This allows passing EGL_PLATFORM_SURFACELESS_MESA to
gl_renderer_display_create(). It is not useful on its own, because the
surfaceless platform has no window surfaces.
This feature will be used by the headless backend.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
In case the base EGLConfig is needed, gl_renderer_display_create() needs to
know it should use EGL_WINDOW_BIT or EGL_PBUFFER_BIT.
The PBUFFER case is added for when the headless backend will grow GL-renderer
support.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
When all shared/ headers are included in the same way, we can drop unnecessary
include seach paths from the compiler.
This include style was chosen because it is prevalent in the code base. Doing
anything different would have been a bigger patch.
This also means that we need to keep the project root directory in the include
search path, which means that one could accidentally include private headers
with
#include "libweston/dbus.h"
or even
#include <libweston/dbus.h>
IMO such problem is smaller than the churn caused by any of the alternatives,
and we should be able to catch those in review. We might even be able to catch
those with grep in CI if necessary.
The "bad" include style was found with:
$ for h in shared/*.h; do git grep -F $(basename $h); done | grep -vF '"shared/'
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
If we don't have the surfaceless_context extension, we create a pbuffer as a
dummy surface to work with. If we also don't have configless_context, then it
is possible the config used for creating the context does not support pbuffers.
Therefore, if both conditions apply, we need to pick a config that supports
both window and pbuffer surfaces.
This makes the "base" config compatible, but it does not yet guarantee that we
actually pick it again when creating the pbuffer surface. Fixing that is
another patch.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Fold more code into the common config choosing, the pbuffer path this time.
Simplifies code and allows gl_renderer_get_egl_config() to grow smarter in the
future to guarantee config compatility in the absence of configless_context
extension.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Now that all backends pass in a list of acceptable DRM formats, that is used to
determine if the EGLConfig has an alpha channel or not. Therefore the
opaque_attribs and alpha_attribs are now useless, and we can remove the whole
config_attribs argument from the API.
gl_renderer_get_egl_config() uses an internal attrib list that matches at least
the union of the opaque_attribs and alpha_attribs matches.
Overall, behaviour should remain unchanged.
The new attribute array becomes variable in the future, so it is left
non-const.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Implement fuzzy EGLConfig pixel format matching, where we ensure that R, G, B
and A channels have the expected number of bits exactly. This is used on EGL
platforms where the EGLConfig native visual ID is not a DRM format code. On EGL
GBM platform, the old exact matching of native visual ID is kept.
As only the DRM backend uses a DRM format list for picking a config, this patch
should not change any behaviour.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
If configless context is supported, we can skip choosing the "base" config
completely as it will never be used.
This simplifies the code a little bit.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Replace a direct call to egl_choose_config() with a higher level function
gl_renderer_get_egl_config(). This will make follow-up work easier when
attribute lists will be generated inside gl_renderer_get_egl_config() instead
of passed in as is.
We explicitly replace visual_id with drm_formats, because that is what they
really are. Only the DRM backend passes in other than NULL/0, and if other
backends start caring about the actual pixel format, drm_format is the lingua
franca.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
In an attempt to pull more of EGLConfig choosing into one place, refactor code
into the new gl_renderer_get_egl_config(). The purpose of this function is to
find an EGL config that not only satisfies the requested attributes and the
pixel formats if given but also makes sure the config is generally compatible
with the single GL context we have.
All this was already checked in gl_renderer_create_window_surface(), but
gl_renderer_create_pbuffer_surface() is still missing it. This patch is
preparation for fixing the pbuffer path.
We explicitly replace visual_id with drm_formats, because that is what they
really are. Only the DRM backend passes in other than NULL/0, and if other
backends start caring about the actual pixel format, drm_format is the lingua
franca.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Start a new source file for EGL glue stuff, for the EGL platform Weston runs
on. gl-renderer.c is getting too long, and I want to add even more boring code
(config pretty-printing etc.).
This pure code move, no changes.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Some extensions (such as EGL_KHR_partial_update) add functions to EGL.
When the extension is present, GetProcAddress must return usable
function pointers for those entrypoints.
Assert that GetProcAddress returns a non-NULL function pointer in these
cases.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Some drivers support EGL_EXT_image_dma_buf_import_modifiers for format
enumeration, but don't have any modifiers. In this case, on platforms where
malloc(0) returns non-NULL, we would leak that allocation to the caller.
Handle this by noticing when the number of supported modifiers is 0 and
returning early.
No caller ever used anything but NULL here, so just use NULL to simplify code.
In fact, no EGL platform defined today even defines any platform attributes
except the X11 platform for choosing a non-default SCREEN.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Contents on an ouput are captured when screenshooter/recorder/screen
sharing is enabled. In such cases the protected content must
be censored to ensure that it is not recorded along with unprotected
content. This is a required only when the surface protection is in
enforced mode.
