llvm-project/llvm/lib/IR/BasicBlock.cpp
tentzen 1a949c871a [Windows SEH]: HARDWARE EXCEPTION HANDLING (MSVC -EHa) - Part 2
This patch is the Part-2 (BE LLVM) implementation of HW Exception handling.
Part-1 (FE Clang) was committed in 797ad701522988e212495285dade8efac41a24d4.

This new feature adds the support of Hardware Exception for Microsoft Windows
SEH (Structured Exception Handling).

Compiler options:
  For clang-cl.exe, the option is -EHa, the same as MSVC.
  For clang.exe, the extra option is -fasync-exceptions,
  plus -triple x86_64-windows -fexceptions and -fcxx-exceptions as usual.

NOTE:: Without the -EHa or -fasync-exceptions, this patch is a NO-DIFF change.

The rules for C code:
For C-code, one way (MSVC approach) to achieve SEH -EHa semantic is to follow three rules:
  First, no exception can move in or out of _try region., i.e., no "potential faulty
    instruction can be moved across _try boundary.
  Second, the order of exceptions for instructions 'directly' under a _try must be preserved
    (not applied to those in callees).
  Finally, global states (local/global/heap variables) that can be read outside of _try region
    must be updated in memory (not just in register) before the subsequent exception occurs.

The impact to C++ code:
  Although SEH is a feature for C code, -EHa does have a profound effect on C++
  side. When a C++ function (in the same compilation unit with option -EHa ) is
  called by a SEH C function, a hardware exception occurs in C++ code can also
  be handled properly by an upstream SEH _try-handler or a C++ catch(...).
  As such, when that happens in the middle of an object's life scope, the dtor
  must be invoked the same way as C++ Synchronous Exception during unwinding process.

Design:
A natural way to achieve the rules above in LLVM today is to allow an EH edge
added on memory/computation instruction (previous iload/istore idea) so that
exception path is modeled in Flow graph preciously. However, tracking every
single memory instruction and potential faulty instruction can create many
Invokes, complicate flow graph and possibly result in negative performance
impact for downstream optimization and code generation. Making all
optimizations be aware of the new semantic is also substantial.

This design does not intend to model exception path at instruction level.
Instead, the proposed design tracks and reports EH state at BLOCK-level to
reduce the complexity of flow graph and minimize the performance-impact on CPP
code under -EHa option.
One key element of this design is the ability to compute State number at
block-level. Our algorithm is based on the following rationales:

A _try scope is always a SEME (Single Entry Multiple Exits) region as jumping
into a _try is not allowed. The single entry must start with a seh_try_begin()
invoke with a correct State number that is the initial state of the SEME.
Through control-flow, state number is propagated into all blocks. Side exits
marked by seh_try_end() will unwind to parent state based on existing SEHUnwindMap[].
Note side exits can ONLY jump into parent scopes (lower state number).
Thus, when a block succeeds various states from its predecessors, the lowest
State triumphs others.  If some exits flow to unreachable, propagation on those
paths terminate, not affecting remaining blocks.
For CPP code, object lifetime region is usually a SEME as SEH _try.
However there is one rare exception: jumping into a lifetime that has Dtor but
has no Ctor is warned, but allowed:

Warning: jump bypasses variable with a non-trivial destructor

In that case, the region is actually a MEME (multiple entry multiple exits).
Our solution is to inject a eha_scope_begin() invoke in the side entry block to
ensure a correct State.
Implementation:
Part-1: Clang implementation (already in):
Please see commit 797ad701522988e212495285dade8efac41a24d4).

Part-2 : LLVM implementation described below.

For both C++ & C-code, the state of each block is computed at the same place in
BE (WinEHPreparing pass) where all other EH tables/maps are calculated.
In addition to _scope_begin & _scope_end, the computation of block state also
rely on the existing State tracking code (UnwindMap and InvokeStateMap).

For both C++ & C-code, the state of each block with potential trap instruction
is marked and reported in DAG Instruction Selection pass, the same place where
the state for -EHsc (synchronous exceptions) is done.
If the first instruction in a reported block scope can trap, a Nop is injected
before this instruction. This nop is needed to accommodate LLVM Windows EH
implementation, in which the address in IPToState table is offset by +1.
(note the purpose of that is to ensure the return address of a call is in the
same scope as the call address.

