This reverts commit 5f5a0648.

Reason for revert: Breaks Android build

 ld.lld: error: undefined symbol: __aeabi_idivmod
 >>> referenced by LLVMJIT.cpp:547 (external/swiftshader/src/Reactor/LLVMJIT.cpp:547)
 >>> LLVMJIT.o:((anonymous namespace)::JITRoutine::JITRoutine(std::__1::unique_ptr<llvm::Module, std::__1::default_delete<llvm::Module> >, llvm::Function**, unsigned int, rr::Config const&)::'lambda'(llvm::StringRef const&)::operator()(llvm::StringRef const&) const) in archive <omitted>/libswiftshader_llvm.a

Change-Id: I153ac3a3a9bc395201024d4b38e43c064691a5ad
Bug: None - build fix.
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41088Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

SwiftShader segfaults when trying to resolve missed symbol. The symbol is requested at least on ARM32 (armv7l)..

Change-Id: I96dd157afc1832ef39daaf46d818c8033820d86b
Bug: b/149229646
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41049Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Kokoro-Presubmit: Lingfeng Yang <lfy@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

`DEBUG` has been defined by some projects as a preprocessor macro.
Rename to fix builds.

Fixes: b/149238449
Change-Id: I9a77321504c7097a5c59dde8b918b52650a90c0f
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41050Reviewed-by: Alexis Hétu <sugoi@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

0aef6454 added a new dependency on Reactor/EmulatedReactor.cpp which was not referenced by Android.bp. Add it to fix the build.

Bug: b/145754674
Change-Id: Id9bac3c9e94edc97959b2a8c1d4abe10f7b78719
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40976Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

As was done with LLVM (https://swiftshader-review.googlesource.com/c/SwiftShader/+/35988),
implement Sinh and Cosh in terms of Exp for Subzero. Both backends now
call into EmulatedReactor. This is not the right solution, as Reactor
should be implemented in terms of std cosh/sinh or intrinsics, as per
b/149110874.

Bug: b/145754674
Change-Id: I8f513e46fb4af270a847343e4b561927aaad1560
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41009Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>

Bug: b/144825072
Change-Id: I17def9659c0d0bc4761e4b1b0673fe248ab31cf5
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/41008Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>

Delete Reactor/DebugAndroid.cpp and Reactor/DebugAndroid.hpp for the same reasons explained in the parent change.

The System / Reactor debug implementations are still not merged, but at least they're now consistent.
System/Debug requires `UNIMPLEMENTED()` to have a bug number for each message, but reactor has a whole bunch of ancient uses without any mention of a bug. I don't have the context to create these bugs, so for now I've created a temporary `UNIMPLEMENTED_NO_BUG()` variant. With this we can easily grep the codebase to locate these usages and create bugs as needed.

Bug: b/146051794
Bug: b/144093703
Change-Id: Id5208f924e63dae7443f95aab5b601a290c297e7
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40969
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Many opcode are not actual statements, but opcodes for declaring types, decorations, capabilities, etc. Single line stepping over these is pointless as they do not actually do anything at execution time.

Just ignore them.

Bug: b/148778720
Change-Id: I0f55fe4b7308f33119dcea4012bc837c375ad5c3
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40830
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

`SWIFTSHADER_LOGGING_LEVEL` controls how much logging we perform.
It defaults to `INFO`, so logs at `DEBUG` level are not printed by default.

Bug: b/146051794
Change-Id: I2037d1c912b24ee8d2e6543074850b65bbbd988a
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40975Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>

Add the missing dependency from WSI -> System.

Bug: b/146051794
Change-Id: I15a48f37d764c339c8c032589625b6d173df76b9
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40974Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Delete Vulkan/VkDebug.hpp and Vulkan/VkDebug.cpp, use System/Debug.hpp from now on.

Also delete System/DebugAndroid.hpp and System/DebugAndroid.cpp.
These declared yet another version of the debug macros, along with the
AndroidEnterDebugger() function. None of this stuff is actually used
for the Android platform builds, and this change brings in logcat support
to the standard set of macros.

Bug: b/146051794
Change-Id: Ib52ec6af34cdf4ac390c73cb331239d81bed6e06
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40968
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Refactor the code so that logging level can be included.

