bgfx/3rdparty/spirv-tools/source/opt/loop_unswitch_pass.cpp

621 lines
23 KiB
C++
Raw Normal View History

2018-04-11 05:44:28 +03:00
// Copyright (c) 2018 Google LLC.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
2018-09-03 07:14:20 +03:00
#include "source/opt/loop_unswitch_pass.h"
2018-04-11 05:44:28 +03:00
#include <functional>
#include <list>
#include <memory>
#include <type_traits>
#include <unordered_map>
#include <unordered_set>
#include <utility>
#include <vector>
2018-09-03 07:14:20 +03:00
#include "source/opt/basic_block.h"
#include "source/opt/dominator_tree.h"
#include "source/opt/fold.h"
#include "source/opt/function.h"
#include "source/opt/instruction.h"
#include "source/opt/ir_builder.h"
#include "source/opt/ir_context.h"
#include "source/opt/loop_descriptor.h"
2018-04-11 05:44:28 +03:00
2018-09-03 07:14:20 +03:00
#include "source/opt/loop_utils.h"
2018-04-11 05:44:28 +03:00
namespace spvtools {
namespace opt {
namespace {
static const uint32_t kTypePointerStorageClassInIdx = 0;
} // anonymous namespace
namespace {
// This class handle the unswitch procedure for a given loop.
// The unswitch will not happen if:
// - The loop has any instruction that will prevent it;
// - The loop invariant condition is not uniform.
class LoopUnswitch {
public:
2018-09-03 07:14:20 +03:00
LoopUnswitch(IRContext* context, Function* function, Loop* loop,
LoopDescriptor* loop_desc)
2018-04-11 05:44:28 +03:00
: function_(function),
loop_(loop),
loop_desc_(*loop_desc),
context_(context),
switch_block_(nullptr) {}
// Returns true if the loop can be unswitched.
// Can be unswitch if:
// - The loop has no instructions that prevents it (such as barrier);
// - The loop has one conditional branch or switch that do not depends on the
// loop;
// - The loop invariant condition is uniform;
bool CanUnswitchLoop() {
if (switch_block_) return true;
if (loop_->IsSafeToClone()) return false;
2018-09-03 07:14:20 +03:00
CFG& cfg = *context_->cfg();
2018-04-11 05:44:28 +03:00
for (uint32_t bb_id : loop_->GetBlocks()) {
2018-09-03 07:14:20 +03:00
BasicBlock* bb = cfg.block(bb_id);
2019-01-05 19:15:11 +03:00
if (loop_->GetLatchBlock() == bb) {
continue;
}
2018-04-11 05:44:28 +03:00
if (bb->terminator()->IsBranch() &&
bb->terminator()->opcode() != SpvOpBranch) {
2019-01-05 19:15:11 +03:00
if (IsConditionNonConstantLoopInvariant(bb->terminator())) {
2018-04-11 05:44:28 +03:00
switch_block_ = bb;
break;
}
}
}
return switch_block_;
}
// Return the iterator to the basic block |bb|.
2018-09-03 07:14:20 +03:00
Function::iterator FindBasicBlockPosition(BasicBlock* bb_to_find) {
Function::iterator it = function_->FindBlock(bb_to_find->id());
2018-04-11 05:44:28 +03:00
assert(it != function_->end() && "Basic Block not found");
return it;
}
// Creates a new basic block and insert it into the function |fn| at the
// position |ip|. This function preserves the def/use and instr to block
// managers.
2018-09-03 07:14:20 +03:00
BasicBlock* CreateBasicBlock(Function::iterator ip) {
2018-04-11 05:44:28 +03:00
analysis::DefUseManager* def_use_mgr = context_->get_def_use_mgr();
2019-01-05 19:15:11 +03:00
// TODO(1841): Handle id overflow.