The handler for catch(...) for -EHa must handle HW exception. So it is
'adjective' flag is reset (it cannot be IsStdDotDot (0x40) that only catches
C++ exceptions).
Suppress push/popTerminate() scope (from noexcept/noTHrow) so that HW
exceptions can be passed through.

Original llvm-dev [RFC] discussions can be found in these two threads below:
https://lists.llvm.org/pipermail/llvm-dev/2020-March/140541.html
https://lists.llvm.org/pipermail/llvm-dev/2020-April/141338.html

Differential Revision: https://reviews.llvm.org/D102817/new/
2022-12-01 23:44:25 -08:00

577 lines
18 KiB
C++

//===-- BasicBlock.cpp - Implement BasicBlock related methods -------------===//
//
// 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 file implements the BasicBlock class for the IR library.
//
//===----------------------------------------------------------------------===//
#include "llvm/IR/BasicBlock.h"
#include "SymbolTableListTraitsImpl.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/Statistic.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Type.h"
using namespace llvm;
#define DEBUG_TYPE "ir"
STATISTIC(NumInstrRenumberings, "Number of renumberings across all blocks");
ValueSymbolTable *BasicBlock::getValueSymbolTable() {
if (Function *F = getParent())
return F->getValueSymbolTable();
return nullptr;
}
LLVMContext &BasicBlock::getContext() const {
return getType()->getContext();
}
template <> void llvm::invalidateParentIListOrdering(BasicBlock *BB) {
BB->invalidateOrders();
}
// Explicit instantiation of SymbolTableListTraits since some of the methods
// are not in the public header file...
template class llvm::SymbolTableListTraits<Instruction>;
BasicBlock::BasicBlock(LLVMContext &C, const Twine &Name, Function *NewParent,
BasicBlock *InsertBefore)
: Value(Type::getLabelTy(C), Value::BasicBlockVal), Parent(nullptr) {
if (NewParent)
insertInto(NewParent, InsertBefore);
else
assert(!InsertBefore &&
"Cannot insert block before another block with no function!");
setName(Name);
}
void BasicBlock::insertInto(Function *NewParent, BasicBlock *InsertBefore) {
assert(NewParent && "Expected a parent");
assert(!Parent && "Already has a parent");
if (InsertBefore)
NewParent->getBasicBlockList().insert(InsertBefore->getIterator(), this);
else
NewParent->getBasicBlockList().push_back(this);
}
BasicBlock::~BasicBlock() {
validateInstrOrdering();
// If the address of the block is taken and it is being deleted (e.g. because
// it is dead), this means that there is either a dangling constant expr
// hanging off the block, or an undefined use of the block (source code
// expecting the address of a label to keep the block alive even though there
// is no indirect branch). Handle these cases by zapping the BlockAddress
// nodes. There are no other possible uses at this point.
if (hasAddressTaken()) {
assert(!use_empty() && "There should be at least one blockaddress!");
Constant *Replacement =
ConstantInt::get(llvm::Type::getInt32Ty(getContext()), 1);
while (!use_empty()) {
BlockAddress *BA = cast<BlockAddress>(user_back());
BA->replaceAllUsesWith(ConstantExpr::getIntToPtr(Replacement,
BA->getType()));
BA->destroyConstant();
}
}
assert(getParent() == nullptr && "BasicBlock still linked into the program!");
dropAllReferences();
InstList.clear();
}
void BasicBlock::setParent(Function *parent) {
// Set Parent=parent, updating instruction symtab entries as appropriate.
InstList.setSymTabObject(&Parent, parent);
}