Signed-off-by: Harish Krupo <harishkrupo@gmail.com>
Signed-off-by: Ankit Nautiyal <ankit.k.nautiyal@intel.com>
Unfortunately, our y_invert helper also forgot to free the region it
transformed to. Clean up our allocation before we exit.
Signed-off-by: Daniel Stone <daniels@collabora.com>
In 55bcb93fef ("gl-renderer: Use helper for conversion to EGL rects"),
we extracted and lovingly commented the transformation from global to
output co-ordinate space used for EGL_KHR_swap_buffer_with_damage, into
a new helper function.
The commenting correctly noted the steps we need to perform the
transformation: shifting by the output's offset into global space,
followed by applying the output's scale and rotation transformations.
Unfortunately, the code did not live up to the high standards of the
comment, and forgot to translate by the output's offset. This meant that
for multiple outputs, we would probably end up with wildly out-of-bounds
co-ordinates.
Fix the code to first translate by the output's offset in global space,
ensuring that both our swap_buffers_with_damage, and our partial_update
co-ordinate sets, can spark joy for those blessed with more than one
output.
Signed-off-by: Daniel Stone <daniels@collabora.com>
The content protection protocol requires that in enforced mode, parts of the
surfaces which lie on outputs with protection level lower than that of the surface
be censored. This patch uses a solid shader to color such regions with
dark red.
Signed-off-by: Harish Krupo <harishkrupo@gmail.com>
partial_update is an EGL extension which allows us to inform the driver
ahead of time the limits of the areas we'll be writing to. This helps
performance for GPU hardware which renders into a local tile buffer:
informing the driver of the rendering extents means it can avoid
fetching unchanged tiles into the tile buffer and subsequently writing
them out.
The extension complements rather than replaces EGL_EXT_buffer_age (used
before partial_update to know which areas we need to update) and
EGL_KHR_swap_buffers_with_damage (used after partial_update to inform
the winsys of the changed region).
Note however that partial_update deals in buffer-damage regions ('what
has changed since the last time I used _this_ buffer?'), whereas
swap_buffers_with_damage deals in surface-damage regions ('what has
changed since the last time I rendered?'). An explanatory diagram can be
found in the specification:
https://www.khronos.org/registry/EGL/extensions/KHR/EGL_KHR_partial_update.txtFixes: #134
Signed-off-by: Daniel Stone <daniels@collabora.com>
Add some comments in the function to make it clear what's going on,
especially as we twist and turn between a lot of things called 'damage'
meaning different things in different co-ordinate spaces.
Signed-off-by: Daniel Stone <daniels@collabora.com>
The buffer_damage variable stores accumulated damage from previous
frames. This is the area that, before considering our current repaint
request, we need to repaint in order to bring the older buffer up to
date with the last buffer we rendered into.
Rename to previous_damage so it's a bit more clear what this refers to.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Technically it is storing which areas of the border are damaged.
However, we already have damage-region variables which need to be
translated by the border region. Rename the variable to not contain the
word 'damage' to reduce confusion.
Signed-off-by: Daniel Stone <daniels@collabora.com>
eglSwapBuffersWithDamage has to convert a damage region from Weston's
global co-ordinate space, into the co-ordinate space for EGL rendering
into a buffer for that output.
The conversion from the global co-ordinate space in logical pixels to
the output space in buffer pixels is slightly long and error-prone,
involving translating by the output's offset within the global
co-ordinate space, multiplying by output scale, and also translating to
allow for any borders we paint around the output.
After this is done, we need to flip the co-ordinates in the Y axis to
account for the lower-left-origin co-ordinate space used by EGL.
Since we want to reuse this for partial_update, but using a different
source region, extract this conversion into a well-commented helper we
can reuse.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Fan debug mode repaints the whole surface in order to clear any 'trails'
left over from previous fan paints. If this happens, fall back to using
regular eglSwapBuffers rather than eglSwapBuffersWithDamageEXT, since
the damage region we would pass will be too small.
Signed-off-by: Daniel Stone <daniels@collabora.com>
This fixes warnings for weston-debug, input, compositor, log and
linux-explicit-sync. Warnings range from swapping '[in]', '[out]' with
the function arguments to wrong parameter names.
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
GL-renderer is expected to grow more files, both by addition and by splitting.
Moving them into a new subdirectory helps people to understand which files are
relevant.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
This is an internal export for GL-renderer, so that it does not need to build
linux-sync-file.c a second time. This follows the example of
linux-explicit-synchronization.c which is also used by GL-renderer.
Signed-off-by: Pekka Paalanen <pekka.paalanen@collabora.com>
Implement the get_release request of the zwp_surface_synchronization_v1
interface.