2018-09-03 07:14:20 +03:00
BasicBlock* bb = &*ip.InsertBefore(std::unique_ptr<BasicBlock>(
new BasicBlock(std::unique_ptr<Instruction>(new Instruction(
2018-04-11 05:44:28 +03:00
context_, SpvOpLabel, 0, context_->TakeNextId(), {})))));
bb->SetParent(function_);
def_use_mgr->AnalyzeInstDef(bb->GetLabelInst());
context_->set_instr_block(bb->GetLabelInst(), bb);
return bb;
}
2019-01-05 19:15:11 +03:00
Instruction* GetValueForDefaultPathForSwitch(Instruction* switch_inst) {
assert(switch_inst->opcode() == SpvOpSwitch &&
"The given instructoin must be an OpSwitch.");
// Find a value that can be used to select the default path.
// If none are possible, then it will just use 0. The value does not matter
// because this path will never be taken becaues the new switch outside of
// the loop cannot select this path either.
std::vector<uint32_t> existing_values;
for (uint32_t i = 2; i < switch_inst->NumInOperands(); i += 2) {
existing_values.push_back(switch_inst->GetSingleWordInOperand(i));
}
std::sort(existing_values.begin(), existing_values.end());
uint32_t value_for_default_path = 0;
if (existing_values.size() < std::numeric_limits<uint32_t>::max()) {
for (value_for_default_path = 0;
value_for_default_path < existing_values.size();
value_for_default_path++) {
if (existing_values[value_for_default_path] != value_for_default_path) {
break;
}
}
}
InstructionBuilder builder(
context_, static_cast<Instruction*>(nullptr),
IRContext::kAnalysisDefUse | IRContext::kAnalysisInstrToBlockMapping);
return builder.GetUintConstant(value_for_default_path);
}
2018-04-11 05:44:28 +03:00
// Unswitches |loop_|.
void PerformUnswitch() {
assert(CanUnswitchLoop() &&
"Cannot unswitch if there is not constant condition");
assert(loop_->GetPreHeaderBlock() && "This loop has no pre-header block");
assert(loop_->IsLCSSA() && "This loop is not in LCSSA form");
2018-09-03 07:14:20 +03:00
CFG& cfg = *context_->cfg();
2018-04-11 05:44:28 +03:00
DominatorTree* dom_tree =
2018-09-03 07:14:20 +03:00
&context_->GetDominatorAnalysis(function_)->GetDomTree();
2018-04-11 05:44:28 +03:00
analysis::DefUseManager* def_use_mgr = context_->get_def_use_mgr();
LoopUtils loop_utils(context_, loop_);
//////////////////////////////////////////////////////////////////////////////
// Step 1: Create the if merge block for structured modules.
// To do so, the |loop_| merge block will become the if's one and we
// create a merge for the loop. This will limit the amount of duplicated
// code the structured control flow imposes.
// For non structured program, the new loop will be connected to
// the old loop's exit blocks.
//////////////////////////////////////////////////////////////////////////////
// Get the merge block if it exists.
2018-09-03 07:14:20 +03:00
BasicBlock* if_merge_block = loop_->GetMergeBlock();
2018-04-11 05:44:28 +03:00
// The merge block is only created if the loop has a unique exit block. We
// have this guarantee for structured loops, for compute loop it will
// trivially help maintain both a structured-like form and LCSAA.
2018-09-03 07:14:20 +03:00
BasicBlock* loop_merge_block =
2018-04-11 05:44:28 +03:00
if_merge_block
? CreateBasicBlock(FindBasicBlockPosition(if_merge_block))
: nullptr;
if (loop_merge_block) {
// Add the instruction and update managers.
2018-09-03 07:14:20 +03:00
InstructionBuilder builder(
2018-04-11 05:44:28 +03:00
context_, loop_merge_block,
2018-09-03 07:14:20 +03:00
IRContext::kAnalysisDefUse | IRContext::kAnalysisInstrToBlockMapping);
2018-04-11 05:44:28 +03:00
builder.AddBranch(if_merge_block->id());
builder.SetInsertPoint(&*loop_merge_block->begin());
cfg.RegisterBlock(loop_merge_block);
def_use_mgr->AnalyzeInstDef(loop_merge_block->GetLabelInst());
// Update CFG.
