Commit graph

225 commits

Author SHA1 Message Date
Geoffrey McRae
d24bc75519 [client] egl: added missing vertex shader file 2021-08-09 23:14:01 +10:00
Geoffrey McRae
92de467edc [client] egl: add ffx_cas post process filter 2021-08-09 23:12:58 +10:00
Quantum
9b1d03fcfe [client] egl: implement #include for shaders with awk 2021-08-09 22:04:07 +10:00
Geoffrey McRae
4eda01949d [client] egl: give pp filters the dimensions of all prior textures 2021-08-09 22:02:07 +10:00
Geoffrey McRae
062d18d5fa [client] egl: don't allocate the texture ringbuffer unless needed 2021-08-09 18:28:52 +10:00
Geoffrey McRae
04a54598b3 [client] egl: set a default scale for textures without filters 2021-08-09 18:27:10 +10:00
Geoffrey McRae
79dcc6d4f1 Revert "[client] egl: set a default scale for textures without filters"
This reverts commit 57a74c156b.
Pushed a ton of changes that should not have been pushed
2021-08-09 18:26:30 +10:00
Geoffrey McRae
57a74c156b [client] egl: set a default scale for textures without filters 2021-08-09 18:24:33 +10:00
Geoffrey McRae
6882e5c59f [client] egl: provide the texture scale to the desktop shader
If the texture has a post-processing filter that has scaled the texture,
the desktop fragment shader needs to know this if it's doing linear
scaling.
2021-08-09 18:22:28 +10:00
Geoffrey McRae
f7f8060447 [client] egl: allow setting an output scale for a post-process shader 2021-08-09 17:57:36 +10:00
Quantum
53461d7515 [client] egl: simplify desktop vertex shader
In GLSL, using the / operator on two vectors of the same size divides the
vector component-wise, i.e. vec2(a, b) / vec2(c, d) == vec2(a / c, b / d).
2021-08-09 17:51:13 +10:00
Quantum
9b87f4ba5e [client] egl: cycle through multiple textures for dmabuf
This avoids race conditions in GL drivers when attempting to render and
call glEGLImageTargetTexture2DOES on the same texture.

Also, when using glEGLImageTargetTexture2DOES, we do not need to allocate
storage for textures.
2021-08-09 17:12:11 +10:00
Geoffrey McRae
f50ef4c23c [client] egl: remove includes from testing 2021-08-09 14:13:03 +10:00
Geoffrey McRae
86d6b67337 [client] egl: rework egl to accomodate post-processing filtering 2021-08-09 14:08:10 +10:00
Geoffrey McRae
91d1b8d2cd [client] egl: refactor egl_texture_free to use project naming standards 2021-08-08 17:52:13 +10:00
Geoffrey McRae
baf9661530 [client] egl: remove texture->ops indirection 2021-08-08 17:31:52 +10:00
Geoffrey McRae
266ad27998 [client] egl: refactor to use project naming standard 2021-08-08 17:16:10 +10:00
Geoffrey McRae
30ed563504 [client] interface: refactor to use camlCase function names 2021-08-08 15:43:42 +10:00
Geoffrey McRae
d347b28481 [client] egl: implement free for texture frambuffer 2021-08-08 15:35:13 +10:00
Geoffrey McRae
f8ae291090 [client] interface: switch to using UPCAST for the renderer's data 2021-08-08 15:32:01 +10:00
Geoffrey McRae
45d1f27fb4 [client] interface: rename LG_Renderer to LG_RendererOps
Part of the standardisation of using the `Ops` suffix for all interfaces
2021-08-08 14:43:04 +10:00
Quantum
b822e255d8 [client] egl: attempt DMABUF import and fallback if it fails
This should deal with drivers not supporting our DMABUF without attempting
to identify the drivers and blacklist them.
2021-08-08 09:53:47 +10:00
Geoffrey McRae
037b76750a [client] egl: revert glsync changes
`process` and `bind` are called from the same thread in order, there is
no need for atomic usage here.

This reverts commit 3d7dbd6371.
This reverts commit b3db1ba10b.
2021-08-08 09:44:59 +10:00
Geoffrey McRae
e949f2f8d2 [client] egl: texture_framebuffer should call the stream init 2021-08-08 09:43:28 +10:00
Geoffrey McRae
88c91d4752 [egl] texture: use more appropriate vairable names (parent & this) 2021-08-08 09:35:56 +10:00
Geoffrey McRae
3d7dbd6371 [client] egl: sync is now an atomic, access it as such 2021-08-08 09:26:36 +10:00
Quantum
b3db1ba10b [client] egl: eliminate GLsync object leaks
It used to be the case that we overwrite this->sync even if it was non-zero
when updating the texture, without deleting the sync object. If we update
faster than we render, the result would be leaking sync objects.

