Both possible paths had bugs:
For OpenGL it passed EGL_CONTEXT_CLIENT_VERSION, which - in older versions
of the standard - was not permitted.
For GLES it always assumed EGL_CONTEXT_FLAGS_KHR to work, which belongs to the
aforementioned extension.
Ironically this was never a problem (probably saved by implementations
not being overly strict) except in 2024 on an emulated Android 14 device
that trips over this edge case. It is a mystery.
When 3cb9119984 introduced AVIF screenshot
support, FILE * in write functions were replaced by filenames. This
resulted in unnecessary duplication of FILE * handling code and the usage
of avio_open API made it hard to use exclusive open with it.
To unify file handling, use avio_open_dyn_buf instead which writes to
memory instead. This way FILE * can be used for the write functions
and file handling code can be deduplicated. Since we now control
the file opening, exclusive open can now be used for AVIF screenshots.
The screenshot command is documented to not overwrite existing files.
However, there is a race window between the filename is generated with
gen_fname and when the file is open to write. Specifically, the
convert_image function in this window can be very time consuming
depending on video and screenshot image format and size. This results
in existing file being overwritten because the file writing functions
don't check for the existance of file.
Fix this be opening the file in exclusive mode. Add overwrite parameter to
write_image for other operations that are documented to overwrite existing
files, like screenshot-to-file. Note that for write_avif, checking
existance is used instead because avio_open does not support exclusive
open mode.
This commit allows image_writer to attach HDR metadata to AVFrames via
the corresponding AVFrameSideData attachments, if present. It does this
by calling pl_avframe_set_color, already used by mp_image_to_avframe.
Signed-off-by: Leo Izen <leo.izen@gmail.com>
Fixes the issue described in https://github.com/mpv-player/mpv/issues/11862
for SDR files for non-d3d11 gpu-api. We currently don't have a smarter
way to get the real on-the-wire bpc for other APIs, so this is the best
that can be done.
Upstream ASS specification says that all subtitles should be rendered
with color primaries and transfer matching their associated video. But
as expected after further discussion the decision has been made to
fallback to SDR mode in case of HDR video.
See-Also: 649a7c2e1f/libass/ass_types.h (L233-L237)
See-Also: libass/libass#297
See-Also: mpv-player#13381
Fixes: mpv-player#13673
Windows 10 top bar height cannot be adjusted individually when
WS_CAPTION is enabled due to buggy DWM NC drawing behavior. The issue is
fixed in Windows 11.
To keep consistent window look remove the border on each side, but only
on Windows 10.
Windows 11 draws border regardless, so we are 1px off on window size,
blending border with one line of video. Fix that by adding the border to
our internal windows size calculation.
Windows 10 on the other hand has a bug where specifying left and top NC
area will trigger title bar drawn in full height always. Keep old
behaviour in this case. Also while there is similar "visible" border
there, it seems to be transparent on Windows 10.
For high contrast themes, don't adjust title bar height and instead
remove WS_CAPTION to have the same border all around the window, as DWM
doesn't make borders invisible in this case.
With runtime geometry change, currently it only results in a
SetWindowPos call to resize the window. However, SetWindowPos doesn't
change the window maximized state, so Windows still thinks that the
window is maximized even though it no longer covers the whole workspace.
This results in visual glitches, and if the window is dragged afterwards
it's "restored" again.
Fix this by correctly setting the window maximized state in this case.
Currently mpv always uses the previous window size when unmaximizing
or exiting fullscreen. This disregards compositor's preferenced size in
the configure event, resulting in wrong window size if changing window
state and size are delivered in the same configure event.
It's better to always respect the preferenced size instead, unless
the state change is due to runtime geometry change, where we want
to use our preference.
In the xdg_toplevel_configure handler, in some cases the geometry is
not equal to the width and height in the event. This can happen
when runtime geometry change is requested (a new size is set),
or in the case of wl->state_change || width == 0 || height == 0
(the old size is used).
However, prepare_resize always uses the width and height in the event
and not the corrected geometry here. This causes
xdg_surface_set_window_geometry using the wrong value resulting in
wrong size of window decoration. Amusingly, the debug message right
above it uses the correct size.
