1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-22 10:42:39 +01:00
llvm-mirror/lib/CodeGen/SelectionDAG/FunctionLoweringInfo.cpp

569 lines
22 KiB
C++
Raw Normal View History

//===-- FunctionLoweringInfo.cpp ------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
//
// This implements routines for translating functions from LLVM IR into
// Machine IR.
//
//===----------------------------------------------------------------------===//
#include "llvm/CodeGen/FunctionLoweringInfo.h"
#include "llvm/ADT/APInt.h"
#include "llvm/Analysis/LegacyDivergenceAnalysis.h"
#include "llvm/CodeGen/Analysis.h"
#include "llvm/CodeGen/MachineFrameInfo.h"
#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/CodeGen/MachineInstrBuilder.h"
#include "llvm/CodeGen/MachineRegisterInfo.h"
#include "llvm/CodeGen/TargetFrameLowering.h"
#include "llvm/CodeGen/TargetInstrInfo.h"
#include "llvm/CodeGen/TargetLowering.h"
#include "llvm/CodeGen/TargetRegisterInfo.h"
#include "llvm/CodeGen/TargetSubtargetInfo.h"
#include "llvm/CodeGen/WasmEHFuncInfo.h"
#include "llvm/CodeGen/WinEHFuncInfo.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/Target/TargetOptions.h"
#include <algorithm>
using namespace llvm;
#define DEBUG_TYPE "function-lowering-info"
/// isUsedOutsideOfDefiningBlock - Return true if this instruction is used by
/// PHI nodes or outside of the basic block that defines it, or used by a
/// switch or atomic instruction, which may expand to multiple basic blocks.
static bool isUsedOutsideOfDefiningBlock(const Instruction *I) {
if (I->use_empty()) return false;
if (isa<PHINode>(I)) return true;
const BasicBlock *BB = I->getParent();
[C++11] Add range based accessors for the Use-Def chain of a Value. This requires a number of steps. 1) Move value_use_iterator into the Value class as an implementation detail 2) Change it to actually be a *Use* iterator rather than a *User* iterator. 3) Add an adaptor which is a User iterator that always looks through the Use to the User. 4) Wrap these in Value::use_iterator and Value::user_iterator typedefs. 5) Add the range adaptors as Value::uses() and Value::users(). 6) Update *all* of the callers to correctly distinguish between whether they wanted a use_iterator (and to explicitly dig out the User when needed), or a user_iterator which makes the Use itself totally opaque. Because #6 requires churning essentially everything that walked the Use-Def chains, I went ahead and added all of the range adaptors and switched them to range-based loops where appropriate. Also because the renaming requires at least churning every line of code, it didn't make any sense to split these up into multiple commits -- all of which would touch all of the same lies of code. The result is still not quite optimal. The Value::use_iterator is a nice regular iterator, but Value::user_iterator is an iterator over User*s rather than over the User objects themselves. As a consequence, it fits a bit awkwardly into the range-based world and it has the weird extra-dereferencing 'operator->' that so many of our iterators have. I think this could be fixed by providing something which transforms a range of T&s into a range of T*s, but that *can* be separated into another patch, and it isn't yet 100% clear whether this is the right move. However, this change gets us most of the benefit and cleans up a substantial amount of code around Use and User. =] llvm-svn: 203364
2014-03-09 04:16:01 +01:00
for (const User *U : I->users())
if (cast<Instruction>(U)->getParent() != BB || isa<PHINode>(U))
return true;
[C++11] Add range based accessors for the Use-Def chain of a Value. This requires a number of steps. 1) Move value_use_iterator into the Value class as an implementation detail 2) Change it to actually be a *Use* iterator rather than a *User* iterator. 3) Add an adaptor which is a User iterator that always looks through the Use to the User. 4) Wrap these in Value::use_iterator and Value::user_iterator typedefs. 5) Add the range adaptors as Value::uses() and Value::users(). 6) Update *all* of the callers to correctly distinguish between whether they wanted a use_iterator (and to explicitly dig out the User when needed), or a user_iterator which makes the Use itself totally opaque. Because #6 requires churning essentially everything that walked the Use-Def chains, I went ahead and added all of the range adaptors and switched them to range-based loops where appropriate. Also because the renaming requires at least churning every line of code, it didn't make any sense to split these up into multiple commits -- all of which would touch all of the same lies of code. The result is still not quite optimal. The Value::use_iterator is a nice regular iterator, but Value::user_iterator is an iterator over User*s rather than over the User objects themselves. As a consequence, it fits a bit awkwardly into the range-based world and it has the weird extra-dereferencing 'operator->' that so many of our iterators have. I think this could be fixed by providing something which transforms a range of T&s into a range of T*s, but that *can* be separated into another patch, and it isn't yet 100% clear whether this is the right move. However, this change gets us most of the benefit and cleans up a substantial amount of code around Use and User. =] llvm-svn: 203364
2014-03-09 04:16:01 +01:00
return false;
}
static ISD::NodeType getPreferredExtendForValue(const Value *V) {
// For the users of the source value being used for compare instruction, if
// the number of signed predicate is greater than unsigned predicate, we
// prefer to use SIGN_EXTEND.
//
// With this optimization, we would be able to reduce some redundant sign or
// zero extension instruction, and eventually more machine CSE opportunities
// can be exposed.