iterator_range<filter_iterator<BasicBlock::const_iterator,
std::function<bool(const Instruction &)>>>
BasicBlock::instructionsWithoutDebug(bool SkipPseudoOp) const {
std::function<bool(const Instruction &)> Fn = [=](const Instruction &I) {
return !isa<DbgInfoIntrinsic>(I) &&
!(SkipPseudoOp && isa<PseudoProbeInst>(I));
};
return make_filter_range(*this, Fn);
}
iterator_range<
filter_iterator<BasicBlock::iterator, std::function<bool(Instruction &)>>>
BasicBlock::instructionsWithoutDebug(bool SkipPseudoOp) {
std::function<bool(Instruction &)> Fn = [=](Instruction &I) {
return !isa<DbgInfoIntrinsic>(I) &&
!(SkipPseudoOp && isa<PseudoProbeInst>(I));
};
return make_filter_range(*this, Fn);
}
filter_iterator<BasicBlock::const_iterator,
std::function<bool(const Instruction &)>>::difference_type
BasicBlock::sizeWithoutDebug() const {
return std::distance(instructionsWithoutDebug().begin(),
instructionsWithoutDebug().end());
}
void BasicBlock::removeFromParent() {
getParent()->getBasicBlockList().remove(getIterator());
}
iplist<BasicBlock>::iterator BasicBlock::eraseFromParent() {
return getParent()->getBasicBlockList().erase(getIterator());
}
void BasicBlock::moveBefore(BasicBlock *MovePos) {
MovePos->getParent()->getBasicBlockList().splice(
MovePos->getIterator(), getParent()->getBasicBlockList(), getIterator());
}
void BasicBlock::moveAfter(BasicBlock *MovePos) {
MovePos->getParent()->getBasicBlockList().splice(
++MovePos->getIterator(), getParent()->getBasicBlockList(),
getIterator());
}
const Module *BasicBlock::getModule() const {
return getParent()->getParent();
}
const CallInst *BasicBlock::getTerminatingMustTailCall() const {
if (InstList.empty())
return nullptr;
const ReturnInst *RI = dyn_cast<ReturnInst>(&InstList.back());
if (!RI || RI == &InstList.front())
return nullptr;
const Instruction *Prev = RI->getPrevNode();
if (!Prev)
return nullptr;
if (Value *RV = RI->getReturnValue()) {
if (RV != Prev)
return nullptr;
// Look through the optional bitcast.
if (auto *BI = dyn_cast<BitCastInst>(Prev)) {
RV = BI->getOperand(0);
Prev = BI->getPrevNode();
if (!Prev || RV != Prev)
return nullptr;
}
}
if (auto *CI = dyn_cast<CallInst>(Prev)) {
if (CI->isMustTailCall())
return CI;
}
return nullptr;
}
const CallInst *BasicBlock::getTerminatingDeoptimizeCall() const {
if (InstList.empty())
return nullptr;
auto *RI = dyn_cast<ReturnInst>(&InstList.back());
if (!RI || RI == &InstList.front())
return nullptr;
if (auto *CI = dyn_cast_or_null<CallInst>(RI->getPrevNode()))
if (Function *F = CI->getCalledFunction())
if (F->getIntrinsicID() == Intrinsic::experimental_deoptimize)
return CI;
return nullptr;
}
const CallInst *BasicBlock::getPostdominatingDeoptimizeCall() const {
const BasicBlock* BB = this;
SmallPtrSet<const BasicBlock *, 8> Visited;
Visited.insert(BB);
while (auto *Succ = BB->getUniqueSuccessor()) {
if (!Visited.insert(Succ).second)
return nullptr;
BB = Succ;
}
return BB->getTerminatingDeoptimizeCall();
}
const Instruction *BasicBlock::getFirstMayFaultInst() const {
if (InstList.empty())
return nullptr;
for (const Instruction &I : *this)
if (isa<LoadInst>(I) || isa<StoreInst>(I) || isa<CallBase>(I))
return &I;
return nullptr;
}
const Instruction* BasicBlock::getFirstNonPHI() const {
for (const Instruction &I : *this)
if (!isa<PHINode>(I))
return &I;
return nullptr;
}
const Instruction *BasicBlock::getFirstNonPHIOrDbg(bool SkipPseudoOp) const {
for (const Instruction &I : *this) {
if (isa<PHINode>(I) || isa<DbgInfoIntrinsic>(I))
continue;