Fix this by using the updated geometry size instead. Since now all
prepare_resize have parameters of 0, the width and height parameters
are no longer needed.
Fixes: 828dd65ef8
8e793bde78 made that changing geometry
while maximized has no effect until the window is unmaximazed. However,
this behavior is inconsistent with setting window-scale on all of win32,
wayland, and x11, which always unmaximizes the window and sets the
window size.
Since setting geometry is conceptually similar to setting window-scale
(both change the window size), they should have the same behavior.
If not fullscreen, unmaximize window on runtime geometry change to
keep the behavior consistent with window-scale.
Currently, setting geometry on runtime only changes the window size
but not the position. This is because reset_size is only set if
the window size is changed, and vo_x11_highlevel_resize doesn't set
the window position without force_window_position enabled. Fix this
by setting the related flags and perform a window move when
geometry is updated.
Fixes 8e793bde78.
As the first aligned format this required a fix to reconfig().
Adding the other component-swapped formats in this group would be trivial
but I checked the DRM database [1] and no driver exists that supports
one of those but not YUYV and this is quite fringe as-is, so I opted not to.
[1] <https://drmdb.emersion.fr/formats>
The VO generic code tries to be helpful and resets this after
each reconfig. However for the simpler VOs the target params
are constant after a reconfig or even for the entire lifetime.
So it's clearly better to let the VO decide.
This also allows the VO to use a static buffer instead.
Segfaults otherwise on uninit because some objects are created while
others are not. Move it to the very top since the purpose of this is to
skip wayland initialization entirely while autoprobing.
Fixes f6f1721101.
Currently, Wayland is above X11 and DRM in probe order. The success
of automatic probing depends on the fact that unsuitable backends
would fail to initialize. For example, X11 backend (which uses Xlib)
fails to initialize if DISPLAY environment variable is not set, so
starting mpv in VT console will pick the DRM backend as expected,
even when an X server is running in another VT.
However, libwayland-client used by the Wayland backend has the
"helpful" behavior of falling back to "wayland-0" if WAYLAND_DISPLAY
is not set. This breaks autoprobing if mpv is started from X server
or VT console while a running Wayland compositor running in another
VT (or even running as an X client) is using "wayland-0" for protocol
socket name: mpv will start playing in the Wayland compsitor instead
of using the X11 or DRM backends.
Similar to DISPLAY for X server, We should consider exporting
WAYLAND_DISPLAY to child processes the responsibility of Wayland
compositors, and any compositor not doing this should be considered
broken. Thus we now require WAYLAND_DISPLAY to be set for the
backend initialization to succeed to make sure that autoprobing
works as intended.
It seems that GLX requires us to explicitly set opts.want alpha before
ra_ctx_create is called. b7fd232524
rearranged the function calls in a way made this not work. Fix this by
rearranging it again so the value is set before ra_ctx is created.
For image formats with 2 or more chroma planes such as YU12, there is
a plane merging pass for these planes calling finish_pass_tex which
does a format conversion. After this conversion, the cscale shader
afterwards works properly.
However, for image formats with only 1 chroma plane (semi-planar formats),
including NV12 and P010, this merging pass is never called, which breaks
the cscale shader afterwards if the video is rotated with subsampled
chroma.
Fix this by adding an explicit conversion pass if this situation is
detected after the pre-scale hooks, so if there are shaders hooking on
CHROMA (like the deband filter) and the conversion is done by pass_hook
already, this conversion won't be called.
replaces the old focus-on-open option with a more generic focus-on
options that can be extended.
adjust the only platform that uses that option.
Fixes#8337
This should only be a problem during initialization. If in a
multi-monitor setup, mpv guesses the wrong scale value and the user
passes --fs, the scaled size will be wrong and you have to unfullscreen
and fullscreen again to fix it. This is because rescale geometry won't
do anything if the value of hidpi-window-scale is false (the default) so
the geometry is never rescaled to the correct value thus the wrong size.
Normally, mpv will just correct itself after subsequent events occur but
because it is considered a locked size (as it should be), we avoid doing
any other resizing events thus it never gets corrected. Fix this by just
always rescaling the geometry in the locked size case. It shouldn't
matter elsewhere because mpv will always have the correct scale value
and the possibility of having the wrong one is only possible on startup.