ISD::NodeType ExtendKind = ISD::ANY_EXTEND;
unsigned NumOfSigned = 0, NumOfUnsigned = 0;
for (const User *U : V->users()) {
if (const auto *CI = dyn_cast<CmpInst>(U)) {
NumOfSigned += CI->isSigned();
NumOfUnsigned += CI->isUnsigned();
}
}
if (NumOfSigned > NumOfUnsigned)
ExtendKind = ISD::SIGN_EXTEND;
return ExtendKind;
}
void FunctionLoweringInfo::set(const Function &fn, MachineFunction &mf,
SelectionDAG *DAG) {
Fn = &fn;
MF = &mf;
TLI = MF->getSubtarget().getTargetLowering();
RegInfo = &MF->getRegInfo();
const TargetFrameLowering *TFI = MF->getSubtarget().getFrameLowering();
DA = DAG->getDivergenceAnalysis();
// Check whether the function can return without sret-demotion.
SmallVector<ISD::OutputArg, 4> Outs;
CallingConv::ID CC = Fn->getCallingConv();
GetReturnInfo(CC, Fn->getReturnType(), Fn->getAttributes(), Outs, *TLI,
mf.getDataLayout());
CanLowerReturn =
TLI->CanLowerReturn(CC, *MF, Fn->isVarArg(), Outs, Fn->getContext());
// If this personality uses funclets, we need to do a bit more work.
DenseMap<const AllocaInst *, TinyPtrVector<int *>> CatchObjects;
EHPersonality Personality = classifyEHPersonality(
Fn->hasPersonalityFn() ? Fn->getPersonalityFn() : nullptr);
if (isFuncletEHPersonality(Personality)) {
// Calculate state numbers if we haven't already.
WinEHFuncInfo &EHInfo = *MF->getWinEHFuncInfo();
if (Personality == EHPersonality::MSVC_CXX)
calculateWinCXXEHStateNumbers(&fn, EHInfo);
else if (isAsynchronousEHPersonality(Personality))
calculateSEHStateNumbers(&fn, EHInfo);
else if (Personality == EHPersonality::CoreCLR)
calculateClrEHStateNumbers(&fn, EHInfo);
// Map all BB references in the WinEH data to MBBs.
for (WinEHTryBlockMapEntry &TBME : EHInfo.TryBlockMap) {
for (WinEHHandlerType &H : TBME.HandlerArray) {
if (const AllocaInst *AI = H.CatchObj.Alloca)
CatchObjects.insert({AI, {}}).first->second.push_back(
&H.CatchObj.FrameIndex);
else
H.CatchObj.FrameIndex = INT_MAX;
}
}
}
if (Personality == EHPersonality::Wasm_CXX) {
WasmEHFuncInfo &EHInfo = *MF->getWasmEHFuncInfo();
calculateWasmEHInfo(&fn, EHInfo);
}
// Initialize the mapping of values to registers. This is only set up for
// instruction values that are used outside of the block that defines
// them.
const Align StackAlign = TFI->getStackAlign();
for (const BasicBlock &BB : *Fn) {
for (const Instruction &I : BB) {
if (const AllocaInst *AI = dyn_cast<AllocaInst>(&I)) {
Type *Ty = AI->getAllocatedType();
Align TyPrefAlign = MF->getDataLayout().getPrefTypeAlign(Ty);
// The "specified" alignment is the alignment written on the alloca,
// or the preferred alignment of the type if none is specified.
//
// (Unspecified alignment on allocas will be going away soon.)
Align SpecifiedAlign = AI->getAlign();
// If the preferred alignment of the type is higher than the specified
// alignment of the alloca, promote the alignment, as long as it doesn't
// require realigning the stack.
//
// FIXME: Do we really want to second-guess the IR in isel?
Align Alignment =
std::max(std::min(TyPrefAlign, StackAlign), SpecifiedAlign);
// Static allocas can be folded into the initial stack frame
// adjustment. For targets that don't realign the stack, don't
// do this if there is an extra alignment requirement.
if (AI->isStaticAlloca() &&
(TFI->isStackRealignable() || (Alignment <= StackAlign))) {
const ConstantInt *CUI = cast<ConstantInt>(AI->getArraySize());
uint64_t TySize =
MF->getDataLayout().getTypeAllocSize(Ty).getKnownMinSize();
TySize *= CUI->getZExtValue(); // Get total allocated size.
if (TySize == 0) TySize = 1; // Don't create zero-sized stack objects.
int FrameIndex = INT_MAX;
auto Iter = CatchObjects.find(AI);
if (Iter != CatchObjects.end() && TLI->needsFixedCatchObjects()) {
FrameIndex = MF->getFrameInfo().CreateFixedObject(
TySize, 0, /*IsImmutable=*/false, /*isAliased=*/true);
MF->getFrameInfo().setObjectAlignment(FrameIndex, Alignment);
} else {
FrameIndex = MF->getFrameInfo().CreateStackObject(TySize, Alignment,
false, AI);
}
// Scalable vectors may need a special StackID to distinguish
// them from other (fixed size) stack objects.
if (isa<ScalableVectorType>(Ty))
MF->getFrameInfo().setStackID(FrameIndex,
TFI->getStackIDForScalableVectors());
StaticAllocaMap[AI] = FrameIndex;
// Update the catch handler information.
if (Iter != CatchObjects.end()) {
for (int *CatchObjPtr : Iter->second)
*CatchObjPtr = FrameIndex;
}
} else {
// FIXME: Overaligned static allocas should be grouped into
// a single dynamic allocation instead of using a separate
// stack allocation for each one.
