We would allocate this texture for devices that don't need it. Instead
do this lazily. This was showing up in the profiles for the MotionMark
benchmark, possibly due to re-creating the surface.

BUG=angleproject:1155

Change-Id: I28b5eda29e21899fc8afef054e1b8063e3cc2e00
Reviewed-on: https://chromium-review.googlesource.com/655479Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This was showing up in profiles of the MotionMark benchmark, possibly
due to re-creating the canvas many times.

BUG=angleproject:1155

Change-Id: Id857b89770b846881fb381f7c3ab70c07b924271
Reviewed-on: https://chromium-review.googlesource.com/655478Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This covers a regression introduced with texture dirty bits.

BUG=angleproject:1387

Change-Id: Ic8112718c185298ef54ec5a6f6ed2cd519e010d6
Reviewed-on: https://chromium-review.googlesource.com/653586Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

The ANGLE_multiview extension can be supported with three possible code
paths - through view being selected in the vertex shader on D3D and
OpenGL, and through the view being selected in the geometry shader on
D3D only. This patch extends the multi-view performance tests to
benchmark these three different code paths.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: I443e4db64a95eede1142718a43a095ee5a03738c
Reviewed-on: https://chromium-review.googlesource.com/652466
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>

Renderer11::createRenderTarget already fixed this issue but it also is
exposed in SwapChain11::resetOffscreenDepthBuffer now that multisampled
surfaces are supported in ANGLE.

BUG=angleproject:2136

Change-Id: I978666ebc1bb3db14ddf69954d7eb750391bf7a8
Reviewed-on: https://chromium-review.googlesource.com/653779
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

The enums introduced by the ANGLE_multiview extension are changed
to use the values reserved specifically for ANGLE.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: Ib668f2e0e9022442a432b1a04050aeb2ff82a3d1
Reviewed-on: https://chromium-review.googlesource.com/654864Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Martin Radev <mradev@nvidia.com>

Until this patch multi-view layered framebuffers used to be cleared by
attaching a single layer of each attachment to a framebuffer and calling
the Clear* command for that internal framebuffer.
According to the GL 4.1+ specifications, Clear* commands clear all of
the layers of an attached 2D texture array. If all of the layers are
active for a multi-view layered framebuffer, then we can directly call
the corresponding Clear* command instead of iterating over each layer
and clearing it.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: Ie4dfd9fff47715b502f358272bfc47c0373c4e91
Reviewed-on: https://chromium-review.googlesource.com/649209
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>

View selection can happen in the vertex shader through the optional
feature VPAndRTArrayIndexFromAnyShaderFeedingRasterizer.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: Iaf65685e04f828b0936295fea867f6f6cbe69bee
Reviewed-on: https://chromium-review.googlesource.com/628419
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:2122

Change-Id: Ia51c139197e1b90e54505278d301ae8ac7dab53d
Reviewed-on: https://chromium-review.googlesource.com/653240
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

The MSDN documentation states that BGRA4, RGB5A1 and RGB565 should all
support 4 samples with DXGI 1.2 and D3D11.1 but some drivers appear to
not have full support.  Fall back to RGBA8 render targets when the
driver cannot support at least 4 samples for ES3 feature levels.

BUG=761489

Change-Id: I6bcd417700f1188945e8032ca6a64c4fbb2bc8a8
Reviewed-on: https://chromium-review.googlesource.com/652828Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

This trybot was broken during the removal of the Linux ChromiumOS
Builder from chromium.gpu.fyi and is blocking all ANGLE CQ jobs.

BUG=762377
TBR=dpranke@chromium.org, jmadill@chromium.org
NOTRY=true

Change-Id: Ibc3ccf3af842ee8b040014732d179573f286afad
Reviewed-on: https://chromium-review.googlesource.com/653244
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>

The patch extends Renderer11 so that a D3D11.1 device and
D3D11.3 context can be created. This is necessary for using
the D3D11.3 feature called
VPAndRTArrayIndexFromAnyShaderFeedingRasterizer.

BUG=angleproject:2062
BUG=angleproject:2145
TEST=angle_end2end_tests

Change-Id: I84c761b2897d7d911686f5b6d79cb93e233015a0
Reviewed-on: https://chromium-review.googlesource.com/591448Reviewed-by: Geoff Lang <geofflang@chromium.org>

Without DXGI, EnumDisplayDevices is the only way to get the
primary device. Previously the code would work on most configs
with a combination of AMD, Intel and NVIDIA GPUs but would fail
on more esoteric system. Like our try bots that have Matrox GPUs.

BUG=angleproject:1874
BUG=angleproject:2137

Change-Id: Ie2dfbb559001ccad2fd5b8a8fd6436e0fba9d003
Reviewed-on: https://chromium-review.googlesource.com/651629Reviewed-by: Geoff Lang <geofflang@chromium.org>

This would drastically slow down some framebuffer updates.

