Change-Id: Ia94eb950d8d3ec4562f9f2e57d8d7ba27a41d16c
Reviewed-on: https://swiftshader-review.googlesource.com/5044Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I0beed96cd68608ce07aec0b11f14d6f61e67e53d
Reviewed-on: https://swiftshader-review.googlesource.com/5034Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Older branches of Android don't support std::unordered_map.

Bug 28006371

Change-Id: I162249db3b7ee9537c7143768c6c1ec216392cba
Reviewed-on: https://swiftshader-review.googlesource.com/5045Reviewed-by: Bo Hu <bohu@google.com>
Tested-by: Nicolas Capens <capn@google.com>
(cherry picked from commit db4fee2752c3b440933fd89078e35b3781c211ce)
Reviewed-on: https://swiftshader-review.googlesource.com/5047Reviewed-by: Nicolas Capens <capn@google.com>

This cl adds 2 new utility function and adds one use of
these in the argument() function.
- getBlockId() finds a block by name and returns its ID
- getTypedMemberInfo() finds which block member matches
  a certain register index and returns the block member's
  packing information and type. If the register falls
  within a block member (as a sub-register of a matrix or
  an array), the base offset of the block member is
  returned, along with the register index, which will then
  be used within the argument() function to compute the
  final parameter index.

Change-Id: Ic0edcb3e6772cdb854301e3d0b400775e7ee72c0
Reviewed-on: https://swiftshader-review.googlesource.com/4682Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: Ia204b708239ae51022f163053e4d054e617695b9
Reviewed-on: https://swiftshader-review.googlesource.com/5031Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I9b46031f57b0cf49428e295b06702099ffef7f01
Reviewed-on: https://swiftshader-review.googlesource.com/5030Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The 2nd operand in a logical and ('&&') operation is evaluated if and only if
the 1st operand evaluates to non-zero. The 2nd operand in a logical or ('||')
operation is evaluated if and only if the 1st operand evaluates to zero.
If an operand is not evaluated, the presence of undefined identifiers in the
operand will not cause an error.

Integer overflow in short-circuited expressions are still and error because
it is part of lexical analysis.

Change-Id: I6ff5e0e9874551d2e40ab4e4ad34dc36cfa703e5
Reviewed-on: https://swiftshader-review.googlesource.com/5020Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: Ifa9db4ab541d9421dddfe6b09571d8b5d38a479d
Reviewed-on: https://swiftshader-review.googlesource.com/5011Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: I9606dfb30693a15a06896d83888022ccbaa8db2e
Reviewed-on: https://swiftshader-review.googlesource.com/5010Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: I5eb797eed31ab233866e5c23d83298d3c57091a9
Reviewed-on: https://swiftshader-review.googlesource.com/4989Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: I277d80be47d2057f31559c5058e2f6ea422dacfd
Reviewed-on: https://swiftshader-review.googlesource.com/4987Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: Ic0e3c5a37174917a541452bab9f92738bc766044
Reviewed-on: https://swiftshader-review.googlesource.com/4986Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: Ic100f0bfe64b1f92f13c4f6a413e30095b7b5a6c
Reviewed-on: https://swiftshader-review.googlesource.com/4985Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 19219444

Change-Id: Id4b209f491b3a3dde716118309cbc8122feb25d0
Reviewed-on: https://swiftshader-review.googlesource.com/4984Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This was causing crashes in Direct3D 9 applications using the "NULL" FOURCC
used as a render target format for rendering to depth only.

Bug 25351344

Change-Id: I69c2e0869f859489321c26dba56b7a80a2ed2f28
Reviewed-on: https://swiftshader-review.googlesource.com/5000Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

The semantic was of size 11, but the 11th member, meant for fog,
was never used, so the array was resized to size 10.

Change-Id: I2afbc2bc52981eb1002252cc3fb89c6954314c23
Reviewed-on: https://swiftshader-review.googlesource.com/4988Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Bug 27460431

Change-Id: Ic030624cc34b31519882e54497457a1b92eaac45
Reviewed-on: https://swiftshader-review.googlesource.com/4981Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 27460431

Change-Id: I90de3285c86c0d3187969270dcbd78a8b02feee9
Reviewed-on: https://swiftshader-review.googlesource.com/4978Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: If0fe05eff397c8d88753fb2b1a1852722c73602e
Reviewed-on: https://swiftshader-review.googlesource.com/4979Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 27460431

Change-Id: Ice854b6faec09dc5f2cbdc2a5c3ffba9a73bfb70
Reviewed-on: https://swiftshader-review.googlesource.com/4977Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 27460431

Change-Id: If11e16597b3fdfbfa6f1f2282e287b302b541f63
Reviewed-on: https://swiftshader-review.googlesource.com/4976Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Added format arguments to DepthBuffer/DepthStencilBuffer
constructors and added existing types to type conversion
functions so that more than one depth/stencil format may
be supported.

Change-Id: Ifd60b896e93a1ba4d05a30f123a1322cdd5254a5
Reviewed-on: https://swiftshader-review.googlesource.com/4766Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

This is not a natively supported pixel format, thus causing a performance
penalty for conversion. It is not supported by Nexus 4 or 5. Android has
deprecated both 4444 and 5551 as HAL formats. Also, UNSIGNED_SHORT_5_5_5_1
has the alpha component in the least significant bit, whereas A1R5G5B5 has
alpha as the most significant bit.

