We only need to call the syncState for samplers when they are dirty.
Also includes changes to refactor out the sampler sync in GL. Adds
observer bindings so sampler sync is handled correctly in resource
sharing scenarios.

Bug: angleproject:2763
Change-Id: I762f0738ee7572ae29ce6bd5384a30aa9588c848
Reviewed-on: https://chromium-review.googlesource.com/1227797Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This patch is the first one in the series of supporting GLSL texture
function textureGather/textureGatherOffset on D3D11.

According to ESSL 3.1 SPEC (Chapter 8.9.3, Page 130), the definition
of textureGather on sampler2D is:
    gvec4 textureGather (gsampler2D sampler, vec2 P[, int comp])

The parameter "comp" is optional, and if it is specified, the value
of "comp" must be a constant integer expression with a value of 0, 1,
2, or 3, identifying the x, y, z, or w postswizzled component of the
four-component vector lookup result for each texel, respectively. If
comp is not specified, it is treated as 0.

According to the definition above, textureGather is equivalent to
Texture2D.Gather[Red|Green|Blue|Alpha] in HLSL, where we can use a
switch-case expression to choose the right HLSL texture function.

The features listed here will be implemented in the following patches:
1. Support textureGatherOffset
2. Support textureGather[Offset] on isamplers and usamplers
3. Support textureGather[Offset] on textures when swizzle is on
4. Support textureGather[Offset] on shadow samplers

BUG=angleproject:2826
TEST=dEQP-GLES31.functional.texture.gather.basic.2d.rgba8.*
     dEQP-GLES31.functional.texture.gather.basic.cube.rgba8.*
     (without texture_swizzle)
     dEQP-GLES31.functional.texture.gather.basic.2d_array.rgba8.*
     (without texture_swizzle)

Change-Id: Iff2ed4f8b65dad613cb0bafdfd19f8f0528e832c
Reviewed-on: https://chromium-review.googlesource.com/1232980
Commit-Queue: Jiawei Shao <jiawei.shao@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

Since this is about to be used for the Q release, and we haven't
receieved permission from the maps team to opt-in their app, we'll
switch to our team's test app.

Bug: angleproject:2794
Change-Id: Idbda050b1130c359cda68066f187aadeb8e949b5
Reviewed-on: https://chromium-review.googlesource.com/1236658
Commit-Queue: Ian Elliott <ianelliott@google.com>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

These are already passing.

Bug: angleproject:2644
Change-Id: Iac3452547e319a709e010431eb8a45f5523bb5f7
Reviewed-on: https://chromium-review.googlesource.com/1235653Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This improves object binding performance. We no longer need to check
two containers when manipulating the bound subjects/observers.

Bug: angleproject:2763
Change-Id: I0ccb2e76be47bd8a28fba5a0c3eff857aa166bb7
Reviewed-on: https://chromium-review.googlesource.com/1227795
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>

Also update a test to trigger the bug.

Bug: angleproject:2763
Change-Id: I8e6735320a34dcc4cc8beee1b7a22b768912f24f
Reviewed-on: https://chromium-review.googlesource.com/1234338Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

The D3D11 backend used to crash to a null pointer dereference if a
glClearBuffer call was done on a nonexistent depth or stencil
attachment. Validate for these conditions so that the backend can't
crash in this case.

BUG=angleproject:2827
TEST=angle_end2end_tests

Change-Id: Iecee78d213d11d492d52f246b4b068e8b6f34244
Reviewed-on: https://chromium-review.googlesource.com/1233675Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

Re-land fixes memory leak and adds inline to a few more methods.

This optimized vector keeps an initial fixed size storage but has
unbounded growth like a normal std::vector. It operates like a
FixedVector initially but then switches the storage to an allocated
pool when the element count exceeds the array limit.

This gives fast performance in the small case since no dynamic
allocation is needed. It also handles the "slow" big case.

