|
cc841237
|
2024-11-29T13:45:50
|
|
Accept framebuffer modifications while PLS is active
The only way for a WebGL implementation to know if PLS is actually
active is to call glGetIntegerv(PIXEL_LOCAL_STORAGE_ACTIVE_PLANES_ANGLE)
(because glBeginPixelLocalStorageANGLE() can fail). So the original
behavior of not allowing glBindFramebuffer() et. al. while PLS was
active created a state scenario that was expensive for the browser to
track.
Instead, just allow glBindFramebuffer() et. al., and implicitly disable
PLS if they are called while it's active.
Bug: angleproject:40096838
Change-Id: Ibd303f9f9950fb5b7f1add2d41882e4379c51e62
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/6060301
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7070a9e9
|
2024-11-20T00:21:16
|
|
Remove draw buffer validation clauses from PLS
It was cumbersome to implement all this validation browser side for
WebGL.
Rather than making it an error to update blend and color mask on
reserved PLS draw buffers, glBeginPixelLocalStorageANGLE() can just
implicitly disable blend, and enable the color mask on overridden draw
buffers. Later calls to enable blend or change the color mask on
overridden planes are silently ignored until
glEndPixelLocalStorageANGLE().
Bug: angleproject:40096838
Change-Id: Ic7e1c5113e7d3fad3b80d0178075df646540d743
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/6045421
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
b16d105f
|
2024-10-03T10:25:32
|
|
Remove Desktop GL front-end support
For Desktop GL applications, please use Zink!
Bug: angleproject:370937467
Change-Id: Ie734634bb62a2e98c80e1b32d8b3d34624da3c04
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5905428
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
966739ac
|
2024-09-19T23:49:06
|
|
Drop PLS support for EXT_shader_pixel_local_storage
Supporting this backend drastically increased the complexity of the
codebase, with little return. We don't support memoryless attachments on
the web anyway, and since this extension requires us to literally draw
the load/store operations, input attachments on Vulkan perform better.
Once this implemention is completely removed, we will delete the PLS
allow list, which isn't required for the other PLS implementations.
Bug: angleproject:7279
Change-Id: Ibb036d36cbd33467e7a94398ce171cda7349e4f4
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5874412
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
d193d51b
|
2024-06-17T22:46:08
|
|
Replace issue ids post migration to new issue tracker
This change replaces anglebug.com/NNNN links.
Bug: None
Change-Id: I8ac3aec8d2a8a844b3d7b99fc0a6b2be8da31761
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5637912
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
5857e0d6
|
2024-01-22T00:00:00
|
|
Metal: Adjust raster order groups AMD workaround condition
Ensured that the raster order groups AMD workaround is
used not only when there is no attachment on slot 0 but
also when an existing attachment on slot 0 is disabled.
Bug: angleproject:8263
Change-Id: I62352eb5d6fb1cce4b6a4723d41862c6c213b9d8
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5230638
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
42d75c7e
|
2024-01-17T00:00:00
|
|
Metal: Fix raster order group AMD workaround
The AMD-specific raster order group workaround relied
on disabled framebuffer attachments being implemented
via all-disabled color masks.
Metal backend no longer includes disabled framebuffer
attachments in render pass descriptors.
Updated the workaround to keep the
placeholder attachment enabled and
to set its color mask explicitly.
Fixed: angleproject:8263
Change-Id: I815ceede84e6ea89eefc10c445768d9d569d0d82
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/5208110
Commit-Queue: Alexey Knyazev <lexa.knyazev@gmail.com>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
959ccb96
|
2023-08-02T13:40:16
|
|
Add EGL_BAD_ACCESS validation into ValidateMakeCurrent
Current code does not generate EGL_BAD_ACCESS in `eglMakeCurrent()`
according to the specification:
- If context is current to some other thread, or if either draw or read
are bound to contexts in another thread, an EGL_BAD_ACCESS error is
generated.
Additional changes:
- replace `Context::getRefCount()` with `isReferenced()` for
consistency with added `Surface::isReferenced()` method.
- remove duplicate `ValidateSurface` call for the
`drawSurface` in case if `readSurface` is the same.
- remove duplicate `ValidateCompatibleSurface()` call for the
`readSurface` in case if `drawSurface` is different.
- Fix EGLContextASANTest.DestroyContextInUse test.
- Fix EGLContextSharingTestNoSyncTextureUploads.NoSync test.
Bug: angleproject:8282
Change-Id: I2af0d7f50019c2445b42279bdd0e0197dd8ad20f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4742382
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Igor Nazarov <i.nazarov@samsung.com>
|
|
|
9962f078
|
2023-07-11T15:39:50
|
|
Pass only context-private state to private entry points
This change ensures that the implementation for these entry points
cannot access anything other than context-private state.
