Implement classifying image enums in utilities and plumb the image
type check to Uniform objects as well. This functionality will be used
to add API support for images.

BUG=angleproject:1442

Change-Id: Idd2ee5ebacd38290aeeace1177165c740bc42c7e
Reviewed-on: https://chromium-review.googlesource.com/441585
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

The context pointer was throwing off the count by one.
Also change the type to void * for ease of use.

BUG=angleproject:1892
BUG=chromium:678870

Change-Id: Iffac667f0aa896d9247e9ffddfeb3bc447a9692f
Reviewed-on: https://chromium-review.googlesource.com/441931Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

The main purpose of this change is to fix a fuzzer bug where we would
trigger undefined behaviour calling between Chrome and ANGLE. It's
not specced how virtual function calls work if the shared objects are
not directly linked together, and ANGLE and Chrome are not linked.
Replace the old class-style API with a C dispatch table.

Follow-up work will make the Platform owned by the Display instead of
using global variables, but fixing this is a bit tricky.

BUG=angleproject:1892
BUG=chromium:678870

Change-Id: Iad188bc2e50f2b5e4a03ce0de233d686f569c705
Reviewed-on: https://chromium-review.googlesource.com/441273
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

OSX10.8 deprecates Gestalt but doesn't expose a replacement in the
public interface so we forward declare it.

BUG=angleproject:1890

Change-Id: Ie4825c9191e81a5894dca4e394275c26a3e27853
Reviewed-on: https://chromium-review.googlesource.com/441309
Commit-Queue: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

BUG=angleproject:1874

Change-Id: Id39c26b806e6a7937517235afe0ca60f5087df5b
Reviewed-on: https://chromium-review.googlesource.com/438940Reviewed-by: Zhenyao Mo <zmo@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

Fuzzer tests were capable of allocating very large chunks of memory by
calling glBufferData with a null pointer.  This sets a limit before the
NULL backend starts returning out of memory GL errors.

BUG=602737

Change-Id: Ic53ebcf999f951b96c1df82e4db57e949d03c908
Reviewed-on: https://chromium-review.googlesource.com/441184
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

These extensions are required by the passthrough command decoder.

BUG=602737

Change-Id: Ifd04fca27ae35b30e4c8ab458904530e3dd44377
Reviewed-on: https://chromium-review.googlesource.com/441204Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

This query was added in both ES3 and ANGLE_framebuffer_multisample but
the code was never updated to accept it for an ES3 context without the
extension present.

Caught by the angle passthrough command decoder fuzzer.

BUG=602737

Change-Id: Ic9a8c67747777ef48e1dd227110482998fee75ce
Reviewed-on: https://chromium-review.googlesource.com/440651Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

glCompressedTexImage3D was missing a check that the format was valid
before trying to compute the expected data size.  This lead to an
assertion failure in the size computation code.

BUG=602737

Change-Id: I74e6ced082e78396a0af92317fb1c0098299a026
Reviewed-on: https://chromium-review.googlesource.com/441185Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

This makes ResourceManagers use more common code. It uses static
polymorphism to implement the various typed resource managers,
and unifies the FramebufferManager into a TypedResourceManager.
The lack of virtual calls and inlining in the header should keep
performance high.

This will make the upcoming refactor to add a destroy hook
for Vulkan objects simpler to implement.

TEST=manual performance testing of BindingPerf.
BUG=angleproject:1684
BUG=angleproject:1639

Change-Id: Ic2f102196020d1005e63a1b2cf0867577fc7c885
Reviewed-on: https://chromium-review.googlesource.com/439488
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

Set the SH_INIT_GL_POSITION bit in WebGL compatibility mode so gl_Position
gets zeroed out.  Add corresponding test.

BUG=angleproject:1825

Change-Id: I47107804abaa83d6aee5cd46e6b69b532c08e6e8
Reviewed-on: https://chromium-review.googlesource.com/439784Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Frank Henigman <fjhenigman@chromium.org>

BUG=angleproject:1639

Change-Id: If9140142ebce89f33921c13d9d212c17d1894162
Reviewed-on: https://chromium-review.googlesource.com/437618Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

This is needed in preparation of implementing DispatchCompute, which
will need to set some of the same state as draw calls, but not
everything.

Also move setTextureCubemapSeamlessEnabled to onMakeCurrent, this way
it doesn't need to be called on every draw call. Overall this patch
doesn't have a measurable performance impact.

BUG=angleproject:1442
TEST=angle_perftests, angle_end2end_tests

Change-Id: I9958bfe7bf895b3cd805e3146a8f74d28ff2835e
Reviewed-on: https://chromium-review.googlesource.com/438349Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>

Suspecting driver bug on Win10 Debug Intel HD 530.

BUG=angleproject:1879

Change-Id: I362ffd83694c7e222826cd3766fa57f785f3ea35
Reviewed-on: https://chromium-review.googlesource.com/438936Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

When copying to and from the same texture, we need to reject only
the feedback loops formed with the same levels of the texture -
copying between different unrelated layers and levels is fine.

