Previously, the type of the return value of all function calls was set to
the type of the return value in the function signature. This did not
carry precision information.

This patch changes this so that the return value precision is set
correctly for built-in functions. For single-argument math functions, it
mostly depends on that addUnaryMath sets the type of the return value to
be the same as the type of the operand. The type is replaced but the
precision information from the operand type is retained when needed. For
multi-argument math functions, precision is determined based on all the
nodes in the aggregate after the type has been set. For texture
functions, the precision is set according the sampler type as per ESSL
1.0 spec. For textureSize, the precision is always highp as per ESSL 3.0
spec.

BUG=angle:787

Change-Id: I48448e3ffe38656b91177dee9b60dd07a03cd095
Reviewed-on: https://chromium-review.googlesource.com/224951Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Nicolas Capens <capn@chromium.org>
Tested-by: Olli Etuaho <oetuaho@nvidia.com>

The Debug and Release configurations are 32-bit only.  The Debug_Base and
Release_Base configurations should be used in these cases.

In the future we should not rely on defines being placed in these
configurations because if ANGLE is included in another project, they may
not have the same configuration names and the defines will not be added.
To combat this, always make sure that there are appropriate #ifdef's in
the headers and don't add defines that would result in poor performance if
they do not exist.

BUG=angle:772

Change-Id: I2b26ffb934f8b3d5e6e556a1e87c1c5ee209c430
Reviewed-on: https://chromium-review.googlesource.com/225051Tested-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

We only ever call this method inside the D3D Renderer, so we can
downcast to BufferD3D and call getData on the D3D-specific type.

Leave a FIXME for handling index range validation, which will
need a CPU-side data cache.

Change-Id: Iaf71bc8055869a8561777b6b36f67e376a1d0b81
Reviewed-on: https://chromium-review.googlesource.com/224654Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:792

Change-Id: I32f32422232bd4cb04c8a70005cb51482224bf3e
Reviewed-on: https://chromium-review.googlesource.com/224655Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

With the invariant handling changes, we're also exporting gl_Position
and other builtin variables from the translator where we were not
previously. This was causing an assert in the XFB code, where we
would inadvertently assign a register to gl_Position, and trip up
logic further in the varying handling logic.

BUG=angle:808

Change-Id: I674902da75c572e969f9516033a85c7a7170d7d1
Reviewed-on: https://chromium-review.googlesource.com/225261Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

This will ensure our currently set pixel and vertex SRVs match with
the underlying D3D state, and reduce the need for resetting them as
often.

BUG=angle:756
BUG=417424

Change-Id: I42e9e25182b279871ac4cdf8da5e2ab62c8cf71f
Reviewed-on: https://chromium-review.googlesource.com/225052Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:520

Change-Id: I5db70189d95babef14d48548054af4c7ff2bfc47
Reviewed-on: https://chromium-review.googlesource.com/219334Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:520

Change-Id: If1c1f7519a84900f594701b6298b64ebf8798073
Reviewed-on: https://chromium-review.googlesource.com/219333Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

This file is a duplicate of angle.gyp, and we can remove it now that
Chrome and Blink are finished with it.

BUG=angle:460

Change-Id: Ida68c8343921ebe9a37b5bcd982e114c91febc3d
Reviewed-on: https://chromium-review.googlesource.com/225053Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

The list of D3D compiler flags and macros used if
compilation of a shader succeeds is added to the
shader debug output.

BUG=angle:751

Change-Id: Ie84a586010787fe8d9c0dedfe9208f881058a588
Reviewed-on: https://chromium-review.googlesource.com/224970Tested-by: Tibor Ouden, den <tibordenouden@gmail.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

If a previous window was not cleaned up properly due to a crash in a
test, the subsequent windows can fail to be created because the class is
not destroyed.  Work around this by always creating a unique class before
creating the window.

