/**************************************************************************** * Copyright (C) 2014-2018 Intel Corporation. All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the "Software"), * to deal in the Software without restriction, including without limitation * the rights to use, copy, modify, merge, publish, distribute, sublicense, * and/or sell copies of the Software, and to permit persons to whom the * Software is furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice (including the next * paragraph) shall be included in all copies or substantial portions of the * Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS * IN THE SOFTWARE. * * @file builder_gfx_mem.cpp * * @brief Definition of the gfx mem builder * * Notes: * ******************************************************************************/ #include "jit_pch.hpp" #include "builder.h" #include "common/rdtsc_buckets.h" #include "builder_gfx_mem.h" namespace SwrJit { using namespace llvm; BuilderGfxMem::BuilderGfxMem(JitManager* pJitMgr) : Builder(pJitMgr) { mpTranslationFuncTy = nullptr; mpfnTranslateGfxAddress = nullptr; mpParamSimDC = nullptr; } void BuilderGfxMem::NotifyPrivateContextSet() { } void BuilderGfxMem::AssertGFXMemoryParams(Value* ptr, Builder::JIT_MEM_CLIENT usage) { SWR_ASSERT(!(ptr->getType() == mInt64Ty && usage == MEM_CLIENT_INTERNAL), "Internal memory should not be gfxptr_t."); } ////////////////////////////////////////////////////////////////////////// /// @brief Generate a masked gather operation in LLVM IR. If not /// supported on the underlying platform, emulate it with loads /// @param vSrc - SIMD wide value that will be loaded if mask is invalid /// @param pBase - Int8* base VB address pointer value /// @param vIndices - SIMD wide value of VB byte offsets /// @param vMask - SIMD wide mask that controls whether to access memory or the src values /// @param scale - value to scale indices by Value *BuilderGfxMem::GATHERPS(Value *vSrc, Value *pBase, Value *vIndices, Value *vMask, uint8_t scale, JIT_MEM_CLIENT usage) { Value *vGather; // address may be coming in as 64bit int now so get the pointer if (pBase->getType() == mInt64Ty) { pBase = INT_TO_PTR(pBase, PointerType::get(mInt8Ty, 0)); } vGather = Builder::GATHERPS(vSrc, pBase, vIndices, vMask, scale); return vGather; } ////////////////////////////////////////////////////////////////////////// /// @brief Generate a masked gather operation in LLVM IR. If not /// supported on the underlying platform, emulate it with loads /// @param vSrc - SIMD wide value that will be loaded if mask is invalid /// @param pBase - Int8* base VB address pointer value /// @param vIndices - SIMD wide value of VB byte offsets /// @param vMask - SIMD wide mask that controls whether to access memory or the src values /// @param scale - value to scale indices by Value *BuilderGfxMem::GATHERDD(Value* vSrc, Value* pBase, Value* vIndices, Value* vMask, uint8_t scale, JIT_MEM_CLIENT usage) { Value* vGather = VIMMED1(0.0f); // address may be coming in as 64bit int now so get the pointer if (pBase->getType() == mInt64Ty) { pBase = INT_TO_PTR(pBase, PointerType::get(mInt8Ty, 0)); } vGather = Builder::GATHERDD(vSrc, pBase, vIndices, vMask, scale); return vGather; } Value* BuilderGfxMem::OFFSET_TO_NEXT_COMPONENT(Value* base, Constant *offset) { return ADD(base, offset); } Value *BuilderGfxMem::GEP(Value *Ptr, Value *Idx, Type *Ty, const Twine &Name) { Ptr = TranslationHelper(Ptr, Ty); return Builder::GEP(Ptr, Idx, nullptr, Name); } Value *BuilderGfxMem::GEP(Type *Ty, Value *Ptr, Value *Idx, const Twine &Name) { Ptr = TranslationHelper(Ptr, Ty); return Builder::GEP(Ty, Ptr, Idx, Name); } Value *BuilderGfxMem::GEP(Value* Ptr, const std::initializer_list &indexList, Type *Ty) { Ptr = TranslationHelper(Ptr, Ty); return Builder::GEP(Ptr, indexList); } Value *BuilderGfxMem::GEP(Value* Ptr, const std::initializer_list &indexList, Type *Ty) { Ptr = TranslationHelper(Ptr, Ty); return Builder::GEP(Ptr, indexList); } Value* BuilderGfxMem::TranslationHelper(Value *Ptr, Type *Ty) { SWR_ASSERT(!(Ptr->getType() == mInt64Ty && Ty == nullptr), "Access of GFX pointers must have non-null type specified."); // address may be coming in as 64bit int now so get the pointer if (Ptr->getType() == mInt64Ty) { Ptr = INT_TO_PTR(Ptr, Ty); } return Ptr; } LoadInst* BuilderGfxMem::LOAD(Value *Ptr, const char *Name, Type *Ty, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(Ptr, usage); Ptr = TranslationHelper(Ptr, Ty); return Builder::LOAD(Ptr, Name); } LoadInst* BuilderGfxMem::LOAD(Value *Ptr, const Twine &Name, Type *Ty, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(Ptr, usage); Ptr = TranslationHelper(Ptr, Ty); return Builder::LOAD(Ptr, Name); } LoadInst* BuilderGfxMem::LOAD(Type *Ty, Value *Ptr, const Twine &Name, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(Ptr, usage); Ptr = TranslationHelper(Ptr, Ty); return Builder::LOAD(Ty, Ptr, Name); } LoadInst* BuilderGfxMem::LOAD(Value *Ptr, bool isVolatile, const Twine &Name, Type *Ty, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(Ptr, usage); Ptr = TranslationHelper(Ptr, Ty); return Builder::LOAD(Ptr, isVolatile, Name); } LoadInst *BuilderGfxMem::LOAD(Value *BasePtr, const std::initializer_list &offset, const llvm::Twine& name, Type *Ty, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(BasePtr, usage); // This call is just a pass through to the base class. // It needs to be here to compile due to the combination of virtual overrides and signature overloads. // It doesn't do anything meaningful because the implementation in the base class is going to call // another version of LOAD inside itself where the actual per offset translation will take place // and we can't just translate the BasePtr once, each address needs individual translation. return Builder::LOAD(BasePtr, offset, name, Ty, usage); } CallInst* BuilderGfxMem::MASKED_LOAD(Value *Ptr, unsigned Align, Value *Mask, Value *PassThru, const Twine &Name, Type *Ty, JIT_MEM_CLIENT usage) { AssertGFXMemoryParams(Ptr, usage); Ptr = TranslationHelper(Ptr, Ty); return Builder::MASKED_LOAD(Ptr, Align, Mask, PassThru, Name, Ty, usage); } Value* BuilderGfxMem::TranslateGfxAddress(Value* xpGfxAddress) { return INT_TO_PTR(xpGfxAddress, PointerType::get(mInt8Ty, 0)); } }