Fixes ded181f642
Convenience to override name if imgfmt is not set.
Allows to create mp_image_params without setting imgfmt. Will be useful
for the next change where mp_imgfmt is not available. This is workaround
that will be remved once all codebase switches to pl_fmt.
the render size cached in ctx->vo->dwidth/dheight can be outdated in
some circumstances at the time the context needs resizing. instead use
the current render size.
MoltenVK itself tries to work around a supposedly Metal problem that
itself causes flicker, black screens or broken rendering. it sets the
drawableSize to 1x1 to forcefully complete the presentation. though if
1x1 resolution frame is presented it causes a visual flicker or rather
a solid coloured frame. it causes even more problems since sometimes it
does not reset the drawableSize to the proper resolution and keeps
rendering everything in 1x1.
work around this workaround by discarding drawableSize that are <=1 in
any direction.
Fixes#13505
c2129c18f8 saves the button down serial to
pointer_serial of the seat so that it can be used for window dragging
later. However, this overwrites the serial saved at the enter event.
Since the serial in wl_pointer_set_cursor must be the latest
wl_pointer_enter serial number sent to the client, if a button down
serial overwrites that, setting cursor no longer works until the cursor
enters the window next time.
Fix this by using separate serials for these two types of events.
refqueue gets the field of the frame from mp_image which almost always(if not
always) assumes bottom field order first. By default this behavior should not
change but specifying the field order should bypass this.
Previously there was no way to specify the field order of interlaced videos
when deinterlacing with inbuilt filters. Lavfi deinterlacers seemed to prefer
top field order while inbuilt ones (vdpaupp, vavpp, d3d11vpp) seemed to prefer
bottom field order. The default "auto" option should work exactly as before
while specifying either "tff" or "bff" should force the specified field order
It's been a longstanding issue in wayland* that the first frame on a
hidpi screen will have wrong scaling. A well behaved client immediately
corrects this, but it's noticeable and also can affect window placement
due to the way resizng works. Preferred scale from the fractional
protocol and preferred buffer scale can actually solve this problem. It
depends on compositors mostly, but one could simply send the event
before the client maps its surface so it knows what the correct scale is
in the first place. I'm not sure if any compositors currently behave
like this (sway seems to still require the client to render before
sending any scaling information at least), but it makes to sense to
account for this possibility.
*: https://gitlab.freedesktop.org/wayland/wayland/-/issues/133
Explicitly send an UP event after the client finishes a resize for
touch event, don't resize if locked_size is set for touch event,
and use the correct type for resize edges.
VOCTRLs are processed in the GUI thread through the mp_dispatch mechanism.
Window dragging requests are asynchronous on x11 and wayland, so the item
is processed quickly without problem. However, currently win32 uses the
SendMessage call for this, which is synchronous. This causes the playback
to stop while the dragging request is being processed because the
dispatch queue is blocked.
Work around this by setting a flag instead if the window dragging is
requested, and immediately starts dragging after processing the dispatch
queue. This doesn't block the dispatch queue while also avoiding any
extra latency added by the Windows message queue.
5736737750 removed the check because it was
not needed to prevent fullscreen window from being dragged. However, this
causes an undesirable behavior: when using a touch screen to drag a window
on Windows 11, DWM shrinks the window content a bit with an acrylic
backdrop to indicate that the window is being dragged. This also happens
when trying to drag the window in fullscreen. Add the check to prevent
this from happening.
Currently, VO dragging logic is hardcoded into each VO, where a left mouse
button down event unconditionally begins dragging if the VO dragging test
passes. This method is extremely unflexible as the VO has no knowledge of
what is happening in the input system: while begin dragging with the second
click of a doubleclick is undesired, it cannot determine whether a click
is a double click or not because it's determined by the input system.
The better way to do it is to handle it somewhere in the downstream
consumers of the events instead, as they have more information to make
this decision. The input system is the perfect place for this as the logic
for checking doubleclick already exists. So just issue a begin-vo-dragging
command if it detects a left mouse button down which isn't also a
doubleclick in this case, and delete all hardcoded VO dragging logic
in win32, x11, and wayland.
