You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Theodosius/Examples/Theodosius-Usermode/asmjit/core/ralocal_p.h

283 lines
9.8 KiB

// AsmJit - Machine code generation for C++
//
// * Official AsmJit Home Page: https://asmjit.com
// * Official Github Repository: https://github.com/asmjit/asmjit
//
// Copyright (c) 2008-2020 The AsmJit Authors
//
// This software is provided 'as-is', without any express or implied
// warranty. In no event will the authors be held liable for any damages
// arising from the use of this software.
//
// Permission is granted to anyone to use this software for any purpose,
// including commercial applications, and to alter it and redistribute it
// freely, subject to the following restrictions:
//
// 1. The origin of this software must not be misrepresented; you must not
// claim that you wrote the original software. If you use this software
// in a product, an acknowledgment in the product documentation would be
// appreciated but is not required.
// 2. Altered source versions must be plainly marked as such, and must not be
// misrepresented as being the original software.
// 3. This notice may not be removed or altered from any source distribution.
#ifndef ASMJIT_CORE_RALOCAL_P_H_INCLUDED
#define ASMJIT_CORE_RALOCAL_P_H_INCLUDED
#include "../core/api-config.h"
#ifndef ASMJIT_NO_COMPILER
#include "../core/raassignment_p.h"
#include "../core/radefs_p.h"
#include "../core/rapass_p.h"
#include "../core/support.h"
ASMJIT_BEGIN_NAMESPACE
//! \cond INTERNAL
//! \addtogroup asmjit_ra
//! \{
// ============================================================================
// [asmjit::RALocalAllocator]
// ============================================================================
//! Local register allocator.
class RALocalAllocator {
public:
ASMJIT_NONCOPYABLE(RALocalAllocator)
typedef RAAssignment::PhysToWorkMap PhysToWorkMap;
typedef RAAssignment::WorkToPhysMap WorkToPhysMap;
//! Link to `BaseRAPass`.
BaseRAPass* _pass;
//! Link to `BaseCompiler`.
BaseCompiler* _cc;
//! Architecture traits.
const ArchTraits* _archTraits;
//! Registers available to the allocator.
RARegMask _availableRegs;
//! Registers clobbered by the allocator.
RARegMask _clobberedRegs;
//! Register assignment (current).
RAAssignment _curAssignment;
//! Register assignment used temporarily during assignment switches.
RAAssignment _tmpAssignment;
//! Link to the current `RABlock`.
RABlock* _block;
//! InstNode.
InstNode* _node;
//! RA instruction.
RAInst* _raInst;
//! Count of all TiedReg's.
uint32_t _tiedTotal;
//! TiedReg's total counter.
RARegCount _tiedCount;
//! \name Construction & Destruction
//! \{
inline RALocalAllocator(BaseRAPass* pass) noexcept
: _pass(pass),
_cc(pass->cc()),
_archTraits(pass->_archTraits),
_availableRegs(pass->_availableRegs),
_clobberedRegs(),
_curAssignment(),
_block(nullptr),
_node(nullptr),
_raInst(nullptr),
_tiedTotal(),
_tiedCount() {}
Error init() noexcept;
//! \}
//! \name Accessors
//! \{
inline RAWorkReg* workRegById(uint32_t workId) const noexcept { return _pass->workRegById(workId); }
inline PhysToWorkMap* physToWorkMap() const noexcept { return _curAssignment.physToWorkMap(); }
inline WorkToPhysMap* workToPhysMap() const noexcept { return _curAssignment.workToPhysMap(); }
//! Returns the currently processed block.
inline RABlock* block() const noexcept { return _block; }
//! Sets the currently processed block.
inline void setBlock(RABlock* block) noexcept { _block = block; }
//! Returns the currently processed `InstNode`.
inline InstNode* node() const noexcept { return _node; }
//! Returns the currently processed `RAInst`.
inline RAInst* raInst() const noexcept { return _raInst; }
//! Returns all tied regs as `RATiedReg` array.
inline RATiedReg* tiedRegs() const noexcept { return _raInst->tiedRegs(); }
//! Returns tied registers grouped by the given `group`.
inline RATiedReg* tiedRegs(uint32_t group) const noexcept { return _raInst->tiedRegs(group); }
//! Returns count of all TiedRegs used by the instruction.
inline uint32_t tiedCount() const noexcept { return _tiedTotal; }
//! Returns count of TiedRegs used by the given register `group`.
inline uint32_t tiedCount(uint32_t group) const noexcept { return _tiedCount.get(group); }
inline bool isGroupUsed(uint32_t group) const noexcept { return _tiedCount[group] != 0; }
//! \}
//! \name Assignment
//! \{
Error makeInitialAssignment() noexcept;
Error replaceAssignment(
const PhysToWorkMap* physToWorkMap,
const WorkToPhysMap* workToPhysMap) noexcept;
//! Switch to the given assignment by reassigning all register and emitting
//! code that reassigns them. This is always used to switch to a previously
//! stored assignment.
//!
//! If `tryMode` is true then the final assignment doesn't have to be exactly
//! same as specified by `dstPhysToWorkMap` and `dstWorkToPhysMap`. This mode
//! is only used before conditional jumps that already have assignment to
//! generate a code sequence that is always executed regardless of the flow.