This commit ensures that sync objects are deleted when they are replaced.
2021-08-08 09:19:04 +10:00
Geoffrey McRae
12d256c7c8 [client] egl: do a full redraw if nightvision is toggled 2021-08-08 08:42:08 +10:00
Quantum
3e32e01c30 [client] egl/imgui: use imgui for night vision gain configuration 2021-08-08 08:42:01 +10:00
Quantum
ac3677d9ae [client] egl: implement partial copies for framebuffer textures
This uses the same line sweep algorithm originally created to copy DXGI
textures to IVSHMEM to implement the copy from IVSHMEM to memory-mapped
pixel buffer objects.
2021-08-08 08:30:11 +10:00
Geoffrey McRae
e755f0befa [client] egl: correct letterbox area clear attempt #3 - sigh 2021-08-07 04:00:18 +10:00
Geoffrey McRae
9f6ad864ed [client] egl: correct letterbox area clear attempt #2 :) 2021-08-07 03:44:47 +10:00
Geoffrey McRae
e5a138d854 [client] egl: properly clear the letterbox areas 2021-08-07 03:28:52 +10:00
Geoffrey McRae
162b1b93db [client] main: don't include the swap into the render timings
If vsync is enabled the swap will block until vblank skewing the timing
metrics.
2021-08-07 01:45:42 +10:00
Quantum
5d3c00717a [client] egl/imgui: use imgui for scaling algorithm selection 2021-08-06 22:49:49 +10:00
Quantum
dc7fd74327 [client] egl: refactor config dialog into main egl module
This will allow other things like scaling to be implemented.
2021-08-06 22:49:49 +10:00
Quantum
65e550a61c [client] egl: pad buffer damage by 1px when rendering desktop
We want to add extra padding to deal with scaling, which may end up blending
with neighbouring pixels.
2021-08-05 08:05:30 +10:00
Geoffrey McRae
6c44bbb53e [client] egl: use a ui switch for damage display instead of a keybind 2021-08-05 00:56:31 +10:00
Quantum
51b9cd4e5a [all] copyright: use unicode copyright sign ©
This is done for consistency with the license strings in appstrings.c.
2021-08-04 21:16:35 +10:00
Geoffrey McRae
f7682c289a [client] egl: always render the black bar areas
Failure to render these causes artifacts/ghosting when the overlays are
using this area, as such we need to always render this area of the
screen.
2021-08-03 22:29:04 +10:00
Geoffrey McRae
4b4a75475a [client] egl: fix out by one error with letterbox rendering
This replaces the scaled `destRect` with a version that uses doubles
correcting the rounding error that is causing a failure to properly
clear the black bar areas.
2021-08-03 22:27:46 +10:00
Quantum
55703b61b7 [client] egl: remove texture copy in DMABUF path
With more efficient rendering due to buffer age support, we no longer need
to copy the texture to avoid excessive dmabuf copies.
2021-08-03 21:37:31 +10:00
Quantum
8545d15c85 [client] egl: warn when EGL_EXT_buffer_age is not supported 2021-08-03 21:37:31 +10:00
Quantum
87aac8cf03 [client] egl: use buffer age extension to render only damaged parts
We avoid rendering any area that has not changed since the buffer was used
and also not covered by an overlay.
2021-08-03 21:37:31 +10:00
Quantum
f9977332a6 [client] egl: convert desktop to use desktop_rects 2021-08-03 21:37:31 +10:00
Quantum
2dca056526 [client] egl: refactor damage mesh generation into desktop_rects
This mesh will later be used to render only damaged portions of the desktop.
We also moved the coordinate transformation for damage overlay into a matrix
and computed by the shader.
2021-08-03 21:37:31 +10:00
Quantum
dd31a7ef93 [client] egl: clean up splash background shader 2021-08-03 21:12:46 +10:00
Geoffrey McRae
4d9ab81ef4 [client] egl: assert the update provdided is a dmabuf 2021-08-03 04:03:37 +10:00
Geoffrey McRae
f3413815a9 [client] egl: re-implement DMABUF (untested) 2021-08-03 03:59:03 +10:00
Quantum
5d5e4ede1a [client] egl: use new EGL damage count semantics
After the damage queue PR, EGL damage count 0 means no change, and -1 means
invalidate the entire window. However, several other places have different
semantics, and we are not handling them correctly:

1. KVMFR uses 0 to signal invalidating the entire frame, so if we receive 0
   rectangles in egl_on_frame, we should set damage count to -1.
2. The damage overlay treated 0 as full damage, which is now incorrect. This
   is fixed, and now it treats 0 as no update, and -1 as full damage.
2021-08-03 00:57:32 +10:00