This change also fixes a couple bugs in our D3D11 CopyTexSubImage3D
implementation. We were missing some "!" operators, and we actually
would hit an ASSERT when trying to blit from a level of a 3D texture.

BUG=angleproject:1685

Change-Id: Id715bebafe8336cf8bb95d0d06275a8b95e522e1
Reviewed-on: https://chromium-review.googlesource.com/425494
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

Tests started failing after directing ERR to Platform

BUG=angleproject:1660, angleproject:1878

Change-Id: I3e6d99bee266c0e7fa541c6e19be7253dcc01cdf
Reviewed-on: https://chromium-review.googlesource.com/438819Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

Passing this through the chain will allow us to have access to the
platform methods, as well as be useful in the future for Vulkan.

BUG=angleproject:1660

Change-Id: I819984fceeb5a2a299aa54e59ef3b428f5f9c91f
Reviewed-on: https://chromium-review.googlesource.com/438684Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Looks like a driver bug on Intel HD 530,
uncovered by directing ERR to Platfrom.

BUG=angleproject:1660, angleproject:1877

Change-Id: I0b8cd1b2bacb2a95c7667be783801686bc8fdb7b
Reviewed-on: https://chromium-review.googlesource.com/438679Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>

BUG=angleproject:1795

Change-Id: Ic25805f99154395c9af28197ceb1218f9c5c3138
Reviewed-on: https://chromium-review.googlesource.com/434457Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

BUG=angleproject:1685

Change-Id: I6019555af69800f07862a0038fd0c25a9bb5aab8
Reviewed-on: https://chromium-review.googlesource.com/425493
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

The absence of the extension should not prevent depth/stencil textures
in FBOs in ES3, since they are core. In fact it shouldn't be possible
to make depth/stencil attachments at all without support, so the check
shouldn't be necessary.

BUG=angleproject:1828

Change-Id: Ibec1e95c78b79cec9654f06ba11beec447017f1c
Reviewed-on: https://chromium-review.googlesource.com/438604
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>

This allows us to retrieve Device handles for Vulkan, and could allow
us to avoid storing some pointers for other back-ends.

BUG=angleproject:1684

Change-Id: Iccd5b97cf1c3a332d62a38ad0fcf21c352270439
Reviewed-on: https://chromium-review.googlesource.com/426400
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

This ASSERT was benign and can be turned into an error check. The
pattern in question is to initialize an array with another array
as the first argument, but dereferencing the array with "." instead
of "[]". This would trip up our error handling.

BUG=chromium:662702

Change-Id: Ie0e44af7b9d1a66cad03cefae9bf931f8e216cd9
Reviewed-on: https://chromium-review.googlesource.com/437599
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

This allows the Vulkan back-end to avoid storing a reference to
the VkDevice. This will extend to all the Vulkan object handle wrapper
types.

BUG=angleproject:1684

Change-Id: I3a98e94bc171ca27f225ce57996c3fdf9581e6e1
Reviewed-on: https://chromium-review.googlesource.com/424229
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

If src/tests/tests.sln is built instead of gyp/ANGLE.sln, the test
executable is placed under src/tests/Release*.

TEST=perf_test_runner on Windows

Change-Id: I72ce55e2caa8270bef089c073840d5e2d6caa14d
Reviewed-on: https://chromium-review.googlesource.com/438348
Commit-Queue: Olli Etuaho <oetuaho@nvidia.com>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

Affects glCompressedTexSubImage and glCopyCompressedTexSubImage calls.

BUG=668223

Change-Id: Ie71faa1fa7dac12cec51a2e29e0ce212ac54e411
Reviewed-on: https://chromium-review.googlesource.com/437605
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

BUG=angleproject:1651

Change-Id: I710df14c2bfa55db2fb9b9715001756d1c1a8f1b
Reviewed-on: https://chromium-review.googlesource.com/434763Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

All logging should be done via ERR() and WARN(),
which call angle::Platform's logError and logWarning,
if there is current Platform which supports logging.
Otherwise, ERR() is directed to std::cerr.

Misc fixes to keep tests passing.

BUG=angleproject:1660, angleproject:1644

Change-Id: I2bca33a021537185d0c236a3083789af3236b5f3
Reviewed-on: https://chromium-review.googlesource.com/434188
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

ScheduleYield was always called after GetData regardless of it's result.  This
meant that we would successfully get the data but still sleep the thread

BUG=667979

Change-Id: Iecba0ae5cacef64a1080bbd89ce352bc4f281359
Reviewed-on: https://chromium-review.googlesource.com/437636Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>

BUG=angleproject:1580

Change-Id: Iceaed896db22dc9eefa3f1bee7d6142fcfb20368
Reviewed-on: https://chromium-review.googlesource.com/412267
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Corentin Wallez <cwallez@chromium.org>

For undefined behavior when point rendering without writing to gl_PointSize.
Even the comment used to say "This is stictly speaking not an error".

BUG=angleproject:1660

Change-Id: Ic8b33689151c9812592dd6b22c83a3355b7d6756
Reviewed-on: https://chromium-review.googlesource.com/437746Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Bundle an error message about whole buffer copies inside gl::Error
instead of delivering it to the Platform.