Bug: angleproject:8224
Change-Id: I988672b138d861db25e91d71ab8c34baa4e8ebee
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4678783
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
|
|
|
40111c68
|
2023-07-11T14:21:21
|
|
Rename context-local to context-private state
Bug: angleproject:8224
Change-Id: I1bb39475043f8fb14d683d11a038b4850692a8c6
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4678781
Auto-Submit: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
44395930
|
2023-07-05T11:59:23
|
|
Make various state setting entry points lockless
These entry points only set context-local state and thus don't require
locking.
Bug: angleproject:8224
Change-Id: I428c23cc862e9356d571bc085b5df0bf48017175
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4661700
Reviewed-by: Igor Nazarov <i.nazarov@samsung.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
9daac2b7
|
2023-07-05T11:36:46
|
|
Make glEnable/Disable entry points lockless
These entry points only set context-local state and thus don't require
locking.
Bug: angleproject:8224
Change-Id: Id4eab729115bd75f82e1ec7a27355c821a7c4320
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4661697
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Igor Nazarov <i.nazarov@samsung.com>
Reviewed-by: Charlie Lao <cclao@google.com>
|
|
|
be41fe29
|
2023-07-04T15:35:46
|
|
Make glColor/DepthMask entry points lockless
These entry points only set state that is entirely accessed by the
owning context (context-local) and thus don't require locking.
glColorMask* functions also affect the cached context state (in
particular draw validity), so the relevant cached state is also modified
to support being locklessly modified.
Bug: angleproject:8224
Change-Id: I221b4efa25fc1c11419d1ac942f1c37e59ec92c0
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4658173
Reviewed-by: Igor Nazarov <i.nazarov@samsung.com>
Reviewed-by: Charlie Lao <cclao@google.com>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
7d4c6d1d
|
2023-05-09T12:19:54
|
|
Allow glDelete* while PLS is active
Banning glDelete* is extremely dangerous. It will almost definitely
cause memory leaks in client code, and it makes JS garbage collection
needlessly complex.
Instead, specify that PLS is implicity deactivated if the client deletes
anything that is attached to the current draw framebuffer during a PLS
rendering pass.
Bug: chromium:1421437
Change-Id: I3a18ee6b5d5567431e6fa3eccea58cb049845502
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4521436
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
45f91a8c
|
2023-05-08T13:11:44
|
|
Deinitialize PLS planes when their texture is deleted
The spec originally called for PLS planes to be converted to memoryless
when their texture was deleted, but this is not compatible with WebGL,
which does not support memoryless planes. Change the behavior to
deinitialize them instead.
This change requires the addition of a new observer message,
angle::SubjectMessage::TextureIDDeleted, which PLS uses to deinitialize
a plane when the app deletes its texture.
Bug: chromium:1421437
Change-Id: I58fd91003747160f0a1abc1a8a7a87668890ba1f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4518565
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
1e166daf
|
2023-04-10T23:49:46
|
|
Finalize PLS support for non-2D texture types
Commit to supporting GL_TEXTURE_2D_ARRAY and GL_TEXTURE_3D, and add
tests. Suppress D3D11 and Metal for now, since shader image support for
non-2D types needs some work on those backends.
Drop support for cube maps, since those will cause complications on an
implementation based on texelFetch.
Bug: chromium:1421437
Bug: angleproject:8124
Bug: angleproject:8125
Change-Id: I403ae157b7d2609f2190a98c2576c92f73441412
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4415150
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
a491bbe3
|
2023-03-18T19:05:08
|
|
Add PLS utilities for interrupting a rendering pass
Adds two more simple commands to ANGLE_shader_pixel_local_storage that
allow WebGL and the command buffer to interrupt rendering passes without
having to either (1) make expensive queries, or (2) track lots of
complex state for validation that they are not currently equipped to
track.
Bug: chromium:1421437
Change-Id: I80eaef3ae6b0b4bbbecb9cd2268ac90b43675d1c
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4355032
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
|
|
|
513ca723
|
2023-03-13T14:30:20
|
|
Delete GL_LOAD_OP_DISABLE_ANGLE from PLS
This load op makes the PLS spec and WebGL implementations unnecessarily
complex.
Bug: chromium:1421437
Change-Id: Iab02a8b02083899c6cc345ecb25b88c5871611c9
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4335148
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
eb1a13d8
|
2022-12-06T21:56:42
|
|
Improve PLS load and store op tokens
Define more new tokens that are more clearly named, as opposed to
recycling tokens that were meant for a different purpose.