This commit implements the zwp_buffer_release_v1 interface. It supports
the zwp_buffer_release_v1.fenced_release event for surfaces rendered by
the GL renderer, and the zwp_buffer_release_v1.immediate_release event
for other cases.
Note that the immediate_release event is safe to be used for surface
buffers used as planes in the DRM backend, since the backend releases
them only after the next page flip that doesn't use the buffers has
finished.
Changes in v7:
- Remove "partial" from commit title and description.
- Fix inverted check when clearing used_in_output_repaint flag.
Changes in v5:
- Use the new, generic explicit sync server error reporting function.
- Introduce and use weston_buffer_release_move.
- Introduce internally and use weston_buffer_release_destroy.
Changes in v4:
- Support the zwp_buffer_release_v1.fenced_release event.
- Support release fences in the GL renderer.
- Assert that pending state buffer_release is always NULL after a
commit.
- Simplify weston_buffer_release_reference.
- Move removal of destroy listener before resource destruction to
avoid concerns about use-after-free in
weston_buffer_release_reference
- Rename weston_buffer_release_reference.busy_count to ref_count.
- Add documentation for weston_buffer_release and ..._reference.
Changes in v3:
- Raise NO_BUFFER for get_release if no buffer has been committed,
don't raise UNSUPPORTED_BUFFER for non-dmabuf buffers,
so get_release works for all valid buffers.
- Destroy the buffer_release object after sending an event.
- Track lifetime of buffer_release objects per commit, independently
of any buffers.
- Use updated protocol interface names.
- Use correct format specifier for resource ids.
Changes in v2:
- Raise UNSUPPORTED_BUFFER at commit if client has requested a
buffer_release, but the committed buffer is not a valid linux_dmabuf.
- Remove tests that are not viable anymore due to our inability to
create dmabuf buffers and fences in a unit-test environment.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>
Implement the set_acquire_fence request of the
zwp_surface_synchronization_v1 interface.
The implementation uses the acquire fence in two ways:
1. If the associated buffer is used as GL render source, an
EGLSyncKHR is created from the fence and used to synchronize
access.
2. If the associated buffer is used as a plane framebuffer,
the acquire fence is treated as an in-fence for the atomic
commit operation. If in-fences are not supported and the buffer
has an acquire fence, we don't consider it for plane placement.
If the used compositor/renderer doesn't support explicit
synchronization, we don't advertise the protocol at all. Currently only
the DRM and X11 backends when using the GL renderer advertise the
protocol for production use.
Issues for discussion
---------------------
a. Currently, a server-side wait of EGLSyncKHR is performed before
using the EGLImage/texture during rendering. Unfortunately, it's not clear
from the specs whether this is generally safe to do, or we need to
sync before glEGLImageTargetTexture2DOES. The exception is
TEXTURE_EXTERNAL_OES where the spec mentions it's enough to sync
and then glBindTexture for any changes to take effect.
Changes in v5:
- Meson support.
- Make explicit sync server error reporting more generic, supporting
all explicit sync related interfaces not just
wp_linux_surface_synchronization.
- Fix typo in warning for missing EGL_KHR_wait_sync extension.
- Support minor version 2 of the explicit sync protocol (i.e., support
fences for opaque EGL buffers).
Changes in v4:
- Introduce and use fd_clear and and fd_move helpers.
- Don't check for a valid buffer when updating surface acquire fence fd
from state.
- Assert that pending state acquire fence fd is always clear
after a commit.
- Clarify that WESTON_CAP_EXPLICIT_SYNC applies to just the
renderer.
- Check for EGL_KHR_wait_sync before using eglWaitSyncKHR.
- Dup the acquire fence before passing to EGL.
Changes in v3:
- Keep acquire_fence_fd in surface instead of buffer.
- Clarify that WESTON_CAP_EXPLICIT_SYNC applies to both backend and
renderer.
- Move comment about non-ownership of in_fence_fd to struct
drm_plane_state definition.
- Assert that we don't try to use planes with in-fences when using the
legacy KMS API.
- Remove unnecessary info from wayland error messages.
- Handle acquire fence for subsurface commits.
- Guard against self-update in fd_update.
- Disconnect the client if acquire fence EGLSyncKHR creation or wait
fails.
- Use updated protocol interface names.
- User correct format specifier for resource ids.
- Advertise protocol for X11 backend with GL renderer.
Changes in v2:
- Remove sync file wait fallbacks.
- Raise UNSUPPORTED_BUFFER error at commit if we have an acquire
fence, but the committed buffer is not a valid linux_dmabuf.
- Don't put buffers with in-fences on planes that don't support
in-fences.
- Don't advertise explicit sync protocol if backend does not
support explicit sync.