Change-Id: Ied6b2818ef03fa12b07111d8204c3c1a6a5bd5ac
Reviewed-on: https://chromium-review.googlesource.com/225080Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:807
TEST=angle_unittests

Change-Id: Ifa4d713deeb25d52a7aafc362a7e4630024fd511
Reviewed-on: https://chromium-review.googlesource.com/225004Tested-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

Change-Id: Ifcff7cba0c16775fdc99812539416ad07d88bf5a
Reviewed-on: https://chromium-review.googlesource.com/225050Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:805
TEST=https://www.khronos.org/registry/webgl/sdk/tests/conformance/glsl/misc/expression-list-in-declarator-initializer.html

Change-Id: I94fbdd0824dbe73034f8bf022aa53627582ad48b
Reviewed-on: https://chromium-review.googlesource.com/224813Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Zhenyao Mo <zmo@chromium.org>

This was broken sometime in the CollectVariables refactor.

BUG=angle:803

Change-Id: Iaa09449f02290c4547f87c1560465dc8998d957c
Reviewed-on: https://chromium-review.googlesource.com/224104Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

BUG=angleproject:804

Change-Id: I4d68f9283299e9fb6b907a6406c37f92f8b08dbe
Reviewed-on: https://chromium-review.googlesource.com/224106Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:733

Change-Id: Iab9a3fbc8aa1e05f4d969016ab28bac868089487
Reviewed-on: https://chromium-review.googlesource.com/224482Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:520

Change-Id: Id18e93b440da64360a6845a42a2664ae531b06f9
Reviewed-on: https://chromium-review.googlesource.com/218769Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

This helps with debugging precision propagation issues.

BUG=angle:787

Change-Id: Ia969481c8d933455cdafef4ae25c6ab3946bc1c2
Reviewed-on: https://chromium-review.googlesource.com/224281Reviewed-by: Nicolas Capens <capn@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Shannon Woods <shannonwoods@chromium.org>

Change-Id: I0cb4262552852ccbc772ca828a0faa7d3e63784b
Reviewed-on: https://chromium-review.googlesource.com/224483Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Chromium builds a std::map with ShBuiltInResources as part of the key.
Comparator for == and < are needed for the map implementation.

Currently Chromium uses memcmp as the comparator. Padding
in ShBuiltInResources causes uninitialized reads.

Fix this by clearing the padding with memset during ShBuiltInResources
initialization.

Change-Id: I78aa3c59ce165503831aa2a67c96cf8af316c152
Reviewed-on: https://chromium-review.googlesource.com/223431Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Shannon Woods <shannonwoods@chromium.org>

Use a helper method isValidIndex to preserve ASSERT checks.

Refactoring patch only.

BUG=angle:741

Change-Id: Ie19fa21db51cd0239a3b391de362584a9fbab2df
Reviewed-on: https://chromium-review.googlesource.com/222268Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

We were being too permissive in some copyImage methods when checking
if we can create a render target or not. This would lead us to trying
to use an inconsistent TexStorage in some cases.

Note that we do need to create inconsistent TexStorage in cases where
we are doing a FBO copy *from* a mipmap-incomplete texture attachment.

BUG=angle:780

Change-Id: I5a849b5d8c53713e38cb5f5052a8bb88b3f56260
Reviewed-on: https://chromium-review.googlesource.com/224480Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

Change-Id: I308956a700df4d440573fbae89e67cab0f2cf482
Reviewed-on: https://chromium-review.googlesource.com/224653Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

* Fix a bug in PreProcessor for STDGL pragma.
* Record all invariant settings and set them in ShaderVariable.
* Write #pragma STDGL invariant(all) in GL

BUG=angle:776
TEST=https://www.khronos.org/registry/webgl/sdk/tests/conformance/glsl/misc/shaders-with-invariance.html

Change-Id: Ie28b75480deed79f0c9f26e3b98f1778d1290182
Reviewed-on: https://chromium-review.googlesource.com/223610Tested-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

We can stop exposing this method on the GL-side, now that we only
use it on the D3D back-end.

