This makes sure the same varying/uniform variables maps to the unique name in vertex/fragment shader.

BUG=
TEST=webgl conformance tests
Review URL: https://codereview.appspot.com/5556065

git-svn-id: https://angleproject.googlecode.com/svn/trunk@950 736b8ea6-26fd-11df-bfd4-992fa37f6226

MapLongVariableNames inherits from TIntermTraverser, and TIntermTraverser uses ANGLE's memory allocator, thus the memory is released per compilation.  Our design is for MapLongVariableNames to be a singleton across all compilations, thus, this is not working.

BUG=
TEST=
TBR=kbr
Review URL: https://codereview.appspot.com/5556053

git-svn-id: https://angleproject.googlecode.com/svn/trunk@949 736b8ea6-26fd-11df-bfd4-992fa37f6226

TRAC #19335
Issue=271
Instead, the texture maintains an internal refcount which causes the pointer to behave
kind of like a "weak" shared pointer-- its reference won't prevent the renderbuffer from
being deleted, but if the renderbuffer's refcount drops to zero, the texture should set
its pointer to the renderbuffer to null so that a new one can be created at the next
getRenderbuffer call.

Signed-off-by: Nicolas Capens
Signed-off-by: Daniel Koch

git-svn-id: https://angleproject.googlecode.com/svn/trunk@948 736b8ea6-26fd-11df-bfd4-992fa37f6226

TRAC #19335
Issue=271

Signed-off-by: Nicolas Capens
Signed-off-by: Daniel Koch

git-svn-id: https://angleproject.googlecode.com/svn/trunk@947 736b8ea6-26fd-11df-bfd4-992fa37f6226

The GLSL to HLSL translator work is not done yet so the extension is disabled in Shader.cpp.
Review URL: https://codereview.appspot.com/5530081

git-svn-id: https://angleproject.googlecode.com/svn/trunk@946 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@945 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@944 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@943 736b8ea6-26fd-11df-bfd4-992fa37f6226

For example: varying variables, uniforms.

This CL makes MapLongVariableNames a ref-counted singleton and therefore, the map is shared by all shaders.

Also, function/variable names changes from Varying to Global because uniforms also need to be consistent, not just varying variables.

ANGLEBUG=279
TEST=webgl conformance tests, especially invalid-passed-params.html and glsl-long-variable-names.html
Review URL: http://codereview.appspot.com/5539046

git-svn-id: https://angleproject.googlecode.com/svn/trunk@942 736b8ea6-26fd-11df-bfd4-992fa37f6226

This is a workaround for an Intel bug.

The crash looks like this:

0x5f816c53	 [d3d9.dll	 + 0x00036c53]	CQuery::~CQuery()
0x5f816bec	 [d3d9.dll	 + 0x00036bec]	CQuery::`vector deleting
destructor'(unsigned int)
0x5f7e8129	 [d3d9.dll	 + 0x00008129]	CBaseObject::~CBaseObject()
0x5f9e19c7	 [libglesv2.dll	 + 0x000319c7]	gl::Fence::`vector deleting
destructor'(unsigned int)
0x5f9de78e	 [libglesv2.dll	 - context.cpp:975]	gl::Context::deleteFence(unsigned
int)
0x5f9e1491	 [libglesv2.dll	 - context.cpp:198]	gl::Context::~Context()
0x5f9e182f	 [libglesv2.dll	 - context.cpp:3936]	glDestroyContext
0x717654ec	 [libegl.dll	 -
display.cpp:749]	egl::Display::destroyContext(gl::Context *)
0x7176a3da	 [libegl.dll	 - libegl.cpp:907]	eglDestroyContext
0x64fbaf33	 [chrome.dll	 - gl_context_egl.cc:75]	gfx::GLContextEGL::Destroy()

The vendor ID is always 8086 (Intel). Not an XP issue - it's happening on Win
7.

With this change, D3D queries are only released when the display is destroyed or reset or if a very high number of D3D queries have been allocated.

Tested by stepping exercising the NV_fence entry points in a debugger.
Review URL: http://codereview.appspot.com/5534065

git-svn-id: https://angleproject.googlecode.com/svn/trunk@941 736b8ea6-26fd-11df-bfd4-992fa37f6226

This is a workaround for an Intel bug.

The crash looks like this:

0x5f816c53	 [d3d9.dll	 + 0x00036c53]	CQuery::~CQuery()
0x5f816bec	 [d3d9.dll	 + 0x00036bec]	CQuery::`vector deleting
destructor'(unsigned int)
0x5f7e8129	 [d3d9.dll	 + 0x00008129]	CBaseObject::~CBaseObject()
0x5f9e19c7	 [libglesv2.dll	 + 0x000319c7]	gl::Fence::`vector deleting
destructor'(unsigned int)
0x5f9de78e	 [libglesv2.dll	 - context.cpp:975]	gl::Context::deleteFence(unsigned
int)
0x5f9e1491	 [libglesv2.dll	 - context.cpp:198]	gl::Context::~Context()
0x5f9e182f	 [libglesv2.dll	 - context.cpp:3936]	glDestroyContext
0x717654ec	 [libegl.dll	 -
display.cpp:749]	egl::Display::destroyContext(gl::Context *)
0x7176a3da	 [libegl.dll	 - libegl.cpp:907]	eglDestroyContext
0x64fbaf33	 [chrome.dll	 - gl_context_egl.cc:75]	gfx::GLContextEGL::Destroy()

This is a crash in deleteFence but the same could happen in flush. Fix for the former to follow.