if (SkipPseudoOp && isa<PseudoProbeInst>(I))
continue;
return &I;
}
return nullptr;
}
const Instruction *
BasicBlock::getFirstNonPHIOrDbgOrLifetime(bool SkipPseudoOp) const {
for (const Instruction &I : *this) {
if (isa<PHINode>(I) || isa<DbgInfoIntrinsic>(I))
continue;
if (I.isLifetimeStartOrEnd())
continue;
if (SkipPseudoOp && isa<PseudoProbeInst>(I))
continue;
return &I;
}
return nullptr;
}
BasicBlock::const_iterator BasicBlock::getFirstInsertionPt() const {
const Instruction *FirstNonPHI = getFirstNonPHI();
if (!FirstNonPHI)
return end();
const_iterator InsertPt = FirstNonPHI->getIterator();
if (InsertPt->isEHPad()) ++InsertPt;
return InsertPt;
}
BasicBlock::const_iterator BasicBlock::getFirstNonPHIOrDbgOrAlloca() const {
const Instruction *FirstNonPHI = getFirstNonPHI();
if (!FirstNonPHI)
return end();
const_iterator InsertPt = FirstNonPHI->getIterator();
if (InsertPt->isEHPad())
++InsertPt;
if (isEntryBlock()) {
const_iterator End = end();
while (InsertPt != End &&
(isa<AllocaInst>(*InsertPt) || isa<DbgInfoIntrinsic>(*InsertPt) ||
isa<PseudoProbeInst>(*InsertPt))) {
if (const AllocaInst *AI = dyn_cast<AllocaInst>(&*InsertPt)) {
if (!AI->isStaticAlloca())
break;
}
++InsertPt;
}
}
return InsertPt;
}
void BasicBlock::dropAllReferences() {
for (Instruction &I : *this)
I.dropAllReferences();
}
const BasicBlock *BasicBlock::getSinglePredecessor() const {
const_pred_iterator PI = pred_begin(this), E = pred_end(this);
if (PI == E) return nullptr; // No preds.
const BasicBlock *ThePred = *PI;
++PI;
return (PI == E) ? ThePred : nullptr /*multiple preds*/;
}
const BasicBlock *BasicBlock::getUniquePredecessor() const {
const_pred_iterator PI = pred_begin(this), E = pred_end(this);
if (PI == E) return nullptr; // No preds.
const BasicBlock *PredBB = *PI;
++PI;
for (;PI != E; ++PI) {
if (*PI != PredBB)
return nullptr;
// The same predecessor appears multiple times in the predecessor list.
// This is OK.
}
return PredBB;
}
bool BasicBlock::hasNPredecessors(unsigned N) const {
return hasNItems(pred_begin(this), pred_end(this), N);
}
bool BasicBlock::hasNPredecessorsOrMore(unsigned N) const {
return hasNItemsOrMore(pred_begin(this), pred_end(this), N);
}
const BasicBlock *BasicBlock::getSingleSuccessor() const {
const_succ_iterator SI = succ_begin(this), E = succ_end(this);
if (SI == E) return nullptr; // no successors
const BasicBlock *TheSucc = *SI;
++SI;
return (SI == E) ? TheSucc : nullptr /* multiple successors */;
}
const BasicBlock *BasicBlock::getUniqueSuccessor() const {
const_succ_iterator SI = succ_begin(this), E = succ_end(this);
if (SI == E) return nullptr; // No successors
const BasicBlock *SuccBB = *SI;
++SI;
for (;SI != E; ++SI) {
if (*SI != SuccBB)
return nullptr;
// The same successor appears multiple times in the successor list.
// This is OK.
}
return SuccBB;
}
iterator_range<BasicBlock::phi_iterator> BasicBlock::phis() {
PHINode *P = empty() ? nullptr : dyn_cast<PHINode>(&*begin());
return make_range<phi_iterator>(P, nullptr);
}
void BasicBlock::removePredecessor(BasicBlock *Pred,
bool KeepOneInputPHIs) {
// Use hasNUsesOrMore to bound the cost of this assertion for complex CFGs.
assert((hasNUsesOrMore(16) || llvm::is_contained(predecessors(this), Pred)) &&
"Pred is not a predecessor!");
// Return early if there are no PHI nodes to update.
if (empty() || !isa<PHINode>(begin()))
return;
unsigned NumPreds = cast<PHINode>(front()).getNumIncomingValues();
for (PHINode &Phi : make_early_inc_range(phis())) {