Bug: angleproject:2763
Change-Id: Ib3cc7b652a14ed346528bf15820023ab0b5317ce
Reviewed-on: https://chromium-review.googlesource.com/1233453
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

This is not needed and results in creating an empty command buffer.

Bug: angleproject:2818
Change-Id: I06e5542cb3cd819c18758c909b8fcb624b074874
Reviewed-on: https://chromium-review.googlesource.com/1232981Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Tobin Ehlis <tobine@google.com>

We add a novel multiview multisampling extension that includes the
requirement to explicitly resolve the multisampled framebuffer. The
explicit resolve is much more straightforward to implement on top of
OpenGL and D3D11 than implicit resolve found in the native extension
OVR_multiview_multisampled_render_to_texture. It also has predictable
performance characteristics.

The extension allows multiview drawing to 2D multisample texture
arrays and is now enabled on both the GL backend and the D3D11
backend. The implementation is fairly simple, as it involves just
small changes in validation to allow multisampled framebuffer
attachments. The multiview rendering logic is exactly the same
regardless of whether multisampling is enabled.

For the most part the same tests are used to test both multisampled
and non-multisampled rendering. The tests will use a different
framebuffer setup depending on the test param. They resolve the
multisampled framebuffer to a non-multisampled framebuffer prior to
any readbacks from the framebuffer.

Some of the tests are adjusted so that they have the correct sub-pixel
positioning of multisampled quads, so there won't be any pixels that
would be just partially covered.

The tests don't have any tolerance for partially covered pixels - if
we find any platforms where the tests run into a sub-pixel positioning
corner case, tolerance may need to be added later.

BUG=angleproject:2775
TEST=angle_end2end_tests

Change-Id: I590d7f300a92ea5439f2720d9db14a7976db2e1d
Reviewed-on: https://chromium-review.googlesource.com/1221214
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

Bug: angleproject:2825
Change-Id: I34a7c520d31cf0a9321d5599046159ef9bf58be5
Reviewed-on: https://chromium-review.googlesource.com/1232973
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

This reverts commit 4d45e7dc.

Reason for revert: Memory leak detected during roll in https://ci.chromium.org/p/chromium/builders/luci.chromium.try/linux_chromium_asan_rel_ng/101750

Original change's description:
> Add FastVector for simple dynamic vector cases.
> 
> This optimized vector keeps an initial fixed size storage but has
> unbounded growth like a normal std::vector. It operates like a
> FixedVector initially but then switches the storage to an allocated
> pool when the element count exceeds the array limit.
> 
> This gives fast performance in the small case since no dynamic
> allocation is needed. It also handles the "slow" big case.
> 
> Bug: angleproject:2763
> Change-Id: I9c002b205bd4ac9fc171d1fdc42bc8ea5fe8dabe
> Reviewed-on: https://chromium-review.googlesource.com/1227794
> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
> Commit-Queue: Jamie Madill <jmadill@chromium.org>

TBR=ynovikov@chromium.org,jmadill@chromium.org,syoussefi@chromium.org

Change-Id: Iaec76244cc736a8469b80bfc8534fb0fd6c49e11
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: angleproject:2763
Reviewed-on: https://chromium-review.googlesource.com/1232978Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

For D3D11 backend, dirtybits are used to sync and update state
for compute.
For GL backend, mProgramUniformBuffersDirty,
mProgramAtomicCounterBuffersDirty, mProgramImagesDirty are newly
applied.

Bug: angleproject:2814

Change-Id: Ia459ca26705b7c685179d64e0c934535bf16741f
Reviewed-on: https://chromium-review.googlesource.com/1215245Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jie A Chen <jie.a.chen@intel.com>

This changes to construct a new ShHandle of compiler for each Shader,
and use it to translate the shader source in a background thread.