// Inform the Frame Information that we have variable-sized objects.
MF->getFrameInfo().CreateVariableSizedObject(
Alignment <= StackAlign ? Align(1) : Alignment, AI);
}
} else if (auto *Call = dyn_cast<CallBase>(&I)) {
// Look for inline asm that clobbers the SP register.
if (Call->isInlineAsm()) {
Register SP = TLI->getStackPointerRegisterToSaveRestore();
const TargetRegisterInfo *TRI = MF->getSubtarget().getRegisterInfo();
std::vector<TargetLowering::AsmOperandInfo> Ops =
TLI->ParseConstraints(Fn->getParent()->getDataLayout(), TRI,
*Call);
for (TargetLowering::AsmOperandInfo &Op : Ops) {
if (Op.Type == InlineAsm::isClobber) {
// Clobbers don't have SDValue operands, hence SDValue().
TLI->ComputeConstraintToUse(Op, SDValue(), DAG);
std::pair<unsigned, const TargetRegisterClass *> PhysReg =
TLI->getRegForInlineAsmConstraint(TRI, Op.ConstraintCode,
Op.ConstraintVT);
if (PhysReg.first == SP)
MF->getFrameInfo().setHasOpaqueSPAdjustment(true);
}
}
}
// Look for calls to the @llvm.va_start intrinsic. We can omit some
// prologue boilerplate for variadic functions that don't examine their
// arguments.
if (const auto *II = dyn_cast<IntrinsicInst>(&I)) {
if (II->getIntrinsicID() == Intrinsic::vastart)
MF->getFrameInfo().setHasVAStart(true);
}
// If we have a musttail call in a variadic function, we need to ensure
// we forward implicit register parameters.
if (const auto *CI = dyn_cast<CallInst>(&I)) {
if (CI->isMustTailCall() && Fn->isVarArg())
MF->getFrameInfo().setHasMustTailInVarArgFunc(true);
}
}
// Mark values used outside their block as exported, by allocating
// a virtual register for them.
if (isUsedOutsideOfDefiningBlock(&I))
if (!isa<AllocaInst>(I) || !StaticAllocaMap.count(cast<AllocaInst>(&I)))
InitializeRegForValue(&I);
// Decide the preferred extend type for a value.
PreferredExtendType[&I] = getPreferredExtendForValue(&I);
}
}
// Create an initial MachineBasicBlock for each LLVM BasicBlock in F. This
// also creates the initial PHI MachineInstrs, though none of the input
// operands are populated.
for (const BasicBlock &BB : *Fn) {
// Don't create MachineBasicBlocks for imaginary EH pad blocks. These blocks
// are really data, and no instructions can live here.
if (BB.isEHPad()) {
const Instruction *PadInst = BB.getFirstNonPHI();
// If this is a non-landingpad EH pad, mark this function as using
// funclets.
// FIXME: SEH catchpads do not create EH scope/funclets, so we could avoid
// setting this in such cases in order to improve frame layout.
if (!isa<LandingPadInst>(PadInst)) {
MF->setHasEHScopes(true);
MF->setHasEHFunclets(true);
MF->getFrameInfo().setHasOpaqueSPAdjustment(true);
}
if (isa<CatchSwitchInst>(PadInst)) {
assert(&*BB.begin() == PadInst &&
"WinEHPrepare failed to remove PHIs from imaginary BBs");
continue;
}
if (isa<FuncletPadInst>(PadInst))
assert(&*BB.begin() == PadInst && "WinEHPrepare failed to demote PHIs");
}
MachineBasicBlock *MBB = mf.CreateMachineBasicBlock(&BB);
MBBMap[&BB] = MBB;
MF->push_back(MBB);
// Transfer the address-taken flag. This is necessary because there could
// be multiple MachineBasicBlocks corresponding to one BasicBlock, and only
// the first one should be marked.
if (BB.hasAddressTaken())
MBB->setHasAddressTaken();
// Mark landing pad blocks.
if (BB.isEHPad())
MBB->setIsEHPad();
// Create Machine PHI nodes for LLVM PHI nodes, lowering them as
// appropriate.
for (const PHINode &PN : BB.phis()) {
if (PN.use_empty())
continue;
// Skip empty types
if (PN.getType()->isEmptyTy())
continue;
DebugLoc DL = PN.getDebugLoc();
unsigned PHIReg = ValueMap[&PN];
assert(PHIReg && "PHI node does not have an assigned virtual register!");
SmallVector<EVT, 4> ValueVTs;
ComputeValueVTs(*TLI, MF->getDataLayout(), PN.getType(), ValueVTs);
for (EVT VT : ValueVTs) {
unsigned NumRegisters = TLI->getNumRegisters(Fn->getContext(), VT);
const TargetInstrInfo *TII = MF->getSubtarget().getInstrInfo();
for (unsigned i = 0; i != NumRegisters; ++i)
BuildMI(MBB, DL, TII->get(TargetOpcode::PHI), PHIReg + i);
PHIReg += NumRegisters;
}
}
}
if (isFuncletEHPersonality(Personality)) {
WinEHFuncInfo &EHInfo = *MF->getWinEHFuncInfo();
// Map all BB references in the WinEH data to MBBs.