if_merge_block->ForEachPhiInst(
2018-09-03 07:14:20 +03:00
[loop_merge_block, &builder, this](Instruction* phi) {
Instruction* cloned = phi->Clone(context_);
2019-01-05 19:15:11 +03:00
cloned->SetResultId(TakeNextId());
2018-09-03 07:14:20 +03:00
builder.AddInstruction(std::unique_ptr<Instruction>(cloned));
2018-04-11 05:44:28 +03:00
phi->SetInOperand(0, {cloned->result_id()});
phi->SetInOperand(1, {loop_merge_block->id()});
for (uint32_t j = phi->NumInOperands() - 1; j > 1; j--)
phi->RemoveInOperand(j);
});
// Copy the predecessor list (will get invalidated otherwise).
std::vector<uint32_t> preds = cfg.preds(if_merge_block->id());
for (uint32_t pid : preds) {
if (pid == loop_merge_block->id()) continue;
2018-09-03 07:14:20 +03:00
BasicBlock* p_bb = cfg.block(pid);
2018-04-11 05:44:28 +03:00
p_bb->ForEachSuccessorLabel(
[if_merge_block, loop_merge_block](uint32_t* id) {
if (*id == if_merge_block->id()) *id = loop_merge_block->id();
});
cfg.AddEdge(pid, loop_merge_block->id());
}
cfg.RemoveNonExistingEdges(if_merge_block->id());
// Update loop descriptor.
2018-09-03 07:14:20 +03:00
if (Loop* ploop = loop_->GetParent()) {
2018-04-11 05:44:28 +03:00
ploop->AddBasicBlock(loop_merge_block);
loop_desc_.SetBasicBlockToLoop(loop_merge_block->id(), ploop);
}
// Update the dominator tree.
DominatorTreeNode* loop_merge_dtn =
dom_tree->GetOrInsertNode(loop_merge_block);
DominatorTreeNode* if_merge_block_dtn =
dom_tree->GetOrInsertNode(if_merge_block);
loop_merge_dtn->parent_ = if_merge_block_dtn->parent_;
loop_merge_dtn->children_.push_back(if_merge_block_dtn);
loop_merge_dtn->parent_->children_.push_back(loop_merge_dtn);
if_merge_block_dtn->parent_->children_.erase(std::find(
if_merge_block_dtn->parent_->children_.begin(),
if_merge_block_dtn->parent_->children_.end(), if_merge_block_dtn));
loop_->SetMergeBlock(loop_merge_block);
}
////////////////////////////////////////////////////////////////////////////
// Step 2: Build a new preheader for |loop_|, use the old one
2019-01-05 19:15:11 +03:00
// for the invariant branch.
2018-04-11 05:44:28 +03:00
////////////////////////////////////////////////////////////////////////////
2018-09-03 07:14:20 +03:00
BasicBlock* if_block = loop_->GetPreHeaderBlock();
2018-04-11 05:44:28 +03:00
// If this preheader is the parent loop header,
// we need to create a dedicated block for the if.
2018-09-03 07:14:20 +03:00
BasicBlock* loop_pre_header =
2018-04-11 05:44:28 +03:00
CreateBasicBlock(++FindBasicBlockPosition(if_block));
2018-09-03 07:14:20 +03:00
InstructionBuilder(
context_, loop_pre_header,
IRContext::kAnalysisDefUse | IRContext::kAnalysisInstrToBlockMapping)
2018-04-11 05:44:28 +03:00
.AddBranch(loop_->GetHeaderBlock()->id());
if_block->tail()->SetInOperand(0, {loop_pre_header->id()});
// Update loop descriptor.
2018-09-03 07:14:20 +03:00
if (Loop* ploop = loop_desc_[if_block]) {
2018-04-11 05:44:28 +03:00
ploop->AddBasicBlock(loop_pre_header);
loop_desc_.SetBasicBlockToLoop(loop_pre_header->id(), ploop);
}
// Update the CFG.
cfg.RegisterBlock(loop_pre_header);
def_use_mgr->AnalyzeInstDef(loop_pre_header->GetLabelInst());
cfg.AddEdge(if_block->id(), loop_pre_header->id());
cfg.RemoveNonExistingEdges(loop_->GetHeaderBlock()->id());
loop_->GetHeaderBlock()->ForEachPhiInst(
2018-09-03 07:14:20 +03:00
[loop_pre_header, if_block](Instruction* phi) {
2018-04-11 05:44:28 +03:00
phi->ForEachInId([loop_pre_header, if_block](uint32_t* id) {
if (*id == if_block->id()) {
*id = loop_pre_header->id();
}
});
});
loop_->SetPreHeaderBlock(loop_pre_header);
// Update the dominator tree.
