interop support to the cuda/nvdec hwdec.
The overall process is mostly straight forward, so the main observation
here is that I had to implement it using an intermediate Vulkan buffer
because the direct VkImage usage is blocked by a bug in the nvidia
driver. When that gets fixed, I will revist this.
Nevertheless, the intermediate buffer copy is very cheap as it's all
device memory from start to finish. Overall CPU utilisiation is pretty
much the same as with the OpenGL GPU backend.
Note that we cannot use a single intermediate buffer - rather there
is a pool of them. This is done because the cuda memcpys are not
explicitly synchronised with the texture uploads.
In the basic case, this doesn't matter because the hwdec is not
asked to map and copy the next frame until after the previous one
is rendered. In the interpolation case, we need extra future frames
available immediately, so we'll be asked to map/copy those frames
and vulkan will be asked to render them. So far, harmless right? No.
All the vulkan rendering, including the upload steps, are batched
together and end up running very asynchronously from the CUDA copies.
The end result is that all the copies happen one after another, and
only then do the uploads happen, which means all textures are uploaded
the same, final, frame data. Whoops. Unsurprisingly this results in
the jerky motion because every 3/4 frames are identical.
The buffer pool ensures that we do not overwrite a buffer that is
still waiting to be uploaded. The ra_buf_pool implementation
automatically checks if existing buffers are available for use and
only creates a new one if it really has to. It's hard to say for sure
what the maximum number of buffers might be but we believe it won't
be so large as to make this strategy unusable. The highest I've seen
is 12 when using interpolation with tscale=bicubic.
A future optimisation here is to synchronise the CUDA copies with
respect to the vulkan uploads. This can be done with shared semaphores
that would ensure the copy of the second frames only happens after the
upload of the first frame, and so on. This isn't trivial to implement
as I'd have to first adjust the hwdec code to use asynchronous cuda;
without that, there's no way to use the semaphore for synchronisation.
This should result in fewer intermediate buffers being required.
by default the pixel format creation falls back to software renderer
when everything fails. this is mostly needed for VMs. additionally one
can directly request an sw renderer or exclude it entirely.