Bug: chromium:849576

Change-Id: Ib49952c7292321ee6aa1c5996f8f7927f40d8f04
Reviewed-on: https://chromium-review.googlesource.com/1177195
Commit-Queue: Jie A Chen <jie.a.chen@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

This optimized vector keeps an initial fixed size storage but has
unbounded growth like a normal std::vector. It operates like a
FixedVector initially but then switches the storage to an allocated
pool when the element count exceeds the array limit.

This gives fast performance in the small case since no dynamic
allocation is needed. It also handles the "slow" big case.

Bug: angleproject:2763
Change-Id: I9c002b205bd4ac9fc171d1fdc42bc8ea5fe8dabe
Reviewed-on: https://chromium-review.googlesource.com/1227794Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This adds a custom logging step to DisplayAndroid and DisplayVkAndroid.
It's easiest to add to these two classes since we have no common info
logging code and other platforms may not want startup spam. It would
also be possible to put this into debug info hooks.

Bug: angleproject:2823
Change-Id: I12a1ea9e27880648d8a73a9e3faa56f318620cbe
Reviewed-on: https://chromium-review.googlesource.com/1230273Reviewed-by: Ian Elliott <ianelliott@google.com>
Reviewed-by: Cody Northrop <cnorthrop@google.com>
Commit-Queue: Cody Northrop <cnorthrop@google.com>

Slight performance increase in binding related tests.

Bug: angleproject:2763
Change-Id: I4d14177adab678d596cd34b8f1b99f5309f017f0
Reviewed-on: https://chromium-review.googlesource.com/1227793
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>

Re-land fixes memory leaks.

Using a custom array instead of std::vector speeds up the resource
manager. One reason is because calls to size() are implemented in many
implementations as a difference between two pointers. This sub size
implementations are slower than storing a simple size variable in a
custom class.

Also includes more inlining of hot spots functions.

Also includes a small unit test class for ResourceMap. And an unrelated
but small test fix for TextureLimisTest. Also a small unrelated fix for
a Transform Feedback test.

Increase the scores of the draw call perf test with texture and buffer
bindings and the buffer binding perf test.

Bug: angleproject:2763
Change-Id: Ic2f0f689107b2bf05c63da2ed6bbc9f0feea63f7
Reviewed-on: https://chromium-review.googlesource.com/1229033Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:2270
TEST=angle_end2end_tests.ComputeShaderTest.DispatchComputeIndirect/ES3_1_D3D11

Change-Id: I29bd888ce1e7e3e09b5635699a36164cd97cd7de
Reviewed-on: https://chromium-review.googlesource.com/1209563Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jie A Chen <jie.a.chen@intel.com>

Cannot have package name with 'google' in it, it is not dependent
on Google, ANGLE is an open source project.

Bug=angleproject:2418

Change-Id: If69be10b2d60789d0cce30b631249927026c5c0e
Reviewed-on: https://chromium-review.googlesource.com/1229434Reviewed-by: Cody Northrop <cnorthrop@google.com>
Commit-Queue: Cody Northrop <cnorthrop@google.com>

This reverts commit 57ff6f95.

Reason for revert: Memory leaks detected during roll in https://ci.chromium.org/p/chromium/builders/luci.chromium.try/linux_chromium_asan_rel_ng/100284

Original change's description:
> Inline and micro-optimize more for perf tests.
> 
> Using a custom array instead of std::vector speeds up the resource
> manager. One reason is because calls to size() are implemented in many
> implementations as a difference between two pointers. This sub size
> implementations are slower than storing a simple size variable in a
> custom class.
> 
> Also includes more inlining of hot spots functions.
> 
> Also includes a small unit test class for ResourceMap. And an unrelated
> but small test fix for TextureLimisTest. Also a small unrelated fix for
> a Transform Feedback test.
> 
> Increase the scores of the draw call perf test with texture and buffer
> bindings and the buffer binding perf test.
> 
> Bug: angleproject:2763
> Change-Id: I41c327987db27ac45e6a62579f01e1cdc22e396c
> Reviewed-on: https://chromium-review.googlesource.com/1171510
> Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Commit-Queue: Jamie Madill <jmadill@chromium.org>