DominatorTreeNode* loop_pre_header_dtn =
dom_tree->GetOrInsertNode(loop_pre_header);
DominatorTreeNode* if_block_dtn = dom_tree->GetTreeNode(if_block);
loop_pre_header_dtn->parent_ = if_block_dtn;
assert(
if_block_dtn->children_.size() == 1 &&
"A loop preheader should only have the header block as a child in the "
"dominator tree");
loop_pre_header_dtn->children_.push_back(if_block_dtn->children_[0]);
if_block_dtn->children_.clear();
if_block_dtn->children_.push_back(loop_pre_header_dtn);
// Make domination queries valid.
dom_tree->ResetDFNumbering();
// Compute an ordered list of basic block to clone: loop blocks + pre-header
// + merge block.
loop_->ComputeLoopStructuredOrder(&ordered_loop_blocks_, true, true);
/////////////////////////////
// Do the actual unswitch: //
// - Clone the loop //
// - Connect exits //
// - Specialize the loop //
/////////////////////////////
2018-09-03 07:14:20 +03:00
Instruction* iv_condition = &*switch_block_->tail();
2018-04-11 05:44:28 +03:00
SpvOp iv_opcode = iv_condition->opcode();
2018-09-03 07:14:20 +03:00
Instruction* condition =
2018-04-11 05:44:28 +03:00
def_use_mgr->GetDef(iv_condition->GetOperand(0).words[0]);
analysis::ConstantManager* cst_mgr = context_->get_constant_mgr();
const analysis::Type* cond_type =
context_->get_type_mgr()->GetType(condition->type_id());
// Build the list of value for which we need to clone and specialize the
// loop.
2018-09-03 07:14:20 +03:00
std::vector<std::pair<Instruction*, BasicBlock*>> constant_branch;
2018-04-11 05:44:28 +03:00
// Special case for the original loop
2018-09-03 07:14:20 +03:00
Instruction* original_loop_constant_value;
2018-04-11 05:44:28 +03:00
if (iv_opcode == SpvOpBranchConditional) {
constant_branch.emplace_back(
cst_mgr->GetDefiningInstruction(cst_mgr->GetConstant(cond_type, {0})),
nullptr);
original_loop_constant_value =
cst_mgr->GetDefiningInstruction(cst_mgr->GetConstant(cond_type, {1}));
} else {
// We are looking to take the default branch, so we can't provide a
// specific value.
2019-01-05 19:15:11 +03:00
original_loop_constant_value =
GetValueForDefaultPathForSwitch(iv_condition);
2018-04-11 05:44:28 +03:00
for (uint32_t i = 2; i < iv_condition->NumInOperands(); i += 2) {
constant_branch.emplace_back(
cst_mgr->GetDefiningInstruction(cst_mgr->GetConstant(
cond_type, iv_condition->GetInOperand(i).words)),
nullptr);
}
}
// Get the loop landing pads.
std::unordered_set<uint32_t> if_merging_blocks;
std::function<bool(uint32_t)> is_from_original_loop;
if (loop_->GetHeaderBlock()->GetLoopMergeInst()) {
if_merging_blocks.insert(if_merge_block->id());
is_from_original_loop = [this](uint32_t id) {
return loop_->IsInsideLoop(id) || loop_->GetMergeBlock()->id() == id;
};
} else {
loop_->GetExitBlocks(&if_merging_blocks);
is_from_original_loop = [this](uint32_t id) {
return loop_->IsInsideLoop(id);
};
}
for (auto& specialisation_pair : constant_branch) {
2018-09-03 07:14:20 +03:00
Instruction* specialisation_value = specialisation_pair.first;
2018-04-11 05:44:28 +03:00
//////////////////////////////////////////////////////////
// Step 3: Duplicate |loop_|.