BUG=angleproject:1660

Change-Id: I13205b74529fbe40af05e7c1896c6315486e90e0
Reviewed-on: https://chromium-review.googlesource.com/437628
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>

This prevents a spurious D3D runtime warning, which was clogging up
some of our test output.

BUG=angleproject:1660

Change-Id: I1a747ba7532323b989dbed1ee7c78b3b457768a6
Reviewed-on: https://chromium-review.googlesource.com/437724Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

This prevents a multiply defined symbol error in ninja. Unforunately
I'm not aware of how to do the equivalent of GN's data_deps in GYP.

BUG=angleproject:1709

Change-Id: I9c616337c668a5fa652b6c084b2a64d83d526e4a
Reviewed-on: https://chromium-review.googlesource.com/437646Reviewed-by: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Noticed this while investigating the tracing code.

BUG=chromium:678870

Change-Id: I016c8532044b8f105ddda67f7e06c6b5ed9c61fb
Reviewed-on: https://chromium-review.googlesource.com/437647Reviewed-by: Geoff Lang <geofflang@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Generate GL_INVALID_VALUE when setting a sampler uniform with a value
outside the range [0, GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS).
Add a test for the new behavior.
Remove TextureLimitsTest.DrawWithTexturePastMaximum, which is made
obsolete by the new test.

BUG=angleproject:1711

Change-Id: I9a4ea13b8cb47742816476689bd3932ce267fd0a
Reviewed-on: https://chromium-review.googlesource.com/430196
Commit-Queue: Frank Henigman <fjhenigman@chromium.org>
Commit-Queue: Geoff Lang <geofflang@chromium.org>
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Reviewed-by: Geoff Lang <geofflang@chromium.org>

BUG=angleproject:1866

Change-Id: I41f1e842b7f64e541f98d466d2b0875412161274
Reviewed-on: https://chromium-review.googlesource.com/437285Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Corentin Wallez <cwallez@chromium.org>

This reverts commit 786ad387.

Reason for revert: 
Seems to have a bug with binding the BlendStates, causing a crash on
Intel.

https://luci-milo.appspot.com/buildbot/chromium.gpu.fyi/Win10%20Release%20%28Intel%20HD%20530%29/141

Failing WebGL 2 tests
WebglConformance_conformance2_reading_read_pixels_from_fbo_test
WebglConformance_deqp_functional_gles3_readpixel

Also generates D3D11 runtime warnings:
D3D11 ERROR: ID3D11DeviceContext::Draw: The renderTarget bound to slot 0 has a format (R8_UINT) that does not support blending. The Pixel Shader output signature indicates this output could be written, and the Blend State indicates blending is enabled for this slot. [ EXECUTION ERROR #376: DEVICE_DRAW_OM_RENDER_TARGET_DOES_NOT_SUPPORT_BLENDING]

BUG=angleproject:1632
BUG=chromium:688419

Original change's description:
> D3D11: Clean up blendState code.
> 
> Masked Clear Draw Changes:
> - Use universal blendstate object
> - Eliminate blendState cache for masked clears
> - Use rasterState and scissor rect for scissoring instead of adjusting vertex positions
> - VB contains only static position data (per vertex color removed)
> - Clear color(s) and depth clear values now passed in using a constant buffer
> - MultiColorclear shader used for float clears to workaround alpha rounding issues
> - Update shader compile script and shader source and bytecode headers
> - Remove unused shaders (source and bytecode headers)
> - Use com pointers where possible for D3D11 objects
> 
> BUG=angleproject:1632
> 
> Change-Id: I98e38451bd453f53b772fe93ec9dcceb4196ea58
> Reviewed-on: https://chromium-review.googlesource.com/413736
> Reviewed-by: Geoff Lang <geofflang@chromium.org>
> Reviewed-by: Jamie Madill <jmadill@chromium.org>
> Commit-Queue: Shahmeer Esmail <shahmeer.esmail@intel.com>
> Commit-Queue: Jamie Madill <jmadill@chromium.org>
> 

TBR=geofflang@chromium.org,jmadill@chromium.org,shahmeer.esmail@intel.com
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=angleproject:1632

Change-Id: Iea537505d8cce7241edaba1f1d9f404abb1d9a10
Reviewed-on: https://chromium-review.googlesource.com/437306Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Although the functionality for these textures is unimplemented, we
still create zero textures for multisample in D3D11. Thus the
constructor and destructor should be marked as implemented.

BUG=angleproject:1590
BUG=angleproject:1660

Change-Id: I44b51dd4e89d492f98c134b8f7dfd62dffc97015
Reviewed-on: https://chromium-review.googlesource.com/437304Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>

Moving it from SetUp to the constructor allows the swapping to work
even if the test decides to not call the SetUp base function. This
fixes the Vulkan extension test, which skips normal test init.

This requires us to change an ERR invocation into a print.

BUG=angleproject:1810

Change-Id: I3be4bdb6df8c02fc1c9bbc7834ba485054272b8e
Reviewed-on: https://chromium-review.googlesource.com/437284Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>