Note that this solution hardcodes left mouse button down for now, but
because the VO dragging is now centralized, it's possible to make more
improvements, such as a deadzone mechanism to fix the conflict with
MBTN_LEFT mouse bind.
This allows begin-vo-dragging command to initialize a vo dragging request
for wayland. The last mouse button press seat and serial is used for the
request if it is not consumed by interactive resizing.
This allows begin-vo-dragging command to initialize a vo dragging request
for win32. Also set dragging to release all keys like for other platforms.
The hard-coded left mouse button down trigger is scheduled to be removed
in a later commit.
This allows begin-vo-dragging command to initialize a vo dragging request
for x11. The last mouse button press event is used for _NET_WM_MOVERESIZE.
The hard-coded left mouse button down trigger is scheduled to be removed
in a later commit.
This command initializes a vo dragging request for VOs that implement
the new VOCTRL_BEGIN_DRAGGING voctrl. This allows scripts to begin vo
dragging for any button press event.
initially we drew on a dedicated render queue to circumvent certain
bottlenecks on the main queue, like concurrent events that could lead
to late drawing.
though due to the way cocoa works we need to draw an the main queue
otherwise we get undefined behaviour that could lead to crashes.
Fixes#10276
The opt validator functions are casted to generic validator, which has
erased type for value. Calling function by pointer of different
definition is an UB.
Avoid that by generating wrapper function that does proper argument type
conversion and calls validator function. Type erased functions have
mangled type in the name.
Fixes UBSAN failures on Clang 17, which enabled fsanitize=function by
default.
Not sure why all of these single line if's were being braced. Probably
some cargo cult from years ago but might as well fix it now since it's
bothering me.
Could have been done in e32554cd57, but I
skipped it there. However, using viewporter is actually a win here.
There's been a longstanding issue in upstream wayland* exactly related
to this. I even forgot about cd7a7a1de8
which was made for this exactly problem and explains the random
spawn_cursor calls. Anyways, just not using buffer scale and instead
scaling the cursor surface via viewporter works just fine and completely
sidesteps this problem. This means we can drop the random looking
spawn_cursor calls and some additional checks.
*: https://gitlab.freedesktop.org/wayland/wayland/-/issues/194
We don't want to use this if we have fractional scaling or version 6 of
the wl_surface interface which has a preferred buffer scale event which
is superior to this. We were checking in the important place (surface
entrance events) but not technically in the output. So just move the
conditional to set_surface_scaling itself. Also through the VO_EVENT_DPI
in there for convenience.
this logic was copied from the cocoa-cb opengl backend, since it was
assumed both layers work similar. apparently they don't and a nil value
on the metal layer means use the display colour profile, while on the
opengl layer it means undefined behaviour.
this also fixes a problem with the target-colorspace-hint option, since
moltenvk needs to set this colour space itself for a proper hint. both
mechanism would compete to set the colour space.
usually the content scale updates automatically, though on init it is
possible the wrong scale is used for the initial rendering leading to a
wrongly sized surface.
properly set the initial content scale of the layer.
cocoa-cb was always pre-allocated in the Application itself because
libmpv needs to be set up before usage, an opengl context has to be set
and because it was decided mac specific code should be kept out of
libmpv.
this means that a completely working libmpv and opengl renderer was set
up even if it wasn't used. leading to unnecessary log message, resources
being used or reserved on the system that might not be used, triggering
of dedicated GPU unnecessarily and many other things.
even if not optimal, this wasn't the biggest problem since we only had
that one working vo on macOS. though now that we have a vulkan
gpu(-next) backend on macOS that was made the default, we always have
that dangling cocoa-cb instance, which is completely unnecessary.
move the cocoa-cb initialisation into libmpv preinit function and only
init cocoa-cb when we are a standalone App and cocoa-cb support is build
into.
The --alpha option currently covers two related but different concepts:
whether or not to ignore the alpha component and possibly blending it
with a background. Because of the way the option currently works, it is
impossible to have a transparent window (which requires setting
--alpha=yes) while blending it with the background at the same time. To
solve this, let's rework it so it it superseded by the background
option.