Bug: angleproject:7279
Change-Id: I840e5ea168235c49eec2693b2231c2d329027b0e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4082057
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
9bd9c1c2
|
2022-11-03T11:49:33
|
|
Workaround for AMD [[raster_order_group()]] bug
Metal [[raster_order_group()]] does not work for read_write textures on
AMD when the render pass doesn't have a color attachment on slot 0. To
work around this we attach one of the PLS textures to
GL_COLOR_ATTACHMENT0, if there isn't one already.
Bug: angleproject:7792
Bug: angleproject:7803
Bug: angleproject:7818
Change-Id: I7567ed28ed30c7e3ca23dc52778ed90f2fdff965
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4004529
Reviewed-by: Kyle Piddington <kpiddington@apple.com>
Commit-Queue: Chris Dalton <chris@rive.app>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
8ee1b89f
|
2022-11-04T13:10:37
|
|
Refactor pixel local storage options
The various different PLS options were getting scattered and unruly. We
are also in need of more backend-specific PLS options that would be
difficult to add as-is. This CL refactors them into a single
"ShPixelLocalStorageOptions" struct that gets initialized all in one
place, and shared between the compiler and the backends.
Bug: angleproject:7279
Change-Id: Ic58dccb8d1ba350a0b6cc5848ce15bd687e30fad
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/4006715
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
9bda9a79
|
2022-10-22T22:05:11
|
|
Add Store Ops to pixel local storage
Browsers will need the ability to pre-empt pixel local storage, which
means every plane will need a backing store to dump to. Store Ops allow
the app to still avoid memory transactions at the end of PLS even if
their plane has a backing texture.
Bug: angleproject:7279
Change-Id: I3a3efa21773f87c03cd346a996e3c638028c68ab
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3974652
Commit-Queue: Chris Dalton <chris@rive.app>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
|
|
|
3605b399
|
2022-10-20T17:00:02
|
|
Move PLS clear values back into context state
The API that required packing raw data into a buffer was un-ergonomic
for developers and difficult to implement for WebGL vendors.
Bug: angleproject:7279
Change-Id: If7c98908c285462c5775e8e2d8811883be139f64
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3972376
Commit-Queue: Chris Dalton <chris@rive.app>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
5e95a4d9
|
2022-10-07T00:52:38
|
|
Add an EXT_shader_pixel_local_storage impl of PLS
Translates ANGLE_shader_pixel_local_storage shaders directly to
EXT_shader_pixel_local_storage.
Polyfills load/store operations using internal fullscreen draws.
Since the ANGLE extension needs the ability to preserve all active PLS
planes to textures, we can only support this extension when the backend
context also has access to ES 3.1 shader images.
Bug: angleproject:7279
Bug: angleproject:7771
Change-Id: Id348bde412efcc081ff29ee05ec59ad652f77569
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3966075
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
|
|
|
9f693aa3
|
2022-10-22T14:45:59
|
|
Implement an allow list for PLS
In order to guarantee no data is lost while using the
EXT_shader_pixel_local_storage extension, we need to restrict
applications to a small subset of commands while pixel local storage is
active. This CL implements the allow list for GL entrypoints using
wildcard matching inside the code generator, and adds custom validation
for the more specific restrictions that go into effect when PLS is
active.
Bug: angleproject:7279
Change-Id: I5dd48bd93c10e8775f32be32a4fcf17855eb2f0e
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3932552
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
a7dc51f2
|
2022-10-01T08:49:11
|
|
Add a framebuffer fetch implementation of PLS
The framebuffer fetch implementation works by attaching PLS backing
textures to the framebuffer, and then rewriting PLS uniforms as "inout"
fragment variables. The compiler's existing machinery takes it from
there and makes it work on GL and Vulkan, and soon Metal.
EXT_shader_framebuffer_fetch is now the preferred backend for pixel
local storage, but we also use EXT_shader_framebuffer_fetch_non_coherent
if shader images can't be coherent. This is especially interesting for
Vulkan, since noncoherent framebuffer fetch is possible without any
extensions.
Bug: angleproject:7279
Bug: angleproject:7683
Bug: angleproject:7684
Bug: angleproject:7724
Change-Id: I33f3b2c6df9a5709969d9165c448ea71b096c9e1
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3900142
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|
|
|
8b2aff28
|
2022-09-12T10:27:28
|
|
Implement the ANGLE_shader_pixel_local_storage API
Implements the OpenGL ES API for ANGLE_shader_pixel_local_storage and
adds thorough validation and testing as outlined in the spec. This
feature is still implemented entirely in the frontend, but the extension
now works end-to-end with a passing test suite, and can be used
externally. Over time we can start gradually moving the implementation
into backends as appropriate.
Bug: angleproject:7279
Bug: angleproject:7647
Change-Id: I1c861a0fca96423be02e17bbe1fb7f57b99ea63f
Reviewed-on: https://chromium-review.googlesource.com/c/angle/angle/+/3886462
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Chris Dalton <chris@rive.app>
|