BUG=angleproject:1155

Change-Id: Ibff428c49bd5127a101bce46cc1df355f4542ccf
Reviewed-on: https://chromium-review.googlesource.com/649986
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

This allows the dirty objects to notify the context if something
needs to be communicated down to the ContextImpl. In this case it
means a dirty Texture can notify the Context that it needs to re-apply
texture bindings.

BUG=angleproject:1387

Change-Id: I162115e51112d1c27c0dab621d1b3d14446af96d
Reviewed-on: https://chromium-review.googlesource.com/648052Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Discovered some of these while investigating Texture dirty bits.

BUG=angleproject:1387

Change-Id: I8b170462bfd283e4b0f9d47b7f7ddbaa7957914d
Reviewed-on: https://chromium-review.googlesource.com/648051Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:2122

Change-Id: I99eed52b1f12004f0bab3f94bd3acddda8dafd69
Reviewed-on: https://chromium-review.googlesource.com/650526Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

This change adds GL_KHR_robust_buffer_access_behavior support.
The old change is in https://chromium-review.googlesource.com/c/angle/angle/+/607413

BUG=755897, angleproject:1393, angleproject:1463

Change-Id: I04a1132c3ae8d3a766194df61c4ff7bf0b084f03
Reviewed-on: https://chromium-review.googlesource.com/640750
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

A branch is added in the geometry shader to select either the
viewport, or texture layer which is being rendered to based on the
value of a uniform in the driver constant buffer. Using this approach
there is no need for separate programs for side-by-side and layered
rendering.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: I66701164ff02a851c13695d5409f8ad350534e69
Reviewed-on: https://chromium-review.googlesource.com/645547
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>

We're removing this bot as redundant now in Chromium.

Change-Id: I35f00bff4e62780609f869c447839aa48b74b46c
R: jmadill@chromium.org
Bug: crbug.com/743212
Reviewed-on: https://chromium-review.googlesource.com/648654Reviewed-by: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:1671
Change-Id: I58dd30d0aaffd1a776aa14a04011cbdd72181bf2
Reviewed-on: https://chromium-review.googlesource.com/648356
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:1387

Change-Id: If3fc67f99716b2e5e6a8b0ffd139a07a06cdcab8
Reviewed-on: https://chromium-review.googlesource.com/648050Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Uses more of the type info table for updates.
Also special-case clamping when uniform count == 1.

Improves the speed of the d3d11 uniform stress test by ~20%.

BUG=angleproject:1390

Change-Id: I6707c67db84c94a28b1519b0bbee5d28fe38b189
Reviewed-on: https://chromium-review.googlesource.com/646828Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=2122

Change-Id: I82bed7215142b62b321c9676972386b74a9efa92
Reviewed-on: https://chromium-review.googlesource.com/604211
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

A code-gen bug was found in 15.3.2 (VC++ 2017 Update 3.2) that causes
test failures due to one loop being skipped. Attempts to disable
optimizations for just the affected function failed so optimizations
must be disabled at the end of the source file, which seems to cause all
template functions and compiler generated functions to have
optimizations disabled. A VS bug has been filed and cross-linked.

BUG=759402

Change-Id: Ida765a47234a63bad48e6a4e910f3b82919d6be9
Reviewed-on: https://chromium-review.googlesource.com/647313Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

Currently most uniform type info is determined by switching on the
uniform type. Some values are computed from other values, which can
result in three or more switch statements plus some multiplies or
other math. This patch attempts to improve the speed by pre computing
necessary values into constant static tables.

Improves performance by about 7% in a uniform stress test.

BUG=angleproject:1390

Change-Id: I29bef259a17f6d6536171ade4950e2d712bfd39c
Reviewed-on: https://chromium-review.googlesource.com/643791Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:1651

Change-Id: I13c1a669d83098e22c0d7fb003d13beacc20c4ae
Reviewed-on: https://chromium-review.googlesource.com/623947Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

This prepares for accepting arbitrary expressions as the "this" node
of the array length method.

BUG=angleproject:2142
TEST=angle_unittests

Change-Id: I728adb6e76d2779dedbabfaeec7d096872e0d00d
Reviewed-on: https://chromium-review.googlesource.com/633945Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

This simplifies uniform management in D3D11. It will also facilitate
further optimizations.

Improves performance in a uniforms stress test by ~13% on a test
machine. (UniformsBenchmark.Run/d3d11_null_400_vec4)

BUG=angleproject:1390

Change-Id: Iba2c15d420396aa8fb4e8c451cba2b4dde7b4b77
Reviewed-on: https://chromium-review.googlesource.com/623930Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This migrates to the new generation style used in GLES2.