Bug: b/144093703
Change-Id: I266384b7f971109214ae8d435209478ba705bc66
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40951
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Time the Fibonacci function in iterations of 8^N, where N=[0..8]

Bug: b/148660286
Change-Id: Ief391a7f6181c5907de128f5c983ac32aeadfae0
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40949Tested-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>

Bug: b/148660286
Change-Id: I790653cd32736fe5d2ff5d511e42fd232032cadb
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40948
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Unpoison all buffers we store to.

Bug: b/140204746
Change-Id: I211d799b294932affe82cb310ed02d0c0ed5149a
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40908Tested-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>

All places this was raised look like safe usage to me.

Also shuffled some CMake code around so that compiler
warning settings all sit next to each other.

Fixes: b/148842417
Change-Id: I262d0fdacc7a88092cad0890504b0e475f750d47
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40835
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

These used to assert.
Instead of asserting, let the compiler warn about unhandled cases - there's already a `return "<unknown>"` statement below the switch.

Credit to Dan Sinclair who did all the hard work here.

Bug: None - maintenance fix for legacy code
Change-Id: I010a1d6228961a80a15523ff1a2c6db522b7756d
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40833
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

The compiler would produce an `OPCODE_ELSE` with no preceeding `OPCODE_IF`, causing hilarity downstream in the VertexProgram and PixelProgram.

Credit to Dan Sinclair who did all the hard work here.

Bug: None - surfaced when rolling a new version of SPIR-V cross.
Change-Id: I8a937a4d24768dbcc125c3fff834342bd8d0ef29
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40832Reviewed-by: Antonio Maiorano <amaiorano@google.com>
Tested-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>

The environment variables were being modified incorrectly, introducing
blank lines, which Windows doesn't like, leading to the spawned deqp
process having a blank env var table.

Change-Id: Id034594219397809667670f9f7fb26f67ee7d87c
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40888Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>

this can break the android build.
Keep the current vk_swiftshader.lds for the other purposes

Bug: b/148822450
Change-Id: I791b216cc05606ebf51f95d051bb658c0be388fe
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40868
Kokoro-Presubmit: Lingfeng Yang <lfy@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Tested-by: Lingfeng Yang <lfy@google.com>

This makes a rough distinction between code that mainly only interacts
with LLVM's IRBuilder interface, and the part that sets up the JIT and
manages module and function creation, memory management, and external
symbol resolution.

The latter much more often requires changes when LLVM is upgraded than
the former, so it's useful to keep them apart as we'll have multiple
versions of LLVM in different build environments.

Bug: b/148561024
Change-Id: I2ac2d00b4f3bcd277289a4ec6856c9eb7bb6aa24
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40688
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Add a single Reactor Coroutine benchmark so there's something to compile.

Bug: b/148660286
Change-Id: Icbc4c44dd5c0f1ab495c3b8eddce6ea90199e987
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40809Tested-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Let's do some benchmarking!

Bug: b/148660286
Change-Id: I1e0e9b0ec7cf1cba689d195a176eebaf418c8b56
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40808Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

We need to keep a copy of the CPU features for 'mattrs' so they don't
get deleted before we pass them into rr::resolveExternalSymbol() by
raw pointer.

Bug: b/139412871
Change-Id: Iedc2d05fe3ec0d903ffa283994ad95a9cd993e22
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40732
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

Copybara creates Gerrit changes containing a PiperOrigin-RevId: line
when importing from Piper, so accept that instead of a Bug: line.

Bug: b/148537471
Change-Id: I2c655c90c2873787054db4de22bf2f9c3526bdce
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40731
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

'v' could be an Ice::Variable or Ice::Constant, both of which derive
from Ice::Operand. This was just a slip, the function was always meant
to cast to the return type, Ice::Operand*.

Bug: chromium:1047150
Change-Id: I88414de1c56504eafae17c483c7bb574b269a587
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40728
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>

This enables running tests and applications directly against the driver
without the use of a loader.

Bug: b/148240133
Change-Id: I7c906ed72a6bd5465a50381d3e91e18383bb23e1
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/28370
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>

Bug: b/139412871
Change-Id: Ia9e49317fa9ce89dcacaa1e5a788518062e4e633
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40730Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

The `ErrTimeout` type was promoted to the OS-common .go file, but I forgot to remove it from the unix .go file.