for (WinEHTryBlockMapEntry &TBME : EHInfo.TryBlockMap) {
for (WinEHHandlerType &H : TBME.HandlerArray) {
if (H.Handler)
H.Handler = MBBMap[H.Handler.get<const BasicBlock *>()];
}
}
for (CxxUnwindMapEntry &UME : EHInfo.CxxUnwindMap)
if (UME.Cleanup)
UME.Cleanup = MBBMap[UME.Cleanup.get<const BasicBlock *>()];
for (SEHUnwindMapEntry &UME : EHInfo.SEHUnwindMap) {
const auto *BB = UME.Handler.get<const BasicBlock *>();
UME.Handler = MBBMap[BB];
}
for (ClrEHUnwindMapEntry &CME : EHInfo.ClrEHUnwindMap) {
const auto *BB = CME.Handler.get<const BasicBlock *>();
CME.Handler = MBBMap[BB];
}
}
else if (Personality == EHPersonality::Wasm_CXX) {
WasmEHFuncInfo &EHInfo = *MF->getWasmEHFuncInfo();
// Map all BB references in the Wasm EH data to MBBs.
DenseMap<BBOrMBB, BBOrMBB> SrcToUnwindDest;
for (auto &KV : EHInfo.SrcToUnwindDest) {
const auto *Src = KV.first.get<const BasicBlock *>();
const auto *Dest = KV.second.get<const BasicBlock *>();
SrcToUnwindDest[MBBMap[Src]] = MBBMap[Dest];
}
EHInfo.SrcToUnwindDest = std::move(SrcToUnwindDest);
DenseMap<BBOrMBB, SmallPtrSet<BBOrMBB, 4>> UnwindDestToSrcs;
for (auto &KV : EHInfo.UnwindDestToSrcs) {
const auto *Dest = KV.first.get<const BasicBlock *>();
UnwindDestToSrcs[MBBMap[Dest]] = SmallPtrSet<BBOrMBB, 4>();
for (const auto P : KV.second)
UnwindDestToSrcs[MBBMap[Dest]].insert(
MBBMap[P.get<const BasicBlock *>()]);
[WebAssembly] Fix incorrect grouping and sorting of exceptions This CL is not big but contains changes that span multiple analyses and passes. This description is very long because it tries to explain basics on what each pass/analysis does and why we need this change on top of that. Please feel free to skip parts that are not necessary for your understanding. --- `WasmEHFuncInfo` contains the mapping of <EH pad, the EH pad's next unwind destination>. The value (unwind dest) here is where an exception should end up when it is not caught by the key (EH pad). We record this info in WasmEHPrepare to fix catch mismatches, because the CFG itself does not have this info. A CFG only contains BBs and predecessor-successor relationship between them, but in `WasmEHFuncInfo` the unwind destination BB is not necessarily a successor or the key EH pad BB. Their relationship can be intuitively explained by this C++ code snippet: ``` try { try { foo(); } catch (int) { // EH pad ... } } catch (...) { // unwind destination } ``` So when `foo()` throws, it goes to `catch (int)` first. But if it is not caught by it, it ends up in the next unwind destination `catch (...)`. This unwind destination is what you see in `catchswitch`'s `unwind label %bb` part. --- `WebAssemblyExceptionInfo` groups exceptions so that they can be sorted continuously together in CFGSort, as we do for loops. What this analysis does is very simple: it creates a single `WebAssemblyException` per EH pad, and all BBs that are dominated by that EH pad are included in this exception. We also identify subexception relationship in this way: if EHPad A domiantes EHPad B, EHPad B's exception is a subexception of EHPad A's exception. This simple rule turns out to be incorrect in some cases. In `WasmEHFuncInfo`, if EHPad A's unwind destination is EHPad B, it means semantically EHPad B should not be included in EHPad A's exception, because it does not make sense to rethrow/delegate to an inner scope. This is what happened in CFGStackify as a result of this: ``` try try catch ... <- %dest_bb is among here! end delegate %dest_bb ``` So this patch adds a phase in `WebAssemblyExceptionInfo::recalculate` to make sure excptions' unwind destinations are not subexceptions of their unwind sources in `WasmEHFuncInfo`. But this alone does not prevent `dest_bb` in the example above from being sorted within the inner `catch`'s exception, even if its exception is not a subexception of that `catch`'s exception anymore, because of how CFGSort works, which will be explained below. --- CFGSort places BBs within the same `SortRegion` (loop or exception) continuously together so they can be demarcated with `loop`-`end_loop` or `catch`-`end_try` in CFGStackify. `SortRegion` is a wrapper for one of `MachineLoop` or `WebAssemblyException`. `SortRegionInfo` already does some complicated things because there discrepancies between those two data structures. `WebAssemblyException` is what we control, and it is defined as an EH pad as its header and BBs dominated by the header as its BBs (with a newly added exception of unwind destinations explained in the previous paragraph). But `MachineLoop` is an LLVM data structure and uses the standard loop detection algorithm. So by the algorithm, BBs that are 1. dominated by the loop header and 2. have a path back to its header. Because of the second condition, many BBs that are dominated by the loop header are not included in the loop. So BBs that contain `return` or branches to outside of the loop are not technically included in `MachineLoop`, but they can be sorted together with the loop with no problem. Maybe to relax the condition, in CFGSort, when we are in a `SortRegion` we allow sorting of not only BBs that belong to the current innermost region but also BBs that are by the current region header. (This was written this way from the first version written by Dan, when only loops existed.) But now, we have cases in exceptions when EHPad B is the unwind destination for EHPad A, even if EHPad B is dominated by EHPad A it should not be included in EHPad A's exception, and should not be sorted within EHPad A. One way to make things work, at least correctly, is change `dominates` condition to `contains` condition for `SortRegion` when sorting BBs, but this will change compilation results for existing non-EH code and I can't be sure it will not degrade performance or code size. I think it will degrade performance because it will