BUG=angle:781

Change-Id: I3a9ecc277dcd21147881b8780849782f68b3e62e
Reviewed-on: https://chromium-review.googlesource.com/222924Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>

This method returned a sampler state as normal, with the base level
automatically offset by the one or two levels we use in the D3D
workaround for small compressed textures. Since this is D3D-only,
we can move all of the logic into the D3D classes themselves.

BUG=angle:781

Change-Id: Ie0a60877efebb41f02f4e57625a44e5fb5cce074
Reviewed-on: https://chromium-review.googlesource.com/222923Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>

It's reasonable to use the serial logic on both the GL and all other
translated Renderers, since we check the logic on the GL-side.
Serials give us a way to identify Textures that aren't raw pointers
and aren't "id". "id" has issues when Textures are deleted, then
re-allocated with the same value.

BUG=angle:781

Change-Id: I1a2a8b6f4ea3db915574c957c143a81e0210ec7d
Reviewed-on: https://chromium-review.googlesource.com/222922Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>

Instead of storing a binary flag indicating if we've created
immutable storage, store a level count. This indicates if or if
not we've created immutable storage, as well as avoiding the
call to the native storage to return the level count.

BUG=angle:781

Change-Id: I08a6bd086d00ddf3255262a7711628be5212b4b7
Reviewed-on: https://chromium-review.googlesource.com/222921Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Brandon Jones <bajones@chromium.org>

Change-Id: I6dd7fef72a73572d4a3deda7ce36a11da3a75c81
Reviewed-on: https://chromium-review.googlesource.com/224366Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:520

Change-Id: I5e51fec38c513419beea5ff612a6ec2fb0e81a26
Reviewed-on: https://chromium-review.googlesource.com/218768Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:520

Change-Id: Ia35103059526aff2fc1b29a160d8471f4b5cbd42
Reviewed-on: https://chromium-review.googlesource.com/223107Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Found while working on BUG=angle:751

BUG=angle:751

Change-Id: I8a5c5383f88ecb379395d1a23e36148145ebe34d
Reviewed-on: https://chromium-review.googlesource.com/224090Tested-by: Tibor Ouden, den <tibordenouden@gmail.com>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

BUG=angle:520

Change-Id: If1f4f71972b669704eff70b5f60927d8e6ac07b3
Reviewed-on: https://chromium-review.googlesource.com/218767Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:741

Change-Id: I67380748ee875cc2dea61eb82a9b0d600144dc85
Reviewed-on: https://chromium-review.googlesource.com/222267Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

Causing regressions in the build: http://build.chromium.org/p/chromium.gpu.fyi/builders/GPU%20Mac%20Builder/builds/20182

This reverts commit 756aebfc.

Change-Id: I2f4bdb5aeb429c9bbc5e655a1761704f33737841
Reviewed-on: https://chromium-review.googlesource.com/224221Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I34e443b1e194800460e441ac6cee42cf68430564
Reviewed-on: https://chromium-review.googlesource.com/224302Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

D3D11 requires us to NULL the update region parameter when updating
depth stencil textures. For these textures, we can't always use the
subdata workaround, so disable it entirely for these textures.

BUG=angle:729
BUG=365078

Change-Id: I44258dd1b8937b1aebcb3a73de835698805537e0
Reviewed-on: https://chromium-review.googlesource.com/222911Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

We were checking readTarget twice, instead of both read and write.

BUG=angle:800

Change-Id: I4096a9c4e0fdec9c77bfc25250923c188d2b43db
Reviewed-on: https://chromium-review.googlesource.com/224102Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Our previous change to use static_assert leads the new wave of c++11
features.

BUG=

Change-Id: Id6d382475daa7029109dab1017a59b5ef07c18df
Reviewed-on: https://chromium-review.googlesource.com/221267Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>