Change-Id: Ic30675bd7bc1c5289080c45e21ce9cf19966d3e8
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40708
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

vk_swiftshader_icd.json is located under build/Linux. If an out/
directory exists, it was assumed to contain the file instead. Verify the
file exists under build/, as well as the ICD's shared library.

Change-Id: Iea60dcea6d247d8eb7d984dedae4f4079976fc98
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40649Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Ben Clayton <bclayton@google.com>

Also correct uses of UNIMPLEMENTED() for functionality that is not
advertised as supported, and add spec quotes to justify and document
them.

This change also enforces requiring that the UNIMPLEMENTED() macro's
format string starts with a bug reference of the form b/###. Note we
still have to mention the bug ID in a FIXME() comment as well for
'go/todo' issue tracking (until b/148221114 gets fixed).

Bug: b/131243109
Change-Id: Ieb28e31f2c6c80a27b833626538e7e223b5fda08
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40509
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>

Leak was introduced in 5ba2a5b9.
There are calls to setUses(operand, nullptr) in the algorithm, that,
with the latest changes, would simply replace the existing allocated
Uses object with a new one. Fix it so that setUses(operand, nullptr)
deletes an existing Uses object on the Operand before allocating and
assigning a new one.

Bug: b/145754674
Change-Id: Ie1438378595401619831a2da81d4d0d54a606f80
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40628Reviewed-by: Ben Clayton <bclayton@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Antonio Maiorano <amaiorano@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>

Change-Id: Iff9400cabd5cbf8e7fcffb042c993e29a3acfdc7
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40608
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Coroutines are emulated by using fibers.

Bug: b/145754674
Change-Id: I3f4bf29d26a75a2386ed812dd821d8a7a8276305
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40548Tested-by: Antonio Maiorano <amaiorano@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>

Previously, the builtins were only exposed in the SPIR-V scope.
This change utilizes the new `VariableContainer::extend()` functionality so that a single `VariableContainer` is created to hold all the builtins, and this is used as a base for the SPIR-V locals VC, as well as the high-level local scope VCs.

As part of this change, I've ditched the concept of non-per-lane builtins.
While some builtins are uniform across all lanes, having some at the top level of the watch tree, and others in the `Lane N` tree was just messy and complex to test.

Finally, a handful of builtins are also exposed for vertex shaders.

Bug: b/148401179
Change-Id: I53fedb716ab9337b0aca515768e114a8eef20f28
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40533
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

This allows VariableContainer to 'inherit' from one another.
Variables traversed and searched will try the direct VariableContainer, before falling back to searching the extended VariableContainers.

This allows for variable scopes to nest one another.

Bug: b/148401179
Change-Id: I58b317386d74ec54b9c1cc4fa223ef9cc50b160c
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40532
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Ben Clayton <bclayton@google.com>

Bug: b/139412871
Change-Id: I7c6b59a19b6f2fdc6feb7420605c9e7935790a90
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40588
Presubmit-Ready: Nicolas Capens <nicolascapens@google.com>
Tested-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

See b/148445528 for more information.

Fixes: b/148445528
Change-Id: I0740a5b0ce18db4c3ef0a55a9c679e1b64395b46
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40536Tested-by: Ben Clayton <bclayton@google.com>
Kokoro-Presubmit: kokoro <noreply+kokoro@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>

This reverts commit 23b26ef1.

Reason for revert: Suspecting this cl of breaking the roll

Change-Id: I8a1c0e474de6bf0b04e37c0c3c15febf9ee137da
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40569Reviewed-by: Alexis Hétu <sugoi@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>

The sample locations were reordered "for fragment offset
computation". The reordering was causing some sample location tests
to fail.

Fixes SwANGLE tests:
dEQP.GLES31/functional_texture_multisample_samples_*_sample_position

Bug: b/147387937 b/139793135
Change-Id: I1a62d96517f2be0d625b9e9f8082d6f9e4c6a3f2
Reviewed-on: https://swiftshader-review.googlesource.com/c/SwiftShader/+/40508Tested-by: Alexis Hétu <sugoi@google.com>
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Reviewed-by: Antonio Maiorano <amaiorano@google.com>