With dynamic shaders we may have multiple shader executables per
program binary. We must store the uniforms outside the executable,
in the program binary, to be consistent between variations.

Change-Id: I1b29a5d78c72dede8562d4878569b609536ba074
Reviewed-on: https://chromium-review.googlesource.com/183586Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

With dynamic shaders we may have multiple shader executables per
program binary. We must store the uniforms outside the executable,
in the program binary, to be consistent between variations.

Change-Id: I1bec83dfb83ee9bb422448b157338a46e8e7ba56
Reviewed-on: https://chromium-review.googlesource.com/183585Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

In preparation for transform feedback, BufferStorage11 needs to be able to
handle a non-staging buffer being updated.  Each D3D11 buffer now has an
incrementing data revision associated with it so the most up-to-date
buffer is always known.

Staging buffers are now represented like any other DirectBuffer.

Change-Id: I6e881867cb2bd02d600213d08cce3ebba316c525
Reviewed-on: https://chromium-review.googlesource.com/184395Tested-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

With certain selection statements with a vertex input in the
condition and side-effects in the else-block, we'd run in to
a D3D9 compiler bug which would cause incorrect results.

We can work around this bug in D3D9 by selectively rewriting
these statements to use an 'else if' clause instead of 'else'.

BUG=322794

Change-Id: I93c96fb201ff4959c00d9a36321faac7e0343278
Reviewed-on: https://chromium-review.googlesource.com/184681Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I5e5aca5cda2bf1a96db05d594c64970ebf54fd29
Reviewed-on: https://chromium-review.googlesource.com/184572Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Change-Id: I83e3091243be442eafa4d16e4a4ce247a326c493
Reviewed-on: https://chromium-review.googlesource.com/184570Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Change-Id: I28a7654d152865c236831636f00554a71784667b
Reviewed-on: https://chromium-review.googlesource.com/184394Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Also add a few missing DXGI formats related to vertex inputs.

Change-Id: Ia39234a164000d11eef3806ba4f06b6426f1cbb7
Reviewed-on: https://chromium-review.googlesource.com/183584Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

The GL mapping depends on context client version (for certain depth
textures), but the DXGI base format info is consistent across
versions. Also add DXGI component type information to the table,
and add a few missing formats related to vertex inputs.

Change-Id: I2d1a8ca92b1cb39f4592d94d7d435ae0ada1de65
Reviewed-on: https://chromium-review.googlesource.com/183583Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

With dynamic shaders, we can have multiple vertex shaders in one
program binary. Hence the binary serial by itself isn't enough to
determine when we need to apply new shaders to D3D.

Change-Id: I5a8b8616d353b506790ffc809e2d459bfc2b831d
Reviewed-on: https://chromium-review.googlesource.com/184370Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

By shuffling some parameters and calls around we can reduce the
number of function calls we run for texture operations.

Change-Id: If6d14cb6dff668cdfcb2f1fe6705e0d488dbc03f
Reviewed-on: https://chromium-review.googlesource.com/183582Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

This enum is only availabe in GLES 3. We were failing a dEQP state
query test (fbo.draw_framebuffer_default_framebuffer)

Change-Id: I4cfb07b8597899561b8cc7956a9eb4cd4b0a99a8
Reviewed-on: https://chromium-review.googlesource.com/181782Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

The gyp files already define NOMINMAX and WIN32_LEAN_AND_MEAN.  windows.h
is also included via the D3D includes.

This ends up speeding up the build process significantly by not having
windows.h included in nearly every file.

Also fixes issues on non-windows systems by not having stdarg.h included
in debug.cpp.

Change-Id: Id32d9aac37a5a29df832a062dd8024302a798a1c
Reviewed-on: https://chromium-review.googlesource.com/183251Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

We were not checking for an invalid renderbuffer when attaching
to a framebuffer.

Change-Id: Ie29dca1833cc8c2b974f746d51dbee848f52a741
Reviewed-on: https://chromium-review.googlesource.com/181558Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I347583ae1adcaaae9feeb1276f6150ceb8739885
Reviewed-on: https://chromium-review.googlesource.com/181557Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Consolidate a lot of duplicated code into helper methods. Also
move GetTargetTexture to the context class.

Change-Id: I8e3990a5e3c1976cb533941d0a0946466accec30
Reviewed-on: https://chromium-review.googlesource.com/181556Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I424cb538808ec64137c85af3bbe8742043e1c573
Reviewed-on: https://chromium-review.googlesource.com/183260Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Changed ANGLE_DISABLE_* to ANGLE_ENABLE_* for perf and tracing defines so
they are disabled by default.  Updated the gyp files to only turn on perf
by default for windows debug builds.

Change-Id: I71706674e6d12fbf4208acc8f100d963b82c7674
Reviewed-on: https://chromium-review.googlesource.com/183250Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

D3D11 can only mask 8 bits of a stencil reference value, so the
upper bits are ignored. GL, however, expects the reference value
to be able to scale to 32-bits.

We will expect to fail certain edge cases where the reference value,
and mask, both use particular 9+ bit values.

Change-Id: I8c7451270d9e40310f4154955d38e51bbf7de4df
Reviewed-on: https://chromium-review.googlesource.com/181555Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

D3D11 supports sampling other channel sizes of backing object
and declaring them as 4-vectors in HLSL. This allows us to
simplify the logic for fast unpack shaders significantly.