Phi.removeIncomingValue(Pred, !KeepOneInputPHIs);
if (KeepOneInputPHIs)
continue;
// If we have a single predecessor, removeIncomingValue may have erased the
// PHI node itself.
if (NumPreds == 1)
continue;
// Try to replace the PHI node with a constant value.
if (Value *PhiConstant = Phi.hasConstantValue()) {
Phi.replaceAllUsesWith(PhiConstant);
Phi.eraseFromParent();
}
}
}
bool BasicBlock::canSplitPredecessors() const {
const Instruction *FirstNonPHI = getFirstNonPHI();
if (isa<LandingPadInst>(FirstNonPHI))
return true;
// This is perhaps a little conservative because constructs like
// CleanupBlockInst are pretty easy to split. However, SplitBlockPredecessors
// cannot handle such things just yet.
if (FirstNonPHI->isEHPad())
return false;
return true;
}
bool BasicBlock::isLegalToHoistInto() const {
auto *Term = getTerminator();
// No terminator means the block is under construction.
if (!Term)
return true;
// If the block has no successors, there can be no instructions to hoist.
assert(Term->getNumSuccessors() > 0);
// Instructions should not be hoisted across exception handling boundaries.
return !Term->isExceptionalTerminator();
}
bool BasicBlock::isEntryBlock() const {
const Function *F = getParent();
assert(F && "Block must have a parent function to use this API");
return this == &F->getEntryBlock();
}
BasicBlock *BasicBlock::splitBasicBlock(iterator I, const Twine &BBName,
bool Before) {
if (Before)
return splitBasicBlockBefore(I, BBName);
assert(getTerminator() && "Can't use splitBasicBlock on degenerate BB!");
assert(I != InstList.end() &&
"Trying to get me to create degenerate basic block!");
BasicBlock *New = BasicBlock::Create(getContext(), BBName, getParent(),
this->getNextNode());
// Save DebugLoc of split point before invalidating iterator.
DebugLoc Loc = I->getDebugLoc();
// Move all of the specified instructions from the original basic block into
// the new basic block.
New->getInstList().splice(New->end(), this->getInstList(), I, end());
// Add a branch instruction to the newly formed basic block.
BranchInst *BI = BranchInst::Create(New, this);
BI->setDebugLoc(Loc);
// Now we must loop through all of the successors of the New block (which
// _were_ the successors of the 'this' block), and update any PHI nodes in
// successors. If there were PHI nodes in the successors, then they need to
// know that incoming branches will be from New, not from Old (this).
//
New->replaceSuccessorsPhiUsesWith(this, New);
return New;
}
BasicBlock *BasicBlock::splitBasicBlockBefore(iterator I, const Twine &BBName) {
assert(getTerminator() &&
"Can't use splitBasicBlockBefore on degenerate BB!");
assert(I != InstList.end() &&
"Trying to get me to create degenerate basic block!");
assert((!isa<PHINode>(*I) || getSinglePredecessor()) &&
"cannot split on multi incoming phis");
BasicBlock *New = BasicBlock::Create(getContext(), BBName, getParent(), this);
// Save DebugLoc of split point before invalidating iterator.
DebugLoc Loc = I->getDebugLoc();
// Move all of the specified instructions from the original basic block into
// the new basic block.
New->getInstList().splice(New->end(), this->getInstList(), begin(), I);
// Loop through all of the predecessors of the 'this' block (which will be the
// predecessors of the New block), replace the specified successor 'this'
// block to point at the New block and update any PHI nodes in 'this' block.
// If there were PHI nodes in 'this' block, the PHI nodes are updated