--background controls what kind of background to set for the image if
possible. It can be blended with the set background color, with tiles,
or not blended at all (the last one is still broken on X11/mesa except
for GLX, *sigh*). In this new paradigm, --alpha=no has no real purpose
because you can simply set the background to color and whatever color
you want for exactly the same effect. So the option is removed. Instead,
the hint set by windowing backends (i.e. setting
ra_ctx->opts.want_alpha) can by done with the --background option.
As an aside, the colors in vo_gpu are currently bugged due to not
pre-multiplying the alpha and it seems no one ever noticed. The next
commit fixes that. vo_gpu_next support happens latter since it requires
new things from libplacebo.
Fixes#9615.
Currently this was being duplicated on init and on runtime updates
between both VOs. Since the conditions for setting this will get more
complicated in the future commits, it's better to just handle it in one
place. We could combine the vo_gpu and vo_gpu_next handling into one
single function but vo_gpu_next has an additional condition for
determining alpha that vo_gpu_does not, so we'll leave these separate
for simplicity. Also this technically fixes a memory leak where gl_opts
weren't being freed.
As it turns out OES_EGL_image is only defined for OpenGL ES.
OpenGL drivers implement this extension anyway because it used to be
the only way of importing EGLImages into GL.
An equivalent extension for OpenGL was defined with EXT_EGL_image_storage.
The only difference is the interaction with immutability, which requires
textures to be recreated since they can be bound only once.
Note: this commit can in theory cause certain systems to lose vaapi / drmprime
support. Since EXT_EGL_image_storage is 7 years old this hopefully doesn't happen.
If it does, the init checks can be relaxed to still permit OES_EGL_image.
We generally try to avoid that due to the #ifdef mess.
The equivalent format is defined in ffmpeg 4.4 while our interop code
requires a higher version, but that doesn't cause any problems.
It's actually been like this for years, but wlroots doesn't keep track
of resizes a client does independent of the compositor. When using sway,
this leads to weird behavior with floating clients resizing themselves
back to the old size if you unfocus it. mpv has been working around this
for a long time, but it's really annoying to selectively ignore events
based on a weird heuristic. Since Sway finally fixed this bug, let's go
ahead and drop this crap. Note that other wlroots compositors may
possibly experience a regression if they didn't correct for this like
sway does, but it's for their own good.
This commit adds a DND_INSERT_NEXT action option for drag-and-drop,
allows for selecting it through the --drag-and-drop=insert-next option,
and adds the necessary plumbing to make that happen when something is
dragged onto the player.
Wayland was the only backend that attempted this, but it can be done in
a centralized place for anything that supports this. hidpi-window-scale
is just the same as a normal window scale but with the OS DPI as the
factor.
Several related things but in a nutshell makes it more like wayland.
1. Remove unneeded --hidpi-window-scale checks. The backend should
always report the actual dpi value regardless of what this option
says.
2. Remove dpi_scale factors from UNFS_WINDOW_SIZE VOCTRLs. It makes
things more complicated and unintuitive for no reason. A window scale
of 1 should mean 1. It annoyed a few years ago in #9437, and I agree
with them (wayland was never implemented like this).
3. Change the dpi log messages to be more brief and remove the unneeded
comments about prescaling.
Previous fix only worked when the video output doesn't have vertical
black bars. This fixes the cases like fullscreen, video-zoom etc.
Fixes: https://github.com/mpv-player/mpv/pull/13528
The purpose of osdep/io is to provide a wrapper for POSIX I/O functions
on Windows. The last 2 functions are utility functions which don't serve
this purpose.
Move them to a separate file.
When doing the dithering pass, the fragment coordinate is queried, but
doesn't take the fbo texture flipped property into account. This results
in different dithering patterns when toggling between fbo flipped and
non-flipped state. This can be reproduced with --vo=gpu --gpu-api=opengl
and easily seen with --dither-depth=1 when toggling between playing and
pausing.
Fix this by flipping the fragment y coordinate if the fbo is flipped
when calculating dithering coordinate.
When redrawing cached frames while the fbo texture is flipped,
the texture after drawing to screen pass will be flipped when blitting.