Source:
http://msdn.microsoft.com/en-us/library/windows/desktop/bb509700(v=vs.85).aspx
(Default Values for Missing Components in a Texture)

Change-Id: I7f645372f266f57bd94cdb1c795f1d30bf2a60a4
Reviewed-on: https://chromium-review.googlesource.com/181901Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

We would sometimes attempt to use a NULL staging texture when
initializing a zero size texture when using unpack buffers.

Change-Id: I2e03141c19d9c47eeffb0d56c03bcfe476a9e48d
Reviewed-on: https://chromium-review.googlesource.com/181781Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

A breaking change in VS2013 necessitates we include <algorithm>
to use std::min/max.

BUG=angle:547

Change-Id: Ia1b21e61d22148ab6d268b34dbfe9a707ca3e59d
Reviewed-on: https://chromium-review.googlesource.com/182512Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

We could overflow when casting from max unsigned int to integer, and
return a negative mask.

See dEQP functional.state_query.stencil_value_mask_getfloat

Change-Id: I59189a40b662ad9d8ded79cdfeded923b98f0b85
Reviewed-on: https://chromium-review.googlesource.com/181780Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: Idb4c2ca318c84433f2b06bd4f0c66045f43e56de
Reviewed-on: https://chromium-review.googlesource.com/182191Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Change-Id: If04bc249b9da8306896c89c28bb057c2cf4da511
Reviewed-on: https://chromium-review.googlesource.com/181771Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Change-Id: I76d568ce7b1acb84aba99adf8e6ca921e0933ac4
Reviewed-on: https://chromium-review.googlesource.com/182190Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Works around performance regressions when index and vertex buffer bind flags are used on the same buffer.

Change-Id: I28bc0d3147c6bd70cec507f20e41d97ec4cc45a5
Reviewed-on: https://chromium-review.googlesource.com/181911Tested-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

Change-Id: I8cf650073ead5b52bac44e31c8c3c894bc31e01c
Reviewed-on: https://chromium-review.googlesource.com/182320Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Consolidate a lot of the code duplication in GetIntegerv, GetFloatv,
GetBooleanv and GetInteger64v.

BUG=angle:540

Change-Id: I7b4a5803ea76e5eeb0ebf25c5caadec4c647b485
Reviewed-on: https://chromium-review.googlesource.com/180971
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I335c9346ac268beab25184e2b18ccfdd0b15c7b1
Reviewed-on: https://chromium-review.googlesource.com/182150Reviewed-by: Jamie Madill <jmadill@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

Since swizzled textures have multiple SRVs and the same serial,
it was possible that the bound SRV would not be updated if the
texture became swizzled while already bound.

Change-Id: Id06e82f4002aa7f7661c7a69243692aa65a83b48
Reviewed-on: https://chromium-review.googlesource.com/181772Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

We were incorrectly calling Texture::getBaseLevelDepth() which was
correctly reporting the number of 3D slices of the texture (1).

Also rename Texture2DArray::getDepth to getLayers for clarity.

Change-Id: I1968cbcb0a3528bb76ec3e4ea4148e5dc0f302c7
Reviewed-on: https://chromium-review.googlesource.com/181553Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

Change-Id: I2390ace1fe5297f9fcad0c873c55dae564a09c43
Reviewed-on: https://chromium-review.googlesource.com/181773Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Geoff Lang <geofflang@chromium.org>

BUG=angle:536

Change-Id: I72287f5ce9736cd48fc241022e09446db717cf92
Reviewed-on: https://chromium-review.googlesource.com/181770Tested-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

When doing a masked clear with only a depth-stencil buffer, we would
look for the zero-th element of an empty vector.

Change-Id: I17d61cf582c6e91fd46f8a8e13e35c6d32dfa109
Reviewed-on: https://chromium-review.googlesource.com/181554Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

BUG=angle:541

Change-Id: I63dd47c40d693724aa6bed93e9967e3b1f8535bd
Reviewed-on: https://chromium-review.googlesource.com/181521Tested-by: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

BUG=angle:428

Change-Id: I282a6bf138b4e2a2c17baf428cbe1e636f43bdd6
Reviewed-on: https://chromium-review.googlesource.com/181640Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Nicolas Capens <nicolascapens@chromium.org>
Commit-Queue: Nicolas Capens <nicolascapens@chromium.org>
Tested-by: Shannon Woods <shannonwoods@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/181670
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

BUG=angle:541

Change-Id: Ied96edd073017a507369319a781b8ed23f60790b
Reviewed-on: https://chromium-review.googlesource.com/181520Tested-by: Nicolas Capens <nicolascapens@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>

The GL spec states that by default a vertex attribute has a size of 4.

Change-Id: Ic4c977fb936cc0b232ed4742eeb061d6ba6843be
Reviewed-on: https://chromium-review.googlesource.com/181480Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Commit-Queue: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>

We defautled to zero max bit depth, which breaks the swizzle format lookup.

Change-Id: I4d1cd8bcfb79db3460950fa22630c259f90fa432
Reviewed-on: https://chromium-review.googlesource.com/181481Reviewed-by: Jamie Madill <jmadill@google.com>
Commit-Queue: Jamie Madill <jmadill@google.com>
Reviewed-by: Shannon Woods <shannonwoods@chromium.org>
Tested-by: Jamie Madill <jmadill@chromium.org>