//////////////////////////////////////////////////////////
LoopUtils::LoopCloningResult clone_result;
2018-09-03 07:14:20 +03:00
Loop* cloned_loop =
2018-04-11 05:44:28 +03:00
loop_utils.CloneLoop(&clone_result, ordered_loop_blocks_);
specialisation_pair.second = cloned_loop->GetPreHeaderBlock();
////////////////////////////////////
// Step 4: Specialize the loop. //
////////////////////////////////////
{
2019-01-05 19:15:11 +03:00
SpecializeLoop(cloned_loop, condition, specialisation_value);
2018-04-11 05:44:28 +03:00
///////////////////////////////////////////////////////////
// Step 5: Connect convergent edges to the landing pads. //
///////////////////////////////////////////////////////////
for (uint32_t merge_bb_id : if_merging_blocks) {
2018-09-03 07:14:20 +03:00
BasicBlock* merge = context_->cfg()->block(merge_bb_id);
2018-04-11 05:44:28 +03:00
// We are in LCSSA so we only care about phi instructions.
2019-01-05 19:15:11 +03:00
merge->ForEachPhiInst(
[is_from_original_loop, &clone_result](Instruction* phi) {
uint32_t num_in_operands = phi->NumInOperands();
for (uint32_t i = 0; i < num_in_operands; i += 2) {
uint32_t pred = phi->GetSingleWordInOperand(i + 1);
if (is_from_original_loop(pred)) {
pred = clone_result.value_map_.at(pred);
uint32_t incoming_value_id = phi->GetSingleWordInOperand(i);
// Not all the incoming values are coming from the loop.
ValueMapTy::iterator new_value =
clone_result.value_map_.find(incoming_value_id);
if (new_value != clone_result.value_map_.end()) {
incoming_value_id = new_value->second;
}
phi->AddOperand({SPV_OPERAND_TYPE_ID, {incoming_value_id}});
phi->AddOperand({SPV_OPERAND_TYPE_ID, {pred}});
2018-04-11 05:44:28 +03:00
}
}
2019-01-05 19:15:11 +03:00
});
2018-04-11 05:44:28 +03:00
}
}
function_->AddBasicBlocks(clone_result.cloned_bb_.begin(),
clone_result.cloned_bb_.end(),
++FindBasicBlockPosition(if_block));
}
2019-01-05 19:15:11 +03:00
// Specialize the existing loop.
SpecializeLoop(loop_, condition, original_loop_constant_value);
BasicBlock* original_loop_target = loop_->GetPreHeaderBlock();
2018-04-11 05:44:28 +03:00
/////////////////////////////////////
// Finally: connect the new loops. //
/////////////////////////////////////
// Delete the old jump
context_->KillInst(&*if_block->tail());
2018-09-03 07:14:20 +03:00
InstructionBuilder builder(context_, if_block);
2018-04-11 05:44:28 +03:00
if (iv_opcode == SpvOpBranchConditional) {
assert(constant_branch.size() == 1);
builder.AddConditionalBranch(
condition->result_id(), original_loop_target->id(),
constant_branch[0].second->id(),
if_merge_block ? if_merge_block->id() : kInvalidId);
} else {
2018-09-03 07:14:20 +03:00
std::vector<std::pair<Operand::OperandData, uint32_t>> targets;
2018-04-11 05:44:28 +03:00
for (auto& t : constant_branch) {
targets.emplace_back(t.first->GetInOperand(0).words, t.second->id());
}
builder.AddSwitch(condition->result_id(), original_loop_target->id(),
targets,
if_merge_block ? if_merge_block->id() : kInvalidId);
}
switch_block_ = nullptr;
ordered_loop_blocks_.clear();
context_->InvalidateAnalysesExceptFor(
2018-09-03 07:14:20 +03:00
IRContext::Analysis::kAnalysisLoopAnalysis);
2018-04-11 05:44:28 +03:00
}
private:
using ValueMapTy = std::unordered_map<uint32_t, uint32_t>;
2018-09-03 07:14:20 +03:00
using BlockMapTy = std::unordered_map<uint32_t, BasicBlock*>;
2018-04-11 05:44:28 +03:00
2018-09-03 07:14:20 +03:00
Function* function_;
Loop* loop_;
LoopDescriptor& loop_desc_;
IRContext* context_;
2018-04-11 05:44:28 +03:00
2018-09-03 07:14:20 +03:00
BasicBlock* switch_block_;
2018-04-11 05:44:28 +03:00
// Map between instructions and if they are dynamically uniform.
std::unordered_map<uint32_t, bool> dynamically_uniform_;
// The loop basic blocks in structured order.
