This cl adds the code that actually performs the copy of the
vertex shader outputs into transform feedback buffers. It
also contains a fix for symmetricNormalizedDepth, which must
be computed after the information was copied into the
transform feedback buffers, when transform feedback is active.

Change-Id: I418f94a15b9425bba0905c840f8cf4828233d0fb
Reviewed-on: https://swiftshader-review.googlesource.com/5172Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The loopCount function was expecting the left side of a binary op
to always be a symbol, which isn't necessarily the case, so a null
pointer check was added to prevent the crash.

Change-Id: I1fe6626bf52ecbb05664d1d2fa18f7ed830a7ee6
Reviewed-on: https://swiftshader-review.googlesource.com/5344Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: Ic078acae24dd2b2361a32498b49238b98e0ac0d1
Reviewed-on: https://swiftshader-review.googlesource.com/5386Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I2c618c03d00718951907e81fcd600155751aac89
Reviewed-on: https://swiftshader-review.googlesource.com/5385Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I17ae53e5274232e9e3b482daac56d507788e822c
Reviewed-on: https://swiftshader-review.googlesource.com/5383Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Previously the 12'th generic varying's w component would overwrite the
clipFlags member field.

Change-Id: I499979a2bbc653c0d95b35512a7a9e36667c2e4d
Reviewed-on: https://swiftshader-review.googlesource.com/5382Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: Iaa6cf9367f291a4bc3bde899484ca323ff54a0d4
Reviewed-on: https://swiftshader-review.googlesource.com/5381Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Implemented EGL_KHR_create_context as described here:
www.khronos.org/registry/egl/extensions/KHR/EGL_KHR_create_context.txt

A small part of it was already implemented, which was to add
support for the EGL_OPENGL_ES3_BIT bit in the EGL_RENDERABLE_TYPE
bitfield. Note that the EGL_OPENGL_ES3_BIT is explicitly disabled
on Android right now.

Change-Id: I10e6222511b29f2d91bd55bfeb0f39bc5b884f89
Reviewed-on: https://swiftshader-review.googlesource.com/5380Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 27225594

Change-Id: Ice38ce8ca8ef157e24d309fe6ac9ce2f69e9e816
Reviewed-on: https://swiftshader-review.googlesource.com/5363Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I47447de635055f9fadf43a71f5423e8fcba884aa
Reviewed-on: https://swiftshader-review.googlesource.com/5358Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: If473b7d83e54ccdca710dffe353b539213ae2046
Reviewed-on: https://swiftshader-review.googlesource.com/5357Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Implemented switch/case for glsl in OpenGL ES 3.0.
For simplicity, it is implemented as a loop without a condition,
so break statements work properly like so:

begin switch
  if(...) // 1st case
  ...
  else if(...) // other cases
  ...
  else // default case
  ...
end switch // Anchor point for break statements

All related dEQP tests pass, except 7 tests where vertex shaders
contain a switch or a loop within another switch. These 7 failures
have only about 5% of bad pixel and seem to be related to an issue
with int(floor(...)), since the equivalent tests inside the fragment
shader pass.

KNOWN ISSUE: If a switch is within a loop and one of the cases
             contains a "continue" statement, this will not be
             handled correctly at the moment. There are no dEQP
             tests for this at the moment, AFAIK.

Change-Id: I3ba34ab06a759d07e8520f6a87d75036a5cdaef5
Reviewed-on: https://swiftshader-review.googlesource.com/5272Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Added a few types:
- Half float
- 2_10_10_10_INT
- 2_10_10_10_UINT

Related dEQP tests pass.

Change-Id: I2fb04fee89c3d69367dc2e401ac53d9758add9bf
Reviewed-on: https://swiftshader-review.googlesource.com/5356Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I093be383c0c5b111592acad4cc9301711c75ab88
Reviewed-on: https://swiftshader-review.googlesource.com/5362Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I3aaf4819fa629282ebeb9f7e70fbae53c261c25d
Reviewed-on: https://swiftshader-review.googlesource.com/5355Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I1444a3d15977c2617866f293e0ac0ccbd253d878
Reviewed-on: https://swiftshader-review.googlesource.com/5353Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: Ifb474e6ab12b254599b6d1713f9f2a01b869e09d
Reviewed-on: https://swiftshader-review.googlesource.com/5361Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I7c854c0c357f480b0686b0ed12e371adca8a738f
Reviewed-on: https://swiftshader-review.googlesource.com/5354Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: Iea408eb9beb6aa3eec4d4e64a3bd34510411dfaa
Reviewed-on: https://swiftshader-review.googlesource.com/5360Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

This fixes issues on Android where applications use these signals in their
normal course of execution. The Mono C# framework uses SIGXCPU for garbage
collection, and ART uses SIGSEGV, SIGQUIT, and SIGUSR1.