Error switchToAssignment(
PhysToWorkMap* dstPhysToWorkMap,
WorkToPhysMap* dstWorkToPhysMap,
const ZoneBitVector& liveIn,
bool dstReadOnly,
bool tryMode) noexcept;
inline Error spillRegsBeforeEntry(RABlock* block) noexcept {
return spillScratchGpRegsBeforeEntry(block->entryScratchGpRegs());
}
Error spillScratchGpRegsBeforeEntry(uint32_t scratchRegs) noexcept;
//! \}
//! \name Allocation
//! \{
Error allocInst(InstNode* node) noexcept;
Error spillAfterAllocation(InstNode* node) noexcept;
Error allocBranch(InstNode* node, RABlock* target, RABlock* cont) noexcept;
Error allocJumpTable(InstNode* node, const RABlocks& targets, RABlock* cont) noexcept;
//! \}
//! \name Decision Making
//! \{
enum CostModel : uint32_t {
kCostOfFrequency = 1048576,
kCostOfDirtyFlag = kCostOfFrequency / 4
};
inline uint32_t costByFrequency(float freq) const noexcept {
return uint32_t(int32_t(freq * float(kCostOfFrequency)));
}
inline uint32_t calculateSpillCost(uint32_t group, uint32_t workId, uint32_t assignedId) const noexcept {
RAWorkReg* workReg = workRegById(workId);
uint32_t cost = costByFrequency(workReg->liveStats().freq());
if (_curAssignment.isPhysDirty(group, assignedId))
cost += kCostOfDirtyFlag;
return cost;
}
//! Decides on register assignment.
uint32_t decideOnAssignment(uint32_t group, uint32_t workId, uint32_t assignedId, uint32_t allocableRegs) const noexcept;
//! Decides on whether to MOVE or SPILL the given WorkReg, because it's allocated
//! in a physical register that have to be used by another WorkReg.
//!
//! The function must return either `RAAssignment::kPhysNone`, which means that
//! the WorkReg of `workId` should be spilled, or a valid physical register ID,
//! which means that the register should be moved to that physical register instead.
uint32_t decideOnReassignment(uint32_t group, uint32_t workId, uint32_t assignedId, uint32_t allocableRegs) const noexcept;
//! Decides on best spill given a register mask `spillableRegs`
uint32_t decideOnSpillFor(uint32_t group, uint32_t workId, uint32_t spillableRegs, uint32_t* spillWorkId) const noexcept;
//! \}
//! \name Emit
//! \{
//! Emits a move between a destination and source register, and fixes the
//! register assignment.
inline Error onMoveReg(uint32_t group, uint32_t workId, uint32_t dstPhysId, uint32_t srcPhysId) noexcept {
if (dstPhysId == srcPhysId) return kErrorOk;
_curAssignment.reassign(group, workId, dstPhysId, srcPhysId);
return _pass->emitMove(workId, dstPhysId, srcPhysId);
}
//! Emits a swap between two physical registers and fixes their assignment.
//!
//! \note Target must support this operation otherwise this would ASSERT.
inline Error onSwapReg(uint32_t group, uint32_t aWorkId, uint32_t aPhysId, uint32_t bWorkId, uint32_t bPhysId) noexcept {
_curAssignment.swap(group, aWorkId, aPhysId, bWorkId, bPhysId);
return _pass->emitSwap(aWorkId, aPhysId, bWorkId, bPhysId);
}
//! Emits a load from [VirtReg/WorkReg]'s spill slot to a physical register
//! and makes it assigned and clean.
inline Error onLoadReg(uint32_t group, uint32_t workId, uint32_t physId) noexcept {
_curAssignment.assign(group, workId, physId, RAAssignment::kClean);
return _pass->emitLoad(workId, physId);
}
//! Emits a save a physical register to a [VirtReg/WorkReg]'s spill slot,
//! keeps it assigned, and makes it clean.
inline Error onSaveReg(uint32_t group, uint32_t workId, uint32_t physId) noexcept {
ASMJIT_ASSERT(_curAssignment.workToPhysId(group, workId) == physId);
ASMJIT_ASSERT(_curAssignment.physToWorkId(group, physId) == workId);
_curAssignment.makeClean(group, workId, physId);
return _pass->emitSave(workId, physId);
}
//! Assigns a register, the content of it is undefined at this point.
inline Error onAssignReg(uint32_t group, uint32_t workId, uint32_t physId, uint32_t dirty) noexcept {
_curAssignment.assign(group, workId, physId, dirty);
return kErrorOk;
}
//! Spills a variable/register, saves the content to the memory-home if modified.
inline Error onSpillReg(uint32_t group, uint32_t workId, uint32_t physId) noexcept {
if (_curAssignment.isPhysDirty(group, physId))
ASMJIT_PROPAGATE(onSaveReg(group, workId, physId));
return onKillReg(group, workId, physId);
}
inline Error onDirtyReg(uint32_t group, uint32_t workId, uint32_t physId) noexcept {
_curAssignment.makeDirty(group, workId, physId);
return kErrorOk;
}
inline Error onKillReg(uint32_t group, uint32_t workId, uint32_t physId) noexcept {
_curAssignment.unassign(group, workId, physId);
return kErrorOk;
}
//! \}
};
//! \}
//! \endcond
ASMJIT_END_NAMESPACE
#endif // !ASMJIT_NO_COMPILER
#endif // ASMJIT_CORE_RALOCAL_P_H_INCLUDED