force many BBs dominated by a loop, which don't have the path back to the header, to be placed after the loop and it will likely to create more branches and blocks. So this does a little hacky check when adding BBs to `Preferred` list: (`Preferred` list is a ready list. CFGSort maintains ready list in two priority queues: `Preferred` and `Ready`. I'm not very sure why, but it was written that way from the beginning. BBs are first added to `Preferred` list and then some of them are pushed to `Ready` list, so here we only need to guard condition for `Preferred` list.) When adding a BB to `Preferred` list, we check if that BB is an unwind destination of another BB. To do this, this adds the reverse mapping, `UnwindDestToSrc`, and getter methods to `WasmEHFuncInfo`. And if the BB is an unwind destination, it checks if the current stack of regions (`Entries`) contains its source BB by traversing the stack backwards. If we find its unwind source in there, we add the BB to its `Deferred` list, to make sure that unwind destination BB is added to `Preferred` list only after that region with the unwind source BB is sorted and popped from the stack. --- This does not contain a new test that crashes because of this bug, but this fix changes the result for one of existing test case. This test case didn't crash because it fortunately didn't contain `delegate` to the incorrectly placed unwind destination BB. Fixes https://github.com/emscripten-core/emscripten/issues/13514. Reviewed By: dschuff, tlively Differential Revision: https://reviews.llvm.org/D97247
2021-02-17 14:08:52 +01:00
}
EHInfo.UnwindDestToSrcs = std::move(UnwindDestToSrcs);
}
}
/// clear - Clear out all the function-specific state. This returns this
/// FunctionLoweringInfo to an empty state, ready to be used for a
/// different function.
void FunctionLoweringInfo::clear() {
MBBMap.clear();
ValueMap.clear();
VirtReg2Value.clear();
StaticAllocaMap.clear();
LiveOutRegInfo.clear();
VisitedBBs.clear();
ArgDbgValues.clear();
[SelectionDAGBuilder] Add restrictions to EmitFuncArgumentDbgValue Summary: This patch fixes PR40587. When a dbg.value instrinsic is emitted to the DAG by using EmitFuncArgumentDbgValue the resulting DBG_VALUE is hoisted to the beginning of the entry block. I think the idea is to be able to locate a formal argument already from the start of the function. However, EmitFuncArgumentDbgValue only checked that the value that was used to describe a variable was originating from a function parameter, not that the variable itself actually was an argument to the function. So when for example assigning a local variable "local" the value from an argument "a", the assocated DBG_VALUE instruction would be hoisted to the beginning of the function, even if the scope for "local" started somewhere else (or if "local" was mapped to other values earlier in the function). This patch adds some logic to EmitFuncArgumentDbgValue to check that the variable being described actually is an argument to the function. And that the dbg.value being lowered already is in the entry block. Otherwise we bail out, and the dbg.value will be handled as an ordinary dbg.value (not as a "FuncArgumentDbgValue"). A tricky situation is when both the variable and the value is related to function arguments, but not neccessarily the same argument. We make sure that we do not describe the same argument more than once as a "FuncArgumentDbgValue". This solution works as long as opt has injected a "first" dbg.value that corresponds to the formal argument at the function entry. Reviewers: jmorse, aprantl Subscribers: jyknight, hiraditya, fedor.sergeev, dstenb, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D57702 llvm-svn: 353735
2019-02-11 20:23:30 +01:00
DescribedArgs.clear();
ByValArgFrameIndexMap.clear();
RegFixups.clear();
[FastISel] Sink local value materializations to first use Summary: Local values are constants, global addresses, and stack addresses that can't be folded into the instruction that uses them. For example, when storing the address of a global variable into memory, we need to materialize that address into a register. FastISel doesn't want to materialize any given local value more than once, so it generates all local value materialization code at EmitStartPt, which always dominates the current insertion point. This allows it to maintain a map of local value registers, and it knows that the local value area will always dominate the current insertion point. The downside is that local value instructions are always emitted without a source location. This is done to prevent jumpy line tables, but it means that the local value area will be considered part of the previous statement. Consider this C code: call1(); // line 1 ++global; // line 2 ++global; // line 3 call2(&global, &local); // line 4 Today we end up with assembly and line tables like this: .loc 1 1 callq call1 leaq global(%rip), %rdi leaq local(%rsp), %rsi .loc 1 2 addq $1, global(%rip) .loc 1 3 addq $1, global(%rip) .loc 1 4 callq call2 The LEA instructions in the local value area have no source location and are treated as being on line 1. Stepping through the code in a debugger and correlating it with the assembly won't make much sense, because these materializations are only required for line 4. This is actually problematic for the VS debugger "set next statement" feature, which effectively assumes that there are no registers live across statement boundaries. By sinking the local value code into the statement and fixing up the source location, we can make that feature work. This was filed as https://bugs.llvm.org/show_bug.cgi?id=35975 and https://crbug.com/793819. This change is obviously not enough to make this feature work reliably in all cases, but I felt that it was worth doing anyway because it usually generates smaller, more