However, when rendering the checkerboard, the fragment coordinate
doesn't take this into account, so the coordinate before flipping is used,
resulting in different checkerboard location when toggling between
flipped and non-flipped state. This can be reproduced with --vo=gpu
and --gpu-api=opengl: the checkerboard patterns when playing and
pausing are different (vertically flipped).
Fix this by flipping the fragment y coordinate if the fbo is flipped
when calculating checkerboard coordinate.
Begin the _NET_WM_MOVERESIZE window dragging in ButtonPress event
to match the behavior of win32 and wayland, simplify logic by dropping
the win_drag_button1_down hack required by the old method, and fix a race
condition described in commit 19f101db68,
which happens when moving the mouse and releasing the button at the same time.
The race condition can be easily triggered when using a touch screen
(tested with libinput driver), where a tap is translated to MotionNotify,
ButtonPress, MotionNotify, and ButtonRelease in sequence, with the last 2
events having the same timestamp. This has caused some window managers to
not stop dragging after the ButtonRelease, resulting in window being stuck
in dragging state after a single tap.
this partially reverts commit 7b5a258. back then the only properly
working vo on macOS was cocoa-cb (libmpv). it would always use the
deprecated opengl cocoa backend or no vo at all. because of that libmpv
was moved to the top of the auto-probing order, so the preferred vo
was used on macOS only.
we now have a working vulkan gpu/gpu-next backend on macOS which should
be the new default vo. though disabling the auto-probing again for
libmpv would probably cause the undesired behaviour on macOS that
cocoa-cb would never be auto selected again. especially if not build
with vulkan support or without vulkan driver on macOS, this would lead
to no video output at all. so instead of completely reverting the
mentioned commit, we instead move libmpv to the bottom of the
auto-probing order but only auto select it when mpv was built with
cocoa-cb support. this restores the previous behaviour on all other
platforms besides macOS, but also lets us auto select cocoa-cb if
supported.
this fixes a crash on quit, when a CATransaction from a system owned
thread/event is happening at the same time. locking the context
synchronises these access and prevents the race condition.
the draw operation induced by any display call from the CAOpenGLLayer
doesn't need that lock, since the display function already does lock
that current context.
Fixes#11681
Since 03cf150ff3, the only purpose of this
VOCTRL was to signal a redraw to the vo. It actualy could have been
removed in 531868fe0d, but this was
missed. The UPDATE_VIDEO flag is better anyway because it allows us to
handle a wide variety of options scattered around that require the VO to
update itself and redraw. We can remove both of the custom callbacks in
vo.c and only leave the VOCTRL_VO_OPTS_CHANGED one. Additionally, that
commit also introduced vo_set_want_redraw, but this is redundant and not
needed. The VOs that use VOCTRL_UPDATE_RENDER_OPTS already set
vo->want_redraw, and those are the only VOs where these options are
relevant in the first place. So we can remove this as well and just let
the big callback in player/command do everything.
It's only necessary to check the existance of cursor_shape_manager.
Also drop the pointer check to handle multi-seat since a removed seat
sets the cursor_seat to NULL.
For VOCTRL_SET_CURSOR_VISIBILITY, set cursors visibility for all seats.
The return is VO_NOTAVAIL is none of the seats have cursor, and
VO_FALSE if setting visibility failed for at least one seat.
The cursor_shape_device is per pointer, which is per seat. Handle it
together with other seat objects. This in turn also handles cursor
visibility per seat.
On a multi-seat setup, wl_registry_global advertises wl_seat multiple times,
one for each seat. if wl->seat is already set, the previous value will be
overwritten, preventing it from being freed.
Additionally, if the latter seat doesn't have the capabilities of the
former seat, the keyboard/pointer/touch input devices will be destroyed
because the current wl_seat_capabilities doesn't distinguish between
different seats. This casues keyboard/pointer/touch not working if the
latter seat has no such capabilities.
Fix this by implementing multi-seat support.
This allows receiving inputs from all devices on all seats.
All seat-specific objects (wl_keyboard/pointer/touch, wl_data_device)
are grouped with their respective seats so they can be managed together.
Note that currently inputs from all seats are interleaved with each other
and modify the same global input states.
If libplacebo manages to create a vulkan swapchain that allows
alpha composition mode to be controlled by the DWM API,
this makes it runtime changeable.