2018-09-03 07:14:20 +03:00
std::vector<BasicBlock*> ordered_loop_blocks_;
2018-04-11 05:44:28 +03:00
// Returns the next usable id for the context.
2019-01-05 19:15:11 +03:00
uint32_t TakeNextId() {
// TODO(1841): Handle id overflow.
return context_->TakeNextId();
2018-04-11 05:44:28 +03:00
}
// Simplifies |loop| assuming the instruction |to_version_insn| takes the
// value |cst_value|. |block_range| is an iterator range returning the loop
// basic blocks in a structured order (dominator first).
// The function will ignore basic blocks returned by |block_range| if they
// does not belong to the loop.
// The set |dead_blocks| will contain all the dead basic blocks.
//
// Requirements:
// - |loop| must be in the LCSSA form;
2019-01-05 19:15:11 +03:00
// - |cst_value| must be constant.
void SpecializeLoop(Loop* loop, Instruction* to_version_insn,
Instruction* cst_value) {
2018-04-11 05:44:28 +03:00
analysis::DefUseManager* def_use_mgr = context_->get_def_use_mgr();
std::function<bool(uint32_t)> ignore_node;
ignore_node = [loop](uint32_t bb_id) { return !loop->IsInsideLoop(bb_id); };
2018-09-03 07:14:20 +03:00
std::vector<std::pair<Instruction*, uint32_t>> use_list;
def_use_mgr->ForEachUse(to_version_insn,
[&use_list, &ignore_node, this](
Instruction* inst, uint32_t operand_index) {
BasicBlock* bb = context_->get_instr_block(inst);
2018-04-11 05:44:28 +03:00
2018-09-03 07:14:20 +03:00
if (!bb || ignore_node(bb->id())) {
// Out of the loop, the specialization does not
// apply any more.
return;
}
use_list.emplace_back(inst, operand_index);
});
2018-04-11 05:44:28 +03:00
// First pass: inject the specialized value into the loop (and only the
// loop).
for (auto use : use_list) {
2018-09-03 07:14:20 +03:00
Instruction* inst = use.first;
2018-04-11 05:44:28 +03:00
uint32_t operand_index = use.second;
2019-01-05 19:15:11 +03:00
// To also handle switch, cst_value can be nullptr: this case
// means that we are looking to branch to the default target of
// the switch. We don't actually know its value so we don't touch
// it if it not a switch.
assert(cst_value && "We do not have a value to use.");
inst->SetOperand(operand_index, {cst_value->result_id()});
def_use_mgr->AnalyzeInstUse(inst);
2018-04-11 05:44:28 +03:00
}
}
// Returns true if |var| is dynamically uniform.
// Note: this is currently approximated as uniform.