TBR=fjhenigman@chromium.org,ynovikov@chromium.org,jmadill@chromium.org,syoussefi@chromium.org

Change-Id: Ie047289c9bf23a842c3cbb9692c811da0534991c
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: angleproject:2763
Reviewed-on: https://chromium-review.googlesource.com/1228893Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

Support for multisampled EGL surfaces was added in
151d5de6 (2017-04-13, "Enable MSAA for texture client buffers")
but the resolve step was missing because it was not needed for
EGL_ANGLE_d3d_texture_client_buffer. However, when using MSAA
with a regular EGL surface then resolving is required to get
an antialiased image.
Please note that the new test case CreateSurfaceWithMSAA does
actually not test the newly added resolve step because the
resolve is performed by the glReadPixels() call there.
So it is rather a general test if MSAA works for EGL surfaces.

TEST=angle_end2end_tests.EGLSurfaceTest.CreateSurfaceWithMSAA

Change-Id: Ieafd6877fa510d5e16c0d9c6872c31fa73efa86c
Reviewed-on: https://chromium-review.googlesource.com/1181138Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

Using a custom array instead of std::vector speeds up the resource
manager. One reason is because calls to size() are implemented in many
implementations as a difference between two pointers. This sub size
implementations are slower than storing a simple size variable in a
custom class.

Also includes more inlining of hot spots functions.

Also includes a small unit test class for ResourceMap. And an unrelated
but small test fix for TextureLimisTest. Also a small unrelated fix for
a Transform Feedback test.

Increase the scores of the draw call perf test with texture and buffer
bindings and the buffer binding perf test.

Bug: angleproject:2763
Change-Id: I41c327987db27ac45e6a62579f01e1cdc22e396c
Reviewed-on: https://chromium-review.googlesource.com/1171510Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Currently we limit the maximum of State bits to 64 for performance
consideration. To accommodate new ES31 states, we have to pick and
squash some existing bits which are less hot.

Bug: angleproject:2814

Change-Id: I8fc19286bbf8d8273949da8d986a8b1587fa105e
Reviewed-on: https://chromium-review.googlesource.com/1227069Reviewed-by: Jie A Chen <jie.a.chen@intel.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jie A Chen <jie.a.chen@intel.com>

This change adds shader type checking before entering
initializeOutputVariables.

BUG: angleproject:2821
Change-Id: Ib931031f2fc187f1f2a1821a09664bbe172a5e90
Reviewed-on: https://chromium-review.googlesource.com/1226229
Commit-Queue: Jiajia Qin <jiajia.qin@intel.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

glBufferSubData is calling vkMapMemory with a size of zero is invalid.
Check for that and exit early if found.

Bug: angleproject:2790

Change-Id: I965badeb3aa2cec1adc24dd7ff5695f8aa3e553d
Reviewed-on: https://chromium-review.googlesource.com/1194610
Commit-Queue: Courtney Goeltzenleuchter <courtneygo@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

This removes the temporary defines now that Chromium is up-to-date.

Bug: angleproject:2806
Change-Id: I8e3308d6b80e49c2b2a7a698ee5d8ad40f442475
Reviewed-on: https://chromium-review.googlesource.com/1227031Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

These were both missing the correct layer offset. Cache the layer
inside the RenderTargetVk for easy access.

Bug: angleproject:2470
Change-Id: I690dbf0702d7ec52f44ba0a9429b6ef0e51baf6b
Reviewed-on: https://chromium-review.googlesource.com/1225910
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

Bug: angleproject:2763
Change-Id: Idcb8a714a6dccdff3dd953679e17a7314866a37d
Reviewed-on: https://chromium-review.googlesource.com/1214206
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>

Is much faster for completeness cache checks in syncProgramTextures.