The vendor ID is always 8086 (Intel). Not an XP issue - it's happening on Win
7.

Tested by stepping through a call to glFinish in the debugger and ensuring that
it sleeps but eventually returns.
Review URL: http://codereview.appspot.com/5503094

git-svn-id: https://angleproject.googlecode.com/svn/trunk@940 736b8ea6-26fd-11df-bfd4-992fa37f6226

Originally we only match varying variables in vertex/fragment shaders.  However, we should also match uniform names.  For example, if a uniform variable name is used in both vertex and fragment shader, and after mapping the names are different, then the name mapping table is no longer 1:1.

BUG=106638 (crbug)
TEST=http://seanseefried.com/chrome-bug/simple-broken.html shows a magenta square
Review URL: http://codereview.appspot.com/5520051

git-svn-id: https://angleproject.googlecode.com/svn/trunk@939 736b8ea6-26fd-11df-bfd4-992fa37f6226

BUG=http://code.google.com/p/angleproject/issues/detail?id=275
TEST=WebGL conformance suite
TBR=nicolas
Review URL: http://codereview.appspot.com/5520049

git-svn-id: https://angleproject.googlecode.com/svn/trunk@938 736b8ea6-26fd-11df-bfd4-992fa37f6226

TRAC #19330
Signed-off-by: Nicolas Capens
Signed-off-by: Daniel Koch
Authored-by: Shannon Woods

git-svn-id: https://angleproject.googlecode.com/svn/trunk@930 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19330
Signed-off-by: Nicolas Capens

In some cases we were trying to retrieve a serial from
a rendertarget texture which didn't have the storage allocated yet.
This was resulting in a 0 serial which was indistinguishable from
the case where context was just marked dirty and we assumed
the target was up-to-date. This resulted in
failing to apply the correct rendertarget and as a result
some d3d clear calls were failing (picked up by debug runtime).

git-svn-id: https://angleproject.googlecode.com/svn/trunk@929 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19356
Issue=273
Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@928 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19356
Issue=273
Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@927 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19356
Issue=273
Signed-off-by: Nicolas Capens
Based on preliminary patch from Ken Russell

Images are never renderable directly, so renaming the Image::isRenderable function clarifies that we are
only checking on the renderability of the format.
Also factor out the lookup to a static helper function (it will be used elsewhere in the next patch)

git-svn-id: https://angleproject.googlecode.com/svn/trunk@926 736b8ea6-26fd-11df-bfd4-992fa37f6226

TRAC #19349
ANGLEBUG=272
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@925 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19246
Issue=262

Some drivers don't properly return device lost

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@924 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

The problems on d3d9/winxp have been fixed.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@923 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@922 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Use it to avoid duplicate calls to getTexturePool

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@921 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@920 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@919 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Instead of looking this up every time, we'll just compute it when the type/format changes.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@918 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@917 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@916 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

managed textures need to be set on the images so they can be updated
properly and to avoid the additional system memory copy of the surfaces.
Also make sure track the managed state properly.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@915 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=268

The image we were trying to get the d3dformat from hadn't been initialized yet.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@914 736b8ea6-26fd-11df-bfd4-992fa37f6226

Add a static helper function to convert a GL format/type to d3dformat and call this from Image::getD3DFormat

Trac #19259
Issue=268

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@913 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19259
Issue=269

git-svn-id: https://angleproject.googlecode.com/svn/trunk@912 736b8ea6-26fd-11df-bfd4-992fa37f6226

git-svn-id: https://angleproject.googlecode.com/svn/trunk@911 736b8ea6-26fd-11df-bfd4-992fa37f6226

(Landing this on behalf of groby@chromium.org.  Review the CL at http://codereview.appspot.com/5482053/)

Fix issue-242 (failure to compile any shader using cos() with
SH_EMULATE_BUILT_IN_FUNCTIONS)

ANGLEBUG=242

git-svn-id: https://angleproject.googlecode.com/svn/trunk@910 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19246
Issue=262

After a device lost the "mSceneStarted" variable was getting out
of sync and thus we were never restarting the scene before drawing.
It's probably best to ensure that any default states we need are
also set after a reset.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@909 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19246
Issue=262

FBO rendertargets are explicitly refcounted.

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@908 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/5484055

git-svn-id: https://angleproject.googlecode.com/svn/trunk@907 736b8ea6-26fd-11df-bfd4-992fa37f6226

Trac #19246
Issue=262

Signed-off-by: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@906 736b8ea6-26fd-11df-bfd4-992fa37f6226

ANGLEBUG=265
TRAC #19196
Signed-off-by: Daniel Koch
Author: Nicolas Capens

git-svn-id: https://angleproject.googlecode.com/svn/trunk@905 736b8ea6-26fd-11df-bfd4-992fa37f6226

Review URL: http://codereview.appspot.com/5449113

git-svn-id: https://angleproject.googlecode.com/svn/trunk@904 736b8ea6-26fd-11df-bfd4-992fa37f6226