This forces clang to avoid use init_array for all initialization rather than
splitting the initialization between init_array and ctors.

Local builds do this anyway, but for some reason the builds on go/ab do not.
Setting this flag makes the builds consistent (and unbroken on JB-MR1).

I already checked this in on cloud-android-current-release because I wanted
to test it before submitting.

Bug 25597090

Change-Id: I5a7195ba53531835fb8333b698ed39d21ff847a6
(cherry picked from commit 95c7182f118cc686df27c3f49db7d3c7908e0e59)
Reviewed-on: https://swiftshader-review.googlesource.com/4271Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>

Bug 25638876

Change-Id: If6d84c6e0778e0e45af27b74c92c861001669409
Reviewed-on: https://swiftshader-review.googlesource.com/4252Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Also added the exports to allow enabling ES3 entirely on Linux.

Change-Id: Ia13d229b9ce1104a74a2bbd1f620296686cf18d4
Reviewed-on: https://swiftshader-review.googlesource.com/3064Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 25597090

Change-Id: Ie6bc4f55b5105a3e75cdc1b636f3e5716c10cc61
Reviewed-on: https://swiftshader-review.googlesource.com/4240Tested-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 25563277

Change-Id: I1721f060e313c135e713915ddd2dbdbca5239380
Reviewed-on: https://swiftshader-review.googlesource.com/4221Reviewed-by: Keun Soo Yim <yim@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

Bug 24300960

Change-Id: I10a97536bb6559ea7b302b475d0b6936a495b65a
Reviewed-on: https://swiftshader-review.googlesource.com/4210Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I2940ff65144f001d811b069b3e017b0849f310d1
Reviewed-on: https://swiftshader-review.googlesource.com/4201Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 25432320

Change-Id: I3fd6d123cc1c15f04092a888a8e5c996d1bd4231
Reviewed-on: https://swiftshader-review.googlesource.com/4200Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>

On some platforms the default is undefined if neither RTLD_GLOBAL
nor RTLD_LOCAL is specified.

Bug 25282950

Change-Id: I0aea6a9f604afa0412a8f6d9c614b18987fbf514
Reviewed-on: https://swiftshader-review.googlesource.com/4190Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: Ic54eefe422f5d0478b7cdd8dfaa19ccb2e1a8a8e
Reviewed-on: https://swiftshader-review.googlesource.com/4131Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25390254

Change-Id: I434e6d319335be00204eff88f09c98b2dd4f2acc
Reviewed-on: https://swiftshader-review.googlesource.com/4188Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25390254

Change-Id: I6ac37912435e2c73f5e905db7c78af7884f0c93c
Reviewed-on: https://swiftshader-review.googlesource.com/4187Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25390254

Change-Id: Ibd8fc3a957f9ecf5480490e9504fa7041c7c9fa2
Reviewed-on: https://swiftshader-review.googlesource.com/4186Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Only configs with formats that match the system's framebuffer format should
advertised the EGL_FRAMEBUFFER_TARGET_ANDROID attribute.

Bug 25390254

Change-Id: I81e1a0365e151ed7082e3315399748e93c531ccc
Reviewed-on: https://swiftshader-review.googlesource.com/4185Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25390254

Change-Id: I21ed5b6d4a918bb6c3ac9ccc31ee5d47e193bd7f
Reviewed-on: https://swiftshader-review.googlesource.com/4184Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I8fb7499ac9d7d858a785c81d1e8e0a8511cab73c
Reviewed-on: https://swiftshader-review.googlesource.com/4182Reviewed-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Definitions of variables inside functions should be located in a scope
different from function's arguments scope.

Change-Id: I89fe1f527681c68963f4a69c71b78975787a410c
Reviewed-on: https://swiftshader-review.googlesource.com/4180Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Veranika Liaukevich <veranika@google.com>

For this patch lexer and parser files were regenerated with newer versions
of Flex(2.5.39) and Bison(3.0.2). Because YYLEX_PARAM was deprecated and removed
from Bison, it was replaced with %param. The change was inspired by Angle source
code.

Change-Id: I7ac60041aabb78567860653a0663ec9adb205e8a
Reviewed-on: https://swiftshader-review.googlesource.com/4115Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Veranika Liaukevich <veranika@google.com>

Bug 25351037

Change-Id: I0c758a218d15116fa5593eb9ddb783141ff93146
Reviewed-on: https://swiftshader-review.googlesource.com/4181Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25322177

Change-Id: I4bf200e534cc1100db1ae777502f794047df81e5
Reviewed-on: https://swiftshader-review.googlesource.com/4170Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

- Added cases required to load integer type textures into
  swiftshader's internal formats.
- Cleaned up copy functions that simply perform a memcpy
  for a given number of bytes per row.
- Removed unused functionality for mini floats. Right now,
  all we need is the ability to import these formats, so
  there's currently no need to keep the export code.

Change-Id: I61500c41e668f885d3ec1e687b6f888c117221f2
Reviewed-on: https://swiftshader-review.googlesource.com/4140Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Added new read/write functionality for integer types. Also added cases for integer types in some utility functions.

Change-Id: I5efc7c4957d3a1591b47a5df888a9534776033be
Reviewed-on: https://swiftshader-review.googlesource.com/4114Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 25161109

This reverts commit 810f99be.
It was only a workaround, which we no longer need after removing
the libandroid dependency which caused circular symbol resolves.

Change-Id: I0b486b5acfd328c0f64379e2c2dc2db00626738f
Reviewed-on: https://swiftshader-review.googlesource.com/4130Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The blitter now supports integer types. This means:
- It supports conversions to/from integer types from/to
  other already supported types.
- It supports integer to integer conversions without going
  to an intermediate float format.