Bug: angleproject:2763
Change-Id: Iffdacbb8a4f6640caa5051643c379a7b4c3311b6
Reviewed-on: https://chromium-review.googlesource.com/1171508
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

This will more easily allow us to use a cached completeness comparison
value. The cached value only gets updated on certain setters.

Bug: angleproject:2763
Change-Id: Ib80db8517560617087ae8360f7af69d6c2392b76
Reviewed-on: https://chromium-review.googlesource.com/1171507Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Android expects to use eglSwapBuffersWithDamage so implement that
for Vulkan backend.

BUG=angleproject:2464

Change-Id: I8d599e70e6c8fee92b530d472dc97e3996630458
Reviewed-on: https://chromium-review.googlesource.com/1220066
Commit-Queue: Courtney Goeltzenleuchter <courtneygo@google.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

The following enums had a small typo causing them to be takend from an
unused block:

EGL_CONTEXT_WEBGL_COMPATIBILITY_ANGLE
EGL_CONTEXT_BIND_GENERATES_RESOURCE_CHROMIUM
EGL_DISPLAY_TEXTURE_SHARE_GROUP_ANGLE

Fix this by temporarily defining a secondary enum. Once Chromium is
fixed upstream we can remove the temporary enum.

Reported by dan.glastonbury@gmail.com.

Bug: angleproject:2806
Change-Id: I2ce1dec0a30d5e83f2bec0ae243fd860cbcea4fc
Reviewed-on: https://chromium-review.googlesource.com/1220371
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

One test was working but the suppression was not lifted. The other test
had no issue filed.

Bug: angleproject:2802
Change-Id: I88cd082b2edeca30ec4a8406809af749bb6ce152
Reviewed-on: https://chromium-review.googlesource.com/1200367Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Mostly inlining checks on the hot draw call path.
Slight increase in draw call validation performance. (Up to 13%)

Bug: angleproject:2747
Change-Id: I34c4d7f412c3bca5e559e9bfb5689c0618bb7536
Reviewed-on: https://chromium-review.googlesource.com/1171506Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This optimizes some of our dynamic switching. It uses packed enum maps.

Bug: angleproject:2747
Change-Id: Ibd2f9306d066f2fd9eb64c99a25668b7ba5c009c
Reviewed-on: https://chromium-review.googlesource.com/1171505
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>

New test coverage for robust initialization of multisampled textures
is added in angle_end2end_tests. Some missing functionality that was
affecting the initialization of multisampled 2D array textures is now
fixed.

BUG=angleproject:2775
TEST=angle_end2end_tests

Change-Id: I1d9d3cd154ca2910159941fe8b1bef4ae6320bdd
Reviewed-on: https://chromium-review.googlesource.com/1224530Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

This can be used for fast lookups with InvalidEnum. Will be useful
for optimizing valid primitive mode lookups.

Bug: angleproject:2747
Change-Id: I5d668863eb23d3c1eb85a775b3e9e46f25a0a335
Reviewed-on: https://chromium-review.googlesource.com/1171504Reviewed-by: Shahbaz Youssefi <syoussefi@chromium.org>
Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Each TextureVk now stores vectors of RenderTargetVks and ImageViews
associated with the image faces. They are initialized lazily when the
RenderTarget is queried in getAttachmentRenderTarget.

There's still one missing edge case for handling clear with the
Framebuffer when using cube maps. Also one additional test failure on
Android.

Bug: angleproject:2470
Change-Id: Ib959a3434a992cef010a11940cf2ee49e118ac17
Reviewed-on: https://chromium-review.googlesource.com/1220727Reviewed-by: Frank Henigman <fjhenigman@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

These are suspected of causing an out-of-memory failure.

BUG=angleproject:2655,angleproject:2549

Change-Id: I92bb4290c79ae8d7b572034666ec181c50c2a587
Reviewed-on: https://chromium-review.googlesource.com/1222370Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>