Bug 27690616

Change-Id: I350f0e780cd87b8a626e2c8914a65c61dbf63258
Reviewed-on: https://swiftshader-review.googlesource.com/4960Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Added 32-bit signed and unsigned vertex attribute

Change-Id: Ibbf9c035294584db71713ae7aeca36b32a8564c0
Reviewed-on: https://swiftshader-review.googlesource.com/4974Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

There were a few issues in unary operators:
- Many were not compiling because the promote function had not
  been adjusted to take the new builtin functions into account
- abs and sign had not been implemented for int
- For the integer abs version, used pabsd. Removed the extra
  argument, which seemed unnecessary (abs should have 1 input,
  1 output, AFAIK).

Change-Id: If02c5040438e8c45c99fc7b3c55107448c85cf58
Reviewed-on: https://swiftshader-review.googlesource.com/4970Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: I0c0a9bc99e8cf9ea92919c77ef6c05b1439f40da
Reviewed-on: https://swiftshader-review.googlesource.com/4941Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: Iec28db332fe328603a68c3073403bfbe6067bf44
Reviewed-on: https://swiftshader-review.googlesource.com/4735Reviewed-by: Greg Hartman <ghartman@google.com>
Tested-by: Greg Hartman <ghartman@google.com>
(cherry picked from commit 5d0bad6eea31ee25437877c11e91c1b39be07030)
Reviewed-on: https://swiftshader-review.googlesource.com/4734Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Change-Id: Idcf7c19c2467edf941a8a426078b6f7d7779ab12
Reviewed-on: https://swiftshader-review.googlesource.com/4943Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

SIGXCPU is used by Mono, the C# framework used by the Unity game engine.
LLVM does not make any setrlimit(RLIMIT_CPU) calls, so when the SIGXCPU
signal is raised it is not up to LLVM to decide how to handle it.

Bug 23017372

Change-Id: I292b0eac527c5efd66022e6f75bbcb68879a30a4
Reviewed-on: https://swiftshader-review.googlesource.com/4940Reviewed-by: Greg Hartman <ghartman@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

This is a first implementation of the transform feedback
primitives written query.

It passes available related dEQP tests (more tests will be
available when transform feedback is fully implemented).

Change-Id: Iaee97e3e2e853174b7f22836f72dabede1ed04c1
Reviewed-on: https://swiftshader-review.googlesource.com/4912Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Implemented rasterizer discard by not using the SetupRoutine
and always returning 0 when rasterizer discard is enabled.
Also guarded all clear calls with a check for rasterizer
discard. Passes all rasterizer discard related dEQP tests.

Change-Id: I1e5c107e3dba550f7a5b01eb302ff51bdac303dc
Reviewed-on: https://swiftshader-review.googlesource.com/4876Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 27227050

Change-Id: Id8b6357f25888801dec9bbedc5855faa58def62d
Reviewed-on: https://swiftshader-review.googlesource.com/4920Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Daniel Toyama <kenjitoyama@google.com>

Bug 27454381

Change-Id: Ie59adfa5e5faf08a09427c993775e7d2892d311b
Reviewed-on: https://swiftshader-review.googlesource.com/4913Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Change-Id: I1eb8764c622ffd99a5937fe539bf78dc01d77f23
Reviewed-on: https://swiftshader-review.googlesource.com/4911Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>

Only the 1st sample's buffer of multisampled buffers was getting
cleared instead of clearing all the samples' buffers, which led
to incorrect results. Added a loop to fix the behavior.

Change-Id: If22ebf2df61ae6afbc4c2975a9baee3c18a2492b
Reviewed-on: https://swiftshader-review.googlesource.com/4910Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Intel's reciprocal approximation instruction is not exact for power-of-two
values. It provides 12 bits of mantissa precision and keeps a balance between
positive and negative errors, but the reciprocal of 2^x is not 2^-x. This
affects conformance tests which expect varyings not to be affected by the
perspective division. Correct for this by multiplying by the inverse.

Bug 27165393

Change-Id: Ie52ec511a14a4f447adc47ce9c875bbad03cd274
Reviewed-on: https://swiftshader-review.googlesource.com/4903Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Ported some changes from Angle. Added 4 new functions to
ParserHelper to remove some code from glslang.y. Added
some extra checks regarding qualifiers and precision.

Change-Id: I2856a764749bef0df500891eb4c003211e634673
Reviewed-on: https://swiftshader-review.googlesource.com/4900Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

Bug 27351438

Change-Id: If15f0f91931f34af6e78ad773036b91ff5d57d01
Reviewed-on: https://swiftshader-review.googlesource.com/4902Tested-by: Nicolas Capens <capn@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <capn@google.com>

FenceSync objects are ref counted objects, so the are deleted
using the release() method. This means that they need to be
refed at creation (since we only store the pointer inside the
map), otherwise we'll hit the assert at deletion.

FenceSync objects are currently only ever refed once (at creation)
and unrefed once (at deletion), but are not fully implemented yet.

Change-Id: Ie693ace8895b6140ae96f9325d5c48951c7e9730
Reviewed-on: https://swiftshader-review.googlesource.com/4901Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>

Change-Id: I827b663c9e3018162b192f267fc038ada4b5ac36
Reviewed-on: https://swiftshader-review.googlesource.com/4879Reviewed-by: Nicolas Capens <capn@google.com>
Tested-by: Nicolas Capens <capn@google.com>