Also added a Blitter::GetScale() function to avoid having
2 instances of the same switch statement in the code and
added the read/write utility functions. The final Blitter
code is not longer peppered with switch statements and is,
hopefully, easier to read that way.

Change-Id: I80de519aaaa768f8cedd98f97dc4414dda75bf54
Reviewed-on: https://swiftshader-review.googlesource.com/4113Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This was causing dlsym to return the pointer to the wrapper's function when
SwiftShader didn't support an entry point.

Bug 25152138

Change-Id: I92d1c9258c823acf0dc60dca9dc3d7dd552512b3
Reviewed-on: https://swiftshader-review.googlesource.com/4120Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>

Change-Id: I604b2f755ece6a0a411562e9229a39040b3b0153
Reviewed-on: https://swiftshader-review.googlesource.com/4112Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Added initialization for swizzle related members of Sampler

Change-Id: I81f7aad5274696390e73c32ff1b424687e9f5943
Reviewed-on: https://swiftshader-review.googlesource.com/4111Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>

Added new enums for:
- 8/16/32 bit integers
- R, RG, RGB, RGBA
- Normalized and unnormalized

Also fixed FORMAT_LAST.

Change-Id: Ie3eb5286820c44490630e302907350358a1d85dc
Reviewed-on: https://swiftshader-review.googlesource.com/4110Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Initial TextureSize parsing and implementation

Change-Id: I8b9b1808366b1013a5001e2dfa15a26d8471ab6a
Reviewed-on: https://swiftshader-review.googlesource.com/3753Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 24111119

This is a trivial implementation which does not actually optimize
compositing using the damage rects. That can be implemented later.

Change-Id: Ibf511ef166a33b1d2f1826bfc5618765062fe2a2
Reviewed-on: https://swiftshader-review.googlesource.com/4014Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 21662268

Change-Id: Iefdc18303c42967ccc25fd6580851e05520dc2b9
Reviewed-on: https://swiftshader-review.googlesource.com/4071Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The new parameters are: swizzleR, swizzleG, swizzleB, swizzleA.

Also, parameters sent from the context that can be either
originating from the Texture object or the Sampler object
are now sent from the correct origin.

An implementation for the swizzleR, swizzleG, swizzleB, swizzleA
parameters was done in SamplerCore. All related dEQP tests pass.

Change-Id: I45405a0d241d0e70a91a3c56357c7bc04c7dc75a
Reviewed-on: https://swiftshader-review.googlesource.com/4017Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Mipmap levels used in texture sampling must be limited by the
texture's base level (default: 0) and max level (default: 1000).
This has an effect on texture completeness, since only the
required levels are required to be present in the texture to be
considered complete.

Change-Id: I31dd87d5a4306622e469c2546107b59f6e51c7dd
Reviewed-on: https://swiftshader-review.googlesource.com/4039Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The unpack row length was supposed to replace the width, not the
pitch, of the input image, which explains why related tests were
failing.

Change-Id: I7bd46e057bae4113b31d072fd6748d9dd6043feb
Reviewed-on: https://swiftshader-review.googlesource.com/4080Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Added 2 new constructors for the Int4 class in Nucleus,
which are for conversion to Int4 from Short4 or UShort4.
Some of the required code was extracted from Float4
constructors, which were cleaned up to remove the
duplicate code.

All 4 constructors (2 new Int4 and 2 modified Float4)
behaved properly in dEQP tests.

Change-Id: Ib400ea27a867d31901f4a9ca08bf6093f186b277
Reviewed-on: https://swiftshader-review.googlesource.com/4060Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I12c262fbdc2b2b489c02871fb9477bf25a0cfb6b
Reviewed-on: https://swiftshader-review.googlesource.com/4050Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>

Change-Id: I14673f17ef0f299a7ead5bc55bd6729f37759b19
Reviewed-on: https://swiftshader-review.googlesource.com/3870Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Greg Hartman <ghartman@google.com>

- A new routine has been added to convert from any component of
  8bit sRGB data to 8bit linear RGB, using a precomputed array.
- Two new classes have been added to easily convert to and from
  the RGB9E5 format and the R11G11B10F format.

Change-Id: I85ca58bed30bcd5a9130bca5040d351badabb19e
Reviewed-on: https://swiftshader-review.googlesource.com/3990Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I201ffae8193cfcac48a7ad3446f4b79da6c3d217
Reviewed-on: https://swiftshader-review.googlesource.com/4032Tested-by: Greg Hartman <ghartman@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

A new ETC2 decoder was added to SwiftShader, based on the
OpenGL ETC2 specification. The decoder is fully standalone
and does not rely on any outside code (there are no files
included in the header and the source file only include the
header file, so it can easily be ported to any other project).

Things to note:
- In Surface.cpp, signed ETC2 images are decoded to full 32FP
  images, because of the lack of support for signed 8 bit R
  and RG internal formats. This should be fixed as soon as
  these formats are made available.
- sRGB conversion is not performed within the decoder, so it has
  been added as a loop inside Surface::decodeETC2 after the ETC2
  decoding is performed. This is to make sure that there is no
  loss of precision, should we choose to do the conversion to a
  higher bit precision format. The loop is fairly straightforward
  and does the conversion in place, so the impact on performance
  compared to doing the sRGB conversion in the decoder should be
  minimal.

Change-Id: I3a1af623353344bf35818ba9c9f4cf349b587e2f
Reviewed-on: https://swiftshader-review.googlesource.com/3960Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>