comprehensible -O0 code. I measured a 0.12% regression in code generation time with LLC on the sqlite3 amalgamation, so I think this is worth doing. There are some special cases worth calling out in the commit message: 1. local values materialized for phis 2. local values used by no-op casts 3. dead local value code Local values can be materialized for phis, and this does not show up as a vreg use in MachineRegisterInfo. In this case, if there are no other uses, this patch sinks the value to the first terminator, EH label, or the end of the BB if nothing else exists. Local values may also be used by no-op casts, which adds the register to the RegFixups table. Without reversing the RegFixups map direction, we don't have enough information to sink these instructions. Lastly, if the local value register has no other uses, we can delete it. This comes up when fastisel tries two instruction selection approaches and the first materializes the value but fails and the second succeeds without using the local value. Reviewers: aprantl, dblaikie, qcolombet, MatzeB, vsk, echristo Subscribers: dotdash, chandlerc, hans, sdardis, amccarth, javed.absar, zturner, llvm-commits, hiraditya Differential Revision: https://reviews.llvm.org/D43093 llvm-svn: 327581
2018-03-14 22:54:21 +01:00
RegsWithFixups.clear();
[Statepoints 3/4] Statepoint infrastructure for garbage collection: SelectionDAGBuilder This is the third patch in a small series. It contains the CodeGen support for lowering the gc.statepoint intrinsic sequences (223078) to the STATEPOINT pseudo machine instruction (223085). The change also includes the set of helper routines and classes for working with gc.statepoints, gc.relocates, and gc.results since the lowering code uses them. With this change, gc.statepoints should be functionally complete. The documentation will follow in the fourth change, and there will likely be some cleanup changes, but interested parties can start experimenting now. I'm not particularly happy with the amount of code or complexity involved with the lowering step, but at least it's fairly well isolated. The statepoint lowering code is split into it's own files and anyone not working on the statepoint support itself should be able to ignore it. During the lowering process, we currently spill aggressively to stack. This is not entirely ideal (and we have plans to do better), but it's functional, relatively straight forward, and matches closely the implementations of the patchpoint intrinsics. Most of the complexity comes from trying to keep relocated copies of values in the same stack slots across statepoints. Doing so avoids the insertion of pointless load and store instructions to reshuffle the stack. The current implementation isn't as effective as I'd like, but it is functional and 'good enough' for many common use cases. In the long term, I'd like to figure out how to integrate the statepoint lowering with the register allocator. In principal, we shouldn't need to eagerly spill at all. The register allocator should do any spilling required and the statepoint should simply record that fact. Depending on how challenging that turns out to be, we may invest in a smarter global stack slot assignment mechanism as a stop gap measure. Reviewed by: atrick, ributzka llvm-svn: 223137
2014-12-02 19:50:36 +01:00
StatepointStackSlots.clear();
StatepointRelocationMaps.clear();
PreferredExtendType.clear();
}
/// CreateReg - Allocate a single virtual register for the given type.
2020-04-08 18:26:32 +02:00
Register FunctionLoweringInfo::CreateReg(MVT VT, bool isDivergent) {
return RegInfo->createVirtualRegister(
MF->getSubtarget().getTargetLowering()->getRegClassFor(VT, isDivergent));
}
/// CreateRegs - Allocate the appropriate number of virtual registers of
/// the correctly promoted or expanded types. Assign these registers
/// consecutive vreg numbers and return the first assigned number.
///
/// In the case that the given value has struct or array type, this function
/// will assign registers for each member or element.
///
2020-04-08 18:26:32 +02:00
Register FunctionLoweringInfo::CreateRegs(Type *Ty, bool isDivergent) {
const TargetLowering *TLI = MF->getSubtarget().getTargetLowering();
SmallVector<EVT, 4> ValueVTs;
ComputeValueVTs(*TLI, MF->getDataLayout(), Ty, ValueVTs);
2020-04-08 18:26:32 +02:00
Register FirstReg;
for (unsigned Value = 0, e = ValueVTs.size(); Value != e; ++Value) {
EVT ValueVT = ValueVTs[Value];
MVT RegisterVT = TLI->getRegisterType(Ty->getContext(), ValueVT);
unsigned NumRegs = TLI->getNumRegisters(Ty->getContext(), ValueVT);
for (unsigned i = 0; i != NumRegs; ++i) {
2020-04-08 18:26:32 +02:00
Register R = CreateReg(RegisterVT, isDivergent);
if (!FirstReg) FirstReg = R;
}
}
return FirstReg;
}
2020-04-08 18:26:32 +02:00
Register FunctionLoweringInfo::CreateRegs(const Value *V) {
return CreateRegs(V->getType(), DA && DA->isDivergent(V) &&
!TLI->requiresUniformRegister(*MF, V));
}
/// GetLiveOutRegInfo - Gets LiveOutInfo for a register, returning NULL if the
/// register is a PHI destination and the PHI's LiveOutInfo is not valid. If
/// the register's LiveOutInfo is for a smaller bit width, it is extended to
/// the larger bit width by zero extension. The bit width must be no smaller
/// than the LiveOutInfo's existing bit width.
const FunctionLoweringInfo::LiveOutInfo *
2020-04-08 18:26:32 +02:00
FunctionLoweringInfo::GetLiveOutRegInfo(Register Reg, unsigned BitWidth) {
if (!LiveOutRegInfo.inBounds(Reg))
return nullptr;
LiveOutInfo *LOI = &LiveOutRegInfo[Reg];
if (!LOI->IsValid)
return nullptr;
if (BitWidth > LOI->Known.getBitWidth()) {
LOI->NumSignBits = 1;
LOI->Known = LOI->Known.anyext(BitWidth);
}
return LOI;
}
/// ComputePHILiveOutRegInfo - Compute LiveOutInfo for a PHI's destination
/// register based on the LiveOutInfo of its operands.