2018-09-03 07:14:20 +03:00
bool IsDynamicallyUniform(Instruction* var, const BasicBlock* entry,
2018-04-11 05:44:28 +03:00
const DominatorTree& post_dom_tree) {
assert(post_dom_tree.IsPostDominator());
analysis::DefUseManager* def_use_mgr = context_->get_def_use_mgr();
auto it = dynamically_uniform_.find(var->result_id());
if (it != dynamically_uniform_.end()) return it->second;
analysis::DecorationManager* dec_mgr = context_->get_decoration_mgr();
bool& is_uniform = dynamically_uniform_[var->result_id()];
is_uniform = false;
dec_mgr->WhileEachDecoration(var->result_id(), SpvDecorationUniform,
2018-09-03 07:14:20 +03:00
[&is_uniform](const Instruction&) {
2018-04-11 05:44:28 +03:00
is_uniform = true;
return false;
});
if (is_uniform) {
return is_uniform;
}
2018-09-03 07:14:20 +03:00
BasicBlock* parent = context_->get_instr_block(var);
2018-04-11 05:44:28 +03:00
if (!parent) {
return is_uniform = true;
}
if (!post_dom_tree.Dominates(parent->id(), entry->id())) {
return is_uniform = false;
}
if (var->opcode() == SpvOpLoad) {
const uint32_t PtrTypeId =
def_use_mgr->GetDef(var->GetSingleWordInOperand(0))->type_id();
2018-09-03 07:14:20 +03:00
const Instruction* PtrTypeInst = def_use_mgr->GetDef(PtrTypeId);
2018-04-11 05:44:28 +03:00
uint32_t storage_class =
PtrTypeInst->GetSingleWordInOperand(kTypePointerStorageClassInIdx);
if (storage_class != SpvStorageClassUniform &&
storage_class != SpvStorageClassUniformConstant) {
return is_uniform = false;
}
} else {
if (!context_->IsCombinatorInstruction(var)) {
return is_uniform = false;
}
}
return is_uniform = var->WhileEachInId([entry, &post_dom_tree,
this](const uint32_t* id) {
return IsDynamicallyUniform(context_->get_def_use_mgr()->GetDef(*id),
entry, post_dom_tree);
});
}
2019-01-05 19:15:11 +03:00
// Returns true if |insn| is not a constant, but is loop invariant and
// dynamically uniform.
bool IsConditionNonConstantLoopInvariant(Instruction* insn) {
2018-04-11 05:44:28 +03:00
assert(insn->IsBranch());
assert(insn->opcode() != SpvOpBranch);
analysis::DefUseManager* def_use_mgr = context_->get_def_use_mgr();
2018-09-03 07:14:20 +03:00
Instruction* condition = def_use_mgr->GetDef(insn->GetOperand(0).words[0]);
2019-01-05 19:15:11 +03:00
if (condition->IsConstant()) {
return false;
}
if (loop_->IsInsideLoop(condition)) {
return false;
}
return IsDynamicallyUniform(
condition, function_->entry().get(),
context_->GetPostDominatorAnalysis(function_)->GetDomTree());
2018-04-11 05:44:28 +03:00
}
};
} // namespace
2018-09-03 07:14:20 +03:00
Pass::Status LoopUnswitchPass::Process() {
2018-04-11 05:44:28 +03:00
bool modified = false;
2018-09-03 07:14:20 +03:00
Module* module = context()->module();
2018-04-11 05:44:28 +03:00
// Process each function in the module
2018-09-03 07:14:20 +03:00
for (Function& f : *module) {
2018-04-11 05:44:28 +03:00
modified |= ProcessFunction(&f);
}
return modified ? Status::SuccessWithChange : Status::SuccessWithoutChange;
}
2018-09-03 07:14:20 +03:00
bool LoopUnswitchPass::ProcessFunction(Function* f) {
2018-04-11 05:44:28 +03:00
bool modified = false;
2018-09-03 07:14:20 +03:00
std::unordered_set<Loop*> processed_loop;
2018-04-11 05:44:28 +03:00
2018-09-03 07:14:20 +03:00
LoopDescriptor& loop_descriptor = *context()->GetLoopDescriptor(f);
2018-04-11 05:44:28 +03:00
bool loop_changed = true;
while (loop_changed) {
loop_changed = false;
2018-09-03 07:14:20 +03:00
for (Loop& loop :
make_range(++TreeDFIterator<Loop>(loop_descriptor.GetDummyRootLoop()),
TreeDFIterator<Loop>())) {
2018-04-11 05:44:28 +03:00
if (processed_loop.count(&loop)) continue;
processed_loop.insert(&loop);
LoopUnswitch unswitcher(context(), f, &loop, &loop_descriptor);
2019-01-05 19:15:11 +03:00
while (unswitcher.CanUnswitchLoop()) {
2018-04-11 05:44:28 +03:00
if (!loop.IsLCSSA()) {
LoopUtils(context(), &loop).MakeLoopClosedSSA();
}
modified = true;
loop_changed = true;
unswitcher.PerformUnswitch();
}
if (loop_changed) break;
}
}
return modified;
}
} // namespace opt
} // namespace spvtools