Linux was complaining about ambiguous conditional expressions.
Added some explicit casts to RValue<Int> to fix these warnings.

Change-Id: Ib8e01ee69bb1794ea60b7954f1520a15e383cefa
Reviewed-on: https://swiftshader-review.googlesource.com/8111Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

A warning treated as an error was causing the
official build to fail. Added -Wno-unused-result
to the LLVM build file to solve the issue.

Also ran 'git cl format' on the BUILD.gn files
to fix all formatting issues.

Change-Id: I235c7d9d3e60f788e1dc694a40991c214e6c77e3
Reviewed-on: https://swiftshader-review.googlesource.com/8110Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>

Added support for depth and stencil formats for glBlitFramebuffer:
- Blitter now supports quad layout and stencil
- Device::stretchRect() now supports specific buffers, so that a
  caller can specifically choose which buffer to copy

Change-Id: Iae0898df11e0a1d3c006113486ed15a3fd2f90a9
Reviewed-on: https://swiftshader-review.googlesource.com/7510Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This is a new compilation only test which can be used
to trigger swiftshader compilation on test bots. More
unit tests can be added later.

Change-Id: I46e605b3e7a67438f8f36346e4fcd4e06a2b00ec
Reviewed-on: https://swiftshader-review.googlesource.com/8032Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

A few warnings were fixed:
- 2 sets of virtual flip/blit functions were colliding, so
  I added pure virtual function overrides so that both
  definitions appear within FrameBufferWin
- Moved a few variables within ASSERT inside the ASSERT
  statement in order to remove unused variable warnings
- Removed stack option from glslang.l and removed comments
  handling code, which is actually already done by the
  preprocessor (tested in dEQP)
- Removed unused yyscanner variable from glslang.l
- Ifdefed debug only code in main.cpp

Removed all related muted warnings from BUILD.gn files.

Change-Id: Idf9e7eed00431cc747b689b5d1931fd0c1e8d506
Reviewed-on: https://swiftshader-review.googlesource.com/8010Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nico Weber <thakis@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:6

Change-Id: I13f9c6d5702ff8debc562ab0c7dc593b4a4a9de5
Reviewed-on: https://swiftshader-review.googlesource.com/7991Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I440c71fb57f439ab6461609810d32cfb74a892c7
Reviewed-on: https://swiftshader-review.googlesource.com/7590Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: If2a5c51991bd60e1b6545799cd50f9520c85f21e
Reviewed-on: https://swiftshader-review.googlesource.com/7570Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I8ebd26ce53930b60892bcf4e2378fb587df08da6
Reviewed-on: https://swiftshader-review.googlesource.com/7550Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

It caused building Subzero individually to fail. Instead I've added
a header which will get included when the submodule is not present
and generates an error. The LLVM dependencies were split off into a
separate project to make this work.

This reverts commit d5d886eb.

Change-Id: Ie80a02e20ac67d6a4e0984e1fbab84b880a03469
Reviewed-on: https://swiftshader-review.googlesource.com/7611Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Removed Visual Studio compilation of Subzero and related tests from
the SwiftShader solution so that building the Solution no longer
attempts to build Subzero, which currently doesn't build in Visual
Studio.

Change-Id: Icde9f9ceb90b46681ce8d6019c8151317f84ad17
Reviewed-on: https://swiftshader-review.googlesource.com/7491Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>

Two minor features are required for the blitFramebuffer
implementation in OpenGL ES 3:
- lockStencil with x,y coordinates
- getting the stencil format (new getStencilFormat function)

Change-Id: I83ccccefe70dbbedd05b5d82bc75d30f689c4abe
Reviewed-on: https://swiftshader-review.googlesource.com/7490Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:15

Change-Id: I6cfb3218a9962ca9fe271e9595ecf56aaca6375d
Reviewed-on: https://swiftshader-review.googlesource.com/7398Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:15

Change-Id: Iaa626705220e4bd9b3c744c7623f8b204022b716
Reviewed-on: https://swiftshader-review.googlesource.com/7397Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:15

Change-Id: I81f88999e183ce3617d923249335ff02d6a9f1b1
Reviewed-on: https://swiftshader-review.googlesource.com/7396Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Subzero does not know the element types of vectors, so pass it as an
argument. Also deprecate createSwizzle() and createMask() from
Nucleus since we only need Float4 Swizzle().

Bug swiftshader:15

Change-Id: I38b630f48f8f43e1248338d564a7406d7c8cd5ef
Reviewed-on: https://swiftshader-review.googlesource.com/7395Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:11

Change-Id: I50bb9c6bf7bbe21630bd5f1eb5e0aa0aeb06fdb3
Reviewed-on: https://swiftshader-review.googlesource.com/7393Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:6

Change-Id: I12e1c109c5954ef65ea00750c2d5b027c0c7765a
Reviewed-on: https://swiftshader-review.googlesource.com/7394Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Unlike LLVM, Subzero does not store the type of value a pointer
references as part of the pointer's type information. So add a type
argument to createGEP(). Also add the type to createStore(), to
enable smaller types to be emulated by larger ones.