// to reflect that the incoming branches will be from the New block and not
// from predecessors of the 'this' block.
// Save predecessors to separate vector before modifying them.
SmallVector<BasicBlock *, 4> Predecessors;
for (BasicBlock *Pred : predecessors(this))
Predecessors.push_back(Pred);
for (BasicBlock *Pred : Predecessors) {
Instruction *TI = Pred->getTerminator();
TI->replaceSuccessorWith(this, New);
this->replacePhiUsesWith(Pred, New);
}
// Add a branch instruction from "New" to "this" Block.
BranchInst *BI = BranchInst::Create(this, New);
BI->setDebugLoc(Loc);
return New;
}
void BasicBlock::splice(BasicBlock::iterator ToIt, BasicBlock *FromBB,
BasicBlock::iterator FromBeginIt,
BasicBlock::iterator FromEndIt) {
#ifdef EXPENSIVE_CHECKS
// Check that FromBeginIt is befor FromEndIt.
auto FromBBEnd = FromBB->end();
for (auto It = FromBeginIt; It != FromEndIt; ++It)
assert(It != FromBBEnd && "FromBeginIt not before FromEndIt!");
#endif // EXPENSIVE_CHECKS
getInstList().splice(ToIt, FromBB->getInstList(), FromBeginIt, FromEndIt);
}
BasicBlock::iterator BasicBlock::erase(BasicBlock::iterator FromIt,
BasicBlock::iterator ToIt) {
return getInstList().erase(FromIt, ToIt);
}
void BasicBlock::replacePhiUsesWith(BasicBlock *Old, BasicBlock *New) {
// N.B. This might not be a complete BasicBlock, so don't assume
// that it ends with a non-phi instruction.
for (Instruction &I : *this) {
PHINode *PN = dyn_cast<PHINode>(&I);
if (!PN)
break;
PN->replaceIncomingBlockWith(Old, New);
}
}
void BasicBlock::replaceSuccessorsPhiUsesWith(BasicBlock *Old,
BasicBlock *New) {
Instruction *TI = getTerminator();
if (!TI)
// Cope with being called on a BasicBlock that doesn't have a terminator
// yet. Clang's CodeGenFunction::EmitReturnBlock() likes to do this.
return;
for (BasicBlock *Succ : successors(TI))
Succ->replacePhiUsesWith(Old, New);
}
void BasicBlock::replaceSuccessorsPhiUsesWith(BasicBlock *New) {
this->replaceSuccessorsPhiUsesWith(this, New);
}
bool BasicBlock::isLandingPad() const {
return isa<LandingPadInst>(getFirstNonPHI());
}
const LandingPadInst *BasicBlock::getLandingPadInst() const {
return dyn_cast<LandingPadInst>(getFirstNonPHI());
}
Optional<uint64_t> BasicBlock::getIrrLoopHeaderWeight() const {
const Instruction *TI = getTerminator();
if (MDNode *MDIrrLoopHeader =
TI->getMetadata(LLVMContext::MD_irr_loop)) {
MDString *MDName = cast<MDString>(MDIrrLoopHeader->getOperand(0));
if (MDName->getString().equals("loop_header_weight")) {
auto *CI = mdconst::extract<ConstantInt>(MDIrrLoopHeader->getOperand(1));
return Optional<uint64_t>(CI->getValue().getZExtValue());
}
}
return None;
}
BasicBlock::iterator llvm::skipDebugIntrinsics(BasicBlock::iterator It) {
while (isa<DbgInfoIntrinsic>(It))
++It;
return It;
}
void BasicBlock::renumberInstructions() {
unsigned Order = 0;
for (Instruction &I : *this)
I.Order = Order++;
// Set the bit to indicate that the instruction order valid and cached.
BasicBlockBits Bits = getBasicBlockBits();
Bits.InstrOrderValid = true;
setBasicBlockBits(Bits);
NumInstrRenumberings++;
}
#ifndef NDEBUG
/// In asserts builds, this checks the numbering. In non-asserts builds, it
/// is defined as a no-op inline function in BasicBlock.h.
void BasicBlock::validateInstrOrdering() const {
if (!isInstrOrderValid())
return;
const Instruction *Prev = nullptr;
for (const Instruction &I : *this) {
assert((!Prev || Prev->comesBefore(&I)) &&
"cached instruction ordering is incorrect");
Prev = &I;
}
}
#endif