void FunctionLoweringInfo::ComputePHILiveOutRegInfo(const PHINode *PN) {
Type *Ty = PN->getType();
if (!Ty->isIntegerTy() || Ty->isVectorTy())
return;
SmallVector<EVT, 1> ValueVTs;
ComputeValueVTs(*TLI, MF->getDataLayout(), Ty, ValueVTs);
assert(ValueVTs.size() == 1 &&
"PHIs with non-vector integer types should have a single VT.");
EVT IntVT = ValueVTs[0];
if (TLI->getNumRegisters(PN->getContext(), IntVT) != 1)
return;
IntVT = TLI->getTypeToTransformTo(PN->getContext(), IntVT);
unsigned BitWidth = IntVT.getSizeInBits();
2020-04-08 18:26:32 +02:00
Register DestReg = ValueMap[PN];
if (!Register::isVirtualRegister(DestReg))
return;
LiveOutRegInfo.grow(DestReg);
LiveOutInfo &DestLOI = LiveOutRegInfo[DestReg];
Value *V = PN->getIncomingValue(0);
if (isa<UndefValue>(V) || isa<ConstantExpr>(V)) {
DestLOI.NumSignBits = 1;
DestLOI.Known = KnownBits(BitWidth);
return;
}
if (ConstantInt *CI = dyn_cast<ConstantInt>(V)) {
APInt Val = CI->getValue().zextOrTrunc(BitWidth);
DestLOI.NumSignBits = Val.getNumSignBits();
DestLOI.Known = KnownBits::makeConstant(Val);
} else {
assert(ValueMap.count(V) && "V should have been placed in ValueMap when its"
"CopyToReg node was created.");
2020-04-08 18:26:32 +02:00
Register SrcReg = ValueMap[V];
if (!Register::isVirtualRegister(SrcReg)) {
DestLOI.IsValid = false;
return;
}
const LiveOutInfo *SrcLOI = GetLiveOutRegInfo(SrcReg, BitWidth);
if (!SrcLOI) {
DestLOI.IsValid = false;
return;
}
DestLOI = *SrcLOI;
}
assert(DestLOI.Known.Zero.getBitWidth() == BitWidth &&
DestLOI.Known.One.getBitWidth() == BitWidth &&
"Masks should have the same bit width as the type.");
for (unsigned i = 1, e = PN->getNumIncomingValues(); i != e; ++i) {
Value *V = PN->getIncomingValue(i);
if (isa<UndefValue>(V) || isa<ConstantExpr>(V)) {
DestLOI.NumSignBits = 1;
DestLOI.Known = KnownBits(BitWidth);
return;
}
if (ConstantInt *CI = dyn_cast<ConstantInt>(V)) {
APInt Val = CI->getValue().zextOrTrunc(BitWidth);
DestLOI.NumSignBits = std::min(DestLOI.NumSignBits, Val.getNumSignBits());
DestLOI.Known.Zero &= ~Val;
DestLOI.Known.One &= Val;
continue;
}
assert(ValueMap.count(V) && "V should have been placed in ValueMap when "
"its CopyToReg node was created.");
2020-04-08 18:26:32 +02:00
Register SrcReg = ValueMap[V];
if (!SrcReg.isVirtual()) {
DestLOI.IsValid = false;
return;
}
const LiveOutInfo *SrcLOI = GetLiveOutRegInfo(SrcReg, BitWidth);
if (!SrcLOI) {
DestLOI.IsValid = false;
return;
}
DestLOI.NumSignBits = std::min(DestLOI.NumSignBits, SrcLOI->NumSignBits);
DestLOI.Known = KnownBits::commonBits(DestLOI.Known, SrcLOI->Known);
}
}
/// setArgumentFrameIndex - Record frame index for the byval
/// argument. This overrides previous frame index entry for this argument,
/// if any.
void FunctionLoweringInfo::setArgumentFrameIndex(const Argument *A,
int FI) {
ByValArgFrameIndexMap[A] = FI;
}
/// getArgumentFrameIndex - Get frame index for the byval argument.
/// If the argument does not have any assigned frame index then 0 is
/// returned.