BUG=angleproject:1309

Change-Id: I43e9d33a0d7c5b1786452895855ff2bfbf82f139
Reviewed-on: https://chromium-review.googlesource.com/638311Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=None

Change-Id: I915c99f21dc879a8e63955cd6a3dd892d05bbf5f
Reviewed-on: https://chromium-review.googlesource.com/646590Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This moves the uniform query to the back-end. In D3D, this requires
a bit more redesign, especially for matrix uniforms.

Gives about a 10% speed improvement in the GL/NULL uniforms stress
test on Windows (UniformsBenchmark.Run/gl_null_400_vec4).

BUG=angleproject:1390

Change-Id: Idac22a77118e9e94d2f28c585e31ff0bc785ba94
Reviewed-on: https://chromium-review.googlesource.com/623929
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

Non-existent XFB varyings with the gl_ prefix used to pass linking on
the D3D11 backend. On a debug build they would cause an assert. Fix
these issues.

BUG=angleproject:2141
TEST=angle_end2end_tests

Change-Id: Iecc3d03823d02700d6b28c44d77df7a2f9e70a5b
Reviewed-on: https://chromium-review.googlesource.com/645747Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

The prefix is unnecessary now that user-defined names are prefixed in
both GLSL and HLSL output. Removing the prefix makes compiler output
a bit simpler to read.

BUG=angleproject:2038
TEST=angle_unittests

Change-Id: I9ffc508f50d6146a2d85798875c88e2c385b83fe
Reviewed-on: https://chromium-review.googlesource.com/508730Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

BUG=602737

Change-Id: I8c24985358dcd297cb437c501b7a3944e36d98de
Reviewed-on: https://chromium-review.googlesource.com/644210Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

The patch adds D3D11 as a testable platform to the multiview performance
tests.

BUG=angleproject:2062
TEST=angle_perftests

Change-Id: I2266073860fd53f546ce9a414bc343b582c77f70
Reviewed-on: https://chromium-review.googlesource.com/645849Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

Now user-defined names are prefixed by _u in GLSL output in case name
hashing is not on. Internal names such as names of temporary variables
created in AST transformations are written out as such.

This makes handling of internal function names and internal variable
names consistent. It also removes the possibility of name conflicts
between user-defined names and internal names in case name hashing is
not on. In the same vein, it makes it safe to use GLSL reserved words
that are not reserved in ESSL as variable names in case name hashing
is not on.

This also makes the GLSL output more consistent with how names are
handled in HLSL output. Name hashing code is shared between
VariableInfo and OutputGLSLBase to ensure names are handled
consistently in both. The name that's used in the shader source for a
given interface variable is written out to ShaderVariable::mappedName.

An exception needs to be made for identifiers close to the length
limit, since adding any prefix would take them over the limit. But
they can be just written out as such, since we don't have any builtins
or ANGLE internal variables that have as long names and could create a
conflict.

BUG=angleproject:2139
BUG=angleproject:2038
TEST=angle_unittests, angle_end2end_tests, WebGL conformance tests

Change-Id: Id6ed052c4fab2d091227dc9a3668083053b67a38
Reviewed-on: https://chromium-review.googlesource.com/507647
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Because the ANGLE_multiview extension uses instancing to multiply
geometry for each view, Draw* calls with an active multiview program
have to be handled in the follwing way:
1) Convert non-instanced Draw calls to their instanced versions.
2) Multiply the number of instances in an instanced Draw call by the
number of views.
The patch also applies the viewport offsets to the viewport and scissor
rectangle and propagates the computed viewports and scissors to the
D3D11 runtime.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: I8b4295c95c2cc0c1046c67e1fb1a782a46703292
Reviewed-on: https://chromium-review.googlesource.com/618331
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Relinking the active program can change its number of views and the
state has to be correspondingly adjusted.

BUG=angleproject:2062
TEST=angle_end2end_tests

Change-Id: I20102a428d7566a8cec5d81eeaa55980665812f4
Reviewed-on: https://chromium-review.googlesource.com/637994
Commit-Queue: Martin Radev <mradev@nvidia.com>
Reviewed-by: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

In some cases, when the app would call glBufferSubData to do a small
data update in an existing vertex buffer, the sync code would not
flush out the data to the native D3D11 buffer from the temporary
staging buffer.

Fix this problem by notifying the VertexArray11 class when buffer
data is updated. Note that in the future we can improve this by
using a different update notification for when the buffer data changes
and when the underlying storage changes. For now take a very broad
approach.

BUG=angleproject:1156

Change-Id: I2e0fabc97c1f1d5a14d609247e61c602e9a5a85f
Reviewed-on: https://chromium-review.googlesource.com/644208Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>