Bug 27555932

Change-Id: I4977b8f0419da660a57a8eeef20a7fe747921a63
Reviewed-on: https://swiftshader-review.googlesource.com/5345Reviewed-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I8158c0bb2c19d6cdee12a970f24785fae609d5fc
Reviewed-on: https://swiftshader-review.googlesource.com/5352Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

https://msdn.microsoft.com/en-us/library/windows/desktop/bb172511(v=vs.85).aspx

Change-Id: I2bc8251bce77aacb55fbad1c677f865395f5a986
Reviewed-on: https://swiftshader-review.googlesource.com/5351Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I192d8fe104c3734c9c37108312f1dc93470d8d9e
Reviewed-on: https://swiftshader-review.googlesource.com/5342Reviewed-by: Shannon Woods <shannonwoods@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I9fe9bd339f2aacb04394259ee8753dbc8e041430
Reviewed-on: https://swiftshader-review.googlesource.com/5340Reviewed-by: Shannon Woods <shannonwoods@google.com>
Tested-by: Nicolas Capens <capn@google.com>

When depth and stencil are both present, they must be the same image,
or return GL_FRAMEBUFFER_UNSUPPORTED, according to the GLES 3.0 spec,
section 4.4.4, Framebuffer Completeness.

Change-Id: I607c05617fb55743635eda987da6de523385d92b
Reviewed-on: https://swiftshader-review.googlesource.com/5262Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I4a7aeefedcd2d891093520d5a10ebefadcddb5be
Reviewed-on: https://swiftshader-review.googlesource.com/5320Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I72439a60850add0600face13c94cc481a7083792
Reviewed-on: https://swiftshader-review.googlesource.com/5221Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 28616401

Change-Id: Ia54134e6e1dc6b101c7f81d9047e9bcc9fab3061
Reviewed-on: https://swiftshader-review.googlesource.com/5311Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 28616401

Change-Id: I888ac3dd3186206c7745c23a542b91510a461f45
Reviewed-on: https://swiftshader-review.googlesource.com/5310Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 28474248

Change-Id: I61bf28687623437c6b9b383dc189ffa28ad24c22
Reviewed-on: https://swiftshader-review.googlesource.com/5330Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I7f62eae32d2122ec14f400841eaefa3712cf582d
Reviewed-on: https://swiftshader-review.googlesource.com/5291Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The spec states that "If <sync> is not a valid sync object for <dpy>,
EGL_FALSE is returned and an EGL_BAD_PARAMETER error is generated."

Change-Id: I4bbd2ddfefd7baba9c301decb644bfb545e66f01
Reviewed-on: https://swiftshader-review.googlesource.com/5290Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Implemented modf as trunc + sub
Passes all related dEQP tests

Change-Id: I43656c51a670d235153e5fac390a8db311b14f8d
Reviewed-on: https://swiftshader-review.googlesource.com/5280Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 28474248

Change-Id: Id5b93f3494c918cb344e5da7010dfa48a25cdde1
Reviewed-on: https://swiftshader-review.googlesource.com/5263Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This reverts commit 0637eac3.
It's considered a spec bug:
https://cvs.khronos.org/bugzilla/show_bug.cgi?id=8656

Change-Id: Id49121d5adf91ecab3d261c3c12e2f879f47e856
Reviewed-on: https://swiftshader-review.googlesource.com/5192Reviewed-by: Shannon Woods <shannonwoods@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I1d1a716bd9bd5b636c5b263a7094ff13ac9aa549
Reviewed-on: https://swiftshader-review.googlesource.com/5260Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: Iddd9acf8496abe4dfa147f8a59b4cfb2f23df3a7
Reviewed-on: https://swiftshader-review.googlesource.com/5261Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

This cl adds a utility function that assigns the correct buffer
for each linked transform feedback variable. All the information
about location and size is sent to the VertexProcessor for use
during rendering.

Change-Id: I942805250804f56805de1fc117024c20976e83a0
Reviewed-on: https://swiftshader-review.googlesource.com/5174Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Previously we would mask instruction execution after hitting a LEAVE,
but still jump back on every loop iteration (even if infinite). This
change applies the enableLeave mask to the loop test condition so we
break out of the loop when all strands have hit LEAVE.

Change-Id: Ia331a91fb363d20743d5b87a468c3cfcaa7b481e
Reviewed-on: https://swiftshader-review.googlesource.com/5250Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The not equal folding logic was wrong. It would end the loop
as soon as it found equal components, which was checking if
ALL components are not equal, but that's not how not equal
works, it should return true if ANY component is not equal.

I also refactored it because it was hard to follow, with the
boolNodeFlag variable being used upside down of the result.

Change-Id: I704f26f2fd31a1cc637f5d7601409de8161c5b80
Reviewed-on: https://swiftshader-review.googlesource.com/5211Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>