int FunctionLoweringInfo::getArgumentFrameIndex(const Argument *A) {
auto I = ByValArgFrameIndexMap.find(A);
if (I != ByValArgFrameIndexMap.end())
return I->second;
LLVM_DEBUG(dbgs() << "Argument does not have assigned frame index!\n");
return INT_MAX;
}
2020-04-08 18:26:32 +02:00
Register FunctionLoweringInfo::getCatchPadExceptionPointerVReg(
const Value *CPI, const TargetRegisterClass *RC) {
MachineRegisterInfo &MRI = MF->getRegInfo();
auto I = CatchPadExceptionPointers.insert({CPI, 0});
2020-04-08 18:26:32 +02:00
Register &VReg = I.first->second;
if (I.second)
VReg = MRI.createVirtualRegister(RC);
assert(VReg && "null vreg in exception pointer table!");
return VReg;
}
const Value *
2020-04-08 18:26:32 +02:00
FunctionLoweringInfo::getValueFromVirtualReg(Register Vreg) {
if (VirtReg2Value.empty()) {
AMDGPU: Fix various issues around the VirtReg2Value mapping Summary: The VirtReg2Value mapping is crucial for getting consistently reliable divergence information into the SelectionDAG. This patch fixes a bunch of issues that lead to incorrect divergence info and introduces tight assertions to ensure we don't regress: 1. VirtReg2Value is generated lazily; there were some cases where a lookup was performed before all relevant virtual registers were created, leading to an out-of-sync mapping. Those cases were: - Complex code to lower formal arguments that generated CopyFromReg nodes from live-in registers (fixed by never querying the mapping for live-in registers). - Code that generates CopyToReg for formal arguments that are used outside the entry basic block (fixed by never querying the mapping for Register nodes, which don't need the divergence info anyway). 2. For complex values that are lowered to a sequence of registers, all registers must be reflected in the VirtReg2Value mapping. I am not adding any new tests, since I'm not actually aware of any bugs that these problems are causing with trunk as-is. However, I recently added a test case (in r346423) which fails when D53283 is applied without this change. Also, the new assertions should provide most of the effective test coverage. There is one test change in sdwa-peephole.ll. The underlying issue is that since the divergence info is now correct, the DAGISel will select V_OR_B32 directly instead of S_OR_B32. This leads to an extra COPY which affects the behavior of MachineLICM in a way that ends up with the S_MOV_B32 with the constant in a different basic block than the V_OR_B32, which is presumably what defeats the peephole. Reviewers: alex-t, arsenm, rampitec Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits Differential Revision: https://reviews.llvm.org/D54340 llvm-svn: 348049
2018-11-30 23:55:29 +01:00
SmallVector<EVT, 4> ValueVTs;
for (auto &P : ValueMap) {
AMDGPU: Fix various issues around the VirtReg2Value mapping Summary: The VirtReg2Value mapping is crucial for getting consistently reliable divergence information into the SelectionDAG. This patch fixes a bunch of issues that lead to incorrect divergence info and introduces tight assertions to ensure we don't regress: 1. VirtReg2Value is generated lazily; there were some cases where a lookup was performed before all relevant virtual registers were created, leading to an out-of-sync mapping. Those cases were: - Complex code to lower formal arguments that generated CopyFromReg nodes from live-in registers (fixed by never querying the mapping for live-in registers). - Code that generates CopyToReg for formal arguments that are used outside the entry basic block (fixed by never querying the mapping for Register nodes, which don't need the divergence info anyway). 2. For complex values that are lowered to a sequence of registers, all registers must be reflected in the VirtReg2Value mapping. I am not adding any new tests, since I'm not actually aware of any bugs that these problems are causing with trunk as-is. However, I recently added a test case (in r346423) which fails when D53283 is applied without this change. Also, the new assertions should provide most of the effective test coverage. There is one test change in sdwa-peephole.ll. The underlying issue is that since the divergence info is now correct, the DAGISel will select V_OR_B32 directly instead of S_OR_B32. This leads to an extra COPY which affects the behavior of MachineLICM in a way that ends up with the S_MOV_B32 with the constant in a different basic block than the V_OR_B32, which is presumably what defeats the peephole. Reviewers: alex-t, arsenm, rampitec Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits Differential Revision: https://reviews.llvm.org/D54340 llvm-svn: 348049
2018-11-30 23:55:29 +01:00
ValueVTs.clear();
ComputeValueVTs(*TLI, Fn->getParent()->getDataLayout(),
P.first->getType(), ValueVTs);
unsigned Reg = P.second;
for (EVT VT : ValueVTs) {
unsigned NumRegisters = TLI->getNumRegisters(Fn->getContext(), VT);
for (unsigned i = 0, e = NumRegisters; i != e; ++i)
VirtReg2Value[Reg++] = P.first;
}
}
}
AMDGPU: Fix various issues around the VirtReg2Value mapping Summary: The VirtReg2Value mapping is crucial for getting consistently reliable divergence information into the SelectionDAG. This patch fixes a bunch of issues that lead to incorrect divergence info and introduces tight assertions to ensure we don't regress: 1. VirtReg2Value is generated lazily; there were some cases where a lookup was performed before all relevant virtual registers were created, leading to an out-of-sync mapping. Those cases were: - Complex code to lower formal arguments that generated CopyFromReg nodes from live-in registers (fixed by never querying the mapping for live-in registers). - Code that generates CopyToReg for formal arguments that are used outside the entry basic block (fixed by never querying the mapping for Register nodes, which don't need the divergence info anyway). 2. For complex values that are lowered to a sequence of registers, all registers must be reflected in the VirtReg2Value mapping. I am not adding any new tests, since I'm not actually aware of any bugs that these problems are causing with trunk as-is. However, I recently added a test case (in r346423) which fails when D53283 is applied without this change. Also, the new assertions should provide most of the effective test coverage. There is one test change in sdwa-peephole.ll. The underlying issue is that since the divergence info is now correct, the DAGISel will select V_OR_B32 directly instead of S_OR_B32. This leads to an extra COPY which affects the behavior of MachineLICM in a way that ends up with the S_MOV_B32 with the constant in a different basic block than the V_OR_B32, which is presumably what defeats the peephole. Reviewers: alex-t, arsenm, rampitec Subscribers: kzhuravl, jvesely, wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits Differential Revision: https://reviews.llvm.org/D54340 llvm-svn: 348049
2018-11-30 23:55:29 +01:00
return VirtReg2Value.lookup(Vreg);
}