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>
Introduce an internal API for dealing with linux sync files,
and use it in the codebase to replace ad-hoc sync file management.
The linux_sync_file_is_valid function is not currently used, but will be
utilized in upcoming commits to implement the
zwp_linux_explicit_synchronization_unstable_v1 protocol.
Changes in v5:
- Meson support.
Changes in v3:
- Use parameter name in function documentation.
- Move kernel UAPI to separate header file.
Changes in v2:
- Add function documentation
- Remove linux_sync_file_wait()
Signed-off-by: Alexandros Frantzis <alexandros.frantzis@collabora.com>
Introduce a helper function to disconnect the client on unhandled
buffer types, and use it in the gl and pixman renderers. The function
is modeled after linux_dmabuf_buffer_send_server_error.
Also print the egl error state in the gl renderer, in case the
unrecognized buffer error happens when querying an egl buffer.
https://gitlab.freedesktop.org/wayland/weston/issues/148
Add new API to gl-renderer interface for providing fence sync fd. the
backend can wait for GPU rendering by this API.
Signed-off-by: Tomohito Esaki <etom@igel.co.jp>
In the case where CreateContext/MakeCurrent fails, we still want to
know what the EGL driver is capable of.
Move the EGL info printing, just after the eglInitialize() call to
ensure that.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Split the two into separate functions. Former requires an initialized
EGL display, while the latter a current context.
We will use that distinction with the next patch.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Not having swap_buffers_with_damage could cause a performance impact on
some backends, but at least on GBM it causes no issues. It also seems to
confuse users into thinking it's a legitimate error which could explain
session slowness.
Similarly with buffer_age, whilst we do lose a little bit of performance
by not being able to do partial renders, it is not a great deal, and the
user is unlikely to be able to do anything about it in any event.
Remove the warning; we print the full extension list at startup, so we
already have enough information from the logs to easily diagnose any
real errors.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Add an is_opaque property that is set to true if the attached buffer does not
have an alpha component, or if the solid color is non-transparent.
Signed-off-by: Philipp Zabel <philipp.zabel@gmail.com>
we have to set num_images after import_simple_dmabuf
call. Otherwise, egl_images will not be correctly
referenced in gl_renderer_attach_dmabuf.
(Found by clang source code analyzer)
Signed-off-by: Emre Ucan <eucan@de.adit-jv.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
EGL_IMG_context_priority allows the client to request that their
rendering be considered high priority. For ourselves, this is important
as we are interactive and any delay in our rendering causes input-output
jitter; a less than smooth user interactive. So if the driver supports
setting the context priority, try and create our EGLContext as high
priority. The driver may reject our request due to system restrictions,
in which case it will fallback to normal priority, but if successful it
will reschedule our rendering and all of its dependencies to execute
earlier, especially important when the GPU is being hogged by background
clients.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
... to get the user_data. Like everywhere else through weston.
Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Since formats is an out parameter, we need to copy to the alloc'ed
memory and not over the pointer address.
Signed-off-by: Philipp Kerling <pkerling@casix.org>
Reviewed-by: Daniel Stone <daniels@collabora.com>
If the EGL_EXT_image_dma_buf_import_modifiers extension is not
supported, let gl_renderer_query_dmabuf_formats return a hardcoded
fallback list. That list contains ARGB8888, XRGB8888, and if the
GL_EXT_texture_rg extension is supported, YUYV, NV12, YUV420, and
YUV444.
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
modifier_hi and modifier_lo are set to 0 by clients,
which are not supporting modifiers. Modifier attributes
of buffers of these clients set to 0 too in linux-dmabuf.c
import_simple_dmabuf function in gl-renderer.c compares
modifier attribute of the buffer with DRM_FORMAT_MOD_INVALID.
DRM_FORMAT_MOD_INVALID is equal to ((1ULL<<56) - 1).
Therefore, modifer 0 is accepted as valid. Then, the function
checks support for eglQueryDmaBufModifiersEXT.
If it is not supported import_simple_dmabuf function is returning
NULL.
This patch sets the modifier attribute to DRM_FORMAT_MOD_INVALID
for clients which are not supporting modifiers. Without this patch
linux-dmabuf protocol is not working for not supporting clients.
Fixes: b138d7afb3 ("gl-renderer: Ignore INVALID modifier")
Signed-off-by: Emre Ucan <eucan@de.adit-jv.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Tested-by: Tomohito Esaki <etom@igel.co.jp>
Pekka Paalanen
Daniel Stone
Adam Jackson
Harish Krupo
Marius Vlad
Alexandros Frantzis
Vasilis Liaskovitis
Tomohito Esaki
Emil Velikov
Philipp Zabel
Matteo Valdina
Emre Ucan
Chris Wilson
Philipp Kerling
Philipp Zabel