Bug swiftshader:11

Change-Id: I08173dd0ba07362d2b27baff4a8fba0ecce093d2
Reviewed-on: https://swiftshader-review.googlesource.com/7392Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I43c782a244996aac7bbafd9cb857a3063a68b5a1
Reviewed-on: https://swiftshader-review.googlesource.com/7391Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:6

Change-Id: Ib60cdecf28027687271d2456c6aae79065fb9804
Reviewed-on: https://swiftshader-review.googlesource.com/7364Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug swiftshader:13

Change-Id: I054d0521ecbb1959c5c19b6fa54f76707f4e49c9
Reviewed-on: https://swiftshader-review.googlesource.com/7358Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Subzero has unrelated types for constants and variables. Both are
operands to instructions, but we can't express this relationship at
the Nucleus level. We'd require a large number of Value to Operand
conversions. Instead, an Assign instruction is provided to convert
a Constant into a Value.

Bug swiftshader:12

Change-Id: Ie35a2cea3e485c4012ed949f92825a41caca3367
Reviewed-on: https://swiftshader-review.googlesource.com/7370Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This avoids having to store the type as a member field.

Bug swiftshader:12

Change-Id: I53ea4869a5b590027217d1e9cc015c0502ab530f
Reviewed-on: https://swiftshader-review.googlesource.com/7361Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

This refactoring paves the way for the next patch to support
constructing RValues from Constant pointers as well.

Bug swiftshader:12

Change-Id: I26df3da13314de556dc1af40388278d144560b97
Reviewed-on: https://swiftshader-review.googlesource.com/7371Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

"filter" option was ignored. Simply added the argument
to the context function to fix the issue.

Change-Id: I108c4128bbc4f319091f082df96e056057f926fe
Reviewed-on: https://swiftshader-review.googlesource.com/7363Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

When x was flipped, the entire row would be copied in the
reverse byte order, which was wrong for any type with pixels
larger than 1 byte. Now the full pixels are copied in the
reverse order instead of the bytes, so pixels are no longer
byte flipped.

Change-Id: Ie0e0516546a49d0f890a4abe24d44ccb3ffb2350
Reviewed-on: https://swiftshader-review.googlesource.com/7362Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The code was not allowing to unset a sampler object from a texture,
so the faulty condition was removed to fix the issue.

Fixes 30 tests in:
dEQP.functional.samplers.*

Change-Id: Ibbc3389d3d6135db7bc5a2f87f0df5527b78c7fd
Reviewed-on: https://swiftshader-review.googlesource.com/7360Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:11

Change-Id: I794ef54a7c2ecde71ee6344c63955d2f838ff189
Reviewed-on: https://swiftshader-review.googlesource.com/7331Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:6

Change-Id: I1de2fe8983c7b8d29e0295e5b4705697e5dd6df1
Reviewed-on: https://swiftshader-review.googlesource.com/7292Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I5ab054a5622c9f9dee099f5a3a30c3bb4079ffeb
Reviewed-on: https://swiftshader-review.googlesource.com/7291Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I7eb124ce596d2c7e2419a01a995ea6fa871c8a88
Reviewed-on: https://swiftshader-review.googlesource.com/7290Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

All LLVM instructions are LLVM values, but in the case of terminator
instructions their value is meaningless, and we don't use them.

Bug swiftshader:10

Change-Id: I9eb71e1dcea2614cb810021dfbf3c8908a7e2510
Reviewed-on: https://swiftshader-review.googlesource.com/7355Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This removes all Common headers from Reactor headers, making it
easier to use Reactor outside of SwiftShader.

Bug swiftshader:10

Change-Id: I35c5d138e781d4e911d206954d38e392714d4659
Reviewed-on: https://swiftshader-review.googlesource.com/7330Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:10

Change-Id: I258fc7d66988c5bb9a20f98ac33e87ae26379855
Reviewed-on: https://swiftshader-review.googlesource.com/7280Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Just a refactoring, to make it clearer that Reactor types are the
higher level abstraction, with Nucleus the intermediate interface.

Bug swiftshader:10

Change-Id: I87d8a03cd05d7847511692c5ac9862e77d670801
Reviewed-on: https://swiftshader-review.googlesource.com/7279Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug swiftshader:10

Change-Id: I29b1de8c1adb67449a380c307d12e2aea21f32cc
Reviewed-on: https://swiftshader-review.googlesource.com/7251Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

It has bitrotted and should be reimplemented with support from the
Reactor back-end if needed again.

Bug swiftshader:10

Change-Id: Ie926903ccd8117e91b9d13bf031a7b88287e7276
Reviewed-on: https://swiftshader-review.googlesource.com/7250Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

It's only used for creating intrinsics calls, which is an
implementation detail.

Bug swiftshader:10

Change-Id: Ieac1303e08571e265e239e74c6ea84ece99b53e1
Reviewed-on: https://swiftshader-review.googlesource.com/7278Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

Bug swiftshader:10

Change-Id: Ide80a6d97f67ae04d9118307984235f8f4c2e3ad
Reviewed-on: https://swiftshader-review.googlesource.com/7277Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>